Molecular systematics of the genera Laurencia , Osmundea and Palisada ( Rhodophyta ) from the Canary Islands-Analysis of rDNA and RUBISCO spacer sequences

Lewis, Sh., Gacesa, P., Gil-Rodríguez, M.C., Valdés, F. & Frías, I. 2008. Molecular systematics of the genera Laurencia, Osmundea and Palisada (Rhodophyta) from the Canary Islands Analysis of rDNA and RUBISCO spacer sequences. Anales Jard. Bot. Madrid 65(1): 97-109. The molecular systematics of Laurencia, Osmundea and Palisada (Rhodomelaceae, Ceramiales) species from the Canary Islands has been determined by analysis sequences of the ribulose,1-5, bisphosphate carboxylase (RUBISCO) spacer from the plastid genome and the internal transcribed spacers (ITS1 and ITS2) and the rDNA 5.8S coding region from the nuclear genome. Comparison of sequence data showed an identity of 72-83 % between the species. Three taxonomic group were identified that correspond to established phylogenetic taxons. Phylogenetic trees using both parsimony and maximum-likelihood methods were derived from the sequence data; the trees indicate that O. pinnatifida appears to be the most distantly related species from the Laurencia and Palisada species. The exact phylogenetic position of Laurencia sp. A (“amarilla”) need additional studies.

Both, the small and large subunits of RUBISCO, are encoded by the red algae plastid genome.The RU-BISCO genes of Rhodophyta, Cryptophyta and Chromophyta are co-transcribed (Zetsche & al.,1991) and are separated by a small non-coding spacer region (Fig. 1).
In addition to RUBISCO, the comparison of DNA sequences of rDNA, has proved to be a useful systematic tool.rDNA consists of genes that encode for the large (28S) and small ribosomal sub-unit RNA (18S and 5.8S) plus transcribed and non-transcribed spacer regions.The intergenic spacer region (non-coding) has evolved most rapidly while the coding regions are the most evolutionary conserved sequences of the cistron (Druehl & Saunders, 1992).Interspecific and intraspecific sequence variation of these regions has been examined in plants (Baldwin, 1992), fungi (O'Donnell, 1992), diatoms (Zechman & al., 1994), green algae (Coleman & al., 1994;Bakker & al., 1995), ahermatypic corals (Beauchamp & Powers, 1996) and red algae (Steane & al., 1991).rDNA has also been utilised for systematic and taxonomic studies of algae (Goff & al., 1994;Rumpf & al., 1996;Stache-Grain & al., 1997).Goff & al. (1994) and Stache-Grain & al. (1997) examined the value of both the ITS and RUBISCO spacer sequences in delineating relationships of populations, species and genera in Gracilaria Greville, Gracilariopsis Dawson, Ectocarpus Lyngbye and Kuckuckia Hamel.They found that both were highly conserved at species and population levels.In this work we initiate a preliminary molecular analysis of this complex group by means of the amplification and sequencing of the RUBISCO spacer and rDNA internal transcribed spacers (ITS1 and ITS2) including the intervening 5.8S coding region of the nuclear genome of six species in the genus Laurencia, Osmundea and Palisada.The results obtained were used to derivate a phylogenetic tree that indicate the molecular taxonomy between species.

Materials and Methods
Treatment of algal material and nuclear DNA extraction: Individual plants were collected from various locations (Table 1) in Tenerife and La Palma (Canary Islands) and New Zealand.A voucher specimen from each species from the Canary Islands used for sequencing and amplification were deposited in the La Laguna University Herbarium TFC.Samples of Laurencia thyrsifera J. Agardh from New Zealand were provided by Dr Ruth Falshaw (Industrial Research Limited, New Zealand).Only algae that were visibly clear of epiphytes were used for DNA extraction.DNA was extracted from the apical parts of the thallus, frozen by immersion in liquid nitrogen and stored at -20 ºC.
For extraction and amplification of plastic DNA fresh/frozen thallus (5 g) was suspended in 15 ml of 100 mM Tris-HCl (pH 8.0) containing 50 mM EDTA (pH 8.0) and 0.5 M NaCl to which 10% (w/v) SDS was added.Ribonuclease A (0.15 mg) and Proteinase K (1.5 mg) were added to the sample and the tube incubated at 37 o C for 1.5 h.DNA was isolated by phenol and chloroform / isoamylalcohol (24:1 v/v) precipitation and extracted in cold absolute alcohol.Plastid, nuclear and mitochondrial DNA was isolated by differential centrifugation in caesium chloride / Hoechst (33258) dye density gradient at 174,000 × g for 16 h.RUBISCO DNA Amplification: The RUBISCO spacer was amplified using forward primer: 5' TGTG-GACCTCTACAAACAGC and reverse primer: 5' CC-CATAGTTCCCAAT.The reactions were performed after initial denaturation at 95 o C for 10 min.The samples were then cycled sequentially using a temperature regime of rDNA Amplification: Target regions of rDNA were amplified using forward primer: 5' GTTTCCGTAG-GTGAACCTGC and reverse primer: 5' ATATGCT-TAAGTTCAGCGGGT.The reactions were performed after initial denaturation at 95 o C for 10 min.The samples were then cycled sequentially using a temperature regime of 94 o C/1.25min, 60 o C/2min and 72 o C/4min repeated for 30 cycles followed by a final period of 72 o C/10 min.
Cloning of purified PCR products: Purified PCR products (Qiagen gel extraction kit) were ligated using the pGEM-T vector system (Promega) according to the company's instructions.These ligated products were transformed either by a heat shock method into E. coli XL1 or electrophorated using electro-competent E. coli JS5 high efficiency cells (Bio-Rad).
Sequencing of plasmid DNA: Chain-termination sequencing of purified PCR products was performed using the USB Sequenase R* Version 2.0 kit according to the company's protocol.Some clones were sequenced using forward and reverse M13 primers labeled with fluorescein (Pharmacia LKB), using a Pharmacia LKB Automated Laser Fluorescent (ALF) DNA sequencer and others sequenced using a Hybaid LI-COR 4000LS infra-red automated sequencer.
Analysis of sequence data: Sequence data were analyzed using the UWGCG (University of Wisconsin Genetics Computer Group, Version 7.1 UNIX, Devereux & al. 1984) package at Daresbury.Alignments of sequences were produced using GAP, which uses the algorithm of Needleman.The default parameters used were "Gap weight" (penalty for introducing a new gap): 1.0 and "Gap length" weight (maximum length of internal gaps): 0.1.Phylogenies were recovered using PHYLIP package (version 3.5c) (Felsenstein, 1993).The branch and bound algorithm (DNAPENNY) was used to find the most parsimonious tree and a phylogeny was also recovered by maximum-likelihood (DNAML).Bootstrap values were calculated for each tree using additional routines in the package (SEQBOOT, CONSENSE).
DNA sequences showed in this work are deposited in the GenBank with the following accession num- bers: AF081268-AF081272 and AF082340-AF082344.

Results
Amplification of the RUBISCO spacer using primers conserved for the 3' end of the rbcL and the 5' end of the rbcS genes (Fig. 1) produced products ranging in size between 350-375 bp depending on the Laurencia, Osmundea and Palisada species.The sequence data revealed a spacer region of 93-112 bp in all of the Laurencia, Osmundea and Palisada samples analyzed (Table 2).
Although the spacer is a non-coding region there is quite a high percentage identity amongst the Laurencia Osmundea and Palisada samples (72-83%) and L. viridis and L. thyrsifera have 95% identity.Additional comparison with spacer sequence from the far related genera: Gracilariopsis lemaneiformis (Bory de Saint-Vincent) Dawson, Acleto & Foldvik and Antithamnion sp., in order to compare the Laurencia, Osmundea and Palisada differences, showns an equivalent 62-73% of identity.Multiple alignments of these sequences were produced using CLUSTALV (Fig. 2).
There are no insertions / deletions in the coding regions of rbcL and rbcS but there are point mutations with approximately equal numbers of transitions and transversions.Of the 379 alignable positions (Fig. 2) 94 were possible phylogenetically important locations and over half of these were in the non-coding spacer region (i.e.55/118 positions = 46.6%informative variability in this region); nine informative positions occurred in the rbcL (9/105 nucleotides = 8.6% variability), and 31 occurred in the rbcS region (31/156 nucleotides = 19.9%variability).
The length of the ITS regions and 5.8S rDNA is relatively well conserved in the Laurencia species.ITS1 is of similar size in all the species studied (160-170 bp) with the exception of O. pinnatifida (211 bp).The latter species produced the largest amplification product and the increase in size resulted from insertions into ITS1 (Fig. 1).There is less variation in the length of the ITS2 regions than ITS1 of Laurencia (202-222 bp) and ITS2 is larger than ITS1 in all of the Laurencia and Osmundea species examined.The percentage identity of the sequence data obtained from each region of these amplified products was calculated using GAP (UWGCG, Daresbury) (Table 3).
There is a high percentage of identity within the species in all three regions.The identity in the ITS1 is the lowest in all the species with values ranging from 70.3-97%, while the identity in the ITS2 is slightly higher at 74.1-98.2%.The 5.8S has the highest percentage identity amongst the Laurencia and Osmundea where L. viridis has a 100% identity to L. sp.A ("amarilla").Sequence data of the 5.8S coding region of the Laurencia and Osmundea species were compared with each other and with the 5.8S data of Gracilariopsis lemaneiformis (Bory de Saint-Vincent) Dawson, Acleto & Foldvik which was used as an outgroup (data kindly supplied by Dr. Lynda Goff, UCSC), and the identity calculated using GAP (Table 4).The multiple alignments of these sequences as produced by CLUSTALV are shown in Figure 2.  Phylogenetic analysis of the data from RUBISCO and rDNA were carried out using both parsimony and maximum-likelihood methods.The phylogenetic trees produced for the different regions using both methods were similar and indicated that O. pinnatifida was the most distantly related of the Laurencia and Osmundea species (Fig. 3a).The phylogenetic trees from the sequence data for the ITSs and 5.8S secuences, and using G. lemaneiformis as the outgroup shown similar results and indicated that O. pinnatifida was the most distantly related of the Laurencia and Osmundea species (Fig. 3b).

Discussion
Amplification of the RUBISCO spacer and its flanking coding regions from the Laurencia, Osmundea and Palisada species studied produced a sin-Sh.Lewis & al. gle product of approximately 350-375 bp in length.The non-coding spacer region in these species ranged between 93-112 bp, which are within the range reported for RUBISCO spacers from other algae (Destombe & Douglas, 1991;Goff & al., 1994).In addition, amplification of the ITS1, 5.8S and ITS2 regions in the nuclear genome gave consistently reproducible results similar to those found for other red algae (680-1450 bp) as examined by Steane & al. (1991).In that study, DNA from the Australian alga Laurencia filiformis (C.Agardh) Montagne was amplified and the product was similar in size to the Laurencia and Osmundea species studied in this work.Even though the size differences of the PCR products within the Laurencia and Osmundea species studied here were quite small, (~ 5-50 bp), this was sufficient to distinguish between them.
The comparison of the sequence data obtained shows a relatively high percentage identity of the spacer region between the Laurencia and Osmundea species (72 -83 %) and across other genera (62.1-73.2%) even though it is a non-coding region.Laurencia viridis and L. thyrsifera have very different geographical locations (Tenerife and New Zealand respectively), but still have a high sequence similarity for this spacer (95.4 %).The two species are morphologically very similar with the exception that L. viridis is green in color and L. thyrsifera reddish/purple.This high degree of conservation of primary sequence of the RUBISCO spacer has been noted in other algae (Valentin & Zetche, 1990;Destombe & Douglas, 1991;Goff & al., 1994;Stache-Grain & al., 1997) and it has been hypothesized that this conservation of sequence is as a result of the secondary structures in the plastid spacer perhaps having a role in tRNA maturation (Destombe & Douglas, 1991).The AT content  for the spacer in the Laurencia, Osmundea and Palisada species ranges between 76-80%, which is similar to other algal species as Cryptomonas sp., Olisthodiscus luteus and Pylaeilla littoralis (Goff & al., 1994;Douglas & Durnford, 1989;Delaney & Cattolico, 1989;Assali & al., 1991).Towards the 3' end of the spacer in all the Laurencia, Osmundea and Palisada species examined, there is a purine rich sequence ( 5' AAGGAG 3' ), corresponding to the ribosome binding site (Shine & Dalgarno, 1974), upstream of the rbcS coding region.This sequence, which has been found in all algae studied so far, contributes to the high conservation of the spacer region (Destombe & Douglas, 1991).The initiation codon of the rbcS in these algae GTG (GUG) is a sequence used at only 3-4% of the frequency of AUG in bacteria (Reddy & al., 1985).The open reading frames (ORF's) of rbcL and rbcS from the sequence data obtained from the algae were compared and there is a high percentage identity of these regions in all the algae with no insertions or deletions of bases (Fig. 4).There are equal numbers of transitions and transversions overall but there are a higher number of substitutions in the rbcS than the rbcL, a situation also observed by Goff & al. (1994).
Analysis of the multiple alignment of the RUBIS-   closely related Gracilariales genera Gracilaria Greville and Gracilariopsis Dawson (Goff & al., 1994).Even though the plastid and nuclear spacers studied in the Laurencia, Osmundea and Palisada species have similar percentage identities (72-95.4% and 70.3-98.7%respectively) the RUBISCO spacer is quite highly conserved across genera, Our results confirm the position of Palisada cf.perforata in the taxon P. perforata.
In addition, Laurencia sp.A ("amarilla") is separated from Osmundea but close to Palisada, these results are different to found in the "classical" morphological approach.Additional analysis is necessary to evaluate the exact taxonomical positions of these groups.
With respect to the nuclear sequences, we found that with the exception of O. pinnatifida, the ITS length ranged between 160 and 170 base pairs for ITS1 while ITS2 was larger by 37-62 base pairs, resulting in lengths between 202 -222 base pairs (Fig. 5).This greater length for ITS2 is seen in other red algae such as species of Gracilariopsis Dawson and Gracilaria Greville (Goff & al., 1994).The comparison of se-    quence data obtained for each species of Laurencia and Osmundea shows a relatively high percentage identity within each of the three ITS/5.8Sregions.Where differences between sequences occurred, 63% of the changes were substitutions.ITS1 was more variable than ITS2, with 52.2% variable positions compared with 42% in ITS2.Furthermore, nucleotide deletions, which accounted for 37% of all differences, were most common in ITS1.Baldwin Phylogenetic of the Laurencia, Osmundea and Palisada spp.
(1992) reported that overall, ITS2 sequences were usually less variable than ITS1 sequences in some members of the Asteraceae, but there is also documentation to the contrary (Zechman & al., 1994;O'Donnell, 1992).Zechman & al. (1994) reported that in Stephanodiscus clones (diatoms), the ITS1 had 7.02% variability including insertions/deletions) while the ITS2 had 8.85% variability; however, this difference is not great considering that these are spacer regions.
The 5.8S coding region has the highest similarity between species which might be expected as the sequence is more likely to be conserved than the noncoding internal transcribed spacers (Table 3).The degree of identity of the 5.8S between the genera Laurencia and Osmundea (91.7-99.3%) is consistent with values for the percentage identity within the genus Gracilaria of 94-98.7%, the lowest identities being between O. pinnatifida and the Laurencia species (Table 4).Hence, the 5.8S is highly conserved within a genus but less so between genera.However, the non-coding  regions displayed a comparatively high degree of similarity between species implying that there may be constraints on these regions as well.
The multiple alignment of the 5.8S sequence data of the four Laurencia species (Fig. 6) shows that the majority of the differences in sequence are towards the 3' end of the coding region.This concurs with results from other genera in which the 3' region of 5.8S is less conserved than the 5' end (Steane & al., 1991, based on Mindell & Honeycut, 1990).It is interesting to note that in Osmundea pinnatifida the 5.8S coding region ends in the sequence TTC rather than GTC which is seen in the Laurencia species, as well as the majority of red algae.
In conclusion, the phylogenetic trees derived from the plastid and nuclear sequences of the Laurencia, Osmundea and Palisada species were different, which would imply that the three genomes are evolving at different rates (Goff & al., 1994).However, all of the trees and sequence data do indicate that O. pinnatifida has the least similarity to the Laurencia species and hence is the most distantly related.Nam & al. (1994) proposed that O. pinnatifida, along with certain other Laurencia species, should be moved from the genus Laurencia into the resurrected genus Osmundea.The data presented here would support this proposal.

Fig. 1 .
Fig. 1.Plastid RUBISCO spacer region in red algae.Amplification primers are represented by the forward and reverse arrows.

Fig. 2 .
Fig. 2.Aligned sequences of the RUBISCO spacer and its flanking regions.Sequences were aligned using the CLUSTALV program.Dots denote the same nucleotide as that of Laurencia sp.A ("amarilla") in row one, a * denotes the same nucleotide in all the species, a x represents unknown nucleotides and a hyphen represents a gap with the exception of L. viridis where for positions 248-375 the sequence is not known.The stop and start codons of rbcL and rbcS respectively are highlighted as are the Shine-Dalgarno sequences.Antithamnion sp. is used as an outgroup and the forward and reverse primers are underlined.

Fig. 3 .
Fig. 3. Phylogenetic tree result from the DNAPENNY analysis of the RUBISCO spacer and its flanking coding regions (a) and of the ITS1, 5.8S and ITS2 regions (b).

Fig. 4 .
Fig.4.Aligned sequences of ITS1, 5.8S and ITS2 regions from Laurencia and Osmundea species.Sequences were aligned using the CLUSTALV programme and in some areas adjusted by eye.Dots denote the same nucleotide as that of Laurencia sp.A ("amarilla") in row one, a * denotes the same nucleotide in all the Laurencia and Osmundea species, a hyphen represent a gap while a x represents an unknown nucleotide.The 5.8S coding region is highlighted in bold.Lower case letters at the beginning and end of the alignment represent the 18S and 28S coding regions respectively.The forward and reverse primers are underlined.
CO spacer sequence data by maximum-likelihood (Fig.3) indicates that the Laurencia, Osmundea and Palisada species appear to split into two branches with L. viridis and L. thyrsifera grouped together and O. pinnatifida, P. perforata, P. cf. perforata and L. sp.A ("amarilla") forming the other.The evolution of the two spacers at different rates was also seen in the two Phylogenetic of the Laurencia, Osmundea and Palisada spp.

Fig. 5 .
Fig. 5. Aligned sequences of rbcL and rbcS regions from Laurencia and Osmundea species.Sequences were aligned using the CLUSTALV programme same that figure 4.

Fig. 6 .
Fig. 6.Multiple alignment of 5.8S sequences produced by CLUSTALV.A dot denotes the same nucleotide as that of Laurencia sp.A ("amarilla"), * represents the same nucleotide in all seven species, a hyphen represents a gap and a x represents an unknown nucleotide.

Table 2 .
Sizes and percentage identity of the rbc spacers in the genera Laurencia, Palisada, Osmundea, Antithamnion and Gracilariopsis

Table 4 .
Percentage identity matrix of the 5.8S coding region.