Introduction

Natural products (NPs) are used in folk medicine since many thousands year, due to their biological origin, better ADME/T (absorption, distribution, metabolism, and excretion/toxicity) characteristics and high chemical diversity. Presently, NPs are considered as a valuable source of lead structures for new pharmaceutical agents. Over 70% of New Chemical Entities (NCEs) introduced into medical practice from 1981–2006 were obtained on the basis of NPs (Newman and Cragg, 2007). NPs have always represented a significant, though often underappreciated resource for the development of new medicines. Even now, in combinatorial chemistry era, drugs from plants or microbial origin account for more than 30% of worldwide sale. Moreover, NPs have been notably successful in the past in opening up new avenues of exploration and in producing entirely new therapeutic classes (Grabley and Thiericke, 1999; Sneader, 2005).

Virtual screening is of particular significance to understand pharmacological behavior and receptor interactions of plant compounds (Rollinger et al., 2008). A lot of empirical knowledge about pharmacotherapeutic properties of NPs is accumulated in Traditional Indian Medicine (TIM) “Ayurveda”, which is known earlier than 1000 years BC. Currently, this empirical knowledge could be analyzed using up-to-date computational and experimental approaches. Such studies could give information about the basic mechanisms of TIM actions, providing the basis for rational design of new medicinal plant combinations, individualization of therapy taking into account particular geno- & phenotypes, and identification of new lead compounds for future pharmaceuticals. Today hundreds of known pharmacological targets are used in medicine and NPs exhibits a pleiotropic action by interacting with these multiple targets. All these facts emphasizes that computer-aided methods could be extremely useful in pharmacological evaluation of NPs (Rollinger et al., 2009). The global research scenario suggests the use of virtual screening techniques for the discovery of bioactive phytoconstituents. Currently in India, herbal drug research has been focused to standardize herbal drugs via the principles of “reverse pharmacology” leading to the integration of Ayurveda with modern medicine (Patwardhan et al., 2004). But all these strategies end up with the identification of phytochemical lead confined to its ethnic use only. In this regard, an effort has been made to explore more comprehensive pharmacological profile of phytoconstituents by application of a computer program PASS (Prediction of Activity Spectra for Substances). The proposed in-silico approach extends further to reveal novel biological activities of selected phytochemical leads, their mechanisms and related side-effects. PASS applicability to NPs has been demonstrated in some investigations (Dembitsky et al., 2005; Zotchev et al., 2006). The current version of PASS predicts around 3750 pharmacological effects, biochemical mechanisms of action, specific toxicities and metabolic terms on the basis of structural formulae of drug-like substances with average accuracy ~95%. This can be further validated in in-vitro as well as in-vivo assays (Filimonov and Poroikov, 2008; Poroikov et al., 2009). This study involves the use of PASS for exploring the hidden pharmacological potential of selected traditional Indian medicinal plants based on their main phytoconstituents.

Materials and methods

Withania somnifera, Curcuma longa, Boerhaavia diffusa, Piper longum, and Alium sativum were selected for the study. Withanolide A from Withania somnifera (Malik et al., 2007), Curcumin from Curcuma longa (Jurenka, 2009), Liriodendrin from Boerhaavia diffusa, Piperine from Piper longum (Bhardwaj et al., 2002), and Allicin from Allium sativum (Zhang et al., 2008) (Table 1) were selected as main phytoconstituents based on literature reports. The structures of these phytoconstituents were obtained from Dictionary of Natural Products (DNP) and reported literature. An extensive literature search was carried out to collect information about the common biological activities of these plants and their individual phytoconstituents (Supplement Table 1) using various databases (PubMed, ScienceDirect, DNP, etc.).

Table 1 Names of plants and their main bioactive phytoconstituent selected for PASS prediction in this study
Full size table

The biological activity spectra of these phytoconstituents were obtained by PASS version (version 9.1, http://www.ibmc.msk.ru/PASS). This software estimates the predicted activity spectrum of a compound as probable activity (Pa) and probable inactivity (Pi). Prediction of this spectrum by PASS is based on SAR analysis of the training set containing more than 205,000 compounds exhibiting more than 3750 kinds of biological activities. Being probabilities, the Pa and Pi values vary from 0.000 to 1.000 and, in general, Pa + Pi ≠ 1, since these probabilities are calculated independently.

The PASS prediction results were interpreted and used in a flexible manner: (i) only activities with Pa > Pi are considered as possible for a particular compound; (ii) if Pa > 0.7, the chance to find the activity experimentally is high; (iii) if 0.5 < Pa < 0.7, the chance to find the activity experimentally is less, but the compound is probably not so similar to known pharmaceutical agents; (iv) if Pa < 0.5, the chance to find the activity experimentally is less, but the chance to find a structurally new compound, that is, NCEs is more (Marwaha et al., 2007). In this study, PASS prediction results (P2) were analyzed and compared with the reported activities of plant (P1), to obtain prediction coefficient (P) for each plant’s main phytoconstituent. Then the unpredicted but already reported activities were matched with the PASS-predicted biological spectrum (P3) of other known phytoconstituents in order to obtain the corrected prediction coefficient (P*) of the particular plant for further consideration of PASS applicability (Table 2; Supplement Table 1). Finally the unexplored but PASS predicted activities having score Pa > 0.7 for particular structure were listed as a hidden potential of the plant (Table 3).

Table 2 PASS prediction coefficient based on the main bioactive phytoconstituent from five selected medicinally active herbs
Full size table
Table 3 PASS predicted but not reported activities of selected phytoconstituents (hidden potential of medicinal plants)
Full size table

Results and discussion

Withania somnifera

Withania somnifera is commonly known as Ashwagandha or Winter Cherry (Andallu and Radhika, 2000). It is a green shrub (family Solanaceae) found throughout the dry parts of India, Baluchistan, Pakistan, Afghanistan, Sri Lanka, Congo, South Africa, Egypt, Morocco, and Jordan (Dafni and Yaniv, 1994). The practitioners of the traditional system of medicine in India regard Withania somnifera as the “Indian Ginseng” (Singh et al., 2001). Various parts of the plant have been used for centuries to alleviate variety of ailments (Bhattacharya et al., 1987, 2001; Kulkarni et al., 1998). Many pharmacological studies have been carried out in past to describe its multiple biological properties, many of which are indexed in the dictionary of natural products (Supplement Table 1). The major phytoconstituents of the plant are steroidal alkaloids and lactones, a class of constituents together known as withanolides (steroidal lactones with ergostane skeleton) (Elsakka et al., 1990). Various alkaloids present in the plant include withanine, somniferine, somnine, somniferinine, withananine, psuedo-withanine, tropine, psuedotropine, 3-α-gloyloxytropane, choline, cuscohygrine, isopelletierine, anaferine, and anahydrine (Kulkarni et al., 2008). Since withanolide A is the principal alkaloid present in the plant, therefore was selected to predict the biological spectrum of this plant by PASS. It was found that out of total 18 reported activities of the plant 13 were predicted by PASS for withanolide A (prediction coefficient 0.72) (Table 2). The remaining reported activities of Wathania somnifera which were not predicted by the PASS for withanolide A were correlated with the PASS predicted spectrum of the other reported phytoconstituents of this plant. It was found that out of remaining 05 activities all were predicted by PASS for the other phytoconstituents present in the plant further correcting the prediction coefficient to 1.00 (Table 2, Supplement Table 2).

Curcuma longa

Curcuma longa Linn., commonly known as Haldi, Turmeric, or Indian saffron, belongs to family Zingiberaceae. It is cultivated mostly in Ceylon, Belgium, Indonesia, France, and in many parts of India, especially in Bengal, Tamil Nadu, and Andhra Pradesh. Curcuma longa is a perennial herb with simple and large leaves. Its rhizomes and oil have great medicinal importance which has been summarized in Supplement Table 1. Curcuma longa contains a yellow compound curcumin, an essential oil (5%), an alkaloid, starch grains, curcuminoids, turmeric oil, turmerol, a coporioc acid, and veleric acid. The oil contains α- and β-curcumins, d-sallinene, α-phellandrene, cineol, zingiberene, small amount of sequiterpenes, α-and β-pinene, camphor and camphene (Jain et al., 2007). In this study, PASS predicted 21 activities for curcumin out of total 29 reported activities of the plant (prediction coefficient 0.71) (Table 3). While considering the PASS spectrum of other constituents of Curcuma longa, we found that out of remaining 08 activities 02 were depicted by its other constituents, therefore further correcting its prediction coefficient to 0.79 (Table 2, Supplement Table 2).

Boerhaavia diffusa

Boerhaavia diffusa L. (Nyctaginaceae), commonly known as “Punarnava” in the Indian system of medicine, is a perennial creeping herb found throughout the waste land of India (Rawat et al., 1997). It is used for its medicinal activities from antiquity, many of which have been experimentally validated (Supplement Table 1). Recently, we have evaluated its anticonvulsant activity in experimental animal models based upon its ethnomedical use (Kaur and Goel, 2009). Boerhaavia diffusa contains a large number of phytoconstituents such as, flavonoids, alkaloids, steroids, triterpenoids, lipids, lignins, carbohydrates, proteins, and glycoproteins (Sahu et al., 2008). In the present study liriodendrin was selected as reference for the plant in PASS assessment. It was found that out of total 22 reported activities of the plant 10 were predicted by PASS for liriodendrin (prediction coefficient 0.45) (Table 2). The remaining activities which were not predicted by the PASS for liriodendrin but were reported for the plant were correlated with the PASS spectrum of other constituents present in the plant. It was found that out of remaining 12 activities 07 were predicted by PASS for the other phytoconstituents present in the plant raising the prediction coefficient to 0.77 (Table 2, Supplement Table 2).

Piper longum

Piper species are distributed widely in the tropical and subtropical regions of the world and have multiple therapeutic utilities in different folk medicine. Piper longum Linn. is an important member of genus Piper, which is widely used in traditional medicine by the population of Asia and Pacific islands. Piper longum is implicated in several disorders like, gonorrhea, menstrual pain, tuberculosis, sleeping problems, etc. (Supplement Table 1) (Sunila and Kuttan, 2004; Park et al., 2007). Piperine is an amide derivative which was first isolated from piper species (Sunila and Kuttan, 2004) and is one of the major phytoconstituent of Piper longum (Bhardwaj et al., 2002), therefore was selected for the present study. In this study, PASS predicted 20 activities for piperine in comparison to 26 reported activities of the plant, giving a prediction coefficient of 0.76 (Table 2). Further prediction coefficient was raised to 0.80 by correlating the remaining reported activities with the PASS spectrum of other phytoconstituents of Piper longum (Table 2, Supplement Table 2).

Allium sativum

Allium sativum Linn. (family Alliaceae), commonly known as garlic, is one of the best-researched, best-selling herbal remedies and is commonly used as a food and a spice (Pittler and Ernst, 2007). Active phytoconstituents present in Allium sativum include enzymes (e.g., alliinase), sulfur-containing compounds (e.g., alliin), and compounds synthesized enzymatically from alliin (e.g., allicin). Traditionally, it has been employed to treat infections, wounds, diarrhea, rheumatism, heart diseases, diabetes, and many other, majority of them have been experimentally validated (Supplement Table 1). As allicin is the major phytoconstituent of the plant (Zhang et al., 2008), it was selected for the study. It was found that out of total 22 reported activities of Allium sativum, 15 were predicted by PASS for allicin (prediction coefficient 0.68) (Table 2). The remaining reported activities of Allium sativum that were not predicted by the PASS for allicin were correlated with the PASS-predicted activities for the other reported phytoconstituents of Allium sativum. It was found that out of remaining total 07 activities, 03 were predicted by PASS for the other phytoconstituents present in the Allium sativum (Supplement Table 2).

The results of this study justified the applicability of PASS program for the prediction of biological activities of a plant, based on its main phytoconstituent, as evidenced by an average prediction coefficient of 0.66 for the selected five plant’s phytoconstituent. Further, it was found that some of the reported activities of the plant were not predicted by the PASS. The probable reason for this may be that reports available in the literature consisted whole plant or extract and the PASS predictions were based on the structure of the main phytoconstituent only. As the plant or its extract contains a number of phytoconstituents that themselves shows varied biological activities, therefore the reported activities can be due to the other phytoconstituents present in the plant or its extract. In this regard, some of the reported but not predicted activities of the selected plants was justified by this approach by comparing these reported activities with the PASS spectrum of other phytoconstituents. This approach improved prediction coefficient for each plant raising the average corrected prediction coefficient to 0.83 (Table 2, Supplement Table 2).

Though plant phytoconstituents exhibit enormous structural diversity, only a small proportion of that diversity has been seriously explored for its pharmacological potential so far and there is, therefore, a little reason to believe that this potential has now run dry. As a phytochemical entity is capable of depicting a lot of biological activities but only a bit of it had been explored based upon its traditional use or serendipitously (Grabley and Thiericke, 1999; Elvin-Lewis, 2001). Moreover, it is also not feasible to explore all the biological activities based on hit and trial basis. PASS can be a possible approach to determine the novel possible activities of the existing phytoconstituents. In the study on examination of the PASS spectrum of the selected phytoconstituent of the selected plants, it was found that there were a significant number of unexplored pharmacological activities in each case. As the PASS-predicted pharmacological activities with a score of Pa > 0.7 have very high chances to be obtained experimentally, therefore only the unexplored pharmacological activities with a score of Pa > 0.7 have been summarized and listed as hidden pharmacological potential of these plant (Table 3).

From the results of this study, it is concluded that PASS predictions well correspond to the reported activities of the main and some other phytoconstituents of five selected medicinal plants. Previously unexplored but PASS-predicted activities provide the basis for evaluation of hidden potential of these medicinal plants.