The suitability of the recently described red fluorescent protein dsRED from reef corals for use as a reporter in plant molecular biology was investigated. Based on the clone pDSRED (Clontech), plant expression vectors were constructed for constitutive dsRED expression in the cytosol, the endoplasmic reticulum and the vacuole. Fluorescence microscopy of tobacco BY2 suspension culture cells transiently expressing the plant vectors generated proved that cytosolic expression of the dsRED gives rise to readily detectable levels of red fluorescence, whereas expression in the ER was poor. Vacuolar dsRED expression did not result in any significant fluorescence. dsRED transgenic tobacco SR1 plants were generated to test the sensitivity of dsRED as a reporter in an autofluorescent background, and to identify the possible impact of the introduced fluorescent protein on morphogenesis, plant development and fertility. During the transformation and regeneration phase plants did not show any abnormalities, indicating that dsRED is not interfering with plant development and morphogenesis. Regenerated plants were analysed by PCR, Western blot and fluorescence microscopy for the presence and expression of the transferred genes. The filter sets chosen for fluorescence microscopy proved to be able to block the red chlorophyll fluorescence completely, allowing specific dsRED detection. Best expression levels were obtained with dsRED targeted to the cytosol or chloroplasts. ER-targeted expression of dsRED also gave rise to readily detectable fluorescence levels, whereas vacuolar expression yielded no fluorescence. dsRED transgenic plant lines expressing the protein in the cytosol, ER or chloroplast proved to be fertile. Seed set and germination were normal, except that the seeds and seedlings maintained the red fluorescence phenotype.
Reporter proteins and the genes encoding them are essential tools of molecular biology, frequently used to monitor developmental and spatial gene expression patterns. Popular reporter proteins include β-glucuronidase (GUS), luciferase (LUC), chloramphenicol acetyltransferase (CAT), and the more recently introduced green fluorescent protein (GFP). Whereas monitoring of CAT, GUS and LUC expression (activity) requires preparation of protein extracts, addition of suitable substrates and performance of enzyme assays, GFP has become very attractive due to the fact that it allows non-invasive, non-destructive detection.
In the past much effort has been spent on the improvement and optimization of GFP as a reporter protein, and on the identification of GFP mutants with altered spectral properties (e.g. altered emission peak wavelength) in order to create tools for simultaneous expression monitoring of separate genes or to obtain suitable donor–acceptor pairs to investigate fluorescence resonance energy transfer (Chalfie, 1995; Cubitt et al., 1995; Más et al., 2000; Pollok and Heim, 1999). Currently, GFP mutants emitting blue, cyan and yellow light are available (Cubitt et al., 1995): BFP, CFP and YFP, respectively. However, discrimination of these GFP mutants can be complicated due to their closely related spectral properties (Jakobs et al., 2000). Despite the fact that GFP has established itself as a very important and versatile reporter in plant molecular biology, at least in our hands, detection of GFP in plants sometimes suffers from the difficulty in distinguishing GFP signals from the autofluorescent background of green plant tissue, which is due to chlorophyll. The very strong red fluorescence signals appearing upon excitation of green tissues with blue light (470 nm) masks GFP fluorescence (Chiu et al., 1996). Hence, for a fluorescent protein to be of any use as a reporter in plant molecular biology, the ability to separate its fluorescence from this background is crucial.
Very recently a novel red fluorescent protein, dsRED, was isolated from reef corals (Discosoma sp.) possessing an excitation peak wavelength (553 nm) just above the excitation peak of chlorophyll (Matz et al., 1999). Thus, by using the appropriate filter sets, it should be possible to excite dsRED specifically, thereby avoiding the chlorophyll autofluorescence. Más et al. (2000) provided the first evidence for the successful use of dsRED in plant cells by showing nuclear localized dsRED expression in tobacco BY2 protoplast cells. However, only transient expression assays were performed. Nothing is known about the performance of dsRED as a reporter in stably transformed plants, and knowledge is lacking about possible interference of dsRED expression in planta with growth, development and metabolism of a target plant.
The experiments described here were performed to investigate the recently described red fluorescent protein dsRED from Discosoma sp., with respect to its ability to act as reporter protein in plant cells. A series of plant-expression vectors was constructed to test for functional expression of dsRED in different subcellular compartments such as ER, cytosol, vacuole and chloroplast. The vectors were used to perform transient expression studies in tobacco BY2 protoplasts, and to generate dsRED transgenic tobacco plants. Using this material we investigated the specificity and sensitivity of dsRED detection in non-autofluorescent and autofluorescent plant cells, as well as plant development, morphogenesis, fertility and germination of dsRED transgenic plants.
Results and discussion
Cloning of dsRED plant expression vectors
Site-directed mutagenesis was used to insert an NcoI and a BamHI site at the 5′ and 3′ prime ends, respectively, of the dsRED coding region from plasmid pDSRED (Clontech), and to remove internal EcoRI and HindIII sites from the gene (for details see Experimental procedures). The resulting modified dsRED gene was subcloned into a previously generated plant expression vector to generate dsRED expression cassettes allowing cytosolic and secretory expression of the protein (Reichel et al., 1996; G.J., unpublished results). Subsequently, site-directed mutagenesis was used to further modify the dsRED coding region to direct ER-targeted and vacuolar expression.
Jang et al. (1999) fused a chloroplast transit peptide derived from the small subunit gene of ribulose bisphosphate carboxylase/oxygenase (rbcS) of rice to the green fluorescent protein, and demonstrated successful plastid-localized GFP expression. In addition, the authors showed that chloroplast targeting of GFP results in a marked increase of GFP protein levels compared to control expression of GFP in the cytosol. To construct a dsRED expression vector allowing for chloroplast-localized dsRED expression, we therefore cloned a synthetic coding region for the rice rbcS transit peptide and generated a translational fusion with dsRED (for cloning details see Experimental procedures). Based on the resulting expression cassettes (presented schematically in Figure 1), two sets of vectors were cloned: for transient gene expression studies, pGJ1425 (35S-dsRED), pGJ1476 (35S-dsRED-ER), pGJ1477 (35S-dsRED-vac), pGJ1862 (35S-dsRED-Chl); and for plant transformation work the binary vectors pGJ1485 (35S-dsRED), pGJ1486 (35S-dsRED-ER), pGJ1487 (35S-dsRED-vac), pGJ1878 (35S-dsRED-Chl).
Transient dsRED expression in tobacco BY2 protoplasts
Tobacco BY2 protoplasts transfected with constructs pGJ1425 (35S-dsRED), pGJ1476 (35S-dsRED-ER) and pGJ1477 (35S-dsRED-vac) and incubated for 18 and 40 h under constant environmental conditions (26°C, dark, K3 medium) were analysed for red light emission by fluorescence microscopy using a dsRED specific filter set specially assembled in close co-operation with AF Analysentechnik (AHF Analysentechnik AG, Túbingen, Germany), Germany. Cells harbouring 35S-dsRED (driving cytosolic expression) showed readily detectable red fluorescence with a distribution similar to that for cytosolically expressed GFP (Figure 2), thereby proving the correct localization of the protein. However, the total number of fluorescent cells was at least fivefold lower than seen in control cells transfected with a similar expression vector bearing GFP as reporter gene (data not shown). In contrast, few cells expressing extremely weak signals were obtained for ER-targeted dsRED (35S-dsRED-ER), and no significant fluorescence signals were found in cells transfected with 35S-dsRED-vac, driving vacuolar expression of dsRED. In consequence, correct localization of the modified dsRED proteins could not be verified in these experiments. When investigated 40 h post-transfection, the total number of cells showing red fluorescence was significantly higher for both cytosolic and ER-targeted dsRED, but was still below the numbers obtained for the control construct bearing GFP. Differences in signal strength between dsRED and dsRED-ER remained the same.
The maturation rate of purified dsRED protein was found to be very slow, requiring days at room temperature to reach full light emission. Half-maximum red fluorescence was obtained after approximately 27 h (Baird et al., 2000). The authors also provided several lines of evidence indicating that dsRED occurs as an obligate oligomer in vitro and in vivo, strongly supported by recent structural data (Wall et al., 2000; Yarbrough et al., 2001). Although knowledge about dsRED maturation is growing rapidly (Gross et al., 2000; Terskikh et al., 2000), it is still not known whether oligomerization is required for maturation or efficient light emission. Based on these results, it was calculated that after 18 and 40 h (incubation times of the BY2 cells), approximately 30 and 60% of the freshly synthesized dsRED protein in the BY2 cells already emits red light. Unfortunately, nothing is known about dsRED maturation rates in planta. Assuming that the lower number of fluorescent cells (compared to the GFP control) is due to the slow maturation of the dsRED protein, the differences found are only partly explained by these calculations. We concluded that the in planta maturation rate of cytosolically expressed dsRED is probably even slower than that in vitro. A lower fluorescence yield of the dsRED protein would be an alternative explanation; the quantum yield and extinction coefficient of dsRED were first described to be about threefold lower with respect to GFP (Matz et al., 1999). However, more recent data indicate that the optical properties of dsRED and GFP are comparable (Baird et al., 2000; Gross et al., 2000), making this explanation less likely.
In the case of vacuolar targeted dsRED, the relative acid pH of this compartment (pH 5–6) may play an important role, although this appears to be unlikely as dsRED fluorescence was found to be quite stable at pH values ranging from 4.5 to about 10 (Baird et al., 2000). In addition, detrimental effects on protein folding, chromophore formation or light emission due to the presence of the additional amino acids at the C-terminus of dsRED can be ruled out, as recombinant dsRED-ER and dsRED-vac produced in Escherichia coli showed spectral properties similar to the unmodified dsRED (data not shown). Hence the poor results for the ER- and vacuole-targeted dsRED versions are probably due to impacts of these environments on the maturation rate or oligomerization of dsRED protein. Unfortunately, immunoblotting experiments did not result in conclusive data about protein amounts and oligomerization states of the dsRED, dsRED-ER and dsRED-vac produced in BY2 protoplasts (data not shown). More experiments are needed to clarify this point. In particular, knowledge about the maturation of dsRED at low pH is needed, and future work will address this point in more detail.
Since the introduction of the green fluorescent protein into molecular biology, attempts have been made to modify the gene to give GFP emission colours other than green, in order to allow simultaneous multicolour tracking of separate genes or to create donor–acceptor pairs for fluorescence resonance energy transfer. Today GFP mutants are available emitting blue, cyan and yellow light. However, due to their largely overlapping emission spectra, discrimination between GFP mutants in a mixture requires high-tech equipment such as confocal laser microscopes, and is complicated. In contrast, using similar equipment, discrimination between dsRED and GFP appears to be quite easy (Jakobs et al., 2000; Más et al., 2000).
We raised the question whether the spectral properties of the proteins are sufficiently different to allow easy discrimination, even by conventional (low-tech) fluorescence microscopy. To test this, tobacco BY2 protoplasts were transfected either with the plasmid 35S-dsRED or pCATgfp (35S-GFP), or with a combination of both plasmids. Cells were incubated and analysed as described above using dsRED- and GFP-specific filter sets. Using the dsRED filter, only cells emitting red fluorescence could be detected (Figure 2). No significant fluorescent signals were obtained when analysing cells solely expressing GFP with these filters. In the reciprocal situation (dsRED-expressing cells, GFP filters), dsRED expression could be detected as (weak) orange-red fluorescence. This is due to the fact that dsRED can be excited partially with blue light (470 nm) and that the chosen GFP filters do not exclude red light (long-pass filters). Under the same conditions, GFP gave rise to very clear signals.
These experiments demonstrate that conventional fluorescence microscopy is sufficient to distinguish between dsRED and GFP co-expressed in plant cells (Figure 2). Nevertheless, it has to be taken into account that monitoring of GFP cannot be completely independent due to the partial excitation of dsRED at the excitation peak wavelength of GFP.
dsRED expression in stably transformed tobacco plants
To test whether dsRED affects the development, morphogenesis, fertility and germination of plants, we decided to produce transgenic tobacco plants via Agrobacterium-mediated plant transformation using the binary vectors pGJ1485 (35S-dsRED), pGJ1486 (35S-dsRED-ER), pGJ1487 (35S-dsRED-vac) and pGJ1878 (35S-dsRED-Chl) harbouring the dsRED expression cassettes (Figure 1). Trans formation experiments with the empty vector (pBIN19) served as negative controls. Leaf-disk transformation experiments were monitored under a fluorescence stereomicroscope. As early as 5–7 days post-infection with Agrobacterium, microcalluses could be identified displaying red fluorescence. As expected, no signals were obtained for plant material transformed with the empty vector control (pBIN19). For both constructs, 35S-dsRED and 35S-dsRED-ER, numerous dsRED expressing calluses were easily detectable by 9–12 days post-infection (Figure 3), whereas transformation with constructs 35S-dsRED-vac and 35S-dsRED-Chl did not give rise to any detectable red fluorescence (not shown).
Transgenic lines expressing dsRED constructs regenerated normally and were not phenotypically altered. From the regenerated plants, 10 were randomly chosen and analysed by Southern hybridization experiments for the presence and numbers of intact copies of transferred T-DNAs in the genomic DNA of the tobacco plants (data not shown). RT–PCR was used to check for proper transcription of transferred dsRED genes in regenerated plants carrying single T-DNA insertions (Figure 4a). All tested plants contained dsRED transcripts, thereby proving function of the inserted genes.
Western blot experiments were performed to verify the presence of dsRED proteins in extracts of the regenerated transgenic plants, thereby proving translation of the detected transcripts (Figure 4b). The first experiments proved dsRED to be easily detectable in total protein extracts from plants expressing the cytosolic and chloroplast-targeted version of the protein, whereas not even a faint signal could be obtained for both ER- and vacuole-targeted dsRED. In particular, the negative results for dsRED-ER were very surprising, as fluorescent microscopy has already demonstrated dsRED expression in these plants (Figure 4c). Further experiments investigating soluble and insoluble protein fractions showed that dsRED-ER and dsRED-vac protein is present only in the insoluble fraction (Figure 4b). These two modified versions of the dsRED protein show a strong tendency to form large, insoluble, yet fluorescent (dsRED-ER) complexes upon expression in plant cells. The C-terminal four amino acids were recently found be part of the dsRED dimer interface (Yarbrough et al., 2001), and it is likely that the additional amino acids present in dsRED-ER and dsRED-vac will cause interference, apparently causing enhanced oligomerization. Work is currently in progress to investigate this in more detail.
Positive signals (including the recombinant dsRED used as positive control) showed a double band, with the size of the upper band being in good accordance with the size expected for mature dsRED. According to published data (Baird et al., 2000), the fragmentation seen is probably due to the sample preparation (boiling), and does not reflect the in planta status of the protein. Data obtained for the chloroplast-targeted dsRED indicate correct post-translational processing of the protein leading to removal of the N-terminal fused 49 amino-acid rbcS transit peptide. Cytosolically expressed dsRED showed the highest expression levels, closely followed by the chloroplast targeted protein. With respect to these proteins, both ER- and vacuole-targeted dsRED showed drastically reduced expression levels, with dsRED-ER being two- to threefold higher than dsRED-vac. The reasons for the very poor expression levels obtained with the vacuolar-targeted dsRED are unknown and need to be investigated in more detail.
Subsequently, transgenic plant lines were investigated by fluorescence microscopy (Figure 4c). As expected, leaf tissue of wild-type control plants did not give rise to any significant red fluorescence, whereas regenerated plants expressing 35S-dsRED, 35S-dsRED-ER and 35S-dsRED-Chl fluoresced brightly. In accordance with the results of our transient expression studies, no significant signals were found for the vacuolar targeted dsRED, which is probably due to the very low expression level obtained with this construct (Figure 4b).
The results for 35S-dsRED-ER transgenic plants are in contrast to the results obtained in transient expression assays. Although the reasons are unknown, it seems unlikely that the differences found are due to physiological differences between BY2 and SR1 cells, as BY2 cells can express other proteins, such as GFP, to high levels in the ER. We assume that the relatively long regeneration phase of the transgenic plants gives the dsRED protein more than enough time to complete maturation and to accumulate, even in compartments less favourable for dsRED maturation than the cytosol, which is probably the case for the ER. Fluorescence signals of plants expressing dsRED-ER were often found to be strongest at the tip of the leaves, gradually decreasing towards the petioles, and were generally much stronger than expected based on the Western blot results. As release of dsRED from the insoluble matter proved to be difficult, we assume that this discrepancy is caused by underestimations of the dsRED-ER amounts, due to only partial solubilization of the protein during extraction.
As mature plants expressing 35S-dsRED-Chl clearly show dsRED expression and red fluorescence, the lack of fluorescent calluses during the early stages of the plant transformation is probably due to the lack of chloroplasts or suitable chloroplast precursors in these cells.
All transgenic plants were found to be fertile, and seeds from self-pollinated 35S-dsRED and 35S-dsRED-ER transgenic plants showed bright red fluorescence (Figure 4c). A closer look revealed significant differences in brightness, with a minor proportion of the seeds showing either no dsRED expression or being almost twice as bright as the average value. These differences are probably due to the segregation of the transferred dsRED gene in this R1 population leading to homozygous and heterozygous dsRED transgenic seeds and non-transgenic seeds in ratios of 1 : 2 : 1. Future experiments will address this point in more detail.
During germination, dsRED transgenic seeds showed normal growth and were indistinguishable from other transgenic or wild-type seeds. An examination by fluorescence microscopy revealed that the seedlings of dsRED transgenic plants maintain the red fluorescence phenotype (Figure 4d).
In previous experiments with BY2 protoplasts, we have not been able to verify the localized expression of the modified proteins dsRED-ER and dsRED-Chl. Therefore mesophyll protoplasts were prepared from leaf tissue of transgenic plants and investigated under the fluorescent microscope (Figure 5). Distribution of red fluorescence within the various protoplasts meets expectations in all cases, indicating that the dsRED proteins are localized in the targeted cell compartments, thus proving the functionality of the targeting signals added to the protein. Despite the enhanced oligomerization of dsRED-ER detected in the Western blot experiments, fluorescence microscopy of mesophyll protoplasts did not reveal either any artefactual protein aggregation or the presence of abnormal ER structures or mysterious organelles as described for ER-targeted GFP (Haseloff et al., 1997). Future experiments will investigate the properties of dsRED-ER in more detail.
From a technical point of view, the experiments described in this study clearly demonstrate that the optimized dsRED filter set abolishes background fluorescence completely, allowing for specific dsRED detection even in autofluorescent green plant tissue (Figures 3 and 4). In this respect, dsRED protein is certainly superior to the various GFP mutants currently used in our laboratory. In practical terms, detection of dsRED expression in green tissue proved to be easier than detection of GFP, at least in our hands. This applied particularly when faint expression signals were being analysed.
In summary, the recently described dsRED protein from Discosoma sp. was found to be a well suited, easy-to-detect reporter protein for plant molecular biology, especially for cytosolic chloroplast-targeted expression in plant cells. dsRED expression in other compartments, such as ER, appears to be possible, whereas vacuolar expression failed. In addition, dsRED proved to be suitable for double-labelling experiments, as red light emission can be easily discriminated from that of GFP even by conventional fluorescence microscopy. dsRED expression had no negative effects on growth, development, fertility or germination of the plants examined.
Further work appears to be necessary to test the suitability of dsRED as a fluorescent tag to be used in localization studies. The results presented here suggest that C-terminal extension of the protein can cause enhanced oligomerization and formation of huge, insoluble protein complexes.
Cloning of plant expression vectors
All cloning work was carried out according to standard procedures (Maniatis et al., 1982). All site-directed mutagenesis experiments were done using the Stratagene chameleon kit (Stratagene Europe, Amsterdam, The Netherlands) and the following oligonucleotides:
Site-directed mutagenesis was used to modify plasmid pDSRED (Clontech) with primers P280GJ, P281GJ, P283GJ and P284GJ simultaneously. Thus plasmid pGJ1420 was created carrying a modified dsRED gene characterized by a novel NcoI site marking the translational start of the coding region, removal of an internal EcoRI – and HindIII – site and a novel BamHI-site downstream and adjacent to the stop codon. To avoid confusion, the resulting modified gene was named gjRFP for internal use.
Plant expression vectors pGJ1425 and pGJ1448, allowing for cytosolic and secretory expression of dsRED protein, were generated by subcloning of the NcoI/BamHI fragment of pGJ1420 into vectors pCK-GFP-S65C (Reichel et al., 1996) and pGJ1235 (G.J. unpublished results). Primers P292GJ and P293GJ were used in site-directed mutagenesis experiments to insert coding sequences for an ER retention signal (KDEL) and a vacuolar targeting signal (GLLVDTM, Neuhaus et al., 1991) at the 3′ end of the gjRFP gene of plasmid pGJ1235. The resulting clones were named pGJ1476 (coding for dsRED-ER) and pGJ1477 (coding for dsRED-vac).
To clone a plant expression vector driving plastid localized expression of dsRED, we rationally designed a DNA sequence encoding a chloroplast transit peptide similar to one from the small subunit of ribulose bisphosphate carboxylase/oxygenase (rbcS) from rice (Jang et al., 1999):
In addition to an optimized translational start site, the designed sequence carried 19 nucleotide exchanges (12.4%; with respect to the published sequence from rice) to optimize codon usage of this synthetic gene for dicot plants. The following sense and antisense primers were used to clone a synthetic coding region for the chloroplast transit-peptide from:
P246GJ: ACGGCGGCAGGATCAGGTGCATGCAGGCCATGGCT TAT
P250GJ: CTTGAGGCCCTGGAATGGAGCAACGGTGGTGGCGG AGG
Phosphorylated primers were mixed in equimolar amounts (10 pmoles each in single-fold T4-ligase buffer, 100 µl end volume), heated to 95°C for 4 min and then incubated for 10 min at room temperature for annealing. Subsequently, 1 unit T4 DNA ligase was added and the reaction was incubated for additional 30 min at room temperature. PCR was performed with primers P242 and P251, and 5 µl aliquots of the resulting reaction mix to amplify intact DNA fragments. The PCR program was: 2 min at 95°C, followed by 30 cycles with 40 sec 95°C, 30 sec 50°C and 30 sec 72°C. Following purification (PCR purification kit, Qiagen GMBH, Hilden, Germany), the amplification product was cut with XhoI and NcoI and inserted into plasmid pGJ1425 (35S-dsRED) resulting in construct pGJ1862 (35S-dsRED-Chl). Correct assembly of the synthetic transit peptide coding region was verified by DNA sequencing.
Binary vectors for Agrobacterium-mediated plant transformation were cloned by inserting the HindIII fragments of plasmids pGJ1425, pGJ1476, pGJ1477 and pGJ1862 into the common vector pBIN19, finally giving rise to plasmids pGJ1485 (35S-dsRED), pGJ1486 (35S-dsRED-ER), pGJ1487 (35S-dsRED-vac) and pGJ1878 (35S-dsRED-Chl).
Transient expression studies in tobacco BY2 cells
All experiments were performed as described by Reichel et al. (1996), except that commercially available degraded herring sperm DNA (Sigma-Aldrich) was used as carrier-DNA.
Agrobacterium-mediated leaf disk transformation of tobacco SR1 plants and regeneration of kanamycin-resistant transgenic plants was carried out according to standard protocols.
Analysis of transgenic plants
After isolating genomic DNA from leaf tissue of regenerated kanamycin-resistant plants (following the protocol of Murray and Thompson, 1980), Southern-hybridization experiments were used to identify plants harbouring single, intact T-DNA copies of the different constructs used for plant transformation (data not shown). To confirm expression of the transferred genes, RT–PCR was performed on total RNA extracted from the identified plant lines using a Stratagene RT–PCR kit. RNA isolation followed the protocol of Logemann et al. (1987). All RT–PCR experiments were performed according to the protocols provided by the manufacturer. Final amplification of the dsRED specific band was achieved using the primers PRFP1: GAGATATCCATGGGG TCTTCCAAG and P32: CTGGTGATTTGCGGACTCTAGAGG (pA-primer). PCR was run on a MiniCycler (MJ Research, Waltham, MA, USA) with the following program: 5 min at 95°C, 30 cycles with 1 min 94°C, 45 sec 50°C, 45 sec 72°C, and finally 10 min at 72°C. Amplification products were analysed by agarose gel electrophoresis according to standard procedures using PstI-digested lambda-DNA as size marker.
To extract soluble protein, 100 mg samples of fresh leaf material from non-transgenic and transgenic tobacco plants were homogenized in 500 µl GUS buffer (50 mm phosphate buffer pH 7, 10 mm EDTA, 5 mmβ-mercaptoethanol, 0.1% Triton X100, 0.1% N-lauroylsarcosine), centrifuged (14 200 g, 14 000 r.p.m., 5 min) and supernatants transferred to fresh tubes. The debris was washed with the same buffer and supernatants were combined with the first extract to give the final soluble protein samples, which were then either used directly or stored at −20°C. Insoluble protein samples were prepared by homogenizing the washed insoluble matter from the soluble protein preparations in GUS buffer supplemented with SDS to a final concentration of 1% and heating to 80°C for 5 min. Following centrifugation (14 000 r.p.m., 5 min) supernatants were transferred to fresh tubes and either used directly or stored at +20°C.
For Western blot analysis, protein samples were separated by SDS–PAGE (10% PAA gels) then blotted onto Immobilon membrane-filters (Amersham–Pharmacia Biotech, Freiburg, Germany) using a semi-dry blotter (Trans-Blot SD, Bio-Rad Laboratories GmbH, München, Germany). Immunological detection of dsRED was done using the Living Colors D.s. Peptide Antibody (Clontech, 8370-1) according to the specifications of the manufacturer, together with the BM Chemiluminescence Western blotting kit and horseradish peroxidase-conjugated secondary antibodies (Roche, Mannheim, Germany). Signal detection was performed in a LUMI-Imager system (Roche). Protein concentrations were determined using the method of Bradford and BSA as standard (Bradford, 1976).
For all fluorescence microscopy work, Zeiss Axiophoto and Leica MZ FL3 fluorescence stereomicroscopes equipped with filter sets purchased from AF Analysentechnik, Tübingen, Germany, were used. The dsRED filters were optimized in collaboration with this manufacturer and finally contained the following filters: excitation filter 545 nm (10 nm bandwidth; 30 nm for Zeiss microscope); dicroic mirror 565 nm LP (only on the Zeiss microscope); and a 600 nm (40 nm bandwidth) emission filter. The GFP-longpass-filter set was composed of: excitation filter 470 nm (20 nm bandwidth), dicroic mirror at 510 nm, and a 520-nm LP emission filter.
All pictures were taken using a video image system mounted on the microscope, consisting of a Hitachi CCD video camera operated by the diskus software package.