Journal of Experimental Botany, Vol. 53, No. 367, pp. 351– 359, February 2002
Linear amplification coupled with controlled extension as a means of probe amplification in a cDNA array and gene expression analysis during cold acclimation in alfalfa (Medicago sativa L.) Sergey Ivashuta1,2,3, Kazuhiro Uchiyama2, Mitsuru Gau2 and Yoshiya Shimamoto1 1 2
Graduate School of Agriculture, Hokkaido University, North-9 West-9, Kita-ku, Sapporo 060-8589, Japan National Agricultural Research Center for Hokkaido Region, Hitsujigaoka 1, Sapporo 062-8555, Japan
Received 29 May 2001; Accepted 19 September 2001
Abstract This study describes a rapid and simple way to amplify limited amounts of probes used for cDNA array hybridization while maintaining the original representation of transcripts in the samples. The approach is based on linear amplification of cDNAcoupled controlled extension of amplified products and yielded a 50 –75-fold increases in hybridization signal intensity. Controlled extension of products is achieved either by adjusting the amplification conditions or by using a digested template. Linear amplification with controlled extension generates a population of fragments consisting mainly of 39-end portions of original transcripts and ranging in length from 200 to 800 nucleotides. cDNA array analysis revealed that amplified and non-amplified probes generate expression profiles with correlations ranging from r ¼ 0.857 to 0.895. Up to 90% of cDNA clones, differentially expressed during cold acclimation in alfalfa, could be detected with both types of probes. This amplification method should increase the utility of cDNA arrays for identifying novel differentially expressed genes as well as expression profiling in specialized tissues or cells when the amount of analysed material is limited. The possibility of diminishing cross-hybridization of long genes sharing high sequence homology and improving the hybridization kinetics of complex probes after amplification is also discussed. 3
Key words: cDNA array, cold acclimation, controlled extension, linear PCR, probe amplification.
Introduction Differential hybridization of a total population of transcripts derived from physiologically different samples to DNA molecules arrayed on a solid support is one of the most widely used approaches for identifying novel differentially regulated genes or for analysing genomewide expression profiles. There are many techniques based on this approach and they can involve hybridization of either RNA or cDNA to plasmid DNA, PCR products or oligonucleotides arrayed at various densities on a support, which may be glass, nylon or silicon (Baldwin et al., 1999). A cDNA array technique that utilizes either ESTs or unidentified cDNA clones immobilized on a solid support has been used successfully in plant biology to identify genes specifically expressed in different organs or tissues, at various developmental stages or in response to stresses (Schena et al., 1995; Ruan et al., 1998; Aharoni et al., 2000; Reymond et al., 2000). Although this technique has a number of known strengths (Harmer and Kay, 2000) it also has several drawbacks associated with the hybridization of highly complex probes to arrayed DNA. These limitations include, amongst others, the need for a relatively large amount of RNA for labelling, significant variation in hybridization kinetics among individual
To whom correspondence should be addressed at the National Agricultural Research Center for Hokkaido Region. Fax: q11 859 2178. E-mail:
[email protected]
ß Society for Experimental Biology 2002
352
Ivashuta et al.
transcripts in a population, and the possibility of crosshybridization among close family members, which may lead to misinterpreted results (Byrne et al., 2000; Ruan et al., 1988; Trenkle et al., 1998; Watson et al., 1998; Wetmur, 1991). Usually, even the most sensitive methods require about 1–2 mg poly Aq RNA in a microarray experiment to create a high concentration of the probe in the hybridization buffer in order to maximize detection of rare messengers (Watson et al., 1998). Such an amount is prohibitively large when expression profiles of unique tissues, cell lines or even some organs of small plants are to be analysed. To overcome this limitation, several methods that rely on amplification of RNA in vitro using RNA polymerase (Phillips and Eberwine, 1996; Byrne et al., 2000), the PCR amplification of cDNA (Froussard, 1993; Spirin et al., 1999; Endege et al., 1999) or PCR amplification of cDNA biased towards the 39 ends (Dixon et al., 1998; Hertzberg et al., 2001) have recently been developed. An alternative to probe amplification is the dendrimer method, which improves the sensitivity of signal detection and therefore requires less RNA for probe synthesis (Stears et al., 2000). Although these methods are useful, they are labourintensive and too expensive for routine use in the average laboratory or they require optimization of amplification conditions to avoid the ‘plateau’ effect. Therefore, a simple and inexpensive way to amplify a probe would provide wider application of the DNA array technique in gene expression analysis. The present study reports on the linear amplification of double-stranded cDNA coupled with the controlled extension of amplified products. This process is denoted as LACE (Linear Amplification with Controlled Extension). The results, obtained by means of LACE of cDNA, indicate that this technique is suitable for rapid, inexpensive, and robust amplification of a probe up to 50–75-fold while maintaining the original representation of the transcripts in samples. This technique was applied to analyse the regulation of gene expression during cold acclimation in alfalfa (M. sativa L.).
Materials and methods Plant material and total RNA isolation Alfalfa seeds (M. sativa L. cv. Rambler) were sown in artificial soil and incubated for 6 weeks in a chamber at 22 8C, 12u12 h dayunight periods. For cold treatment plants were moved to 2 8C for 3 d. Total RNA was isolated from the aerial parts of control and cold-stressed plants with TRIzol reagents (Life Technologies, Maryland, USA). cDNA library construction
Poly(A) RNA was selected from total RNA of cold-acclimated plants using Dynobeads Oligo(dT)25 (Dynal AS, Oslo, Norway). ds cDNA was constructed according to the manual
for a cDNA synthesis kit (Stratagene, La Jolla, CA, USA) and size-fractionated on a Sepharose CL-2B column. cDNAs larger than 800 bp were ligated into Uni-Zap XR Vector (Stratagene) and packaged with the Gigapack III Gold packaging extract, yielding 2.6 3 106 plaque-forming units ( pfu).
cDNA array preparation and hybridization In total, 5 3 105 pfu of the lambda phage were excised in vitro (exAssistuSOLR system; Stratagene), and individual colonies were transferred to LBuAmp plates. Phagemid inserts were PCR-amplified using primers that were complementary to the vector sequences flanking both sides of the cDNA insert. Following amplification, 3 ml of PCR products was examined by agarose gel electrophoresis, and reactions in which a clear single band was detected, were used for array preparation. Four hundred and fifty-three anonymous cDNA clones with sizes ranging from 0.7 to 4.3 kb (average 1.8 kb) and 17 clones representing genes with known functions were arrayed as diluted PCR products (approximately 150 –200 ng) on duplicate Hybond Nq membranes (Amersham Pharmacia). Two control clones were spotted on each membrane. The cold-inducible Cas18 cDNA (Wolfraim et al., 1993) was used as a positive control and a cDNA clone for an unknown gene that demonstrated the same expression level in preliminary experiments as an internal control. For a negative control, a DNA fragment of pBluescript SK phagemid (Stratagene) containing multiple cloning sites was spotted. DNA was denatured by placing the membrane for 1 min onto two sheets of Whatman 3MM paper what had been saturated with 0.5 M NaOH, followed by rinsing in neutralization solution (1.5 M TRIS-HCl, pH 7.5, 1.5 M NaCl) for 1 min and then cross-linking by baking at q80 8C for 2 h.
Probe preparation and amplification
Total RNA from control and cold-stressed plants was treated with deoxyribonuclease I (Life Technologies), and poly(A) RNA was selected with Dynobeads Oligo(dT)25 (Dynal). First-strand cDNA synthesis was conducted with MMLV reverse transcriptase and linker-primer at 37 8C for 1 h, essentially as described in the manual for the cDNA synthesis kit (Stratagene) except that normal dCTP was used instead of a 5-methyl dCTP analogue. For second-strand synthesis, a second-strand mix including RNase I and DNA polymerase I was added, followed by incubation for 2.5 h at 16 8C. cDNA was purified with a MicroSpin S- 400 HR column (Amersham Pharmacia) and either digested with MseI (TTAA) or directly used for amplification. About 50 ng of the cDNA was digested with 1 unit of MseI for 1.5 h at 37 8C, heat-treated at 75 8C for 10 min to inactivate MseI, and then purified with a Suprec-02 column (Takara, Japan). Twenty-five ng of undigested or digested template was linearly amplified with Ddif1 primer (59-GAGAGAGAGAGAGAACTAGTCTCGAGT; concentration in master mix, 1 mM), complementary to the portion of oligo(dT) linker-primer used for cDNA synthesis. Amplification was carried out in a volume of 25 ml with either AmpliTaqGold (Applied Biosystems, Japan) or Pfu polymerase (Promega, Madison, WI, USA) in a GeneAmp PCR System 9600 (Applied Biosystems). PCR was carried out in a buffer provided with DNA polymerase including MgCl2 (2.5 mM) and dNTPs (0.2 mM). Thermal cycling conditions consisted of 99 cycles of 94 8C for 20 s and 65 8C for 10 s for the undigested template. In the present experiments, the ramping time for transition from 65 8C to 94 8C was sufficient for extension. For the
Amplification of probes digested template 99 cycles of 94 8C for 20 s, 65 8C for 10 s, and 72 8C for 20 s were used. Probe labelling, hybridization and detection Non-amplified and amplified probes were labelled according to the manual for the ECL nucleic acid labelling system (Amersham Pharmacia), except that the amplified probes were not denatured prior to labelling. Membranes were hybridized with probes overnight, washed in primary washing buffer (6 M urea, 0.4% SDS, 0.25 3 SSC) and then in secondary washing buffer (2 3 SSC), after which signals were detected with ESL detection reagents. The data were collected from X-ray films using GS-700 Imaging Densitometer (Bio-Rad Laboratories Inc., CA, USA) or directly using a Typhoon 8600 Variable Mode Imager (Molecular Dynamics, Amersham Pharmacia Biotech, Uppsala, Sweden). The signals were quantified with Image Master TotalLab ver. 1.00 software (Amersham Pharmacia Biotech). Hybridization signals produced in different experiments were normalized according to controls by calculating the ratios of intensities of control spots on different membranes. When the ratio was other than 1 : 1, the ratio value was applied to normalize for the differences in intensities. Alkaline gel electrophoresis and transfer
Two ml of amplification reaction was mixed with alkaline 2 3 loading buffer (200 ml glycerol, 750 ml water, 46 ml saturated BPB, and 5 ml 5N NaOH) and loaded on 1.2% alkaline agarose gel. The gel was run in alkaline buffer (30 mM NaOH, 2 mM EDTA) at 100 mA for 3.5 h. The separated products were transferred under alkaline conditions to Hybond Nq membrane (Amersham Pharmacia). Following transfer, the membrane was rinsed in neutralization solution for 10 min, baked at q80 8C for 2 h, and hybridized with ESL-labelled probes.
353
single-stranded fragments, which could improve labelling efficiency. Twenty-five ng of double-stranded cDNA, undigested or digested with MseI, was subjected to linear PCR in a 25 ml reaction volume using a primer complementary to the portion of the sequence of the oligo-dT linker-primer used for cDNA synthesis. Pfu polymerase was used for amplification of the undigested template to generate a LACEpfu probe. AmpliTaqGold polymerase was used to amplify both the digested and undigested templates to generate LACEmse and LACEtag probes, respectively. Maintenance of the original representation of the RNA samples
The maintenance of the original representation of transcripts in samples is the main concern of any probe amplification procedure. Comparative analysis was performed by hybridization of LACE-amplified and non-amplified probes to cDNA clones spotted as PCR products on a nylon membrane. A total of 17 clones with known sequences and 453 anonymous cDNA clones randomly selected from a cDNA library were arrayed on duplicate membranes, and six independent sets of hybridization experiments with each LACEpfu, LACEtag and non-amplified probes were carried out. Additionally, two experiments, consisting of a total of 188 cDNA clones, were conducted with LACEmse probes. The reproducibility of signals was evaluated by two or three independent experiments. Each hybridization used
Northern analysis Seven mg of total RNA from control and cold-stressed plants were subjected to electrophoresis in denaturating 1% formaldehyde agarose gel and capillary transferred to Hybond Nq membrane (Amersham, Pharmacia). Equal loading of RNA was assessed by EtBr staining. The blots were probed with PCR amplified cDNA fragments, which were ESL-labelled.
Results Amplification of probes
cDNAs derived from control and cold-stressed alfalfa plants were used as templates for linear amplification. Primer extension in the LACE procedure was restricted by either changing the duration of extension time or digesting the template with a restriction enzyme (Fig. 1). Such an approach generates a population of truncated fragments lacking a portion of the original template sequence at the 59-ends. The procedure was designed with the following aims: (1) to accelerate the process of linear amplification when a large number of cycles are required; (2) to generate shorter fragments, which may reduce cross-hybridization of close family members and improve hybridization kinetics, and (3) to generate
Fig. 1. Outline of LACE. Details of the protocol are given in Materials and methods. Briefly, 25 ng of double-stranded cDNA (undigested or digested with MseI ) was linearly amplified with primer, complementary to the portion of oligo(dT) linker-primer used for cDNA synthesis. Thermal cycling conditions consisted of 99 cycles of 94 8C for 20 s, 65 8C for 10 s and 72 8C for 0 or 20 s for the undigested and digested template, respectively.
354
Ivashuta et al.
1 ml of LACE-amplified products, including 1 ng of original cDNA or 100 ng of non-amplified cDNA. Such amounts of DNA were sufficient to generate detectable signals in more than 80% of arrayed cDNAs under conditions of this study. Membranes were hybridized with labelled probes overnight and then exposed to X-ray films for 7 min, 30 min, and 4 h. To ensure that only newly amplified single-stranded products participated in the development of the hybridization signal, the LACE-amplified probes were not heat-denatured before labelling. However, heating a DNA sample at 55 8C for 1–2 min may be useful in removing secondary structures. In the control, when the heat-denaturation step was
Fig. 2. cDNA array analysis of gene expression during cold acclimation. An array containing 96 cDNA clones was hybridized with probes from control or cold-stressed tissues. The probes were either linearly amplified using AmpliTaqGold polymerase (LACEtaq, top row) and Pfu polymerase (LACEpfu, bottom) or not-amplified (middle row). One hundred ng of non-amplified ds cDNA or 1 ml of LACE products (including 1 ng of non-amplified ds cDNA) was used in each hybridization experiment.
skipped in the labelling of double-stranded cDNA, almost no signals were observed. Experiments with LACEpfu, LACEtag and nonamplified probes generated quite similar but not identical expression profiles (Fig. 2; Table 1). To assess the correlation between profiles generated with non-amplified and LACE-amplified probes, the experiments were partially repeated three times. In total, 188 cDNA clones were used; however, 34 clones were eliminated from further analysis because of weak expression or high background. Correlation coefficients were 0.873 and 0.895 for LACEpfu and LACEtag, respectively. Figure 3 shows a scatter plot of gene expression ratios generated with non-amplified versus LACEpfu probes. Detailed analysis of the results, based on a comparison of signal intensities at various exposure times, showed that about 80 –90% of the clones that had been differentially expressed with non-amplified probes were also discerned by hybridization with LACEpfu and LACEtaq probes. Most of the arrayed clones (about 93%) showed nearly proportional enhancement of hybridization signals when hybridized with LACE-amplified probes, suggesting that the rates of transcript amplification were similar. Hybridization of cDNA arrays with LACEmse-amplified probes produced expression profiles with more distinct patterns. Furthermore, an average of 24% of clones that had each been clearly detected with a non-amplified probe showed weak or absent hybridization signals. Meanwhile, the expression ratios of most cDNA clones were similar to those obtained with non-amplified probes (correlation coefficient ¼ 0.857). Northern blot analysis of randomly selected cDNA clones identified as differentially expressed with either all three probes (16 clones) or only one of the probes (4 clones for each probe) was carried out to calculate the average percentage of ‘false’ positives in each experiment. Out of 16 overlapped clones, 12 (75%) were corroborated by Northern analysis. Two, one, and none of the clones identified with only a certain probe were confirmed as being differentially expressed in experiments with LACEpfu, LACEtag, and non-amplified probes, respectively. Thus, the average rate of corroborated differentially expressed clones was similar among the three, but highest in the LACEpfu experiments (Table 1).
Table 1. Summary of results of cDNA array experiments Characteristic of clones
Differentially expressed in each cDNA array experiment a, b Corroborated by Northern analysis a b c
Based on two replicate experiments. Having a ratio )2.5. ND, not determined.
Number of clones detected in experimentsunumber of analysed clones (%) Non-amplified
LACEpfu
LACEtaq
LACEmse
59u470 (12.5) 12u20 (60.0)
58u470 (12.3) 14u20 (70.0)
64u470 (13.6) 13u20 (65.0)
14u188 (7.5) ND c
Amplification of probes
Fig. 3. A scatter plot of gene expression ratios (cold-acclimated against non-acclimated plants) calculated from experiments with non-amplified versus LACE-pfu amplified probes. Overall, 154 expression ratios were estimated in three replicates and average ratio values were plotted.
While generally the rates of true differentially expressed cDNA were high enough, some inconsistencies were noticed between cDNA array and RNA gel blot results on several positive clones (for example, Fig. 6D, F). Quantity and size variations of amplified cDNA
The ultimate product of LACE is a complex population of single-stranded DNA fragments that corresponded to the 39 end of mRNAs. Various methods were used to determine the yield of LACE-products after 99 cycles of linear PCR. The purified product was quantified by measuring the optical density using a conversion factor for single-stranded DNA, by means of the EtBr-plate assay and based on the intensity of hybridization signals. The results obtained by the various approaches differed, while they were reproducible within the same amplification experiments. The yield depended on the brand of DNA polymerase used, the kind of template (digested versus undigested), the number of PCR cycles and the duration of extension (data not shown). On average, a 30–45-fold increase was achieved in the yield of LACEpfu products as determined by the EtBr plate assay or by reading the optical density and calculated as a ratio of initial template to single-stranded products of amplification. However, hybridization experiments with a series of diluted ‘targets’ showed at least a 50–75-fold enhancement of signal intensity. For most of the cDNA clones analysed, 100 ng of the labelled cDNA usually generated a signal intensity similar to that achieved with only 1 ml of amplification product drawn from a 25 ml reaction in which 25 ng of double-stranded cDNA was used as the initial template. To analyse the size variations in LACE products, 2 ml of amplification reaction was run on alkaline agarose gel,
355
Fig. 4. (A–E) Size variation of individual transcripts in LACE-amplified probes. Two ml of each LACE amplification reaction was run on alkaline agarose gel, transferred to a nylon membrane, and probed with (A) Msdh2 (1.3 kb), (B) MsftsH1 (2.4 kb), (C) Msnat36 (3.8 kb) and (D) Msptf11 (1.2 kb) cDNA clones which are up-regulated by cold. (E) Hybridization with cDNA used as internal control in cDNA array experiments shows equal loading of DNA.
transferred DNA to a nylon membrane, and independently probed with several novel cold-induced genes ranging from 1.2 to 3.8 kb. Figure 4 shows the results of hybridization of the clones Msdh2 (1.3 kb), MsftsH1 (2.4 kb), Msnat36 (3.8 kb), and Mspts11 (1.2 kb) with LACE-amplified products. To assess the equal loading, the membrane was probed with cDNA that was used as the internal control in our experiments (Fig. 4E). The results suggest that both polymerases used in this study generated a relatively broad range of lengths of products, although, as expected, Pfu polymerase tended to generate a narrower range and shorter fragments under the same conditions. Unlike AmpliTaqGold polymerase, which generated a sufficient quantity of full-sized products of short transcripts (Fig. 4A), Pfu polymerase produced fragments within 25–65% of the length of the original transcript (Fig. 4A, B, C). Figure 4 also demonstrates that the size of amplified products does not depend on the length of original transcripts, and therefore, the LACE-amplified probes are somewhat ‘normalized’ in terms of length in contrast to non-amplified probes. When the lengths of products were restricted by template digestion rather than by duration of extension period, the length did not vary much and a distinct band of the predicted size was observed with some ‘smearing’ forward to products of lower molecular weight (Fig. 4D). While Mse1 (TTAA) was selected so that the restriction site was more likely to be within AU-rich plant 39 UTR, it is evident that a combination of several enzymes would be needed to ensure that most transcripts in the population are digested as well as that the 39-end fragments are long enough to be used as probes. To analyse the preferential representation of the 39-ends of mRNA in the amplified population further, various parts of the Msdh2 gene were amplified, resolved
356
Ivashuta et al.
Fig. 5. Southern hybridization of various fragments of Msdh2 clone with a cold-stressed LACEpfu probe. Different portions of the Msdh2 cDNA clone were PCR-amplified from plasmid DNA using specific primers, except for fragment 1, which was amplified using universal primers and contains short portions of the vector sequence. Products were electrophoresed, transferred to a membrane, and probed with a LACEpfu probe. The gel stained with EtBr (bottom) demonstrates equal loading of PCR products.
on agarose gel, transferred to a nylon membrane, and probed with non-denatured LACEpfu products. Under stringent conditions, only fragments that contained the 39-end of the gene sequence were hybridized with the probe (Fig. 5). However, hybridization under lessstringent conditions and with a longer exposure time also showed faint signals in fragments corresponding to the 59-end of a gene that lacks a significant portion of the 39-end of cDNA (data not shown). The latter can be explained either by the presence of some amount of full-length products in amplified probes or by crosshybridization between different regions of the gene sequence or with other closely related genes. Gene expression analysis during cold acclimation of alfalfa using LACEpfu-amplified probes
cDNA was defined as cold-regulated if it had an expression ratio more than 2.5-fold that of the control clone. Of 470 cDNA screened in the cDNA array analysis with LACEpfu-amplified probes, 35 clones (7.4%) were
up-regulated and 23 clones (4.9%) were down-regulated after 3 d of cold acclimation at 2 8C. Partial sequence analysis of 14 clones that were corroborated by Northern hybridization revealed that some clones shared a high degree of sequence similarity, and that two clones were each represented twice, reflecting some level of redundancy in the arrayed clones. Both clones were highly expressed at a low temperature (Fig. 6D, I). After correction for redundancy, 12 unique clones were analysed. In most cases, the expression level was detected appropriately on the cDNA array as compared with the Northern blot (Fig. 6A–L). A similarity search was carried out with BLASTX at NCBI (www.ncbi.nlm.nih.govublastu ), and the results of the search are shown in Table 2. Of the 12 clones, four had no significant similarities to the database sequences, and the other eight showed similarities to genes with known or predicted functions. One cDNA clone that was up-regulated by low temperatures (Fig. 6L) showed strong sequence similarity to the gene encoding chloroplast FtsH protease from tobacco (Seo et al., 2000). It has been reported that in eukaryotes the FtsH protein has protease and chaperone functions ( Nakai et al., 1995; Rep et al., 1996). In plants, the chloroplast FtsH protease is involved in the degradation of inactivated D1 protein of the photosystem II (Lindahl et al., 2000). These functions suggest that this gene may play an important role in cold acclimation and maintenance of functionality of the photosystem II under low temperatures conditions. Another gene that was found to be strongly induced by the cold (Fig. 6D) has no match in the database entries. Further analysis of the expression showed that both of these clones were up-regulated by cold stress, but not in response to NaCl, ABA, dehydration or heat-shock (data not shown). Results of a more-detailed analysis of these genes during cold acclimation of alfalfa will be given in subsequent papers. Expression patterns of clones that shared similarity with stress-induced protein (S40947), dehydrin-like protein (L07516) and ribulose biphosphate carboxylase small chain precursor (AF056315) were consistent with previous reports on their expression profiles (Luo et al., 1992; Wolfraim et al., 1993; Strand et al., 1997).
Discussion Tissue-specific expression of particular transcripts plays an important role in plant responses to environmental stimuli. However, systematic analysis of gene expression in specific tissues or cells of miniaturized or unique plants is severely limited by the large amount of RNA needed for standard probe labelling. The results of this study demonstrate that linear amplification of cDNA, when coupled with controlled extension, enables rapid
Amplification of probes
357
Table 2. Sequence analysis of differentially expressed cDNA clones identified on a cDNA array using LACEpfu probes and corroborated by Northern blot analysis No.b
Identification of RNA blots as in Fig. 5
Protein similarity a
1 2 3 4 5 6 7 8 9 10 11 12
A B C D E F G H I J K L
No similarity Ribulose biphosphate carboxylase small chain precursor Arabidopsis hypothetical protein No similarity Plastidic aldolase No similarity Stress-induced protein ATP-dependent Clp protease regulatory subunit Dehydrin-like protein, CAS18 NADH dehydrogenase like protein No similarity Cloroplast FtsH protease
Accession no.a
E-valuea
AF056315 AC012562
0.0 2e-30
AB027002
1e-84
S40947 AF036328 L07516 AL022603
0.70 1e-05 7e-48 2e-40
AB017480
1e-47
a
Results of BLASTX program search. The sequences have been deposited in GenBank under following accession numbers: 1-AF411549, 2-AF411547, 3-AF411550, 4-AF411545, 5-AF411551, 6-AF411546, 7-AF411552, 8-AF411553, 9-AF411554, 10-AF411555, 11-AF411548, 12-AF332134. b
Fig. 6. (A–J ) Comparison of RNA blot and cDNA array hybridization results. cDNA clones, identified as being differentially expressed on a cDNA array after hybridization with LACEpfu products, were used as probes in Northern hybridization. Expression ratios for RNA gel blot and cDNA array are indicated below the RNA gel blots. Equal loading of RNA was assessed by EtBr staining. Expression ratio values for cDNA array results are the mean of two or three ("standard deviation) experiments. Sequence similarity results for each clone are given in Table 2. ND, not determined.
amplification of a complex population of cDNAs while maintaining the original representation of individual transcripts. In contrast to transcriptional amplification of mRNA in vitro (Philips and Eberwine, 1996; Wodicka et al., 1997; Byrne et al., 2000), the LACE procedure is simple and inexpensive. On the other hand, unlike various permutations of PCR-based amplification of cDNA (Froussard, 1993; Spirin et al., 1999; Endege et al., 1999; Dixon et al., 1998; Hertzberg et al., 2001), this procedure does not require time-consuming optimization of amplification conditions. A recently developed method of PCR-based amplification of cDNA using the SMART technique (Clontech, Palo Alto, CA, USA) requires longdistance PCR and high-quality RNA (Spirin et al., 1999; Endege et al., 1999). Such an approach may be biased against decapped mRNA. Due to the nature of linear amplification, the results of LACE amplification do not depend on size variation of individual transcripts in a population and seem to be less sensitive to the partial 59-end degradation of unstable transcripts that may occur during sampling. Additionally, the LACE product is single-stranded cDNA and can therefore be directly used in a labelling reaction without the requirement of a denaturation step, therefore reducing the probability that a double-stranded cDNA template or contaminating DNA participating in detection. About 25 ng of LACE-amplified cDNA generally proved to be sufficient for 20–25 hybridizations (each membrane size was 8 3 12 cm). In these experiments, 25–50 ng of cDNA could be obtained from less than 2 mg of total RNA. Identification of several novel coldregulated cDNAs (Table 2; Fig. 6) demonstrated that the LACE approach is useful for finding differentially expressed genes. The correlation (ranging from r ¼ 0.857 to 0.895) between expression ratios generated with non-amplified and amplified probes suggests that this
358
Ivashuta et al.
approach can be used for systematic expression profiling; however, more-comprehensive analysis in various systems may be required. Although these results depended on specific techniques ( probe labelling, hybridization, detection and image processing) it is believed that the LACE method can be extended to accommodate more general settings. One of the limitations of linear amplification is the low rate of accumulation of the amplified product. However, in these experiments, a comparison of the average intensities of hybridization signals generated with nonamplified or LACE-amplified probes suggested increases of at least 50–75-fold in the concentration of labelled DNA that can provide a sufficient amount of probe for most applications. Meanwhile, it was found that LACE amplification of an undigested template led to a nonproportional increase in the signal intensities of about 7% of cDNA clones analysed, while the expression ratios between control and cold-stressed conditions for these clones were unchanged. The results of partial sequencing of several cDNA clones and Northern hybridization analysis provided no additional clues, but it is speculated that the lower rate of signal enhancement might have been due to a decrease of cross-hybridization of long transcripts sharing sequence homology at the 59-end. Another possibility is that the primer extension of some clones is restricted by the presence of highly stable secondary structures near the 39-end that can affect the amount of amplified products. It was demonstrated that linear PCR with limited extension time tends to generate fragments that consist mainly of the 39-end part of the sequence of an original transcript (Fig. 5) and that under certain conditions the majority of products ranged from 200 to 800 nucleotides (Fig. 4A, B, C) which is the optimal range for hybridization. By contrast, diffusion rates and optimal annealing conditions of the individual transcripts, the sizes of which in original cDNA samples may differ by factors greater than ten, can vary significantly. Thus, it is quite possible that the smaller range of variation and reduced average length of DNA fragments in the LACE probe could improve both hybridization kinetics and labelling efficiency. It is also supposed that the LACE approach can reduce cross-hybridization of family members, because the 39-ends of even closely related genes may be sufficiently divergent. In eukaryotes, 39 UTR may play an essential role in differential regulation of expressions of closely related genes by affecting transcript stability, localization and translation efficiency (Conne et al., 2000; Hoopes et al., 2000). Furthermore, in plant biology, biochemical studies suggest that growth at low temperature will result in the expression of low-temperature isoforms of enzymes involved in ‘houskeeping’ functions (Hughes and Dunn, 1996). The coding region of mRNA of such isoforms would be highly homologous, and the
differential expressions of some of them are unlikely to be detected in conventional differential screening. An example of such a gene is barley elongation factor-1a, which is a member of a multigene family in the barley genome. The low- temperature induction of this gene can be seen in Northern analysis only if the 39 UTR of the mRNA is used as a probe (Hughes and Dunn, 1996). To date, except for gel-based fingerprinting techniques, only a gene-specific oligonucleotide array (Wodicka et al., 1997; Ramsay, 1998) and SAGE (Velculescu et al., 1995) theoretically enable systematic analysis of the expression profiles of genes sharing significant sequence homology. However, these approaches are either expensive or require extensive sequence information. Amplification with controlled extension of 39-ends of mRNAs could provide an inexpensive alternative in probe modification for such expression profiling. The method described here tends to produce fragments with a relatively large average size and a wide range of length variation. This suggests that even Pfu polymerase, which has quite a low rate of dNTP incorporation, is too ‘fast’ for this application. While the LACE procedure is already an improvement over existing protocols, it is technically possible to improve its efficiency through shorter products with less variation. Ideally, products after amplification should be within 150–250 nucleotides, which corresponds to the average size of plant 39 UTR (based on 130 M. sativa and Arabidopsis mRNA sequences randomly selected from the GenBank database (www.ncbi.nlm.nih.govuentrez)), and various approaches toward this goal are being analysed. Linear amplification has been used in several applications, including sequencing (Murray, 1989). Linear pre-amplification in gel-based expression profiling has been used successfully (Ivashuta et al., 1999). Here it is demonstrated that linear PCR, when coupled with controlled extension, is useful for probe amplification in cDNA array expression analysis. One of the most biologically significant finding in this experiment is that up-regulation of the chloroplast FtsH protease was detected. As far as is known, this is the first example of a membrane-bound chloroplast protease whose expression is linked to cold acclimation. In addition, three novel differentially expressed cDNA clones were identified with LACE-amplified probes in this study. The mechanism underlying detection in these cases is unknown. It is speculated that various factors are involved, including reduced cross-hybridization and improved hybridization kinetics. In summary, the utilization of linear amplification of cDNA with controlled extension provides a rapid method for amplification of a probe for use in analysis of gene expression in unique tissues and cells. It will provide an inexpensive and robust complement to current approaches.
Amplification of probes
Acknowledgements SI was funded by the Japan Society of Science Promotion. We thank the anonymous referees for helpful and constructive comments.
References Aharoni A, Keizer LC, Bouwmeester HJ, et al. 2000. Identification of SAAT gene involved in strawberry flavor biogenesis by use of DNA microarray. The Plant Cell 12, 647–661. Baldwin D, Crane V, Rice D. 1999. A comparison of gel-based, nylon filter and mycroarray techniques to detect differential RNA expression in plants. Current Opinion in Plant Biology 2, 96–103. Byrne MC, Whitley MZ, Follettie MT. 2000. Preparation of mRNA for expression monitoring. In: Ausubel et al., eds. Current protocols in molecular biology. John Wiley & Sons, Inc., 22.2.1–22.2.13. Conne B, Stutz A, Vassalli JD. 2000. The 39 untranslated region of messenger RNA: a molecular ‘hotspot’ for pathology? Nature Medicine 6, 637–641. Dixon AK, Richardson PJ, Lee K, Carter NP, Freeman TC. 1998. Expression profiling of single cells using 3 prime end amplification (TPEA) PCR. Nucleic Acids Research 26, 4426 – 4431. Endege WO, Steinmann KE, Bordman LA, Thibodeau SN, Schlegel R. 1999. Representative cDNA libraries and their utility in gene expression profiling. BioTechniques 26, 542–550. Froussard P. 1993. rPCR: a powerful tool for random amplification of whole RNA sequences. PCR Methods and Applications 2, 185–190. Harmer SL, Kay SA. 2000. Microarray: determining the balance of cellular transcripts. The Plant Cell 12, 613–615. Hertzberg M, Sievertzon M, Aspeborg H, Nilsson P, Sandberg G, Lundeberg J. 2001. cDNA microarray analysis of small plant tissue samples using a cDNA tag target amplification protocol. The Plant Journal 25, 585–591. Hughes MA, Dunn MA. 1996. The molecular biology of plant acclimation to low temperature. Journal of Experimental Botany 47, 291–305. Hoopes BC, Bowers GD, Bowers MJ, DiVisconte MJ. 2000. The two Saccharomyces cerevisiae SUA7 (TFIIB) transcripts differ at the 39-end and respond differently to stress. Nucleic Acids Research 28, 4435– 4443. Ivashuta S, Imai R, Uchiyama K, Gau M. 1999. The coupling of differential display and AFLP approaches for non-radioactive mRNA fingerprinting. Molecular Biotechnology 12, 137–141. Lindahl M, Spetea C, Hundal T, Oppenheim AB, Adam Z, Andersson B. 2000. The thylakoid FtsH protease plays a role in the light-induced turnover of the photosysetem II D1 protein. The Plant Cell 12, 419– 431. Luo M, Liu JH, Mohapatra S, Hill RD, Mohapatra SS. 1992. Characterization of a gene family encoding abscisic acid- and environmental stress-inducible proteins of alfalfa. Journal of Biological Chemistry 267, 15367–15374. Murray V. 1989. Improved double-stranded DNA sequencing using the linear polymerase chain reaction. Nucleic Acids Research 17, 88–89.
359
Nakai T, Yasuhara T, Fujiki Y, Ohashi A. 1995. Multiple genes, including a member of the AAA family, are essential for degradation of unassembled subunit 2 of cytochrome c oxidase in yeast mitochondria. Molecular and Cellular Biology 8, 4441– 4452. Phillips J, Eberwine JH. 1996. Antisense RNA amplification: a linear amplification method for analysing the mRNA poulation from single living cells. Methods 10, 283–288. Ramsay G. 1998. DNA chips: state-of-art. Nature Biotechnology 16, 40– 44. Rep M, van Dijl JM, Suda K, Schatz G, Grivell LA, Suzuki CK. 1996. Promotion of mitochondrial membrane complex assembly by a proteolytically inactive yeast Lon. Science 274, 103–106. 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, Gilmor J, Conner T. 1998. Toward Arabidopsis genome analysis: monitoring expression profiles of 1400 genes using cDNA microarrays. The Plant Journal 15, 821–833. Schena M, Shalon D, Davis RW, Brown PO. 1995. Quantitative monitoring of gene expression patterns with a complementary DNA microarray. Science 270, 467– 470. Seo S, Okamoto M, Iwai T, Iwano M, Fukui K, Isogai A, Nakajima N, Ohashi Y. 2000. Reduced levels of chloroplast FtsH protein in Tobacco Mosaic Virus-infected tobacco leaves accelerate the hypersensitive reaction. The Plant Cell 12, 917–932. Spirin KS, Ljubimov AV, Castellon R, Wiedoeft O, Marano M, Sheppard D, Kenney MC, Brown DJ. 1999. Analysis of gene expression in human bullous keratopathy corneas containing limiting amounts of RNA. Investigative Ophthalmology and Visual Science 40, 3108–3115. Stears RL, Getts RC, Gullans SR. 2000. A novel, sensitive detection system for high-density microarrays using dendrimer technology. Physiological Genomics 3, 93–99. Strand A, Hurry V, Gustafsson P, Gardestrom P. 1997. Development of Arabidopsis thaliana leaves at low temperatures releases the suppression of photosynthesis and photosynthetic gene expression despite the accumulation of soluble carbohydrates. The Plant Journal 12, 605–614. 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. Velculescu V, Zhang L, Vogelstein B, Kinzler KW. 1995. Serial analysis of gene expression. Science 270, 484 – 487. Watson A, Mazumder A, Stewart M, Balasubramanian S. 1998. Technology for microarray analysis of gene expression. Current Opinion in Biotechnology 9, 609–614. Wetmur JG. 1991. DNA probes: applications of the principles of nucleic acid hybridization. Critical Review in Biochemistry and Molecular Biology 26, 227–259. Wodicka L, Dong H, Mittman M, Ho MN, Lockhart DJ. 1997. Genome-wide expression monitoring in Saccharomyces cerevisiae. Nature Biotechnology 15, 1359–1367. Wolfraim LA, Langis R, Tyson H, Dhindsa RS. 1993. cDNA sequence, expression and transcript stability of a cold acclimation-specific gene, cas18, of alfalfa (Medicago falcata) cells. Plant Physiology 101, 1275–1282.
