Putative homologue sequences of 1 8S rRNA,
�
-actin and GAPDH for Sophora andClianthus were isolated using the same strategy as for that of the other target genes, and
named STl8S, CM18S, STACT, CMACT, STGAP and CMGAP, respectively (Appendix 3).
Each sequence was confirmed by BLAST searching in the GenBank database using
deduced amino acid sequence. The confirmed housekeeping gene sequences were used
4.4
DiscussionIn the present study, floral identity gene homologues STLFY, STAP 1, STPI and STAG for
Sophora, and CMLFY, CMAP 1, CMPI and CMAG for Clianthus were isolated and
characterized. The combined evidences provided by BLAST search against the GenBank, sequence comparison and phylogenetic analysis strongly suggested that these genes are most probably homologues of LFYIFLO, AP lISQUA, PIIGLO and AGIPLE for
A rabidopsis and Antirrhinum, respectively. Thus, the hypothesis that there would be homologues present of the herbaceous floral identity genes in the reproductive organs of both Sophora and Clianthus was supported.
Although more than one homologue gene can often be found in the same species, such as
LFY homologues AFLl and AFL2 from Malus (Wada et al., 2002), AP 1 homologues EAP 1 and EAP2 from Eucalyptus (Kyozuka et al., 1997), PI homologues RbPI-I and
RbPI-2 from Ranunculus (Kramer et al., 1998) and AG homologues PTAGl and PTAG2
from Populus (Brunner et al., 2000), only one putative gene sequence was identified for each floral identity gene group in the present study. However, it cannot be said with certainty that there exists only a single copy of each floral identity genes for Sophora and
Clianthus, since no Southern hybridization assay was carried out, and only PCR products of appropriate size were sequenced. However, it was beyond the aim of this study to confirm the copy numbers of these genes. The objective to isolate one true homologue gene for LFY and each of the ABC class genes for further gene expression study was fulfilled.
As genetic cloning from New Zealand native species is strictly regulated, all target gene sequences in the present study were determined through direct sequencing of RT-PCR product using tissue specific cDNA as PCR template. Degenerate primers were used for peR amplification and sequencing because there was no previous molecular information for floral genes and housekeeping genes available for these two species, and only a little such information was available for leguminous species and woody species to date. The major disadvantage of using degenerate primers was that more than one PCR product could be produced and all of these products with appropriate size could be sequenced. Furthermore, the sequencing efficiency is much lower when using degenerate primers than specific primers. To overcome these problems, and to efficiently determine the final sequence of the target gene, at least two separate sequencing reactions for both DNA
strands were conducted in the present study.
In order to safely confirm the identity of the newly isolated putative fragments, at least 1 5 reported homologues from a broad taxonomic range of plant species, consisting of dicot, monocot angiosperms and even gymnosperms, were selected and included in the present study for direct sequence comparison and phylogenetic analysis of each separate gene group. Whenever possible, the representative gene sequences of paralogues or genes from other sub-families were included for direct comparison of each putative group of gene homologues. In all cases, the similarities of putative gene sequences, at the amino acid level, were remarkably lower to paralogues or other sub-family gene sequences than to their homologues from species across wide taxonomic ranges.
The isolated putative LFY homologue fragments STLFY and CMLFY showed high sequence similarity to other reported LFYIFLO homologues from a broad range of plant species across angiosperm and gymnosperm. Since these fragments had 93-95% sequence similarity to LFY and FLO, were well placed in the middle of the eudicot LFYIFLO gene group, and there was no sequences from any other gene family found by the BLAST search against the GenBank, it was sound to consider that STLFY and CMLFY are partial LFYlFLO homologues for Sophora and Clianthus, respectively. It is evident that STLFY and CMLFY had the highest homology (95-97%) at the amino acid level to FLOILFY-like protein (BLFL) for Brownea leucantha (Archambault and Bruneau, 2004) and
UNIFOLIATA (UNI) protein for Pisum sativum (Hofer and Ellis, 1998) which was similar to that between STLFY and CMLFY (97%). This was consistent with their taxonomic relationship since they all belong to the family Fabaceae.
It is evident that STAP 1 and CMAP 1 had the highest sequence similarity at the amino acid level (9 1 % and 98%) to PEAM4 (Berbel et al., 200 1 ), the APl homologue from Pisum sativum, supporting a model for AP i -like genes controlling both floral meristem and floral organ identity. Given that STAP 1 and CMAP 1 shared much more sequence similarity to AP 1 and SQUA (79-83%) than to AG and PI (35-42%), it could be safely concluded that STAP 1 and CMAP 1 are partial AP 1 ISQUA homologues for Sophora and
Clianthus, respectively.
The fact that STPI and CMPI had over 70% sequence similarity to PI and GLO, and 91 % and 94% sequence similarity to the functional PI homologue PEAMl from Pisum sativum
Genbank, strongly support the interpretation that they are indeed fragments of the PIIGLO gene homologues for Sophora and Clianthus. Since gene sequences from the PIIGLO paralogues from the other sub-class of B-class genes, AP3 and DEF, were included for direct sequence comparison and phylogenetic analysis, and STPI and CMPI were grouped far away from these paralogues, there was little chance for STPI and CMPI to belong to this paralogue class.
Although relatively low amino acid sequence similarities were observed among AGIPLE homologues, the BLAST search in the GenBank still yielded only AGIPLE homologues. Together with the fact that the sequence similarities of STAG and CMAG to all AGIPLE homologues (61 -80%) was remarkably higher than to APl and PI homologues (29-34%), this evidence strongly indicates that STAG and CMAG are partial homologues of Sophora
and Clianthus.
In comparison between different gene groups or subfamilies, the sequence similarity of the putative genes to their homologues varied considerably. The similarity of STLFY and CMLFY to LFYIFLO were the highest and those of STAG and CMAG to AGIPLE were the lowest, being 93-95% and 7 1 -72%, respectively. This pattern of variation was also the case between other homologues from each gene group. For instance, the similarity
between LFY and FLO, AGl and SQUA, PI and GLO, AG and PLE, were 9 1 %, 77%, 73% and 64%, respectively. This suggests that gene sequences for LFYIFLO homologues were the most conserved and those for AGIPLE were the least conserved, at least within the region covered by the putative gene sequences. However, this might not reflect the actual variation among these gene groups at the whole gene level since the putative sequences covered different regions of the genes, which are more or less conserved in general. Results from the present study also revealed that the structures (exon-intron boundaries, exon numbet and size) of the floral identity genes were very conserved across a broad taxonomic range. For instance, even though the amino acid sequences of STAG and
CMAG differed considerably from that of the Populus AG homologue PTAGl (Brunner et al., 2000), with sequence similarities of only 76% and 75%, the position and size of all comparable exons were identical or very similar among these species. The same was true for STPI and CMPI to the Malus PI homologue MdPI (Yao et al., 200 1 ) . This result is consistent with the fact that the MADS box gene structure is conserved across angiosperm and gymnosperm species (Huang et al., 1 995; Montag et al., 1 995 ; Sundstroem et al.,
1 999). This conservation of gene structures has been observed even in MADS box genes from non-flowering plant species. In comparing the exon-intron structures of the genomic
loci of a moss (Physcomitrella patens) MADS box genes with that of SQUA from the
model plant Antirrhinum majus, Henschel et al. (2002) observed highly conserved or even
identical exon-intron structures among all tested entries, and stated that this conservation in some regions might exist for the whole MADS box gene family. However, future work
is required to definitely determine the intron I exon structure of the putative gene
fragments isolated in the current study. This could be achieved by comparing the genomic sequences with the cDNA sequences.
In the present study, two different nucleotide signals presented at the same nucleotide site in several cases. This might be caused by either site mutation at different tissue cells or by allelic variation and the heterozygous nature of the individual plant. To solve this problem, separate samples were taken from several plants, and the higher probability of nucleotide w as used for the final sequence.
It was noticed, during the multiple sequence comparison and phylogenetic analysis process in the present study, that the sequence similarity between two entry genes would be notably different if the length of the two sequences were different or covered different regions of the homologue genes. This could also affect their positions in the phylogenetic tree. Therefore, the result of phylogenetic analysis would presumably be somewhat biased if the reported sequences were included in the analysis without considering their
differences in length and covering regions of the genes in question. For this reason, all sequences, either isolated in the present study or obtained from the GenBank database, were cut to the same length and covered the same region of the homologue gene before being used for sequence comparison and phylogenetic analysis. In the case when a sequence in the GenBank had particular significance but was shorter than the putative gene sequences isolated in the present study, it was included in the analysis for reference
only and was not used for direct comparison. This was especially the case of MEL and
MESAP 1 for Metrosideros excelsa (Sreekantan et al., 2004) and of ALF and AAP 1 for
Actinidia deliciosa (Walton et al., 200 1 ) . Also for the same reason, the results of sequence similarities and phylogenetic position obtained in the present study for a specific gene could be different from that in other studies.
It also provided useful information for molecular studies of other leguminous species, especially woody legumes. However, further investigations toward their functions and expression patterns in different organ tissues and across the different developmental stages should be carried out to confirm unequivocally whether they have the same roles as their homologues in Arabidopsis and other plant species. While function studies of these homologues were beyond the reach of the current project, studies of the detailed
expression characteristics of each of these genes were subsequently conducted and are presented in Chapter 5 .
Chapter
5
Temporal and developmental expression of floral identity genes inClianthus
andSophora
using real-timeRT-PCR
5.1 Introduction
Genes controlling the specification of floral organ identity and development have been isolated and characterized in a variety of plant species. Intensive studies and analyses of identity mutations in Arabidopsis and Antirrhinum have led to the ABC model of floral organ identity specification (Bowman et aI., 199 1 ; Coen and Meyerowitz, 199 1 ; Weigel and Meyerowitz, 1 994). Although analyses of genes encoding the ABC functions have largely substantiated the applicability of this model, a number of refinements have been made (Zik and Irish, 2003). For instance, homologues of the A function genes AP 1 and
AP2 play a minor role in the specification of sepals and petals in a number of species, such as Antirrhinum, Petunia, maize and apple, although they have a role in floral meristem identity (Chuck et aI., 1 998; Maes et al., 200 1 ; Muller et al., 200 1 ).
Expression of these floral identity genes have also been studied in a number of woody perennial species, including Eucalyptus (Kyozuka et aI., 1 997), Populus (Rottmann et al.,
2000), Actinidia (Walton et al., 200 1 ) , Malus (Wada et al., 2002), Vztis (Carmona et aI.,
2002) and the New Zealand genus Metrosideros (Sreekantan et al., 2004). However, most of these studies focused on the initiation of the floral meristem and floral organs. Little is known about the detailed expression profiles of floral identity genes during different developmental stages, and in different floral organs. Study of the expression profile of all three classes of the floral organ identity genes and their upstream controller genes for the same species and at the same time has not been reported to date, even for model
herbaceous species.
Partial homologues of all three classes (A, B and C) of Arabidopsis floral identity genes,
APETALA1 , PISTALLATA and AGAMOUS, and their up-stream controller gene LEAFY
were successfully isolated from Sophora tetraptera and Clianthus maximus (Chapter 4). These partial homologues were named STAP1, STPI, STAG and STLFYfor Sophora, and
CMAP 1, CMPI, CMAG and CMLFY, respectively, for Clianthus. The high sequence similarity of these genes to their Arabidopsis homologues lent support to the hypothesis
that the Arabidopsis based floral gene model would be applicable to these two woody perennial leguminous species. To test this hypothesis, real-time RT-PCR was used to quantify the temporal and developmental expression of these genes .
Real-time quantitative RT-PCR i s an extremely powerful and preferred method for gene expression studies, which can generate reliable, reproducible, and biologically meaningful results (Bustin and Nolan, 2004). The method is based on the detection of a fluorescent signal produced and monitored during the amplification process and can quantitate the initial amount of template in a dynamic range of up to 107 -fold without post-PCR processing of PCR products (Dorak, 2004). The incorporation of S YBR Green I into real-time RT -PCR makes this technique a relatively economical method for the quantification of gene expression (Ramos-Payan et al., 2003). The non-specific DNA detecting feature of SYBR Green I can be overcome by analyzing the melting-curve of PCR product and/or by using the intron spanning primer designing strategy. Accurate normalization of experimental data is an absolute prerequisite for correct measurement of gene expression using real-time RT-PCR. The most commonly used normalization
strategy involves standardization to a constitutively expressed housekeeping gene as the internal control (Aerts et al., 2004; Andersen et al., 2004; Dheda et al. , 2004). However, such housekeeping genes are usually used without a thorough investigation of how invariant their rnRNA levels are under the experimental conditions being investigated. If housekeeping genes are to be used, they must be validated for the specific experiment (Bustin and Nolan, 2004)
In this chapter, the detailed spatial (developmental) and temporal expression profiles of
LEAFY, APETAlAl, PISTILLATA and AGAMOUS homologues in Sophora and Clianthus
were investigated, using real-time PCR in combination with the SYBR Green I detection system, and multiple housekeeping genes.
5.2 Materials and methods