Pathway-focused bioassays and transcriptome analysis contribute to a better activity monitoring of complex herbal remedies

Herbal drug combinations have long been used in traditional medicinal concepts, and multicomponent preparations, such as plant-derived pharmaceuticals, botanicals and dietary supplements, are becoming increasingly important in modern medicine and lifestyles. The concept of effect potentiation by synergistic actions has opened new approaches for phytopharmaceutical and nutraceutical product development, and suggests the (re-)utilization of some traditional formulations for therapeutic issues. However, this development is hampered by the difficulties that accompany the assessment of molecular mechanisms and pharmacological actions of multicomponent preparations that contain an enormous chemical diversity of substances [1–3].

One major beneficial effect of many multicomponent preparations is their antioxidant activity. Reactive oxygen species (ROS) are involved in the development and the progression of diseases such as cardiovascular disorders [4], neurodegeneration [5] and atherosclerosis [6]. The capacity of dietary antioxidants to modulate the cellular redox balance is discussed to be favourable in both the treatment and prevention of ROS-related disorders [7, 8]. An antioxidant capacity may not only originate from the direct radical scavenging activity of a compound, but also from indirectly induced effects such as interference with the endogenous antioxidant machinery, the transcription induction of detoxifying and cytoprotective enzymes or the increase of immunoprotective mechanisms [9, 10]. Several of these actions are mediated by the nuclear-factor-E2-related factor (Nrf)-2 that transactivates the transcription of protective genes upon binding of antioxidant response element (ARE) promoter sequences to maintain homeostasis during oxidative stress [11]. However, an extensive input of antioxidants can also promote adverse responses such as allergies or toxicity-related phenomena [12, 13].

Studies into the interference of phytochemicals with cellular signaling pathways and gene expression have revealed many examples of potential biological activities that go far beyond an antioxidant action [14, 15]. Further, the use of large-scale data acquisition strategies, such as gene expression analysis, is suggested to be helpful in deciphering complex molecular information flows [16–18]. Currently, DNA microarray technology is the most widely used genome-scale data in pharmacological and clinical research. Transcriptomics allows the simultaneous detection of multiple transcriptional events in a non-biased manner, enabling standards for analysis and interpretation to be continuously developed [19]. Microarray expression data may provide key insights into gene function and interactions within and across metabolic pathways. However, it appears necessary to go beyond simple clustering by integrating data into biological interaction databases [20].

In general, the therapeutic efficacy of botanical multicomponents is suggested to be based on the interference of a multitude of phytochemicals with multiple cellular targets [21, 22]. This enormous complexity of interactions represents a challenge for activity monitoring. Often, the active principles of such remedies cannot be identified, as the pharmacological effects are suggested to be generated from the joint actions of several substances. Thus, reductionist and hypothesis-driven research strategies may prove insufficient to deal with such pleiotropic activity profiles. The combination of pathway or molecule focused and large-scale data assessment strategies, such as “omics” technologies, has proved useful in monosubstance and drug combination activity assessment and is also considered helpful for multicomponents [23–25].

Based on these assumptions, we attempted to explore the mode of action of the Indo-Tibetan remedy Padma 28 at the molecular and transcriptional level in human hepatoma HepG2 cells [26]. Padma 28 is a polyherbal preparation containing a variety of secondary plant substances with potential bioactivity, such as essential oils, flavonoids, tanning agents and sesquiterpenes. It contains 20 different powdered plants including natural camphor and calcium sulphate. The formulation has been subjected to extensive in vitro and in vivo studies and has been approved by the Swiss drug authorities (Swissmedic, Nr. 58436) for symptoms associated with circulatory disorders such as tingling sensation, formication, feeling of heaviness and tension in the legs and arms, numbness of the hands and feet and calf cramps [27]. Studies have been performed where Padma 28 was applied in the treatment of intermittent claudication, a hallmark of peripheral arterial occlusive disease (PAOD) [28, 29]. HepG2 cells were used because the liver is the main organ for drug metabolism after the ingestion of compounds and because of their suitability for studying the mechanism of action of drugs, dietary genotoxicants, as well as the cytoprotective and antigenotoxic activities of agents [30, 31]. Furthermore, this cell line is useful for the evaluation of pro- and antioxidant agents as they express many related enzymes such as Mn-superoxide dismutase, catalase, glutathione reductase and thioredoxin reductase [32, 33]. A HepG2 derived reporter gene cell line, CellSensor® ARE-bla HepG2, has been applied to investigate ARE-mediated transcriptional response.

In traditional medical systems, combinations of herbs and natural materials are often used in the treatment of multifactorial diseases [35]. However, evaluation of the molecular mechanisms modulated by complex multicomponent preparation in living cells is often limited because of the chemical diversity and multiple effects on a variety of molecular targets. Functional redundancy and regulatory loops add further complexity to the interaction networks responsible for biological actions [36, 37]. Re-establishment of a healthy state following disease development requires interventions on multiple targets at different levels [38–40].

Here, we employed the effects of the well-studied Indo-Tibetan polyherbal remedy Padma 28 on HepG2 cells using a microarray approach to gain deeper insights into multi-target effects of polyherbal preparations.

We used the ORAC method to reveal a peroxyl-radical scavenging activity of 2006 ± 235 μmol TE/g, which is much higher than that of other herbal preparations. Liao et al. reported an ORAC range of 40–940 μmol TE/g for 35 herbs used in traditional Chinese medicine, with a further eight showing antioxidant capacities of 1000–1500 μmol TE/g and only two approaching the level of Padma 28 [41]. Of note, the antioxidant capacity of herbal preparations may depend not only from the plant species used, but also from other factors e.g. the extract processing procedure or the storage under non oxidizing conditions. Also, in cultivated HepG2 cells, Padma 28 ethanolic extract showed a considerable antioxidant capacity by dose-dependently decreasing peroxyl-radical-induced intracellular ROS levels. These results are consistent with previous reports about the concentration dependent antioxidant and pro-oxidant ability of Padma 28 [42, 43]. Beside peroxyl radical, other ROS such as superoxide, hydrogen peroxide, hydroxyl and alkoxyl radicals, as well as reactive nitrogen species (RNS) are produced under (patho-)physiological conditions and may be neutralized by antioxidant compounds [44, 45].

To investigate genome-wide transcriptional alterations in treated HepG2 cells, we used a concentration of the extract equal to the IC50 value of growth after 72 h (218 μg/ml), but limited the incubation period to 18 h, where cell viability was not yet affected (data not shown). At this concentration, AAPH-induced ROS level in HepG2 cells was significantly decreased. A total of 578 genes were more than two-fold differentially regulated following Padma 28 treatment compared to the solvent control. However, the extent of regulation was mostly moderate and only 24 genes showed a four-fold or higher increase in gene expression, whereas nine were down-regulated at least four-fold. Notably, despite the weakness of regulation, the gene expression change of 12 selected transcripts was confirmed by qPCR in five independent experiments, implicating the robustness of the experimental design. According to the biological response fingerprinting (BioReF) approach of Rong et al. [46], this set of validated transcripts may be used as a signature pattern, which may be specific for the biological response of Padma 28 in HepG2 cells. Thus, qPCR analysis of marker genes might serve as a sensitive and cost-effective method for quality assessment and batch-to-batch comparison of polyherbal formulations. However, these results should be confirmed by further experiments using different incubation times, treatment concentrations and other cell models.

One of the major findings of this microarray analysis was the differential expression of phase I and II metabolic enzymes upon treatment of HepG2 cells with Padma 28. The CYP superfamily plays an essential role in detoxification phase I metabolism reactions of a wide range of xenobiotics including therapeutic agents, food additives, cruciferous vegetables and herbal components [47–50]. The majority of CYP-mediated drug biotransformation reactions are catalysed by CYP1A2, 2C9, 2C19, 2D6 and 3A4, and, to a limited extent, CYP2C8, 2B6 and 3A5 [47, 51]. Other CYP isoenzymes such as CYP1A1, 1B1, 2A6, and 2E1 are modulated by only a few, if any, therapeutics [51].

In the present study, the isoenzyme CYP1A1 was differentially expressed to the greatest extent. It is of a particular interest because of its implication in the metabolic activation of some exogenous pro-carcinogens or endogenous molecules [52]. Interestingly, CYP1A1 also exerts important functions in the detoxification of environmental carcinogens and the metabolic conversion of dietary compounds into substances with cancer preventive properties [53, 54]. As many phytochemicals are known to interact with CYP1A1 and AhR signaling, a deeper examination of the regulation of CYP1A1 by herbal components is of pharmacological relevance [55]. In general, CYP1A1 expression is mediated by the aryl hydrocarbon receptor (AhR) transcription factor. Additionally, AhR signaling is involved in many cellular processes such as cell growth, apoptosis, hypoxia signaling, cell adhesion and cell matrix metabolism [56–58]. AhR induces the expression of several phase II xenobiotic metabolizing enzymes, such as glutathione S-transferase alpha 1 (GSTA1), NAD(P)H: quinone oxidoreductase 1 (NQO1), microsomal glutathione S-transferase 1 (MGST1) or aldehyde dehydrogenase 1 family member L1 (ALDH1L1). Interestingly, in the present study, several AhR regulated enzymes were up-regulated by Padma 28 treatment in HepG2 cells (Table 2).

The effect of Padma 28 on cell growth and apoptosis has been illustrated by the >six-fold increase of nuclear protein 1 (NUPR1), which responds to proapoptotic stimuli and promotes cellular growth to counteract diverse injuries [59]. In addition, up-regulation of NQO1 by Padma 28 indicates an involvement of apoptosis and hypoxia signalling, since NQO1 is able to stabilize the p53 tumour suppressor protein, especially under oxidative stress conditions [60].

As expected from a pleiotropic acting antioxidant remedy, Nrf2-regulated stress response genes were also differentially expressed following Padma 28 treatment (Table 2). Nrf2 is a transcriptional activator for the ARE-mediated induction of antioxidant and phase II-detoxifying genes such as HMOX or glutathione S-transferase (GST), which protect cells and tissues from oxidative damage [11]. To validate this result, we showed that ARE-promoter driven β-lactamase expression was dose-dependently induced in a reporter cell line transactivation assay (Figure 5). We also observed a significant up-regulation of further antioxidant and detoxification genes such as HMOX, glutamate-cysteine ligase modifier (GCLM) and catalytic (GCLC) subunits, aldo-keto reductase family 1, member C1 (AKR1C1) and member C2 (AKR1C2) in the microarray analysis (Table 2). The induction of heme oxygenase expression upon treatment of HepG2 cells with Padma 28 could be confirmed at both the transcript (Table 1) and protein level (HO-1, Figure 6). The expression of heme oxygenase-1 is associated with therapeutic benefits in a number of pathologic conditions such as atherosclerotic vascular disease and inflammation [61]. A number of natural and synthetic molecules have been reported to induce HO-1 as additive mechanism responsible for their therapeutic effects [62].

The network visualization of the interaction of differentially expressed genes provides a link between transcriptional changes and their functional consequences in biological processes. We observed interference of the herbal preparation with molecules relevant to cardiovascular system development and function, cell-to-cell signaling and interaction and inflammatory responses (Figure 3 and Additional file 1: Table S1, network 7) within a centrally arranged subnetwork. Within this subnetwork, tumor necrosis factor (TNF) and interferon gamma (IFNG) have a central position. TNF is a principal mediator of the acute inflammatory response and coordinates the proliferation and protective functions of pathogen-reactive cells [63]. TNF is engaged in the activation of various pathways under both pathological and physiological conditions, and shows a remarkable functional duality, thus being also responsible for many of the systemic complications of severe infections [64]. The pro-inflammatory cytokine interferon gamma (IFNγ) plays a central role in the cellular immune response, as it induces several immuno-regulatory pathways and cellular response signals e.g. via the Janus activated kinase (JAK)-signal transducer and activator of transcription (STAT) pathway [65, 66]. Beside anti-microbial and inflammatory responses, IFNγ also interferes with growth suppression, cell death, tumor immunity and autoimmunity [66].

The strong interconnection of network 7 with other subnetworks might indicate that the represented interactions are involved in the regulation of some important cell behavior, as the removal of this part would affect the whole interaction network [38]. However, it should be kept in mind that transcriptional signatures are time and dosage dependent. Therefore, these results should be further validated using additional experimental approaches in order to design a more comprehensive activity model of Padma 28 that reflects also low-dose and long term effects. The application of “omics” technologies in multicomponent activity assessment is suggested to be most useful when applied in an iterative strategy in combination with focused bioassay [23]. Moreover, the interaction databases used for network generation might influence the result. Thus, our experimental findings must be interpreted with caution and a broader experimental design is necessary to use such data in the context of a comprehensive risk-benefit assessment.

In the present study, we show that the Indo-Tibetan polyherbal remedy Padma 28 displays both direct and indirect antioxidant capacities and exerts a variety of actions including the transcriptional activation of cytoprotective genes. The identification of a number of potential key molecules and signaling cascades that are differentially regulated upon cell treatment in vitro, supports the hypothesis for a pleiotropic mode of action. The results agree with several previous studies that described antioxidant, anti-inflammatory and anti-atherosclerotic properties using different experimental approaches.

The combination of molecular and pathway-based bioassays with unbiased transcriptome analysis has proved to be a valuable strategy in multicomponent activity assessment. Nevertheless, the selection of an appropriate experimental design, the cellular model system and the verification of findings in an iterative manner is necessary to generate valuable results.