Introduction

Sparstolonin B (SsnB) is a compound with unique anti-inflammatory effects and was first reported in 2011 [1]. SsnB is extracted from the Chinese herb Spaganium Stoloniferum (Rhizoma Sparganii), and its structure was determined using nuclear magnetic resonance (NMR) spectroscopy and X-ray crystallography [1]. Subsequently, Tang et al. developed a novel and straightforward method to obtain SsnB in high yield from inexpensive commercial compounds [2]. Toll-like receptors (TLRs) are an important family of receptors that constitute the first line of defense system against microbes [3], and function as the sensors of our immune system [4]. Liang et al. suggested that SsnB may serve as a promising precursor for the development of a new toll-like receptor (TLR)-2/4 antagonists for the treatment of inflammatory diseases [1], and this possibility was demonstrated consistently in further experimental verification [5, 6].

Based on existing studies, SsnB has shown various effects on different inflammatory diseases. It has exhibited multiple activities, including anti-tumor [7,8,9,10], anti-inflammation [6, 11,12,13,14,15,16,17,18,19,20,21,22,23], anti-oxidation [24, 25], anti-coagulation [26], anti-angiogenesis [27, 28], anti-fibrosis[29, 30], anti-migration [22], anti-virus [31], etc. Moreover, SsnB was observed to improve neurological outcomes following intracerebral hemorrhage in mice [6], and prevent lumbar intervertebral disc degeneration [24]. Interestingly, SsnB could exert bidirectional regulation effects, i.e., either anti-apoptotic [8, 24, 25, 32] or pro-apoptotic [33] effects in various diseases, depending on the different pathological microenvironment and the drug dose of SsnB. Despite the discovery of these effects, a comprehensive understanding of the effects of SsnB in different diseases is still lacking. Therefore, this study aims to explore the common effects and core related proteins of SsnB in different inflammatory diseases using data mining methods and network analysis of literature, which include frequency description, cluster analysis, association rule mining, functional enrichment, and protein–protein interaction mining.

The significance of this study lies in exploring the therapeutic potential of SsnB and assessing its therapeutic potential in related diseases, with the employment of various data mining methods and algorithms. The associations and interactions in the network interpret the action mechanisms of SsnB in different diseases, which may enlighten the potential leading to the discovery of new therapeutic pathways and pharmacological effects. Thus, data mining in this study provides a more comprehensive understanding and evaluation of the potential value and clinical applications of the novel herb-derived compound SsnB, which may provide a basis for further pre-clinical investigations and clinical trials.

Methods

Data Source and Screening Process

A comprehensive search of the main databases including Pubmed, Embase, Web of Science (WOS), and Cochrane Library was conducted to explore the existing literature on the research of SsnB. The library database was meticulously constructed using Endnote software, and any duplicate studies were eliminated. The remaining reports were then screened based on their titles and abstracts, followed by a thorough assessment of their eligibility according to the predefined inclusion and exclusion criteria. Full-text articles meeting the criteria were included for the final data analysis. To provide a visual representation of the study selection process, a PRISMA flow diagram was generated.

Inclusion and Exclusion Criteria

Inclusion criteria: First, the literature must conform to the standards of clinical or basic experimental research. Second, the study must have utilized SsnB as an interventional drug. Third, the research must have investigated the effects and mechanism of SsnB.

Exclusion criteria: The literature belongs to secondary literature, such as review, and/or the full text cannot be found.

Assessment of Literature Quality

The literature quality was assessed following the guidelines provided by ARRIVE, which consisted of 10 necessary entries divided into 23 sub-entries, as well as 11 recommended entries divided into 16 sub-entries. To evaluate the 27 included reports, we utilized the 10 necessary entries and 23 sub-entries as assessment criteria. Each report was assessed for each of the 23 sub-entries, with options such as “low risk,” “medium risk,” and “high risk” being chosen. The risk levels were defined as follows: if all the information required by a sub-entry was fulfilled, it was categorized as “low risk”; if no information fulfilled the sub-entry, it was categorized as “high risk”; and if there was partial or ambiguous information related to the sub-entry, it was categorized as “medium risk.” The number of reports conforming to each of the 23 sub-entries was calculated, and the percentage of compliance for each sub-entry and main entry was determined. Additionally, a bar chart illustrating the risk levels of each necessary entry was plotted.

Descriptive Analysis

To gain a comprehensive understanding of the study, we conducted frequency description on the 27 included reports using Microsoft Excel and IBM SPSS statistics. The frequency description included the analysis of disease distribution, research objects, pharmacological effects, and related proteins.

Network Analysis

The Gephi software was used to visualize the distribution of disease types and pharmacological effects of SsnB in the diseases from the 27 reports. Network analysis and visualization were performed, where the commonly investigated diseases and effects were represented as “nodes” connected by “lines” in the network. The size of nodes and thickness of the lines corresponded to the frequency of items mentioned in the reports.

Sankey Diagram for Path Analysis

Sankey diagram provides the illustration of different kinds of flows like studies on various diseases, and it provides the summary of all the paths involved in a process. It was considered as the best map depiction method between different domains for identifying the different paths. In Sankey diagram, thickness of the lines represent the quantity of this flow, and direction of the path toward the flow direction. The nodes represent the events of each path, and refers to the link that enters or exits the path. Thus, we employed the Sankey diagram plot for the components involved in the SsnB studies, which was composed by R software to display the connections among key elements of investigation sequencing.

Cluster Analysis

Cluster analysis is a method of data mining that categorizes similar samples or parameters together with specific algorithms. The aim of cluster analysis is to divide a dataset into clusters which categorize data more similar to each other within groups than to data in external groups. Thus, to obtain categorical relationships of high-frequency related proteins involved in SsnB studies, we employed IBM SPSS Statistics with the Intergroup linkage algorithm in the clustering method. The icicle chart and hierarchical plot were performed for depicting the cluster analysis.

Association Analysis with Apriori Algorithm

Apriori algorithm is the most popular algorithm for mining association rules, which enables finding the most frequent combinations in a database and identifies association rules between the items. Furthermore, we analyzed the commonly checked related proteins of SsnB using IBM SPSS Modeler from the 27 reports. The Apriori algorithm was applied to analyze the association rules of the included related proteins, and the support and confidence levels were obtained. Association rules serve to reflect the dependent or related information between items, while the support and confidence levels measure the strength of the associations. This approach allows for a better exploration of the hidden relationships between related proteins.

Statistical Analysis

Data collection and analysis were performed using various software tools, including IBM SPSS suites (SPSS statistics v21.0, SPSS clementine, v12.0; Inc. Chicago, Illinois, USA), Cytoscape (v3.10.0, http://cytoscape.org), Gephi (v0.10.1, http://gephi.org), and R (v3.6.2, http://www.r-project.org). Categorical variables were presented as frequencies and proportions (%). For cluster analysis, a dendrogram was generated using the average linkage rule between groups, and a threshold of 20 was selected as the rescaled distance between cluster combinations for cluster formation. In the association analysis for network construction, the apriori rule was employed to screen core related proteins, with a support setting of 25-50% and a confidence setting of 80-100%. For the functional enrichment analysis of gene ontology (GO), an adjusted P-value of less than 0.05 was considered statistically significant.

Results

Summary of Included Studies

Initially, a total of 149 reports related to SsnB research were identified from major databases, including Pubmed, Embase, and WOS. After removing duplicates, 77 reports underwent a preliminary screening based on title and abstract. Subsequently, 59 reports were selected for further eligibility assessment, and 28 reports were excluded based on predefined inclusion and exclusion criteria. Ultimately, 31 reports were deemed eligible for full-text analysis, with four reports being excluded at this stage. Consequently, a total of 27 experimental reports were included for data analysis. The literature processing workflow is depicted in Fig. 1 using the PRISMA flow diagram.

Fig. 1
figure 1

The PRISMA flow diagram for literature processing. The diagram displayed the literature searching, screening, and eligibility assessment and inclusion

Full size image

Quality Assessment of Animal Studies on SsnB

The quality of the included animal studies was evaluated according to the ARRIVE 2.0 guideline. The assessment of risk for each report based on the 10 necessary entries is summarized in Table 1, and the categorization of risk levels is illustrated in Fig. 2. Notably, a majority of the reports were classified as “high risk” in terms of randomization and blinding, while “medium risk” was predominantly associated with result reporting. The entries of research design, statistical methods, and experimental animals mostly received a classification of “low risk.” These findings provide an overview of the quality of the included animal studies on SsnB.

Table 1 The literature quality of animal experimental reports on sparstolonin B was evaluated according to the ARRIVE 2.0 guideline
Full size table
Fig. 2
figure 2

The assessment of literature quality according to the ARRIVE 2.0 guideline. The risk levels (high, medium, low levels) of included animal reports in this study were assessed by research design, sample size, inclusion and exclusion criteria, randomization, blinding, outcome measures, statistical methods, experimental animals, experimental procedures, and results items. Each report included in this study was assessed for each of the sub-entries, with options such as “low risk,” “medium risk,” and “high risk” being chosen. The risk levels were defined as follows: if all the information required by a sub-entry was fulfilled, it was categorized as “low risk”; if no information fulfilled the sub-entry, it was categorized as “high risk”; and if there was partial or ambiguous information related to the sub-entry, it was categorized as “medium risk”

Full size image

Characteristics of Frequency Distribution for Study Elements

In the initial analysis, the distribution of disease types, research objects, pharmacological effects, and proteins related to SsnB were examined (Tables 2, 3, 4, and 5). The frequency analysis of disease types revealed the inclusion of 13 different types in SsnB investigations, with the cardio-cerebrovascular system being the most commonly studied (23.53%) (Fig. 3A, Table 2). Both cells and animals were utilized as research objects, with mice being the most frequently employed animal model for SsnB studies (38.24%) (Table 3). Regarding pharmacological effects, a total of 12 categories were identified. The most prominent effect of SsnB was found to be anti-inflammation (53.85%), surpassing the anti-tumor effect (10.26%) and other effects (Fig. 3B, Table 4). In terms of related proteins, a comprehensive collection of 67 related proteins from the 27 studies was compiled, with the top four frequently investigated indicators being TLR4, interleukin (IL)-1β, IL-6, and tumor necrosis factor (TNF)-α (each accounting for 7.64%) (Table 5).

Table 2 Frequency distribution of disease types involved in sparstolonin B studies
Full size table
Table 3 Frequency distribution of research objects involved in sparstolonin B studies
Full size table
Table 4 Frequency distribution of pharmacological effects involved in sparstolonin B studies
Full size table
Table 5 Frequency distribution of therapeutic targets involved in sparstolonin B studies
Full size table
Fig. 3
figure 3

Distribution of disease types and pharmacological effects involved in sparstolonin B studies. The commonly investigated diseases (A) and pharmacological effects (B) were represented as “nodes,” which were connected by “lines” with the central compound of sparstolonin B in the network. The size of nodes and thickness of the lines corresponded to the frequency of items mentioned in the reports. Network diagram was constructed in the Gephi software with connections and degree parameters

Full size image

Sankey Diagram for the Investigation Sequencing of SsnB Studies

To gain insights into the interplay between key elements in SsnB studies, such as research objects, diseases, and pharmacological effects, we employed a sankey diagram to illustrate the sequencing of investigations (Fig. 4). In the case of the frequently studied cardio-cerebrovascular diseases in SsnB research, it is evident that investigations involved the use of cells, mice, and human samples, with pharmacological effects encompassing anti-inflammation, migration inhibition, anti-coagulation, and anti-apoptosis.

Fig. 4
figure 4

The Sankey plot showed the connections among key elements of investigation sequencing in sparstolonin B studies. The connective relationship among compound, research objects, dieases, and pharmacological effects were analyzed. The sankey plot was constructed by the R software, using the connections and degree parameters calculated. The thickness of lines represent the quantity of this flow, and the nodes represent the events of each path with the link that enters or exits the path. The flow diagram represents the sequential connections between the modules, and colors indicate different categories of diseases involved in this sequential diagram

Full size image

Cluster Analysis of the Related Proteins for SsnB

To establish homogeneous categories for the related proteins of SsnB, we conducted cluster analysis. By employing the average linkage method, we generated an icicle chart (Fig. 5A) and a hierarchical dendrogram (Fig. 5B). The results demonstrated that the 67 related proteins of SsnB could be classified into four primary categories at a distance scale of 20. These categories are as follows: (1) Nuclear factor kappa B (NF-κB); (2) IL-1β, IL-6, and TNF-α; (3) TLR2, TLR4, and myeloid differentiation primary response protein (MyD88); and (4) other related proteins. This finding underscores the significance of the TLR2/TLR4-MyD88-NF-κB-IL-1β/IL-6 and TNF-α pathways as related proteins for SsnB in inflammatory diseases.

Fig. 5
figure 5

The icicle chart (A) and hierarchical plot (B) for the cluster analysis of related proteins involved in sparstolonin B studies. The cluster analysis were performed to obtain categorical relationships of high-frequency related proteins of SsnB with the intergroup linkage algorithm. The rescaled distance at 20 was selected as the threshold for clustering in this study. These data indicated four categories as follows: (1) Nuclear factor kappa B (NF-κB); (2) IL-1β, IL-6, and TNF-α; (3) TLR2, TLR4, and myeloid differentiation primary response protein (MyD88); and (4) other related proteins. The icicle chart and hierarchical plot was constructed by SPSS statistics software, with the visualized clustered and hierarchical data to explore the relationships within data such as interactive features

Full size image

Association Analysis of Related Proteins for SsnB

To gain further insights into the interrelationships among the related proteins of SsnB, we initially constructed a network of related proteins, with the degree of each node calculated (Fig. 6A). Given that SsnB has been recognized as a TLR2 and TLR4 antagonist, we specifically examined the relationship between TLR2 and other high-degree nodes. Notably, TLR2 exhibited close connections with MyD88, ILs, caspases, and other factors (Fig. 6B). Subsequently, we highlighted the connections of related proteins with top links (no fewer than 23) in Fig. 6C, with particular emphasis on phosphatase and tensin homolog deleted on chromosome ten (PTEN), p21, signal transducer and activator of transcription 3 (STAT3), endothelin-1 (ET1), Ca2+, protein kinase C (PKC), Smad2, phosphoinositide-specific phospholipase C (PLC), and murine double minute 2 (mdm2).

Fig. 6
figure 6

The association network of related proteins involved in sparstolonin B studies. (A) The network plot with the connections among related proteins in the sparstolonin B studies. The dot colors indicate node degree for connections, and lines indicated the connections between these proteins. (B) The highlighted dots indicated the association between TLR2 and nodes with top degree. Network diagrams in (A) and (B) was constructed in the Gephi software with connections and degree parameters. (C) The association network plot with the connections among related proteins based on the top links (no less than 23). The dot colors indicate the individual proteins involved in this study, and lines indicated the connections between the highlighted proteins. The network diagram was constructed in the SPSS modular software

Full size image

Additionally, we employed the Apriori algorithm to screen for statistical associations among all related proteins, setting the support range at 25–50% and the confidence range at 80–100%. The association pairs with the highest support and confidence were as follows: (1) TNF-α, IL-6, and IL-1β; and (2) TLR2, TLR4, and MyD88 (support range 33.33–50%, confidence range 83.33–88.89%) (Table 6).

Table 6 Apriori association rule mining of the therapeutic targets involved in sparstolonin B studies
Full size table

Functional Enrichment of the Related Proteins for SsnB

The functions of the related proteins for SsnB were comprehensively analyzed using GO enrichment, which categorized the related proteins into biological process (BP), cellular component (CC), and molecular function (MF) items. Based on the gene ratio and adjusted P value, the top enriched items in BP primarily involved the response to lipopolysaccharide (LPS), response to molecule of bacterial origin, positive regulation of cytokine production, LPS-mediated signaling pathway, positive regulation of IL-6 production, regulation of inflammatory response, and positive regulation of TNF production (Fig. 7A and Table 7). To visualize the key genes involved in these top BP functions, we employed cnetplot (Fig. 7D). In terms of CC, the top enriched items included the inflammasome complex, membrane raft, membrane microdomain, external side of plasma membrane, endoplasmic reticulum lumen, RNA polymerase II transcription regulator complex, transcription repressor complex, and caveola (Fig. 7C and Table 7). Regarding MF, the top enriched items comprised cytokine receptor binding, cytokine activity, receptor ligand activity, signaling receptor activator activity, phosphatase binding, TLR-binding, DNA-binding transcription factor binding, cysteine-type endopeptidase activity involved in apoptotic process, RNA polymerase II-specific DNA-binding transcription factor binding, and growth factor receptor binding (Fig. 7B and Table 7).

Fig. 7
figure 7

GO analysis for the functional enrichment of related proteins involved in sparstolonin B studies. The bubble plots of biological process (BP) (A), molecular function (MF) (B) and cellular components (CC) (C) items involved in GO analysis. The bubble size indicate frequency of related proteins mentioned in the included studies, and bubble color indicated the statistical P value for the association between related proteins and GO items. (D) The cnetplot of top 10 items of cell signaling pathways, and lines indicate the related key genes in GO analysis. The bubble size indicate frequency of related proteins mentioned in the included studies. The bubble plots and cnetplot were constructed by R software

Full size image
Table 7 GO analysis for the functional enrichment of therapeutic targets involved in sparstolonin B studies
Full size table

Protein–Protein Interaction (PPI) Network and Hub Genes Selection

To explore the interactions among the related proteins of SsnB at the functional protein level, we utilized the STRING database to construct a PPI network. The network included genes that were identified in at least three individual studies, as depicted in Fig. 8A. Subsequently, we employed 11 different algorithms to select hub genes, and the top-ranked proteins were TNF-α, IL-1β, IL-6, protein kinase B (PKB/AKT1), peroxisome proliferator-activated receptor (PPAR)-γ, TLR4, C–C motif chemokine ligand (CCL)-2, and TLR2 (Fig. 8B and Table 8). These findings further support the central role of the TLR2/4-MyD88-ILs inflammatory pathways in mediating the potential therapeutic effects of SsnB.

Fig. 8
figure 8

The protein–protein interaction (PPI) network of top related proteins of sparstolonin B. The total network constructed by STRING database (A), and the hub protein network was selected by cytohubba from the PPI network (B), which indicates the key related proteins in the protein network of sparstolonin B. The network diagram was constructed by STRING database, and reconstructed by the Cytoscape software with the calculated degrees and connections

Full size image
Table 8 Hub protein network analysis by 11 algorithms based on the STRING network of therapeutic targets involved in sparstolonin B studies
Full size table

Discussion

SsnB, a versatile monomeric drug derived from Chinese herbal sources, was initially discovered for its anti-inflammatory effects. Subsequent investigations have shed light on its potential in targeting a range of diseases involving inflammation to a certain extent, including diseases of cardiovascular and cerebrovascular systems, liver system, respiratory system, digestive system, nervous system, orthopedic muscular system, immune system, and endocrine system, and tumors, skin diseases, congenital diseases, and systemic inflammatory diseases. Current research applications have primarily focused on the actions of SsnB within cardiovascular and cerebrovascular systems [6, 22, 23, 32].

The mechanisms underlying the effects of SsnB in cardiovascular and cerebrovascular diseases entail the modulation of key signaling pathways, such as the TLR4-MyD88-NF-κB pathway and the ERK1/2 and AKT pathways. These pathways play crucial roles in mitigating the risk of conditions such as atherosclerosis, myocardial ischemia-reperfusion injury, stroke, and cerebral hemorrhage. TLRs, specifically TLR2 and TLR4, are expressed by various immune cells and contribute to different aspects of the inflammatory response. TLR2 and TLR4 are particularly implicated in the initiation and progression of atherosclerosis, which is a known precursor to coronary artery disease. The recruitment of adaptor MyD88 leads to the activation of MAPK and NF-κB pathways downstream of TLRs [34]. Additionally, the ERK1/2 pathway, downstream of immune receptors such as TLRs, plays a critical role in triggering the expression of inflammatory genes in response to infection or tissue damage [35]. Targeting the ERK1/2 pathway in inflammation holds potential therapeutic benefits for a broad range of diseases, including cancer and neurodegenerative diseases. SsnB has been shown to exhibit a dose-dependent inhibition of the activation of TLR4-mediated stress-related transcription factor NF-κB. It also inhibits the recruitment of MyD88 by TLR4, resulting in a significant suppression of pro-inflammatory cytokines TNF-α, IL-6, and IL-1β in activated macrophages. Consequently, SsnB effectively suppresses the occurrence of inflammatory responses [15].

Besides the conventional inflammation, the inflammasomes are multiprotein complexes that are assembled by pattern-recognition receptors, and TLR signaling triggers the transcriptional activation of pro-IL-1β and pro-IL-18 that are processed into their active forms by the inflammasomes [36]. The effects of SsnB on inflammasomes activation also triggered great interest in elaborating anti-inflammatory functions in various diseases. Sun et al. tested the therapeutic effects of SsnB on collagen-induced rheumatoid arthritis (RA) and the association with nod-like receptor protein 3 (NLRP3) inflammasome, and they found that SsnB relived inflammation by inhibiting NLRP3 inflammasome and downregulating the TLR4-Myd88-NF-κB signaling pathway. Bose et al. indicated SsnB was effective in suppressing TLR4-induced NLRP3 inflammasome activation in astrocytes [11]. These data suggested that SsnB could be a modulator of TLR4-induced NLRP3 inflammasome activation. However, a greater understanding of the balance between beneficial and detrimental inflammasome activation is also needed [37] to get a balanced effects in reliving cardiovascular disease (CVD) by modulating inflammasomes. Moreover, the endothelial dysfunction is an initial and crucial mechanism for CVD. Our team investigated the anti-inflammatory role of SsnB in human umbilical vein endothelial cells, and our findings showed that SsnB can suppress endothelial cell inflammation [23].

Furthermore, SsnB has emerged as a promising potential therapeutic agent for a wide range of diseases that share common underlying mechanisms. For example, it has shown efficacy in alleviating symptoms of autoimmune systemic diseases such as multiple sclerosis and systemic lupus erythematosus. Additionally, SsnB has demonstrated potential in addressing systemic diseases, including skin pruritus, spinal cord injury, intervertebral disc degeneration, and spontaneous cerebral hemorrhage. These therapeutic effects are mediated through the inhibition of the TLR2/4-MyD88-NF-κB pathway, leading to improvements in disease symptoms and reductions in associated inflammatory markers [1, 6, 19, 20, 24].

In a comprehensive analysis utilizing 11 different algorithms, this study revealed that SsnB primarily related to proteins such as TNF-α, IL-1β, IL-6, AKT1, PPAR-γ, TLR4, CCL2, and TLR2. These findings provide further evidence highlighting the central functional role of the TLR2/TLR4-MyD88-ILs inflammatory pathway in elucidating the therapeutic effects of SsnB.

In summary, this study emphasizes the significance of the TLR2/TLR4-MyD88-NF-κB-IL-1β/IL-6/TNF-α pathways as crucial related proteins for SsnB in the context of inflammatory diseases. Ongoing efforts are focused on clarifying and refining the therapeutic pathways and related proteins associated with SsnB. We will continue to monitor the latest research findings to further explore and evaluate additional related proteins and pathways that contribute to a comprehensive understanding of the efficacy of SsnB in treating inflammatory diseases.

Limitations and Prospects

When conducting literature mining and network analysis, it is important to acknowledge certain limitations that may impact the interpretation of the results. First, it should be noted that the number of research papers suitable for meaningful analysis was relatively small, with a majority of them being preclinical studies. Upon further examination, it was observed that while some studies included in vivo experiments, the majority of them focused on in vitro experiments using SsnB.

Second, the statistical analysis of the risk assessment table indicated that entries categorized as “high risk” were primarily associated with random selection and blind methodology, while “medium risk” was predominantly related to result reporting. These risk assessments suggest that the quality of the reviewed literature may not have been optimal, and researchers may have unintentionally introduced biases in the conduct and reporting of results due to perceived positivity within the trial group. Therefore, it is necessary to enhance the quality of animal research in future studies.

Furthermore, it is crucial to increase the number of in vivo experiments to facilitate a deeper exploration of relevant animal models and potential translation into clinical practice. This approach is expected to yield significant advancements in the field of clinical application research.

Last, although TLRs are among the ideal targets for exploitation in immunotherapy as central components of both innate and adaptive arms of the immune system [38], their biology still needs to be better understood in the context of target diseases. Hopefully, more selective antagonist on each TLR or the downstream targets on TLR signaling may achieve better clinical outcome in the future.

Conclusion

The results of this study shed light on the mechanisms and targets involved in the effects of SsnB in various systemic diseases. Notably, they underscore the importance of the TLR2/TLR4-MyD88-NF-κB-IL-1β/IL-6/TNF-α pathway in mediating the effects of SsnB in inflammatory diseases. Although the precise pharmacological activity of SsnB remains incompletely understood, ongoing research holds immense potential for uncovering its vast therapeutic potential. We will continue to closely monitor the progress of SsnB research.