Journal of Experimental Botany, Vol. 53, No. 379, pp. 2315-2323,
December 1, 2002
© 2002 Oxford University Press
Using array hybridization to monitor gene expression at the single cell level
Received 4 July 2002; Accepted 10 July 2002
Max Planck Institut für Molekulare Pflanzenphysiologie, Department Willmitzer, Am Mühlenberg 1, D-14476 Golm, Germany
1 Present address: School of Biological Sciences, University of Wales Bangor, Bangor, Gwynedd LL57 2DG, UK.
2 To whom correspondence should be addressed. Fax: +49 331 567898213. e-mail: kehr{at}mpimp-golm.mpg.de
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and methodsResultsDiscussionReferences |
|---|
Advances in high-throughput genome sequencing demand the development of more efficient ways of examining gene expression at a cellular level. During recent years, polymerase chain reaction (PCR)-based methods have been developed that allow the amplification of mRNA from small amounts of material, even from single animal cells. In parallel, several analytical tools permit a global monitoring of gene expression. To date, high throughput analysis methods have not been accessible for single plant cell samples. In the protocol described here, cDNA array hybridization (expression profiling) and an amplification strategy using reverse transcriptase PCR are merged with high spatial resolution sampling from undamaged plant tissue. This protocol gives us a new tool to examine tissue-specific gene expression patterns on a comprehensive scale. To demonstrate the usefulness of this tool, gene expression patterns in samples from Arabidopsis thaliana L. cv. C24 leaf epidermal and mesophyll cells were measured; several differentially expressed genes were identified when single cell samples were compared. The protocol described has the potential of increasing the efficiency of tissue-specific expression analysis by combining high-throughput profiling with straightforward sampling and amplification procedures.
Key words: Array hybridization, differential display, gene expression profiling, single cell analysis.
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
Following recent advances in the sequencing of complete genomes, one major challenge will be discovering the function of gene products at all morphological levels. The analysis of samples from entire plants or even from a whole organ results in data that represent an average of multiple cell types. Because physiological processes are often restricted to specific tissues, a high resolution analysis would lead to a better understanding of biological processes, such as adaptation to environmental conditions (Ishitani et al., 1996), intercellular communication (Dzelzkalns et al., 1992; Mezitt and Lucas, 1996), wound response (O’Donnell et al., 1996; Titarenko et al., 1997), or signal transduction (Lamb, 1994). Although knowledge about the expression pattern of a given mRNA does not directly indicate the function of the gene it encodes, information about the temporal and spatial expression of a specific gene can provide valuable information about its potential role. A view of the transcriptional activity within a single cell or a specific tissue type would substantially increase current understanding of the physiology of whole organisms (Kehr, 2001).
While in situ hybridization (Shu et al., 1999) and in situ RT-PCR (Koltai and McKenzie Bird, 2000) in fixed tissue sections can be used to determine the cell-specific expression of individual transcripts, few protocols describe an analysis of the expression of specific genes at the single cell level of living plants (Karrer et al., 1995; Brandt et al., 1999). One major problem is the sampling of cells from a (nearly) undamaged plant, which can be accomplished using glass microcapillaries. Following sampling, the minute amount of mRNA from a single plant cell (<1 pg, Dresselhaus et al., 1994) is not applicable to standard techniques like northern hybridization and it has to be amplified for in vitro analysis. There are multiple amplification procedures for specific single sequences available, through which enrichments of one million-fold from as few as one starting molecule are theoretically possible (Compton, 1991; Saiki et al., 1988). It has been demonstrated that PCR amplification can be applied to individual transcripts from single plant cell samples (Brandt et al., 1999).
The desire to examine not only single genes, but to extend the analysis to complete genome expression patterns, led to the development of techniques like nucleic acid microarrays. Such techniques have been applied in the study of thousands of genes in parallel in mammalian systems (DeRisi et al., 1996; Schena et al., 1996), yeast (Posas et al., 2000), and plants (Schena et al., 1995; Ruan et al., 1998). Nevertheless, one serious drawback of these hybridization-based methods is the relatively high amount of starting material required for probe generation (Kehoe et al., 1999), which is typically in the high nanogram to low microgram range. Consequently, samples from individual and even several pooled cells cannot be directly subjected to array hybridization. In animals, this hurdle can be circumvented by collecting thousands of cells of one specific type by laser capture microdissection (Luo et al., 1999). But since this technique has not yet been shown to work with plant samples (Kehr, 2001), gene expression studies have so far been limited to entire plants (Desprez et al., 1998), or at least to whole organs like leaves or roots (Reymond et al., 2000; Schena et al., 1995). Such experiments have enabled the identification of genes that are differentially expressed in the light and the dark (Desprez et al., 1998), during stress (Reymond et al., 2000), and in roots and leaves (Schena et al., 1995).
There have been few protocols described that deal with the amplification of more than one gene at a time. Multiplex RT-PCR is a forthcoming technique that allows the amplification of up to nine genes in one reaction (Ponce et al., 2000). Other protocols deal with sets of cDNAs independently from their sequences. For instance, adaptor ligation attaches identical 5' ends to each cDNA fragment (Karrer et al., 1995) while tailing synthesizes homomers to all 5' ends by terminal deoxynucleotidyl transferase (Belyavsky et al., 1989; Brady et al., 1995) prior to PCR amplification. A third method, linear antisense RNA amplification (Wang et al., 2000), can also be employed. All these protocols have been successfully adapted to the non-specific amplification of diverse cDNAs from small amounts of starting material from animal or human sources. However, only very few experiments have up to now described the non-specific amplification of plant derived nucleic acids (Dresselhaus et al., 1994; Karrer et al., 1995).
Here a method is described for monitoring gene expression patterns in samples taken from a few Arabidopsis thaliana L. cv. C24 leaf cells. By contrast with animal or human-derived samples (Brady et al., 1995), research on gene expression patterns of single cells from higher plants is relatively difficult, mainly because of the small sample size, rigid cell walls and the fact that most of the cell volume is taken up by large vacuoles. Cell contents were sampled using glass microcapillaries and the mRNA was amplified by arbitrary primers derived from differential display and hybridized to array filters containing 16 000 cDNAs from expressed sequence tags (ESTs). Multiple differences were observed when the expression profiles of leaf epidermal and mesophyll cells were compared. As far as is known, this is the first successful expression profiling attempt starting from a few plant cells.
|
Materials and methods |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
Plant material
Arabidopsis thaliana L. cv. C24 plants were grown in a greenhouse at an average photon flux density 120 µmol m–2 s–1, 22 °C temperature, 65% relative humidity, and with 16/8 h light/dark cycles. 8–10-week-old plants were used for microsampling as described below.
Microsampling
Sampling was performed as described by Brandt et al. (1999). In short, borosilicate glass capillaries (WPI, Berlin, Germany) were pulled on a List pipette puller (Darmstadt, Germany) with a tip aperture of 1–10 µm. The capillaries were mounted on a micromanipulator and cauline leaves of an intact plant were fixed under an Optiphot 2 microscope (Nikon, Duesseldorf, Germany). For sampling and pooling epidermal cells, a capillary was inserted into a single cell, withdrawn, and inserted into the next surface cell. Mesophyll cells were sampled by pushing the capillary through the epidermis. Contamination from the epidermis during impalement was prevented by applying positive pressure and by using silanized capillaries. Immediately after withdrawal, the content of the capillary was released into a sterile, 0.5 ml reaction vial containing 1 µl DEPC water with 5 U RNase inhibitor (RNAsin, Promega, Heidelberg, Germany). Each capillary was used for only one sampling process.
PCR
As described by Brandt et al. (1999), precautions to prevent contamination or degradation were taken. In brief, sterile disposable microcapillaries, filter tips, and tubes were used. All RT-PCR reactions were prepared under a sterile UV hood without air circulation and all post-PCR work was done in a separate laboratory.
Reverse transcription: Reverse transcriptions were performed in the vials containing the extracted cell samples. The final concentrations in 10 µl volume were: 1xreverse transcription buffer (Life Technologies, Eggenstein, Germany), 10 mM DTT (Life Technologies), 50 U superscript II reverse transcriptase (Life Technologies), 0.5 mM of each dNTP (Pharmacia, Freiburg, Germany), and 0.1 µM of a poly (dT)15 primer (Roche, Mannheim, Germany). This reaction mix was incubated for 75 min at 48 °C, followed by 5 min at 95 °C for the inactivation of reverse transcriptase in a Primus thermocycler (MWG Biotech, Ebersberg, Germany).
Gene specific PCR: PCR was performed in a volume of 50 µl. Final concentrations of the components were: 1xreaction buffer (Clontech, Heidelberg, Germany), 0.5 µM of each primer (forward 5'- AAA AAT GGC TGA GGC TGA TG, reverse 5'- TTC TCG ATG GAA GAG CTG GT), 200 µM of each dNTP (Pharmacia), and 0.5 µl Advantage cDNA Polymerase Mix (Clontech). 40 µl of a master mix were added to the entire 10 µl of the reverse transcription mix. The PCR cycle programme was as follows: 5 min at 94 °C, 40 cycles of 90 s at 94 °C, 90 s annealing at 62 °C, 90 s at 70 °C with 5 s time increment per cycle, followed by 5 min at 70°C for the final extension. 20% of each PCR reaction was separated on a 1% agarose gel.
Arbitrarily primed PCR: The PCR conditions were modified as follows: Final concentrations: 1xreaction buffer (Biometra, Goettingen, Germany), 0.5 µM of each primer (forward 5'- GGA ACC AAT C, reverse 5'- TTT TTT TTT TTT VN), 100 µM of each dNTP (Pharmacia), 100 µM of digoxygenin labelled dNTPs (DIG Labeling Mix, Roche), and 2 U Dynazyme DNA polymerase (Biometra), with a cycling programme of 5 min at 94 °C, 40 cycles of 30 s at 94 °C, 120 s annealing at 40 °C, 30 s at 7 4°C with 5 s time increment per cycle, followed by 10 min at 74 °C for final extension.
Control reactions: To confirm the reliability of the results, the following control reactions were performed. The templates for all types of controls were also handled by microcapillaries and expelled into sterile reaction vials as described for microsampling (see above).
Positive controls: RNA was used to ensure the performance of all components of the RT-PCR reaction. RNA was isolated from leaves of 8–10-week-old plants by the Trizol reagent method (Life Technologies) following the manufacturers instructions.
Negative controls: To ensure that signals on the gels and Southern blots originated from RNA and not from genomic DNA, water was added to some samples instead of reverse transcriptase. Other cell extracts were treated for 30 min at 37 °C with RNase A (Roche) prior to reverse transcription. To exclude false positive results caused by contamination, at least one mock experiment was performed in parallel to every experiment using water instead of cell extracts. If one of the negative controls showed any PCR products, the entire experiment was discarded.
Detection of PCR products: After size fractionation of a 20% aliquot of the products on a 1% agarose gel, Southern blots were performed. The membranes (Porablot NyAmp, Macherey and Nagel, Dueren, Germany) were washed for 2 min in washing buffer (0.1 M maleic acid, 0.15 M NaCl, 0.3% Tween 20), 30 min in blocking reagent (Roche), 30 min in blocking reagent plus an anti-DIG antibody coupled with alkaline phosphatase (diluted 1:20 000, Roche), washed twice in washing buffer for 15 min and 5 min in detection buffer (0.1 M Tris–HCl, pH 9.5; 0.1 M NaCl). Finally, the membranes were incubated in detection buffer plus CDP Star (diluted 1:100, Roche) as a substrate for 5 min. Chemiluminescence was visualized with x-ray film (X-omat AR, Kodak, Stuttgart, Germany) after 5–60 min exposures.
Probe generation
The PCR products were purified by spin columns (PCR Purification Kit, Qiagen, Hilden, Germany) and recovered in 50 µl EB buffer (see the manufacturer’s instructions). 20 µl of the products were denatured at 95 °C for 10 min and labelled with 60 µCi 33P dCTP (NEN, Zaventem Belgium) using the Random Primed Labelling Kit (Roche) following the instructions of the manufacturer.
After purification by Nick columns, 50 (Pharmacia) aliquots of the probes were counted for incorporation into the products. To enable grid orientation for evaluating the data after hybridization, human desmin (Accession No. NM_001927) with a specific activity of 140 000 cpm was added to the samples (Fig. 3).
|
Array filter generation
Arabidopsis cDNA clones: Arabidopsis cDNA plasmid clones were obtained from the Arabidopsis Biological Resource Center (ABRC, Ohio, USA) where most of the clones were generated as part of the Michigan State EST effort (Newman et al., 1994).
PCR amplification of cDNA inserts: Inserts of the cDNA clones were amplified by PCR using primers that were complementary to vector sequences flanking both sides of the cDNA insert (forward 5'-AAA GGG GGA TGT GCT GCA AGG CG-3', reverse 5'-GCT TCC GGC TCG TAT GTT GTG TG). PCR was performed from E. coli colonies using standard parameters.
Array preparation: Arabidopsis arrays were prepared on 22.2x 22.2 cm nylon membranes (Hybond N+, Amersham, Freiburg, Germany), which were soaked in 0.5 M NaOH before spotting. The spotting was performed by a BioGrid robot from BioRobotics (Cambridge, UK) with a 384-pin spotting tool. Each PCR-product was transferred 10x in duplicates in a 4x4 spotting pattern (Fig. 3). Membranes were neutralized (1 M Tris–HCl pH 7.2, 1.5 M NaCl) and the spotted DNA was fixed by baking the membranes for 2 h at 80 °C followed by irradiation in a Stratalinker model 2400 (Stratagene, Amsterdam, Netherlands) at 254 nm.
Array hybridization and evaluation
Reference hybridization: A reference hybridization was performed on a shaker in 200 ml 1xSSarc (24% Sarcosyl NL30, 0.6 M NaCl, 0.06 M sodium citrate, 4 mM EDTA) at 5 °C overnight using a 33P end-labelled oligomer (5'-TTC CCA GTC ACG) unique to all PCR products. Membranes were washed for 30 min at 5 °C in 0.5 l SSarc and exposed. Membrane regeneration was done by stripping twice in 0.1 l 0.1xSSarc at 65 °C for 30 min.
Complex hybridization: Filters were permutated for different hybridizations. Prehybridization was done for at least 2 h at 65 °C in a buffer consisting of 250 mM sodium phosphate at pH 7.2, 10 mM EDTA, 7% SDS, and 1% BSA. The hybridization was performed in the same buffer for 30 h at 65 °C. The filters were washed for 20 min in 2x SSC, 0.5% SDS, 4 mM sodium phosphate (pH 7.2) at 65 °C and for 20 min in 0.2x SSC, 0.5% SDS, 4 mM sodium phosphate (pH 7.2) at 65 °C. Fuji BAS intensifying screens (Raytest, Straubenhardt, Germany) were exposed overnight and scanned at 50 µm resolution (16 bits per pixel) with a Fuji BAS 1800 II phosphor imager (Raytest). The image was analysed and converted into a table of signal intensities using the Array Vision software (Imaging Research, St Catherine’s, Canada).
Data processing: The obtained ‘raw data set’ was submitted to an in-house database. Within this database, the raw data were annotated, rearranged, and normalized. In the first normalization step, the local background was subtracted from each value, which was than divided by the average signal obtained for all spots on the filter (except for desmin). A second normalization step involved the values of each individual spot obtained by the reference hybridization based on the assumption that, under the chosen conditions, the signal from the complex hybridization is linearly proportional to the amount of bound PCR product (Thimm et al., 2001). No normalization was done to counterbalance possible variations during the RT-PCR reaction, but 2–3 times more epidermal than mesophyll cells were sampled to equalize the different amounts of cytoplasm in both types of samples (Fig. 2a). The clone functions were determined by homology searches using the TAIR database (http://www. arabidopsis.org/).
|
|
Results |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
General considerations
The intention of this study was to use a simple and fast mRNA amplification strategy to make the content of single plant cells amenable for comprehensive gene expression analyses, as it is realized in the differential display technique (Liang and Pardee, 1992; Bauer et al., 1993). In this PCR-based strategy arbitrarily chosen, 10 nucleotide long primers are used to amplify cDNA randomly. Up to 100 bands on a sequencing gel (Liang and Pardee, 1992) result, with each band potentially representing several commensurate products (Baldwin et al., 1999).
Sampling and mRNA amplification
The contents of single cells from the epidermis and mesophyll of undamaged Arabidopsis plants were sampled by inserting glass microcapillaries into the cells of interest. Cell selection and sampling was carried out using a microscope and micromanipulation tools (Fig. 1; Brandt et al., 1999). Since it is known that epidermal cells contain a larger vacuole than mesophyll cells (Leidreiter et al., 1995), it was tested whether or not microcapillary-derived samples from both cell types contain similar amounts of cytoplasm and, consequently, equal levels of mRNA. With this question in mind, cell sap from five epidermal or five mesophyll cells was subjected to RT-PCR with specific primers raised against the housekeeping gene actin, since it is likely that this gene is constitutively expressed in mature leaves. Agarose gel electrophoresis revealed that different amounts of PCR products were amplified from both sample types (Fig. 2a). When stained with ethidium bromide, weak signals emerged from the epidermal samples while samples from the mesophyll resulted in intense bands. When the number of sampled epidermal cells was increased to 15, the signal intensity was approximately equal to that of five pooled mesophyll cells (Fig 2a). Depending on the plant and the leaf sampled, a ratio of epidermal to mesophyll cells between 3:1 and 2:1 appeared to be most suitable to yield a comparable product amount after RT-PCR.
|
To generate probes for array hybridization, the contents of about 20 mesophyll and 50 epidermal cells were collected and pooled. For the three independent experiments, samples were collected from three different plants at approximately the same time in the afternoon on three different days. In one experiment, epidermal and mesophyll samples were taken from the same leaf of one A. thaliana plant. After amplification via RT-PCR with one arbitrary and one degenerate Oligo d(T15) primer, only weak signals in the size range from 500 bp to 1.5 kb were visible on a 1% agarose gel (Fig. 2b). Southern blotting and digoxygenin detection unveiled products in the range of 200 bp up to 2 kb (Fig. 2b).
Probe synthesis and array hybridization
To generate probes, PCR products were labelled with radioactive dCT33P using random primed labelling. Labelling rates of up to 90% of total activity could be achieved, whereas negative controls incorporated less than 20% radioactivity. Microarray filters carrying 16 000 ESTs (expressed sequence tags) spotted in duplicate and arranged in a 4x4 pattern were hybridized with either the radioactively labelled epidermis or mesophyll probes, respectively (Fig. 3).
For evaluation of the arrays, the local background of each 4x4 spotting pattern was determined by measuring the signal intensity in position 7, which contained no spotted cDNA (see Fig. 3). Only signals that exhibited a value higher than twice the local background were accepted as positive signals (for details of the normalization procedure see materials and methods). ESTs with intensities smaller than 2-fold the local background were replaced by the value 0.1. When the ratios of epidermis to mesophyll were calculated, all ESTs with a resulting value of exactly 1 were deleted to exclude all signals not significantly above background in both tissues from further analyses. In addition, only ESTs were taken into account that showed a less than 4-fold variation in the epidermis to mesophyll ratios in at least two independent experiments. Values with higher variation were regarded as artefacts caused by PCR or hybridization.
Figure 4 shows the reproducibility of the method. All ESTs that fulfilled the above-mentioned criteria in at least two independent sets of epidermal- and mesophyll-derived probe hybridizations are plotted. The ascending slope of the resulting compensation curve is near to 1. The ‘extreme’ values at each end of the compensation curve represent the preferentially different expressed genes, whereas the equally expressed genes are found in between them.
|
About 680 ESTs fulfilled the above-mentioned criteria and were annotated using the TAIR database (www. arabidopsis.org/ABRC). Signal intensity ratios of epidermis to mesophyll were between 0.2 and 5 for most of these ESTs. 98 ESTs had values lower than 0.2 and 22 had ratios higher than 5. The corresponding genes were regarded as being preferentially expressed in the mesophyll or epidermis, respectively. Table 1 summarizes the functional categories to which these ESTs belong. As expected, the mesophyll-specific subset of ESTs contained many genes that are involved in photosynthesis, because epidermal cells are generally regarded as being non-photosynthetic. In fact, ESTs homologous to a Rubisco binding protein, carbonic anhydrase, and some chloroplastidic proteins seem to be preferentially expressed in mesophyll cells (Fig. 3). Most of the remaining genes are involved in regulation/transcription, primary biosynthesis or stress response (Table 1). For 51 ESTs no homology could be assigned and they remained as unknowns. Some examples of differentially expressed transcripts are shown in Fig. 3. Database analysis of the genes expressed predominantly in the epidermis did not result in the identification of one of the few known epidermis-specific genes (Fig. 3). The epidermis-specific ESTs belong to the functional classes of protein metabolism and regulation/transcription as well as stress response. Four ESTs revealed no homology. In the group of housekeeping genes not supposed to be expressed in a tissue-specific manner, this study shows that ß-tubulin mRNA (Seki et al., 2001) is equally present in both tissues analysed (Fig. 3), although the signal intensity was comparably low.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
To gain a better understanding of how plants interact with their environment and how organs and tissues communicate, it is important to increase the spatial (and temporal) resolution of analyses. Unfortunately, not much is known about cell type-specific gene expression so far. Tools like in situ PCR and in situ hybridization are time-consuming, difficult, and restricted to the study of one specific gene at a time. For a more comprehensive study of cell type-specific gene expression, sampling by laser capture microdissection has been successfully combined with array hybridization in animal tissues (Simone et al., 1998), but there are as yet no examples in plants that have been published. For all these analyses, material has to be fixed and, consequently, time-courses or developmental processes cannot be followed in a single plant. Neither tissue peels (Bret-Harte, 1993) nor protoplasts (Dresselhaus et al., 1994) can serve as good alternatives because the destructive sampling processes involved may alter gene activity (Grosset et al., 1990).
The minimally invasive glass capillary sampling technique can be applied in experiments that require sampling from living plants (Brandt et al., 1999; Fig. 1). By combining this technique with specifically primed RT-PCR, it became possible to study one gene in a given single cell sample (Brandt et al., 1999; Fig. 2a). Ponce et al. (2000) published a multiplex approach for expression monitoring of nine genes in just one reaction and, in principle, this should be applicable also to picolitre samples. Yet even this approach only opens a limited view to global gene expression.
In this study, a method is presented for gene expression profiling from a few pooled individual plant cells collected with glass microcapillaries. Due to the very low mRNA amount even in pooled single cell samples, the RNA could not be directly subjected to hybridization without preceding amplification. For that purpose, one set of primers derived from differential display strategies was used for RT-PCR. In contrast to other amplification protocols, arbitrarily primed PCR is easy to perform (no need of purification, no loss of RNA), fast (completed in about 5–6 h), and cheap (since no additional enzymes are required). The PCR products were hybridized to array filters, which contained about 16 000 ESTs. With the primer pair used, about 680 ESTs showed clear signals. Depending on the number of primers (see below), it should be theoretically possible to monitor nearly the entire transcriptome of pooled single cells.
The reproducibility of the method is demonstrated by the correlation of all ESTs which fulfilled this study’s criteria that results in an ascending slope near to 1 (Fig. 4). The variation of the spots was higher than in bulk experiments that do not require any amplification (for an example using the same filters and procedures see Thimm et al., 2001). However, this is most probably due to the sampling and PCR amplification procedures, both of which can cause imbalances in transcript representation in the radiolabelled probes (see below). Due to the comparably low signal intensities achieved in this set of experiments, minor differences in the transcript representation can lead to a relatively high variation in the signal intensities compared to experiments with plenty of starting material (Desprez et al., 1998; Schenk et al., 2000; Thimm et al., 2001). Moreover, for transcripts with low expression major differences can be caused by hybridization artefacts (Desprez et al., 1998).
To demonstrate the applicability of the method to biological samples, the gene expression of epidermal and mesophyll cells were compared. These cell types are functionally (non-photosynthetic versus photosynthetic) as well as anatomically (surface versus non surface tissue) very different and therefore, a comparatively large amount of differentially expressed genes could be expected. The experiments highlighted 120 differentially expressed genes in both sample types, 98 with increased activity in the mesophyll and 22 preferentially expressed in epidermal cells. These genes can be assigned to different functional classes (Table 1). Among the predominantly mesophyll-expressed genes, photosynthesis-related genes (for example, a Rubisco-binding protein, carbonic anhydrase, and some chloroplastidic proteins) were identified (Fig. 3). These results were expected, since epidermal cells are not believed to be photosynthetically active. Only some of the 98 genes have been previously demonstrated to exhibit mesophyll-specific or chloroplast-specific expression, as for example carbonic anhydrase (Jacobson et al., 1975), Rubisco-binding protein subunit (Martel et al., 1990) or aldolase (Tsutsumi et al., 1994). For the other mesophyll-specific genes as well as for the 22 genes showing epidermis-specific expression, no literature concerning spatial expression is available. From the group of genes that have been presumed to be equally expressed in nearly all plant tissues, the hybridization results revealed that tubulin is constitutively expressed. Actin, another presumed housekeeping gene, and Rubisco that is expected to be expressed differentially between the epidermis and the mesophyll, could not be detected in the hybridizations. This is probably due to the primer combination used that amplifies only a subset of the transcriptome (see below). Two presumably false positives caused by hybridization artefacts could be observed: two out of 16 spotted clones of chlorophyll-binding protein and 1 out of 50 Rubisco small subunit 3b precursor clones exhibited signals exclusively in the epidermis samples, all the other clones spotted do not show any expression in the epidermis or mesophyll and, consequently, seem not to be covered by the primers used for amplification. The observed signals thus might originate from hybridization artefacts or a high local background. This stresses the necessity to confirm the data by a second, independent method, since it cannot be ruled out that false positives are found among the differentially expressed genes, as long as there are no literature data concerning the expression pattern available.
In addition to the inherent limitations of differential display and array hybridization (DeRisi et al., 1997; Trenkle et al., 1998), the following critical aspects should be considered.
Sampling: The sampling method is not quantitative. Although it is possible to normalize the samples by volume, this does not necessarily mean that samples contain equal proportions of cytoplasm and, consequently, mRNA (Fig. 2a). For example, in potato leaves, 76.1% of the mesophyll cell volume is taken up by the vacuole, while in epidermal cells the vacuole accounts for more than 97% of cell volume (Leidreiter et al., 1995). Since the applied sampling procedure favours the extraction of vacuolar content, epidermal cell samples contain less mRNA than mesophyll cell samples. For this reason 2–3 times more epidermal than mesophyll cells had to be used to gain equal signals for actin by specifically primed PCR (Fig. 2a) and tubulin by array hybridization (Fig. 3). This underlines the necessity of identifying constitutively expressed genes to normalize hybridization data. This can be difficult because there is very limited information about the tissue-specific expression of genes available.
RT-PCR: It is commonly known that PCR prefers smaller template molecules, and, therefore, amplifies longer molecules less effectively (Phillips and Eberwine, 1996; Luo et al., 1999). Moreover, differential display is prone to generate false positive signals caused by PCR amplification, which can reach unacceptable levels (Baldwin et al., 1999; Luce and Burrows, 1998). The origin of the false positives is probably the PCR reaction, which is sensitive to minor variations in the different templates (Liang and Pardee, 1994) and to mispriming favoured by the relatively low annealing temperature. This can lead to imbalances in the observed gene expression levels (Mohr et al., 1997; Luce and Burrows, 1998; Baldwin et al., 1999). Therefore, it is important to confirm the results by a second method such as in situ PCR, in situ hybridization, or specifically-primed single cell RT-PCR (Brandt et al., 1999) to exclude false positives and other PCR artefacts.
Primer selection: For the present study, one arbitrary primer in combination with one degenerated Oligo dT primer (Kleber-Janke and Krupinska, 1997) was chosen. This set of primers can theoretically be expected to amplify about 11% of all mRNAs present in a cell. At least 25 different primers would be necessary to amplify nearly all the transcripts (Bauer et al., 1993). Therefore, it is likely that, in this experiment, several genes were not covered by the primers used in the PCR reaction. The application of a combination of several different arbitrary primers should lead to a more comprehensive coverage of the transcriptome (Bauer et al., 1993).
Array hybridization: Hybridization is not necessarily uniform over the entire filter and differences in the local hybridization as well as the different local background and the efficiency of spotting can lead to differences in observed gene activity. Although some effort is spent performing mathematical corrections in order to compensate for these irregularities, they remain minor factors when compared to sampling and RT-PCR. Non-specific hybridization could also be a problem and lead to false positives. Highly stringent washing after the hybridization could solve this problem, but could also lead to the loss of some signals.
In conclusion, as far as is known this is the first report demonstrating a successful combination of tissue-specific sampling from undamaged plants and array hybridization. Although confirmation of data by other methods remains invaluable, the described hybridization allows fast and comprehensive gene expression analysis.
|
Acknowledgements |
|---|
We are thankful to Sabine Fischer and Dr Bernd Essigmann for providing the array filters and to Megan McKenzie for critically reading the manuscript. Professor Lothar Willmitzer is gratefully acknowledged for many fruitful discussions.
|
References |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
Baldwin D, Crane V, Rice D. 1999. A comparison of gel-based, nylon filter and microarray techniques to detect differential RNA expression in plants. Current Opinion in Plant Biology 2, 96–103.[Web of Science][Medline]
Bauer D, Müller H, Reich J, Riedel H, Ahrenkiel V, Warthoe P, Strauss M. 1993. Identification of differentially expressed mRNA species by an improved display technique (DDRT-PCR). Nucleic Acids Research 21, 4272–4280.
Belyavsky A, Vinogradova T, Rajewsky K. 1989. PCR-based cDNA library construction: general cDNA libraries at the level of a few cells. Nucleic Acids Research 17, 2919–2932.
Brady G, Billia F, Knox J et al. 1995. Analysis of gene expression in a complex differentiation hierarchy by global amplification of cDNA from single cells. Current Biology 5, 909–922.[Web of Science][Medline]
Brandt S, Kehr J, Walz C, Imlau A, Willmitzer L, Fisahn J. 1999. A rapid method for detection of plant gene transcripts from epidermal, mesophyll and companion cells of intact leaves. The Plant Journal 20, 245–250.[Web of Science][Medline]
Bret-Harte S. 1993. Total epidermal cell walls of pea stems respond differentially to auxin than does the outer epidermal wall alone. Planta 190, 379–386.
Compton J. 1991. Nucleic acid sequence-based amplification. Nature 350, 91–92.[Medline]
DeRisi JL, Penland L, Brown PO, Bittner ML, Meltzer PS, Ray M, Chen Y, Su YA, Trent JM. 1996. Use of cDNA microarray to analyse gene expression patterns in human cancer. Nature Genetics 14, 457–460.[Web of Science][Medline]
DeRisi JL, Iyer V, Brown PO. 1997. Exploring the metabolic and genetic control of gene expression on a genomic scale. Science 278, 680–686.
Desprez T, Amselem J, Caboche M, Höfte H. 1998. Differential gene expression in Arabidopsis monitored using cDNA arrays. The Plant Journal 14, 643–652.[Web of Science][Medline]
Dresselhaus T, Lörz H, Kranz E. 1994. Representative cDNA libraries from few plant cells. The Plant Journal 5, 605–610.[Web of Science][Medline]
Dzelzkalns VA, Nasrallah JB, Nasrallah ME. 1992. Cell–cell communication in plants self-incompatibility in flower development. Developmental Biology 153, 70–82.[Web of Science][Medline]
Grosset J, Marty I, Chartier Y, Meyer Y. 1990. Messenger RNAs newly synthesized by tobacco mesophyll protoplasts are wound inducible. Plant Molecular Biology 15, 485–496.[Web of Science][Medline]
Ishitani M, Majumder AL, Bornhouser A, Michalowski CB, Jensen RG, Bohnert H. 1996. Coordinate transcriptional induction of myo-inositol metabolism during environmental stress. The Plant Journal 9, 537–548.[Web of Science][Medline]
Jacobson BS, Fong F, Heath RL. 1975. Carbonic anhydrase of spinach. Studies on its location, inhibition, and physiological function. Plant Physiology 55, 468–474.
Karrer EE, Lincoln JE, Hogenhout S, Bennett AB, Bostock RM, Martineau B, Lucas WJ, Gilchrist DG, Alexander D. 1995. in situ isolation of mRNA from individual plant cells: creation of cell-specific cDNA libraries. Proceedings of the National Academy of Sciences, USA 92, 3814–3818.
Kehoe DM, Villand P, Sommerville S. 1999. DNA microarrays for studies of higher plants and other photosynthetic organisms. Trends in Plant Science 4, 38–41.[Web of Science][Medline]
Kehr J. 2001. High resolution spatial analysis of plant systems. Current Opinion in Plant Biology 4, 197–201.[Web of Science][Medline]
Kleber-Janke T, Krupinska K. 1997. Isolation of cDNA clones for genes showing enhanced expression in barley leaves during dark-induced senescence as well as during senescence under field conditions. Planta 203, 332–340.[Web of Science][Medline]
Koltai H, McKenzie Bird D. 2000. High throughput cellular localization of specific plant mRNAs by liquid-phase in situ reverse transcription-polymerase chain reaction of tissue sections. Plant Physiology 123, 1203–1212.
Lamb CJ. 1994. Plant disease resistance genes in signal perception and transduction. Cell 76, 419–422.[Web of Science][Medline]
Leidreiter K, Kruse A, Heinike D, Robbinson DG, Heldt HW. 1995. Subcellular volumes and metabolite concentrations in potato (Solanum tuberosum cv. Desiree) leaves. Botanica Acta 108, 439–444.
Liang P, Pardee AB. 1992. Differential display of eukaryotic messenger RNA by means of the polymerase chain reaction. Science 257, 967–971.
Liang P, Pardee AB. 1994. Differential display of mRNA by PCR. In: Ausubel FM, Brent R, Kingston, RE, Moore DD, Seidman JG, Smith JA, Struhl K, eds. Current protocols in molecular biology. New York: John Wiley and Sons, 15.8.1–15.8.8.
Luce MJ, Burrows PD. 1998. Minimizing false positives in differential display. Biotechniques 25, 766–770.
Luo L, Salunga RC, Guo H et al. 1999. Gene expression profiles of laser-captured adjacent neuronal subtypes. Nature Medicine 5, 117–122.[Web of Science][Medline]
Martel R, Cloney LP, Pelcher LE, Hemmingsen SM. 1990. Unique composition of plastid chaperonin-60: 
and ß polypeptide-encoding genes are highly divergent. Gene 94, 181–187.[Web of Science][Medline]
Mezitt LA, Lucas WJ. 1996. Plasmodesmal cell-to-cell transport of proteins and nucleic acids. Plant Molecular Biology 31, 251–273.
Mohr S, Cullen P, Schmidt R, Cignarella A, Assmann G. 1997. Avoidance of false positives in PCR-based mRNA differential display during investigation of non-standardized tissues or cells. Clinical Chemistry 43, 182–186.
Newman T, Debruijn FJ, Green P et al. 1994. Genes galore: a summary of the methods for accessing the results of large-scale partial sequencing of anonymous Arabidopsis thaliana cDNA clones. Plant Physiology 106, 1241–1255.[Abstract]
O’Donnell PJ, Calvert C, Atzorn R, Wasternack C, Leyser HM, Bowle DJ. 1996. Ethylene as a signal mediating the wound response of tomato plants. Science 274, 1914–1917.
Phillips J, Eberwine J. 1996. Antisense RNA amplification: a linear amplification method for analysing the mRNA population from single living cell. Methods 10, 283–288.[Medline]
Ponce M, Pérez-Pérez J, Piqueras P, Micol J. 2000. A multiplex reverse transcriptase- polymerase chain reaction method for fluorescence-based semiautomated detection of gene expression in Arabidopsis thaliana. Planta 211, 606–608.[Web of Science][Medline]
Posas F, Chambers JR, Heyman JA, Hoeffler JP, de Nadal E, Arino J. 2000. The transcriptional response of yeast to saline stress. Journal of Biological Chemistry 23, 17249–17255.
Reymond P, Weber H, Damond M, Farmer EE. 2000. Differential gene expression in response to mechanical wounding and insect feeding in Arabidopsis. The Plant Cell 12, 707–719.
Ruan Y, Gilmore J, Conner T. 1998. Torwards Arabidopsis genome analysis: monitoring expression profiles of 1400 genes using cDNA micro arrays. The Plant Journal 15, 821–833.[Web of Science][Medline]
Saiki RK, Gelfand DH, Stoffel S, Scharf SJ, Higuchi R, Horn GT, Mullis KB, Erlich HA. 1988. Primer-directed enzymatic amplification of DNA with a thermostable DNA polymerase. Science 239, 487–491.
Schena M, Shalon D, Davis RW, Brown PO. 1995. Quantitative monitoring of gene expression patterns with a complementary DNA microarray. Science 270, 467–470.
Schena M, Shalon D, Heller R, Cjai A, Brown PO, Davis RW. 1996. Parallel human genome analysis: microarray-based expression monitoring of 1000 genes. Proceedings of the National Academy of Sciences, USA 93, 10614–10619.
Schenk PM, Kazan K, Wilson I, Anderson JP, Richmond T, Sommerville SC, Manners JM. 2000. Coordinated plant defense response in Arabidopsis revealed by microarray analysis. Proceedings of the National Academy of Sciences, USA 97, 11655–11660.
Seki M, Narusaka M, Abe H, Kasuga M, Yamaguchi-Shinozaki K, Carninci P, Hayashizaki Y, Shinozaki K. 2001. Monitoring the expression pattern of 1300 Arabidopsis genes under drought and cold stresses by using a full-length cDNA microarray. The Plant Cell 13, 61–72.
Shu GG, Baum DA, Mets LJ. 1999. Detection of gene expression patterns in various plant tissues using non-radioactive mRNA in situ hybridization. The World Wide Web Journal of Biology 4, http://www.epress.com/w3jbio/vol4/shu.
Simone N, Bonner R, Gillepsie J, Emmert-Buck M, Liotta L. 1998. Laser capture microdissection: opening the microscopic frontier to molecular analysis. Trends in Genetics 14, 272–276.[Web of Science][Medline]
Thimm O, Essigmann B, Kloska S, Altmann T, Buckhout TJ. 2001. Response of Arabidopsis thaliana to Fe defiency stress as revealed by microarray hybridization. Plant Physiology 127, 1030–1043.
Titarenko E, Rojo E, Leon J, Sanchez-Serrano JJ. 1997. Jasmonic acid-dependent and -independent signaling pathways control wound-induced gene activation in Arabidopsis thaliana. Plant Physiology 115, 817–826.[Abstract]
Trenkle T, Welsh J, Jung B, Mathieu-Daude F, McClelland M. 1998. Non- stoichiometric reduced complexity probes for cDNA arrays. Nucleic Acids Research 26, 3883–3891.
Tsutsumi KI, Kagaya Y, Hidaka S, Suzuki JI, Tokairin Y, Hirai T, Hu DL, Ishikawa K, Ejiri SI. 1994. Structural analysis of the chloroplastic and cytoplasmic aldolase-encoding genes implicated the occurance of multiple loci in rice. Gene 141, 215–230.[Web of Science][Medline]
Wang E, Miller LD, Ohnmacht GA, Liu ET, Marincola FM. 2000. High-fidelity mRNA amplification for gene profiling. Nature Biotechnology 18, 457–459.[Web of Science][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  K. Ohtsu, H. Takahashi, P. S. Schnable, and M. Nakazono Cell Type-Specific Gene Expression Profiling in Plants by Using a Combination of Laser Microdissection and High-Throughput Technologies Plant Cell Physiol., January 1, 2007; 48(1): 3 - 7. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. Kurimoto, Y. Yabuta, Y. Ohinata, Y. Ono, K. D. Uno, R. G. Yamada, H. R. Ueda, and M. Saitou An improved single-cell cDNA amplification method for efficient high-density oligonucleotide microarray analysis Nucleic Acids Res., March 17, 2006; 34(5): e42 - e42. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J.-Y. Lee, M. Levesque, and P. N. Benfey High-Throughput RNA Isolation Technologies. New Tools for High-Resolution Gene Expression Profiling in Plant Systems Plant Physiology, June 1, 2005; 138(2): 585 - 590. [Full Text] [PDF]
|
|
|
|
 |
|
 L. C. HO and P. J. WHITE A Cellular Hypothesis for the Induction of Blossom-End Rot in Tomato Fruit Ann. Bot., March 1, 2005; 95(4): 571 - 581. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. P. Brandt Microgenomics: gene expression analysis at the tissue-specific and single-cell levels J. Exp. Bot., February 1, 2005; 56(412): 495 - 505. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
B. C. Meyers, D. W. Galbraith, T. Nelson, and V. Agrawal Methods for Transcriptional Profiling in Plants. Be Fruitful and Replicate Plant Physiology, June 1, 2004; 135(2): 637 - 652. [Full Text] [PDF]
|
|
|
|
|
|
P. J. WHITE and M. R. BROADLEY Calcium in Plants Ann. Bot., October 1, 2003; 92(4): 487 - 511. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. Honys and D. Twell Comparative Analysis of the Arabidopsis Pollen Transcriptome Plant Physiology, June 1, 2003; 132(2): 640 - 652. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
N. M. Kerk, T. Ceserani, S. L. Tausta, I. M. Sussex, and T. M. Nelson Laser Capture Microdissection of Cells from Plant Tissues Plant Physiology, May 1, 2003; 132(1): 27 - 35. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||