Targeting NANOG and FAK via Cx26-derived Cell-penetrating Peptides in Triple-negative Breast Cancer Free

This article is featured in Selected Articles from This Issue, p. 1

Triple-negative breast cancer (TNBC), the most aggressive breast cancer subtype, is associated with high rates of recurrence and metastasis (1–5). To date, unlike other breast cancer subtypes, no specific targeted approaches exist for TNBC, and toxic chemotherapeutic agents are the primary treatment regimen, highlighting the need to develop new targeted therapies. TNBC contains a high degree of cellular heterogeneity and plasticity that has emerged as a hallmark of the malignant state and accounts for the observed persistent tumor growth, therapeutic resistance, and metastasis (6–9). Given the limited treatment options for TNBC, more effective treatment strategies are an immediate priority for therapeutic development.

The elevated cellular density within tumors stimulates cellular programs that are activated by cell–cell contact and close proximity. The gap junction family of proteins is composed of connexin subunits and canonically functions in gap junction plaques at the interface of adjacent cells to facilitate direct cell–cell communication. Connexins can also function noncanonically as single membrane channels (hemichannels, which contain a single assembled connexon hexamer of connexins) or as signaling hubs adjacent to any organelle and/or the plasma membrane (10). In cancer, the classical role of connexins as tumor suppressors has been widely described in many tumor models (11–15). However, the paradigm that connexins have a global tumor-suppressive role has been challenged by recent evidence that connexins can be protumorigenic by facilitating tumor progression and metastasis (16–22). The connexin subunits involved in maintaining homeostasis during the physiologic development and function of the mammary gland are Cx26 and Cx43 (13, 23). In breast cancer, connexins have been described to be both pro- and antitumorigenic by regulating transformation, proliferation, cell survival, and metastasis (22). Most studies to date suggest a tumor-suppressive role for Cx26 early in breast cancer progression based on evidence that Cx26 is frequently absent or downregulated in human breast cancer cell lines and human primary tumors (10, 24–26). However, these studies are inconsistent with clinical observations that demonstrate a strong correlation between increased Cx26 expression in breast cancer tissue samples posttreatment (chemotherapy/surgery) and decreased overall survival (27). Moreover, Cx26 expression has been shown to be associated with increased lymphatic vessel invasion, increased tumor size, and poor prognosis in human breast cancers (27, 28).

We previously demonstrated that Cx26 is elevated in TNBC cell lines and patient-derived xenograft models compared with normal mammary tissue and other subtypes of breast cancer (29). In functional studies, we demonstrated that Cx26 is necessary and sufficient for TNBC cell tumorsphere formation and regulates NANOG protein stability. In TNBC, Cx26 localizes to an intracellular membrane–bound vesicle in complex with the pluripotency transcription factor NANOG and focal adhesion kinase (FAK). In contrast to its role in the normal mammary epithelium, Cx26 functions in a gap junction–independent manner in TNBC by forming an intracellular complex that underlies self-renewal and stabilizes NANOG. We leveraged these initial findings to develop a peptide-based strategy to disrupt the signaling of the complex members. We developed peptides to the intracellular and extracellular regions of Cx26 and identified the binding domains necessary for interaction with NANOG and FAK. Using a cell-penetrating peptide strategy, we show that introduction of the Cx26 C-terminus into TNBC cells inhibits self-renewal, reduces NANOG levels and target gene expression, and attenuates in vivo tumor growth. Together, this study provides proof of concept that targeting TNBC via connexin peptide mimetics represents a viable therapeutic strategy.

All mouse procedures were performed under adherence to protocols approved by the Institute Animal Care and Use Committee at the Lerner Research Institute, Cleveland Clinic (#2018–2101). All human specimens were obtained under adherence to protocols approved by the Cleveland Clinic Foundation Institutional Review Board (approval #17–418).

Annotated cases of invasive ductal carcinoma were retrieved from the surgical pathology files of Cleveland Clinic and included triple-negative breast carcinoma (TNBC), hormone receptor positive (HR⁺ being estrogen receptor positive) and HER2 negative (HR⁺/HER2⁻), hormone receptor negative (HR⁻ being estrogen receptor negative) and HER2 positive (HR⁻/HER2⁺) or hormone receptor positive and HER2 positive (HR⁺/HER2⁺). In total, 12 TNBC, 12 HR⁺/HER2⁻, 5 HR⁻/HER2⁺, 4 HR⁺/HER2⁺ tumors were evaluated. Hematoxylin and eosin–stained slides were reviewed for invasive tumor adequacy and IHC was performed on 4–μm–thick formalin-fixed paraffin-embedded (FFPE) whole tissue serial sections following antigen retrieval (Target Retrieval Solution, pH 6.1; Dako). Immunostains performed included Connexin 26 (Cx26) mouse mAb (clone CX-12H10, Thermo Fisher Scientific) at 1:200 dilution, NANOG rabbit monoclonal antibody (clone EPR2027, Abcam) at 1:250 dilution, and FAK rabbit polyclonal antibody (clone Q05397, Cell Signaling Technology) at 1:500 dilution. All staining was performed on an automated platform (Ventana Benchmark, Ultraview XT, Ventana Medical Systems, Inc) and visualized with ultraView Universal DAB Detection Kit (Ventana Medical Systems, Inc). The cellular compartment staining (membranous, cytoplasmic or nuclear) was evaluated and the intensity in each case for the three antibodies was characterized as weak, moderate, or strong, and the extent of immunoreactivity scored according to the percentage of tumor cells with staining: 0 (0%), 1+ (<5%), 2+ (5%–25%), 3+ (26%–50%), 4+ (51%–75%), and 5+ (76%–100%).

MDA‐MB‐231, MDA-MB-468, HCC70 (BCRJ, catalog no. 0386, RRID:CVCL_1270), and MCF7 breast cancer cells (ATCC) were cultured in log‐growth phase in DMEM supplemented with 1 mmol/L sodium pyruvate (Cellgro) and 10% heat‐inactivated FCS at 37°C in a humidified atmosphere (5% CO₂). All cell lines were authenticated through the Cleveland Clinic Lerner Research Institute Cell Culture Core and routinely tested for Mycoplasma contamination.

All peptides, unless otherwise noted, were synthesized by the LRI Molecular Biotechnology Core through standard solid-phase chemistry, using Rink Amide resin and side chain–protected Fmoc-amino acids, on a Liberty Blue Automated Microwave Peptide Synthesizer (CEM Corporation). Each peptide was purified to >95% by acetonitrile gradient elution (20%–50%) on a Vydac preparative C18 column, using reverse-phase HPLC (Waters Corporation). The quality of the peptide was analyzed by HPLC (Agilent 1100) on a reverse-phase analytical C-18 column. The correct mass corresponding to the amino acid composition of each peptide was then confirmed by mass spectrometry (MALDI-TOF/TOF mass spectrometer; model no. 4800; AB SCIEX). The peptides were lyophilized and stored at −80°C until reconstituted in 1% DMSO/PBS.

AlphaFold2 was used to predict the 3D structure of Cx26, NANOG, and FAK: (https://www.blopig.com/blog/2021/07/alphafold-2-is-here-whats-behind-the-structure-prediction-miracle/). Images were rendered using Visual Molecular Dynamics software (30).

The equilibrium dissociation constants (K_d) between synthesized Cx26 cytosolic tail peptides and either NANOG or FAK proteins were determined using the Biacore S200 (Cytiva) at 25°C. The 1% DMSO/1× PBS-P+ (Cytiva) buffer was filtered through a 0.22-μm membrane filter and degassed immediately before the experiment as running buffer with a flow rate of 30 μL/minute for the instrument system. The purified active NANOG was diluted in 10 mmol/L sodium acetate solution pH 4.5, resulting in a protein concentration of 50 μg/mL, and purified active FAK was diluted in 10 mmol/L sodium acetate solution pH 5.0, resulting in a protein concentration of 20 μg/mL. The diluted protein was immobilized on the surface of a Series S Sensor Chip CM5 via the primary amine group, employing a standard Amine Coupling Kit (1-ethyl-3-(3-dimethylaminopropyl)-carbodiimide hydrochloride (EDC), N-hydroxysuccinimide (NHS) and 1.0 mol/L ethanolamine-HCl (pH 8.5), and the target immobilization level was 5,000 response units (RU). The flow cell 1 without any modification was used as a reference to correct for system error. The RU values were collected, and all the binding affinity data were calculated by kinetic models (1:1 Langmuir interaction) within the Biacore S200 Evaluation Software. To determine the binding affinity between each peptide and NANOG or FAK, a series of peptide dilutions was analyzed by multicycle kinetics/affinity. A concentration gradient of each peptide as the analyte was freshly prepared in running buffer over at least five concentrations (100 nmol/L–100 μmol/L). Each peptide at various gradient concentrations (including a repeat concentration) and one zero concentration (running buffer) flowed over the immobilized NANOG or FAK, with 180 seconds for binding, followed by disassociation for 600 seconds, and the obtained RUs were recorded. After the injection of each concentration of each peptide, the sensor chip surface was regenerated with 10 mmol/L glycine pH 2.5 for 30 seconds to completely remove the residual peptide. The K_d was calculated to evaluate the ability of peptides to interact with NANOG or FAK proteins.

After peptide treatment, MDA‐MB‐231 cells were lifted and washed in 1× PBS followed by staining with UV Live/Dead dye (1:1,000, Thermo Fisher Scientific) for 15 minutes at room temperature. Cells were then fixed and permeabilized using the Annexin V Staining Kit (BD Biosciences) as per the manufacturer's instructions and incubated with FITC-conjugated anti-antennapedia (anti-Antp-FITC) antibody (1:100, LSBio) for 30 minutes. Samples were washed and analyzed using a BD Fortessa flow cytometer. Data analysis was performed using the FlowJo software v10 (FlowJo, RRID:SCR_008520; Tree Star, Inc., Ashland, OR).

For immunofluorescence staining, MDA-MB-231 cells were plated onto glass coverslips coated with Geltrex (Thermo Fisher Scientific) in 24-well plates and treated with aCx26-pep (10 μmol/L) for 16 hours. Subsequently, coverslips were washed with 1× PBS, fixed with 4% paraformaldehyde, and either permeabilized with 0.1% Triton X-100 or left nonpermeabilized. Cells were then stained with anti-antennapedia antibody (anti-Antp 4C3, 1:100, University of Iowa Developmental Studies Hybridoma Bank, DSHB catalog no. anti-Antp 4C3, RRID:AB_528082) for 1 hour at room temperature. Coverslips were washed twice with 1× PBS and stained with Texas Red–conjugated anti-mouse secondary (1:1,000, Invitrogen) for 1 hour at room temperature in the dark, before being mounted in DAPI mount (VWR) onto glass slides and imaged with a confocal microscope at 63× magnification (Leica). Appropriate secondary alone controls were completed to confirm the absence of background staining.

MDA-MB-231 cells were plated in a 96-well (opaque-walled) plates at 1,000 cells/well and treated for 72 hours with Antp peptide alone, aCx26-pep or vehicle (1% DMSO in PBS) at the concentrations shown in the figures. At day 0 and post 72 hours of treatment, cells were lysed with CellTiter-Glo Luminescent Cell Viability Assay reagent (Promega), shaken in the dark for 10 minutes, and luminescence was recorded. Fold change was calculated by normalizing the 72-hour luminescence value to day 0 for each group.

Duplicate rows of cells were cultured in serial dilutions per well (1–20 cells per well) in one 96‐well plate per condition with 200 μL serum‐free DMEM supplemented with 20 ng/mL basic fibroblast growth factor (Invitrogen), 10 ng/mL EGF (BioSource), 2% B27 (Invitrogen), and 10 μg/mL insulin. The frequency of tumorsphere formation was calculated in such a way that a well with a tumorsphere was counted as a positive well and a well with no tumorspheres was counted as a negative well. Tumorspheres were counted under a phase-contrast microscope after 2 weeks. The stem cell frequencies were calculated using an extreme limiting dilution algorithm (http://bioinf.wehi.edu.au/software/elda/; ref. 31).

A total of 1.5×10⁷ cells were plated in a 15-cm dish and treated with 10 μmol/L peptide or 1% DMSO vehicle for 48 hours. Cells were subsequently lysed and subcellular fractions prepared using the Subcellular Fractionation Kit for Cultured Cells (Thermo Fisher Scientific) according to the manufacturer's protocol. A total of 25 μg of each fraction was loaded on a 12% SDS-PAGE gel and transferred to polyvinylidene difluoride membrane. Membranes were blocked in 5% nonfat milk in TBST for 1 hour and incubated with the appropriate primary antibody overnight at 4°C: anti-NANOG [(Cell Signaling Technology, catalog no. 4903 (1:1,000) RRID:AB_10559205), anti-FAK (D2R2E; Cell Signaling Technology, catalog no. 13009, RRID:AB_2798086; 1:1,000), anti-Lamin A/C (ProteinTech, catalog no. 10298–1-AP, RRID:AB_2296961; 1:3,000)]. Membranes were then washed in TBST and incubated with the appropriate horseradish peroxidase–tagged secondary antibody (Sigma-Aldrich) for 3 hours at room temperature. Rhodamine-tagged anti-actin (Bio-Rad #12004163; 1:10,000) was used as a loading control. All immunoblots were imaged on a Bio-Rad ChemiDoc MP.

qPCR was performed using a StepOnePlus Real-Time PCR System (Applied Biosystems) with SYBR‐Green MasterMix (SA Biosciences). RNA was extracted using the RNeasy kit (Qiagen), and cDNA was synthesized using the Superscript III kit (Invitrogen). For qPCR analysis, the threshold cycle (C_t) values for each gene were normalized to expression levels of GAPDH. Dissociation curves were evaluated for primer fidelity. The primers used are detailed in Supplementary Table S1.

NOD.Cg-Prkdc^scid Il2rg^tm1Wjl/SzJ (NSG) female mice were purchased from Jackson Laboratories or the Biological Resource Unit at the Lerner Research Institute, Cleveland Clinic (Cleveland, OH). All mice were maintained in microisolator units and provided free access to food and water. MDA‐MB‐231 cells were transduced with a luciferase lentiviral vector (pHIV-Luciferase, RRID:Addgene_21375). Following puromycin drug selection, 1 × 106 cells were injected into the right subcutaneous flank of 6-week-old female mice. Once tumors became palpable, mice were randomized into groups and received a total of 50 μL of either 0.1 mmol/L peptide or vehicle (1% DMSO) via intratumoral injection every other day for 20 days. Tumor progression was assessed weekly via bioluminescent IVIS imaging (see below). Following treatment regiment, mice were euthanized by CO₂ asphyxiation with subsequent cervical dislocation and necropsied for further analysis.

Bioluminescence images were taken with an IVIS Lumina (PerkinElmer) using d-luciferin as previously described (32). Mice received an intraperitoneal injection of d-luciferin (Goldbio LUCK-1G, 150 mg/kg in 150 μL) under inhaled isoflurane anesthesia. Images were analyzed (Living Image Software), and total flux were reported in photons/second/cm²/steradian for each mouse.

Upon necropsy, subcutaneous tumors were excised, sectioned, and placed in 4% paraformaldehyde (PFA) for 3 days at room temperature, before being transferred into 70% ethanol and sent to the Histology Core at the Lerner Research Institute (Cleveland, OH) for processing, sectioning, and staining. Slides were imaged using the Aperio slide scanner and analyzed.

Values reported in the results represent mean ±  SEM. One way analysis of variance was used to calculate statistical significance, and P values and replicate numbers are detailed in the text and figure legends. All experiments were repeated at least three separate times, with technical replicates performed within each experimental replicate as appropriate.

The data generated in this study are available upon request from the corresponding author.

Previous data from our group identified a protein complex unique to TNBC composed of the gap junction protein connexin 26 (Cx26), the pluripotency transcription factor NANOG, and FAK (29). To assess whether the Cx26/NANOG/FAK complex members are also present in tumors from patients, we stained sections of human triple-negative invasive ductal carcinoma for Cx26, NANOG, and FAK. Hematoxylin and eosin staining showed the invasive nature of the tumors (Fig. 1A). In cultured TNBC cells, we found that Cx26 localizes to an intracellular compartment rather than the canonical plasma membrane localization of connexins (29). Consistently, we observed intracellular localization of staining for Cx26, NANOG, and FAK in adjacent tumor sections of TNBC patient tumor (Fig. 1B–D). In contrast, tumor samples from patients with hormone receptor–positive and HER2-negative tumors showed cell surface localization of Cx26 distinct from NANOG and FAK (Fig. 1E–H; summarized in Fig. 1I). This indicates that Cx26, NANOG, and FAK have similar subcellular localizations in TNBC, suggesting that the complex may be present in TNBC patient tumors as well as in cells, as we previously described (29).

On the basis of the essential role for Cx26 in TNBC cells and its formation of a complex with NANOG and FAK (29), targeting the complex and its members represents a therapeutic opportunity. To aid in therapeutic development, we used surface plasmon resonance (SPR) to interrogate the specific domains through which the proteins interact. We generated peptide fragments representing the intracellular and extracellular domains of Cx26 and determined their ability to bind to either FAK or NANOG by SPR. The strongest binding occurred between NANOG and the Cx26 cytosolic tail and between FAK and both the cytosolic tail and the cytosolic loop, with dissociation constants in the nanomolar and micromolar ranges, respectively (Fig. 2A). Importantly, no binding was observed between the FAK and NANOG proteins alone (Fig. 2A). These observations combined with predictive modeling suggest that complex binding requires Cx26, as we detected binding between Cx26 and NANOG, Cx26 and FAK, and all members of the complex but not between FAK and NANOG alone (Supplementary Fig. S1). These data informed a binding schematic for the Cx26/NANOG/FAK complex (Fig. 2B) where NANOG binds to the C-terminus while FAK interacts with both the C-terminus and the intracellular loop of Cx26. To determine the specific Cx26 cytosolic tail amino acid residues important for binding, several truncations of the C-terminal tail were synthesized and subjected to SPR analysis for binding to both NANOG and FAK. The minimal required region mapped to Cx26 amino acids 216–220 (Supplementary Table S2; Supplementary Figs. S2 and S3).

On the basis of the observation that the Cx26 cytosolic tail is essential for interaction with NANOG and FAK, we sought to understand the effect of targeting members of the complex using a Cx26 cytosolic tail–derived cell-penetrating peptide. We tagged a peptide containing the Cx26 cytosolic tail sequence with an Antp (RQIKIWFQNRRMKWKK) leader sequence (termed hereafter aCx26-pep). Antp is a well-studied leader sequence used to allow peptides and small molecules to penetrate the plasma membrane (33) and has been used to tag other connexin mimetic peptides. aCT1, an antennapedia-tagged Cx43 C-terminal tail mimetic peptide, impairs the malignant progression of luminal breast cancer cells, and enhances the activity of the standard-of-care chemotherapies lapatinib and tamoxifen (34, 35). We first confirmed that the Antp leader sequence did not significantly alter the ability of aCx26-pep to bind to either FAK or NANOG by SPR (Fig. 2C). To determine the optimal treatment conditions, cells were assessed for the presence of aCx26-pep by flow cytometry (Fig. 2D) and intracellular immunofluorescence (IF) staining (Fig. 2E) using an anti-Antp antibody. Optimal treatment conditions were determined to be 10 μmol/L peptide for a minimum of 16 hours.

To assess the functional impact of treatment, we evaluated cell proliferation and self-renewal in TNBC cells treated with aCx26-pep in vitro. MDA-MB-231 TNBC cells were treated with increasing concentrations of Antp alone and aCx26-pep and cell proliferation was measured after 72 hours of treatment. Interestingly, cell proliferation was not significantly affected by treatment with aCx26-pep (Fig. 3A and B). To assess changes in tumorsphere formation, we treated three TNBC cell lines (MDA-MB-231, MDA-MB-468, and HCC70) and one luminal breast cancer cell line (MCF7) with vehicle, Antp, Antp-tagged scrambled aCx26-pep (aScr-pep), and aCx26-pep. Tumorsphere-initiation frequency was assessed using an extreme limiting-dilution assay (ELDA) and subsequently analyzed using ELDA analysis software (31). TNBC cells treated with aCx26-pep had significantly decreased tumorsphere frequency compared with those treated with aScr-pep, whereas the luminal breast cancer cell line showed no significant difference between aScr-pep and aCx26-pep (Fig. 3C–F; Supplementary Table S3; Supplementary Fig. S4). A shortened region of the Cx26 cytosolic tail tagged with Antp, aCx26-pep 216–220, showed similar effects to aCx26-pep (Supplementary Fig. S5). Together, these data show that targeting members of the Cx26/NANOG/FAK complex using aCx26-pep decreases the tumorsphere formation capacity of TNBC.

To evaluate the effect of aCx26-pep on members of the Cx26/FAK/NANOG complex, we first measured the effect of the peptide on pluripotency transcription factors. While we did not observe a change in SOX2 levels, there was a significant decrease in OCT4 mRNA levels in both MDA-MB-231 and HCC70 cells treated with aCx26-pep compared with aScr-pep (Fig. 4A). We next interrogated whether peptide treatment affected the levels and subcellular localization of NANOG and FAK. After 48-hour treatment with aCx26-pep, HCC70 cells were subjected to subcellular fractionation and immunoblotting for FAK and NANOG (Fig. 4B). We observed dramatic decreases in both FAK and NANOG levels in the nuclear fraction of TNBC cells, suggesting that treatment with aCx26-pep may destabilize these proteins. This recapitulates the destabilization of NANOG we previously observed after knockdown of Cx26 (29). To further understand the downstream signaling changes following treatment of TNBC cells with aCx26-pep, we performed a targeted qPCR assessment of NANOG-regulated genes. After 7 days of treatment with vehicle, aScr-pep, or aCx26-pep, expression of NANOG target genes including FABP5, POSTN, NRP2, PIN1, and TAGLN was significantly decreased in MDA-MB-231 TNBC cells treated with aCx26-pep compared with those treated with vehicle or aScr-pep (Fig. 4C). In addition, expression of epithelial cell markers such as cytokeratin 18 and 19 (KRT18 and KRT19) was significantly increased in the aCx26-pep treatment group (Fig. 4D). In contrast, treatment of MCF7 luminal breast cancer cells did not significantly change NANOG target gene or cytokeratin gene expression (Fig. 4D–F). Together, these results indicate decreased levels of FAK and NANOG, reduced transcriptional activity of NANOG, and an increase in epithelial cell signature after treatment with aCx26-pep.

To determine the effect of aCx26-pep on tumor formation in vivo, NSG mice were subcutaneously injected into the flank with TNBC MDA-MB-231 luciferase–expressing cells and randomized into treatment groups once tumors became palpable. Mice were intratumorally injected every other day for 20 days with either vehicle, aScr-pep, or aCx26-pep and monitored by weekly bioluminescence imaging (Fig. 5A). Mice treated with aCx26-pep exhibited significantly reduced luciferase signal (tumor burden) over time when compared to those treated with aScr-pep or vehicle controls (Fig. 5B and C).

We then performed histologic assessment of the resulting tumors. Upon H&E staining of isolated tumors, we observed that the percent tumor viability was significantly decreased in aCx26-pep–treated mice compared with vehicle- or aScr-pep–treated mice (Fig. 6A). Immunofluorescence staining of the endpoint tumors showed that Ki67 (Fig. 6B), a marker for proliferation, and vimentin (Fig. 6C), a marker for epithelial–mesenchymal transition (EMT), were decreased, while cleaved caspase-3 was increased (Fig. 6D), in the aCx26-pep–treated groups, suggesting that peptide treatment led to a more epithelial cell phenotype with increased cell death and diminished proliferation. Taken together, these data support the efficacy of aCx26-pep for specifically targeting TNBC.

Overall, our findings provide proof-of-concept data that a peptide strategy can be used to reduce self-renewal and tumor growth. We leveraged previous observations of the Cx26/NANOG/FAK complex to develop a preclinical peptide-based targeting strategy for TNBC. Our studies focused solely on tumorsphere formation and tumor growth, key cancer phenotypes that likely involve FAK and NANOG. However, given the number of downstream signaling networks associated with FAK and NANOG, there are likely to be other cellular programs that may also be driven by their association with Cx26, including adhesion, migration, and invasion. An additional unanswered question is the possibility of differential functions of free and Cx26 complex–bound NANOG and FAK.

IHC analysis complemented our breast cancer cell line data, with the majority of TNBC specimens having cytoplasmic expression of Cx26 and a subset exhibiting a similar staining pattern of FAK and NANOG. In contrast, the majority of HR⁺/HER2⁻ carcinomas showed membranous expression of Cx26 and cytoplasmic localization of FAK and NANOG. The distinct cellular localization of Cx26 expression in TNBC and HR⁺/HER2⁻ tumors further supports the hypothesis that therapeutic targeting of Cx26 would benefit this aggressive tumor subtype. Although the IHC TNBC data are not as compelling as those seen in the TNBC cell lines, FFPE IHC may suffer from antibody affinity, fixation issues, and antigen retrieval. Clinical outcomes data related to a larger cohort defined with Cx26, FAK, and NANOG will be helpful in further elucidating this complex and its role.

Our findings identify a strategy to treat TNBC by targeting Cx26 and the NANOG-associated signaling network using a cell-penetrating peptide approach. A challenge in targeting connexin family members remains specificity, as there is a high degree of structural homology among connexin family members, apart from the cytoplasmic tail that provides a binding surface for adaptor proteins to initiate signaling programs. In previous work, the cytoplasmic tail of Cx43 was found to associate with zona occludens-1 (ZO-1) to exclude assembled connexins from the gap junction plaque; disruption of the interaction with ZO-1 resulted in an increase in gap junction intracellular communication (35). A peptide mimetic of this region of the Cx43 tail tagged with antennapedia has been successfully leveraged to enhance diabetic would healing and is in advanced stages of clinical testing (36), and a TAT fusion of a region of Cx43 that binds to SRC has been shown to decrease glioblastoma cancer stem cell (CSC) phenotypes (37–41). These findings provided the first example that peptides can be used to modulate intracellular connexin function for a clinically relevant application. Here, we provide the evidence for use of a peptide-based connexin targeting approach for TNBC and show that this strategy also modulates downstream signaling. While these peptide-based approaches show promise, there remain barriers to implementation, including delivery and stability. While we primarily tested aCx26-pep, which contains the full 11 amino acid cytosolic tail sequence, we observed similar effects with the first five amino acids of the cytosolic tail, RYCSG, which corresponds to amino acids 216–220. Further study is required to determine the optimal peptide sequence for both in vivo efficacy and stability. Differences in concentrations of peptide available, as well as the presence of the tumor microenvironment, may have been responsible for slight differences between peptide effects observed in vivo and in vitro, such as on tumor cell viability. An additional question is the effect of our targeting peptide on normal and CSC function. On the basis of our observation that aCx26-pep does not affect luminal breast cancer cells such as MCF7, which also express FAK and NANOG, we would anticipate limited toxicity to normal cells, including normal stem cells. Future studies will interrogate peptide specificity for targeting CSCs and non-CSCs in TNBC. The question of whether aCx26-pep is able to disrupt the Cx26/NANOG/FAK complex remains and will also be the focus of future work. Beyond this mechanistic analysis, development of the peptide-based connexin therapeutic will also require a deep understanding of its on- and off-target toxicity as part of preclinical drug discovery.

Connexins and their main functions of cell–cell communication, hemichannel communication, and initiation of signaling have been largely understudied in the context of cancer and may represent a new therapeutic target. These findings highlight a selective strategy using a peptide approach to target TNBC, leading to new therapeutic options for the unmet need in this patient population.

A. Myers reports grants from U.S. Department of Defense during the conduct of the study. I.N. Smith reports grants from NIH NIGMS K99GM143552 and grants from Ambrose Monell Cancer Genomic Medicine Fellowship during the conduct of the study. J.D. Lathia reports a patent 11136368 issued. O. Reizes reports grants from Department of Defense, grants from NIH/NCI, and grants from VeloSano Bike to Cure during the conduct of the study; in addition, O. Reizes has a patent for US 11,136,368 issued. No disclosures were reported by the other authors.

E.E. Mulkearns-Hubert: Data curation, investigation, methodology, writing–original draft, writing–review and editing. E. Esakov Rhoades: Data curation, formal analysis, validation, investigation, methodology, writing–original draft. S. Ben-Salem: Formal analysis, investigation, methodology. R. Bharti: Validation, investigation, writing–review and editing. N. Hajdari: Investigation, methodology. S. Johnson: Investigation, methodology. A. Myers: Investigation, methodology. I.N. Smith: Formal analysis, visualization. S. Bandyopadhyay: Investigation, methodology. C. Eng: Resources, formal analysis. E. Downs: Data curation, investigation, visualization, methodology, writing–original draft, writing–review and editing. J.D. Lathia: Conceptualization, resources, formal analysis, supervision, funding acquisition, validation, writing–original draft, project administration, writing–review and editing. O. Reizes: Conceptualization, resources, supervision, funding acquisition, writing–original draft, project administration, writing–review and editing.

The authors would like to thank members of the Lathia and Reizes laboratories for insightful comments during implementation of these studies. We appreciate the technical contributions of Chad Braley; members of the Lerner Genomics Shared Lab Resources, specifically Dr. Dean Horton; and the Imaging Core, namely Dr. Judith Drazba. This research was supported by the Department of Defense Grant W81XWH1910503, NIH R21 CA191263, and VeloSano Bike to Cure. Work in the Reizes laboratory is also supported by Cleveland Clinic Center of Excellence, VeloSano Bike to Cure, and The Laura J. Fogarty Endowed Chair for Uterine Cancer Research.

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.