INTRODUCTION

Orchid production has become a world-wide business important in floricultural industry 1. In subtropical and temperate areas, phalaenopsis becomes the most important for orchid production, due to its showy, long-lasting flowers and a large selection of flower colors.

New clones of orchid hybrids are selected annually by different breeding programs and mass propagated through tissue culture using either meristem or inflorescence tip and nodes as starting materials 2, 3, 4, 5. In commercial laboratories, two methods are used to massively produce an elite orchid clone. Shoot multiplication from pre-existed nodal buds or meristems was adopted by many companies. The drawback of the method is the low speed of proliferation of multiple shoots for mass production. To overcome this, others used an alternative approach by inducing protocorm-like bodies (PLBs) from shoot tips or leaves and then further propagating for secondary or tertiary PLBs to obtain a large quantity 2, 3, 6, 7, 8. The PLBs will then differentiate into plantlets in later stage of micropropagation. However there are unpredictable mutations or somaclonal variation occurred during the process of multiplication. Some hybrids are more amenable than others to somaclonal variation. The percentage of the variations ranges from 0-100% depending on varieties, with an average of 10% among phalaenopsis hybrids 9. Ploidy level change was not detected among tested hybrids from tissue culture when analyzed by flow cytometry 9. However, 2,4-D was reported to affect ploidy levels in suspension cultures of Doritaenopsis 10. Peloric flower as well as flower color mutants occurred in many orchid hybrids through tissue culture process 11, 12.

The cause of somaclonal variation in higher plants has been reported during different biochemical and molecular events, including changes in DNA methylation pattern, activation of transposable elements or retroelements, and chromosome remodeling 13, 14, 15, 16, 17, 18, 19. Rice retrotransposons could be activated during the process of tissue culture 14. Activation of some retrotransposons has been found linked to chemical and physical causes, and biotic stresses, such as wounding and pathogen infection 17, 20. Molecular markers have been exploited for the detection of somaclonal variation, including RAPD 21, 22, Methylation sensitive RFLP 18, 23, 24, and microsatellite sequence variation 25. In tissue culture derived plantlets of oil palms, reduced level of DNA methylation in general has been observed 18. Somaclonal variation has also been reported in phalaenopsis by RAPD analysis on regenerated plants, which showed morphological and physiological changes in the flowers 21. The microsatellite instability could be induced by mutation in mismatch repair genes and by pathogen infection in the inflorescence 25, 26, 27.

Epigenetic changes in orchids have not been studied in depth so far. Due to limited information of classical genetic map and genomic sequences for phalaenopsis orchid, conventional molecular biological approach becomes the major tool for analysis of orchid development and developmental regulation. The variable genome sizes of Phalaenopsis among different species 28, 29, 30 make the genetic analysis more complicated. Mutants obtained from different methods, such as chemical mutagenesis, T-DNA insertion, and in vitro culture, provide opportunities to pursue developmental process of orchids. In this report, we described transcript profiling of a phalaenopsis peloric flower mutant derived from tissue culture, using randomly amplified polymorphic cDNAs (cDNA-RAPD) and suppression subtractive hybridization (SSH) methods. Expression levels of selected cDNA clones were then compared to the wild type as well as peloric and semi-peloric mutants using real-time RT-PCR to achieve informative understanding of peloric mutants from transcripts level.

MATERIALS AND METHODS

Plant materials and RNA isolation

For cDNA-RAPD analysis, four different sets of Phalaenopsis plants were used, including wildtype and semi-peloric flowers and flower buds of P. Zuma's Pixie (Obtained from Taida Orchids, Changhwa, Taiwan), P. Little Mary 'F535' and Doritaenopsis Minho Diamond 'F607' (obtained from Sogo Orchids, Pingtung, Taiwan). For cDNA subtraction, P. Little Mary 'F535' and an unknown hybrid (designated as P. D) were used (Fig. 1). For all experiments, flower buds younger than stage 4 (Fig. 2) were harvested and frozen in the liquid nitrogen then stored at -80ºC until use.

Figure 1
figure 1

Morphology of Phalaenopsis flowers used in the experiments. (A) Wildtype flower of P. Zuma's Pixie '#1', (B) Semi-peloric flower of P. Zuma's Pixie '#1', (B) Semi-peloric flower of P. Zuma's Pixie '#1'. (C) Wildtype flower of P. Little Mary, (D) Semi-peloric flower of P. Little Mary, (E) Peloric flower of P. Little Mary. (F) Wildtype flower of Doritaenopsis Minho Diamond 'F607'. (G) Peloric flower of Dtps. Minho Diamond 'F607', (H) Peloric flower of P. hybrid D.

Full size image
Figure 2
figure 2

Developmental stages of flower buds of P. Little Mary. (fb1-fb8) (bar = 2 cm)

Full size image

Total RNAs were isolated and purified using the TRIZOL reagents according to the manufacturer's instructions (Invitrogen). RNA precipitates were resuspended in DEPC-treated sterile water and precipitated with LiCl (a final concentration of 2M). The supernatant was treated with RNase-free DNase (Promega) at 37ºC for 30 min to remove residual genomic DNAs. Purified RNAs were quantified using a spectrophotometer (Hitachi U2000) and quality checked by agarose electrophoresis.

cDNA synthesis, RAPD, and cloning of DNA

Double stranded cDNAs were synthesized from 1 μg total RNAs using the SMART PCR cDNA Synthesis Kit following the manufacturer's instructions (Clontech). For RAPD analysis, the random decamer primers were obtained from Operon. The reaction mixtures contain 200 μg cDNA, 1×PCR buffer, dNTPs, 2 μM random primer, and 1U of Taq DNA polymerase (Takara) in a total volume of 20 μl. The PCR condition was as follow: 94ºC for 5 min, then 35 cycles of 94ºC 30 s, 38ºC 30 s, 72ºC 1 min, and finally an extension of 7 min at 72ºC. The amplified DNA products were separated on 2% agarose gel.

Differentially amplified DNA bands were purified and ligated into pGEM-T Easy vector (Promega). The positive inserts were checked by PCR using universal primers T7 and SP6. The cloned DNA fragments were sequenced with the ABI PRISM 3100 sequencer.

Suppression subtractive hybridization and DNA sequencing

To compare gene expression in flower buds of the peloric mutants and wild type of P. Little Mary, SSH was conducted as described previously 31. For comparison of differential gene expression in flowers and leaves, the peloric unknown hybrid P. D with similar flower color and size was used. Poly A+ RNAs were isolated from 75 μg total RNAs using the Oligo (dT)25 containing Dynabeads following manufacturer's instruction (Dynal Biotech). Equal amounts of poly A+ RNAs (2 μg) from flower buds and leaves were converted into double stranded cDNAs using the SMART PCR cDNA Synthesis Kit (Clontech).

Two subtracted libraries were constructed by using the PCR-Select cDNA Subtraction kit (Clontech). In both libraries, the wild type flower buds of P. Little Mary were used as driver and peloric flower buds as tester, and vice versa. A third subtracted library was constructed to compare the transcripts in flower buds and leaves, the P. D hybrid leaves were used as driver and peloric flower buds as tester. The subtracted clones were purified and cloned into pGEM-T Easy vector and positive inserts were sequenced.

Sequence analysis

The cDNA sequences were compared for similarities against GenBank with the BLASTX 32. The resulted matches were annotated. Classification of annotated sequences was according to the method described in 33.

Real-Time reverse transcriptase PCR assay

Real-time RT-PCR and data analysis were performed in the ABI PRISM 7900 Sequence Detection System using 2×SYBR Green Master Mix to monitor dsDNA synthesis (Applied Biosystems). The gene-specific forward and reverse primers (Tab. 1) were designed based on cloned partial cDNA sequences following the directions of Primer Express 2.0 (Applied Biosystems) and synthesized by Mission Biotechnology (Taipei). Each primer pairs was designed to amplify around 60-80bp length to allow optimized estimation. For controlling the integrity of RNA and normalizing target RNA copy numbers in mutant and wild type flower buds, the housekeeping gene actin (PACT4, AY134752) was amplified by real-time RT-PCR to generate a standard curve of actin mRNA levels. In order to distinguish the expression levels of individual target genes, their standard curves, used as the calibrators, were established using single stranded cDNAs of the wild type flower buds.

Table 1 Primers used for real-time PCR analysis
Full size table

PCR condition was essentially as described in Czechowski et al. 34 except that the total reaction volume was 25 μl for each sample and half amount of the 2×SYBR Green (12.5 μl) was added. We found this amount of the dye sufficient for PCR analysis. PCR condition was as follow: 50ºC for 2 min, 95ºC for 10 min, then 40 cycles of 95ºC 15 s, 60ºC 1 min, after that, one more cycle at 95ºC 15 s, 60ºC 15 s and 95ºC 15 s. Standard curve (Ct value against log ng template) for each target and housekeeping gene was established according the guides provided by the ABI PRISM 7900 Sequence Detection System.

RESULTS

Morphology of peloric mutant flowers

Wild type Phalaenopsis flowers possess three petal-like sepals, with one in the top or dorsal sepal, and two lower lateral sepals. There are two lateral petals and a specialized enlarged flamboyant bottom petal, called lip or labellum (Fig. 1A, C, F). In the middle of the flower there is a pistil/stigma fused together with pollinia to form so called column or gynostemium (Fig. 1). In rare cases some seedlings from sexual hybridization may generate lip-like lateral petals. When a selected orchid plant was propagated by flower stalk nodal culture, some clones showed different degrees of lip-like lateral petals, due to somaclonal variation. Mild change caused the lateral petals bulge slightly near the center, which was named semi-peloric flowers (Fig. 1D). In the severe case of somaclonal variation, enlarged central bulge of the lateral petals led the transition to lip-like, which was named peloric flowers (Fig. 1 B, E, G, H). The degree of the peloric flower formation varies in different Phalaenopsis hybrids as shown in Fig. 1. In severe cases, peloric flowers lost their pollinia (Fig. 1E). When the peloric mutant was re-propagated by tissue culture, some clones may revert back to the wild type flower morphology, an indication of epigenetic changes. Since peloric or semi-peloric mutants share the same genetic background with the wild type of the same variety, we took the advantage to investigate the peloric mutants on molecular level by using cDNA-RAPD and suppression subtractive hybridization techniques.

Differentially expressed ESTs by cDNA-RAPD

Two hundred random primers obtained from Operon (Operon Technologies, Inc., Alameda, CA), including OPAA01-20, OPB01-20, OPD01-20, OPE01-20, OPF01-20, OPT01-20, OPW01-20, OPX01-20, OPY01-20 and OPZ01-20, were used to amplify cDNAs from peloric and wild type flower buds as well as leaves. No PCR products were observed among 45 primers. Products from other 35 primers showed no difference in amplified DNA fragments. The remaining 120 primers produced a total of 510 differential DNA bands, including 93 from wild type flower buds, 366 from peloric flower buds, and 51 from leaves. Among those differentially expressed cDNAs, 191 bands were randomly picked for further characterization. Other bands were not included in the analysis due to noise background of their DNA sequences. After sequencing, only 90 inserts showed unambiguous sequence reading. The length of these cDNAs ranged from about 200 to 3000 bp. Those commonly expressed cDNAs were excluded for analysis. After searching against the GenBank by BlastX, the gene functions of the cDNA fragments were classified according to Bohnert et al. 33. Among them, 28 cDNA clones were preferentially expressed in the wild type flower buds, and 24 expressed in the peloric mutants (Fig. 3A). Annotation of these ESTs was listed in Tab. 2 and Tab. 3. A majority of the cloned cDNA inserts belongs to the unknown function. Only a small proportion of ESTs encoded proteins involved in transcription, metabolism, signal transduction, etc. (Fig. 3A). The efficiency of the cDNA-RAPD approach seems to be low. Interestingly, transoposons and orchid virus (Cymbidium mosaic virus, CyMV; and Odontoglossum ringspot virus, ORSV) were only detected in the peloric mutants in this experiment (Fig. 3A, Tab. 3).

Figure 3
figure 3

Functional category of the ESTs in Phalaenopsis flower buds. (A) ESTs from wild type (WT) and peloric mutant flower buds by cDNA-RAPD. (B) ESTs preferentially enriched in flower buds from cDNA-RAPD flower-to-leaf analysis. (C) ESTs enriched in either wild type or peloric mutants of P. Little Mary after SSH. (D) ESTs enriched in the peloric flower buds of P. Little Mary after flower-leaf SSH.

Full size image
Table 2 Homology search of ESTs enriched in wildtype flowers by cDNA-RAPD flower-flower assay
Full size table
Table 3 Homology search of ESTs specifically enriched in mutant flowers by cDNA-RAPD flower-flower assay
Full size table

When transcripts of peloric flower buds were compared to that of the leaves in the same mutant, 38 ESTs were preferentially expressed in the flowers (Fig. 3B). Again, most of the ESTs (68.4%) had no known function when searched against the GenBank (Tab. 4). One of the ESTs, 607FF03-2, shows homology to MtN3 of the nodules of Medicago truncatula 35 and the NEC1 of Petunia hybrida 36. NEC1 was specifically expressed in the nectary of the Petunia flowers 36.

Table 4 Homology search of ESTs enriched in peloric flower buds in cDNA-RAPD flower-leaf assay
Full size table

Among the identified clones by cDNA-RAPD, most of the ESTs belong to the unknown protein category, partially contributed by unknown genome information of orchid and the incomplete ESTs sequences. Some ESTs were probably representing genes involved in signal transduction, protein-protein interactions and transcription regulation, such as cyclophylin, putative ankyrin 37, auxin-related dual specificity cytosolic kinase, putative DEAD/DEAH box RNA helicase, putative WD-repeat protein, and DNA binding protein TGA1a (Tab. 2, 3). The RNA dependent RNA polymerase of CyMV, as well as several classes of retroelements or transposon has also been detected several times in the peloric flowers (Tab. 3, 4).

Differentially expressed ESTs by SSH

The forward subtractive cloning strategy of SSH (wild type flower buds as driver and peloric mutant as tester) generated 181 clones, and reverse subtractive SSH generated 104 clones (Fig. 3C). The highest percentage in the subtracted clones from both forward and reverse SSH was the unknown EST clones. Among them, 28.1% were from the wild type flower buds, while only 3.5% from the peloric flower buds (Fig. 3C). EST clones responsible for metabolism, cell growth, signal transduction and transcription were present mainly in the wild type (Tab. 5), indicating the biased transcript expression due to either epigenetic changes or virus infection. We observed 80 clones (28.1%) of the ESTs, from peloric flower buds, belonging to the transcripts of the orchid virus, Cymbidium mosaic virus (CyMV). Almost the whole genome of the CyMV, including RNA replicase, RNA-dependent RNA polymerase, movement protein, triple gene blocks, and coat protein, was preferentially enriched in the peloric flower buds after SSH (Tab. 6). In order to overcome the surveillance of host cells by posttranscriptional gene silencing, tombusviruses may preferentially express gene products such as p19 protein to serve as a silencing suppressor so that virus population can be established in host cells 38. The enriched virus in orchid tissues may also adopt similar strategy as tombusviruses to overcome the surveillance of host cells.

Table 5 Annotation of cDNA clones enriched in wildtype flower buds from SSH flower-flower analysis
Full size table
Table 6 Annotation of cDNA clones enriched in peloric flower buds from SSH flower-flower analysis
Full size table

When the peloric flower buds were used as driver, and leaves of the same mutant as tester, 64 EST clones were enriched in the flower buds (Fig. 3D). Again, genes with unknown function and metabolism (such as tocopherol cyclase, polyamine oxidase, arginine decarboxylase, and pectinesterase, Tab. 7) were present in majority (37.5 and 25%, respectively). Two clones (D20, NP-31) containing TCP domain, which may play a role in floral symmetry 39, 40, 41, were detected in the peloric flower buds from the flower-to-leaves SSH (Tab. 7), as well as in the wild type from the flower-to-flower SSH (NP-31, Tab. 5), respectively.

Table 7 Annotation of cDNA clones enriched in peloric flower buds from SSH flower-leaves analysis
Full size table

Some other ESTs, involved in DNA methylation, chromatin remodeling and post-transcriptional regulation, such as DNA methyltransferase, histone acetyltransferase, ERECTA, and DEAD/DEAH RNA helicase, were preferentially up-expressed in wild type flower buds (Tab. 2, 5) in the wild type-flower/mutant-flower SSH. Thus the ESTs involved in epigenetic regulation are down regulated in the mutants, suggesting abnormal gene silencing in the mutants. Multiple copies of cDNA clones were observed in the SSH experiments, such as unspecific monooxygenase, proteasome subunit alpha type 7, cytochrome P450 like-TBP, RNA dependent RNA polymerase (CyMV), triple gene blocks, etc (Tab. 5, 6). Whether they belong to the real differentially expressed clones or due to background interference remains to be elucidated. Elimination of false positive clones during SSH has been reported by using a Mirror Orientation Selection 42.

Expression levels of selected ESTs in wild type and mutant flowers

In order to confirm the differential expression of the cloned ESTs from both cDNA-RAPD and SSH, real-time RT-PCR using SYBR Green as fluorescence source was conducted to analyze selected clones to check their relative expression level in either wild type or peloric or semi-peloric flower buds by comparing to the housekeeping gene. These clones were selected due to their probable role in cell division, hormone induction, and flower development. First, standard curves of the housekeeping gene, actin, and then selected target genes were generated with the ABI Prism 7900 detection system. Coefficients of determination (R2) for the standard curves were more than 0.96 for most ESTs examined, except for the clones ZOPB10M-60-2(1) and LOPW06N-11-3(19), but all above 0.82–0.91 (Fig. 4).

Figure 4
figure 4

Standard curves of selected target genes and the housekeeping Actin for real time RT-PCR analysis. The target nucleic acid concentration was plotted against the Ct value. All values were means of triplicate measurements. (A) LOPW02N-3-2(3); (B) LOPW06N-11-3(19); (C) ZOPAA01M-2(72); (D) LOPW07N-13(29); (E) 607FD12; (F) LOPW02M-4-1(7); (G) ZOPB10M-60-2(1); (H) ZOPAA12M-24(65); (I) D20 (From SSH experiment); (J) Actin.

Full size image

Expression level of the selected clones was shown in Fig. 5, except of one WD-40 repeat clone OPT05M-50 (Tab. 3) which did not show significant difference between wild type and mutants (data not shown). The clone LOPW02N-3-2(3), encoding a TGA1a-like protein belonged to the bZIP family, was 3-fold and 1.4-fold up-regulated in the wild type as compared to the peloric mutant and semi-peloric mutant respectively (Fig. 5A). The clone LOPW06N-11-3(19) was slightly up-regulated in the peloric and semi-peloric mutants than the wild type (Fig. 5B). The Clone LOPW07N-13(29) was down-regulated in the semi-peloric flower buds as compared to the wild type and the peloric (Fig. 5D). Two clones, ZOPAA01M-2(72) and 607FD12, were highly up-regulated in the prloric mutants. The ZOPAA01M-2(72) clone is a CyMV RNA-dependent RNA polymerase, which was expressed only at basal level in the wild type and semi-peloric flower buds, but highly in the peloric mutants (Fig. 5C). The other highly expressed clone 607FD12 is a retroelement, which was also 6-fold expressed in the peloric mutant than the wild type (Fig. 5E). An unknown protein clone LOPW02M-4-1(7) was down-regulated in the semi-peloric flower buds, and about 1.6-fold expressed in the peloric flower buds and 1.3-fold in the wild type (Fig. 5F). ZOPB10M-60-2(1) obtained from cDNA-RAPD represents an auxin-regulated dual specificity cytosolic kinase. It was highly expressed in the peloric mutants, with 3-fold higher than the wild type and about 1.8 fold-in semi-peloric mutants (Fig. 5G). A serine/threonine protein kinase gene APK2a was reported to be negatively regulated by the AGAMOUS protein in flower development 43. The Arabidopsis receptor-like kinase ERECTA also plays a role in inflorescence architecture 44. Finally, clone ZOPAA12M-24(65) from cDNA-RAPD, encoding a cyclophilin-like protein, was slightly up-regulated in the peloric mutants (Fig. 5H). One clone isolated from SSH experiment with 1.6-fold increase in transcripts in the peloric mutants has sequence similarity to a transcriptional factor PCF6 and a TCP family member (Fig. 5I).

Figure 5
figure 5

Real-time RT-PCR analysis for expression levels of selected target genes of normal (wild type) and mutant flower buds of Phalaenopsis orchids from both cDNA-RAPD and suppression subtractive hybridization. (A) LOPW02N-3-2(3), encoding a TGA1a-like protein; (B) LOPW06N-11-3(19), encoding a protein kinase; (C) ZOPAA01M-2(72), encoding CyMV RNA-dependent RNA polymerase; (D) LOPW07N-13(29), encoding a praline iminopeptidase; (E) 607FD12, a retroelement; (F) LOPW02M-4-1(7), encoding an unknown protein; (G) ZOPB10M-6-2(1), encoding auxin-regulated dual specificity cytosolic kinase; (H) ZOPAA12M-24(65), encoding cyclophilin-like protein; (I) D20, encoding a transcriptional factor PCF6 or TCP family protein.

Full size image

DISCUSSION

SSH has been reported as a tool to compare differentially expressed genes during development and in responses to stresses, pests and somaclonal variations 31, 45. Since many cultivated orchids tended to be infected by different viruses, such as the CyMV and ORSV 46), the abundant virus transcripts in the plant cells and tissues might interfere with the SSH or the cDNA-AFLP analysis 47. In this study, although limited numbers of differentially expressed ESTs were obtained, several interesting clones derived from both suppression subtractive hybridization and cDNA-RAPD analyses revealed potential roles in flower development. A big portion of the subtracted clones belonged to the orchid virus, although it is not clear if the virus infection may cause the phenotypic change. There are clones differentially expressed in the mutant flower buds and one of the most interesting ones is D20, which encodes a potential TCP domain protein. The study on bilateral symmetry in snapdragon flower showed that the radial symmetry was caused by mutations in cycloidea and dichotoma genes 48, 49. Both gene products belong to the members of the TCP family proteins that control plant development from different aspects 39, 40, 41. We have attempted to clone the cycloidea (cyc) homolog in Phalaenopsis orchids by using degenerate primers based on conserved amino acid sequence of cyc homologs, but failed (data not shown). The cloning of TCP homolog, D20 from peloric flower buds by SSH analysis (Tab. 7) enables us to further characterize the biological function of D20 in floral symmetry in Phalaenopsis orchids.

Bulge of the central part of lateral petals in the peloric flowers might be epidermal cell origin. The interaction between transcription factor such as the homeobox genes and TCP homolog and some other protein factors may lead to ectopic cell division in the petals to form the bulge structure 45, 50. One homoebox gene was recently cloned from the phalaenopsis inflorescence by using degenerate primers based on multiple nucleotide sequence alignment. The homeobox shows homology to the ovule-specific homeobox gene 51 and a rice GL2-type homeobox gene Roc1 specifically expressed in the protoderm or epidermal layer of rice embryo 52. Stimuli of hormones, such as the auxin-regulated dual specificity cytosolic kinase may induce unusual cell division in the bulge regions of the peloric petals as shown in Fig. 5G. The unequal distribution of auxin may lead to unusual cell division in certain part of a leaf tissue 53, which may imply similar situation in phalaenopsis flowers. The leaf of the asymmetric leaves 1 mutant of Arabidopsis was reverted by applying auxin onto the leaves, an indication of the role of polar auxin transport in leaf patterning 53. The putative auxin-regulated gene also shares high homology to serine/threonine/tyrosine-specific protein kinases, suggesting a role in the signal transduction pathway during cell division or plant development 54, 55, 56.

In conclusion, we have obtained hundreds of ESTs from the wild type and peloric mutant flower buds of the Phalaenopsis hybrids by using techniques of cDNA-RAPD and SSH. Biased as well as redundant clones towards the wild type or the peloric mutants revealed potential differential transcription regulation. Several analysed ESTs, such as the retroelement, TCP family proteins, auxin-regulated kinase may play a role in the orchid flower development. One question we can ask is whether these up-regulated ESTs due to the epigenetic regulation mechanism, such as DNA hypomethylation? DNA methylation has been implied as a mechanism regulating plant development such as vernalization and somaclonal variation 18, 23, 24, 57, 58, 59, 60. Recently, we have cloned the full length cDNA of a DNA cytosine methyltransferase and a receptor-related protein kinase, ERECTA-like gene from the SSH study (Tab. 5). Their function in orchid flower development will be further characterized through complementation of Arabidopsis mutants and via knockout orchids by genetic approach.