Integrated Network Pharmacology and Molecular Docking to Dissect the Molecular Mechanisms of Anti-Periodontitis of Minocycline

Qihong Qiu

About the author： Qiu Qihong， born in September 1988， female， Han nationality， native place of Jieyang City， Guangdong Province， doctor， attending physician， engaged in systematic diagnosis and treatment of periodontal disease.

Abstract：Background： Minocycline is one of the most widely applied tetracycline analogues， which has been used to treat periodontitis for decades. Nevertheless， the targets and mechanisms of minocycline in treating periodontitis are still not completely understood. Therefore， the purpose of this study was to explore the targets and mechanisms of minocycline in treating periodontitis by integrating network pharmacology and molecular docking analysis.

Methods： The targets related to periodontitis were screened from DisGeNET database. The targets of minocycline were retrieved from Drugbank and Swiss Target Prediction databases. A Venn diagram was used to visualize the common targets of periodontitis and minocycline. The protein–protein interaction （PPI） network of common targets was established by using STRING database. Additionally， Gene Ontology （GO） and Kyoto Encyclopedia of Genes and Genomes （KEGG） were used to cluster the common targets. The network among periodontitis， minocycline， targets of periodontitis， and minocycline as well as KEGG pathways were constructed by Cytoscape. The key targets were also sought by Cytoscape and docked with minocycline.

Results： A total of 41 targets of minocycline were obtained and the common targets of minocycline and periodontitis were CASP1， CASP3， MAPK1， MAPK14， MMP2， MMP9， MMP13， IL1B， ALOX5， ESR2， NOS2， VEGFA. Enrichment analysis revealed that minocycline may treat periodontitis by regulating some biochemical reactions which were involved in immuno-inflammation， apoptosis， fibrosis， and angiogenesis. Molecular docking also verified that minocycline had affinity to MAPK1， CASP3， MAPK14， and VEGFA.

Conclusion： Minocycline exerts the therapeutic effect on periodontitis by acting on MAPK1， CASP3， MAPK14， and VEGFA.

Key words： periodontitis; minocycline; network pharmacology; protein–protein interaction; enrichment analysis; molecular docking

【中圖分类号】R781.4+2             【文献标识码】A             【文章编号】2107-2306（2021）07--05

1.  Introduction

Periodontitis is a chronic inflammatory disease caused by plaque microorganism and host immune regulation， which can result in destruction of periodontal tissue， tooth loosening， and even tooth loss eventually. Periodontitis is one of the most common diseases in the oral cavity. For example， nearly 50% of adults suffer from periodontitis in the United States [1]. However， the pathogenesis of periodontitis is still not completely understood. For one thing， bacterial etiology is well established and removal of plaque microorganisms is an effective treatment for periodontitis [2]. For another， some people are more susceptible to periodontitis probably because of carrying some risk genes [3-5]. Therefore， genetic research and drug therapies are appealing for refractory or aggressive forms of periodontitis.

Minocycline is one of the semi-synthetic tetracycline analogs that has been used for periodontal therapy for about 40 years [6]. In periodontal treatment， minocycline displays its powerful antibacteria function， especially effective to reduce anaerobic subgingival organisms [7]. For the patients with periodontitis who received nonsurgical periodontal therapy in combination with minocycline systemically， the plaque index （PI）， sulcus bleeding index （SBI） and probing depth （PD） were reduced more significantly than those receiving just nonsurgical periodontal therapy [8]. Pandit et al. also found that treatment with scaling and root planning （SRP） coupled with minocycline microspheres in localized residual pockets improved PD and clinical attachment level （CAL） in patients with periodontitis compared to SRP alone [9]. In recent years， it has been reported that minocycline possesses many other effects， including anti-inflammation， anti-apoptosis， oxidation resistance， and immunoregulation [10]. As a result of the various effects of minocycline， its detailed mechanisms for treating periodontitis have not been completely illustrated.

Network pharmacology is a new discipline based on the theory of systematic biology， which analyzes the network of biological systems and selects specific signal nodes to optimize the therapeutic effect of drugs [11]. Molecular docking is a method of computer-aided drug research， which is based on the characteristics of the receptor and the interaction between the receptor and the drug molecule [12]. To our knowledge， the present study is the first to integrate the network pharmacology method and molecular docking analysis to investigate the predicted targets and signaling pathways of minocycline against periodontitis. We hope this study will provide bioinformatics data that facilitates drug development and clinical researches of periodontal therapy. The flowchart of the design was shown in Fig. 1.

2. Methods

2.1.  Screening of disease targets and targets of minocycline

DisGeNET database （https：//www.disgenet.org/） was used to filter out periodontitis-associated targets. Additionally， all predicted targets of minocycline were screened out from the databases of Swiss Target Prediction （http：//www.swisstargetprediction.ch/） and Drugbank （https：//go.drugbank.com/）. Only target genes of Homo sapiens were included in this study. Subsequently， a Venn diagram was created by Venny 2.1 （https：//bioinfogp.cnb.csic.es/tools/venny） to visualize the amount of overlap between the genes related to minocycline and periodontitis-related genes. [13]

2.2.  Construction of protein-protein interaction （PPI） networks and cluster analysis.

The obtained intersection genes were uploaded into STRING 11.0 （https：//string-db.org/） to acquire the relationships of PPIs.13） Further， cluster analysis including Gene Ontology （GO） function and Kyoto Encyclopedia of Genes and Genomes （KEGG） pathways were analyzed with STRING 11.0. GO function consisted of cellular component （CC）， biological process （BP）， and molecular function （MF）. In addition， the common genes were uploaded to KEGG database （https：//www.genome.jp/kegg/）， and the predicted pathways were constructed.

2.3.  Topological analysis of minocycline against periodontitis and screening of the key targets

In brief， the targets of both periodontitis and minocycline along with their related signal pathways were imported into Cytoscape 3.8.2 to construct the network among periodontitis， minocycline， targets of periodontitis and minocycline as well as KEGG pathways [14]. Cytoscape is a powerful software employed to display and analyze networks graphically， which is widely used in the field of biological analysis [15].

The plug-in cytoHubba of Cystoscope 3.8.2 was used to screen the key targets. The common targets in the PPI network were ranked by the topological analysis method “Degree”. “Degree” means the number of edges connected to a node; the more edges connected to a node， the greater value of the degree it has [16]. The targets with the highest value of “Degree” were regarded as the key targets.

2.4.  Minocycline-target molecular docking

The operations of molecular docking are as follows： （1） The 2D structure of minocycline was determined from the Drugbank database. The 3D structure of minocycline was constructed by using the ChemOffice software. （2） The 3D structure of the key target protein combined with its original ligand were obtained from the RCSB PDB database （http：//www.rcsb.org/）. Their respective conformations of the key target protein and the original ligand were acquired by separating them with Pymol 2.4.1 software. （3） The processed protein （in PDB format）， the original ligand （in PDB format）， and minocycline （in mol2 format） were converted to PDBQT format with AutoDockTools 1.5.6. The AutoDockTools was also used to determine the size and the X-Y-Z coordinates of the grid box. We used a script to calculate the binding energy. （4） The docking conformation of minocycline and the key target protein was generated in Pymol 2.4.1. [17].

3.Results

3.1.  Identification of targets of minocycline and periodontitis in various databases

We found 956 disease targets related to periodontitis belonging to Homo sapiens in DisGeNET database for further research. A total of 41 drug targets of minocycline belonging to Homo sapiens were picked from Drugbank database and Swiss Target Prediction database. The principal members of drug targets of minocycline covered matrix metalloproteinases （MMPs）， mitogen-activated protein kinases （MAPKs）， cysteinyl aspartate specific proteinases and Protein kinase C （PKCs） and may play important roles in minocycline against periodontitis.

3.2.  Looking for common targets and establishing the PPI network

The Venn diagram showed that there were 12 common targets for the drug targets and the disease-targets （Fig. 2A）. The common targets were regarded as significant targets for the treatment of periodontitis. The PPI networks of the above 12 targets were established by introducing them to the STRING database （Fig. 2B）. A total of 12 nodes and 52 edges were found in the network. These common targets were CASP1， CASP3， MAPK1， MAPK14， MMP2， MMP9， MMP13， IL1B， ALOX5， ESR2， NOS2， VEGFA， including cysteinyl aspartate specific proteinases， MAPKs， and MMPs. These 12 targets were important targets for minocycline in periodontal therapy and were used in the following study.

3.3.  Cluster analysis

A total of 8 items of CC of GO enrichment were extracellular region， ficolin-1-rich granule lumen， secretory granule， secretory granule lumen， extracellular space， extracellular matrix， mitochondrion， intracellular organelle lumen （p<0.05） （Fig. 3A）. The top 10 items in ascending order of p-value of MF were endopeptidase activity， metalloendopeptidase activity， MAPK activity， MAPK kinase activity， cysteine-type endopeptidase activator activity involved in apoptotic process， cysteine-type endopeptidase activity involved in apoptotic process， catalytic activity acting on a protein， serine-type endopeptidase activity， identical protein binding， transition metal ion binding （Fig. 3B）. Similarity， the top 10 items of BP were cellular response to cytokine stimulus， cellular response to organic substance， cytokine-mediated signaling pathway， cellular response to oxygen-containing compound， response to lipopolysaccharide， response to oxygen-containing compound， cell surface receptor signaling pathway， cellular response to lipopolysaccharide， positive regulation of molecular function， response to lipid （Fig. 3C）.

On the basis of p-value， the top 10 pathways screened by KEGG analysis were relaxin signaling pathway， pertussis， IL-17 signaling pathway， AGE-RAGE signaling pathway in diabetic complications， proteoglycans in cancer， salmonella infection， endocrine resistance， TNF signaling pathway， toxoplasmosis， pathways in cancer （Fig. 3D）. The KEGG analysis revealed that minocycline treatment of periodontitis was strongly associated with relaxin signaling pathway （Fig. 4）.

3.4.  Integrated Network Construction

Periodontitis and minocycline were correlated with corresponding targets and the KEGG pathway. The top 10 pathways were used to construct a drug-target-pathway interaction network， which was visualized by Cytoscape. As shown in Fig. 5， the blue lines were mainly connected to the endocrine-related pathways， the red lines were connected to the pathways associated with infectious diseases， the green lines were connected to the inflammation-related pathways， and the purple lines were connected to the cancer-related pathways. Over all， the network analysis revealed that multiple targets of minocycline act synergistically in treating periodontitis. Taken together with the PPI network， CASP3， MAPK1， MAPK14， and VEGFA had the highest value of degree. Therefore， these four targets were considered as the key targets.

3.5.  Molecular docking analysis

Molecular docking analysis was employed to confirm whether minocycline played a crucial role in regulating the top four targets. The result showed that minocycline had affinity to the key targets. Minocycline and MAPK1 had the lowest binding energy， which suggested that they had the best binding effect. Minocycline formed a stable complex with MAPK1 by forming three hydrogen bonds with Ser153， Asn154， and Asp167 amino acid residues （Fig. 6A）. Table 1 displays the binding energy of minocycline and the four key targets.

4.Discussion

Minocycline is comprised of a four-ring nucleus attached to a variety of side groups （Table 1）. The dimethylamino on the upper half of the molecule is crucial for antibiotic properties [18]. Minocycline has the broad-spectrum antimicrobial activity mainly because it can inhibit bacterial protein synthesis by binding to the 30S ribosomal subunit [19]. Minocycline has a better lipophilicity than the other tetracyclines， making it have stronger antibacterial activity against gram-positive bacteria [20]. The lower half of the molecule is important for binding protein targets and has a close relationship with the its effectiveness [20]. The ability of binding to proteins can be strengthened by complexing minocycline with metal ions [21].

Compared with the first generation of tetracycline， minocycline has better pharmacokinetic characteristics when taken orally. It has a longer half-life period， better tissue permeability， and almost complete bioavailability， which makes it be absorbed and quickly and completely [22-24]. However， the side-effects of minocycline， such as hepatotoxicity， pigmentation and lupus-erythematosus-like syndrome cannot be neglected when it was used chronically. Compared with other tetracyclines， minocycline can cause a higher risk of lupus erythematosus-like syndrome and nonreversible pigmentation [23-25].

Even though Drugbank database has shown the targets of minocycline， none of the articles involved in the database dissected the targets and mechanisms of anti-periodontitis of minocycline. In recent years， more researches have paid more attention to the non-antibacterial activity of minocycline. It has been reported that minocycline is conducive to inflammatory diseases， periodontitis serving as an example [26-28]. The result of the Venn diagram showed 12 common targets of minocycline and periodontitis， which indicated minocycline’s anti-inflammatory and immunomodulatory effects. The results of PPI network showed that MAPK1， CASP3， MAPK14， and VEGFA were considered hub targets. The results of molecular docking also verified that minocycline played an important role in the regulation of these four hub targets. Minocycline prolonged the life-span of mouses suffering from Huntington disease by suppressing caspase-3 expression [29]. It suggested that minocycline may also arrest periodontitis by inhibiting caspase-3 activation and protecting the periodontal ligament cells against apoptosis. A study showed that minocycline inhibited p38 MAPK and ERK1/2 MAPK activation， resulting in the decrease of cytokines and chemokines in THP-1 cells when stimulated by LPS [30]. We speculated that minocycline may alleviate the inflammatory reaction of periodontitis by decreasing the phosphorylation of p38 MAPK and ERK1/2 MAPK. In addition， minocycline increased VEGFA protein and facilitated angiogenesis [31]. This study implied that minocycline may promote the repair of periodontal tissue by increasing VEGFA protein. In general， minocycline slowed periodontitis progression and promoted periodontal healing by modulating MAPK1， CASP3， MAPK14， and VEGFA expression. According to the results of molecular docking， minocycline can be structurally modified or new compounds can be designed so that the drug can form a better connection to the active sites， and then the drug activity will be tested. This study has certain directive significance for drug research and development.

GO enrichment analysis showed that minocycline modulated various proteolytic enzymes in our study. Taken together with the results of the PPI network， MMP-2， MMP-9， and MMP-13 were important targets of minocycline against periodontitis. The collagenase MMP-13 cleaves fibrillar collagens （types I and III） and gelatinases including MMP-2 and MMP-9 degrade basement membrane collagen （type IV） [32]. Studies revealed that MMP-2， MMP-9， and MMP-13 were inhibited by minocycline [33-34]， which suggested that minocycline had the potential to arrest periodontitis by preventing damage of periodontal tissue.

KEGG analysis showed that the relaxin signaling pathway was one of the main pathways in which the common targets were centralized. According to the KEGG database， there were several signaling pathways in relaxin signaling pathway， such as NFκB signaling pathway， MAPK signaling pathway， PI3K-Akt signaling pathway， and so on. In our study， relaxin signaling pathway involved VEGFA， ERK1/2 MAPK， p38 MAPK， MMP-2， MMP-9， MMP-13， and iNOS. In accordance with the above results， we speculate that the pathways of minocycline in the treatment of periodontitis are as follows. （1） NFκB signaling pathway： The study by Sarwar et al. revealed that activation of Relaxin Family Peptide Receptor 1 （RXFP1） led to increased nNOS-driven NO generation， along with VEGF and MMP-2 and 9 in human umbilical arterial smooth muscle cells （HUASMCs） and human cardiac fibroblasts （HCFs） [35]. VEGFA， MMP-2/9/13 and iNOS were involved in NFκB signaling pathway. Taken together with the results of GO enrichment analysis， minocycline may prevent the degradation of extracellular matrix through NFκB signaling pathway. （2） MAPK signaling pathway： Recent evidence also points to phosphorylation of ERK1/2 and PKA-induced activation of p38 MAPK [35-36]. Minocycline may treat periodontitis through MAPK signaling pathway.

In conclusion， the present study manifested that minocycline treated periodontitis by regulating apoptotic process and immuno-inflammatory responses probably. The crucial targets of minocycline against periodontitis may be MAPK1， CASP3， MAPK14， and VEGFA. Although we have made some progress in the present study， our study had several limitations. First of all， our results have not been experimentally verified. We will conduct further experiments to verify the results. Second， more comprehensive databases should be incorporated into our study to make the data of network pharmacological analysis more reliable. Third， the accurate therapeutic mechanism of minocycline was still not elucidated completely. A comprehensive understanding of minocycline and periodontitis needs further research.

References：

[1] Eke P.I.， Dye B.A.， Wei L.， Thornton-Evans G.O.， Genco R.J. Prevalence of periodontitis in adults in the United States： 2009 and 2010[J]. J Dent Res， 2012，91（10）：914-920.

[2] Manresa C.， Sanz-Miralles E.C.， Twigg J.， Bravo M. Supportive periodontal therapy （SPT） for maintaining the dentition in adults treated for periodontitis[J]. Cochrane Database Syst Rev， 2018，1（1）： D9376.

[3] Shi Q.， Cai C.， Xu J.， Liu J.， Liu H.， Huo N. Is there an association between IFN-γ +874A/T polymorphism and periodontitis  susceptibility？： A meta-analysis[J]. Medicine （Baltimore）， 2017，96（25）： e7288.

[4] Mashhadiabbas F.， Neamatzadeh H.， Nasiri R.， Foroughi E.， Farahnak S.， Piroozmand P.， Mazaheri M.， Zare-Shehneh M. Association of vitamin D receptor BsmI， TaqI， FokI， and ApaI polymorphisms with susceptibility of chronic periodontitis： A systematic review and meta-analysis based on 38 case -control studies[J]. Dent Res J （Isfahan）， 2018，15（3）：155-165.

[5] Salles A.G， Antunes L.， Carvalho P.A.， Küchler E.C.， Antunes L.S. Association Between Apical Periodontitis and TNF-α -308 G>A Gene Polymorphism： A Systematic Review and Meta-Analysis[J]. Braz Dent J， 2017，28（5）：535-542.

[6] Mashimo P.A.， Yamamoto Y.， Slots J.， Evans R.T.， Genco R.J. In vitro evaluation of antibiotics in the treatment of periodontal disease[J]. Pharmacol Ther Dent， 1981，6（1-2）：45-56.

[7] Ciancio S.G.， Slots J.， Reynolds H.S.， Zambon J.J.， McKenna J.D. The effect of short-term administration of minocycline HCl on gingival inflammation and subgingival microflora[J]. J Periodontol， 1982，53（9）：557-561.

[8] Basegmez C.， Berber L.， Yalcin F. Clinical and biochemical efficacy of minocycline in nonsurgical periodontal therapy：  a randomized controlled pilot study[J]. J Clin Pharmacol， 2011，51（6）：915-922.

[9] Pandit N.， Dahiya R.， Gupta R.， Kathuria A. Comparative evaluation of locally delivered minocycline and metronidazole in the treatment of periodontitis[J]. Contemp Clin Dent， 2013，4（1）：48-53.

[10] Nagarakanti S， Bishburg E. Is Minocycline an Antiviral Agent？ A Review of Current Literature[J]. Basic Clin Pharmacol Toxicol， 2016，118（1）：4-8.

[11] Hopkins AL. Network pharmacology： the next paradigm in drug discovery[J]. Nat Chem Biol， 2008，4（11）：682-690.

[12] Sivakumar K.C.， Haixiao J.， Naman C.B.， Sajeevan T.P. Prospects of multitarget drug designing strategies by linking molecular docking and molecular dynamics to explore the protein-ligand recognition process[J]. Drug Dev Res， 2020，81（6）：685-699.

[13] Li J.， Huang Y.， Zhao S.， Guo Q.， Zhou J.， Han W.， Xu Y. Based on network pharmacology to explore the molecular mechanisms of astragalus membranaceus for treating T2 diabetes mellitus[J]. Ann Transl Med， 2019，7（22）：633.

[14] Qin T.， Wu L.， Hua Q.， Song Z.， Pan Y.， Liu T. Prediction of the mechanisms of action of Shenkang in chronic kidney disease： A network pharmacology study and experimental validation[J]. J Ethnopharmacol， 2020，246：112128.

[15] Shannon P.， Markiel A.， Ozier O.， Baliga N.S.， Wang J.T.， Ramage D.， Amin N.， Schwikowski B.， Ideker T. Cytoscape： a software environment for integrated models of biomolecular interaction networks[J]. Genome Res， 2003，13（11）：2498-2504.

[16] Tao Q.， Du J.， Li X.， Zeng J.， Tan B.， Xu J.， Lin W.， Chen X. L. Network pharmacology and molecular docking analysis on molecular targets and mechanisms of Huashi Baidu formula in the treatment of COVID-19[J]. Drug Dev Ind Pharm， 2020，46（8）：1345-1353.

[17] Yin B， Bi YM， Fan GJ， Xia YQ. Molecular Mechanism of the Effect of Huanglian Jiedu Decoction on Type 2 Diabetes Mellitus Based on Network Pharmacology and Molecular Docking[J]. J Diabetes Res， 2020，2020：5273914.

[18] Yin B.， Bi Y.M.， Fan G.J.， Xia Y.Q. A non-antibacterial chemically-modified tetracycline inhibits mammalian collagenase activity[J]. J Dent Res， 1987，66（8）：1310-1314.

[19] Garrido-Mesa N， Zarzuelo A， Gálvez J. Minocycline： far beyond an antibiotic[J]. Br J Pharmacol， 2013，169（2）：337-352.

[20] Griffin MO， Ceballos G， Villarreal FJ. Tetracycline compounds with non-antimicrobial organ protective properties： possible mechanisms of action[J]. Pharmacol Res， 2011，63（2）：102-107.

[21] Golub L.M.， Ramamurthy N.S.， McNamara T.F.， Greenwald R.A.， Rifkin B.R. Tetracyclines inhibit connective tissue breakdown： new therapeutic implications for an old family of drugs[J]. Crit Rev Oral Biol Med， 1991，2（3）：297-321.

[22] Takahashi M， Altschmied L， Hillen W. Kinetic and equilibrium characterization of the Tet repressor-tetracycline complex by fluorescence measurements. Evidence for divalent metal ion requirement and energy transfer[J]. J Mol Biol， 1986，187（3）：341-348.

[23] Barza M.， Brown R.B.， Shanks C.， Gamble C.， Weinstein L. Relation between lipophilicity and pharmacological behavior of minocycline， doxycycline， tetracycline， and oxytetracycline in dogs[J]. Antimicrob Agents Chemother， 1975，8（6）：713-720.

[24] Kramer P.A.， Chapron D.J.， Benson J.， Mercik S.A. Tetracycline absorption in elderly patients with achlorhydria[J]. Clin Pharmacol Ther， 1978，23（4）：467-472.

[25] Klein NC， Cunha BA. Tetracyclines[J]. Med Clin North Am， 1995，79（4）：789-801.

[26] Williams DN， Laughlin LW， Lee YH. Minocycline： Possible vestibular side-effects[J]. Lancet， 1974，2（7883）：744-746.

[27] Sapadin AN， Fleischmajer R. Tetracyclines： nonantibiotic properties and their clinical implications[J]. J Am Acad Dermatol， 2006，54（2）：258-265.

[28] Soory M. A role for non-antimicrobial actions of tetracyclines in combating oxidative stress in periodontal and metabolic diseases： a literature review[J]. Open Dent J， 2008，2：5-12.

[29] Griffin M.O.， Fricovsky E.， Ceballos G.， Villarreal F. Tetracyclines： a pleitropic family of compounds with promising therapeutic properties. Review of the literature[J]. Am J Physiol Cell Physiol， 2010，299（3）：C539-C548.

[30] Chen M.， Ona V.O.， Li M.， Ferrante R.J.， Fink K.B.， Zhu S.， Bian J. Guo L. Farrell L. A. Hersch S. M. Hobbs W. Vonsattel J. P. Cha J. H. Friedlander R. M. Minocycline inhibits caspase-1 and caspase-3 expression and delays mortality in a transgenic mouse model of Huntington disease[J]. Nat Med， 2000，6（7）：797-801.

[31] Sun J.， Shigemi H.， Tanaka Y.， Yamauchi T.， Ueda T.， Iwasaki H. Tetracyclines downregulate the production of LPS-induced cytokines and chemokines in THP-1 cells via ERK， p38， and nuclear factor-κB signaling pathways[J]. Biochem Biophys Rep， 2015，4：397-404.

[32] Yang Y.， Kimura-Ohba S.， Thompson J. F.， Salayandia V.M.， Cossé M.， Raz L.， Jalal F. Y.， Rosenberg， G. A. Vascular tight junction disruption and angiogenesis in spontaneously hypertensive rat with neuroinflammatory white matter injury[J]. Neurobiol Dis， 2018，114：95-110.

[33] Machado L.S.， Kozak A.， Ergul A.， Hess D.C.， Borlongan C.V.， Fagan S.C. Delayed minocycline inhibits ischemia-activated matrix metalloproteinases 2 and 9 after experimental stroke[J]. BMC Neurosci， 2006，7：56.

[34] Fortier L.A.， Motta T.， Greenwald R.A.， Divers T.J.， Mayr K.G. Synoviocytes are more sensitive than cartilage to the effects of minocycline and doxycycline on IL-1alpha and MMP-13-induced catabolic gene responses[J]. J Orthop Res， 2010，28（4）：522-528.

[35] Sarwar M.， Samuel C.S.， Bathgate R.A.， Stewart D.R.， Summers R.J. Serelaxin-mediated signal transduction in human vascular cells： bell-shaped concentration-response curves reflect differential coupling to G proteins[J]. Br J Pharmacol， 2015，172（4）：1005-1019.

[36] Singh S， Simpson RL， Bennett RG. Relaxin activates peroxisome proliferator-activated receptor γ （PPARγ） through a pathway involving PPARγ coactivator 1α （PGC1α）[J]. J Biol Chem， 2015，290（2）：950-959.

Fig. 1. Flowchart for exploring the mechanisms of minocycline against periodontitis.

Fig. 2. The intersection of minocycline and periodontitis targets （A） and PPI network of 12-common targets （B）. PPI， protein-protein interaction.

Fig. 3. GO and KEGG analysis of 12 common targets. （A） CC of the GO function analysis. （B） MF of the GO function analysis. （C） BP of the GO function analysis. （D） The KEGG pathway enrichment analysis. GO， Gene Ontology; CC， cellular component; MF， molecular function; BP， biological process; KEGG， Kyoto Encyclopedia of Genes and Genomes.

Fig. 4. Distribution of minocycline targets in the relaxin signal pathway. The pink nodes are potential targets of minocycline， while the green nodes are related targets in this pathway.

Fig. 5. Periodontitis-minocycline–target–pathway network. Pink indicates periodontitis; orange indicates minocycline; light blue indicates the common targets; the pathways are classified into four groups including endocrine-related pathways （blue）， pathways associated with infectious diseases （red）， inflammation-related pathways （green） and cancer-related pathways （purple）.

Fig. 6. Three-dimensional molecular docking diagrams. （A） MAPK1 protein-minocycline. （B） CASP3 protein-minocycline. （C） MAPK14 protein-minocycline. （D） VEGFA protein-minocycline.