The genus Hypericum contains around 490 plant species that are classified into 36 sections (Crockett and Robson, 2011; Hypericum Online, 2016). These epitomize a large pool of natural resource for phytochemical investigation, particularly given the well-investigated biological potential of several representatives of the genus accumulating pharmacologically important secondary metabolites. The majority of morphological and phytochemical studies have so far been realized on one well-known species, Hypericum perforatum, commonly known as St. John's wort. A broad spectra of secondary metabolites with diverse biological activities have been found in this species, for example, the naphthodianthrone derivatives hypericin and pseudohypericin, their supposed precursor emodin, and/or emodin anthrone (Bais et al., 2003; Kirakosyan et al., 2004; Zobayed et al., 2006), phloroglucinols hyperforin, adhyperforin and flavonoids quercetin, quercitrin, rutin and hyperoside and/or isoquercitrin, to name a few (Nahrstedt and Butterweck, 1997; Saddiqe et al., 2010; Zorzetto et al., 2015; Hou et al., 2016; Oliveira et al., 2016). Among these compounds, the effectiveness of hypericin as a natural photosensitizer with implications in photodynamic therapy has been rigorously explored (Zhang et al., 2014; Jendželovská et al., 2016), in addition to studies on its antiviral and antimicrobial properties (Kubin et al., 2005; Karioti and Bilia, 2010; Dementavicius et al., 2016). Furthermore, both hyperforin and adhyperforin have been shown to play a role in inhibition of various neurotransmitter receptors, thereby exhibiting antidepressant efficacy (Laakmann et al., 1998; Jat, 2013). Additionally, St. John's wort extracts have been shown to contain a plethora of flavonoids, which further aid the antidepressant potency of the extracts (Butterweck et al., 2000; Saddiqe et al., 2010; Hou et al., 2016).

Numerous studies have been performed in the last decade for qualitative as well as quantitative enumeration of secondary metabolites in plant tissues of Hypericum spp. Analytical methods such as thin layer chromatography (TLC) (Ferraz et al., 2002), high-performance liquid chromatography (HPLC) (Ferraz et al., 2002), liquid-chromatography mass spectrometric analyses (LC-MS as well as LC-HRMS) (Tatsis et al., 2007; Kusari et al., 2009), and nuclear magnetic resonance (NMR) (Porzel et al., 2014) have been successfully used to determine several major and minor compounds in Hypericum plants. However, there is only sporadic information about the distribution and localization (or accumulation) of the aforementioned metabolites in different tissues of plants belonging to the genus Hypericum. For example, Hölscher et al. (2009) examined the localization of various secondary metabolites in the leaves and reproductive organs of H. perforatum and Hypericum reflexum by using matrix-free UV-laser desorption/ionization (LDI) imaging. Thunig et al. (2011) studied the localization of selected secondary metabolites in the leaves and petals of H. perforatum using indirect desorption electrospray ionization mass spectrometry imaging (DESI-MSI) via an imprint on a porous Teflon surface. This work was followed-up by Li et al. (2013) with direct DESI analyses of the leaves and petals of H. perforatum. Recently, we employed matrix-assisted laser desorption/ionization-HRMS (MALDI-HRMS) imaging to reveal, in high spatial resolution with limited sample preparation, the distribution, localization, and subtleties of hypericins and structural analogs and proposed precursors in the secretory as well as non-secretory tissues of three wild-growing Hypericum species (H. perforatum, H. olympicum, and H. patulum) (Kusari et al., 2015). More recently, we explored the interspecific variation in localization of hypericins and phloroglucinols as well as some new compounds using indirect DESI-MSI in the leaves of seventeen different in vitro cultured Hypericum species (Kucharíková et al., 2016). This study provided an extensive range of dataset on the localization of major secondary metabolites within leaves of such a broad spectrum of Hypericum species.

The present study builds upon and enlarges the aforementioned knowledge and provides a deeper insight to the occurrence and localization (accumulation) of the major compounds (Figure 1) within leaves of the same spectrum of seventeen Hypericum species adapted to outdoor conditions using MALDI-HRMS. To the best of our knowledge, this is the first report revealing the occurrence and distribution of important phytochemicals in the leaves of in vitro cultured Hypericum spp. adapted to outdoor conditions. Notably, our study offers a comparative and qualitative evaluation of the distribution of the target compounds within species cultivated in vitro and those adapted to outdoor conditions analyzed by MALDI-HRMS.

MALDI-HRMS analyses were performed to scrutinize the presence and localization of the important secondary metabolites in the leaves of seventeen selected Hypericum species, particularly focusing on the secretory tissues in comparison to the non-secretory tissues. The detailed spatial distribution of naphthodianthrone hypericin, its structural analogs protohypericin and pseudohypericin and potential precursor emodin, emodin anthrone, the phloroglucinol hyperforin, and the flavonoids quercetin, quercitrin, rutin and hyperoside and/or isoquercitrin, along with chlorogenic acid were mapped within fresh, healthy leaves of the seventeen Hypericum spp., classified in eleven sections. In addition to the dark and translucent glands, other glandular-like structures were investigated and compared to the surrounding non-secretory tissues, both from the adaxial as well as abaxial side of the leaves. By performing extensive mass spectrometric optimization based on our previous work (Kusari et al., 2015), we aimed to reduce the emergence of interferences and enhance the detection limits for the investigated compounds. Admittedly, due to the lack of a leaf negative control for each species coupled to the complexity of the biological tissues under investigation, low abundant non-continuous signals for compounds that do not have a reference standard, such as emodin anthrone, should be interpreted carefully keeping in mind possible tissue interferences from the leaf surfaces. Notably, low abundant non-continuous signals for emodin anthrone could not be subjected to MALDI-HRMS/MS. Nevertheless, we interpreted these selected data (e.g., for emodin anthrone) with great care, and emphasized on the “trends” while qualitatively working out the distribution pattern in various Hypericum spp. Furthermore, the flavonoids hyperoside and isoquercitrin are listed together in the results as it is known that certain Hypericum spp. might also contain isoquercitrin (Zorzetto et al., 2015); unfortunately, hyperoside and isoquercitrin could not be distinguished by means of MALDI-HRMS or MALDI-HRMS/MS.

In the present study, we performed extensive MALDI-HRMS analyses to investigate in detail the spatial distribution of the major compounds (Supplementary Figures 14, 15) in the leaves of a broader spectrum of seventeen Hypericum spp. classified into eleven different sections. We designed our study to ensure detailed mapping of target compounds in the secretory as well as non-secretory tissues of leaves from both the adaxial and abaxial sides. We analyzed each of the seventeen in vitro cultured Hypericum spp. adapted to outdoor conditions, section by section, in order to compare our results both from the intra-sectional as well as inter-sectional perspective.

The section Hypericum, represented by H. perforatum, H. maculatum, H. erectum, and H. tetrapterum generally comprises of species accumulating hypericins in specialized dark glands (Robson, 2003), which occur in the aerial parts of the plants (Zobayed et al., 2006). In our present study, hypericin was found to be localized primarily in the dark glands, mainly discerned from the abaxial side of the leaves in all the four representatives of the section, which is consistent with Robson's observation (Robson, 2003). Better detection of hypericins from abaxial side of leaves could be attributed to the absence of a thick epidermis that is present on the adaxial side of leaves, thereby increasing MALDI laser-mediated soft ionization of the compounds. Interestingly, hypericin and pseudohypericin were also detected in low abundance in some of the translucent glands and non-secretory tissues throughout the leaves of all the four species under our study. It is noteworthy that only a few papers dealing with the distribution and localization of secondary metabolites within the genus Hypericum have been published so far, and almost all of them are focused typically on H. perforatum. The presence of hypericin, pseudohypericin, and protohypericin was also observed in our earlier study in minute amounts around and/or just outside the dark glands in the adaxial side of leaves of H. perforatum (Kusari et al., 2015). However, we did not detect these compounds in the translucent glands or in the surrounding non-secretory tissues. Localization of hypericins in the leaves of H. perforatum was also realized by Thunig et al. (2011) and Li et al. (2013) using DESI-MSI analyses. Their results confirm the localization of hypericin in the dark glands of leaves and petals, and pseudohypericin and protohypericin in the dark glands of leaves of H. perforatum. More recently, we confirmed the localization and/or co-localization of hypericin, pseudohypericin and their protoforms in the dark glands of leaves of the four aforementioned in vitro cultured species belonging to the section Hypericum using indirect DESI-MSI (Kucharíková et al., 2016).

Emodin, the potential proposed precursor of hypericin (Mazur et al., 1992) was found to be localized in all of the four species under our study from the section Hypericum. However, emodin was found to typically accumulate in high abundance in the dark glands of H. tetrapterum. In our earlier study (Kusari et al., 2015), emodin was observed to be primarily accumulated in the dark glands, in addition to being dispersed at the vicinity of the glands in the non-secretory tissues, and even throughout other parts of leaves of H. perforatum. Our present study corroborates our earlier observation in that emodin is present not only in the dark glands and surrounding non-secretory leaf tissues, but also in morphologically uncharacterized reddish masses (Supplementary Figure 4) and translucent glands within the leaves. In H. perforatum, emodin anthrone was detected in some dark and translucent glands in lower intensities than emodin. However, in the other three plants, emodin anthrone could be detected sporadically only in the tissues where emodin was detected.

Within the section Hypericum, the phloroglucinol hyperforin could be detected only in H. perforatum after longer, optimized treatment of the tissues with laser. Hyperforin was found to accumulate in the translucent glands when discerned from the adaxial side of the leaves. Sporadic occurrence of hyperforin was also observed in the dark glands. Our results are in agreement with the work of Soelberg et al. (2007) who used UV-HPLC to detect hyperforin in the translucent glands as well as in minute amounts in dark glands of H. perforatum. In our present study, hyperforin was found to be absent in H. maculatum, H. erectum, and H. tetrapterum, which is in accordance with our earlier study (Kucharíková et al., 2016). This further demonstrates that the accumulation pattern of selected phytochemicals in these in vitro cultured Hypericum spp. remain unchanged after acclimatization of the plants to outdoor conditions.

All target flavonoids except rutin were found to be present in the section Hypericum. Rutin was detected only in the leaves of H. perforatum. Quercetin was present in all selected areas of detection; however, higher incidence of quercetin was observed compared both to its glycoside form (quercitrin) as well as its galactoside form (hyperoside). We made similar observations in our earlier study and the discrimination between quercitrin and hyperoside and/or isoquercitrin was made according to Kusari et al. (2015). Thus, far, it is compelling that higher signal(s) obtained for quercetin can result from ion source fragmentation, thanks to the presence of other minor flavonyl glycosides such as rutin or isoquercitrin. The important biosynthetic intermediate, chlorogenic acid, has been known to be present in several Hypericum species since several decades (Seabra and Alves, 1989). In the present study, we could detect chlorogenic acid in all parts of leaves of plants belonging to all species of the section Hypericum, corroborating earlier studies reporting its presence in H. perforatum (Franklin and Dias, 2011), H. maculatum (Kladar et al., 2015), and H. tetrapterum (Danova, 2010).

Our next set of plants belonged to the sections Drosocarpium, Oligostema, and Adenosepalum, all of which include species that demonstrated lack of hyperforin in the leaves. On one hand, absence of hyperforin was also demonstrated by us earlier in the leaves of in vitro cultured species H. rumeliacum (sect. Drosocarpium) and H. tomentosum (sect. Adenosepalum) (Kucharíková et al., 2016). On the other hand, H. humifusum (sect. Oligostema) contained traces of hyperforin in translucent glands and H. annulatum (sect. Adenosepalum) on the upper part of leaf margins (Kucharíková et al., 2016). This reveals that unlike the section Hypericum, the accumulation pattern of selected phytochemicals in in vitro cultured Hypericum spp. belonging to the sections Oligostema and Adenosepalum are prone to selective variation after acclimatization of the plants to outdoor conditions.

It was interesting to compare the occurrence and accumulation of emodin, the proposed precursor of hypericin, between different sections. In the section Hypericum, emodin was detected in the translucent glands and even in the non-secretory tissues in all hypericin-containing species. Strikingly however, in H. rumeliacum known to contain higher amounts of hypericin in leaves, emodin was found to localize solely in the dark glands. Localization of emodin anthrone followed a similar pattern as that of emodin, albeit in lower intensities than emodin. Moreover, the sections Androsaemum, Ascyreia, and Myriandra include species that are all known to lack dark glands on their leaves. Thus far, as expected, we did not detect hypericin, pseudohypericin, or protohypericin in their leaves. Surprisingly, both emodin and emodin anthrone were found to be present in all species of the section Ascyreia. These fascinating results open up the following two questions: Is emodin really the precursor of hypericin? What role does emodin and/or emodin anthrone play in Hypericum species that lack hypericin?

The last two analyzed species under our study were H. bupleuroides (sect. Bupleuroides) and H. pulchrum (sect. Taeniocarpum), which are typically characterized by formation of the dark glands only in reproductive organs. Given this fact, it was remarkable to find even sporadic and/or low abundance of the naphthodianthrones hypericin and pseudohypericin, along with emodin but not emodin anthrone, in some morphologically uncharacterized brownish or reddish spots or masses in the leaves of both species (Supplementary Figures 11–13). This points toward two main possibilities. Firstly, the evolution of the dark glands primarily in the reproductive organs is plausibly to protect the seeds and progenies, given that any tissue-specific regulation of gene expression in the flowers for production of naphthodianthrones should be similar to that in other tissues of the same plant. Secondly, the plausible trafficking of selected bioactive compounds in H. bupleuroides and H. pulchrum to the non-reproductive organs might be for possible eco-specific plant defense against biotic and abiotic factors. Considering the fact that the plants under our study were acclimatized to outdoor conditions where they are amenable to environmental stressors, trafficking of compounds for the purposes of chemical defense is realistic. The phloroglucinol derivative, hyperforin, was found to be absent in H. bupleuroides as well as in H. pulchrum, upholding our previous results (Kucharíková et al., 2016). Finally, the flavonoids showed variable distribution within the species without any dark gland in the leaves, further corroborating our previous study (Kusari et al., 2015). Especially, quercetin in both the species and rutin in H. pulchrum were detected in aforementioned reddish and brownish spots or masses, indicating possible damage of the leaves and the attempts to quench ROSs that are generated in the process of tissue death.

The utilization of MALDI-HRMS allowed us to determine the occurrence and localization of several important phytochemicals directly from the leaf surface of seventeen Hypericum spp. classified into eleven sections. We were able to simultaneously map the naphthodianthrones hypericin, pseudohypericin, protohypericin, their proposed precursor emodin as well as emodin anthrone, and the phloroglucinol derivative hyperforin along with four flavonoids (quercetin, quercitrin, rutin, and hyperoside and/or isoquercitrin) in the secretory and non-secretory leaf tissues of in vitro cultured plants adapted to outdoor conditions. Our broad dataset provides detailed elaboration of plausible biosyntheses and ecological relevance of target compounds in Hypericum spp. that have been first cultured in vitro, and then adapted to outdoor conditions. On one hand, our study reveals archetypal similarities in the accumulation pattern vis-à-vis ecological function of compounds in selected sections (e.g., Hypericum) when comparing in vitro cultured plants (Kucharíková et al., 2016) with those adapted later to outdoor conditions (present study). For an important plant genus such as Hypericum, particularly containing species such as H. perforatum (section Hypericum), our results demonstrate the possibilities of using in vitro cultured plants further adapted to field conditions for commercial purposes. Certainly, further fundamental research on the feasibility and cost-effectiveness of such an approach should be pursued. On the other hand, our results reveal distinct differences between in vitro cultured plants belonging to selected sections (e.g., Bupleuroides and Taeniocarpum) with those adapted later to outdoor conditions, when comparing the occurrence and localization of our target compounds. In practical terms, this can be expected given the fact that the plants growing outdoor are incessantly frazzled by biotic and/or abiotic environmental factors, which in turn activate trafficking of compounds to the tissue (site) of stress for plant chemical defense responses. Admittedly, the well-known facet of phytochemical variation in different Hypericum plant tissues and tissue parts concomitant to morphological, genetic, developmental, population, and environmental factors must be taken into account when evaluating and comparing any data. Nevertheless, the present study exemplifies how MALDI-HRMS can reveal a reliable and robust “pattern” of localization and qualitative abundance of target compounds in selected tissues of various Hypericum species.

SK, MS, and EČ conceived the project, SK and EČ designed the experiments, AK prepared the plants for analyses, SK and SS performed MALDI-HRMS analyses, AK, SK, SS, and EČ analyzed and interpreted the data, AK and SK wrote the manuscript. All co-authors read and approved the manuscript.

The present research was supported by the Slovak Research and Development Agency APVV-14-0154 and the Scientific Grant Agency of Slovak Republic VEGA 1/0090/15. SK and MS thankfully acknowledge the Ministry of Innovation, Science, Research and Technology of the State of North Rhine-Westphalia, Germany, and the German Research Foundation (DFG) for granting a high-resolution mass spectrometer and the MALDI imaging high-resolution mass spectrometer.