The Discovery and Characterization of HBS-101, a Novel Inhibitor of Midkine, as a Therapeutic Agent for the Treatment of Triple-Negative Breast Cancer.

To examine whether alterations occurs in the levels of MDK in BC, we have examined its status using TNMplot analysis platform that enables comparison of gene expression between tumor and normal tissues using a validated data base [ 38 ]. Results showed that MDK is highly expressed in BC compared to normal tissues ( Supplementary Fig. S1A ). Compared to normal breast and ER + BC, MDK expression is high in HER2 + and TNBC sub types ( Supplementary Fig. S1B ). Also, high expression of MDK is observed in several subtypes of TNBC ( Supplementary Fig. S1C ). Importantly, increased expression of MDK was associated with poor progression-free survival (PFS) in BC patients ( Supplementary Fig. S1D ). To test the functional significance of MDK in TNBC model cells, we have silenced MDK expression using 3 different siRNAs. Results showed that silencing MDK significantly reduced the viability and clonogenic survival of TNBC cells ( Fig. 1A – D ). We also confirmed these results with MDK knockout (KO) cell lines using CRISPR/cas9 system ( Fig. 1E – G ). Knockout of MDK also significantly reduced the survival of TNBC cells in colony formaton asays ( Fig. 1F – G ). Collectively these results suggest that MDK is highly expressed in TNBC and plays a critical role in the growth of TNBC cells. Our discovery of the MDK inhibitor started with following the dienogest molecule using insilico simulation studies. Dienogest is a synthetic progestogen and is approved for the treatment of birth control, endometriosis, and menorrhagia. Dienogest exhibits mild MDK inhibitory activity [ 39 ]. The first-in-class molecule HBS-101 was designed based on the dienogest pharmacophore. To modify its activity, the steroidal core has been kept as such and we decided to change the cyano methylene functional group with a structurally similar difluro acetylenic group as seen in HBS-101 ( Supplementary Fig. S2A – D ). MDK modeling was performed using a combination of loop modeling and MD simulations to reconstruct its complete 3D structure. MDK consists of 143 amino acids, with partial structures available in the Protein Data Bank (PDB) as NMR structures for both the N-terminal (PDB ID: 1MKN) and C-terminal (PDB ID: 1MKC) domains [ 40 ]. The N-terminal domain comprises 59 residues (positions 22–81), while the C-terminal domain consists of 42 residues (positions 84–140). Sequence analysis identified a gap of two missing residues between these domains, necessitating structural reconstruction. To bridge this gap, the C-terminal domain was initially positioned in close proximity to the N-terminal domain to maintain structural continuity. The missing residues were then modeled using loop modeling to seamlessly connect the two domains. Following this, additional loop refinement was performed to optimize flexibility, orientation, and steric compatibility while minimizing structural clashes. To achieve a stable conformation, an all-atom MD simulation was conducted for 1 µs ( Supplementary Fig. S2A – D ). The final refined structure was subsequently utilized for docking studies to identify potential binding interactions of HBS-101 with MDK. Binding site prediction using the Site Finder tool revealed a single promising binding pocket for HBS-101 on MDK. This site was composed of three key amino acid stretches: residues 31–36 (N-terminal), 95–100 (C-terminal), and 107–111 (C-terminal) ( Fig. 1H ). Together, these three stretches formed a structural arrangement resembling a fence around a shallow cleft on the protein surface, effectively creating a confined pocket for HBS-101 binding. Initial docking analysis indicated that the binding of HBS-101 was primarily stabilized by hydrophobic interactions ( Fig. 1I ) with free energy of binding −57.11 kcal/mol. However, during MD simulations, both the protein and ligand underwent slight conformational rearrangements, leading to the formation of a stable hydrogen bond between HBS-101 and K129. This hydrogen bond persisted throughout the simulation, suggesting a key stabilizing interaction ( Fig. 1J ). Based on these findings, we propose that the binding of HBS-101 to this site may interfere with MDK’s interaction with its endogenous receptors, potentially disrupting its biological function. MST is flexible, robust, and rapid for dissecting direct molecular interactions by combining fluorescence detection quality with thermophoresis flexibility and sensitivity. Two technical runs with the same samples were performed. No sticking of the target to the capillary walls was observed in the capillary scan. No sample aggregation or precipitation effects were observed in the normalized fluorescence. Binding was detected in both technical runs. MST assays using recombinant MDK showed high-affinity binding (38.4 nM) ( Fig. 1K ). To further confirm the direct interaction between HBS-101 and MDK, we generated biotinylated HBS-101. The addition of biotin did not alter the biological activity of HBS-101 ( Supplementary Fig. S3A ). To assess its interaction with MDK, purified MDK protein was incubated with biotin-HBS-101, and the interaction was evaluated using an avidin pull-down assay followed by Western blot analysis. The results demonstrated that HBS-101 binds to MDK ( Fig. 1L ). Further, using cellular thermal shift assays, we found stabilization of endogenous MDK upon HBS-101 treatment in cell lysates ( Fig. 1M ) confirming the interaction of HBS-101 with MDK. Using MDK knockdown TNBC cells, we tested whether HBS-101 specifically targets MDK. Cell viability assays showed that MDK-KD substantially diminished HBS-101's effectiveness, confirming its target specificity ( Supplementary Fig. S3B – C ). Collectively, the results from these experiments suggest that HBS-101 directly binds to MDK. We treated various TNBC cell lines (SUM-159, BT-549, MDA-MB-468, MDA-MB-231, MDA-MB-231-BrM2–831, HCC70, HCC1937) alongside HER2 + BC (SKBR3) and ER + BC (MCF7) cell lines, with different concentrations of HBS-101, assessing cell viability to evaluate its functional significance ( Fig. 2A ). HBS-101 significantly reduced cell viability of TNBC cell lines, with an IC 50 range of 0.3 to 2.8 µM, while the IC 50 for HER2 + BC cells was much higher at ~16 µM and for ER + BC is at >20 µM ( Fig. 2B ). Given this high IC 50 in HER2 + and ER + BC, we have decided to focus solely on TNBC in this study to further investigate the effects of HBS-101. Furthermore, treatment with HBS-101 resulted in a dose-dependent reduction in the colony-forming ability of TNBC cells ( Fig. 2C – D ). We then tested the effects of the MDK inhibitor HBS-101 on the invasion of TNBC cells. Our results indicate that HBS-101 significantly reduced the invasive properties of these cells ( Fig. 2E – F ). Next, to study whether HBS-101 treatment causes apoptotic cell death, we utilized Annexin V Assays. The results demonstrated that the administration of HBS-101 induced apoptosis in TNBC cells ( Fig. 2G , Supplementary Fig. S4A – B ), but not in normal mammary epithelial cells ( Supplementary Fig. S3D ). Previous studies have shown that MDK binding to its receptors activates mTOR, STAT3, and NF-κB signaling pathways. Western blotting analyses confirmed that TNBC models exhibit measurable levels of MDK and its receptors ( Fig. 3A ), indicating the possibility of autocrine signaling. Treatment with HBS-101 in HCC-70 cells significantly reduced the expression of MDK receptors, specifically Notch2 and inhibited downstream signaling pathways, including mTOR, STAT3, and NF-κB p65 ( Fig. 3B ) and increased cleaved-caspase 3 levels. In addition, HBS-101 treatment also decreased the activation of known MDK downstream signaling molecules such as AKT and PI3 kinase p85 in MDA-MB-231 cells ( Supplementary Fig. S3E ). RT-qPCR analyses further confirmed the downregulation of MDK target genes following HBS-101 treatment ( Fig. 3C ). Moreover, HBS-101 treatment significantly reduced STAT3 and NF-κB reporter activity in TNBC cells that stably expressing these reporters ( Fig. 3D – F ). We conducted an initial PK analysis of HBS-101 in rats following IV administration at 1 mg/kg and oral administration at 10 mg/kg ( Fig. 4A ). The results revealed that HBS-101 had mono-exponential disposition in vivo and had an estimated plasma half-life between 0.7 to 1.6 hours. The drug also had excellent oral bioavailability in rats. To validate these findings in another rodent species, we performed a subsequent PK analysis in female C57BL/6 mice (n=4) following 10 mg/kg i.p. administration of HBS-101 at various time points: 10 minutes, 30 minutes, 1 hour, 2 hours, 6 hours, and 24 hours post-intraperitoneal administration ( Fig. 4B – C ). The blood concentration versus time profile of HBS-101 is shown in Fig. 4B . Following intraperitoneal administration, HBS-101 was absorbed rapidly, resembling an intravenous bolus drug administration. PK model analysis indicated a negligible absorption phase, with maximum blood concentration of HBS-101 reached in less than 30 minutes post-dosing. The estimated maximum mean blood concentration was 6,271 ng/mL, with a half-life of approximately 0.5 hour in each case ( Fig. 4C ). By 6 hours post-dose, measurable drug concentrations (>1.5 ng/mL) were present in the systemic circulation. We then examined whether HBS-101 is capable of brain penetration. Brain samples from C57BL/6 mice (n=3) were collected at various time points after intraperitoneal administration of HBS-101 (10 mg/kg) ( Fig. 4D ). The biodistribution study confirmed that HBS-101 was detected in the brain, with peak detectable levels reaching 1,654 ng/g 10 minutes after administration, suggesting that HBS-101 effectively crosses the BBB ( Fig. 4D ). We also conducted toxicity studies using C57BL/6 mice and five-day HBS-101 treatment with doses of 2, 5, and 10 mg/kg/oral. At the end of the treatment period, mice were euthanized, and vital organs were analyzed using H&E staining. In these studies, no abnormalities were detected during the gross pathological examination of animals ( Fig. 4E ). These results suggested doses up to 10 mg/kg are safe to use for testing efficacy studies. To test the efficacy of HBS-101 on in vivo tumor progression, we initially conducted in vivo studies using xenograft tumor model in nude mice. MDA-MB-231 cells (2 x 10 6 ) mixed with 1:1 volume of Matrigel were injected into the mammary fat pad of mice. After the tumors were formed, a group of mice (n=7 tumors/group) were randomly selected and given HBS-101 treatment five days a week. The treatment consisted of either a vehicle or 2 mg/kg or 5 mg/kg/oral of HBS-101. HBS-101 treatment led to a significant dose-dependent reduction in tumor volume, with a decrease of 60–80% compared to vehicle treatment ( Fig. 5 A , B ). Mouse body weight was not affected by HBS-101 treatment ( Fig. 5C ). Given that PDX models accurately reproduce the intricate structure and diverse characteristics of human TNBC, we have also tested the efficacy of HBS-101 using PDX models. Results showed a significant reduction in PDX tumor growth following HBS-101 treatment compared to vehicle ( Fig. 5F – H ). Ki-67 staining in both CDX and PDX models revealed a significant decrease in Ki-67 positive cells in HBS-101 treated tumors when compared to vehicle-treated groups ( Fig. 5D – E , I – J ), indicating reduced proliferative activity of HBS-101 treated tumors. Since TNBC commonly results in brain metastases, and HBS-101 exhibits good brain penetration, we explored the potential of HBS-101 for treating TNBC brain metastases. TNBC brain tropic cells (MDA-MB-231-BrM2–831 (1.75 x 10 5 )) were injected intracranially, and following tumor establishment, mice were treated with either vehicle or HBS-101 (10 mg/kg/ip/5 days/week). HBS-101 treatment significantly reduced tumor progression ( Fig. 5K – M ) and extended the survival of mice with orthotopic tumors ( Fig. 5L ). Together, these findings indicate that HBS-101 demonstrates BBB permeability, and exhibits potent anti-tumor activity in TNBC CDX, PDX, and brain metastatic models.

MDA-MB-231 (RRID:CVCL_0062), BT-549 (RRID:CVCL_1092), HCC70 (RRID:CVCL_1270), SUM159 (RRID:CVCL_5423), MDA-MB-468 (RRID:CVCL_0419), HCC-1937 (RRID:CVCL_0290), HCC-1806 (RRID:CVCL_1258), SKBR3 (RRID:CVCL_0033), MCF7 (RRID:CVCL_0031) BC cell lines and MCF10A (RRID:CVCL_0598) normal mammary epithelial cells were purchased from the American Type Culture Collection (ATCC, Manassas, VA, USA) and were maintained as per ATCC guidelines. Brain-tropic MDA-MB-231 cells (MDA 231-BrM2–831) were obtained from the Memorial Sloan Kettering Cancer Center (MSKCC) Antibody & Bioresource Core Facility. Cell identity was confirmed with Short Tandem Repeat (STR) polymorphism analysis. Using the Mycoplasma PCR Detection Kit, we confirmed that none of the model cells used were contaminated with mycoplasma (Sigma, St. Louis, MO, USA). CellTiter-Glo Luminescent Cell Viability Assay kit and Luciferase Assay reagents were obtained from Promega (Madison, WI, USA). Annexin V/PI kit was purchased from BioLegend (San Diego, CA, USA). The antibodies (Notch2, p-NF-κB p65 (Ser529), NF-κB p65, p-mTOR(S2448), mTOR, p-S6(S235/236), S6, p-STAT3(Y705), STAT3, pPI3K p85, PI3K p85 and cleaved caspase 3) were obtained from Cell Signaling Technology (Beverly, MA, USA). β-actin antibody was purchased from Sigma (Sigma, St. Louis, MO, USA). Ki67 antibody was purchased from Abcam (Waltham, MA). Midkine polyclonal antibody was purchased from Proteintech (Rosemont, IL, USA) and ALK antibody was purchased from Santa Cruz (Dallas, TX, U.S.A). Supplementary data includes a detailed list of antibodies used in this study ( Supplementary Table S1 ). TNBC cells were transfected using Lipofectamine RNAiMAX Reagent (Thermofisher, cat# 13778150) according to the manufacturer’s protocol. Briefly, cells were transfected with either scrambled-siRNA or MDK-siRNAs (50 nM) for 60 – 72 h before they were subjected to Western blot analysis, cell viability and colony formation assays. siRNA Universal Negative Control #1; MDK siRNA 1 (SASI_Hs01_00094499); MDK siRNA 2 (SASI_Hs01_00094501); and MDK siRNA 3 (SASI_Hs01_00094503) were purchased from Millipore Sigma ( Supplementary Table S2 ). MDA-MB-231-MDK knockout (KO) cells were generated by infecting human specific MDK CRISPR gRNA lentiviral particles (MDK CRISPR gRNA1 (HSPD0000025378) and MDK CRISPR gRNA2 (HSPD0000025379)) and clones were selected with puromycin (1 µg/mL). Lentiviral particles expressing non-targeting gRNA with no complementary sequence in the genome were used to generate control cells. Briefly, the synthesis of HBS-101 (MW: 346.41) starts from commercially available 3,3-(ethylenedioxy)estra-5(10),9(11)-dien-17-one. 17-addition of 3-bromo, 3,3-difluoro-1-trisiopropylsilyl propyne in presence of n-butyl lithium affords 3,3-(ethylenedioxy)-17α-[1,1-difluoro-3-[tris(1-methylethyl)silyl]-2-propyn-1-yl]-17β-hydroxy-estra-5(10),9(11)-diene in 82% yields. The resulting compound was subsequently treated with hydrochloric acid and tetrabutylammonium fluoride to afford the final compound 17α-(1,1-difluoro-2-propyn-1-yl)-17β-hydroxy-estra-4,9-dien-3-one in 85% yields. Structures of the intermediates as well as the final compound were confirmed by 1H and 13C NMR data ( Supplementary Table S3 and Supplementary data ). Potential interactions between the target protein midkine and the HBS-101 were investigated using a label-free (LF) MST assay. Before the MST measurement, a quality control (QC) test was done on the protein using DLS to determine if the protein is soluble, properly folded and monodisperse in solution and if it is suitable for the subsequent MST measurements. The initial measurement set up include testing human midkine at various concentrations including 1000, 500 & 250 nM with an LED excitation of 50%. The ligand, HBS-101 was tested at various concentrations up to 500 µM with an LED excitation of 100%. The instrument used was Monolith NT LabelFree for LF MST signal test for midkine, and NT.115 (UV), 40%, Premium Coated capillaries for comprehensive MST binding assay using LF MST. The buffer used in the experiment was 1x PBS pH 7.4, 0.1% Pluronic F-127, 1% DMSO. Biotin pull-down assays were performed using the established protocol [ 34 ]. NanoLINK ® Streptavidin Magnetic Beads were purchased from Vector Labs. Briefly, in vitro immunoprecipitation assays were performed over a 1-hour incubation period using purified recombinant MDK (rMDK) protein (Cat #450–16). Purified rMDK (10 ng) was incubated with 5 µM Biotin-HBS-101 and 5 µl of NanoLINK ® Streptavidin Magnetic Beads in 300 µl of NETN buffer (20 mM Tris pH 8.0, 50 mM NaCl, 1 mM EDTA, 0.5% NP40, 10% Glycerol) for 1 hour at room temperature. Streptavidin-Biotin beads were washed five times with 500 µl of wash buffer (20 mM HEPES pH7.9, 150 mM KCl, 0.1 mM EDTA, 0.2% Triton X-100, 10% Glycerol) using magnetic stand. Bound protein was eluted using 2X SDS buffer and run on a 12% SDS-PAGE gel. To elucidate the interaction between HBS-101 and MDK, we conducted molecular modeling studies. Initially, the 3D structure of MDK was generated using computational modeling. Putative binding sites for HBS-101 were then identified using the Site Finder tool. Molecular docking was performed at the selected sites using the extra precision (XP) docking method in Schrödinger Maestro, with grid preparation tailored to these regions. Prior to docking, HBS-101 was prepared using the LigPrep module in Schrödinger Maestro. Binding poses were ranked based on free energy profiles, and the best MDK-HBS-101 complex was subjected to a 250 ns molecular dynamics (MD) simulation. Prior to the simulation, the complex was fully immersed in an orthorhombic water box containing TIP3P water molecules, and simulations were conducted using the OPLS 2005 force field. Snapshots were collected every 1.2 ps for residence time analysis of the ligand. CETSA was conducted to evaluate the thermal stabilization of MDK upon binding to HBS-101 in cells. MDA-MB-231 cells were treated with 20 µM HBS-101 or an equal volume of DMSO (vehicle control) for 6 hours. Following treatment, cells were harvested, washed with PBS, and resuspended in PBS. The suspensions were aliquoted into 100 µL portions and subjected to a series of temperatures (44, 48, 52 and 56 °C) for 3 minutes, followed by cooling on ice for 3 minutes. Subsequently, RIPA lysis buffer containing protease inhibitors was added to each sample, and the mixture was incubated on ice for 15 minutes. The lysates were centrifuged at 13,000 rpm for 30 minutes at 4°C, and the supernatants were collected. Protein concentrations in the supernatants were quantified using the Bradford assay (Bio-Rad). A total of 50 μg of protein from each sample was resolved on 15% SDS-PAGE and analyzed by Western blotting using an antibody specific to MDK [ 35 ]. MTT and the Cell Titer-Glo Luminescent Cell Viability Assays were used to determine the cell viability rates of TNBC cells treated with HBS-101 as described previously [ 36 ]. Clonogenic experiments were performed with TNBC cells in 6-well plates as described [ 36 ]. Invasion assays were conducted utilizing the Corning ® BioCoat ™ Growth Factor Reduced Matrigel Invasion Chambers as previously mentioned [ 36 ]. The effect of HBS-101 on apoptosis of normal mammary epithelial cells and TNBC cells was measured using an annexin V/PI assay as previously described [ 36 ]. TNBC cells were treated with vehicle or HBS-101 for 20 hours and RNA was isolated using RNeasy mini kit (Qiagen, Hilden, Germany). RT-qPCR was used to verify the status of the MDK target genes. Data were normalized to GAPDH or β-actin , and the delta-delta-CT method was used to determine the difference in fold change. The list of primers were included in the supplementary data ( Supplementary Table S4 ). Western blotting and IHC conditions were followed as described in our earlier publication [ 36 ]. Reporter gene assays were performed following established protocols [ 36 ]. TNBC model cells, stably transfected with either a STAT3 reporter or an NF-κB reporter, were used for these experiments. The cells were seeded in 24-well plates, serum-starved overnight, and then incubated for 24 hours under different conditions: STAT3 reporter cells were treated with or without 10% FBS, while NF-κB reporter cells were treated with or without TNFα, in the presence or absence of HBS-101. After incubation, the cells were lysed using luciferase lysis buffer, and luciferase activity was measured using a dual-luciferase assay kit (Promega, Madison, WI, USA). Initial PK study of HBS-101 was conducted in SD rats following single intravenous (IV) (1 mg/kg) and oral (PO) (10 mg/kg) administration of the compound. Preparation of IV (1 mg/kg, 5 mL/kg) dosing formulation consists of 0.2 mg/mL solution of "10% HP-β-CD in DI water (w/v)" [Dissolved 0.96 mg of HBS-101 in 4.800 mL "10% HP-β-CD in DI water (w/v)" with vortexing and sonification to obtain a solution of 0.2 mg/mL HBS-101). Preparation of PO (10 mg/kg, 10 mL/kg) dosing formulation consists of 1 mg/mL solution of "10% HP-β-CD in DI water (w/v)" [Dissolved 9.94 mg of HBS-101 in 9.940 mL "10% HP-β-CD in DI water (w/v)" with vortexing and sonification to obtain a solution of 1 mg/mL HBS-101]. Briefly, blood sampling for IV study at 5, 15, 30 min, 1, 2, 4, 8, 24, 48 hours post dose and for oral 30 min, 1, 2, 4, 8, 24, 48 hours post dose were collected. Bioanalytical assay was conducted using AB Sciex Triple Quad 5500 LC/MS/MS instrument equipped with HALO 90A, AQ-C18, 2 µm 3.0×30 mm column with a mobile phase (A) 5% Acetonitrile in Water (0.1%Formic acid) and B 95% Acetonitrile in Water (0.1% Formic acid). The PK parameters were estimated by non-compartmental model using WinNonlin 8.3. The bioavailability (F%) was calculated by following formula: AUClast-PO/AUCINF-PO > 80%: F= (mean AUCINF-PO*DoseIV)/(mean AUCINF-IV*DosePO). A second PK study of HBS-101 was conducted in female C57BL/6 mice (RRID: IMSR_JAX:000664) following intraperitoneal (i.p.) administration of the compound at a dosage of 10 mg/kg body weight. The formulation of HBS-101 for i.p. administration was prepared at a required volume of 0.1 mL, dissolving the compound in 0.3% hydroxypropyl cellulose. This was administered to a group of four mice. Blood samples (10 µL each) were collected via the tail vein at 10 minutes, 30 minutes, 1 hour, 2 hours, 6 hours, and 24 hours after administration. Key PK parameters such as the area under the drug concentration curve (AUClast, and AUCinfinity), the time to reach maximum drug concentration (Tmax, and Cmax) following i.p. administration, biological half-life (T½), apparent clearance (CL), and apparent volume of distribution (V), were estimated using non-compartmental analysis with Phoenix WinNonlin v8.0 software (Pharsight Corporation, Mountain View, CA, USA). To assess the brain bioavailability of HBS-101, another group of female mice (n = 3) received the i.p. formulation at the same dosage of 10 mg/kg. Whole brain samples were collected at the same time points as the blood samples. The concentrations of HBS-101 in the blood and brain homogenates were quantified using validated LC-MS/MS bioanalytical methods developed by the GCC Center for Comprehensive PK/PD & Formulation at Texas Southern University in Houston, TX. Briefly, bio-samples were extracted through protein precipitation using nine times the volume of acetonitrile containing an internal standard (IS), 4-hydroxybenzophenone at 5 ng/mL. After mixing, the samples were vortexed for 2 minutes and then centrifuged for 20 minutes at 14,000 rpm at 4°C. An aliquot of the supernatant was injected into the LC-MS/MS for analysis. HBS-101 was separated using a Synergi Fusion-RP column (50 x 2 mm, 4 μm, 80Å, Phenomenex Inc., Torrance, CA, USA) with a gradient mobile phase composed of solvent A (0.1 % formic acid in water) and solvent B (0.1 % formic acid in acetonitrile) in an Exion LC system. The drug concentrations were determined using a SCIEX TRIPLE QUAD ™ 6500+ Mass Spectrometer with an ESI source, monitored in positive ion mode under multiple reaction monitoring (MRM). The mass transitions were m/z 347.1 → 135.1 for HBS-101 and 199.1 → 121.1 for the internal standard. The assay demonstrated linearity for HBS-101 concentrations in mouse blood from 0.5 ng/mL to 1000 ng/mL and in mouse brain homogenate from 0.75 ng/g to 1500 ng/g of brain tissue. The regression correlation coefficients were better than 0.99 in all validation runs. The accuracy (RE%) was 0.63%, and precision (CV%) was 11.58%. Mean matrix effect and recovery were 106.4% ± 15.8% and 94.8% ± 9.0% respectively, suggesting that the matrix effect was negligible, and the sample extraction was thorough with good recovery in this study. All animal studies were carried out using IACUC approval from UT Health San Antonio. To assess the maximum tolerable dose of HBS-101, we utilized C57BL/6 mice due to their intact immune system. The study included four treatment groups: 1) vehicle control, 2) 2 mg/kg, 3) 5 mg/kg, and 4) 10 mg/kg of HBS-101, administered orally. Each group consisted of three mice (n = 3). The treatment was conducted daily for five consecutive days. The mice were monitored daily for adverse toxic effects. On the sixth day, the mice were euthanized, and vital organs including the heart, liver, kidneys, pancreas, spleen, ovaries, and uterus were harvested. The collected organs were fixed in 10% neutral-buffered formalin for 24 hours, processed, and embedded in paraffin blocks. For histological analysis, 5 µm sections were prepared from the paraffin-embedded tissues and stained with hematoxylin and eosin (H&E). The stained sections were then examined under a microscope for pathology screening to assess any histological changes or abnormalities. Nude mice (8–10 weeks old) were obtained from Charles River (Wilmington, MO, USA). For xenograft studies, MDA-MB-231 cells (2 × 10⁶) were injected into the mammary fat pad of female mice (n = 7 tumors). Once tumors reached a measurable size, the mice were randomly assigned to treatment groups receiving either vehicle control or HBS-101 (2 mg/kg and 5 mg/kg, administered orally, 5 days per week). Tumor volumes were measured every three to four days using a digital caliper and calculated using the modified ellipsoidal formula: tumor volume = 1/2(L × W²), where L represents the longitudinal diameter and W the transverse diameter. At the end of the study, all mice were euthanized, and tumors were excised and processed for histological analysis. The percentage of Ki-67-positive proliferating cells was determined by analyzing five randomly selected microscopic fields [ 36 ]. Female NOD.CB17-Prkdcscid/NCrCrl mice, aged 8–10 weeks, were purchased from Charles River (Wilmington, MO). To assess whether HBS-101 reduces the growth of breast cancer brain metastases, MDA-MB-231-BrM2–831 cells (1.75 × 10 5 ), which stably express a triple-fusion reporter encoding herpes simplex virus thymidine kinase 1, green fluorescence protein and firefly luciferase were injected orthotopically into the right cerebrum of mice using an established protocol [ 37 ]. After tumor establishment, the mice were treated with either a vehicle (0.3% hydroxypropyl cellulose) or HBS-101 at a dosage of 10 mg/kg body weight, administered intraperitoneally five days a week, starting three days after implantation (n = 6 mice per group). Tumor growth was monitored using the Xenogen IVIS system. Throughout the study, the animals were observed for any neurological symptoms. At the conclusion of the experiment, the mice were euthanized, and their brains were isolated and processed for histological studies. TNBC-PDX-96 tumor-bearing mice were obtained from The Jackson Laboratory (TM00096). Female SCID mice were used as recipients for 2 mm³ TNBC-PDX-96 tumor tissue implants to establish subcutaneous PDX tumor models. Once tumors became measurable, the mice were randomized into control and treatment groups (n = 6 tumors per group). The control group received the vehicle, while the treatment group was administered HBS-101 (5 mg/kg, orally, 5 days per week). Mice were monitored daily for signs of toxicity, and tumor volumes were measured every 3–4 days using digital calipers. At the end of the study, mice were euthanized, and tumors were excised, weighed, and processed for histological analysis. IHC staining was performed following a previously published protocol [ 36 ]. GraphPad Prism 10 software (RRID:SCR_002798) was used for all statistical analyses (GraphPad Software, San Diego, CA). Statistics were compared between the control and HBS-101-treated groups using a studenťs t-test and one-way ANOVA. All the data in the bar graphs are displayed as mean ± SEM. Significant results were defined as a p-value less than 0.05. All data supporting the conclusions is included in the paper and/or in the Supplementary Materials . The listed web links https://tnmplot.com/analysis/ , https://ualcan.path.uab.edu/ , https://xenabrowser.net/heatmap/ were used to collect the data for Fig. S1A – D .

MDK is a multifunctional protein secreted into the extracellular space, where it functions as a cytokine or growth factor modulating various signaling pathways [ 3 , 4 ]. MDK overexpression has been reported for at least 20 different cancer types, including the malignancies of breast, brain, ovarian, prostate, lung, colorectal, pancreatic, liver, bladder, and blood, and reported as a potential diagnostic and prognostic biomarker associated with poor survival [ 9 , 41 ]. MDK signaling is implicated in critical hallmarks of cancer such as cell growth, survival, evasion of apoptosis, migration, invasion and metastasis, angiogenesis, antitumor immunity, and inflammation [ 9 ]. This makes MDK a promising target for cancer therapy. However, to date, no published inhibitors of MDK have been shown to directly bind and inhibit its activity. In this study, we developed and evaluated the anti-tumor efficacy of HBS-101, a first-in-class MDK inhibitor. HBS-101 demonstrated direct binding to MDK (EC 50 ~ 38.4 nM) and exhibited potent growth-inhibitory effects across multiple TNBC cell models. Using CDX and PDX models, we further demonstrated the anti-tumor efficacy of HBS-101. Based on the findings from insilico, MST and CETSA studies, it is apparent that HBS-101 is binding to allosteric sites other than active or orthosteric sites of MDK, inducing a conformational change that modifies the protein’s activity to bind target receptors. Future studies such as isothermal calorimetry (ITC) combined with cryoEM are needed to resolve whether HBS-101 is a negative allosteric regulator of MDK to reduce the function of the target protein and its ability to bind with designated receptors. MDK mediates its effects through several membrane receptors, including integrins [ 42 ], protein tyrosine phosphatase ζ (PTPζ) [ 8 ], anaplastic lymphoma kinase (ALK) [ 43 ], and Notch2 [ 44 ]. Additionally, MDK interacts with low-density lipoprotein receptor-related protein (LRP) [ 45 ] and syndecans [ 43 ] in other tissues. MDK signaling modulates downstream components such as PI3K, MAPK, and transcription factors including NF-κB, Hes-1, and STATs. Notably, MDK has been shown to induce temozolomide resistance in glioblastoma by promoting cancer stem-like properties [ 46 ]. In BC, MDK activates the NF-κB pathway, promoting proliferation and migration [ 7 ]. In this study, we demonstrated that HBS-101 functions as an MDK inhibitor in biophysical and cell-based assays. Furthermore, HBS-101 treatment substantially reduced MDK downstream signaling pathways, including STAT3, mTOR, and NF-κB, demonstrating its therapeutic potential. MDK expression is minimal in normal tissues but is markedly elevated in malignant tumors [ 47 ]. High levels of MDK are observed in various solid cancers, where MDK signaling promotes tumor proliferation, metastasis, and therapy resistance [ 9 ]. Tumors often upregulate MDK signaling via autocrine and paracrine mechanisms [ 9 ]. For instance, fibroblasts secrete MDK, driving ovarian metastases in gastric cancer through the MDK-LRP1 axis [ 48 ]. Collectively, these studies underscore the critical role of MDK in cancer progression. Consistent with these findings, our study revealed that TNBC models express MDK and its receptors, exhibiting autocrine signaling. Using genetic knockout and pharmacological inhibition with HBS-101, we provided compelling evidence that MDK signaling is essential for TNBC cell growth and survival. MDK overexpression has been identified as an independent poor prognostic factor for BC patient survival [ 12 ]. Elevated serum MDK levels have been associated with BC carcinogenesis [ 13 ]. Additionally, MDK receptor ALK is overexpressed in BC, with gene amplification observed in 25% of TNBC cases [ 49 ]. MDK-Notch2/JAK2-STAT3 signaling has been shown to induce epithelial-mesenchymal transition (EMT) [ 50 ], and STAT3 represents a potential therapeutic target for TNBC [ 51 ]. MDK-PTPζ signaling plays a critical role in cancer progression, with PTPζ expression identified as an independent risk factor for TNBC recurrence and metastasis [ 52 ]. In our study using TNBC CDX and PDX models, we demonstrated that HBS-101 effectively inhibits TNBC progression, consistent with previously reported roles of MDK signaling. The development of brain metastases in TNBC remains a significant clinical challenge due to the lack of effective therapies and the restrictive nature of the blood-brain barrier (BBB), which limits drug delivery to the brain. Our study demonstrates that HBS-101 exhibits favorable BBB permeability, making it a promising candidate for treating brain metastases in TNBC. Using orthotopic brain metastatic models, we showed that HBS-101 treatment significantly reduced tumor progression and prolonged the survival of mice bearing intracranial MDA-MB-231-BrM2–831 tumors. These findings demonstrate the therapeutic potential of HBS-101 in addressing TNBC brain metastases, which are associated with poor patient outcomes and limited treatment options. In conclusion, we have developed and characterized HBS-101, a first-in-class small-molecule compound capable of binding to MDK and inhibiting its activity. HBS-101 represents a novel and effective strategy to counteract MDK-mediated oncogenic processes. By inhibiting the interaction of MDK with its key receptors, including integrins, PTPζ, ALK, and Notch2, HBS-101 disrupts critical signaling pathways such as NF-κB, mTOR, and STAT3, thereby reducing TNBC cell survival. Its simple route for synthesis, stability, and oral bioavailability makes HBS-101 easily adapted for clinical application in cancers characterized by elevated MDK signaling.

Triple-negative breast cancer (TNBC) is characterized by the absence of estrogen receptor alpha (ER-α), progesterone receptor (PR), and human epidermal growth factor receptor 2 (HER2) expression [ 1 ]. TNBC is an extremely aggressive subtype of breast cancer (BC) that accounts for a substantial proportion of cancer-related fatalities in women with ~150,000 deaths annually worldwide. Most TNBC patients are treated with chemotherapy, but ~ 50% develop resistance, increasing the risk of disease recurrence and decreasing survival [ 2 ]. Despite advances in BC research and therapy, TNBC's propensity for early metastases, limited targeted therapies, and poor prognosis represent a major clinical problem. Midkine (MDK) is a heparin-binding protein that is released into the extracellular space and can function as either a cytokine or a growth factor modulating several signaling pathways [ 3 – 5 ]. MDK is highly expressed in several cancers, including TNBC, and promotes cell migration, proliferation, and survival [ 6 – 14 ]. Additionally, MDK overexpression is associated with metastasis, therapy resistance, and aggressive tumor behavior [ 15 – 19 ]. MDK also modulates the tumor microenvironment by influencing immune cell recruitment, cytokine secretion, and angiogenesis [ 18 , 20 – 25 ]. Preclinical studies indicate MDK inhibition blocks the progression of many cancers [ 25 – 27 ]. MDK levels were higher in TNBC, and high MDK levels were associated with a poor prognosis for TNBC patients. However, the lack of a small molecule inhibitor represents a knowledge gap for clinical translation. Considering the importance of the MDK axis in cancer, several investigators focused on targeting MDK receptors. For example, targeting the MDK/ALK axis with ALK inhibitors impairs in vivo tumorigenicity and chemoresistance in PDAC [ 28 ]. Metformin was shown to block some of the effects of MDK such as activation of PI3K and MAPK in endometrial cancer [ 29 ]. MDK inhibition using shRNA enhances anti-PD-1 immunotherapy in hepatocellular carcinoma by preventing immunosuppressive MDSCs infiltration [ 30 ]. Treatment with functional antibodies against MDK suppressed growth of osteosarcoma cell lines [ 31 ]. A small molecule compound iMDK that inhibits the expression of MDK was reported in the literature. iMDK inhibited lung adenocarcinoma [ 25 , 32 ] and augmented Interferon-γ anti-tumor activity in ovarian cancer [ 33 ]. These results are interesting; however, a weakness is that iMDK only lowers the expression of MDK and does not act directly on MDK and a major target for iMDK seems to be PI3K [ 32 ]. Hence, the lack of a specific MDK inhibitor that blocks MDK oncogenic signaling represents a knowledge gap for its translation. In this study, we report the discovery of HBS-101, a first-in-class MDK inhibitor with a high affinity (KD 38.4 nM). Our studies showed that HBS-101 reduces TNBC cell survival, promotes apoptosis, and decreases tumor growth in TNBC xenograft models. Pharmacokinetic (PK) studies indicated oral bioavailability and toxicity studies reveal no significant organ toxicity. Biodistribution studies showed that HBS-101 is brain permeable. Importantly administration of HBS-101 significantly decreased TNBC tumor growth in CDX and PDX models as well as TNBC brain metastatic tumor growth in orthotopic mice models.