Journal of Methods Microbiological Journal of Microbiological Methods 40 (2000) 125–134 www.elsevier.com/ locate / jmicmeth Widefield deconvolution epifluorescence microscopy combined with fluorescence in situ hybridization reveals the spatial arrangement of bacteria in sponge tissue a , b a a*Werner Manz , Gernot Arp , Gabriela Schumann-Kindel , Ulrich Szewzyk , bJoachim Reitner a ¨ ¨ ¨Technische Universitat Berlin, Institut f ur Technischen Umweltschutz, Fachgebiet Okologie der Mikroorganismen, Sekretariat OE 5, Franklinstraße 29, D-10587 Berlin, Germany b ¨ ¨ ¨ ¨ ¨Institut und Museum f ur Geologie und Palaontologie, Universitat Gottingen, D-37077 Gottingen, Germany Received 8 July 1999; received in revised form 27 July 1999; accepted 8 August 1999 Abstract Widefield deconvolution epifluorescence microscopy (WDEM) combined with fluorescence in situ hybridization (FISH) was performed to identify and characterize single bacterial cells within sections of the mediterranean sponge Chondrosia reniformis. Sponges were embedded in paraffin wax or plastic prior to the preparation of thin sections, in situ hybridization and microscopy. Serial digital images generated by widefield epifluorescence microscopy were visualized using an exhaustive photon reassignment deconvolution algorithm and three-dimensional rendering software. Computer processing of series of images taken at different focal planes with the deconvolution technique provided deblurred three-dimensional images with high optical resolution on a submicron scale. Results from the deconvolution enhanced widefield microscopy were compared with conventional epifluorescent microscopical images. By the application of the deconvolution algorithm on digital image data obtained with widefield epifluorescence microscopy after FISH, the occurrence and spatial arrangement of Desulfovibrionaceae closely associated with micropores of Chondrosia reniformis could be visualized. Ó 2000 Elsevier Science B.V. Keywords: Widefield deconvolution epifluorescence microscopy; FISH; 16S rRNA oligonucleotide probes; Sponge associated bacteria; Chondrosia reniformis 1. Introduction This technique is commonly applied to suspensions of bacteria, bacterial aggregates and biofilms (Amann Fluorescence in situ hybridization (FISH) is a et al., 1995). To detect microorganisms in situ within modern microbiological technique for the identifica- tissues or thick biofilms in their natural spatial tion and characterization of single bacterial cells in distribution, embedding and sectioning of the sam- complex natural samples without prior cultivation. ples prior to the hybridization procedure is neces- sary. For this purpose, paraffin sections have been shown to be suitable for in situ hybridizations *Corresponding author. Tel.: 1 49-30-314-25589; fax: 1 49-30- (Poulsen et al., 1994; Licht et al., 1996; Rothemund314-73461. E-mail address: manz0654@mailszrz.zrz.tu-berlin.de (W. Manz) et al., 1996). Alternatively, hard setting resins (2- 0167-7012/00 Ó 2000 Elsevier Science B.V. PI I : S0167-7012( 99 )00103-7 Open access under CC BY-NC-ND license. Open access under CC BY-NC-ND license. 126 W. Manz et al. / Journal of Microbiological Methods 40 (2000) 125 –134 hydroxyethyl-methacrylate) have been applied for iterative algorithms, such as the nearest-neighbour embedding (Gerrits and Smid, 1983) that are suitable method, are commonly used for a quick image for hybridizations because of their relatively hydro- restoration. However, the nearest-neighbor method is philic properties (Moter et al., 1998). considered as the least accurate and the most artifi- However, the application of in situ probing com- cial of the algorithms available, and is therefore most bined with conventional widefield epifluorescence often used for quick previews (Holmes et al., 1995). microscopy is often impaired by inherent problems, Deconvolution based on a measured point spread e.g. thickness of the samples (tissue sections) or function is considered to be the more accurate image background fluorescence caused by out-of-focus restoration compared to theoretical functions (Carrin- blur, which obliterate specific signals of the fluores- gton et al., 1995) because the distortion by non-ideal cent probes and may limit the applicability of the optics of the microscope and filter sets is taken into FISH technique. Thus, small objects and structures account and the out-of-focus signals are placed back remain invisible and fluorescently labeled cells are to the points from which they came (‘photon reas- hardly detectable within these blurred images. signment processing’). One possible way to circumvent these problems is Deconvolution can be combined for images ac- the combination of FISH with confocal laser scan- quired using transmitted light brightfield, widefield ning microscopy (Wagner et al., 1994; Manz et al., fluorescence, or confocal fluorescence microscopes. 1995; MacNaughton et al., 1996). Using CLSM, The combination of FISH with conventional epi- thickness of the specimen is not limiting since light fluorescence microscopy and subsequent deconvolu- from out-of-focus planes is physically excluded by tion rendering of series of 2D digital images may the detector pinhole. In addition, the series of 2D have useful implications for the in situ identification digital images obtained by CLSM allows numerous of bacteria in a wide range of specimen. digital image processings including 3D rendering and Recently, sponge-associated bacteria (SAB) were reconstructions (Lawrence et al., 1996). Limitations characterized by a top-to-bottom directed in situ of CLSM are that the detector pinhole system hybridization approach using fluorescent rRNA reduces the amount of light that may enter the targeted oligonucleotide probes (Schumann-Kindel et detector system, the strong bleaching effects caused al., 1997). This study resulted in the visualization of by the laser beam used as light source and the a great amount of bacteria displaying remarkable expensive microscopical and computing equipments. metabolic potential, which could be assigned to the In the present approach, we demonstrate that the alpha-, gamma- and delta-subclass of Proteobacteria. combination of FISH with conventional widefield However, strong blurring effects caused by out-of- epifluorescence microscopy and the subsequent ap- focus fluorescence and low average hybridization plication of deconvolution algorithm provides a signal intensities obstructed a more detailed elucida- suitable, low-cost alternative to confocal laser scan- tion of the sponge–bacteria association. In the study ning microscopy. Deconvolution refers to the mathe- reported here, we demonstrate that the combination matical removal, or deblurring of out-of-focus haze of embedding and thin-sectioning of sponge material of a certain two dimensional image obtained with an followed by FISH and deconvolution enhanced optical microscope. For that purpose, a series (stack) widefield microscopy is an excellent tool to surpass of digitalized optical sections from defined depths at these problems and provide new insights into the the same xy location of the sample is required. A microbial ecology of these important invertebrates. computer then compiles the information of the focus plane by reassigning signals which originated from objects located in other focal planes. 2. Material and methods The out-of-focus blur can be mathematically modeled as a point spread function (Shaw and 2.1. Sample preparation Rawlins, 1991) and deconvolution microscopy can therefore be thought of as a method for inverting the Sample fixation was done essentially following the unavoidable and natural blurring effect of the point protocol of Manz et al. (1992). Minor modifications spread function (Holmes et al., 1995). Linear, non- were needed as sponge tissue was used instead of W. Manz et al. / Journal of Microbiological Methods 40 (2000) 125 –134 127 cell suspensions. Complete native sponges were including Desulfobacteriaceae (Rabus et al., 1996), soaked with 3.7% formalin and fixed for 4–12 h at (iii) probe DSV1292, specific for Desulfovibrio 4–78C. Fixed sponges were washed twice with 1 3 desulfuricans and relatives (Manz et al., 1998). The PBS (130 mM NaCl, 10 mM Na-phosphate buffer, Cy3-labeled oligonucleotide non-EUB338, which has pH 7.2) for 15 min, and stored in a 1:1 mixture (v /v) the complementary (antisense) sequence to probe of 1 3 PBS and 96% (v/v) ethanol at 2 208C. EUB338 served as a negative control for nonspecific Tissue blocks (10 3 10 mm) were cut from different binding. Custom synthesized oligonucleotides were parts of the sponges using a sterile razor blade, purchased 59-labeled with the indocarbocyanine dye dehydrated (50, 80 and 96% (v/v) ethanol, 3 min Cy3 (CyDyeE, Amersham Pharmacia Biotech each; 2 3 100% xylene, 15 min each) and sub- Europe, Freiburg, Germany) and Oregon Green sequently embedded in paraffin. Paraffin sections of (Molecular Probes Inc., Eugene, OR) from Biometra ¨15–20-mm thickness were obtained by using a rotary (Gottingen, Germany) and Metabion (Ebersberg, microtome HM 340 E (Microm, Walldorf, Germany) Germany), respectively. All oligonucleotides were with single-use blades. The slices were mounted on stored in TE buffer (10 mM Tris, 1 mM EDTA, pH silanized glass slides, dewaxed by xylene treatment 7.5) at 2 208C. Working solutions were adjusted to for 10 min and subsequently air-dried prior to 50 ng DNA per ml. Pre-warmed hybridization solu- hybridization. Subsamples of the sponge material tion (0.9 M NaCl, 20 mM Tris–HCl [pH 7.2], 0.01% were embedded in Technovit 7100 (2-hydroxyethyl- SDS, 35% [v/v] formamide) was mixed with fluores- methacrylate, Heraeus Kulzer, Wehrheim, Germany) cently labeled oligonucleotide (1 ng per ml hybridi- as described by the manufacturer. Briefly, sponge zation solution). tissue blocks were dehydrated by an ethanol series in Fixed slices of sponge tissue were hybridized by 70, 90 and 96% (v/v) ethanol for 1 h each, and a addition of 30 ml hybridization solution and 3 ng final treatment in 99% ethanol (v /v) for 2 h. The labelled probe applied on spots of 5 mm diameter to pre-infiltration of the dehydrated samples was per- each tissue slice and incubated for at least 5 h at formed in an 1:1 mixture of 99% ethanol (v /v) and 468C. After this, unbound oligonucleotides were Technovit 7100 for at least 2 h. For the final removed by gently rinsing with washing buffer (20 infiltration, 1.0 g of Technovit hardener was solved mM Tris–HCl, 0.01% SDS, 88 mM NaCl) and in 100 ml base-liquid 100% Technovit, and the incubated at 488C for 20 min. Before microscopic samples were subsequently infiltrated over night. For analysis, hybridized tissues slices were carefully the embedding of the sponge samples, the infiltrated rinsed with distilled water, air dried and mounted in material was transferred into embedding solution (1 antifading glycerol medium (Citifluor AF2, Citifluor part of Technovit hardener II and 15 parts of Ltd., London, UK). All hybridization and washing infiltration solution). The polymerization of the steps were performed in the dark. samples was performed by incubation of the embed- ded sponge material for 1 h at room temperature, 2.3. Widefield deconvolution epifluorescence followed by the final incubation for 2 h at 308C in microscopy ( WDEM) the dark. All steps were carried out under vacuum using a desiccator equipment. Polymerized sponge Epifluorescence microscopy was performed with a tissue blocks were trimmed with a small circular widefield Zeiss Axioplan microscope (Oberkochen, saw. Sections of 20-mm thickness were cut with a Germany) equipped with a high-pressure mercury steel knife at the rotary microtome. arc lamp (HBO 50, Zeiss) and Zeiss filter sets no. 09 (excitation 450–490 nm, dichroic mirror 510 nm, 2.2. Hybridization procedure suppression 520 nm), and no. 15 (excitation 546 nm, dichroic mirror 580 nm, suppression 590 nm), corre- For in situ hybridization of sponge sections the sponding to the excitation and emission maxima of following oligonucleotide probes were used: (i) the fluorochromes Oregon Green and Cy3, respec- probe EUB338, specific for the domain Bacteria tively. To reduce photobleaching, illumination was (Amann et al., 1990), (ii) probe SRB385Db, specific controlled by an Uniblitz shutter device D 122 for most members of delta subclass Proteobacteria (Vincent Associates, Rochester, NY). A piezo-mover 128 W. Manz et al. / Journal of Microbiological Methods 40 (2000) 125 –134 (Physik Instrumente, Waldbronn, Germany) was 2.4. Image data acquisition and processing attached to a infinity color-corrected 63 3 Plan- Apochromat, oil immersion objective with a numeri- The general flow for the data acquisition and cal aperture of 1.4 (Zeiss, Oberkochen, Germany). image processing is given in Fig. 1. For the acquisi- Real color images were acquired using a two-stage tion of the image stacks and for their subsequent Peltier cooled CCD-camera (PCO Computer Optics, processing the Metamorph Imaging software (Uni- Kehlheim, Germany) with a spectral range of 290– versal Imaging Corporation, West Chester, PA) in 1000 nm. The 2/30 sized CCD chip consists of 1280 conjunction with the EPR-deconvolution software (H) 3 1024 (V) pixel, each 6.7 3 6.7 mm in size. (Scanalytics, Billerica, MA) were used. Calculations Due to the RGB mosaic filter the real resolution was were performed using a 200-MHz personal computer lowered by a factor of the square root. The dynamic running under Microsoft Windows 95. Image data range was 24 bit (8 bit each colour). When using the collection was carried out by sampling of data stacks 63 3 objective, the voxel size was 100 3 100 nm 3 z comprising up to 21 digital images obtained from distance (z 5 250 or 500 nm). real color images and stored in TIFF format for Fig. 1. Flow chart of acquisition and processing of digital image stacks for deconvolution and three-dimensional reconstruction. W. Manz et al. / Journal of Microbiological Methods 40 (2000) 125 –134 129 subsequent digital image analysis. Spacing along the 3.2. Deconvolution microscopy of sponge z-axes in increments of 0.25 or 0.5 mm turned out to associated bacteria be suitable in most cases. Consequently, 15–20-mm thick sponge sections could be conveniently pro- Fluorescence in situ hybridization (FISH) of cessed. Dependent on the samples and fluorochromes paraffin sections of Chondrosia reniformis using the used for labelling, the exposure times varied between Bacteria specific probe EUB338 (Oregon Green 800 ms to 1.5 s for each single image. Prior to labeled) and subsequent 3D reconstruction resulted photon reassignment and three-dimensional image in a strongly blurred image caused by high levels of restoration, the acquired real color image stacks were out-of-focus fluorescence emitted from the sponge color decoded. For this purpose the red and green tissue. In consequence, the maximum optical res- channel was processed separately and the image data olution which could be obtained by conventional were combined afterwards to form a deblurred, epifluorescence microscopy remained fairly low. three-dimensional real color image stack. Usually, This hindered the visualization of details within the the processing of one channel covering the emission sponge sections and the spatial arrangement of the maximum was by far sufficient. hybridized cell could not be thoroughly resolved The EPR-deconvolution software has to be cali- (Fig. 2A). brated by measured point spread functions of fluores- The application of the calibrated deconvolution cent beads of a size below optical resolution. The algorithm after sampling of a stack of 21 optical fluorescence emission spectra of the beads should be sections through the identical object shown in Fig. comparable to the fluorescent emission spectra of the 2A resulted in a remarkable higher optical resolution sample or fluorophore used. For this purpose, 200 (Fig. 2B). Specific cell features of sponge associated nm-sized Nile red fluorescent polystyrene beads were bacteria could not be shown by classical epifluoresc- analyzed (FluoSpheres, Molecular Probes Inc., ence microscopy (Fig. 2C). In contrast to this, the Eugene, OR) which have broad excitation and emis- application of the EPR deconvolution algorithm for sion bandwidths (excitation maximum 535 nm, emis- stacks of optical sections followed by rendering with sion maximum 575 nm) suitable for the used filter the maximum option using the brightest pixels along sets (Haugland, 1996). the line of sight provided detailed information on the cell shape, inclusion bodies and growth states of the corresponding bacteria (Fig. 2D, arrow indicated). 3. Results and discussion Additionally, the spatial arrangement of the sponge associated bacteria within the sponge tissue could be 3.1. Sample embedding and sectioning determined. Simultaneous hybridization of Chon- drosia reniformis with the Bacteria specific probe Embedding of biological specimen in paraffin and EUB338 (Oregon Green-labeled) and the Cy3- plastics prior to sectioning and subsequent FISH is labeled probe SRB385Db resulted in a highly blurred discussed rather controversially in the literature image obtained from the unprocessed optical sections (Poulsen et al., 1994; Licht et al., 1996; Rothemund (Fig. 3A. Application of the deconvolution algorithm et al., 1996; Moter et al., 1998). In this study, to the identical optical sections shown in Fig. 3A sections of the paraffin and Technovit embedded visualized not only large amounts of small bacterial sponge tissues were obtained with defined thickness cells hybridizing with probe EUB338, but also of 15–20 mm and tissue slices were subsequently revealed dividing cells hybridizing with probe mounted on silanized glass slides. With regard to SRB385Db (Fig. 3B, indicated by arrows). autofluorescence of the matrix used for embedding, The photomicrographs given in Fig. 3C and D no specific improvement by using the Technovit show the maximum pixel enlargement of an un- could be observed. In contrast to the paraffin sponge processed extended image projection obtained by tissue blocks, cutting of Technovit sections with the widefield epifluorescence microscopy (Fig. 3C) com- rotary microtom was more difficult and slices could pared with the maximum resolution achieved for the be hardly mounted even on silanized glass slides. same object after deconvolution rendering of a series 130 W. Manz et al. / Journal of Microbiological Methods 40 (2000) 125 –134 Fig. 2. Comparison of conventional and deconvolution epifluorescence photomicrographs of tissue sections of Chondrosia reniformis after hybridization with the Bacteria specific probe EUB338 ([A, B] Oregon Green-, [C, D] Cy3-labeled). Twenty one optical sections were obtained along the optical axis with increments of 0.25 mm. (A) Extended focus image obtained by projection of all optical sections without prior removal of out-of-focus signals. Fine structures remain almost invisible within the blurred image. (B) Extended focus image of the identical tissue section shown in (A) after deconvolution processing of the image stack using the EPR algorithm. Note numerous small coccoid Bacteria within the sponge mesohyl and rod-shaped bacteria colonizing the channel system of the sponge cortex. (C) Extended focus image obtained by projection of unprocessed image data showing a blurred microcolony of rod shaped bacteria. (D) Identical focus image given in (C) after application of deconvolution algorithm resulting in the visualization of specific cell features such as dividing states (indicated by arrow 1) and inclusion bodies (indicated by arrow 2). Fig. 3. Comparison of conventional and deconvolution epifluorescence photomicrographs of Chondrosia reniformis after FISH with oligonucleotides on different phylogenetic levels. (A) Extended focus image of a series of 21 unprocessed optical sections obtained after hybridization with Bacteria specific probe EUB338 (Oregon Green-labeled) and Cy3-labeled probe SRB385Db. (B) Extended focus image obtained by projection of the identical image stack shown in (A) after deconvolution processing resulting in the detailed population analysis and the visualization of specific cell states (dividing cells indicated by arrows). (C and D) Detailed view of images given in (A) and (B) showing the maximum pixel enlargement for conventional epifluorescence (C) and the improved resolution after application of the deconvolution algorithm (D). (E) Extended focus image generated by projection of unprocessed serial sections obtained with the red channel of the CCD camera showing a blurred microcolony of bacteria hybridizing with the Cy3-labeled probe DSV1292. (F) Extended focus image of the identical tissue section shown in (E) after deconvolution processing. The combined color-encoded green and red channel projection shows the Desulfovibrionaceae stained with the Cy3-labeled probe DSV1292 surrounded by large amounts of coccoid cells within the sponge mesohyl simultaneously hybridizing with the Bacteria specific probe EUB338 (Oregon Green-labeled). All pictures were done at a magnification of 3 630. W. Manz et al. / Journal of Microbiological Methods 40 (2000) 125 –134 131 132 W. Manz et al. / Journal of Microbiological Methods 40 (2000) 125 –134 of 21 digital images acquired with increments of numbers of cells hybridizing with the Bacteria 0.25 mm each (Fig. 3D). Whereas the cell boundaries specific probe EUB338, and 20% (6 4.2) of the total of the bacteria in images obtained by conventional cell counts emitted strong fluorescent signals after epifluorescence microscopy could not be resolved, hybridization with the SRB specific probe application of the deconvolution algorithm to images SRB385Db (Fig. 3B). Again, the application of obtained from the same specimen showed clearly deconvolution processing remarkably improved the distinguishable cell boundaries and allowed the optical resolution and showed the typical morpholo- analysis of dividing cells forming defined mi- gy of dividing cells within the SRB population in the crocolonies. sponge mesohyl (Fig. 3B, indicated by arrows). Further detailed analysis of the sponge associated 3.3. Visualization of the spatial arrangement of SRB population performed with probes on more sponge associated bacteria in Chondrosia narrow phylogenetic levels elucidated the presence reniformis of microcolonies hybridizing with the Cy3-labeled probe DSV1292, indicating members of the family For the first time, great amounts of coccoid cells Desulfovibrionaceae (Fig. 3E,F). The simultaneous (average cell diameter 0.7 mm) located within the application of Oregon Green-labeled probe EUB338 sponge mesohyl could be successfully visualized and Cy3-labeled probe DSV1292 followed by de- (Fig. 2B). Some of the rod shaped cells on the convolution processing showed that the microcolony surface of the sponge cortex were remarkably en- of Desulfovibrionaceae within the sponge channel larged (4 mm in length, 1 mm in width) and showed system was densely surrounded by coccoid cells with significantly stronger fluorescence signal intensities an average cell diameter of 0.7 mm which formed the compared to other sponge associated bacteria (Fig. vast majority of the sponge associated bacteria 2B). Detailed analysis of the distribution of the within the mesohyl (Fig. 3F). Caused by the overlay sponge associated cells within different parts of the of green and red fluorescence, SRB appear yellow in sponge revealed that microcolonies formed by Fig. 3F. strongly fluorescent, rod shaped bacteria were always directly located at micropores of Chondrosia re- 3.4. Key features of widefield deconvolution niformis. The putatively higher metabolic potential microscopy of these cells might be caused by the additional accessibility of electron acceptors (e.g. sulfate from The most critical point during the deconvolution the ambient seawater) and/or the improved availabil- image process is the calibration of the point spread ity of substrates through the sponge channel system. function. Depending on the CCD camera used, a The application of a top-to-bottom directed approach high signal to noise ratio and high optical resolutions using rRNA targeted oligonucleotide probes revealed can be realized. Since no laser beam is necessary for the phylogenetic affiliation and three-dimensional illumination, photobleaching of fluorochromes or arrangement of sulfate-reducing bacteria (SRB) photodamage of the investigated specimen is less among the sponge associated bacterial community. critical than using confocal laser scanning micro- SRB play an important role in pyritization within scopy which require laser beams emitting high marine systems as previously discussed (Reitner and photon densities. Confocal laser scanning micro- Schumann-Kindel, 1997). The presence of culturable scopy offers the possibility to directly acquire 2D SRB originating from sponge material could be imaging of 3D objects, but the maximum signal to recently demonstrated by the application of specific noise ratio and the optical resolution of confocal in situ probes (Schumann-Kindel et al., 1997). laser scanning microscopy is lower than using wide- However, a defined three-dimensional localization of field epifluorescence microscopy coupled with de- the SRB within the natural structure of the sponge convolution rendering. tissue could not be shown yet. In the present study, While mathematical approaches for deblurring hybridization of thin sections of paraffin and Tech- microscopic images are less expensive than hard- novit embedded sponge material revealed large ware-based optical sectioning methods, there are W. Manz et al. / Journal of Microbiological Methods 40 (2000) 125 –134 133 Gerrits, P.O., Smid, L., 1983. A new, less toxic polymerizationlimitations to the number of depths that may be used system for the embedding of soft tissues in glycol-methacrylateduring the calculation as well as to the complexity of and subsequent preparing of serial sections. J. Microsc. 132, the image. For example, the deconvolution of 10–20 81–85. slices per object requires a complex calculation Haugland, R.P., 1996. Handbook of Fluorescent Probes and Research Chemicals, Molecular Probes, Eugene, OR.process, and the time for iterative calculations re- Holmes, T.J., Bhattacharyya, S., Cooper, J.A., Hanzel, D., Kris-quired with the PC system described above up to 8 h hnamurthi, V., Lin, W., Roysam, B., Szarowski, D., Turner, J., of processing. In order to reduce the number of slices 1995. Light microscopic images reconstructed by maximum necessary for optimum deconvolution efficiency, the likelihood deconvolution. In: Pawley, J.B. (Ed.), Handbook of Biological Confocal Microscopy, Plenum Press, New York, pp.combination of thin-sectioning methods and de- 389–402.convolution processing seems to be appropriate for Lawrence, J.R., Korber, D.R., Wolfaardt, G.M., Caldwell, D.E.,the analysis of more complex specimen at present. In 1996. Analytical imaging and microscopy techniques. In: this study, sampling usually comprised up to 21 Hurst, C.J., Knudsen, G.R., McInerney, M., Stetzenbach, L.D., slices, originated from series of z-steps with 0.25– Walter, M.V. (Eds.), Manual of Environmental Microbiology, American Society for Microbiology Press, Washington, DC, pp.0.5-mm wide increments. 29–51.In the near future, the development and applica- Licht, T.R., Krogfelt, K.A., Cohen, P.S., Poulsen, L.K., Urbance, tion of further deconvolution algorithms, e.g. blind J., Molin, S., 1996. Role of lipopolysaccharide in colonization deconvolution algorithms which do not require the of the mouse intestine by Salmonella typhimurium studied by calibration of the point spread function, and the in situ hybridization. Infect. Immun. 64, 3811–3817. MacNaughton, S.J., Booth, T., Embley, T.M., O’Donnell, A.G.,availability of more powerful personal computers 1996. Physical stabilization and confocal microscopy of bac-may provide easier and faster handling of deconvolu- teria on roots using 16S rRNA targeted fluorescent-labeled tion microscopy and will significantly reduce the oligonucleotide probes. J. Microbiol. Methods 26, 279–285. calculation time needed for more complex images. Manz, W., Amann, R., Ludwig, W., Wagner, M., Schleifer, K.-H., 1992. Phylogenetic oligodeoxynucleotide probes for the major subclasses of proteobacteria: problems and solutions. System. Appl. Microbiol. 15, 593–600. Acknowledgements ¨Manz, W., Amann, R., Szewzyk, R., Szewzyk, U., Stenstrom, T.-A., Hutzler, P., Schleifer, K.-H., 1995. In situ identification of Legionellaceae using rRNA-targeted oligonucleotide probesThe authors gratefully acknowledge the support and confocal laser scanning microscopy. Microbiology 141,from the Deutsche Forschungsgemeinschaft (Re 665/ 29–39.12-1 LEIBNIZ-Award). This publication is contribu- Manz, W., Eisenbrecher, M., Neu, T.R., Szewzyk, U., 1998. tion no. 16 of the Collaborative Research Center Abundance and spatial organization of gram-negative sulfate- SFB 468 ‘Wechselwirkungen an geologischen reducing bacteria in activated sludge investigated by in situ probing with specific 16S rRNA targeted oligonucleotides. ¨ ¨Grenzflachen’ at the University of Gottingen, Teil- FEMS Microbiol. Ecol. 25, 43–61.projekt A1, also funded by the Deutsche Forschungs- Moter, A., Leist, G., Rudolph, R., Schrank, K., Choi, B.K.,gemeinschaft. ¨Wagner, M., Gobel, U.B., 1998. Fluorescence in situ hybridiza- tion shows spatial distribution of yet uncultured treponemes in biopsies from digital dermatitis lesions. Microbiology 144, 2459–2467.References Poulsen, L.K., Lan, F., Kristensen, C.S., Hobolth, P., Molin, S., Krogfelt, K.A., 1994. Spatial distribution of Escherichia coli in Amann, R., Krumholz, L., Stahl, D.A., 1990. Fluorescent-oligo- the mouse large intestine inferred from rRNA in situ hybridiza- nucleotide probing of whole cells for determinative, phylo- tion. Infect. Immun. 62, 5191–5194. genetic, and environmental studies in microbiology. J. Bac- Rabus, R., Fukui, M., Wilkes, H., Widdel, F., 1996. Degradative teriol. 172, 762–770. capacities and 16S rRNA-targeted whole cell hybridization of Amann, R.I., Ludwig, W., Schleifer, K.-H., 1995. Phylogenetic sulfate-reducing bacteria in an anaerobic enrichment culture identification and in situ detection of individual microbial cells utilizing alkylbenzenes from crude oil. Appl. Environ. Mi- without cultivation. Microbiol. Rev. 59, 143–169. crobiol. 62, 3605–3613. Carrington, W.A., Fogarty, K.E., Lifschitz, L., Fay, F.S., 1995. Reitner, J., Schumann-Kindel, G., 1997. Pyrite in mineralized Three-dimensional imaging on confocal and wide-field micro- sponge tissue. Product of sulfate reducing sponge-related scopes. In: Pawley, J.B. (Ed.), Handbook of Biological Con- bacteria? Facies 36, 272–276. focal Microscopy, Plenum Press, New York, pp. 151–161. Rothemund, C., Amann, R., Klugbauer, S., Manz, W., Bieber, C., 134 W. Manz et al. / Journal of Microbiological Methods 40 (2000) 125 –134 Schleifer, K.-H., Wilderer, P., 1996. Microflora of 2,4-dichloro- confocal microscope: its measurement and use in deconvolu- phenoxyacetic acid degrading biofilms on gas permeable tion of 3D data. J. Microsc. 163, 151–165. membranes. Syst. Appl. Microbiol. 19, 608–615. Wagner, M., Assmus, B., Hartmann, A., Hutzler, P., Amann, R., Schumann-Kindel, G., Bergbauer, M., Manz, W., Szewzyk, U., 1994. In situ analysis of microbial consortia in activated sludge Reitner, J., 1997. Aerobic and anaerobic microorganisms in using fluorescently labelled, rRNA-targeted oligonucleotide modern sponges: A possible relationship to fossilization pro- probes and confocal laser scanning microscopy. J. Microsc. cesses. Facies 36, 268–272. 176, 181–187. Shaw, P.J., Rawlins, D.J., 1991. The point-spread function of a