Twenty-one strains used in the study were isolated from paddy fields of Eastern Uttar Pradesh. Anabaena sp. PCC7120 was chosen as reference strain. Strains were maintained and cultured in BG-11 medium (Hughes et al., 1958) buffered with Tris/HCl (pH 7.5) under continuous cool white fluorescent light 72 µmol photon m^-2 s^-1 PAR irradiance with a photoperiod of 14:10 h (light: dark) at 24  2^oC. Cultures were manually shaken 2 to 3 times daily.

Morphological characterization of 21cyanobacterial strains were conducted using trinocular microscope (Kyowa Getner, Opto-Plan, 2KT, Japan, Model No. 90116) equipped with high resolution Nikon digital camera (model no. DXm 1200). Morphological characters were described according to traditional criteria (Desikachary, 1959; Geitler, 1932). All the (16) morphological characters used to prepare a matrix for WINCLADA are listed in Table 1 and described in detail in Appendix 1.

Among the characters only the presence and absence of sheath and hormogones were codified as binary data, while the remainings were codified as multivariate data. Composite coding was used in preference to binary coding in order to minimize the occurrence of inapplicable or missing entries. Unknown determinations of the true state or missing data, are represented in the matrix by a “?”. Primary causes of missing data include character states being not visible in the material available. The order of appearance of the characters in the matrix is functional facilitating future data addition.

The phylogenetic tree was constructed using WINCLADA version 1.00.008 (Nixon, 1999). The maximum parsimony analysis of above mentioned characters was done selecting Arthrospira maxima as an outgroup. The data matrix was analysed using WINCLADA applying the following search parameters for Nona’s heuristic search: maximum trees to keep (hold) =100, number of replications (mult* N) =10, starting trees per rep (hold/) =10. A multiple tree bisection- reconnection (TBR) plus TBR (mult* max*) search strategy was employed. All the characters were weighted equally. The distribution of each morphological character and different character states of the isolates was then analyzed to get the most consensus phylogenetic tree through bootstrapping by using consensus compromise following majority fools option.

Figure 1

Table 1: Distribution of 16 morphological characters among exempler species. Character states are scored as 0-9, ? (Indeterminate). Refer to Appendix 1 for character descriptions

Figure 2

Figure 1. Microphotographs illustrating the morphological appearance of

(1) Anabaena doliolum LCR1, (2) Cylindrospermum muscicola LCR6, (3) Nostoc species LCR21,(4) Tolypothrix tenuis LCR7, (5) Aulosira fertilissima LCR4, (6) Anabaena anomala LCR5,(7) Nostoc paludosum LCR16, (8) Nostoc spongiaeforme LCR10, (9 Anabaena oryzae LCR2, (10), Anabaena constricta LCR23, (11) Nostoc piscinale LCR22, (12) Nostoc species LCR17, (13) Anabaena fertilisssima LCR23, (14) Nostoc species LCR12, (15) Anabaena toruolsa LCR24, (16) Nostoc ellipsosporum LCR14, (17) Nostoc carneum LCR19, (18) Nostoc linckia LCR11, (19) Nostoc punctiforme LCR15, (20) Anabaena sp. PCC7120, (21)) Nostoc rivulare LCR13

Protein extraction was performed using a modified protocol of Wagener et al., (2002) as detailed in Bhargava et al.,( 2008). SDS-PAGE was performed in 12% polyacrylamide gel according to Sambrook & Russel (2001). All chemicals were of Sigma grade (Sigma chemical Co., USA) unless and otherwise stated. 15 µg of protein was loaded in the wells for each sample along with the Sigma wide range molecular weight marker in last lane. Gels were initially run at 15 mA (stacking gel run) and successively at 25 mA (resolving gel run) in a vertical elctrophoretic apparatus (Bio-Rad, USA). Gels were stained in Coomassie brilliant blue R-250 and destained in destaining solution. Gel images were recorded using the Chemi Doc EQ gel documentation system (Bio-Rad, USA)

The image analysis was performed using the Quantity one software (Bio-Rad, USA). The protein fingerprinting patterns were compared using the information on apparent molecular masses of bands, band intensity, normalized quantity, trace intensity, and Gauss intensity.

SDS-PAGE patterns were processed as described by Piraino et al., (2002), using a logarithmic transformation of molecular weight (log kDa). Classes (16; class width = 0.078 log kDa) were obtained in the range from 5 kDa (starting class) to 75 kDa (last class); flat range (FR) around the class centre and the membership in the flat range (MFR) were 30% and 99%, respectively, in all cases. Hierarchical cluster analysis (Unweighted Pair Group Method Using Average Linkage, UPGMA) was carried out on the matrix of similarity data, calculated by Pearson product-moment correlation of the logistic weighted SDS-PAGE patterns. Statistical analysis was performed using Systat 11.0 for Windows (Systat Software Inc., Richmond, CA, USA).

Total genomic DNA was extracted by the phenol free method of Srivastava et al., (2007). The amplification of 16S rRNA gene was performed using cyanospecific primers CYA106F (forward) and CYA781R (reverse) (Nübel et al., 1997). These primers were synthesized from Sigma Chemical Co. (USA). PCR reaction mixture (25 µl) contained 100 ng of genomic DNA, 2.5 µl of 10X PCR buffer with 15 mM MgCl₂, 200 µM dNTPs, 10 pmol of each primer and 0.2 U Taq DNA polymerase (Bangalore Genei, India). PCR amplification was perfrormed in a iCycler (Bio-Rad, USA) using the following temperature program: initial denaturation for 3 min at 94°C followed by 35 incubation cycles each consisting of 1.5 min denaturation at 94°C, 1 min annealing at 50°C, 2 min elongation at 72°C and a final 5 min elongation at 72°C. All amplifications were done in triplicates to ensure the reproducibility of results. The amplification products were resolved on 1. 2 % agarose gel and stained with ethidium bromide (1 µg ml^-1). PCR products (500 ng) were lyophilized in a Speed Vac concentrator, Model-SPD111V (Thermo Electron Corporation, USA). These lyophilized PCR products were subjected to purification and sequencing on commercial basis (Macrogen Inc., Korea). Partial 16S rRNA gene sequences of the isolates were deposited in the GenBank database with accession numbers EF066606- EF066611 and EU446006- EU4460021.

The 16S rRNA gene sequences of the twenty-one local isolates were multiple aligned using CLUSTAL X (1.83) (Thompson et al., 1997) and manually corrected by using Jalview. A phylogenetic analysis was performed using MEGA 4.0 platform (molecular evolutionary genetics analysis) (Tamura et al., 2007) an analytical package. With the aligned sequences Maximum-parsimony analysis in conjuction with a maximum mini Branch -&-bound option was carried out. To evaluate the robustness of branches in the inferred tree the bootstrap resampling of Felsenstein was employed using 111 random seed and 100 replicates and a consensus tree was generated.

The cluster analysis carried out on the morphological characters grouped the strains in three major clusters (Fig. 2). Cluster 1 harbours Nostoc species LCR21, Anabaena anomala LCR5, Aulosira fertilissima LCR4, Tolypothrix tenuis LCR7 with basal clade of Anabaena doliolum LCR1 and Anabaena torulosa LCR24. The clustering of Anabaena anomala LCR5, Anabaena doliolum LCR1 and Anabaena torulosa LCR24 is characterized by barrel shaped, 2-15µm wide cells and flattened end spores formed centrifugally adjacent to the heterocyst. Moreover, the clustering together of Aulosira and Tolypothrix can be ascribed on the grounds of the closeness in cell dimensions (5-15µm), cell shape (quadrangular) and most importantly, beginning of false branching in one and the developed branch in the other case (Fig. 1). On the other hand clade 2 is composed of reference strain Anabaena sp. PCC7120, Anabaena fertilissima LCR18, Cylindrospermum muscicola LCR6, Anabaena oryzae LCR2 and Anabaena constricta LCR23. This grouping can be ascribed to the common type of vegetative cell shape i.e. cylindrical, conica,l terminal heterocyst and akinetes positioned as adjacent to the heterocyst.

Further, cluster 3 emerged as a pure Nostoc clade composed of Nostoc piscinale LCR22, Nostoc paludosum LCR16, Nostoc punctiforme LCR15, Nostoc species LCR12, Nostoc spongiaeforme LCR10, Nostoc rivulare LCR13, Nostoc species LCR17, Nostoc linckia LCR11, Nostoc ellipsosporum LCR14, Nostoc carneum LCR19. This clade is characterized by their entangled wave like or tightly coiled aggregation, presence of prominent/diffluent sheath, short spherical/ovoid cells, predominant intercalary heterocyst and akinetes position distant from heterocyst. Probably these characteristic features make their grouping consistent. Contrary to the above, Nostoc species LCR21 did not fall within the Nostoc cluster. This could be attributed to the absence of akinetes in Nostoc species LCR21.

Figure 3

Figure 2: Maximum-Parsimony based phylogram of morphological data showing the clustering of , , , and

The hierarchical cluster analysis (Unweighted Pair Group Method Using Average Linkage, UPGMA) carried out using the logistic weighted SDS-PAGE data set generated five major clusters at a similarity level (Pearson product-moment coefficient) of 0.7 (Fig. 3). Only four strains (position 1a, 2a, 3a and 4a) failed to group in any major cluster because their SDS-PAGE profile was different from the other strains. Most of the strain replicates grouped together at a similarity level ranging from 0.99 to 0.85 showing that the analysis was reproducible. Cluster C1 and C2, in the upper section of the dendrogram, were very near. Cluster C1 grouped Nostoc linckia LCR11 and Anabaena doliolum LCR1 while cluster C2 included the replicates of the species of Nostoc carneum LCR19 and Nostoc piscinale LCR22. On the other hand clusters C3, C4 and C5 in the lower section of the dendrogram grouped most of the isolates. Cluster C3 included the strains Cylindrospermum muscicola LCR6, Anabaena anomala LCR5 and Nostoc rivulare LCR13. The largest cluster C4 grouped most of the Nostoc species such as Nostoc punctiforme LCR15, Nostoc spongiaeforme LCR10, Nostoc ellipsosporum LCR14, Anabaena fertilissima LCR18 and the strains of the species Tolypothrix tenuis LCR7 and Aulosira fertilissima LCR4. The nearest group of cluster C4 was cluster C5 that included

the strains of Anabaena sp. PCC7120, Anabaena oryzae LCR2, Anabaena constricta LCR23 and Nostoc species LCR17. SDS–PAGE analysis confirmed the high similarity level of the species Nostoc punctiforme LCR15, Nostoc spongiaeforme LCR10 and Anabaena fertilissima LCR18 (Cluster C4). However, the species of Nostoc rivulare LCR5, Tolypothrix tenuis LCR7, Aulosira fertilissima LCR4 and Anabaena sp. PCC 7120 included in the nearest cluster C3, C4 and C5 possessed a good level of similarity.

Figure 4

Figure 3: Dendrogram showing the similarity relationship among the SDS-PAGE patterns of whole cell proteins of cyanobacterial strains. Similarities were calculated as Pearson product-moment correlations among the logistic weighted patterns and clustering was performed by using the UPGMA algorithm

In contrast to SDS-PAGE analysis the deepest nodes of the phylogenetic tree with only partial support by the bootstrap values separated the twenty-one strains in two large clades (Fig. 4). Clade 1 that included clusters I and II corresponded to Nostoc and Anabaena with representatives of the species Nostoc spongiaeforme LCR10, Nostoc linckia LCR 11, Nostoc rivulare LCR13 and Anabaena torulosa, Anabaena constricta Nostoc species LCR21, Nostoc species LCR12, Nostoc species LCR17 and Anabaena sp. PCC 7120, while clade 2 consisted of clusters III and IV.

The large cluster I, included maximum number of nostocalean strains except few like Nostoc carneum, Nostoc piscinale, Nostoc punctiforme, Nostoc paludosum and Nostoc ellipsosporum are included in cluster II and III. These nostocalean strains belonging to cluster 1 contained in this clade also showed morphological resemblance with Nostoc strains in terms of entangled filamentous aggregation and predominant intercalary heterocyst. The frequent occurrence of these strains within cluster I and II could be due to distinct morphological features like entangled trichome aggregation, cylindrical/barrel-shaped cells, terminal and intercalary/terminal or intercalary heterocyst and akinete position (adjacent/distant) with respect to heterocyst

Thus the 16S rRNA gene data supports the grouping of the species indicated that Nostoc and Anabaena are not clearly separated at genetic level and may have their common origin. This further suggests that the Nostoc strains were heterogenous and seemed to form monophyletic cluster containing more than one genus.

The small cluster II and III are composed of Nostoc strains. The placement of Anabaena fertilissima within Nostoc cluster II may be due to a lack of genetic distance between the Nostoc and Anabaena strains. This is in accordance with the reported higher degree of genetic relatedness among heterocystous species relative to other cyanobacteria (Lachance M., 1981). The intermingling of Anabaena with Nostoc and other strains like Cylindrospermum and Tolypothrix deciphers lack of correlation between sequence and morphology and suggested polyphyletic origin. In conclusion, the lack of congruency of phylogenies based on morphological criteria may indicate a close genetic relationship in the genera Nostoc and Anabaena.

The lowest clade of the phylogenetic tree corresponded to the wide radiation including the genera Anabaena, Cylindrospermum, Aulosira and Tolypothrix where Aulosira and Tolypothrix grouped together depicting their phenotypic similarity (Geitler, 1932) and Cylindrospermum muscicola and Anabaena anomala were clustered together on the basis of sequence relatedness. However, this clustering displays low sequence similarity and low bootstrap support which may be due to the small numbers of characters supporting each node, further indicating them as closely related strains that have not diverged broadly. This cluster deserves a complete characterization and possibly the addition of more strains for a better interpretation.

Figure 5

Figure 4: Majority rule consensus tree of 16S rRNA gene sequences by using Maximum-parsimony method

All the three approaches were used together not only to find out the congruencies /incongruencies of the three systems in classifying the cyanobacterial isolates but also to check as how far the polyphasic approach is reliable as compared to single approach. Cluster C1 grouped Nostoc linckia LCR11 and Anabaena doliolum LCR1 together. The 16S rRNA gene maximum-parsimony tree not only puts them in distinct clades but indicates morphological delimitation as well. Thus it can be anticipated that the two organisms have divergent evolutionary relationships. Their SDS-PAGE brings them in one clade because they have been acquired from the same culture conditions and hence analogous protein profile. The above observation is in agreement with the reports (Tomitani et al., 2006; Prasanna et al., 2005; Henson et al., 2004; Gugger et al., 2002) that traditional techniques for identification and systematics of cyanobacteria based on morphological traits suffer from various limitations such as environmental pressure, as phenotypic characters are prone to environmental changes. Cluster C2 included the replicates of the species of Nostoc carneum LCR19 and Nostoc piscinale LCR22. This is well attested by high sequence similarity in 16S rRNA gene and high bootstrap values. The two strains appear close morphotypes. Thus the congruence of genotypic and phenotypic data attested the taxonomic coherence of these species. Cluster C3 encompassed Cylindrospermum muscicola LCR6, Anabaena anomala LCR5 and Nostoc rivulare LCR13. Their clustering pattern is in accordance with 16S rRNA gene based MP tree except Nostoc rivulare LCR13 harbouring clade3; however, all the three strains are separated into different clades on morphological grounds, like straight/entangled aggregation, barrel or cylindrical cell morphology, attenuated/conical apical cell, akinete position etc. Thus morphological clustering of species was not supported by 16S rRNA gene sequences. The clade is represented by a phylogenetically coherent but morphologically diverse group of strains.

It may be thus proposed that the strains were phylogenetically linked and are of same genetic make up but appear quite different under different physiological conditions as a result of differential gene expression. Morphological differences suggest that they might have undergone different environmental pressures.

The grouping of Cylindrospermum muscicola and Anabaena anomala was consistently observed in all phylogenetic analyses either SDS-PAGE or 16S rRNA gene analysis. A larger number of strains would be necessary to understand whether they represent a new phylogenetic unit.

The coherent positioning of Aulosira and Tolypothrix at genetic level corresponds to their traditional phenotypic traits and suggested that sometime clear traditional phenotypes can be found that correspond to the genotypes. However, an ambiguous morphology can often be observed which may correspond to the genetic diversity (Whitton, 1987).

The largest cluster C4 was quite congruent with the cluster 3 of 16S rRNA gene based MP tree. This exclusively harboured the species of Nostoc and Anabaena including, Nostoc punctiforme LCR15, Nostoc spongiaeforme LCR10, Nostoc ellipsosporum LCR14, Anabaene fertilissima LCR18 and strains of the species Tolypothrix tenuis LCR7, Aulosira fertilissima LCR4 suggesting that Nostoc strains were heterogeneous. They seemed to form a monophyletic cluster containing more than one genus. Intermingling of Anabaena species indicates that the two taxa are not clearly separated and distinguished at genetic level and may be polyphyletic in origin. It is worth mentioning here that apical cell shape emerged as a diacritical parameter for more clear distinction of Nostoc and Anabaena in morphological tree. Although in the above case a good agreement existed between phenotypic and genotypic features regarding Tolypothrix tenuis LCR7 and Aulosira fertilissima LCR4. They are clustered together in cluster 3 with different strains of Nostoc and Anabaena in MP tree. This strongly indicates their independent evolution than that of other nostocalean strains. The SDS-PAGE, 16S rRNA gene and morphology based classification displayed high similarity level for these species, Thus phenotypic characteristics of this and other strains reflect incongruency with 16S rRNA gene sequences demonstrating that the genotypic plasticity of the heterocystous, entangled/coiled forms, consequently illustrated the adequacy of trichome aggregation, heterocyst and akinete position for Anabeana species characterization.

The species Anabaena oryzae LCR2 and Anabaena constricta LCR23 are close together both in morphological and SDS-PAGE dendrogram showing a low level of similarity with Nostoc species. This is in accordance with the DNA-DNA hybridization data of Lachance (1981) and hybridization patterns with repetitive DNA (STRR) sequences (Mazel et al., 1990). The diversity between Anabaena strains has also been previously demonstrated (Prasanna et al., 2005).

The present study reveals that the phenotypically diverse genera are closely related at their genetic level and represent their realistic phylogenetic relationship at the class and up to the genera level. It is also suggested that the clustering of strains in SDS-PAGE and 16S rRNA gene analyses is congruent and illustrates complementarity. Despite a few incongruities the clustering pattern of taxa in SDS-PAGE and 16S rRNA gene analyses are quite complementary to each other. The SDS-PAGE and 16S rRNA gene based phylogenetic analysis affirm the heterogeneity but monophyletic origin of nostocalean strains. Merging of Anabaena strains with Nostoc indicates their common origin. However, their taxonomic placement seems debatable. For the nostocales it appears that the assignment of clades may be premature due to the limited number of taxa presently available for phylogenetic analysis but high bootstrap values strongly point towards its monophyletism. There were few strains like Cylindrospermum muscicola LCR6, Tolypothrix tenuis LCR7, and Aulosira fertilissima LCR4 taken as a single strain in the study and these least number of strains is not adequate to explore their phylogenetic relationship evidently. Hence large number of strains must be isolated and employed in the study to assess their phylogenetic relationships more precisely.

In conclusion, single-dimensional SDS-PAGE phylogeny may be used as a complementary tool for the 16S rRNA gene based analysis to identify the nitrogen-fixing species at the molecular level. This shows immense importance of the classical cyanobacterial taxonomic systems. Molecular methods should be used in parallel or as a supplement to the phenotypic characterization but never replace it otherwise the more precise and probably more reliable molecular data would produce confusing results concerning the occurrence of cyanobacterial genera and species in nature.

Appendix 1. Morphological character descriptions, arranged in an order conducive to minimal handling of cyanobacterial taxa during character scoring.

Lal Chand Rai is thankful to ICAR, New Delhi for financial support. Poonam Bhargava is thankful to UGC for the award to SRF and Rashmi Rai and Yogesh Mishra to CSIR for JRF and SRF respectively. We are also thankful to Head and Program Coordinator, CAS in Botany, BHU for facilities.