The genus Hypericum L., family Hypericaceae, is composed of approximately 450 species of trees, shrubs, and herbs widely distributed in temperate regions across the globe [1]. Originally native to southern Europe, H. perforatum is commonly found throughout temperate regions of both the northern and southern hemispheres [2]. Classified within the second largest section (Hypericum) of the genus, H. perforatum, commonly known as St. John's wort, is the best known species of the family. Hypericum perforatum has been suggested to have originated from the ancient hybridization and subsequent polyploidization of two diploids (2n = 2× = 16), H. maculatum subsp. maculatum Crantz and H. attenuatum Choisy [3]. It is a facultative apomict, as both sexual and aposporic processes can take place on the same plant [4]. While most H. perforatum individuals generated through apomixis are tetraploid (2n = 4× = 32) there are hexaploid (2n = 6× = 48), diploid (2n = 2× = 16), and aneuploid individuals as well [5], [6], [7].

Hypericum perforatum's biological extracts are widely recognized as valuable phytopharmaceutical agents with antiviral capabilities [8], and the potential to treat maladies such as depression, skin wounds, and burns [9]. Hypericum perforatum contains at least ten classes of biologically active compounds [10], of which two of the more important bioactive compounds, hypericin and hyperforin, are broadly variable in biological activity in humans [11]. Research indicates that these compounds vary in concentration and or constituency depending on species origin, tissue type, genetics, and environmental factors [11]. In addition, concentrations of these compounds can vary widely between accessions derived from the same species [12], [13].

Quality control involved with the production and distribution of phytopharmaceutical medicines has not been highly regulated with respect to species of plants being used in the preparation of commercial products and the concentration of bioactive compounds. Moreover, the technology available for identification of H. perforatum plant material in commercially available products is not standardized and thus variation between products is an issue [14]. Because of the importance of H. perforatum to the phytopharmaceutical industry, it is important to develop a reliable marker system that can be used to affordably and accurately identify plant material purported to be H. perforatum in order to aid producers while protecting consumers from potentially adulterated products.

Studies conducted by Arnholdt-Schmidt [15] and Mayo et al. [7] demonstrated that techniques such as RAPD (random amplification of polymorphic DNA) and AFLP (amplified fragment length polymorphism) analysis, would enable the elucidation of genetic diversity in wild populations of Hypericum spp. In this study, AFLP analysis was used to describe patterns of genetic variation and distribution within and among wild and commercially cultivated accessions of H. perforatum, and additionally, to develop a suite of species-specific markers that can be used to identify H. perforatum plant material. AFLP analysis is a whole-genome approach that has broad applicability in determining genetic variability within and among plant populations [17], crop origins [18], and relationships among cultivars [9]. AFLP markers are highly repeatable [19], provide broad genomic coverage and a virtually limitless number of genetic markers. Using AFLP technology, we identify two monomorphic and 28 polymorphic species-specific markers that can be used to accurately identify plant material purported to be H. perforatum.

Hypericum spp. were obtained from the North Central Regional Plant Introduction Station in Ames, Iowa (Table 1). Fifty-six Hypericum accessions from three different continents were studied, including 11 different species, 38 wild-collected and four cultivated accessions of H. perforatum, and two accessions of the outgroup Triadenum walteri [20]. The taxonomic identities of these accessions follow the systematic treatment used in the Germplasm Resources Information Network database, http://www.ars-grin.gov/npgs, except that accessions of H. perforatum were identified to subspecies on the basis of available herbarium vouchers, digital images, living plants, and geographic origin by following Robson's (2002) key [21]. Leaf material was obtained from three individual plants per accession, flash frozen in liquid nitrogen, and stored at −80°C prior to DNA extraction.

Table 1 Accessions of Hypericum spp. studied

Total genomic DNA was extracted from leaf tissue using the DNeasy Plant Mini kit (Qiagen Inc.; Valencia, CA, USA) in accordance with the supplied protocol and quantified using a Nanodrop (Nanodrop Technologies; Wilmington, DE, USA) spectrophotometer. Amplified fragment length polymorphism (AFLP) analysis was run on each sample and its technical replicate in accordance to Vos et al. [16], with modifications to include slight differences in adapter and primer sequences (Table 2). Digestion, ligation, pre-selective and selective amplifications were performed as in Hawkins et al. [22]. Following amplifications, samples were submitted to the DNA facility of the Iowa State University and run on an ABI 3100 Genetic Analyzer (Applied Biosystems; Foster City, CA, USA).

Table 2 AFLP primer and adapter sequences

AFLP banding patterns were visualized with Genographer 1.6.0 [23]. For analytical purposes, bands of the same size were considered homologous, even though it is possible that some bands of the same size may actually represent non-homologous genomic fragments. Visual comparisons between three biological replicates, as well as two technical replicates, were used to determine reproducibility. Bands absent from two of the three biological replicates and their corresponding technical replicates were excluded from the study. Homologous bands were scored for presence (1) or absence (0).

To visualize relationships among accessions, Neighbor-joining analysis was conducted in Paup* version 4.0 [24], using the 56 accessions of Hypericum spp. and rooting with two accessions of Triadenum walteri. Default settings were employed, except “Break ties” was set to “randomly” and distances were calculated using Nei's [25] restriction-site distances. Branch support was assessed through the implementation of 5000 bootstrap replicates. Principal coordinate analysis (PCO) was performed with NTSYS-pc [26] to obtain an additional visual representation of patterns of genetic variation in the wild and cultivated material and to explore possible relationships with geography. Genetic diversity within H. perforatum was hierarchically partitioned using analysis of molecular variance (AMOVA) [27] in the GenAlEx program [28].

AFLP markers were generated for 56 accessions of Hypericum spp. and 2 accessions of Triadenum walteri. Four AFLP primer combinations produced a total of 298 easily scored and reproducible markers. Within the 42 H. perforatum accessions, 221 markers were generated, of which 204 (92%) are polymorphic and 17 (8%) are present (monomorphic) in all accessions. Of the 17 monomorphic markers, only two (IVC134 and IVC335) were specific to H. perforatum, while the other 15 were present in accessions outside of H. perforatum. However, AFLP analysis generated 28 polymorphic H. perforatum-specific markers (Table 3). Of these, 10 were present at a frequency of 50% or more, 9 were present in 20–49% of the accessions, and 11 were present in less than 20% of the H. perforatum accessions.

Table 3 Markers detected in H. perforatum as revealed by AFLP analysis

Neighbor-joining analysis (Fig.1) revealed a monophyletic H. perforatum clade supported by a bootstrap value of 80. Within the H. perforatum clade there is a basal monophyletic group (clade 3) composed primarily of accessions from Lithuania and supported by a bootstrap value of 99. The remainder of the H. perforatum accessions are sister to this basal Lithuanian group and are divided into two additional major clades (clades 1 and 2) and one minor paraphyletic group (“paraphyletic accessions”). The larger clade (clade 2) predominately contains accessions from the Czech Republic and appears to be divided into 3 groups, each with bootstrap support of 100. Three of the four domesticated H. perforatum accessions studied (Ames 27453–27455) are located within this clade. The remaining major clade (clade 1) is comprised of accessions representing all 4 of the sub-species found in H. perforatum. As expected, the H. perforatum clade and its sister group, the Hypericum spp. clade, are composed of species sharing the characteristic dark leaf glands, these containing hypericin, pseudohypericin and hyperforin.

Fig. 1 Neighbor-joining analysis of Hypericum spp. Bold letters represent different sub-species A = perforatum, B = songaricum, C = veronense, and D = chinense. Accessions in single quotes indicate cultivated accessions. Taxa outside of box contain dark glands. (more ...)

PCO analysis on AFLP data derived from all accessions show a clear delineation between H. perforatum and all other accessions (Fig. 2A). Congruent with the neighbor-joining analysis, the H. perforatum accessions appear in a tight cluster most closely associated with other Hypericum spp. that produce dark glands. When only H. perforatum accessions are included in the PCO analysis, three separate clusters are apparent, consistent with the three major clades recovered in the neighbor-joining analysis (Fig. 2B). Additionally, 5 accessions (Ames 27452, 27510, 27511, and 27512, and PI 325351) occupy an intermediate position in the PCO outside of the three major clusters, indicative of possible introgression. These five accessions are basal to the larger two H. perforatum clades in the neighbor-joining tree.

Fig. 2 A PCO illustrating the relationship of all Hypericum spp. and the outgroup Triadenum walteri. Inner circle is composed of only H. perforatum accessions and outer circle contains all accessions with dark, hypericin-containing glands. B PCO illustrating (more ...)

The distribution of genetic diversity within and between Hypericum spp. populations was explored using AMOVA. Accessions were grouped together based on region of origin and/or domestication, only polymorphic markers were employed, and the analysis examined both the global and locus-by-locus partitioning of genetic diversity (Table 4). First, AMOVA was used to evaluate partitioning of genetic diversity within and among accessions from different geographic areas. Results indicate that the majority of variation (64%) present in the H. spp. used for this study can be attributed to within-population differences, while 36% of the variation can be ascribed to among-population genetic variation. When only H. perforatum accessions are included in the AMOVA, 88% of the variation can be attributed to within-population differences due to geographic collection locations, while only 12% of the variation can explain among-populations differences. Second, if the populations are segregated by domestication, the measure of variation within-populations increased to 94%, while variation among-populations decreased to 6%. A closer look at the PCO indicates that there is a domesticated accession (Ames 27452 Elixir; Richters, Goodward, Canada) central to the 3 main groups. This could be a result of hybridization and introgression of genetic material from the other primary clusters. With this in mind, the same comparison was made while excluding Elixir. The resulting analysis indicated that the within-population variance is 11% and the among-population variance is 89%. A fourth AMOVA, consisting of only H. perforatum accessions and segregated into 4 populations based on relatedness as indicated by neighbor-joining analysis (accessions in clade 1 = population 1 etc.), was conducted. The among-population variance is 33% and the within-population variance is 67%.

Table 4 Analysis of molecular variance (AMOVA) for Hypericum spp.

One obstacle facing breeders, horticulturists, researchers, and oversight agencies working with medicinals is the inability to genetically determine the source of plant material. Markers generated in this study may aid in overcoming this obstacle. Of 298 polymorphic markers generated, 17 markers are present in all Hypericum accessions and 30 markers are present in only H. perforatum. Two markers (IVC134, IVC335), are specific for H. perforatum, and present in all accessions studied; these may prove particularly useful for identification of H. perforatum plant material. Collectively, the 30 unique H. perforatum markers may aid breeders in determining genetic identity and source, can be employed as a tool by producers to accurately diagnose the identity of individual plant lots, and could be useful to agencies or consumer groups as a means to evaluate end-user H. perforatum “St. John's wort” preparations. Additionally, this molecular marker study provides the foundation for future work focused on developing species-specific primers that could be used to identify material purported to be H. perforatum with a single PCR reaction.

Neighbor-joining analysis supports the delineation of Hypericum spp. that either have or lack hypericin-containing dark glands. It is also evident from both neighbor-joining and PCO analysis that H. perforatum clusters tightly and separately from other species of Hypericum. Within H. perforatum, three distinct clades and one minor paraphyletic group are observed. Three of the four domesticated accessions belong to the same clade (clade 2), and two of those accessions are phylogenetically sister to one another and share boot-strap support of 100%. Interestingly, clade 2 is comprised entirely of subspecies perforatum, suggesting that the domesticated accessions originate from within this group. None of the domesticated accessions in our study belong to clades 1 or 3, which contain members of subspecies perforatum (clade 3) and a mixture of subspecies perforatum, songaricum, and veronense (clade 1). It is within these clades that breeders may choose to look in order to identify new traits or increase genetic diversity within the domesticated accessions.

Accessions from regional geographic areas tend to be more closely related. However, there are multiple instances where H. perforatum accessions from one location are more closely associated with those from different locations. Additionally, only 12% of the total amount of genetic diversity observed can be attributed to among-population difference, indicative of high levels of gene flow between populations. For example, the presence of the California accession Ames 27490 in clade 2; given that wild populations from California are naturalized from foreign introductions, it is not surprising that this accession groups with the European, domesticated species in clade 2. Observations such as this are consistent with results previously shown in H. perforatum illustrating that populations from different geographic areas can and often times are more closely related [29].

Analysis of molecular variance between H. perforatum accessions and Hypericum accessions from other species indicates that there is a high level of among-population variation (36%). This indicates an abundance of variation at the genus level, which the phylogenetic and clustering analyses readily partitioned into distinctive groupings. When comparing only H. perforatum accessions by geographic region of collection, 88% of the variance occurs within populations. The variance due to geographic distribution was similar to variance attributed to domestication if the cultivated accession Elixir was excluded from the analysis. Elixir, which is centrally located in the PCO, is responsible for 5% of the among-population variance within the domesticated varieties. The high within-population variation exhibited in the analysis when the populations are distinguished by either geographic location or domestication, along with the findings of Maron et al. [25], encouraged us to re-analyze the data with the populations segregated in accordance with the neighbor-joining analysis. When analyzed under these conditions, the level of among-population variation is substantially increased. These findings imply dispersal of plant material outside of their original range, most likely with human assistance.

Genetic distance analysis of the AFLP data revealed that the cultivated populations studied share higher genetic identity with the Western and Central European populations (0.925) than with populations from East Europe and Asia (0.828). This could be attributed to the fact that the cultivated varieties used in this study were developed in Germany and Denmark. Additional studies involving a larger sampling of domesticated material will help distinguish these possibilities, and may shed additional light on the source(s) and number of times that H. perforatum has been domesticated.

While other studies have utilized a molecular approach to place H. perforatum within a phylogenetic framework [14], [20], [29], [30], this is the first study placing an emphasis on the relationships and diversity between both wild and cultivated accessions of H perforatum, within the overall phylogenetic framework of the genus Hypericum. This study demonstrates that there is a great deal of genetic diversity among Hypericum species as well as within H. perforatum, and that this diversity is structured phylogenetically and geographically. These data provide the foundation for future work characterizing the evolutionary history, genetic relationships, and recent domestication of St. John's wort and its closely related species.

We would like to thank Nathan Johnson for his technical assistance. Mention of commercial brand names does not constitute an endorsement of any product by the U.S. Department of Agriculture or cooperating agencies. This publication was made possible by grant number P01ES012020 from the National Institute of Environmental Health Sciences (NIEHS) and the Office of Dietary Supplements (ODS), NIH. Its contents are solely the responsibility of the authors and do not necessarily represent the official views of the NIEHS, NIH.