Quantification of Target Molecules Needed To Detect Microorganisms

APPLIED AND ENVIRONMENTAL MICROBIOLOGY, Aug. 2008, p. 5068–5077 0099-2240/08/$08.00⫹0 doi:10.1128/AEM.00208-08 Copyright © 2008, American Society for Microbiology. All Rights Reserved.

Vol. 74, No. 16

Quantification of Target Molecules Needed To Detect Microorganisms by Fluorescence In Situ Hybridization (FISH) and Catalyzed Reporter Deposition-FISH䌤 Tatsuhiko Hoshino,1 L. Safak Yilmaz,2 Daniel R. Noguera,2 Holger Daims,1* and Michael Wagner1 ¨ kologie, Universita Department fu ¨r Mikrobielle O ¨t Wien, Althanstrasse 14, A-1090 Wien, Austria,1 and Department of Civil and Environmental Engineering, University of Wisconsin, 1415 Engineering Dr., Madison, Wisconsin 537062 Received 23 January 2008/Accepted 8 June 2008

Fluorescence in situ hybridization (FISH) with rRNA-targeted oligonucleotide probes is a method that is widely used to detect and quantify microorganisms in environmental samples and medical specimens by fluorescence microscopy. Difficulties with FISH arise if the rRNA content of the probe target organisms is low, causing dim fluorescence signals that are not detectable against the background fluorescence. This limitation is ameliorated by technical modifications such as catalyzed reporter deposition (CARD)-FISH, but the minimal numbers of rRNA copies needed to obtain a visible signal of a microbial cell after FISH or CARD-FISH have not been determined previously. In this study, a novel competitive FISH approach was developed and used to determine, based on a thermodynamic model of probe competition, the numbers of 16S rRNA copies per cell required to detect bacteria by FISH and CARD-FISH with oligonucleotide probes in mixed pure cultures and in activated sludge. The detection limits of conventional FISH with Cy3-labeled probe EUB338-I were found to be 370 ⴞ 45 16S rRNA molecules per cell for Escherichia coli hybridized on glass microscope slides and 1,400 ⴞ 170 16S rRNA copies per E. coli cell in activated sludge. For CARD-FISH the values ranged from 8.9 ⴞ 1.5 to 14 ⴞ 2 and from 36 ⴞ 6 to 54 ⴞ 7 16S rRNA molecules per cell, respectively, indicating that the sensitivity of CARD-FISH was 26- to 41-fold higher than that of conventional FISH. These results suggest that optimized FISH protocols using oligonucleotide probes could be suitable for more recent applications of FISH (for example, to detect mRNA in situ in microbial cells).

different reasons (for a review, see reference 43). One prevalent problem occurs when targeted organisms have a low cellular rRNA content, which causes a probe-stained microbial cell to emit only weak fluorescence that may be invisible to the naked eye when the sample is viewed under a microscope. Electronic signal amplification, as performed by digital cameras or confocal laser scanning microscope (CLSM) detectors, can overcome this obstacle if the signal-to-noise ratio (S/N ratio) is high enough to distinguish the amplified fluorescent signals of the cells from the (also amplified) background autofluorescence. Unfortunately, many kinds of environmental samples contain large amounts of autofluorescent matter, which hampers the detection of dim probe-stained cells. Cells with low rRNA contents are common in samples from oligotrophic environments (31), where many microorganisms either have low metabolic activity or persist in a dormant state. Several methods to improve the sensitivity of FISH in such situations have been developed (3, 17, 28, 40, 49). A very elegant, albeit relatively time-consuming, technique is FISH combined with tyramide signal amplification (TSA), which exploits peroxidase-labeled probes. In combination with an additional cell permeabilization step this method has found wide application as catalyzed reporter deposition (CARD)-FISH (30, 37, 44) in microbial ecology. Despite these technical improvements, the minimal numbers of rRNA target molecules required to obtain a visible fluorescence signal after FISH or CARD-FISH with rRNA-targeted probes have not been determined yet. These numbers are especially relevant for further modifications and new applica-

Natural ecosystems teem with microbial life, and most microbes have not yet been isolated and studied by traditional microbiology methods. Among the molecular approaches that enable cultivation-independent characterization of microorganisms, fluorescence in situ hybridization (FISH) with rRNAtargeted oligonucleotide probes (4, 14) has been one of the most powerful and widely used techniques (1). Thanks to its intimate link with rRNA-based phylogeny, FISH can specifically detect, identify, and quantify microbes at a wide range of phylogenetic levels, ranging from whole domains to specific genera and species. Furthermore, FISH has been key to analyzing the spatial organization of microbial communities (24, 26) and has been combined with other techniques for studying the physiology of uncultured microbes in situ (18, 22, 23, 27, 38). The success of FISH experiments in determining microbial community structures is often measured as the fraction of detectable microbial cells. Usually, this fraction is determined by comparing the number of microorganisms detected by FISH with a universal or Bacteria-specific probe labeled with a single fluorescent dye to the total number of cells labeled by a nonspecific nucleic acid stain (16). In many ecosystems this fraction is well below 50% (see reference 11 and references therein). FISH can fail to detect microbes for a number of * Corresponding author. Mailing address: Department fu ¨r Mikro¨ kologie, Universita¨t Wien, Althanstrasse 14, A-1090 Wien, bielle O Austria. Phone: 43 1 4277 54392. Fax: 43 1 4277 54389. E-mail: daims @microbial-ecology.net. 䌤 Published ahead of print on 13 June 2008. 5068

VOL. 74, 2008

NUMBERS OF TARGET MOLECULES NEEDED FOR FISH

tions of FISH. Different FISH protocols are usually evaluated and compared based on fractions of detectable cells (see above) and on the fluorescence intensities, which can be expressed as relative units (RU) when digital image analysis is used (11, 31). The main disadvantage of these relative measures is that the same samples and imaging setup must be used to compare the results obtained. In contrast, the required minimal rRNA content would be a direct and absolute measure of sensitivity. After thorough experimental verification, this parameter could be used to compare different FISH and signal amplification protocols even between laboratories using different samples and imaging equipment. Furthermore, the required rRNA content would indicate whether a particular FISH protocol has the potential to detect targets other than rRNA, such as mRNA, whose copy numbers are much lower than those of rRNA in bacterial cells (only about 2% of the total RNA in Escherichia coli cells is mRNA [6]). Determining the rRNA content of probe-stained cells is not a straightforward task. In this study we developed a novel approach to measure, based on FISH experiments and the theory of hybridization thermodynamics, the numbers of 16S rRNA copies needed to detect the cells of a bacterial probetargeted population. Our approach does not distinguish between 16S rRNA in mature ribosomes and 16S rRNA in precursor transcripts, because both types of molecules are targets for FISH (36). This method was used with E. coli cells to determine and compare the sensitivities of conventional FISH and CARD-FISH. MATERIALS AND METHODS Bacterial strains, growth conditions, and cell fixation. E. coli strain JM109 was grown in 250-ml flasks in Luria-Bertani medium for up to 26 h at 37°C. The cultures were shaken at 200 rpm. Growth was monitored by measuring the optical density at 600 nm, and cells were harvested at different stages during growth. Verrucomicrobium spinosum was grown as described by Daims et al. (11). Cells of each species were collected by centrifugation (10 min, 7,000 ⫻ g) and fixed for 8 h in a 3% formaldehyde solution as described elsewhere (13). Fixed cells were suspended in a 1:1 mixture of 1⫻ phosphate-buffered saline (PBS) and 96% (vol/vol) ethanol and stored at ⫺20°C. Oligonucleotide probes, FISH, and CARD-FISH. Probe EUB338-I (2) targets most known Bacteria and has been used in many studies where FISH was used to detect environmental bacteria. Therefore, we chose this probe to determine the required rRNA copy numbers in most experiments; the exceptions were the experiments performed with activated sludge, where probe Eco681 (15) was used to specifically detect E. coli. Probe EUB338-I was 5⬘ labeled with the sulfoindocyanine dye Cy3 or Cy5 or with horseradish peroxidase (HRP). Probe Eco681 was labeled with Cy3 or Cy5. Labeled probes were obtained from Thermo Hybaid (Interactiva Division, Ulm, Germany). Conventional FISH of biomass immobilized on slides was performed as described by Daims et al. (13) in hybridization buffers containing 40% (vol/vol) formamide (FA) (probe EUB338-I) or 10% (vol/vol) FA (probe Eco681). The optimal FA concentrations for the probes were determined by recording the probe dissociation profiles with increasing FA concentrations (see Fig. 3; data not shown for Eco681). The hybridization time was 24 h (EUB338-I) or 3 h (Eco681). For conventional FISH of suspended cells (45), fixed E. coli cell suspensions were dehydrated in 50, 80, and 96% (vol/vol) ethanol. Between the ethanol steps the cells were centrifuged, and the supernatant was removed. The dehydrated cells were resuspended in 200-␮l PCR tubes in hybridization buffer with 40% (vol/vol) FA. The probe concentration in the hybridization buffer was adjusted to values between 0.01 and 0.5 ␮M. After 24 h of hybridization at 46°C, the cells were centrifuged (5 min, 7,000 ⫻ g), resuspended in prewarmed (48°C) washing buffer, and incubated for 15 min at 48°C. The composition of the washing buffer was the same as the composition of the buffer for FISH on slides. Subsequently, the cells were centrifuged again, the supernatant was discarded, and the cells were washed once in ice-cold 1⫻ PBS. Following another centrifugation step, the cells were finally resuspended in ice-cold 1⫻ PBS. To measure the fluorescence

5069

intensity of suspended cells after FISH, the cells were dispersed by sonication on ice. Almost complete dispersion (i.e., disruption of cell aggregates) was confirmed by microscopy. The fluorescence intensity of the sonicated suspension was measured with an ND-3300 fluorospectrometer (NanoDrop Technologies, Wilmington, DE). The fluorescence intensities of hybridized and sonicated cell suspensions were also measured in an independent experiment that was performed without any washing step after the hybridization. CARD-FISH of E. coli cells was performed on slides. To improve the comparability to conventional FISH, the hybridization temperature for CARD-FISH was adjusted to 46°C instead of the 35°C used in the original CARD-FISH protocol with HRP-labeled oligonucleotide probes (30). This higher temperature, which also required a shorter hybridization time to avoid a loss of HRP activity, was used with success in a previous study (41). The CARD-FISH protocol used was the protocol described by Sekar et al. (39), with the following modifications. All CARD-FISH experiments were performed on glass slides instead of membrane filters. The cells were permeabilized in a 3-mg/ml lysozyme solution at 4°C for 15 min, and achromopeptidase was not used. Subsequently, the cells were incubated in 0.01 M HCl at room temperature for 30 min. The hybridization time was 3 h, and the hybridization buffer contained 20% (vol/vol) FA. The slides were washed at 48°C for 15 min in the same washing buffer that was used for conventional FISH. The cells were incubated in 1⫻ PBS on ice for 10 min and then in “substrate mixture” (1 part of Cy3-labeled tyramide and 200 parts of amplification buffer) at 46°C for 45 min. To determine dissociation profiles of Cy3- and Cy5-labeled probe EUB338-I, FISH was performed in hybridization buffers containing 30 to 90% (vol/vol) FA using a hybridization time of 3 h. To determine the dissociation profile of HRP-labeled EUB338-I, CARD-FISH was performed in hybridization buffers containing 20 to 50% FA. To determine the dissociation profile of Cy3- and Cy5-labeled probe Eco681, FISH was performed with 0 to 60% FA. Microscopy, cell counting, and digital image analysis. The cell concentrations of fixed E. coli cell suspensions were determined by direct visual cell counting. For this purpose, the suspended cells were stained at room temperature for 5 min with a 0.1% (wt/vol) solution of 4,6-diamidino-2-phenylindole (DAPI). Subsequently, the suspension was diluted 1:500, and 5 ml of the diluted suspension was filtered onto a polycarbonate filter (pore size, 0.2 ␮m; diameter, 47 mm; type GTTP; Millipore Corp., Bedford, MA). The immobilized cells on the filter were observed by confocal microscopy (see below). In each of four replicate experiments, 20 images were recorded, and the fluorescent cells in each image were counted (every image contained approximately 200 cells). Finally, the cell concentration in the original suspension was calculated from the average number of cells per image, the known area of each image (in ␮m2), the known area of the filter, the filtered volume of the diluted suspension, and the dilution factor. Images of probe-stained cells on glass microscope slides and on polycarbonate filters were recorded with a CLSM (LSM 510 Meta; Zeiss, Oberkochen, Germany) by using two HeNe lasers (543 and 633 nm) for detection of Cy3 and Cy5, respectively. The fluorescence intensity of the probe-stained cells (mean pixel brightness per cell) in confocal images was measured, and probe dissociation curves were obtained by using the digital image analysis program daime (12). Competitive hybridization with Cy3- and Cy5-labeled probes for measuring the sensitivity of conventional FISH. The sensitivity of conventional FISH (in terms of the required 16S rRNA copy number) was determined by using a competitive hybridization approach. The theoretical basis of this approach is described in the Results and Discussion. First, the ratio of the equilibrium constants for Cy3- and Cy5-labeled probes (KCy3/KCy5) was determined by carrying out competitive FISH of the E. coli pure culture diluted 1:100 on glass microscope slides. The hybridization buffers contained Cy3- and Cy5-labeled EUB338-I at Cy3/(Cy3 ⫹ Cy5) ratios of 0.2 to 1. The total probe concentration in the buffers was 0.5 ␮M. Following FISH, images of Cy3 probe-labeled cells were recorded, and the fluorescence intensity conferred by the Cy3-labeled probe was measured by image analysis. At least 200 cells were measured for each ratio of the Cy3- and Cy5-labeled probes. In a subsequent experiment, aliquots (10 ␮l) of a mixture of fixed E. coli (diluted 1:100) and V. spinosum (diluted 1:50) cells were immobilized on glass slides. Competitive FISH was then carried out with Cy3- and Cy5-labeled EUB338-I at Cy3/(Cy3 ⫹ Cy5) ratios of 0.001 to 0.1. The total probe concentration in the hybridization buffers was 0.5 ␮M. After FISH the slides were observed by using the CLSM in epifluorescence mode with the optical filter set for Cy3. The lowest probe ratio (i.e., the smallest amount of Cy3-labeled probe) that yielded a clearly visible fluorescence signal for E. coli was noted. Activated sludge samples were obtained from an animal waste-rendering wastewater treatment plant (Plattling, Germany) and fixed immediately in a 3% formaldehyde solution. A suspension of fixed E. coli cells was diluted 1:50 with fixed activated sludge, and 10-␮l aliquots of this mixture were immobilized on

5070

HOSHINO ET AL.

FIG. 1. Optical density at 600 nm (䢇) and intensity of probeconferred fluorescence (per cell) after FISH with Cy3-labeled probe EUB338-I (䡺) during the growth phases of an E. coli batch culture. The arrows indicate the growth phases (arrow 1, lag phase; arrow 2, logarithmic phase; arrow 3, stationary phase). One experiment was performed; for each data point the fluorescence of at least 200 cells was measured. OD600, optical density at 600 nm.

glass microscope slides for competitive FISH with Cy3- and Cy5-labeled Eco681 at Cy3/(Cy3 ⫹ Cy5) ratios of 0.001 to 0.1 and with a total probe concentration of 0.5 ␮M in the hybridization buffers. Following FISH, confocal images of the Cy3 signal were recorded by using the same detector settings that were used in the experiments with E. coli pure cultures and probe EUB338-I. The intensity of the fluorescence signal was measured by image analysis in the images where the probe-stained E. coli cells were clearly visible against the autofluorescent background of activated sludge. Competitive hybridization with Cy- and HRP-labeled probes for measuring the sensitivity of CARD-FISH. To determine the ratio of equilibrium constants for Cy3- and HRP-labeled probes (KCy3/KHRP) (see Results and Discussion), we performed competitive hybridizations with E. coli pure-culture cells diluted 1:100 and Cy3- and HRP-labeled EUB338-I. The Cy3/(Cy3 ⫹ HRP) probe ratio was adjusted to values ranging from 0.01 to 1. These competitive hybridizations were carried out using the CARD-FISH protocol, except that the detection step of CARD-FISH (i.e., addition of tyramide) was omitted, because only the fluorescence intensity conferred by the Cy3-labeled probe was measured after the hybridizations. The measured fluorescence intensities were plotted, and curve fitting based on equation 13 (see below) was performed with Sigma Plot 8.0 (Systat Software, San Jose, CA). To determine the detection limit of CARD-FISH, probe EUB338-I labeled with Cy5 instead of Cy3 was used as a competitor for the HRP-labeled probe. Cy5 was required because Cy3-labeled tyramide had to be used for CARD-FISH to maintain comparability to the results of the conventional FISH experiments, which had been performed by using Cy3-labeled probes. The Cy5/(Cy5 ⫹ HRP) probe ratio was adjusted to values ranging from 0.003 to 0.06. CARD-FISH and the detection step were carried out as described above.

RESULTS AND DISCUSSION Dependence of fluorescence intensity (i.e., 16S rRNA content) on the growth phase of E. coli. FISH with Cy3-labeled EUB338-I and E. coli cells which had been harvested at different stages in the growth curve showed that the intensity of probe-conferred fluorescence decreased during the log phase and reached a constant, low value in the stationary phase (Fig. 1). Theoretically, the lower intensities might be explained by reduced permeability of the cells due to structural changes in the cell wall in the stationary phase. However, this seems unlikely as E. coli is a gram-negative bacterium and problems with fluorochrome-labeled oligonucleotide probes and cell permeabilization are restricted mostly to gram-positive organ-

APPL. ENVIRON. MICROBIOL.

FIG. 2. Intensity of probe-conferred fluorescence of a sonicated E. coli cell suspension with (E) and without (䡺) the washing steps after FISH with Cy3-labeled probe EUB338-I. The fluorescence intensities were measured by fluorimetry. (Inset) Data for probe concentrations between 0 and 0.1 ␮M, showing that the fluorescence without washing also increased linearly at these low probe concentrations. The error bars indicate the standard deviations (n ⫽ 3), and error bars smaller than the symbols are not shown.

isms and Archaea (9, 44). Thus, the decrease in fluorescence intensity was most likely caused by a decrease in the cellular 16S rRNA content. This is consistent with previous studies that demonstrated the use of FISH with oligonucleotide probes for monitoring changes in the 16S rRNA content of bacteria (7, 33). All subsequent experiments were performed with stationary-phase E. coli cells that were collected at 26 h. After fixation and resuspension of the cells, the concentration of E. coli was 1.26 ⫻ 109 ⫾ 0.15 ⫻ 109 cells/ml. Average cellular 16S rRNA content of E. coli. Aliquots of stationary-phase E. coli cells were hybridized in suspension with different concentrations (0.01 to 0.5 ␮M) of Cy3-labeled EUB338-I. After the hybridization and washing steps, the cells were sonicated, and the fluorescence intensities of the suspensions were measured. Parallel experiments were performed without the washing steps after the hybridizations. The fluorescence measured in the experiments with the washing steps was conferred only by probes bound to 16S rRNA, whereas the fluorescence measured in the experiments without the washing steps was emitted by probes that had bound to 16S rRNA and also by remaining probe molecules in the supernatant. The fluorescence intensities measured with and without the washing steps were plotted against the known initial probe concentrations in the hybridization buffers. The curve obtained in the experiment without washing was linear over the whole range of probe concentrations (Fig. 2). The linearity of this curve demonstrates that there was no significant difference between the intensities of fluorescence emitted by hybridized and nonhybridized probe molecules and that possible quenching or other negative effects of cell material were negligible in this experiment. This curve was used as a standard curve for determining the 16S rRNA content, as follows. In the experiment with washing, the fluorescence intensity

VOL. 74, 2008


reached a constant value (239 ⫾ 15 RU) at the higher probe concentrations, indicating that all available binding sites on 16S rRNA molecules were occupied by probes (Fig. 2). This intensity corresponds to a total probe concentration of 0.076 ⫾ 0.005 ␮M in the standard curve (i.e., in the experiment without washing). Based on Avogadro’s number (6.022 ⫻ 1023 molecules per mol), this probe concentration could be converted to the absolute number of probe molecules that had bound to 16S rRNA (4.56 ⫻ 1013 ⫾ 0.28 ⫻ 1013 molecules ml⫺1). Then, dividing this probe concentration by the cell concentration of the suspension, we found that the average 16S rRNA content per cell was 37,000 ⫾ 4,500 copies. Bremer and Dennis (8) inferred the ribosome content of E. coli from other measured parameters and found that cells from a fast-growing culture (doubling time, 24 min) contained about 72,000 ribosomes/cell, whereas cells from a slow-growing culture (doubling time, 100 min) contained only about 6,800 ribosomes/cell. Based on these numbers, one would expect stationary-phase cells to contain a lower number of 16S rRNA copies than the number determined by our method. However, the rate of ribosome degradation in starved E. coli cells depends strongly on the composition of the nutrient medium (19). Ramagopal (34) isolated ribosomes from E. coli grown in Luria-Bertani medium and showed (by directly quantifying the ribosomes) that the ribosome content at 24 h after inoculation was as high as 82% of the ribosome content in the logarithmic growth phase. In our study, E. coli was also grown in Luria-Bertani medium, and the cells used to determine the 16S rRNA content were harvested after 26 h. Furthermore, ribosome degradation leads to a transient accumulation of high-molecular-weight RNAcontaining particles in starved E. coli cells (25). We cannot exclude the possibility that rRNA-targeted probes also hybridized to such degradation intermediates. The spatial structure of partially degraded rRNA may differ from the structure in intact ribosomes, and structural modifications, such as dimerization (42), occur in mature ribosomes at different growth stages of E. coli. Therefore, it is difficult to predict which fraction of the 16S rRNA molecules was accessible to the oligonucleotide probe, and the 16S rRNA copy number reported here should be considered a minimum estimate of the average copy number in the cells analyzed. Theoretical framework for competitive hybridizations. Our approach for quantifying the sensitivity of FISH relied on the competition between differently labeled probes targeting the same site on the 16S rRNA. Below we describe the thermodynamic model for these competition experiments, which is based on models used previously for predicting the affinity of FISH probes (47, 48) and for simulating probe dissociation profiles (46). The competitive hybridization of two probes can be described by equations 1 and 2, where P, T, and H are the probe, target, and probe/target hybrid, respectively (subscripts 1 and 2 indicate the competing probes). In FISH, the probe and target can have folded and unfolded conformations (47), both of which are included in the definitions of P and T in equations 1 and 2 but are not shown for the sake of simplicity. If probes are added in excess of the target {i.e., [P1]0 ⬇ [P2]0 ⬎⬎ [T]0}, then [P1]0 ⫽ [P1] ⫹ [H1] ⬇ [P1] and [P2]0 ⫽ [P2] ⫹ [H2] ⬇ [P2], where [P1]0 and [P2]0 are the initial probe concentrations in the hybridization buffer. Thus, the equilibrium constants, K1 and

5071

K2, of the hybridizations can be described by equations 3 and 4. These constants correspond to the overall equilibrium constants of the main processes that occur during in situ hybridization (47). Dividing equation 3 by equation 4 and rearranging result in equation 5, which describes the ratio of the hybrids. P1 ⫹ T 7 H1

(1)

P2 ⫹ T 7 H2

(2)

K1 ⫽

关H1兴关P1兴0关T兴

(3)

K2 ⫽

关H2兴关P2兴0关T兴

(4)

关H1兴 K1关P1兴0 ⫽ 关H2兴 K2关P2兴0

(5)

In competitive hybridizations, the experimental response is obtained as the fluorescence of one of the probes. Let this probe be probe P1. To link the response of P1 to ribosome quantity, we started with the mass balance of probe and target as described by equations 6 and 7, where [PT] and [TT] are the total probe and target concentrations, respectively. The partitioning of total target into free (T) and hybridized (H1 and H2) forms (equation 7) does not always allow a unique solution for the intended derivation of ribosome quantity from the experimental response. However, in the ideal case, when the rRNA targets are nearly saturated {i.e., [T] ⬇ 0; hence, [T] ⬍⬍ [H1] ⫹ [H2]}, equation 7 simplifies to equation 8. Then it is possible to determine the fraction of targets hybridized to the fluorescent probe ([H1]/[TT]) as a function of equilibrium constants and probe concentrations only, as shown in equation 9. The rightmost term of this equation is obtained by substituting [H1] and [H2] from the corresponding expressions in equations 3 and 4. 关PT兴 ⫽ 关P1兴0 ⫹ 关P2兴0

(6)

关TT兴 ⫽ 关T兴 ⫹ 关H1兴 ⫹ 关H2兴

(7)

关TT兴 ⬇ 关H1兴 ⫹ 关H2兴

(8)

关H1兴关H1兴 K1关P1兴0 ⬇ ⫽ 关TT兴关H1兴 ⫹ 关H2兴 K1关P1兴0 ⫹ K2关P2兴0

(9)

Equation 9 can be rearranged in two steps. First, dividing both the numerator and the denominator on the rightmost side by [PT] results in equation 10, where R is the probe ratio (i.e., Ri ⫽ [Pi]0/[PT]). Second, equation 11 is obtained by substituting R2 ⫽ 1 ⫺ R1 in equation 10, dividing the nominator and denominator by K2, and rearranging. This equation allows quantification of rRNA targets from competitive hybridization experiments if the K1/K2 ratio is known. K 1R 1 关H1兴 ⬇ 关TT兴 K1R1 ⫹ K2R2 关H1兴 ⬇ 关TT兴

K1 R K2 1 K1 1⫹ ⫺ 1 R1 K2

冉

冊

(10)

(11)

Finally, assuming that fluorescence intensity (F) is propor-

5072

HOSHINO ET AL.


FIG. 3. Dissociation profiles of Cy3-labeled (䢇 and solid line), Cy5-labeled (E and dashed and dotted line), and HRP-labeled (䡺 and dashed line) probe EUB338-I after conventional FISH (Cy3 and Cy5) or CARD-FISH (HRP) with a hybridization temperature of 46°C and a washing temperature of 48°C. The profiles were determined with E. coli cells in hybridization experiments in which increasingly stringent hybridization and washing conditions were used. For each data point the mean fluorescence intensity of at least 1,000 cells was determined. Error bars (standard deviations; n ⫽ 3) are not shown as they are smaller than the symbols. Fluorescence intensities were normalized by defining the intensity measured with the lowest FA concentration used in each experiment as 1.

tional to the fraction of targets hybridized to the fluorescent probe (i.e., [H1]/[TT]), it can be described by equation 12, where ␤ is a proportionality factor and ε is the background fluorescence. Combining equations 11 and 12 results in equation 13, which allows quantification of the K1/K2 ratio from competitive hybridization experiments. However, the critical assumption that nearly all rRNAs are hybridized with probes (equation 8) must be fulfilled when equations 11 and 13 are used. This assumption is true if two conditions are met: (i) the amount of at least one probe is greater than the amount of the target, and (ii) this probe is at the high-fluorescence plateau of its dissociation profile. F⫽␤

关H1兴 ⫹ε 关TT兴

K1 R K2 1 F⫽␤ ⫹ε K1 1⫹ ⫺ 1 R1 K2

冉

冊

(12)

(13)

FA dissociation profiles. The molecular weight of HRPlabeled oligonucleotide probes is approximately 10 times higher than that of Cy-labeled probes. Therefore, the K values of the probes were expected to differ significantly, which would lead to different probe dissociation profiles. Since the hybridization stringency had to be adjusted to ensure that the probes were at their high-fluorescence plateaus, FA dissociation profiles of Cy- and HRP-labeled EUB338-I probes were determined and compared after conventional FISH (Cy3 and Cy5) and CARD-FISH (HRP). As shown in Fig. 3, a high-fluorescence plateau for both Cy-labeled probes was observed with FA concentrations up to 50%. In contrast, the signal of the HRP-labeled probe decreased rapidly at FA concentrations

above 30% and disappeared at an FA concentration of 50%. To verify that the rapid decrease in signal intensity was not caused by inhibition of HRP by higher FA concentrations, we also determined the dissociation profile of the HRP-labeled probe at room temperature. In this experiment the signal intensities were higher than those at 46°C, confirming that HRP was not inhibited by FA and that the observed decrease in the signal intensity at 46°C was caused by probe-target dissociation (data not shown). Consequently, to ensure that the mathematical modeling was valid, competition between Cy3 and Cy5 was examined using 40% FA, while competition between Cy-labeled probes and HRP-labeled probes was examined using 20% FA. Note that the required minimal numbers of 16S rRNA copies found in this study for conventional FISH and CARD-FISH could be accurately determined despite the different FA concentrations used (40% versus 20%), because in all experiments all probes were at the plateaus of their dissociation profiles. Estimation of equilibrium constant ratios. Since the molecular weights and the chemical structures of Cy3- and Cy5labeled probes are comparable, it was anticipated that the KCy3 and KCy5 equilibrium constants would be similar. According to equation 13, if this condition is met, then the intensity of fluorescence conferred by a Cy3-labeled probe should linearly depend on the fraction of the Cy3-labeled probe (R1 ⫽ Cy3/ [Cy3 ⫹ Cy5]) in the probe mixture. Figure 4A shows the results of competitive hybridizations with E. coli cells and mixtures of Cy3- and Cy5-labeled EUB338-I probes, confirming that there is a linear correlation between R1 and fluorescence intensity. Therefore, we consider the ratio of KCy3 to KCy5 to be 1. From equation 5, it also follows that in all competitive hybridization experiments with Cy3- and Cy5-labeled probes, the ratio of hybridization products was equal to the ratio of the initial concentrations of the two probes. The KCy3/KHRP ratio was also determined experimentally by competitive FISH with Cy3- and HPR-labeled EUB338-I probes (Fig. 4B). A coefficient of determination of 0.995 indicates that the curve fit based on equation 13 matched the experimental data very well. Accordingly, the KCy3/KHRP ratio was found to be 16.2. This is consistent with the probe dissociation profiles shown in Fig. 3 since the lower melting point of the HRP-labeled probe indicates lower stability and hence a smaller equilibrium constant. Sensitivity of conventional FISH. The minimal rRNA content required for FISH depends on the type of environmental sample being analyzed. Planktonic cells immobilized on glass slides or by filtration are observed, after FISH, against a background with a relatively low level of autofluorescence (if no inorganic particles or phototrophic microbes exhibiting strong autofluorescence are present). In contrast, samples such as activated sludge or sediment samples can contain relatively large amounts of fluorescing matter, which makes the observation of probe-stained cells more difficult. Hence, cells with a low rRNA content may be detectable on glass slides or filters but may be overlooked in more complex samples. To determine the required minimal 16S rRNA content of planktonic cells immobilized on glass slides, we performed a competitive FISH experiment with Cy3- and Cy5-labeled EUB338-I probes and a mixture of E. coli and V. spinosum cells. V. spinosum, which is not detected by EUB338-I (11), was used to obtain a

VOL. 74, 2008


5073

FIG. 4. (A) Fluorescence intensities measured in a competitive hybridization experiment (conventional FISH) with E. coli cells and mixtures of Cy3- and Cy5-labeled EUB338-I. The mean intensity of fluorescence (per cell) conferred by the Cy3-labeled probe was quantified. The equation and the coefficient of determination of the regression line are indicated. The experiment was performed to verify that KCy3/KCy5 was 1 (see the text for details). For each data point the fluorescence intensity of at least 200 cells was determined. Error bars (standard deviations) are not shown as they were smaller than the symbols. (B) Fluorescence intensities measured in a competitive hybridization experiment (conventional FISH) with E. coli cells and mixtures of Cy3- and HRP-labeled EUB338-I. The mean intensity of fluorescence (per cell) conferred by the Cy3-labeled probe was quantified. For each data point the fluorescence intensity of at least 200 cells was determined. The equation and the coefficient of determination for the curve fitted based on equation 13 are indicated. Data points obtained for probe ratios (Cy3/[Cy3 ⫹ HRP]) of ⱕ0.02 (‚) were excluded from the regression, because at these ratios the Cy3-labeled probe was not in high excess of the target molecules. Note that no background fluorescence was recorded, because the excitation laser of the CLSM had been set to a very low intensity to prevent bleaching of the sample. Therefore, the regression curve was forced to pass through the origin. The experiment was performed to determine KCy3/KHRP. The error bars indicate the standard deviations (n ⫽ 3). The gray lines indicate the 95% confidence limits.

weakly autofluorescent background like that often seen in nontargeted organisms after FISH on slides or filters. The Cy3 probe-stained E. coli cells could clearly be distinguished by epifluorescence microscopy from the unlabeled V. spinosum cells if the probe ratio (R1 ⫽ Cy3/[Cy3 ⫹ Cy5]) was ⱖ0.01. According to equation 11 and with KCy3 ⫽ KCy5, the R1 ratio also corresponds to the ratio of 16S rRNA molecules hybridized to the Cy3-labeled probe. Thus, based on the previously determined average 16S rRNA content of E. coli (i.e., 37,000 ⫾ 4,500 copies per cell), we found that 370 ⫾ 45 16S rRNA molecules per cell had hybridized to the Cy3-labeled probe when R1 was 0.01. This value can be considered the minimal 16S rRNA content of planktonic cells required for conventional FISH if the cells are observed with a common epifluorescence microscope. The detection limit may be slightly lower under optimal conditions if the background fluorescence is very low and highest-quality microscope equipment is used. To determine the minimal 16S rRNA content required for FISH with a complex environmental sample, activated sludge was spiked with E. coli cells. Subsequently, these cells were detected by FISH and observed against the relatively strong autofluorescence of the activated sludge biomass. However, for this purpose, a probe specific for E. coli (Eco681) had to be used instead of EUB338-I. The Cy3 probe-stained E. coli cells became visible against the fluorescent background with an R1 ratio of 0.07, but a ratio of 0.1 was required to easily spot E. coli cells that were embedded in thick sludge flocs. With an R1 value of 0.1, the intensity of fluorescence conferred by Cy3labeled Eco681 was 59.4 ⫾ 0.8 RU. In general, the fluorescence signal of Eco681 was dimmer than that of EUB338-I, as shown in Fig. 5 as a function of R1. Thus, from Fig. 5 it follows that the fluorescence observed with Eco681 at an R1 value of 0.1 would be achieved with EUB338-I at an R1 value of 0.038. Based on this ratio (0.038), we calculated that 1,400 ⫾ 170 16S

rRNA molecules per cell would be needed to unambiguously detect E. coli cells in activated sludge with the EUB338-I probe. As mentioned above, a probe ratio of at least 0.01 was

FIG. 5. Fluorescence intensities measured in competitive hybridization experiments (conventional FISH). One experiment (E) was done with mixtures of Cy3- and Cy5-labeled EUB338-I hybridized to mixed E. coli and V. spinosum cultures. Data points obtained for probe ratios (Cy3/[Cy3 ⫹ Cy5]) of ⱕ0.02 (‚) were excluded from the regression, because at these ratios the Cy3-labeled probe was not in high excess of the target molecules. In the other experiment (䡺), mixtures of Cy3- and Cy5-labeled Eco681 were hybridized to activated sludge spiked with E. coli cells. The mean intensity of fluorescence (per cell) conferred by the Cy3-labeled probe was quantified. The dashed lines indicate the lowest signal intensities needed to visually detect E. coli cells in the presence of the autofluorescence of V. spinosum (S/N ⫽ 1.5) and to detect E. coli cells against the fluorescent background of activated sludge (S/N ⫽ 3). The equation and the coefficient of determination for the regression line are indicated. For each data point the fluorescence intensity of at least 200 cells was determined. The error bars (standard deviations) were smaller than the symbols and are not shown.

5074

HOSHINO ET AL.

needed to clearly distinguish E. coli from V. spinosum by FISH with probe EUB338-I on glass slides. This probe ratio corresponded to a fluorescence intensity of approximately 30 RU. Based on this value and on a background fluorescence intensity of 20.1 (Fig. 5), an S/N ratio of at least 1.5 was required to detect E. coli. This value is consistent with the results of Pernthaler et al. (32), who found that an S/N ratio of at least 1.3 was needed to detect and automatically count probe-stained planktonic cells on filters. The intensity of the E. coli cells which were detected by probe Eco681 in the activated sludge was approximately 60 RU, which would have corresponded to an S/N ratio of 3 if FISH had been done with a pure culture on glass slides. These S/N ratios were used to estimate the 16S rRNA copy numbers needed for CARD-FISH as described below. Sensitivity of CARD-FISH. As HRP-labeled probe molecules are larger than probes labeled with Cy fluorochromes, microbial cells must be permeabilized by incubating the sample with enzymes such as lysozyme prior to CARD-FISH. This permeabilization step might cause leakage of ribosomes and rRNA precursor molecules from the cells. To check this, FISH with Cy3-labeled EUB338-I was performed using the conventional FISH and CARD-FISH protocols with the same amount of FA (40%) in the hybridization buffer. Only the TSA step of the CARD-FISH protocol was omitted, because a Cy3-labeled probe was used in this experiment. Subsequently, the fluorescence intensities of the probe-stained E. coli cells were measured and compared. Indeed, after the CARD-FISH protocol the average intensity of probe-conferred fluorescence was 67% ⫾ 9% of the value obtained with conventional FISH. Theoretically, the decrease could have been caused by the shorter hybridization time (24 h for conventional FISH versus 3 h for CARD-FISH) and/or by the permeabilization with lysozyme. Therefore, to assess the impact of hybridization time, we performed conventional FISH for 3 and 24 h and found that the shorter hybridization time resulted in only a 10% decrease in the fluorescence intensity. Thus, the larger decrease observed with the CARD-FISH protocol was most likely caused by a reduced 16S rRNA content due to the permeabilization step. The loss of target molecules must be considered in the CARDFISH sensitivity calculations. Therefore, we determined two detection limits for this method. The “practical” detection limit indicates how many target molecules that a cell must contain prior to the permeabilization step to be detectable by CARDFISH. This value was calculated based on the original 16S rRNA content (i.e., 37,000 ⫾ 4,500 copies per E. coli cell). The “ideal” detection limit is the number of remaining target molecules per cell (after the permeabilization step) required to obtain a visible fluorescence signal. This value is the maximal sensitivity of CARD-FISH that could theoretically be achieved if no permeabilization was needed. It was determined based on a 16S rRNA content that was 67% of the original content (i.e., 24,800 ⫾ 4,100 16S rRNA molecules per E. coli cell). As with conventional FISH, the minimal 16S rRNA content required for CARD-FISH was determined by performing a competitive FISH experiment. In this case, competitive hybridizations were performed with Cy5- and HRP-labeled EUB338-I and E. coli cells on glass slides. As KCy3 and KCy5 are identical (see above), the experimentally determined KCy3/ KHRP ratio of 16.2 was assumed to be applicable to the com-


FIG. 6. Fluorescence intensities measured in a competitive hybridization experiment (CARD-FISH) with E. coli cells and mixtures of Cy5- and HRP-labeled probe EUB338-I. The mean intensity of fluorescence (per cell) conferred by the HRP-labeled probe after TSA was quantified. The dashed lines indicate the signal intensities that correspond to S/N ratios of 1.5 and 3. The equation and the coefficient of determination for the regression line are indicated. Note that the x axis does not show the initial probe concentration ratio (as in Fig. 4 and 5) but instead shows the percentage of HRP-labeled probe that bound to 16S rRNA inside E. coli cells. This percentage was calculated based on equation 11 and the experimentally determined KCy3/KHRP (Fig. 4B). For each data point the fluorescence intensity of at least 200 cells was determined. Error bars (standard deviations) are not shown as they were smaller than the symbols.

petition experiment with Cy5- and HRP-labeled probes (i.e., KCy5/KHRP ⫽ 16.2). Then, using equation 11, the percentages of 16S rRNA molecules hybridized to the HRP-labeled probe in these competitive hybridizations were calculated (i.e., 100 ⫻ [HHRP]/[TT]). These calculated percentages were then plotted against the measured intensities of HRP probe-conferred fluorescence (Fig. 6). As with conventional FISH (Fig. 5), a clear correlation was obtained. This demonstrates that the quantification of target molecules by CARD-FISH is not hampered by effects such as fluorescence quenching or exhaustion or saturation of HRP in the signal amplification step. The conventional FISH experiments had revealed that an S/N ratio of 1.5 was needed to detect E. coli cells on glass slides and to distinguish E. coli from a probe nontarget population. According to Fig. 6, an S/N ratio of 1.5 corresponded to a fluorescence intensity of 39 RU after CARD-FISH. At this signal intensity, 0.036% of the 16S rRNA molecules had hybridized to the HRP-labeled probe (Fig. 6). From this it follows that the ideal detection limit of CARD-FISH of planktonic cells is as low as 8.9 ⫾ 1.5 target molecules per cell. The practical detection limit, which considers the necessity of cell permeabilization causing a loss of target molecules, is slightly higher, 14 ⫾ 2 target molecules per cell. The same calculations repeated for an S/N ratio of 3 showed that 36 ⫾ 6 target molecules per cell (ideal detection limit) to 54 ⫾ 7 target molecules per cell (practical detection limit) would be required for successful CARD-FISH in activated sludge. Consequently, CARD-FISH was 26- to 41-fold more sensitive than conventional FISH in our experiments. These values are similar to the results obtained by Lebaron et al. (21) (12-fold) and by Scho ¨nhuber et al. (37) (20-fold), who determined the increase in sensitivity due to CARD-FISH by measuring fluorescence intensities without quantifying 16S rRNA copy numbers.

VOL. 74, 2008


Concluding remarks and outlook. In this study, we determined how many 16S rRNA molecules are required for two applications of FISH and CARD-FISH with oligonucleotide probes: the detection of planktonic cells on glass slides and the detection of single cells in activated sludge flocs. The value obtained here for conventional FISH on glass slides (approximately 370 16S rRNA copies per cell) is low compared to the detection limit of flow cytometers (at least those which were used before 1995), which were used to sort probe-stained bacterial pure cultures (several thousand ribosomes per cell) (4). This is consistent with the findings of Zheng and Harata (50), who reported that a CLSM could detect as few as 82 rhodamine molecules in an area of 1.4 ␮m2, which is similar to the area of a single bacterial cell in a two-dimensional image. Moreover, the detection of the E. coli cells in our FISH experiments on glass slides was facilitated by the almost complete absence of fluorescent background. The great influence of background autofluorescence on the sensitivity of FISH is shown by the much higher 16S rRNA content needed for FISH in activated sludge (approximately 1,400 copies per cell), which is closer to the range indicated by Amann et al. (4). Activated sludge represents a structurally relatively complex type of environmental sample and resembles other complex samples, such as aquatic biofilms. Thus, the minimal 16S rRNA copy numbers determined for activated sludge can serve as reference values for many standard applications of FISH and CARD-FISH. However, the sensitivity of these methods might be significantly lower with extreme samples, such as biofilms containing large numbers of phototrophic microbes or algae, whose pigments show much stronger autofluorescence than the activated sludge biomass and matrix. In addition to the type of sample, the sensitivity of FISH and CARD-FISH depends on technical factors, such as the optical quality of the microscope and the fluorochrome used for probe labeling or the detection step. For example, Cy3 is much brighter than the green dyes fluorescein isothiocyanate and FLUOS [5(6)carboxyfluorescein-N-hydroxysuccinimide ester] (35). Furthermore, the rRNA content determines the brightness of FISH signals in combination with other parameters, like the cell wall structure, the cell fixation method, quenching effects, probe affinity, and the accessibility of the probe binding sites on the rRNA (43, 44, 47, 48). Experiments should be performed to identify the relative importance of these factors with the goal of further optimizing FISH and tailoring diagnostic FISH protocols to specific target organisms. For quantification of rRNA copy numbers, our work provides the theoretical and technical backbone for such studies. Another important factor, which affects the sensitivity of FISH, is the hybridization efficiency of the oligonucleotide probe used. Some probes yield dim fluorescence signals because of low probe affinity, which is a function of not only the probe sequence but also the accessibility of the probe target site due to structural constraints in the ribosome (47). Hence, to yield visible signals after FISH, such dim probes may require 16S rRNA copy numbers for their target organisms higher than those determined here for probe EUB338-I, which yields moderate to high fluorescence intensities (15). On the other hand, Yilmaz et al. (48) showed that the brightness of so-called dim probes can be enhanced significantly by longer hybridization times, which accounts for the slower hybridization kinetics of

5075

such probes. Therefore, it should be noted that if probes other than EUB338-I are used in the competitive hybridization approach, the hybridization time and stringency must be adjusted to ensure that the hybridization efficiency is optimal, because only then would the results be independent of the probe used. Interestingly, the dissociation profiles of Cy- and HRP-labeled probes were significantly different, although the nucleotide sequences of the probes were identical (Fig. 3). These results imply that, despite the signal amplification, CARDFISH may yield only dim fluorescence signals if the hybridization stringency (e.g., FA concentration) is the same as that used for conventional FISH with Cy-labeled probes. One could argue that this effect may be less pronounced if CARD-FISH is performed at temperatures lower than those used in this study. However, direct comparison with the same hybridization and washing temperatures clearly showed that the stability of the rRNA-probe hybrid was lower for the HRP-labeled probe. One reason for this could be that the large molecule HRP negatively affects the base pairing between the nucleotides close to the 5⬘ end of the probe and the complementary nucleotides at the probe target site. This would decrease the equilibrium constant (i.e., K value) of probe target hybridization and, therefore, contribute to the altered probe dissociation profile. Moreover, it might reduce the specificity of an HRP-labeled probe if the base mismatches with nontarget organisms are located close enough to the 5⬘ end of the probe. Our observations strongly suggest that the specificity of HRPlabeled probes should be carefully verified and that the optimal hybridization conditions should be determined separately for HRP- and Cy-labeled probes. Applications of our competitive FISH technique are not necessarily restricted to determining the sensitivity of FISH. Previous research has shown that cellular metabolic activity and rRNA content correlate in some bacteria, whereas in other bacteria this is not the case (7, 8, 36). This was found by using the signal intensity after FISH as a simple indicator of ribosome content. Determining absolute rRNA copy numbers instead of only relative fluorescence units could improve the comparability of such studies. Several studies have shown that CARD-FISH enables in situ detection of mRNA in microbial cells. Most of these studies relied on means to further enhance the sensitivity, such as polynucleotide probes labeled with many HRP molecules (29, 44) or two rounds of TSA-mediated signal amplification (20). These approaches suffer from disadvantages such as the reduced specificity of polynucleotide probes (compared to oligonucleotides) and the need for additional time-consuming steps. However, based on the high sensitivity of “normal” CARDFISH with oligonucleotide probes, this method alone could theoretically be sufficient for detecting mRNA in bacterial cells. Nevertheless, detection of bacterial mRNA by CARDFISH with oligonucleotide probes and only one signal amplification step has been reported only once (5). Recently, even conventional FISH using probes labeled with a single nearinfrared dye was used to detect mRNA in pure bacterial cultures (10). Future studies may show whether this approach, which does not include any signal amplification step, is also suitable for analyzing uncultured microbes in environmental samples. General problems with mRNA detection by FISH could be (i) the low stability of mRNA and its degradation

5076

HOSHINO ET AL.

during the relatively long CARD-FISH or conventional FISH procedures, which take place at temperatures far above room temperature, and (ii) leakage of mRNA out of the cells. Experiments carried out in this study revealed that 16S rRNA molecules may leak out of cells permeabilized for CARDFISH. As mRNA is often smaller than 16S rRNA or the sizes of 16S rRNA and mRNA are similar, it appears likely that the permeabilization can also result in a significant loss of mRNA molecules. Furthermore, the lower stability of the probe-target duplex with the HRP label (Fig. 3) also suggests that hybridization conditions should be specifically optimized for HRPlabeled probes. Therefore, developing new protocols that minimize mRNA loss and maximize probe-target duplex stability might be the key to successful, specific, and efficient FISH detection of mRNA in microorganisms. ACKNOWLEDGMENTS T.H. was supported by a JSPS postdoctoral fellowship for research abroad. Support from National Science Foundation grant CBET0636533 to D.R.N. and L.S.Y. is also acknowledged. REFERENCES 1. Amann, R., B. M. Fuchs, and S. Behrens. 2001. The identification of microorganisms by fluorescence in situ hybridisation. Curr. Opin. Biotechnol. 12:231–236. 2. Amann, R. I., B. J. Binder, R. J. Olson, S. W. Chisholm, R. Devereux, and D. A. Stahl. 1990. Combination of 16S rRNA-targeted oligonucleotide probes with flow cytometry for analyzing mixed microbial populations. Appl. Environ. Microbiol. 56:1919–1925. 3. Amann, R. I., L. Krumholz, and D. A. Stahl. 1990. Fluorescent-oligonucleotide probing of whole cells for determinative, phylogenetic, and environmental studies in microbiology. J. Bacteriol. 172:762–770. 4. Amann, R. I., W. Ludwig, and K.-H. Schleifer. 1995. Phylogenetic identification and in situ detection of individual microbial cells without cultivation. Microbiol. Rev. 59:143–169. 5. Bakermans, C., and E. L. Madsen. 2002. Detection in coal tar waste-contaminated groundwater of mRNA transcripts related to naphthalene dioxygenase by fluorescent in situ hybridization with tyramide signal amplification. J. Microbiol. Methods 50:75–84. 6. Baracchini, E., and H. Bremer. 1987. Determination of synthesis rate and lifetime of bacterial mRNAs. Anal. Biochem. 167:245–260. 7. Binder, B. J., and Y. C. Liu. 1998. Growth rate regulation of rRNA content of a marine Synechococcus (cyanobacterium) strain. Appl. Environ. Microbiol. 64:3346–3351. 8. Bremer, H., and P. P. Dennis. 1996. Modulation of chemical composition and other parameters of the cell by growth rate, p. 1553–1569. In F. C. Neidhardt, R. Curtiss III, J. L. Ingraham, E. C. C. Lin, K. B. Low, B. Magasanik, W. S. Reznikoff, M. Riley, M. Schaechter, and H. E. Umbarger (ed.), Escherichia coli and Salmonella: cellular and molecular biology, 2nd ed. ASM Press, Washington, DC. 9. Burggraf, S., T. Mayer, R. Amann, S. Schadhauser, C. R. Woese, and K. O. Stetter. 1994. Identifying members of the domain Archaea with rRNAtargeted oligonucleotide probes. Appl. Environ. Microbiol. 60:3112–3119. 10. Coleman, J. R., D. E. Culley, W. B. Chrisler, and F. J. Brockman. 2007. mRNA-targeted fluorescent in situ hybridization (FISH) of Gram-negative bacteria without template amplification or tyramide signal amplification. J. Microbiol. Methods 71:246–255. 11. Daims, H., A. Bru ¨hl, R. Amann, K.-H. Schleifer, and M. Wagner. 1999. The domain-specific probe EUB338 is insufficient for the detection of all Bacteria: development and evaluation of a more comprehensive probe set. Syst. Appl. Microbiol. 22:434–444. 12. Daims, H., S. Lu ¨cker, and M. Wagner. 2006. daime, a novel image analysis program for microbial ecology and biofilm research. Environ. Microbiol. 8:200–213. 13. Daims, H., K. Stoecker, and M. Wagner. 2005. Fluorescence in situ hybridisation for the detection of prokaryotes, p. 213–239. In A. M. Osborn and C. J. Smith (ed.), Molecular microbial ecology. Bios-Garland, Abingdon, United Kingdom. 14. DeLong, E. F., G. S. Wickham, and N. R. Pace. 1989. Phylogenetic stains: ribosomal RNA based probes for the identification of single cells. Science 243:1360–1363. 15. Fuchs, B. M., G. Wallner, W. Beisker, I. Schwippl, W. Ludwig, and R. Amann. 1998. Flow cytometric analysis of the in situ accessibility of Escherichia coli 16S rRNA for fluorescently labeled oligonucleotide probes. Appl. Environ. Microbiol. 64:4973–4982.

APPL. ENVIRON. MICROBIOL. 16. Hicks, R. E., R. I. Amann, and D. A. Stahl. 1992. Dual staining of natural bacterioplankton with 4⬘,6-diamidino-2-phenylindole and fluorescent oligonucleotide probes targeting kingdom-level 16S rRNA sequences. Appl. Environ. Microbiol. 58:2158–2163. 17. Hoshino, T., N. Noda, S. Tsuneda, A. Hirata, and Y. Inamori. 2001. Direct detection by in situ PCR of the amoA gene in biofilm resulting from a nitrogen removal process. Appl. Environ. Microbiol. 67:5261–5266. 18. Huang, W. E., K. Stoecker, R. Griffiths, L. Newbold, H. Daims, A. S. Whiteley, and M. Wagner. 2007. Raman-FISH: combining stable-isotope Raman spectroscopy and fluorescence in situ hybridization for the single cell analysis of identity and function. Environ. Microbiol. 9:1878–1889. 19. Kaplan, R., and D. Apirion. 1975. Decay of ribosomal ribonucleic acid in Escherichia coli cells starved for various nutrients. J. Biol. Chem. 250:3174– 3178. 20. Kubota, K., A. Ohashi, H. Imachi, and H. Harada. 2006. Visualization of mcr mRNA in a methanogen by fluorescence in situ hybridization with an oligonucleotide probe and two-pass tyramide signal amplification (two-pass TSAFISH). J. Microbiol. Methods 66:521–528. 21. Lebaron, P., P. Catala, C. Fajon, F. Joux, J. Baudart, and L. Bernard. 1997. A new sensitive, whole-cell hybridization technique for detection of bacteria involving a biotinylated oligonucleotide probe targeting rRNA and tyramide signal amplification. Appl. Environ. Microbiol. 63:3274–3278. 22. Lee, N., P. H. Nielsen, K. H. Andreasen, S. Juretschko, J. L. Nielsen, K.-H. Schleifer, and M. Wagner. 1999. Combination of fluorescent in situ hybridization and microautoradiography—a new tool for structure-function analyses in microbial ecology. Appl. Environ. Microbiol. 65:1289–1297. 23. Li, T., T. D. Wu, L. Mazeas, L. Toffin, J. L. Guerquin-Kern, G. Leblon, and T. Bouchez. 2008. Simultaneous analysis of microbial identity and function using NanoSIMS. Environ. Microbiol. 10:580–588. 24. Maixner, F., D. R. Noguera, B. Anneser, K. Stoecker, G. Wegl, M. Wagner, and H. Daims. 2006. Nitrite concentration influences the population structure of Nitrospira-like bacteria. Environ. Microbiol. 8:1487–1495. 25. Maruyama, H. B., and S. Okamura. 1972. Ribosome degradation and the degradation products in starved Escherichia coli. V. Ribonucleoprotein particles from glucose-starved cells. J. Bacteriol. 110:442–446. 26. Moter, A., G. Leist, R. Rudolph, K. Schrank, B. K. Choi, M. Wagner, and U. B. Go ¨bel. 1998. Fluorescence in situ hybridization shows spatial distribution of as yet uncultured treponemes in biopsies from digital dermatitis lesions. Microbiology 144:2459–2467. 27. Orphan, V. J., C. H. House, K. U. Hinrichs, K. D. McKeegan, and E. F. DeLong. 2001. Methane-consuming archaea revealed by directly coupled isotopic and phylogenetic analysis. Science 293:484–487. 28. Ouverney, C. C., and J. A. Fuhrman. 1997. Increase in fluorescence intensity of 16S rRNA in situ hybridization in natural samples treated with chloramphenicol. Appl. Environ. Microbiol. 63:2735–2740. 29. Pernthaler, A., and R. Amann. 2004. Simultaneous fluorescence in situ hybridization of mRNA and rRNA in environmental bacteria. Appl. Environ. Microbiol. 70:5426–5433. 30. Pernthaler, A., J. Pernthaler, and R. Amann. 2002. Fluorescence in situ hybridization and catalyzed reporter deposition for the identification of marine bacteria. Appl. Environ. Microbiol. 68:3094–3101. 31. Pernthaler, A., C. M. Preston, J. Pernthaler, E. F. DeLong, and R. Amann. 2002. Comparison of fluorescently labeled oligonucleotide and polynucleotide probes for the detection of pelagic marine bacteria and archaea. Appl. Environ. Microbiol. 68:661–667. 32. Pernthaler, J., A. Pernthaler, and R. Amann. 2003. Automated enumeration of groups of marine picoplankton after fluorescence in situ hybridization. Appl. Environ. Microbiol. 69:2631–2637. 33. Poulsen, L. K., G. Ballard, and D. A. Stahl. 1993. Use of rRNA fluorescence in situ hybridization for measuring the activity of single cells in young and established biofilms. Appl. Environ. Microbiol. 59:1354–1360. 34. Ramagopal, S. 1984. Metabolic changes in ribosomes of Escherichia coli during prolonged culture in different media. Eur. J. Biochem. 140:353–361. 35. Randolph, J. B., and A. S. Waggoner. 1997. Stability, specificity and fluorescence brightness of multiply-labeled fluorescent DNA probes. Nucleic Acids Res. 25:2923–2929. 36. Schmid, M., S. Schmitz-Esser, M. Jetten, and M. Wagner. 2001. 16S-23S rDNA intergenic spacer and 23S rDNA of anaerobic ammonium-oxidizing bacteria: implications for phylogeny and in situ detection. Environ. Microbiol. 3:450–459. 37. Scho ¨nhuber, W., B. Fuchs, S. Juretschko, and R. Amann. 1997. Improved sensitivity of whole-cell hybridization by the combination of horseradish peroxidase-labeled oligonucleotides and tyramide signal amplification. Appl. Environ. Microbiol. 63:3268–3273. 38. Schramm, A., L. H. Larsen, N. P. Revsbech, N. B. Ramsing, R. Amann, and K.-H. Schleifer. 1996. Structure and function of a nitrifying biofilm as determined by in situ hybridization and the use of microelectrodes. Appl. Environ. Microbiol. 62:4641–4647. 39. Sekar, R., A. Pernthaler, J. Pernthaler, F. Warnecke, T. Posch, and R. Amann. 2003. An improved protocol for quantification of freshwater Actinobacteria by fluorescence in situ hybridization. Appl. Environ. Microbiol. 69:2928–2935.

VOL. 74, 2008


40. Trebesius, K. H., R. Amann, W. Ludwig, K. Mu ¨hlegger, and K.-H. Schleifer. 1994. Identification of whole fixed bacterial cells with nonradioactive 23S rRNA-targeted polynucleotide probes. Appl. Environ. Microbiol. 60:3228– 3235. 41. Tujula, N. A., C. Holmstrom, M. Mussmann, R. Amann, S. Kjelleberg, and G. R. Crocetti. 2006. A CARD-FISH protocol for the identification and enumeration of epiphytic bacteria on marine algae. J. Microbiol. Methods 65:604–607. 42. Wada, A., Y. Yamazaki, N. Fujita, and A. Ishihama. 1990. Structure and probable genetic location of a “ribosome modulation factor” associated with 100S ribosomes in stationary-phase Escherichia coli cells. Proc. Natl. Acad. Sci. USA 87:2657–2661. 43. Wagner, M., M. Horn, and H. Daims. 2003. Fluorescence in situ hybridization for the identification of prokaryotes. Curr. Opin. Microbiol. 6:302–309. 44. Wagner, M., M. Schmid, S. Juretschko, K. H. Trebesius, A. Bubert, W. Goebel, and K.-H. Schleifer. 1998. In situ detection of a virulence factor mRNA and 16S rRNA in Listeria monocytogenes. FEMS Microbiol. Lett. 160:159–168.

5077

45. Wallner, G., R. Erhart, and R. Amann. 1995. Flow cytometric analysis of activated sludge with rRNA-targeted probes. Appl. Environ. Microbiol. 61: 1859–1866. 46. Yilmaz, L. S., and D. R. Noguera. 2007. Development of thermodynamic models for simulating probe dissociation profiles in fluorescence in situ hybridization. Biotechnol. Bioeng. 96:349–363. 47. Yilmaz, L. S., and D. R. Noguera. 2004. Mechanistic approach to the problem of hybridization efficiency in fluorescent in situ hybridization. Appl. Environ. Microbiol. 70:7126–7139. 48. Yilmaz, L. S., H. E. Okten, and D. R. Noguera. 2006. Making all parts of the 16S rRNA of Escherichia coli accessible in situ to single DNA oligonucleotides. Appl. Environ. Microbiol. 72:733–744. 49. Zarda, B., R. Amann, G. Wallner, and K. H. Schleifer. 1991. Identification of single bacterial cells using digoxigenin-labelled, rRNA-targeted oligonucleotides. J. Gen. Microbiol. 137:2823–2830. 50. Zheng, X. Y., and A. Harata. 2001. Confocal fluorescence microscope studies of the adsorptive behavior of dioctadecyl-rhodamine B molecules at a cyclohexane-water interface. Anal. Sci. 17:131–135.