Separation of the genera in the subtribe Cassiinae (Leguminosae: Caesalpinioidae) using molecular markers

(Separation of the genera in the subtribe Cassiinae (Leguminosae: Caesalpinioidae) using molecular markers). Random amplified polymorphic DNA (RAPD), Inter simple sequence repeat (ISSR) and Amplified fragment length polymorphism (AFLP) markers were used to verify the segregation of the genus Cassia L. senso lato into three distinct genera namely Chamaecrista Moench., Senna P. Mill. and Cassia L. sensostricto Eighteen representatives of the three taxa were characterized using the molecular markers. 25 RAPD, six ISSR primers and six AFLP primer combinations resulted in the amplification of 612, 115 and 622 bands ( loci ) respectively. Most of the loci are found to be polymorphic, showing high degrees of genetic diversity among the different taxa studied. The dendrogram constructed on the basis of the RAPD, ISSR and AFLP data using SHAN clustering, divided Cassia L. senso lato. into three different clusters as Chamaecrista Moench. Senna P. Mill. and Cassia L. senso stricto High bootstrap value revealed that all the clusters were stable and robust. It was observed from the present investigation that these genera have their identity at molecular level, which supports the elevation of the genus Cassia L. senso lato to the level of subtribe Cassiinae and segregation into three distinct genera instead of intrageneric categories.


Introduction
Cassia L. senso lato (Leguminosae: Caesalpinioideae) is one of the twenty-fi ve largest genera of dicotyledonous plants in the world. Th e genus extends in all terrestrial habitats from the equator to the edges of dry and cold deserts, but much of its diversity is centred in areas of varied topography with seasonal climates. There has been considerable divergence of opinion concerning the delimitation and taxonomic status of its three constituent subgenera. Irwin & Barneby (1981, 1982 proposed an improved classifi cation and raised the genus Cassia L. to the level of subtribe Cassiinae; the latter comprised of three genera, viz Cassia L. senso stricto, Senna P. Mill. and Chamaecrista Moench. Th is concept had found wide recognition in recent years (Randel 1988(Randel , 1989(Randel ,1990Lock 1988, 89;Larsen & Hou 1996).
Despite several studies by taxonomists, either on the whole subfamily Caesalpinioideae in restricted areas or of certain genera throughout the world, there is still a great deal of taxonomic work to be done at the level of genus and tribe. Th e taxonomic treatment of the genus Cassia L. senso lat. has been done in some countries in Asia, namely Malaysia (De Wit 1955) and Pakistan (Ali & Quraishi 1967). But nowadays taxonomy is not simply based on fl oral and vegetative characters but many other types of data, like anatomical, cytological, serological and molecular. Th ese are regarded as very important in determining relationships and affi nities of the plants. New trends and new information on various aspects had prompted modern taxonomists to propose a new system of classifi cation. Th ese have also led to some changes in the criteria for determining the evolutionary status of the various taxa in the subtribe Cassinae. Barneby (1981, 1982) raised the genus Cassia L. sensu lato to the level of subtribe and elevated previous subgenera to generic rank viz. Senna Mill and Chamaecrista Moench under the tribe Cassieae Bronn ex Irwin and Barneby of Caesalpiniaceae (today Leguminosae-Caesapinioideae). However there is little work done on the molecular phylogeny of this genus and conformation of the trifurcation of this group. Th ere are very few reports on other aspects like biochemical and cytological on this taxa. Th e reports that are present contradict each other in the cytological data (Singh 2001).
Keeping all these facts the present investigation was carried out with the objectives to derive an authentic relationship among the studied genera of the subtribe Cassiinae. Justification of trifurcation of the subtribe into three distinct categories like Cassia, Senna and Chamaecrista as suggested by Irwin and Barneby (1982), taking the molecular data obtained from RAPD, ISSR and AFLP analyses.

Plant material
All plants were raised from seed. Eighteen species of the genus Cassia were taken for the present study. Th eir taxonomic rank and changed names are given in Tab.1. Th ey were grown in the nursery of Regional Plant Resource Centre, Bhubaneswar, Orissa, India. Very young and tender leaves were taken for genomic DNA isolation. Vouchers were deposited at Herbarium of Regional Plant Resource Centre.

Molecular Analysis
DNA was isolated from young, fresh leaves using the CTAB method as described by Saghai-Maroof et al. (1984). For RAPD analysis, PCR amplifi cation of 25ng of genomic DNA was carried out using standard 30 decamer oligonucleotide primers, out of which in 25 primers reproducible amplifi cation was found. So they are taken  for the present RAPD analysis and those are OPA02,  OPA03, OPA04, OPA10, OPA16, OPA18, AF14, OPC02,  OPC05, OPD02, OPD03, OPD07, OPD08, OPD18, OPD20,  OPN02, OPN04, OPN05, OPN06, OPN08, OPN10, OPN11, OPN12, OPN16 and OPN18 (Operon Tech. Alameda, CA. USA). The RAPD analysis was performed as per the standard methods of Williams et al. (1990). PCR products were separated on a 1.5% agarose gel containing ethidium bromide solution (@0.5μg/ml of gel solution). The size of the amplicons was determined using size standards (100bp ladder plus or DNA ladder mix, MBI Fermentas, Lithuania). DNA fragments were visualized under UV light, documented in Gel Doc (Bio-Rad, USA) and photographed. Inter-simple sequence repeats have recently been developed which access the variation in the numerous microsatellite regions distributed throughout different genomes (basically the nuclear genome) and bypass the challenges of characterizing individual loci that other molecular techniques require. Th e PCR products were separated in polyacrylamide gel. Amplifi ed fragment length polymorphism analysis was done as per the standard protocol of Vos et al. (1995) and the protocol supplied by the manufacturer (Invitrogen, USA). All the reagents and chemicals were procured from Invitrogen (Invitrogen life technology, CA, USA). Aft er the completion of the gel run it was stained with 0.0002% of ethidium bromide and destained in distilled water. Gel was documented in a gel doc (Gel Doc 2000, Bio Rad, USA).

Data analysis
The bands amplified from RAPD, ISSR and AFLP were scored as '1' and '0' for presence and absence of band respectively. All the bands whether monomorphic or polymorphic were used for similarity calculation in order to avoid over estimation of distance (Gherardi et al. 1998). Jaccard's coeffi cient of similarity (Jaccard 1908) was calculated and a dendrogram based on similarity coeffi cient was obtained through unweighted pair group method using arithmetic averages (UPGMA) (Sneath & Sokal 1973) and SHAN clustering. All the analysis was done using the computer package NTSYS-PC-2.02e (Rohlf 1997). Resolving power (Rp) of the RAPD, ISSR and AFLP was calculated as per Prevost & Wilkinson (1999). Resolving power is: Rp=ΣIB (IB (Band informative ness)= 1-[2×(0.5-P)], P is the proportion of the 18 species containing the band. Primer Index (PI) was calculated from the polymorphic index. A polymorphic index (PIC) was calculated as PIC = 1-×P 2 i , P i is the band frequency of the ith allele (Smith et al. 1997). In the case of RAPDs, ISSRs and AFLP the PIC was considered to be 1-p 2 -q 2 , where p is band frequency and q is no band frequency (Ghislain et al. 1999 calculate the RAPD primer index (RPI). RPI is the sum of the PIC of all the markers amplifi ed by the same primer. Principal coordinate analysis (PCA) was used to retrieve information on the clustering pattern of the analyzed populations. PCA was performed based on the RAPD, ISSR and AFLP data, for all the primers.

Random Amplifi ed Polymorphic DNA
All the 18 species produced distinct reproducible amplifi cations. Th e DNA profi les as observed in RAPD are represented in (Fig. 1). A total of 612 numbers of bands were amplifi ed (summary in Tab.2). All the bands were found to be polymorphic. Maximum and minimum amplifi cation was observed in the case of the species Chamaecrista absus L. (139) and Cassia spectabilis L. (57) respectively. Th e highest numbers of bands were amplifi ed with primer OPA02 and OPAO3 (36 each) and lowest numbers of amplifi cation were observed with the primers OPC02 and OPN11 (17 loci each). Among the polymorphic bands 124 were exclusive bands, specifi c for a single species. Highest number of exclusive bands were observed for the primer OP N10 (10) while in the case of OP D02 and OP D20 only two bands were found to be exclusive. In Chamaecrista there were highest unique bands (20), only two bands were found to be unique in the case of Ch. pumila as well as for C. grandis. Th e dendrogram constructed using SHAN clustering shows three major groups among the 18 studied taxa (fi gure not shown).

Inter Simple Sequence Analysis
Six ISSR primers resulted in the amplification of 115 fragments. The ISSR banding pattern in Cassiinae is represented in Fig. 1. The primer (CAA)5 and (GACA)4 produced maximum number of bands (21), while with the primer(GA)9T only 14 loci could be amplified. Of these, only 8 were found to be unique bands. The bands were amplified in the range of 200-3000 base pairs. All the loci amplified with the ISSR primers were polymorphic and prominent. Maximum number of unique bands were observed with primer T (GA)9 (4), while there was no unique band with the use of primer (AGG)6, (GTG)5 and (CAA)5. Among these ISSR primers, maximum resolving power was obtained for (AGG)6 (12) and the minimum Rp was for (CAA) (5.78). Maximum primer index was calculated for (GACA)4 and minimum for (CAA)5. Highest numbers of 36 bands were resolved for the species C. javanica and the least number of amplicons were amplified for the species Ch. mimosoides. Details of ISSR banding pattern and bands amplified in different species with different primers have been represented in Tab. 3.

AFLP Analysis
Six AFLP primer combinations were used for the present work. This primer combination had amplified 622 loci among which five loci were found to be monomorphic in nature and the rest were polymorhic. Banding pattern in AFLP for 18 species of Cassiinae is shown in Fig. 1. From the 594 polymorphic bands amplified, only 28 were found to be unique that to in a single species. Maximum numbers of 139 bands were resolved for the primer combination EACA/MCAG and the minimum for EACCA/MCAT (62). The average numbers of bands amplified per primer combination was calculated to be as high as 103.6. The highest numbers of bands (65) were noted for C. javanica var. indochinensis for primer combination EACT/MCAG and the lowest for the primer combination EACC/MCTC (19) in the case of the species C. javanica. Bands resolved between 1500bp to 20 bp were taken into consideration for the present investigation. Maximum numbers of 3 monomorphic bands were scored for the primer combination EACC/ MCAT, whereas no band was found to be common to all the species with the primer combinations EACC/MCTC, EACT/MCAG, EACA/MCAG and EAAG/MCAC. The resolving power was maximum for the primer combination EAAG/MCAG (78.889) and minimum for EACC/MCTC (42.778). However, the primer index was found to be highest for EACA/MCAG (46.944) and lowest for EACC/MCAT (22.895). The details of AFLP analysis are presented in Tab. 4.

Similarity for RAPD, ISSR and AFLP
Jaccard's similarity was calculated from the data generated from RAPD, ISSR and AFLP analysis. It was observed that C. grandis and C. javanica var. indochinensis were most closely related with Jaccard's similarity coeffi cient of 0.473, whereas Ch. mimosoides and S. sulfurea were distantly placed with the lowest Jaccard's similarity coeffi cient of 0.064. Among all the species, there was average similarity of 0.1834. Jaccard's similarity among the taxa is represented in Tab. 5.

Tree generated from the combined markers
On the basis of the data obtained from all the molecular marker systems, a dendrogram was constructed using UPGMA and SHAN clustering in NTSYS-pc 2.02e (Fig. 2). All the 18 taxa of Cassinae studied were separated into three different clusters each containing members of a particular genus. The species of Senna were further divided into two subclusters each containg 5 species. While the first subcluster contained S. tora, S. occidentalis, S. alexandrina, S. siamea and S. pallida, the second had S. sulfurea, S. alata, S. spectabilis,    S. auriculata and S. septemtrionalis and both the subclusters shared a common node at approximately 23% level of similarity. All the species of Chamaecrista had a similarity of nearly 35.5% among themselves and Cassia at 32% similarity level. The tree had a common node at 11% similarity level for the three different genera in the subtribe Cassiinae.

Cophenetic correlation
Cophenetic correlation was calculated for diff erent phenograms generated from three different marker systems. It was found that the maximum cophenetic correlation existed between C. grandis and C. javanica var. indochinensis (0.473). Th e cophenetic correlation for diff erent markers is represented in Tab. 6. Th e average cophenetic correlation between any two species was found to be 0.183392.

Bootstrap analysis
Bootstrap analysis revealed that all the clusters were stable and had bootstrap value of approximately 100. It indicates that the clades containing Cassia javanica and C. fi stula among the members of Cassia, Chamaecrista pumila and Ch. mimosoides in Chamaecrista and Senna tora and S. occidentalis in Senna were closely linked with bootstrap value of 100 or nearly 100. Th e bootstrap values among diff erent clusters and subclusters were presented in the bootstrap tree. (Fig. 2).

Principal coordinate analysis
In the PRINCORD analysis, all the species were separated in three diff erent plains each containing the representatives of a particular genus. All the species of Chamaecrista were grouped together at a corner in the PCA fi gure while all the members of Cassia were grouped in the opposite corner in the fi gure. However, the elements of Senna were found in the middle of the diagram maintaining equal distance from Cassia and Chamaechrista groups. Th e PCA diagram is represented in Fig. 3.

Discussion
The genomic relations among different taxa of Cassiineae were studied on the basis of RAPD, ISSR and AFLP fi ngerprinting. Th e relationship obtained among the different taxa of Cassinnae using RAPD was in agreement with the conventional taxonomic classifi cation of the subtribe. Th e trifurcation of Cassiinae into three distinct genera as suggested by Irwin and Barneby (1981) proved to be justifi ed. Whitty et al. (1994) worked on Cassiinae and justifi ed the grouping of Cassia, Senna and Chamaecrista. We observed similar type of result. However, the intrageneric relationship among species of Senna does not follow the sequence that Irwin and Barneby (1981) worked out. Th e three species represented here of sect. Peiranisia namely, S. pallida, S. auriculata and S. spectabilis formed a cluster. Similarly, S. tora, S. occidentalis, S. septemtrionalis and S. siamea all belonging to sect. Chamaefi stula, came together in a distinct clade. Non-inclusion of S. alexandrina in the above cluster could not be reasoned out. Of the three elements of the genus Chamaecristra, the lone species of sect. Grimaldia i.e. Ch. absus got separated from the other two, which belong to sect. Chamaecrista. Th e infra-generic arrangement of species in Chamaecrista was in agreement with Irwin and Barneby (1981). Th e deviations with regard to intrageneric relationships may be due to selection of a small number of species from such a large taxon for the present investigation and amplifi cation a small portion of the entire genome. Souza & Benko-Iseppon (2004) found signifi cant diff erences in chromosome size, morphology and condensing behavior among members of the controversial tribe Cassieae (Cassia, Chamaecrista and Senna), revealing the tribe to be a heterogeneous group from the karyological point of view.
In contrast, the species relationship in the genus Cassia as observed from RAPD tree was confusing. Th e distant placement of C. javanica and C. roxburghii, both belonging to the series Obolospermae (Irwin & Barneby 1981) and even separation of two varieties of C. javanica could not be suitably explained. Inclusion of more species and resorting  to other molecular techniques in conjunction with RAPD are likely to alter the arrangement of taxa under the genus and bring about the expected order.
The dendrogram obtained from the ISSR data segregated the subtribe into two clusters; the first with the three species of Chamaecrista and the second having all the species of Cassia and Senna. The second clade was further bifurcated to two subclusters; one with all 10 species of Senna and the second subcluster had five species belonging to the genus Cassia. Thus, broadly there was a segregation of the subtribe Cassiinae into three genera. This was in conformity with the classification proposed by Irwin & Barneby (1981). Though the division of the subtribe on the basis of ISSR data appears justified, the clustering and arrangement of species under Senna and Cassia were not at par with the traditional grouping made earlier. As in the case of RAPD, in the genus Chamaecrista, Ch. absus that comes under sect. Grimaldia got separated from the remaining two and Ch. mimosoides and Ch. pumila formed a subcluster justifying their inclusion under sect.

Chamaecrista.
In the Cassia clade, C. javanica var. indochinensis and C. roxburghii, which are closely related came in a single cluster with distantly related C. grandis. But the other variety C. javanica var. javanica remained isolated and formed a clade with C. fi stula. Th e placement of the four species of Cassia under the three series Cassia, Grandes and Obolospermae was not possible. Similarly, from the dendrogram generated from ISSR data, segregation of species of Senna into traditionally recognized sections like Psilorhegma, Peiranisia, Chamaefi stula and Senna also could not be done. Th ough only six ISSR primers were used for the present investigation, the result was comparable with the RAPD.
As both RAPD and ISSR were dominant markers and arbitrarily amplified the loci the need for application of better marker system was felt and AFLP was used. Six AFLP primer combinations were used for deciphering the genetic relationship among the 18 species of Cassiinae and a total of 622 bands were obtained, most of which were polymorphic in nature. A high degree of genetic diversity among the species of Cassiinae was noted. The dendrogram constructed on the basis of the data obtained from the AFLP analysis segregated the subtribe into three distinct groups. As observed in RAPD and ISSR analysis, all members of Senna, Cassia and Chamaecrista formed distinct clusters. The arrangement of species under the genera Chamaecrista and Cassia was exactly similar as determined from RAPD and ISSR analysis and described earlier. However, the species of Senna were clustered in two groups; the first cluster containing S. tora, S. occidentalis, S. alexandrina, S. siamea and S. pallida and the second cluster with the remaining five species. All the five species of the first subcluster shared a common node at 34% similarity level; S. tora and S. occidentalis were the closest. The second subcluster consisted of S. sulfurea, S. alata, S. spectabilis, S. auriculata and S. septemtrionalis and shared a node with the first subcluster at about 27% level of similarity. The species arrangement did not follow any logical pattern and the data obtained was not discernible. Marazzi et al. (2006) studied phylogenetic relationships within Senna based on parsimony analyses of three chloroplast regions (rpS16, rpL16, and matK) and provided new insights on the evolution of floral symmetry and extrafl oral nectaries. Th eir results supported the monophyly of only one sect. Psilorhegma of the six currently recognized sections, while Chamaefi stula, Peiranisia, and Senna were paraphyletic and monotypic Astroites and Paradictyon were nested within two of the seven major clades identifi ed by molecular phylogeny. Th eir investigation further suggested that fl owers in Senna were ancestrally monosymmetric with seven fertile stamens and three adaxial staminodes, switched to asymmetry later, and reverted to monosymmetry in most     advanced clade. Fertility of all 10 stamens was considered to be a derived state, characterizing the Psilorhegma subclade.
The phenogram constructed using combined data from RAPD, ISSR and AFLP analyses exhibited similar relationships among the genera and species. The bootstrap value obtained for different groups was fairly good which indicated that the branching in the tree was stable. When the correlation among all the markers was calculated it was highly encouraging and all markers showed high degree of correlation with each other and with the combined data.
Data obtained from analysis through molecular markers revealed high degree of genetic diversity among the diff erent taxa of Cassiinae. Similar observations were made by other authors who worked on the subtribe taking diff erent morphological markers (Bhattacharya & Maheshwari, 1971;Lasseigne 1979); Malik & Krishna 1978;Shyam et al. 1983;Mathur 1985;Shyam & Vartak, 1985;Bhattacharya & Saha, 1992;Sahai et al. 1997). In the present study taking the molecular markers into account the trifurcation of the subtribe Cassiinae could be re-established but the intrageneric classifi cation and phylogeny in diff erent genera of the subtribe needs to be worked out in detail taking large number of species and using sophisticated molecular marker systems such as SSR and sequence-based markers like cp-DNA, nr DNA and ITS regions.