Identification of tetracycline combinations as EphB1tyrosine kinase inhibitors for treatment ofneuropathic painMahmoud S. Ahmeda,1 , Ping Wanga,1 , Ngoc Uyen Nhi Nguyena, Yuji Nakadaa, Ivan Menendez-Montesa,Muhammad Ismailb , Robert Bachooa, Mark Henkemeyerc, Hesham A. Sadeka,d,e,f,2, and Enas S. Kandilg,2 aDepartment of Internal Medicine, University of Texas Southwestern Medical Center, Dallas, TX 75390; bDepartment of Pharmaceutical Chemistry, TheBritish University in Egypt, Cairo 11837, Egypt; cDepartment of Neuroscience, University of Texas Southwestern Medical Center, Dallas, TX 75390;dDepartment of Molecular Biology, University of Texas Southwestern Medical Center, Dallas, TX 75390; eDepartment of Biophysics, University of TexasSouthwestern Medical Center, Dallas, TX 75390; fCenter for Regenerative Science and Medicine, University of Texas Southwestern Medical Center, Dallas, TX75390; and gDepartment of Anesthesiology, University of Texas Southwestern Medical Center, Dallas, TX 75390Previous studies have demonstrated that the synaptic EphB1receptor tyrosine kinase is a major mediator of neuropathic pain,suggesting that targeting the activity of this receptor might be aviable therapeutic option. Therefore, we set out to determine ifany FDA-approved drugs can act as inhibitors of the EphB1 intracellular catalytic domain. An in silico screen was first used to identifya number of tetracycline antibiotics which demonstrated potentialdocking to the ATP-binding catalytic domain of EphB1. Kinase assays showed that demeclocycline, chlortetracycline, and minocyclineinhibit EphB1 kinase activity at low micromolar concentrations. Inaddition, we cocrystallized chlortetracycline and EphB1 receptor,which confirmed its binding to the ATP-binding domain. Finally,in vivo administration of the three-tetracycline combination inhibitedthe phosphorylation of EphB1 in the brain, spinal cord, and dorsalroot ganglion (DRG) and effectively blocked neuropathic pain in mice.These results indicate that demeclocycline, chlortetracycline, and minocycline can be repurposed for treatment of neuropathic pain andpotentially for other indications that would benefit from inhibitionof EphB1 receptor kinase activity.EphB receptor drug repurposing neuropathic pain crystallographyDownloaded at UT SOUTHWESTERN MEDICAL CTR on February 24, 2021Chronic pain is a common debilitating condition which typically results in prescriptions of high doses of opioids in anattempt to control the pain severity (1). In the United States,over 100 million patients are affected by a form of chronic pain,which results in significant morbidity and costs over 600 billionin combined lost wages and medical expenses annually. In fact, itis estimated that over 25 million Americans experience pain on adaily basis (2). Due to these staggering numbers and its link toopioid use, chronic pain has been dubbed “the silent epidemic”by the NIH (3). As a result of this chronic pain epidemic, the useof opioids has increased exponentially in the United States sincethe late 1990s (4) and has reached the level of a national crisis.Although opioids are commonly used for treatment of chronicpain, they are only partially efficacious for short-term painmanagement, and the response to their long-term use is widelyvariable (5). Despite the magnitude of the pain epidemic and theopioid crisis, there has been little progress in the development ofnonopioid alternative therapies. Therefore, there is an urgentneed for developing novel nonopioid and nonaddicting therapiesthat are effective in management of chronic pain, as well ascounteracting the adverse, highly addictive effects of opioid use,in particular opioid dependence and opioid withdrawal-mediatedpain (6). Importantly, the path to new drug development is lengthyand costly. For example, it takes an average of 12 y and over 2billion for a new drug to reach the market. These staggering statistics underscore the importance of developing faster and cheaperstrategies to counteract the growing opioid epidemic.PNAS 2021 Vol. 118 No. 10 e2016265118Peripheral neuropathic pain (PNP), which is defined by neuralgia and painful polyneuropathy, is a highly prevalent type ofpain that results in significant morbidity and disability (7–9).PNP is one of the common associated manifestations for a seriesof different diseases, such as autoimmune diseases, diabetes,different types of cancers, neurofibromatosis, viral infections,and toxin exposure, among others (10–16). Current therapies fortreatment of PNP include over-the-counter drugs such as ibuprofen and acetaminophen, in addition to prescription painmedication including opioids, anticonvulsants, and antidepressants (17, 18). However, these therapies are seldom effective, inparticular opioids, which, as outlined above, are ineffective fortreatment of chronic pain, requiring dose escalation which further contributes to the ongoing opioid crisis.The large family of Eph (erythropoietin-producing hepatocellular carcinoma) receptor tyrosine kinases have been implicated in numerous pathologies, including Alzheimer’s disease,anxiety, neuropathic pain, malignancies, fibrotic diseases, andviral infections, among others (19–22). Eph receptors are highlyconserved proteins that are divided into two subfamilies of nineEphA and five EphB receptors based on sequence similarity (23).As both Eph receptors and ephrins are membrane anchored,SignificanceChronic pain is one of the most pressing national healthcareproblems, and as a result, chronic pain was named by the NIHas the “silent epidemic.” Consequentially, chronic pain has directly contributed to the disproportionate increase in opioiduse in America. However, thus far, there has been little progress in discovery of alternative noninvasive therapeutics forchronic pain. The current report outlines the repurposing ofthree tetracyclines for treatment of neuropathic pain by targeting the catalytic domain of EphB1 kinase. These results canbe readily applicable clinically for treatment of neuropathicpain and provide proof of concept for the development ofmolecules for targeting Eph/ephrin-mediated pathologies.Author contributions: H.A.S. and E.S.K. designed research; M.S.A., P.W., N.U.N.N., Y.N.,I.M.-M., M.I., R.B., M.H., and E.S.K. performed research; M.S.A., P.W., N.U.N.N., Y.N.,I.M.-M., R.B., M.H., and H.A.S. analyzed data; and H.A.S. and E.S.K. wrote the paper.The authors declare no competing interest.This article is a PNAS Direct Submission.Published under the PNAS license.1M.S.A. and P.W. contributed equally to this work.2To whom correspondence may be addressed. Email: [email protected] [email protected] article contains supporting information online at .2016265118/-/DCSupplemental.Published February 24, 2021. 1 of 10BIOPHYSICS ANDCOMPUTATIONAL BIOLOGYEdited by Robert J. Lefkowitz, HHMI, Durham, NC, and approved December 29, 2020 (received for review July 31, 2020)

receptor–ligand interactions generally occur upon cell–cell contact,and this leads to the transduction of bidirectional intracellular signals into both the Eph-expressing cell and ephrin-expressing cell.Previous studies using EphB1 / knockout mice and otherapproaches demonstrated that the EphB1 receptor tyrosine kinaseprotein is essential for neuropathic pain induced by different typesof experimental nerve damage and for the related painful effectsof opioid withdrawal (24–28). These studies also demonstratedthat the EphB1 receptor mediates neuropathic pain directly in thespinal cord through activation of the n-methyl-d-aspartate receptor (NMDAR) which results in immediate early gene expression(c-Fos) and long-term potentiation, which are well documented toparticipate in chronic pain states and opioid withdrawal via themechanism of central sensitization in the spinal cord. A keyfeature is that, following peripheral nerve damage, presynapticephrin-B2 ligand protein is up-regulated in nociceptive peripheral nerve fibers, and postsynaptic EphB1 receptor proteinbecomes up-regulated on dorsal horn neurons in the spinal cord.This model of ephrin-B2 and EphB1 in neuropathic pain and theephrin-B:EphB:NMDAR protein interactions has been validatedby numerous groups (29–43). Furthermore, what makes theEphB1 receptor a highly significant and important therapeutictarget for neuropathic pain is that the knockout mouse shows nomajor developmental abnormalities and exhibits normal perception of acute/normal pain stimuli. Intriguingly, even the EphB1 / heterozygote animals are refractory to neuropathic pain. This indicates reducing EphB1 activity 50% will likely be sufficient tomount a positive response. We therefore postulated that inhibitorycompounds do not need to completely block EphB1 activity; theyonly need to reduce it. Therefore, we performed in silico analysisof Food and Drug Administration (FDA)–approved drugs for inhibitors of the EphB1 kinase domain. These studies allowed us toidentify a group of tetracycline antibiotics as candidate FDAapproved drugs to be repurposed for treatment of neuropathic pain.Despite their high relevance to a wide range of diseases, thereare currently no approved FDA drugs targeting any of the Ephreceptors. Conceptually speaking, drug repositioning can be aplausible approach toward screening of FDA-approved drugswith well-documented safety profiles and therapeutic indices toidentify any with the potential to target Eph receptors. Herein,we present a drug-repositioning platform coupled with in vitroand in vivo evaluation for FDA-approved drugs that inhibit EphBreceptors. Taking advantage of in silico screening, molecular biology, structural biology, biochemistry, and in vivo mouse models,we identify members of the tetracycline family including chlortetracycline, demeclocycline, and minocycline, used as antibioticsto treat a wide variety of bacterial infections (44), as potentialdrugs to repurpose for the treatment of PNP.Downloaded at UT SOUTHWESTERN MEDICAL CTR on February 24, 2021ResultsIn Silico Screen for EphB Kinase Inhibitors. The EphB receptors havehighly conserved intracellular tyrosine kinase catalytic domain (45,46). Crystal structures for EphB kinase domains have been determined and can be accessed in the protein data bank (PDB), such as3ZFX (apo EphB1) and 5MJA (EphB1 bound with quinazolinebased inhibitor), 3ZFM (apo EphB2), 3ZFY (apo EphB3), and3ZEW (EphB4 bound with staurosporine) (47, 48). We havecoupled two different approaches related to drug discovery viaemployment of ligand-based and structure-based in silico screening,starting with staurosporine, which is bound with EphB4. Staurosporine was selected as a starting query for a ligand-based approach because of its broad ability to bind within many differentkinase domains and the possessing of different functional moietiesthat will allow including all the potential hits (Fig. 1A). TheMMFF94 energy-minimized library of FDA-approved drugs wasscreened for molecular similarity compared to staurosporine. Thetop 100 drugs based on Tanimoto chemical similarity score werefurther filtered based on literature survey and exclusion of drugs2 of 10 PNAS undesirable clinical indications/side effects. This was followedby an in silico docking study along with the pocket of EphB1 kinase domain for ATP and/or inhibitor binding (PDB code: 5MJA)ending up with top 10 drugs (Fig. 1 B and C and SI Appendix,Table S1). Demeclocycline, chlortetracycline, rolitetracycline,oxytetracycline, ertapenem, darifenacin, quinagolide, minocycline,ramelteon, and galantamine showed proper binding affinity withrespect to energy scoring (49), Tanimoto scores (50), and bindingmode toward EphB1 kinase domain, suggesting hydrophobic–hydrophobic and a network of hydrogen bond interactions(Fig. 1 D and E and SI Appendix, Fig. S1).Biochemical Evaluation of EphB Kinase Inhibitors. Seven of theidentified FDA-approved drugs were first tested for their inhibitory profiles using an in vitro EphB1 kinase assay. Ramelteon,oxytetracycline, galantamine, and darifenacin failed to inhibit theEphB1 kinase activity at concentrations up to 100 μM (SI Appendix, Fig. S2). However, chlortetracycline, demeclocycline, andminocycline inhibited the EphB1 kinase activity with inhibitionconcentration at 50% (IC50) calculated to be 39, 44, and 56 μM,respectively (Fig. 2). The ability of demeclocycline, chlortetracycline, and minocycline to inhibit EphB1 kinase activity promptedus to determine if they are also able to inhibit the related EphB2,EphB3, and EphB4 kinase domains, revealing all could beinhibited with IC50 ranging from 37 to 92 μM (Fig. 2). Curiously,chlortetracycline, demeclocycline, and minocycline all showed asimilar biphasic influence on the EphB3 catalytic activity, in thelow nanomolar ranges they slightly elevated activity, whereas athigher concentrations, they inhibited activity. Such biphasic responses were not observed with the EphB1, EphB2, and EphB4kinase domains.We determined the IC50 against EphB kinases for two-drugand three-drug combinations for the demeclocycline (D), chlortetracycline (C), and minocycline (M) to minimize the therapeuticdose per drug by having equimolar ratios of DC (demeclocycline chlortetracycline), DM (demeclocycline minocycline), and MC(minocycline chlortetracycline) which demonstrated improvement in the IC50 against EphB1 at 16, 13, and 15 μM, respectively(SI Appendix, Fig. S3), compared to the staurosporine inhibitoryprofile (SI Appendix, Fig. S4). Exploring the triple-drug strategyby having equimolar ratio of MCD (minocycline chlortetracycline demeclocycline) improved the IC50 towards EphB1kinase to 8 μM. Furthermore, we performed a kinase screen on88 tyrosine kinase and tyrosine kinase-like kinases for MCD todetermine its selectivity. The results indicate that only FGF-R1had a r