Abstract

   Astragaloside IV (AGS-IV) is a main active ingredient of Astragalus
   membranaceus Bunge, a medicinal herb used for cardiovascular diseases
   (CVD). In this work, we investigated the therapeutic mechanisms of
   AGS-IV at a network level by computer-assisted target identification
   with the in silico inverse docking program (INVDOCK). Targets included
   in the analysis covered all signaling pathways thought to be implicated
   in the therapeutic actions of all CVD drugs approved by US FDA. A total
   of 39 putative targets were identified. Three of these targets,
   calcineurin (CN), angiotensin-converting enzyme (ACE), and c-Jun
   N-terminal kinase (JNK), were experimentally validated at a molecular
   level. Protective effects of AGS-IV were also compared with the CN
   inhibitor cyclosporin A (CsA) in cultured cardiomyocytes exposed to
   adriamycin. Network analysis of protein-protein interactions (PPI) was
   carried out with reference to the therapeutic profiles of approved CVD
   drugs. The results suggested that the therapeutic effects of AGS-IV are
   based upon a combination of blocking calcium influx, vasodilation,
   anti-thrombosis, anti-oxidation, anti-inflammation and immune
   regulation.

Introduction

   Astragalus membranaceus Bunge, the dried root of a low shrub, has been
   widely prescribed in traditional Chinese medicine (TCM) for the
   treatment of cardiovascular disorders. Over 25% of the 90 recipes in
   the 2010 edition of the Chinese Pharmacopoeia for cardiovascular
   diseases (CVD) contain Astragalus membranaceus Bunge.

   Astragaloside IV (AGS-IV) is one of the main active compounds of
   Astragalus membranaceus Bunge [37][1]. Experimental studies from
   several laboratories, including ours, have provided abundant evidence
   demonstrating the explicit cardiovascular-protective effects of AGS-IV
   [38][2]–[39][6]. In vivo animal studies have shown that AGS-IV is
   protective against isoproterenol-induced myocardial injury [40][3] and
   isoproterenol-induced cardiac hypertrophy [41][2], elevates coronary
   blood flow and reduces the size of myocardial infarcts after coronary
   occlusion [42][4]. In vitro experiments suggested that AGS-IV could
   improve post-ischemic heart function and ameliorate reperfusion
   arrhythmias [43][4], attenuate hypoxia-induced cardiomyocyte damage
   [44][5], and relax smooth muscle in the aorta from both normal rats and
   stroke-prone spontaneously hypertensive rats [45][6]. AGS-IV could also
   enhance the activity of antioxidant enzymes [46][2], [47][4], [48][5],
   [49][7]; reduce the levels of phenylephrine and angiotensin II [50][6],
   block calcium influx and intracellular calcium release, and stimulate
   the NO–cGMP pathway [51][8]. AGS-IV also exhibits significant
   anti-inflammatory effects in vivo [52][9].

   Dysfunctions of multiple genes and/or their products are implicated in
   the pathogenesis of complex chronic diseases (e.g., CVD)
   [53][10]–[54][13]. Thus, targeting the entire network should be much
   more effective than targeting a single protein [55][14], [56][15].

   In this study, we attempted to investigate the therapeutic mechanism of
   Astragaloside IV against CVD using a network-based methodology that
   integrates data of drugs, targets and pathways. The in silico program
   INVDOCK [57][16] was used to search for putative binding sites for
   AGS-IV in the 3D structures of the proteins in the signaling pathways
   known to be affected by FDA-approved drugs for CVD. This analysis
   revealed 39 putative targets of AGS-IV. Experiments at the protein
   level were carried out to validate three of these targets. Experiments
   were also carried out in cultured cardiomyocytes to examine protective
   action of AGS-IV at the cellular effect level. The protein-protein
   interactions, drug-target and target-pathway association networks of
   the 39 putative targets were then constructed to probe for
   relationships between known drugs, targets and pathways and to evaluate
   how collective actions arising from these relationships contribute to
   the therapeutic effects of AGS-IV.

Results

Identification of key pathways and candidate protein targets associated with
CVD therapy

   CVD is a series of diseases with complex etiology involving multiple
   biological processes or pathways [58][17], [59][18]. Using existing
   cardiovascular drugs as a starting point, we applied a network-based
   approach to probe possible key pathways involved in the therapeutic
   actions. The DrugBank [60][19] includes 174 FDA-approved small
   molecular CVD drugs that act on 188 protein targets. By mapping these
   targets onto the KEGG pathways [61][20], it was found that 131 of the
   188 targets appear in a total of 120 pathways, corresponding to 133
   FDA-approved small molecular CVD drugs. The target-pathway network and
   drug-pathway network were constructed to reflect the target-pathway and
   drug-pathway interactions. The distribution of pathway nodes in both
   target-pathway and drug-pathway networks obeyed power laws [62][21]
   ([63]Figure S1), suggesting that a large number of these pathways are
   influenced by only a small number of CVD drugs and targets, whereas
   most CVD drugs and targets operate in only a few pathways, which could
   be the key pathways involved in the therapy of cardiovascular diseases.
   We applied pathway enrichment analysis [64][22] to identify these key
   pathways.

   For each of the 120 pathways owning CVD drug targets, we computed the
   p-values according to the definition of pathway enrichment,
   respectively. As there are different statistic methods to calculate
   this value, we adopted three strategies: distinct protein enrichment of
   CVD drug targets, protein node enrichment of CVD drug targets and CVD
   drug enrichment (see Methods section for details). The analysis
   generated 33 pathways (28% of all pathways) with at least one
   p-value<0.05 ([65]Table S1). The significant differences in the
   percentage of distinct protein targets and regulating points between
   all proteins and CVD drug targets on the target-enriched pathways are
   shown in [66]Figure S2. Enrichment type III pathways are regulated by
   many more CVD drugs than other categories of drugs ([67]Figure S3). In
   total, 129 drugs and 103 targets for CVD were associated with these 33
   pathways, making up 97% and 79% of all CVD drugs and targets that are
   related to KEGG pathways, respectively. Therefore, it is reasonable to
   regard these 33 pathways as key pathways involved in the therapy of
   cardiovascular diseases. All together, we identified 1,619 proteins
   involved in these 33 pathways. These proteins could potentially
   contribute to cardio-protective action of CVD drugs. Hence they were
   used as candidate protein targets in our study.

Putative protein targets for AGS-IV

   Small-molecule drugs generally function by binding to proteins and/or
   nucleic acids. Ligand-protein inverse docking is one approach to search
   for multiple putative protein targets. Base on the 33 key
   CVD-associated pathways identified by enrichment analysis, we used
   INVDOCK [68][16] to identify putative protein targets for AGS-IV.

   The candidate 3D dataset included all known information for the 1,619
   proteins in the 33 key pathways (total entries: 3,475 PDB). The 3D
   structure of AGS-IV ([69]Figure S4) was entered into INVDOCK to search
   for potential interaction. The analysis identified 39 distinct proteins
   (54 PDB) as putative targets for AGS-IV ([70]Table S2).

   Putative complexes of AGS-IV with calcineurin (CN) and cyclophilin
   (Cyp) are shown in [71]Figure 1. The immunosuppressive drugs
   cyclosporin A (CsA) and FK506 bind to Cyp and FK506-binding protein,
   respectively, forming drug–protein complexes that in turn recruit CN
   and inhibit its activity [72][23]. Crystal structures of the Cyp/CsA/CN
   ternary complex are deposited in PDB database as entries 1mf8 and 1m63
   [73][23], [74][24]. INVDOCK removes CsA from the complex to generate an
   active cavity. The program then attempts inverse-docking with the small
   molecule of interest. Such an analysis indicated that AGS-IV could form
   a reasonable drug-protein complex with CN and Cyp at the same site as
   CsA ([75]Figure 1).

Figure 1. INVDOCK-predicted binding model of AGS-IV molecule (shown in ball
and stick model) with cyclophilin (Cyp) and calcineurin (Cn).

   [76]Figure 1
   [77]Open in a new tab

   CnA and CnB are the catalytic subunit and the calcium-binding
   regulatory subunit of calcineurin, respectively. In the protein
   secondary structure, red, blue and grey colors represent alpha-helices,
   beta-sheets and loops, respectively; for the ligand structure, red and
   grey colors correspond to oxygen and carbon atoms, respectively.

   The Gene Ontology (GO) terms of the 39 putative protein targets, their
   implications for CVD therapy, and corresponding references are listed