Laboratory of Tropical Crop Improvement
Catholic University Leuven
Kasteelpark Arenberg 13
An ultrasensitive camera system consisting of a slow-scan, liquid nitrogen-cooled CCD camera connected to powerful image analysis software can detect very low levels of light emission from several signal sources, and has been used for the successful recording of results for different applications in banana biotechnology. Sensitive detection and quantification of bioluminescence in transgenic banana cells following transformation with the luciferase (LUC) gene permitted reliable transient gene expression studies and screening for LUC activity in putative promoter-tagged cultures. Chemiluminescence signals were detected faster and with more flexibility using the imaging system than with film, which simplifies Southern hybridisation analysis. The CCD camera was also capable of capturing fluorescence signals, as shown by monitoring green fluorescent protein expression in transgenic banana cultures.
Keywords: Agrobacterium, CCD camera, chemiluminescence, luciferase, Musa, particle bombardment, promoter tagging
Scientific imaging with digital cameras has recently become a common tool for various research applications, and has now been introduced to medium-size plant biotechnology laboratories. When a charge-coupled device (CCD) camera incorporates an analogue-to-digital (A/D) converter on the sensor or in close proximity, it is called a digital camera. During exposure, the analogue CCD chip produces a stream of varying voltages, which are subsequently digitised in the camera and converted to a computer-compatible output. As many scientific applications involve low light emission, a highly sensitive CCD camera is often required for detection. But not all the photons that reach the CCD generate electron-hole pairs in the chip to produce a signal. Quantum efficiency (QE) is the percentage of incident photons that interact with the CCD chip. A QE of up to 90% can be realized in a back-illuminated CCD where the light falls onto the back of the CCD, which is made transparent. This kind of sensor is employed in scientific-grade, slow-scan cameras. Read-out noise and dark noise are the two major sources of noise in CCD cameras, and reduce the overall sensitivity . The amplifier on the CCD chip that converts the stored charge of each photodiode (i.e. pixel) into an analogue voltage to be quantified by the A/D converter is responsible for the read-out noise. Since the read-out noise is proportional to the read-out speed, operation at slower speed reduces this read-out noise substantially. Read-out noise of less than ten electrons per pixel is reached in modern CCD cameras. Sensitivity can be increased by longer exposure, but this is limited by the dark noise or charge generated within the CCD by leakage current (dark current), which accumulates even in the dark, resulting in false-positive white dots on the image. The lower the CCD temperature, the lower the dark current. Ideally, a liquid nitrogen cooling system should be used, since it reduces the dark current to less than one electron per pixel per hour, allowing long exposures of up to tens of minutes.
Reporter genes are frequently used for analysing gene expression at the cellular level and in tissues. In plant biotechnology, the gusA gene  encoding b-glucuronidase (GUS) is most widely used. However, assessment of GUS activity is destructive and usually lethal, which excludes in vivo monitoring of gene expression. In contrast, the luciferase gene (luc)  from the firefly Photinus pyralis allows continuous detection of reporter gene activity in plants using low-light imaging techniques [4-6]. Luciferase (LUC) catalyses the oxidative decarboxylation of the substrate luciferin, causing the release of a photon at 562 nm with a quantum yield of almost 90% . Since the LUC enzyme is only slowly regenerated after reacting with luciferin  and has a short half-life of about 3 h in animal cells [9-10], each LUC molecule can only react once and emit one photon. Provided that luciferin, oxygen and ATP are constantly available, LUC will not accumulate, allowing accurate in vivo detection of this bioluminescent reporter with sensitivity largely depending on the type of CCD camera employed. As the emphasis in our work with the luc reporter gene is on promoter tagging, in which expression is often weak, we decided to use a liquid nitrogen-cooled slow-scan CCD camera to ensure maximum sensitivity for in vivo monitoring.
In this overview article, we demonstrate the usefulness of an ultrasensitive digital CCD camera in different applications in banana biotechnology research in our laboratory. In combination with powerful image analysis software it can detect and process a very wide range of light emission levels from both transgenic cell cultures and chemiluminescent hybridisation membranes.
2. MATERIALS AND METHODS
2.1. Plant material
Embryogenic cell suspensions of the plantain 'Three Hand Planty' (accession no. of International Transit Centre ITC.0185, AAB) and the dessert banana 'Grande Naine' (ITC.1256, AAA) were maintained and subcultured in half-strength MS medium  supplemented with 5 µM 2,4-D and 1 µM zeatin, called ZZ medium .
2.2. Plasmids and bacterial strains
In plasmids pAL41 and pAL25  the maize ubiquitin promoter plus first intron (Ubi1)  controls expression of the wild-type luciferase gene (luc)  and the codon-modified luciferase gene (luc+) , respectively. T-DNA in p35Sluc19  contains the luc gene under the control of the CaMV 35S promoter, while the neo gene encoding neomycin phosphotransferase is driven by the nopaline synthase (nos) promoter. In pluc19  the T-DNA consists of a promoterless luc gene with its start codon positioned 662 bp from the left border, and the neo gene under the control of the nos promoter. The T-DNA of the binary vector pBINUbi-sgfpS65T  contains the neo gene driven by the nos promoter, while the Ubi1 promoter controls expression of the sgfpS65T-encoded green fluorescent protein (GFP) . Plasmid pIGNPAce contains the Ace-AMP1 (first antimicrobial protein from Allium cepa)  coding region comprising the mature Ace-AMP1 with its natural signal sequence and the neo gene under control of the nos promoter. Agrobacterium strains EHA105 (p35Sluc19), EHA105 (pluc19) and AGLO (pBINUbi-sgfpS65T) were used for transformation.
2.3. Genetic transformation
Embryogenic suspension cells were transformed by particle bombardment , with the following modifications for in vivo transient luciferase (LUC) expression (TLE) analyses. 100 µl of a 33% settled cell volume suspension (approximately 25 mg fresh weight cells) was evenly dispersed onto a small 50 µm polyester mesh and bombarded at -0.8 bar vacuum with a home-made, helium-driven particle inflow gun. The mesh with cells was then transferred to a well of a black 24-well plate (662160, Greiner, Germany) containing ZZ medium.
Agrobacterium-mediated transformation of embryogenic suspension cells was done as described by Pérez Hernández  with some modifications. Induction medium consisted of YEP medium (10 g/l Bacto yeast extract, 10 g/l Bacto peptone, 5 g/l NaCl, pH 7.5) supplemented with the appropriate antibiotics and 200 µM acetosyringone (AS). Induced bacterial cultures were diluted to an OD600 of approximately 0.4, and then transferred to antibiotic-free ZZ medium (pH 5.6) containing 200 µM AS. Each sample of embryogenic suspension cells of 200 µl at 33% settled cell volume (approximately 50 mg fresh weight of cells) was mixed with 1 ml induced agrobacteria in a well of a 24-well plate, and incubated in the dark at 25°C on a rotary shaker at 25 r.p.m. for 6 h.
2.4. Digital imaging system
Figure 1 Ultrasensitive CCD camera imaging system used for the recording of results of different applications in banana biotechnology. Light signals are detected with a liquid nitrogen-cooled slow-scan CCD camera, A (Versarrayä 512 B LN, Roper Scientific, USA), provided with a light-sensitive camera lens, B (Nikkor F 50 mm f1.2, Nikon, Japan), in a light-tight box, C (Progress Control, The Netherlands). Digitised images are transferred to the computer for display on the monitor and further processing using image analysis software, D (MetaMorph® 4.5, Universal Imaging, USA)
Light emission was captured with a back-illuminated, slow-scan, liquid nitrogen-cooled CCD camera (Versarrayä 512 B LN, Roper Scientific, USA) (Figure 1, A) attached to a light-sensitive camera lens (Nikkor F 50mm, f1.2, Nikon, Japan) (Figure 1, B) and placed in a light-tight box (Progress Control, The Netherlands) (Figure 1, C) to eliminate background illumination. Quantification of photon emission was done by computer (MetaMorph® 4.5 software, Universal Imaging, USA) (Figure 1, D). For a region of interest, total light intensity output is given in relative grey levels, while average light intensity output is given in relative grey levels/pixel. Signals can be enhanced and noise suppressed by proper setting of the grey scale adjustment parameters, using the software. Each acquired image contains 16 bits per pixel or 65536 grey levels. The software automatically assigns a background cut-off threshold and a high signal cut-off threshold for each image. These thresholds are applied to the range of 65536 grey levels. The background lies at the low end of this range and the signal(s) at the high end of the range. The range between these threshold values is then divided into 8 bit 256 levels, because a computer monitor is usually 8 bit and can only display 256 grey shades. To obtain a clearer image the range of grey levels in each image can be modified by adjusting the cut-off threshold values.
2.4.1. In vivo LUC detection
For TLE analysis, each sample of bombarded cells was saturated with 50 µl luciferin solution (0.1 or 0.4 mM beetle luciferin potassium salt, Promega, The Netherlands) in half-strength MS medium. After placing in the light-tight box, a reference image was first taken under normal light conditions. Then in complete darkness, LUC activity was measured 5 min after adding luciferin for 1 min (integration time). LUC activity of each sample (region of interest) was corrected for background due to read-out noise as determined for control, non-transformed cells. Images of LUC activity are depicted in pseudocolours (black-purple indicating low activity, yellow-red indicating high activity). For screening stable transformants, luciferin solution (20-60 µl depending on the size) was applied onto proliferating cell cultures and LUC activity was determined after an integration time of 20 min as described above.
2.4.2. In vitro LUC detection
Cells were collected in a siliconised Eppendorf tube using forceps, quickly submerged in liquid nitrogen and stored at -80°C, or immediately transferred to a glass tube containing 200 µl Cell Culture Lysis Reagent (E1561, Promega, The Netherlands). Following sonication in ice-cold water for 10 min at 50 Hz (ultrasonic bath, Bransonic 220), the lysate was transferred to a siliconised Eppendorf tube, stored for 5 min on ice, and centrifuged (13,000 r.p.m.) for 10 min at 4°C. The supernatant was transferred to a fresh siliconised Eppendorf tube and centrifuged again to remove remaining debris.
Prior to storage at -80°C an aliquot of the extract was removed for determination of total protein content with a modified Lowry assay , the DC Protein Assay (500-0112, Bio-Rad, Belgium), following the manufacturer's instructions. Absorbance at 750 nm of a mixture of 5 µl test solution, 25 µl reagent A (20 µl detergent per millilitre of an alkaline copper tartrate solution), and 200 µl diluted Folin reagent was measured after 15 min incubation at room temperature. Bovine serum albumin was used as reference protein.
To determine LUC activity, 10 µl extract was mixed with 50 µl Luciferase Assay Reagent (LAR, E1501, Promega, The Netherlands) at room temperature in a black microtitre plate (655076, Greiner, Germany). After 1 min, LUC activity was measured as described above with an integration time of 10 s. The average intensity of two replicates was calculated. Using recombinant luciferase (E170, Promega, The Netherlands), a standard curve (0-100 ng LUC/ml) was generated, which permitted the calculation of the LUC concentration (nanograms of LUC per millilitre) in the extract after correction for background detected in extracts of control, non-transformed cells. Following a second correction for protein concentration (micrograms of protein per millilitre of extract), LUC activity was expressed in nanograms of LUC per microgram of extracted protein.
2.5. Southern blot hybridisation
Total DNA isolation and digestion, gel electrophoresis, blotting and hybridisation were done as described previously . Immunochemiluminescent detection of bound probe using CSPD® (1655884, Roche, Belgium) was done according to the manufacturer's instructions. Chemiluminescent light emission was captured with the liquid nitrogen-cooled CCD camera as described above.
3. RESULTS AND DISCUSSION
To assess the sensitivity and usefulness of the LUC reporter gene system in banana, we first investigated in vivo TLE in bombarded embryogenic suspension cells. LUC activity could be detected 80 min after bombardment, and was clearly visible 120 min after bombardment with the codon-modified luc+ gene  (data not shown). In wheat scutellar tissue, LUC activity was detectable somewhat sooner after bombardment, i.e. 35-45 min after bombardment . Figure 2A shows TLE between 3 and 48 h after bombardment with the wild-type luc gene and luc+ gene. For the luc+ gene, a greater than tenfold increase in LUC activity was first detected 14 h after bombardment (from about 20 to about 240 relative grey levels/pixel at 3 h and 14 h, respectively) resulting in maximum LUC activity 24 h after bombardment (approximately 300 relative grey levels/pixel) followed by a clear decrease (about 100 relative grey levels per pixel at 48 h). An increase in LUC activity, though less pronounced, was observed with the luc gene up to 36 h after bombardment, with about 17 relative grey levels/pixel. Since the peroxisomal translocation sequence is absent in luc+ , this difference might be due to a difference in stability of the LUC enzyme in the cytoplasm (luc+) and peroxisomes (luc). These time courses of LUC activity were already evident from the captured images (Figure 2B). Whereas 3 h after bombardment the luc+/luc ratio was 4 (17.1/4.2), it increased to 24 (300.1/12.5) at 24 h after bombardment. These results clearly show that the luc+ gene is more efficiently expressed than the wild-type luc gene in banana cells. Bombarding wheat scutella followed by in vivo LUC detection revealed a luc+/luc ratio of 7 , which is also a considerable enhancement of LUC activity in this monocot. Further monitoring revealed a continuous decrease in LUC activity until only a few light-emitting cell groups could be detected 20 days after bombardment (data not shown). As a wide range of LUC activities could be detected (from less than 5 to more than 300 relative grey levels/pixel), it is clear that this simple, fast and sensitive in vivo reporter gene assay could become a valuable tool in gene expression studies of banana.
A similar TLE study was undertaken by measuring LUC activity in cell extracts using the CCD camera (Figure 3). Interestingly, for both luc+ and luc, maximum in vitro LUC activity was obtained 24 h after bombardment. The discrepancy in the time of maximum TLE observed with the luc gene between the in vivo (36 h) and in vitro (24 h) assay requires further investigation, but is probably not due to decreased luciferin availability in the in vivo assay, as it is readily taken up by plant cells [5, 24]. At the time of maximum TLE (24 h) the in vitro luc+/luc ratio was 18, which is comparable to that obtained with the in vivo LUC detection, i.e. 24, as mentioned above. Despite less precise quantification, this demonstrates the reliability of the in vivo LUC detection assay for assessing TLE in banana.
The in vitro results also confirm the enhancing effect of the luc+ gene over the wild-type luc gene in this non-cereal monocot. Using the same two plasmid constructs (pAL25 and pAL41, respectively) Lonsdale et al.  obtained similar results in maize suspension cells and wheat scutella, where they measured in vitro luc+/luc ratios of 23 and 60, respectively, 48 h after bombardment.
Figure 2 In vivo transient LUC activity in 'Grande Naine' embryogenic suspension cells 3 to 48 h after particle bombardment with luc+ (pAL25) and luc (pAL41). 50 µl of 0.4 mM luciferin in half-strength MS medium was applied to each sample of 25 mg fresh weight cells. Five minutes later the image was acquired in a light-tight box using a liquid nitrogen-cooled CCD camera with an integration time of 1 min. (A) For each time point three samples were measured, and each value shown is the average (± SD) of three independent samples after correction for background detected in control, non-bombarded cells.
Figure 2 In vivo transient LUC activity in 'Grande Naine' embryogenic suspension cells 3 to 48 h after particle bombardment with luc+ (pAL25) and luc (pAL41). 50 µl of 0.4 mM luciferin in half-strength MS medium was applied to each sample of 25 mg fresh weight cells. Five minutes later the image was acquired in a light-tight box using a liquid nitrogen-cooled CCD camera with an integration time of 1 min. (B) Images of one sample at different time points for luc+ (upper row) and luc (lower row) are shown in pseudocolours. The time after bombardment of each image is shown in hours
Although a destructive assay, LUC activity can also be measured and quantified in tissue extracts using a luminometer with great sensitivity; for example, as few 2000 molecules of LUC have been detected . These sensitive in vitro measurements make it a valuable marker not only for studies of transient gene expression [26, 27] but also for more basic studies on the effects of mRNA structure on gene expression [28, 29]. To detect in vitro LUC activity in banana extracts, the low-light sensitive CCD camera was also used successfully, making it an extremely versatile tool for LUC detection.
In conclusion, the ultrasensitive CCD camera system allows both in vivo and in vitro detection of bioluminescence in banana with high sensitivity. As it is less time-consuming and easier to perform, we feel that in vivo detection is the method of choice for TLE studies.
Figure 3 In vitro transient LUC activity in 'Grande Naine' suspension cells 2-48 h after particle bombardment with luc+ (pAL25) and luc (pAL41). Each value is the average (± SD) of four independent replicates. One minute after adding 50 µl of the LAR reagent to a 10 µl sample extract, LUC activity was measured for 10 s
To the best of our knowledge, all promoters driving transgene expression in banana are heterologous promoters. In addition, a few tissue-specific, wound- and pathogen-responsive promoters have been isolated and characterized in monocots. Such promoters are required to express antifungal genes to control several diseases that threaten banana production. Direct isolation of promoters can be done via T-DNA tagging with a promoterless reporter gene. Although the most commonly used reporter gene for this kind of tagging has been the gusA gene [30-34], the ideal reporter gene should have a sensitive, non-destructive and non-toxic assay allowing multiple in vivo screening rounds to identify simultaneously developmental-specific, tissue-specific or stress-responsive patterns of expression. In this way several promoter types of diverse genes can be identified and characterized without a priori isolation and characterization of the corresponding gene. A highly sensitive detection system is important to visualize the suboptimal reporter gene expression level in most transformants due to the random integration of the promoterless reporter gene. The luc gene fulfils this condition and has already been successfully used in tobacco to isolate a functional meristem-specific promoter .
Since the LUC reporter gene system proved to be functional in banana, and in vivo detection could be done with great sensitivity using a liquid nitrogen-cooled CCD camera, as shown above, the identification of banana promoters via T-DNA tagging using a T-DNA border linked promoterless luc gene has been initiated. Recent advances in Agrobacterium-mediated transformation of banana in our laboratory resulted in a high expression frequency , allowing us to generate more than 700 independent putative promoter-tagged cell cultures from 18 cocultured plates, or more than 42 independent cell cultures per 50 mg fresh weight of cocultured cells three months after Agrobacterium-mediated transfer of pluc19. Part of each cell culture was transferred to regeneration medium for later screening for LUC activity at the young plantlet stage, and the remainder was first screened for baseline LUC activity, i.e. detection of LUC activity under standard conditions without prior application of any stress (Table 1). Sixty-four (8.3%) out of 774 geneticin-resistant cell cultures showed LUC activity when assayed with the liquid nitrogen-cooled CCD camera. However, for all except four (Figure 4A) of these 61 putative promoter-tagged cell cultures, adjustment of the grey level scale was necessary (Figure 4B) to visualize LUC activity, indicating that the activity was very low. The latter might occur when the promoterless luc gene integrates a long distance from a promoter. The resulting frequency of cell cultures with strong baseline LUC activity (0.52% or 4/774) is much lower than that obtained in tobacco shoots using the same tagging vector (3.3% ) indicating that the promoter trap vector might have to be improved for efficient use in banana. As the luc+ gene significantly increases expression levels in banana (Figures 2 and 3), it might be worth using this gene for tagging. Alternatively, the start codon of the luc(+) gene could be positioned closer to the left border, as it is 662 bp from the left border in pluc19 . As a positive control, suspension cells were transformed with the vector p35Sluc19 containing the luc gene under the control of the constitutive CaMV 35S promoter (Figures 4C and 4D). All the selected cell cultures following transformation with this vector showed LUC activity, which was visible without scaling in 99% of them and varied between 1 and 142 relative grey levels (data not shown). For comparison, the clearly visible putative promoter-tagged cell culture (Figure 4A) displayed a LUC activity of three relative grey levels.
Table 1 Screening of independent putative promoter-tagged and geneticin-resistant banana cell cultures with baseline LUC activity three months after Agrobacterium-mediated transformation of 'Three Hand Planty' embryogenic suspension cells with pluc19 and p35Sluc19, respectively
Number of independent cell cultures screened
Cell cultures with baseline LUC activity a, b
a in vivo detection of LUC activity using the liquid nitrogen cooled CCD camera and 20 min integration time after saturation of cell cultures with 0.1 mM luciferin in half-strength MS medium.
b detection was done without prior application of any stress.
d p35S-luc-tnos, positive control.
Figure 4 In vivo detection of LUC activity in geneticin-resistant cell cultures 3 months after Agrobacterium-mediated transformation of 'Three Hand Planty' embryogenic suspension cells with the vector pluc19 containing a promoterless luc gene (A and B) or the vector p35Sluc19 containing the luc gene under control of the CaMV 35S promoter (C and D). Following application of 20-60 µl of 0.1 mM luciferin in half-strength MS medium to each cell culture, LUC activity was detected in a light-tight box using a liquid nitrogen-cooled CCD camera with an integration time of 20 min. The image B required adjustment of the grey level scale using the MetaMorph® image analysis software to visualize putative LUC expression. Images are shown in pseudocolours.
The putative promoter-tagged cell cultures are currently being subjected to different abiotic stresses (heat, salt, wounding, etc.) to screen further for useful banana promoter sequences. The regenerated plants will also be screened for tissue-specific and pathogen-responsive expression patterns. The preliminary results described here clearly demonstrate that the ultrasensitive luminescent detection system will allow highly sensitive in vivo characterization of promoter-tagged lines in banana.
In non-radioactive hybridisation analysis a chemiluminescent signal is captured on X-ray film, which is then developed for visualization. This has been the method of choice in our laboratory for several years , and although it is a sensitive technique, it is time-consuming and costly, especially when multiple exposures are needed to evaluate a wide range of signal intensities on the same membrane. It has been shown that digital imaging using a cooled CCD camera offers an alternative, non-film-based method for image acquisition with comparable sensitivity of detection, a greater dynamic range (five orders of magnitude, 16 bits or 65536 grey levels), enhanced flexibility and faster results than film . Therefore, some of these features were tested in Southern blot hybridisation analysis of transgenic banana using the liquid nitrogen-cooled Versarrayä 512 B CCD camera connected to the MetaMorph® image analysis software.
Figure 5A demonstrates that, within less than one hour after application of the substrate CSPD®, signals were detected using an integration time of 20 min. To obtain a similar image on X-ray film, a significantly longer incubation and/or exposure time is required (data not shown). Although some hybridising bands are clearly visible, most bands were still faint. With film-based detection, the membrane would have to be exposed to a new film for a longer time to obtain a clear result with these faint signal intensities. In contrast, due to its wide dynamic range, the 16-bit CCD camera is able to process a wide range of signal intensities after a single exposure. By adjusting the grey scale, the resulting image can simulate a longer exposure time with faint bands becoming intense (Figure 5B). Naturally, it is also possible to lower the intensity of saturated signals (data not shown). Images captured after a prolonged (several hours) incubation of the membrane in the presence of the substrate also yielded clearer bands (data not shown). The benefits attributed to the use of a CCD camera system for recording non-radioactive hybridisation results clearly also apply to our imaging system, which is now widely used in our laboratory for this type of application.
Figure 5 Southern blot images captured with the liquid nitrogen cooled Versarrayä 512 B CCD camera connected to the MetaMorph® image analysis software in a light-tight box. Total DNA of a control, non-transgenic plant (c) and four different generations (4, 5, 6 and 7) of a 'Williams' transformant was digested with Dra I, electrophoresed and hybridised with a digoxigenin-labelled Ace-AMP1 probe. Mr = digoxigenin-labelled DNA molecular marker. 1c, 5c and 10c are 1, 5 and 10 copy number reconstructions of Dra I digested plasmid pIGNPAce. e = empty lane. (A) a Southern blot image was captured using a 20 min exposure, 30 min after application of the substrate CSPD®, and the contrast inverted for display using the software without adjustment of the grey scale.
Figure 5 Southern blot images captured with the liquid nitrogen cooled Versarrayä 512 B CCD camera connected to the MetaMorph® image analysis software in a light-tight box. Total DNA of a control, non-transgenic plant (c) and four different generations (4, 5, 6 and 7) of a 'Williams' transformant was digested with Dra I, electrophoresed and hybridised with a digoxigenin-labelled Ace-AMP1 probe. Mr = digoxigenin-labelled DNA molecular marker. 1c, 5c and 10c are 1, 5 and 10 copy number reconstructions of Dra I digested plasmid pIGNPAce. e = empty lane. (B) The same image as in A, but the grey scale has been adjusted to simulate a longer exposure with the CCD camera.
In addition to bioluminescence and chemiluminescence, a CCD camera can also detect fluorescent signals. However, the liquid nitrogen-cooled Versarrayä 512 B camera is designed for capturing ultra low-light emission, and the chip consists of large pixels (24 × 24 µm), resulting in a relatively low spatial resolution for a given magnification. Large pixels allow long exposure times without becoming saturated with charge, as has been shown above. For fluorescent signals, which are usually more intense than luminescent signals, long exposures are not required and even not recommended when, for instance, excitation is done with UV light. Smaller pixels would suffice and result in a higher resolution. Despite the suboptimal resolution, the liquid nitrogen-cooled CCD camera could be used to monitor green fluorescent protein (GFP ) expression in transgenic banana cultures (Figure 6). Identical software analysis of the acquired images to that used for luminescent images allows quantification of GFP fluorescence, opening up the possibility of performing semi-quantitative GFP expression studies in plants in a fast and simple way.
Figure 6 In vivo detection of GFP expression in a transgenic banana cell culture with the liquid nitrogen-cooled Versarrayä 512 B CCD camera connected to a Leicaä MZ FL III stereomicroscope equipped with a GFP3 plant fluorescence filter. The image was acquired in a 1 min exposure, and was taken four months after Agrobacterium-mediated transformation of 'Three Hand Planty' embryogenic suspension cells with the vector pBINUbi1-sgfpS65T containing the codon-modified sfgpS65T gene under control of the maize Ubi1 promoter. Scale bar = 1 mm.
The superior performance of a cooled CCD camera in terms of sensitivity, quantum efficiency, read-out noise, dark charge, dynamic range and others make it a valuable tool to acquire scientific images. This is clearly supported by the wide range of applications for which the ultrasensitive digital imaging system presented here has been used successfully in our banana biotechnology research. Since sensitivity is of paramount importance for our main activity involving digital image acquisition, i.e. promoter tagging using the bioluminescent luc gene, a back-illuminated, slow-scan, liquid nitrogen-cooled CCD camera has been employed. This type of camera can also detect chemiluminescent and fluorescent signals, as we have demonstrated, although a less sensitive but cheaper CCD camera would suffice for applications involving these kinds of signal.
The authors wish to thank Dr. D.M. Lonsdale, John Innes Centre, Norwich, UK, for providing plasmids pAL41 and pAL25, and Dr. R.G. Birch, University of Queensland, Brisbane, Australia for plasmids pluc19 and p35Sluc19. Plasmid pIGNPAce was a gift from Dr. C. Bird, Syngenta, Bracknell, UK, while Dr. A. Elliot, CSIRO Tropical Agriculture, Brisbane, Australia, kindly provided us with pBINUbi-sgfpS65T. We also thank W. Dillemans for taking the pictures of the imaging system. This work has been supported by the International Network for Improvement of Bananas and Plantains (INIBAP) through a grant from DGIS. We acknowledge the INIBAP Transit Centre for supplying banana germplasm.
 SPRING, K.R., Scientific imaging with digital cameras, BioTechniques 29 (2000) 70-76.
 JEFFERSON, R.A., et al., b-glucuronidase from Escherichia coli as a gene fusion marker, Proc. Natl. Acad. Sci. USA 83 (1986) 8447-8451.
 DE WET, J.R., et al., Cloning of firefly luciferase cDNA and the expression of active luciferase in Escherichia coli, Proc. Natl. Acad. Sci. USA 82 (1985) 7870-7873.
 OW, D.W., et al., Transient and stable expression of the firefly luciferase gene in plant cells and transgenic plants, Science 234 (1986) 856-859.
 MILLAR, A.J., et al., Firefly luciferase as a reporter of regulated gene expression in plants, Plant Mol. Biol. Rep. 10 (1992) 324-337.
 CHIA, T.-F., et al., The firefly luciferase gene as a non-invasive reporter for Dendrobium transformation, Plant J. 6 (1994) 441-446.
 AFLALO, C., Biologically localised firefly luciferase: a tool to study cellular processes, Int. Rev. Cytol. 130 (1991) 269-323.
 DENBURG, J.L., et al., Substrate-binding properties of the firefly luciferase, Arch. Biochem. Biophys. 134 (1969) 381-394.
 NYGEN, V.T., et al., Protein denaturation during heat shock and related stress, J. Biol. Chem. 264 (1989) 10487-10492.
 THOMPSON, J.F., et al., Modulation of firefly luciferase stability and impact on studies of gene regulation, Gene 103 (1991) 171-177.
 MURASHIGE, T., SKOOG, F., A revised medium for rapid growth and bioassays with tobacco tissue cultures, Physiol. Plant. 15 (1962) 473-497.
 HED'A, D., et al., Plant regeneration in cell suspension cultures of the cooking banana cv. 'Bluggoe' (Musa spp., ABB group), Fruits 46 (1991) 125-135.
 LONSDALE, D.M., et al., The effect of altered codon usage on luciferase activity in tobacco, maize and wheat, Plant Cell Rep. 17 (1998) 396-399.
 CHRISTENSEN, A.H., et al., Maize polyubiquitin genes: structure, thermal perturbation of expression and transcript splicing, and promoter activity following transfer to protoplasts by electroporation, Plant Mol. Biol. 18 (1992) 675-689.
 SHERF, B.A., WOOD, K.V., Firefly luciferase engineered for improved genetic reporting, Promega Notes 49 (1994) 14-21.
 MUDGE, S.R., et al., Comparison of Vibrio and firefly luciferases as reporter gene systems for use in bacteria and plants, Aust. J. Plant Physiol. 23 (1996) 75-83.
 ELLIOT, A.R., et al., Green fluorescent protein facilitates rapid in vivo detection of genetically transformed plant cells, Plant Cell Rep. 18 (1999) 707-714.
 CHIU, W.-I., et al., Engineered GFP as a vital reporter for plants, Curr. Opin. Biol. 6 (1996) 325-330.
 CAMMUE, B.P.A., et al., A potent antimicrobial protein from onion (Allium cepa) seeds showing sequence homology to plant lipid transfer proteins, Plant Physiol. 109 (1995) 445-455.
 REMY, S., Genetic transformation of banana (Musa spp.) for disease resistance by expression of antimicrobial proteins, PhD thesis, Catholic University Leuven, Belgium, (2000).
 PéREZ HERNáNDEZ, J.B., Development and application of Agrobacterium-mediated genetic transformation to increase fungus-resistance in banana (Musa spp.), PhD thesis, Catholic University Leuven, Belgium, (2000).
 LOWRY, O.H., et al., Protein measurement with the Folin phenol reagent, J. Biol. Chem. 193 (1951) 265-275.
 LONSDALE, D.M., et al., Using firefly luciferase to identify the transition from transient to stable expression in bombarded wheat scutellar tissue, Physiol. Plant. 102 (1998) 447-453.
 VAN LEEUWEN, W., et al., The use of the luciferase reporter system for in planta gene expression studies, Plant Mol. Biol. Rep. 18 (2000) 143a-143t.
 WOOD, K.V., "Recent advances and prospects for use of beetle luciferases as genetic reporters", Bioluminescence and Chemiluminescence: Current Status (STANLEY, P., KRICKA, L., Eds), John Wiley & Sons, Chichester, UK (1991) 543-546.
 LUEHRSEN, K.R., et al., Transient expression analysis in plants using firefly luciferase reporter gene, Methods Enzymol. 216 (1992) 397-414.
 BARUAH-WOLFF, J., et al., Luciferase as a reporter gene for transformation studies in rice (Oryza sativa L.), Plant Cell Rep. 18 (1999) 715-720.
 CALLIS, J., et al., Introns increase gene expression in cultured maize cells, Genes Dev. 1 (1987) 1183-1200.
 GALLIE, D.R., The cap and poly(A) tail function synergistically to regulate mRNA translational efficiency, Genes Dev. 5 (1991) 2108-2116.
 FOBERT, P.R., et al., Detection of gene regulatory signals in plants revealed by T-DNA mediated fusions, Plant Mol. Biol. 17 (1991) 837-851.
 TOPPING, J.F., et al., Identification of molecular markers of embryogenesis in Arabidopsis thaliana by promoter trapping, Plant J. 5 (1994) 895-903.
 TOPPING, J.F., LINDSEY, K., Promoter trap markers differentiate structural and positional components of polar development in Arabidopsis, Plant Cell 9 (1997) 1713-1725.
 CAMPISI, L., et al., Generation of enhancer trap lines in Arabidopsis and characterisation of expression patterns in the inflorescence, Plant J. 17 (1999) 699-707.
 FOSTER, E., et al., A tobacco cryptic constitutive promoter, tCUP, revealed by T-DNA tagging, Plant Mol. Biol. 41 (1999) 45-55.
 MUDGE, S.R., BIRCH, R.G., T-DNA tagging and characterisation of a novel meristem-specific promoter from tobacco, Aust. J. Plant Physiol. 25 (1998) 637-643.
 SáGI, L., et al., Genetic transformation of banana and plantain (Musa spp.) via particle bombardment, BioTechnology 13 (1995) 481-485.
 BUDOWLE, B., et al., Using a CCD camera imaging system as a recording device to quantify human DNA by slot blot hybridisation, Biotechniques 30 (2001) 680-685.
 SHEEN, J., et al., Green-fluorescent protein as a new vital marker in plant cells, Plant J. 8 (1995) 777-784.