Background

The genus Lippia comprises 150 species, most of which are distributed in the Neotropical ecozone [1]. Brazil stands out as the centre of diversity of the genus with 98 species presenting high degrees of endemism. More than half of these species are concentrated in the Espinhaço Range, which stretches 1000 km through the Brazilian states of Minas Gerais and Bahia [2]. However, 18 species are considered rare or endangered, and nine are under threat of extinction due to the destruction of their natural environments in the Cerrado region (Brazilian type of Savana) [3].

The Brazilian Ministry of Health has developed an extensive phytotherapy program over the last decade with the aim of providing access to herbal medicines for the entire population. One of the target species of this program is Lippia sidoides Cham. (syn. L. origanoides) (Verbenaceae), a plant that was included in the Formulário de Fitoterápicos da Farmacopéia Brasileira [4, 5] based on its strong antimicrobial activity, against Candida albicans [6, 7], Staphylococcus aureus, and Escherichia coli [8] were included due to the presence of terpenoids in the essential oil. It is well known that terpenoids are produced as part of the plant defense system and have been considered a promising source of biological compounds [9,10,11,12]. Several essential oil compounds such as linalool, eugenol, carvone, vanillin, carvacrol, and thymol have been accepted by the European Commission to be used in food preservation or flavorings [13].

The morphological similarities between this and other species within the genus tend to complicate the accurate botanical identification, leading to difficulties in the production of certified herbal medicines.

Based on the differential morphological characteristics, the genus Lippia was classified in seven sections [14]. The Zapania Schauer section is the most complex and exhibits inflorescences with flat bracts, spirally arranged, globose or hemispheric type, capituliform, with varying numbers of chromosomes (2n = 10–28). L. alba (Mill.) N.E.Br., L. aristata Schauer, L. brasiliensis (Link) T.R.S. Silva, L. corymbosa Cham., L. diamantinensis Glaz., L. duartei Moldenke, L. filifolia Mart. & Schauer, L. hermannioides Cham., L. lacunosa Mart. & Schauer, L. rotundifolia Cham. and L. rubella (Moldenke) T.R.S. Silva & Salimena [15, 16] are among the representatives of this section in the Brazilian flora.

The Goniostachyum Schauer section presents tetrastic inflorescences formed by four series of keeled bracts aligned in rows. This section is considered monophyletic and is characterized by small variations (2n = 12) in the number of chromosomes [15, 17]. A recent revision of the species belonging to Goniostachyum resulted in the validation of only four representatives, namely: L. grata Schauer, L. origanoides Kunth, L. sericea Cham. and L. stachyoides Cham. [17]. Thus, some nominations of species or varieties must be considered synonyms of L. origanoides including, amongst others, L. sidoides, L. graveolens Kunth, L. microphylla Cham., L. salviifolia Cham., L. velutina Schauer, and Lantana origanoides Martens & Galeotti. Additionally, L. dumetorum Herzog, L. gracilis Schauer ex DC, L. hickenii Tronc., L. laxibracteata Herzog, and others have received the synonym L. grata. [17]. The Rhodolippia Schauer section comprises species with numbers of chromosomes that are intermediate between those of sections Zapania and Goniostachyum [15, 18], including L. bradei Moldenke, L. felippei Moldenke, L. florida Cham., L. hederaefolia Mart. & Schauer, L. lupulina Cham., L. pseudothea Schauer, L. rhodocnemis Mart. & Schauer, and L. rosella Moldenke.

However, the taxonomic classification of Lippia remains incoherent mainly due to the morphological variability within the genus and the existence of a great number of nomenclatures for this species resulting in classification dualism, both of which can be explained if we consider the interaction between the genotype and the environment [19]. In this context, studies aimed at evaluating the genetic structure of the genus through analysis of molecular markers could be useful in classifying species into clusters according to their genetic similarities.

A number of reports confirm that the association of molecular markers such as amplified fragment length polymorphism (AFLP) and internal transcribed spacer 2 (ITS2) can contribute significantly to the analysis of genetic variability and phylogenetic inferences [20, 21].

Besides molecular markers, chemical markers can also be used to help the correct plant characterization. WinK [22] developed a phylogenetic classification based on the secondary metabolites produced by Fabaceae, Solanaceae and Lamiacea families. The author considered that the ability or inability to produce a specific metabolite—shown by different members of related phylogenetic groups, are the result of differential expression patterns that reflect specific plant strategies for adaptation that were incorporated into the phylogenetic structure.

Therefore, the aim of the present study was to assess the genetic and chemical variability of species of Lippia spp. using molecular and chemical markers, to draw inferences regarding the phylogenetic relationships within the genus, and to identify inconsistencies in the current taxonomic classification for the correct use of those plants in phytomedicine.

Methods

Plant materials, DNA extractions, PCR amplifications and sequencing

We used 141 accessions (Table 1) comprising six Lippia species; although L. sidoides and L. origanoides are synonymous, they were considered, for the purposes of this study, as they were classified. Thirty-seven of these accessions were from the medicinal plants germplasm bank (Ribeirão Preto University, Brazil) and 104 were collected in the medicinal botanical garden of Nature Pharmacy, Brazil, with voucher numbers; 1340; 1350;1351; 1353; 1355; 1359; 1360; 1362–1365; 1368–1376; 1378–1380; 2000–2015; 2017–2112; 2114; 2471; 2473–2475. Sampling permission, for both locations, were obtained from by the Brazilian Council for the Administration and Management of Genetic Patrimony (CGEN) of the Brazilian Ministry of the Environment (MMA) by the National Council for Scientific and Technological Development (CNPq—CGEN/MMA Process #: 02001.005059/2011-71). Fátima R. G. Salimena (Juiz de Fora Federal University, Brazil) identified all samples. Total genomic DNA was extracted from 0.15 g of frozen leaves using the cetyltrimethylammonium bromide (CTAB) method [23]. The DNA integrity was determined by electrophoresis on 0.8% agarose gels and the concentration and quality of the isolated nucleic acid was determined by a NanoPhotometer® P360 spectrophotometer (Inplen, Westlake Village, CA, USA).

Table 1 Location, Geographical coordinates and voucher number of Lippia species
Full size table

Reactions and analysis of AFLP data

Samples from all 141 genotypes were analyzed according to the method of Vos et al. [24]. Briefly, genomic DNA (300 ng) was digested with EcoRI/MseI enzymes (New England Biolabs, Ipswich, MA, US) at 37 °C for 3 h, followed by inactivation at 70 °C for 5 min. Resulting DNA fragments were ligated to adaptors complementary to the restriction enzymes recognition sites and the ligation products were then diluted 6× with deionized water. In the first round of polymerase chain reaction (PCR), pre-selective amplification was achieved with primers EcoRI + 1 (50 µM) and MseI + 1 (50 µM). The pre-selective products were diluted 10× with deionized water and a second round of PCR was carried out using marker primers fluorescently tagged with IRDye® (LI-COR Biosciences, Lincoln, NE, USA). The selected marked primers were those that generated the largest number of polymorphic bands. Genotyping of individuals was performed using a 4300 DNA Analyzer (LI-COR Biosciences, Lincoln, NE, USA) while data alignment was accomplished with the aid of SagaMX Automated AFLP Analysis software version 3.3 guided by molecular weight markers in the range 50–700 bp. A binary matrix was constructed based on a 1/0 score for the presence/absence of each electrophoretic band. The genetic distance was calculated from the binary matrix using Jaccard indices, whereas the dendrogram was constructed using the unweighted pair group method with arithmetic average (UPGMA) clustering technique with 1000 permutations and Free Tree software version 0.9.1.50 [25] and visualized through TreeView X program [26]. The genetic structure of genotypes was established by principal coordinates analysis (PCoA) using the software GenAlEx version 6.5 [27] and STRUCTURE version 2.2.4 [28], which generated a posterior distribution based on Bayesian and admixture models. Each analysis comprised a “burn-in” of 200,000 interactions followed by a run length of 500,000 interactions and five independent runs for each K value (K = 1 to 7). The most probable number of genetic groups was determined from the ΔK value [29]. The correlation between genetic and geographical data was performed using the Mantel test and the POPGENE 32 [30] and GENES version 2009.7.0 [31] programs with 1000 simulations.

Sequencing and phylogenetic analysis of the ITS2 gene

The primers employed in the amplification reactions ITS2F-5′AATTGCAGAATCCCGTGAAC3′ and ITS2R-5′GGTAATCCCGCCTGACCT3′ were designed based on ITS2 sequences of some Verbenaceae species from the GenBank database at the National Center for Biotechnology Information (NCBI), namely Aloysia gratissima (DQ463782.1), A. gratissima var. schulziae (AY178651.1), A. triphylla (EU761080.1), Lippia alba (EU761076.1), L. alba (EU761078.1), L. salsa (FJ867399.1), and Phyla dulcis (EU761079.1). Polymerase chain reaction was performed as described by Chen et al. [32] and the resulting amplified fragments were sequenced using a Thermo Sequenase™ Cycle Sequencing kit (Affymetrix, Inc, Cleveland, USA), following manufacturer recommendations, with e-Seq™ DNA Sequencing version 3.1 (LI-COR Biosciences). Consensus sequences were assembled with the aid of LI-COR Biosciences AlignIR software (version 2.1) and aligned with ClustalW. The sequence alignments were edited using the BioEdit software (version 7.2) [33]. Phylogenetic trees were inferred by the NJ method based on the Kimura-2 parameter using PHYLIP software version 3.69 [34]. The alignment quality of the final phylogenetic tree was verified by the presence of saturation of the nucleotide substitutions, and sequences exhibiting high genetic similarity were excluded from the phylogenetic analysis using DAMBE software version 4.0.36 [35]. Thirty-three sequences of the ITS2 region deposited in the NCBI GenBank were selected as references (Table 2).

Table 2 Accession number for ITS2 references of region from NCBI and used code
Full size table

Extraction and analysis of essential oils

The essential oils of L. origanoides, L. origanoides × velutina, L. velutina, L. sidoides, L. salviifolia, and L. grata were extracted from dried leaves and flowers by steam distillation in a Clevenger apparatus. A mixture of essential oil/ethyl acetate (v/4v) was analysed using gas chromatograph Varian, model 3900 (Palo Alto, CA, USA), coupled with a Saturn 2100T ion trap mass selective detector and equipped with a non-polar DB-5 fused silica capillary column (30 m × 0.25 mm i.d.; 0.25 μm). The analytical conditions were: carrier gas helium at 1 mL/min; oven temperature 60 to 240 °C at 3 °C/min; injector temperature 240 °C; detector temperature 230 °C; injector split ratio 1:20; injection volume 1 μL; ionization voltage 70 eV. Individual components of oil samples were identified from their Kovats retention indices [36] and by comparison of their electron impact spectra with entries in the NIST62 mass spectral library embedded in the GC/MS system. Data were submitted for principal component analysis (PCA) using the program GENES version 2009.7.0 [31] in order to determine which of the chemical descriptors contributed most to the variability.

Results

Analysis based on AFLP markers

The set of six primers selected for AFLP analysis of the 141 genotypes amplified 273 loci, of which 267 (97.8%) were polymorphic (Table 3). The dendrogram constructed from these amplified loci (Fig. 1) enabled the 141 genotypes to be discriminated into three distinct genotypic groups, namely group 1 (L. alba), group 2 (L. lupulina) and group 3 (L. origanoides, L. origanoides × velutina, L. velutina, L. sidoides, L. salviifolia, and L. grata). Interestingly, L. alba appeared to be more closely related to L. lupulina (boostrap 100%) than to L. origanoides.

Table 3 Sequences of selected primers IRDye 700/800 and number of amplified fragments
Full size table
Fig. 1
figure 1

UPGMA dendrogram constructed using data obtained AFLP polymorphic markers (1000 permutations). Individuals featured: Black circle: L. grata (LT9, LT16, LT44, LT47, LU142, LU143, LU144); white circle: L. salvifolia (LT118); black small circle: L. sidoides (LT116; LT117); lozenge: L. velutina (LT42, LT46, LT78, LT89, LU145, LU146, LU148)

Full size image

The cluster formed by group 3 indicated the absence of significant differentiation between L. origanoides, L. origanoides × velutina, L. velutina, L. sidoides, L. salviifolia, and L. grata. However, only 29% of the hybrid individuals clustered together, whereas 71% assembled with other species. Furthermore, only 37.5% of L. grata individuals clustered together, while 62.5% clustered with other species, demonstrating the occurrence of intra- and inter-specific similarities in Lippia.

The results generated by PCoA analysis also revealed three groups (Fig. 2), but the Bayesian approach using the STRUCTURE software indicated that the genotypes could be organized into two main groups (K = 2), suggesting that L. lupulina (group 1) occupied an intermediate position between groups 1 and 3 (Fig. 3).

Fig. 2
figure 2

Population structure as determined by principal coordinates analysis (PCoA) of 141 individuals of Lippia spp. Group 1—(Alb) L. alba; Group 2—(Lup) L. lupulina; Group 3—(Lor) L. origanoides, (Orv) L. origanoides × velutina, (Lv) L. velutina, (Sid) L. sidoides, (Sal) L. salviifolia and (Lgr) L. grata

Full size image
Fig. 3
figure 3

Population structure as determined by Bayesian analysis of 141 individuals genotypes of Lippia spp. Individual genotypes are represented by columns while the clusters (K = 2) are represented by the colors green and red. Two colors shown for the same individual indicate the percentages of the genome shared between the two groups

Full size image

The measure of shared variance between the genetic and geographic variables for individuals in group 3 showed a significant positive correlation (r = 0.80; p = 0.46), while the isolation by distance showed the existence of gene flow across group 3 (Nm = 1.6), although gene flow between groups 1 and 3 was lower (Nm = 1.3).

Analysis based on ITS2 genotyping

Primers ITS2F and ITS2R amplified DNA fragments of approximately 340 bp. The saturation test revealed that the ITS2 region presents significant genetic variability among the Lippia spp.

The Neighbor-Joining (NJ) of the phylogenetic tree was rooted using the Phyla canescens species identified in France (Fig. 4: Table 4). The use of a outgroup species from a different geographic location favors a more robust separation of the tree branches confirming the separation of the phylogenetic groups.

Fig. 4
figure 4

Evolutionary relationships between Lippia individuals generated from NJ analysis of ITS2 sequences (Kimura-2 model: PHYLIP software version 3.69). Reference sequences (see Table 2): Lamicr, Laangu, Lascab, Lacama, LaspX1, Lastri, Lahodg, LastrA, Glsubi, Glgvgo, Glguar, Glmend, Gldiss, Glaris, Glchei, Glbipi, Glchir, Glgvne, Glwrig, Glaura, GlbipT, Glarau, Glmicr, Jumicr, Jucaes, Jusela, Juavlo, Juspat, Julvlo, Juunif, Juaspa, Juaspe, Phylla canensis. Samples grouped by high genetic similarity: L2, L3, L4, L9, L11, L69, L118, L120, L129, L142 (see Table 4). Capital letters adjacent to code numbers 142 and 144 refer to the amplified bands of 340 bp (A) and 360 bp (B)

Full size image
Table 4 Lippia individual grouped by genetic similarity (ITS2) by DAMBE program version 4.0.36
Full size table

The phylogenetic analysis based on the species from the genus Lantana (A), Glandularia (B), Junellia (C), and Lippia (D) demonstrated separation of the three branches into four principal clusters with 83%, 93%, 85%, and 96% bootstrap, respectively. In the Lantana group, Lippia lupulina (L165) and Lippia alba (L120, L121, L122, L128), divided into subgroups with a bootstrap of 71% and 83%, respectively, were also identified. The group of Glandularia and Junellia was clearly subdivided into two groups: one belonging to the species of Glandularia and another to the Junellia subgroup.

Most of the analyzed species were separated within the Lippia group as a monophyletic group. Samples LU145 (L. velutina) and LT118 (L. salviifolia) were identical to the sample classified as L. grata (LU164). Furthermore, a sample classified as L. velutina (LT78) was identical to one of L. sidoides (LT117), as well as to samples of L. origanoides and L. origanoides × velutina. Additionally, a L. grata individual (LT47) was identical to one L. origanoides × velutina (LU156) and to some L. origanoides (LT2, LT31, LT34, LT36).

Principal Components Analysis (PCA) of essential oil profiles

The application of PCA analysis allowed individuals to be grouped according to their different chemical profiles and enabled the seven original chemical descriptors, namely carvacrol, endo-fenchol, thymol, β-caryophyllene, isoborneol, trans-caryophyllene, and bicyclogermacrene, to be reduced to the first three (Fig. 5). Endo-fenchol (PC1) and carvacrol (PC2) accounted for most of the total variation (86.36%), with the first and second components contributing factors of 0.69 and 0.17, respectively, while the contribution of thymol was minimal (only 0.063). Considering all the analyzed individuals, 72% contained carvacrol and 16% contained endo-fenchol; since no individuals contained both carvacrol and endo-fenchol, the quantification of these two components would cover 88% of the analyzed samples (Fig. 5).

Fig. 5
figure 5

Principal component analysis of the chemical constituents of Lippia essential oil

Full size image

Discussion

AFLP analysis

The employed AFLP technique distributed the 141 Lippia genotypes into three groups (Fig. 1) that were compatible with the existing taxonomic sections, namely Zapania (L. alba), Rhodolippia (L. lupulina) and Goniostachyum (L. origanoides, L. sidoides, L. salviifolia, L. origanoides × velutina, and L. grata) [16, 17]. The efficiency of dominant AFLP markers to regroup genetically similar species has been also demonstrated in a number of studies [37,38,39], having been attributed to the large numbers of amplified loci that are generated [40]. Additionally, PCoA analysis (Fig. 2) confirmed the distribution of the studied genotypes into three groups, a separation likely related to the reduced gene flow between the groups [41] as demonstrated by the values of Nm (1.3–1.6) obtained for Lippia species.

However, Bayesian analysis performed using the program STRUCTURE identified only two genetic groups (K =2), demonstrating that L. lupulina shares 50% of the genome of each group (Fig. 3), for more detail see Additional file 1. This result corroborates the results of Campos et al., [18], which classified Rhodolippia section (Group 2) as an intermediary between Zapania (Group 1) and Goniostachyum (Group 3) sections.

A recent study by O’Leary et al. [17] grouped L. origanoides × velutina, L. velutina, L. sidoides, and L. salviifolia, but not L. grata, within L. origanoides. Our results showed that individuals classified as L. origanoides, L. origanoides × velutina, L. velutina, L. sidoides, L. salviifolia, and L. grata formed a single group due to their strong genetic similarity, and therefore should be recognized as a single taxon to be named L. origanoides.

Nuclear ribosome ITS2

The results presented herein show that species in the genus Glandularia and Junellia may be considered genetically similar as were forming one group (Fig. 4), thus confirming former results [42]. Furthermore, the species used as an outgroup, Phyla canescens, showed clear genetic divergence from Lantana, Glandularia, Junellia and Lippia, even though the separation of these genus has been proposed based on increased morphological descriptors [43, 44].

Lippia alba and L. lupulina are closely related to members of the genus Lantana and, together, they can be considered sister-groups [45,46,47], attesting the genetic similarity between the genera Lippia and Lantana [18, 48, 49].

Additionally, L. alba and L. lupulina exhibit longer branches in comparison with other Lippia species, suggesting that they underwent a more accelerated evolutionary rate and that they are older species [20, 43, 50].

The results of the phylogenetic analysis performed with ITS2 markers confirmed the results obtained with AFLP markers, suggesting the existence of only three species, namely L. alba, L. lupulina and L. origanoides. Of these, L. alba (section Zapania) can be considered the most divergent within the genus, whereas L. lupulina (section Rhodolippia) represents an intermediate between sections Zapania and Goniostachyum, for more detail see Additional files 2 and 3. In this aspect, the findings from the molecular-based analyses corroborate those based on cytogenetic and morphological characteristics [15, 16, 18].

Chemical markers

The PCA analysis of the terpenoid composition from L. origanoides L. origanoides × velutina, L. velutina, L. sidoides, L. salviifolia and L. grata showed no specific grouping by species (Fig. 5), suggesting that they are different chemotypes. Conversely, Sandasi et al. [51], when investigating the chemotaxonomic differentiation of South-African Lippia species, namely L. javanica, L. scaberrima, L. rehmannii and L. wilmsii, were able to separate the species into distinct clusters. These results, paired with AFPL and ITS, suggest that L. origanoides, L. origanoides × velutina, L. velutina, L. sidoides, L. salviifolia, and L. grata belong to the same species, but present different chemotypes, for more detail see Additional file 4.

The chemotypes may be associated with the diverse biotic and abiotic stimuli to which each of the individuals had been subjected, which led to the creation of a complex biological system [52]. It is clear that nowadays the taxonomic identification of plants frequently rely on molecular biology techniques, especially when plants exhibit very similar morphological characters. In regards to medicinal plants, the use of chemical markers becomes essential if we consider that the biological activity can, most of the time, be related to a specific chemotype. Therefore, when any species is employed in the production of certified phytomedicines, the plant material must be identified taxonomically and the chemotype identified to assure the biological activity of the extract.

Conclusions

The molecular markers AFLP and ITS2 were efficient in separating L. alba and L. lupulina, and in grouping L. origanoides, L. origanoides × velutina, L. velutina, L. sidoides, L. salviifolia, and L. grata. Moreover, the markers revealed the existence of intra- and inter-specific variability within the genus, as well as the close phylogenetic relationship between L. alba and L. lupulina. Since individuals grouped in L. origanoides exhibit morphological diversity and variability regarding the major constituents of the essential oils, the selection of genotypes for the production of certified phytomedicines must be based on the chemical profile of the oil produced.