The Src inhibitor peptide TAT-Cx43 improves survival in an intracranial murine model of lung cancer brain metastasis.

Methods
For detailed protocols, see Supplementary Information.

Animals
Equal number of male and female C57BL/6 mice were obtained from Charles River Laboratories, Jackson Laboratories, or the Experimental Animal Service (SEA) of the University of Salamanca. All animal procedures were previously approved by the ethics committee of the University of Salamanca and the Junta de Castilla y León (Spain) (CBE 697 and 895) and performed agreeing with the European Community Council directives (2010/63/UE), and Spanish law (RD 53/2013 BOE 34/11370-420, 2013) for the use and care of laboratory animals.

Cells
Mouse lung cancer Lewis Lung Carcinoma (LLC), mouse lung cancer LSZ4 cells, human lung cancer A549 cells, and human lung cancer H441 cells were selected as KRAS-mutated NSCLC model and cultured in 2D adherent or 3D non-adherent cancer stem-like cell (CSC) conditions. For 2D adherent conditions, cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% FBS, while for 3D non-adherent CSC conditions FBS was gradually decreased and once cells were detached from the petri dish, they were cultured in stem-cell medium containing 20 ng/mL EGF, 20 ng/mL bFGF and 1% B27 in non-treated flasks for culturing cell in suspension until tumorspheres were formed.27,33

Treatments
Peptides were obtained from GenScript with purity higher than 95%. Control peptide TAT sequence was YGRKKRRQRRR and TAT-Cx43266-283 sequence was TAT-AYFNGCSSPTAPLSPMSP (patent ID: WO2014191608A1).20 For cell treatments, peptides were used at 50 μM and selumetinib at 10 µM in culture medium. For intracranial injections, TAT-Cx43266-283 was injected at 100 μM in 1 µl of saline, together with 5,000 LLC-CSCs. One week after implantation, mice were intraperitoneally injected with 4 nmol/g TAT-Cx43266-283, twice per week until the end of the experiment.

Stereotactic Surgery for Intracranial LLC-CSC Implantation
LLC-CSCs were intracranially implanted into the brains of 8-week-old C57BL/6 mice. Equal number of female and male mice were used. Mice were anaesthetized using isoflurane while placed in a stereotaxic frame. Stereotactic coordinates were used (1.5 mm AP, +1.5 mm LM to bregma) to trephine the skull and make the brain tissue accessible for injection. Saline (1 μL) containing 5,000 cells was injected per animal at -1 mm deep into the brain tissue using a Hamilton microsyringe coupled to a 26G needle. At the indicated times, brains were processed and sliced into coronal sections.

H&E Staining and Immunohistochemistry
Paraffin-embedded brain sections were used for H&E staining, and cleaved Caspase-3 and Ki-67 immunohistochemistry. These procedures were performed by the Compared Molecular Pathology service of Cancer Research Center (IBMCC, Salamanca). Bright-field images were acquired, and H&E staining images were evaluated by the pathologist Dr Carmen García Macías (IBMCC, University of Salamanca).

Phosphoproteomic Analysis
For phosphoproteomic analysis, extraction was carried out in protein extraction buffer (supplemented with protease and phosphatase inhibitor cocktails) or phosphoproteomic protein extraction buffer (supplemented with NaF, Na3VO4, Sodium β-Glycerophosphate, and Na2H2P2O7) were used.34 For frozen brain tissue, coronal slices were obtained and the tumor area together with the closest surrounding microenvironment was isolated in addition to the contralateral side to the tumor tissue. Protein extraction buffer was added, until tissue was completely homogenized. Samples were then transferred into clean Eppendorf tubes, heated, or thawed in ice in the case of future phosphoproteomic analysis, and sonicated before centrifugation. In all cases, protein quantification was performed.
For sample preparation, protein lysates from each sample were transferred to clean Eppendorf Protein LoBind S/L Tubes and topped up to a defined final volume. Then, cysteine reduction and alkylation were performed by sequentially adding dithiothreitol and iodoacetamide solutions and incubating with agitation in the dark for 1 h and 30 min, respectively. Then, tryptic digestion was performed overnight, and the supernatant was transferred to clean Eppendorf Protein LoBind S/L Tubes and stored at −80°C. Samples were transferred to 96-well plates and subsequently acidified adding trifluoroacetic acid. Desalting and phosphopeptide enrichment procedures were performed and the eluted samples were dried in a SpeedVac. Peptide pellets were stored at −80°C.

LC-MS/MS and Phosphoproteomic Data Analysis
For peptide sample reconstitution, samples were thawed in ice, reconstituted, and sonicated.35 Samples were then left on ice and centrifuged. The sample supernatants were transferred to LC-MS/MS vials and injected, one at a time, into an LC-MS/MS system consisting of a nano flow ultra-high pressure liquid chromatography UltiMate 3000 RSLC nano directly coupled to an Orbitrap Q-Exactive Plus mass spectrometer using an EASY-Spray system. The LC system used two mobile phases. Peptides were trapped in a μ-precolumn and separated in an analytical column. Then, the column was washed and equilibrated. Peptides eluted from the LC system and were infused into the online connected Q-Exactive Plus system. The instrument acquired full scan survey spectra followed by data-dependent acquisition in which the 15 most intense ions were selected for dissociation and MS/MS scanning. A dynamic exclusion period of 30 s was enabled with m/z window of ±10 ppm. The Q Exactive Plus operated a duty cycle of 2.1 s.
The resulting MS RAW data files were loaded into Mascot Daemon (v2.8.0.1) for automated peptide identification.35 Peak list files from RAW data files were generated with Mascot Distiller and loaded into the Mascot search engine. Searches were performed against the SwissProt Database (SwissProt_2021_02.fasta) specifically for Mus musculus. Pescal (Peak Statistic Calculator)36 was used to generate the extracted ion chromatograms (XICs) for all the peptides identified in at least one sample. Peak area values were quantified and individual peptide intensity values in each sample were normalized.
Pescal outputs were further processed with the protools2 package (v0.2.9) in a R environment for statistical analysis and data representation. Phosphoproteomic data was normalized, and statistical differences computed using limma37 and Student’s t-test. Kinase-Substrate Enrichment Analysis (KSEA) and Pathway Enrichment Analysis (PEA) were carried out as described before35,38 to estimate kinase activity and pathway enrichment. PhosphositePlus® v6.7.7 (Cell Signaling) was also used to conduct the search for the phosphosites of interest and their functional relevance. As the aim of the project was to generate hypotheses from the phosphoproteomics data, P-values were not adjusted for multiple testing and therefore there is a risk of false positives. Consequently, to confirm these results and their biological relevance, conclusions derived from the data were corroborated using targeted immunochemical methods and functional assays.

Functional and Biochemical Assays
Alamar blue viability assay, Matrigel invasion assay, WB and Immunofluorescence analyses were performed as described in detailed in the Supplementary Information.

Image Analysis
Image processing package Fiji software39 was used for image analysis. For immunohistochemistry a common threshold was applied to the 8 bit images, and for the images of cleaved Caspase-3 and Ki-67 the number of particles was counted, while for CD31 the area of the segmented region was measured.

Cell Random Movement Analysis
LSZ4 cells were plated at a low density in adherent conditions in DMEM supplemented with 10% FBS or in stem cell medium.4 TAT-Cx43266-283 was added at 50 μM and random cell movement was recorded by time-lapse live-cell imaging every 10 min with an inverted Zeiss Axio Observer Z1 microscope for live-cell imaging (Carl Zeiss Microscopy, LLC, USA) coupled to an AxioCam MRm camera. Cell movement was manually tracked using Fiji software. Average cell velocity was calculated for up to seven cells per field in three replicates and three independent experiments.

Patient Protein Level Analysis
The Cancer Genome Atlas (TCGA) NSCLC adenocarcinoma (TCGA, Firehose Legacy) database40 (584 patients) containing reverse phase protein array (RPPA) data was used. cBioportal41 was used for downloading data, plots, and analyses.

Statistical Analysis
In addition to the above-mentioned statistical analysis of phosphoproteomic data, results were represented in the plots as means ±SEM of at least 3 independent experiments. For comparisons between two groups, Shapiro-Wilk normality test was performed. For normal distribution of the data, a paired or unpaired two-tailed Student’s t-test was applied. When normality could not be assumed, Mann-Whitney test was performed. For comparisons between more than two groups, one- and two-way ANOVA, followed by Tukey’s multiple comparisons test, were used. For survival analysis, data were represented in a Kaplan-Meier curve and differences were calculated using a Log-rank (Mantel-Cox) test. In all cases, differences between groups were considered significant when P-value ≤.05. Statistical analysis and results representation were carried out using GraphPad Prism 8.

Results

Effect of TAT-Cx43266-283 in NSCLC Cells, LLC, In Vitro
Our previous studies showed that TAT-Cx43266-283 affected GSCs but not differentiated glioblastoma cells.27 Therefore, we aimed to investigate whether this was also the case in LLC cells, a well-studied model of NSCLC adenocarcinoma, which carries Kras mutations—one of the most common driver mutations in NSCLC.42,43 To do so, we isolated the subpopulation of CSCs from LLC cell cultures, as previously described.27 Briefly, LLC cells were cultured in adherent conditions, decreasing one half of the FBS concentration every 3 days until cells detached from culture surface. After approximately 3 passages (1-2 weeks), cells were cultured as spheres in non-adherent conditions, in stem cell medium containing B27, EGF and b-FGF (Figure S1A) for 3-4 weeks (3-4 passages) until non-adherent small cell aggregations were formed. As previously reported,27 non-adherent LLC cells grew as round cells, some of them forming sphere-like structures, in contrast to the mixed morphology observed in adherent LLC cells (Figure 1A and B and Figure S1B). Once the selection of both subpopulations was performed, we evaluated the levels of the TAT-Cx43266-283 target, Src and Src active form (p-Src Y416) (Figure S1C). Similarly to GSCs,24 we found increased levels of Src and its activity in non-adherent LLC-CSCs compared to adherent LLCs. Intriguingly, Src showed a small change in its molecular weight between both cell types, suggesting that a different member of the Src family might be expressed in non-adherent LLC-CSCs. We also analyzed the levels of Cx43 and Cx26. As expected, Cx26 is upregulated and Cx43 is downregulated in non-adherent LLC-CSCs compared to adherent LLC cells (Figure S1C and D), which is consistent with the key role of Cx26 in promoting CSC phenotype44 and the opposite role of Cx43.45
Next, we analyzed the effect of TAT-Cx43266-283 on the cell growth and viability of these cells for 144 h. As shown in Figure 1A and B, while adherent LLC cells did not change their morphology or proliferation rate between conditions, non-adherent LLC-CSCs showed some cell death signs, as soon as 48 h after adding TAT-Cx43266-283, and their growth is reduced compared with control conditions. Viability assays (Figure 1C and D) were consistent with the photomicrographs (Figure 1A and B) and showed that cell growth curves in adherent LLC cells displayed no significant changes among conditions (Figure 1C), whereas LLC-CSCs cells exhibited a decrease in cell viability when cultured in the presence of TAT-Cx43266-283, starting at 48 h and being statistically significant at 144 h (Figure 1D). In contrast to the inhibition of invasion promoted by TAT-Cx43266-283 on GSCs,24 our results showed that TAT-Cx43266-283 did not significantly affect the invasive capacity of LLC-CSCs cultured in the presence of TAT-Cx43266-283 for 15 h using Matrigel invasion assays (Figure S2).
Because TAT-Cx43266-283 effects were most notable in the non-adherent LLC-CSCs subpopulation as shown in GBM models, next experiments were performed in non-adherent LLC-CSCs (hereafter termed LLC-CSCs). To address the molecular mechanism involved in the effects of TAT-Cx43266-283 in LLC-CSCs in vitro, we performed a phosphoproteomic analysis. To do so, LC-MS/MS techniques together with computational analysis were used to analyze these samples. Comparing the phosphopeptides identified and quantified in LLC-CSCs cultured in control conditions with those cultured in the presence of TAT-Cx43266-283, we found that 233 phosphopeptides were increased and 66 phosphopeptides were decreased after TAT-Cx43266-283 treatment (Figure 1E and Table S1). These phosphosites were manually interrogated using the PhosphoSitePlus® database and were found to be present in proteins related to a great variety of cellular processes, such as protein homeostasis (Psmd2, Hspa12b), intracellular trafficking (Dync1li1), cell cycle and mitosis (Ccne2, Epb41l2, Fmnl1, Rgcc), cell migration and adhesion (Anks1a, Arhgef2), apoptosis (Ciapin1), or cytoskeleton organization (Iqsec2, Vasp). In order to narrow the search, we carried out a Kinase-Substrate Enrichment Analysis (KSEA),39 which assesses statistically significant enrichment (Z-score) of phosphorylation events for specific kinase substrates in treated vs control cells, providing insights into the potential kinase activities driving these phosphoproteomic changes. Figure 1F and Table S2 show the KSEA results of 16 kinases that play important roles in signaling pathways controlling survival, proliferation, migration, adhesion, and other processes related to cancer. As shown, the activities of two of these kinases, Mapk1 (Extracellular signal-regulated kinase 2, ERK2) and Map2k1 (Mitogen-activated protein kinase kinase, MEK1), were significantly decreased by TAT-Cx43266-283 treatment in LLC-CSCs. It should be mentioned that the activities of Src (SRC Proto-Oncogene Non-Receptor Tyrosine Kinase, Src), mTor (Mammalian Target of Rapamycin Kinase, mTOR), Mapk14 (Mitogen-Activated Protein Kinase 14, p38), Csnk2a1 (Casein kinase II, Cka2), and Cdk1 (Cyclin Dependent Kinase 1, Cdk1) were also decreased by TAT-Cx43266-283, although the extent of inhibition did not reach statistical significance. As this is the first evidence of the involvement of the ERK-MEK pathway in TAT-Cx43266-283 effect, we performed a WB analysis of the expression of ERK and its level of activity (p-ERK) in control and treated with TAT-Cx43266-283 LLC-CSCs. Figure 1G and Figure S3A showed the reduction in ERK phosphorylation on the residues affected by MEK (Thr202 and Tyr204), which are indicative of its activity. The quantification of these blots confirmed the significant decrease in ERK activity promoted by TAT-Cx43266-283 (Figure 1H), suggesting the involvement of MEK-ERK pathway in the effects of TAT-Cx43266-283 in LLC-CSCs in vitro. To confirm this suggestion, the effect of the MEK/ERK inhibitor selumetinib, alone or in combination with TAT-Cx43266-283 was analyzed (Figure 1I and Figure S3B). As expected,46 MEK/ERK inhibition reduced NSCLC cell proliferation, but we did not find a significant additive or synergistic effect in combination with TAT-Cx43266-283, supporting the participation of MEK/ERK inhibition in the antitumor effect of the Src inhibitor TAT-Cx43266-283 in LLC-CSCs.

Effect of TAT-Cx43266-283 on the Development of LLC-CSC-Derived Brain Tumors
Once we found that TAT-Cx43266-283 affects LLC-CSCs in vitro, we decided to explore its effect in vivo, specifically in brain tumors derived from LLC-CSCs, as a model of lung cancer brain metastasis. To do so, C57BL/6 mice were intracranially injected with LLC-CSCs in saline with or without TAT-Cx43266-283. Fifteen days post-implantation the brains were sectioned and analyzed (Figure 2A). Out of six animals per group, 5 control, and 3 TAT-Cx43266-283-treated mice developed visible tumors at this stage. The H&E and GFAP staining of the brains without visible macroscopic tumor confirmed the absence of alterations in brain histology caused by TAT-Cx43266-283 treatment (Figure S4A), as well as the lack of significant tumors; only a small group of tumor cells appeared in a TAT-Cx43266-283-treated mice (Figure S4B). Therefore, only those brains with visible tumors were used for subsequent analysis (Figure 2B). Brain slices were dissected to separate tumor tissue including the microenvironment from the contralateral region, as illustrated in Figure 2C and phosphoproteomic analysis was performed following the guidelines previously described.34,36  Figure 2C shows the phosphopeptides significantly decreased by TAT-Cx43266-283 (161; in blue) and those increased (836; in orange). The top 15 phosphopeptides decreased and increased were labelled by their names. In Table S3, all differences between the detected phosphopeptides are displayed.
As these phosphopeptides are related to different pathways, we carried out a PEA to identify the main cellular and molecular pathways affected by TAT-Cx43266-283 treatment in brain tumors derived from LLC-CSCs after 15 days of development. Figure 2D shows the top 15 pathways that were significantly up- or down-regulated by TAT-Cx43266-283 and Table S4 includes all the pathways increased or decreased after TAT-Cx43266-283 treatment. We found that several pathways related to the cytoskeleton—such as cytoskeletal regulatory protein binding, astrocyte cell migration, and regulation of microtubule depolymerization—were affected by TAT-Cx43266-283  in vivo, suggesting that this treatment affects processes regulated by the cytoskeleton. To test this idea, we analyzed the expression of the microtubule-associated protein 2 (MAP2), a cytoskeleton protein found to be differentially phosphorylated by TAT-Cx43266-283 treatment in the phosphoproteomic analysis (Table S4), together with other cytoskeleton-related proteins that are also relevant in cancer progression and metastasis. This was carried out by immunofluorescence and WB using brain sections from the same mice analyzed by phosphoproteomic analysis. Among them, MAP2 (Figure S4E) and the class III β-tubulin (TuJ1) (Figure S4D) did not show evident variations between groups, suggesting that TAT-Cx43266-283, while affecting MAP2 phosphorylation, did not significantly affect their level of expression. In contrast, the expression of α-tubulin appeared to be increased in TAT-Cx43266-283-treated samples as evidenced by immunofluorescence (Figure 3A and Figure S4C) and WB analyses (Figure 3B and C and Figure S5). However, β-actin decreased its expression with the treatment. In addition, other cytoskeleton proteins modified by TAT-Cx43266-283 in GBM models, such as p120 and β-catenin,23,47 did not exhibit significant changes in brain tumors derived from LLC-CSCs (Figure 3B and C and Figure S5).
As our in vitro results unveiled that TAT-Cx43266-283 reduced ERK activity, whose signaling pathway is tightly linked to Src, the canonical target of TAT-Cx43266-283, we explored their activities by WB in the brain tumor samples from the mice used for phosphoproteomic analysis. As evidenced in Figure 3B and C and Figure S5, both ERK and Src activities were significantly reduced by TAT-Cx43266-283 in LLC-CSC-derived tumors after 15 days of development (Figure 3C).

Effect of TAT-Cx43266-283 on the Survival of Mice Bearing LLC-CSC-Derived Brain Tumors
To further explore the effect of TAT-Cx43266-283 on metastatic lung brain tumors, we analyzed the effect of this treatment on the survival of mice bearing lung cancer brain metastasis, following the procedure used in previous studies.25,27,28 Thus, we performed the procedure to generate LLC-CSC-derived brain tumors described in previous sections and extended the study until humane endpoint was reached instead of 15 days (Figure 4A).
Our results showed that the median survival time was significantly higher in TAT-Cx43266-283-treated mice (23 days) compared to control mice (18 days) (Figure 4B). Although, brains and their H&E staining did not show noticeable differences between both experimental conditions (Figure 4C and D and Figure S6A and B), three main tumor features were found: proliferative foci present in the tumor border near vascular structures (left column images), highly vascularized tumors with no necrosis detected (middle column images) and cancer cells with pyknotic nuclei (right column images), ie, highly stained nuclei due to chromatin condensation. Because these pyknotic cells are associated with apoptotic processes, and chromatin condensation is one of the main events that take place during this process,48 we analyzed the levels of cleaved Caspase-3, together with the levels of Ki-67 staining, to derive indexes of cell apoptosis and proliferation, respectively. As shown in Figure 4E and F and Figure S7, there were no significant differences in Ki-67 levels between conditions, while TAT-Cx43266-283 significantly increased cleaved Caspase-3 in the tumor, indicating a higher degree of apoptosis in these treated tumors compared to the control groups.
In order to explore if the effect of TAT-Cx43266-283 on the proteins found in LLC-CSC-derived brain tumors after 15 days of development persisted until humane endpoint, WB analysis was carried out. Figure 4G and H and Figure S8, showed that changes in β-actin expression were no longer significant at humane endpoint. However, p120 exhibited a decrease in its expression in tumors from TAT-Cx43266-283-treated mice. The changes in β-catenin and α-tubulin expressions were similar to those observed in 15-day-treated mice, with β-catenin displaying no differences between groups and α-tubulin increasing its expression in the treated group. We also found that the decrease in ERK activity was slightly but not significantly reduced, while the inhibition of Src activity persisted in TAT-Cx43266-283-treated animals at humane endpoint (Figure 4H).

Effect of TAT-Cx43266-283 on Vascularization in LLC-CSC-Derived Brain Tumors
The PEA analysis performed on the tumor samples (Figure 2D and Table S4) showed that the regulation of blood vessel endothelial cell migration was downregulated in TAT-Cx43266-283-treated mice. Indeed, the vascularization appeared to be reduced by TAT-Cx43266-283 treatment in LLC-CSC-derived tumors at 15 days after tumor implantation (Figure 2B).
As TAT-Cx43266-283 normalized vascularization and reduced VEGF expression in GBM models,28 we studied the role of TAT-Cx43266-283 on LLC-CSC-derived brain tumors vascularization. To do this, brain sections from mice bearing lung cancer brain metastasis at endpoint treated and not treated with TAT-Cx43266-283, were used to analyze the expression of the platelet and endothelial cell adhesion molecule 1 (PECAM-1), also known as CD31, a marker of the blood vessel endothelium. As expected, the tumor tissue was highly vascularized compared to the rest of the tissue (Figure 5A and Figure S9A). Focusing on the tumor center (magnified images), CD31 staining suggested a reduction and a change in the tumor vascularization pattern in TAT-Cx43266-283-treated tumors. Thus, the quantification of these images indicated that TAT-Cx43266-283 reduced the area occupied by CD31 in these tumors, suggesting a reduction in the vascularization of these tumors (Figure 5B). Because of the key role of the vascular endothelial growth factor (VEGF) in angiogenesis to support brain metastatic tumors derived from lung cancer,1 we analyzed the effect of TAT-Cx43266-283 on VEGF in LLC-CSC-derived brain tumors by WB. Figure 5C and D and Figure S9B showed that VEGF expression was significantly reduced in LLC-CSC-derived brain tumors from mice treated with TAT-Cx43266-283 when compared with controls.

Clinical Relevance of the TAT-Cx43266-283 Targets in KRAS-Mutated NSCLC
To address the clinical relevance of our results, we analyzed the dataset of human NSCLC adenocarcinoma samples provided by The Cancer Genome Atlas (TCGA) Firehose Legacy, which includes genomic, transcriptomic, and proteomic information.40 As KRAS mutations are the most frequent oncogenic driver in NSCLC,49 we used cBioportal41 to compare protein levels among KRAS altered (104 samples) and unaltered (414 samples) NSCLC adenocarcinoma groups. Interestingly, we found that the TAT-Cx43266-283 targets revealed in this study using murine in vivo and in vitro models, such as the active forms of MEK1 and ERK2 and ERK1, as well as CD31, are among the top five proteins whose abundance is significantly increased in KRAS-altered NSCLC samples compared with unaltered ones (Figure 6A and B and Table S5). The activity of other important proteins in this study, such as mTOR or Src is also increased, although not significantly, in KRAS-altered NSCLC samples (Table S5). Altogether, these data suggest the clinical interest of TAT-Cx43266-283 for future studies in patients with KRAS-mutated NSCLC. As a proof of concept, we tested the effect of TAT-Cx43266-283 in two human (A549 and H441) and one additional mouse (LSZ4) lung cancer cell lines, all of which are KRAS-mutated NSCLC cell lines.50 To do so, the non-adherent CSC subpopulations were selected, as previously described, and the effect of TAT-Cx43266-283 on cell survival was analyzed. Our results confirmed that TAT-Cx43266-283 did not have significant effects on the adherent populations (Figure 6C and E), while reduced specifically the growth of the non-adherent A549-CSC, H441-CSC, and LSZ4-CSC subpopulations (Figure 6D and F). In addition, we found a significant decrease in the A549-CSC invasion capacity (Figure 6G and Figure S10), as well as a reduction in LSZ4 cell random movement, which was statistically significant when these cells were cultured in stem cell medium in the presence of TAT-Cx43266-283 (Figure 6H and Supplementary Movies S1-S4).

Discussion
TAT-Cx43266-283 is a new Src inhibitor with notable antitumor effects in preclinical models of GBM.20,22,24–28 In this study, we show that TAT-Cx43266-283 can also impact lung cancer brain metastasis derived from KRAS-mutated NSCLC cells in vitro and in vivo. Some of the effects exerted by TAT-Cx43266-283 in GBM, such as Src inhibition, reduced tumor cell viability and migration, cytoskeletal disruption, decreased aberrant vascularization, and the improved survival of mice bearing brain tumors, were also observed in lung cancer brain metastasis. Surprisingly, our unbiased phosphoproteomic analysis unveiled that TAT-Cx43266-283 inhibited the MEK/ERK signaling pathway in lung cancer brain metastasis models. Interestingly, these targets of TAT-Cx43266-283 are elevated in NSCLC patients with KRAS alterations compared with unaltered ones, supporting the clinical interest of TAT-Cx43266-283 in lung cancer brain metastasis derived from KRAS-mutated NSCLC, which are very frequent and difficult to treat brain tumors.6
As tumors that develop in the brain, GBMs and brain metastases share several features, such as the relationship of tumor cells with the brain tumor microenvironment and the relevance of the subpopulation of CSCs, which are highly tumorigenic, infiltrate the brain parenchyma and are resistant to chemotherapy and radiotherapy.16 In this study, we found that the viability of the non-adherent stem cell-like subpopulation of murine LLC and LSZ4, as well as human A549 and H441 NSCLC cells was specifically reduced by TAT-Cx43266-283, as previously observed in different GBM models, including primary GBM patient derived stem cells.24,27,28 These results support the specificity of TAT-Cx43266-283 for the non-adherent CSC-like subpopulation, which is highly dependent on Src activity.16
TAT-Cx43266-283 reduces GSC migration and invasion in GBM models in vitro and in vivo.24,27,28 In this study we found a significant reduction in A549-CSC invasion, as well as in the random movement of adherent LSZ4 cells when cultured in stem cell medium containing B27, EGF and bFGF. Intriguingly, we did not observe significant effects of TAT-Cx43266-283 on LLC-CSC invasion in vitro, although, the PEA analysis revealed that several pathways involved in cell migration related to the cytoskeleton function were altered in TAT-Cx43266-283-treated brain tumors derived from LLC-CSCs in the in vivo models. Given that Src is a node of numerous signals involved in these processes, these data suggest that the effects of TAT-Cx43266-283 on NSCLC cell migration and invasion are likely context- and cell-specific and may further depend on the methodological approach used to evaluate them. Thus, WB confirmed that TAT-Cx43266-283 disrupted LLC-CSC cytoskeleton, by increasing the levels of α-tubulin and inducing a transient reduction in β-actin, which may affect cellular protrusions that promote cell migration.51 It should be mentioned that actin can also form tunneling nanotubes (TNTs), which are actin-rich and tubulin-free thin membrane tubes described in metastasis,52 including brain metastasis,53 which correlate with more aggressive phenotype.54 In line with this, the expression of p120, a cytoskeleton protein highly relevant in cancer cell invasiveness,55 was reduced in TAT-Cx43266-283-treated mice bearing lung cancer brain metastasis, as we had previously found in GBM models.23 Collectively, these data suggest that changes in the cytoskeleton organization can contribute to the increase in the survival of mice bearing lung cancer brain metastases after TAT-Cx43266-283 treatment.
Similarly, decreased vascularization and VEGF expression, found after TAT-Cx43266-283 treatment in GBM models28 and in brain metastasis found in this study, can be part of the antitumor effects promoted by TAT-Cx43266-283. Mechanistically, we propose that decreased vascularization might be caused by VEGF reduction, which in turn may be due to the inhibition of Src activity56 promoted by TAT-Cx43266-283. The reduction of VEGF-mediated angiogenesis with agents such as bevacizumab represents a well-established approach to restrict the growth of numerous types of tumors, including recurrent brain metastasis from lung cancer.57 In fact, it has been proposed that bevacizumab may benefit NSCLC patients with synchronous brain metastases more than it does in patients without intracranial disease, possibly through inhibiting angiogenesis and minimizing vasogenic oedema.58
As previously mentioned, phosphoproteomic analysis unveiled the involvement of MEK/ERK signaling pathway in the antitumor effect of TAT-Cx43266-283 in LLC-CSCs and in brain metastasis derived from these cells. TAT-Cx43266-283 significantly reduced ERK2 and MEK1 activities in both in vitro and in vivo models of lung cancer brain metastasis, which was confirmed by WB of ERK phosphorylation at sites catalyzed by MEK, which coincided with Src inhibition, suggesting a crosstalk between these signaling pathways. Indeed, Src can phosphorylate and activate Raf-1 on tyrosine residues 340 and 341 in a Ras-dependent manner, and this is required for full activation of the downstream MEK/ERK signaling.59 The activation of MAPK/ERK pathway promoted by Src increases survival and proliferation in several tumor cells,60 including NSCLC.61 In addition, ERK can mediate cell adhesion and migration when activated by Src62 and the aberrant tumor vascularization and VEGF levels are also Src- and ERK-dependent cellular processes.63 Although we cannot exclude a direct interaction between TAT-Cx43266-283 and MEK/ERK, the lack of additive effects between TAT-Cx43266-283 and the MEK/ERK inhibitor selumetinib, together with the existing literature, suggest that Src inhibition promoted by TAT-Cx43266-283 can reduce MEK/ERK activity, thereby causing the antitumor effects of TAT-Cx43266-283 found in lung cancer models. Importantly, we found that the most elevated proteins in NSCLC patients with KRAS alterations compared with unaltered ones are the targets of TAT-Cx43266-283 described in this study, ie, MEK1, ERK2, and CD31, supporting the clinical interest of TAT-Cx43266-283 for future studies in patients with KRAS-mutated NSCLC. These results are not unexpected as it is well described that ERK activity supports KRAS-dependent cancer growth,64 but they are very relevant, as KRAS mutations are the most frequent oncogenic driver in NSCLC49 and might serve as a biomarker of TAT-Cx43266-283 response in NSCLC.
The success of targeted therapies in certain lung cancer brain metastases supports the development of new candidates within this therapeutic approach.6 In line with this, direct targeting of the KRAS-G12C-mutant protein using covalent inhibitors is a promising therapeutic strategy in KRAS-mutated NSCLC. However, drug resistance is an emerging concern in this approach. Interestingly, it has been recently shown that the inhibition of SRC with dasatinib avoids the resistance to KRAS-G12C inhibitors.65 Our previous studies showed that the inhibition of Src with TAT-Cx43266-283 exhibits reduced toxicity in healthy neurons and astrocytes compared with dasatinib,24 which together with this study support further exploration on the combination of TAT-Cx43266-283 with new KRAS inhibitors, particularly in the context of brain metastasis. In addition, KSEA of LLC-CSCs informed us about several kinases involved in cancer that were upregulated, although not significantly, in LLC-CSCs treated with TAT-Cx43266-283. The combination of these kinase inhibitors with TAT-Cx43266-283 treatment might enhance their antitumor effect. Furthermore, future studies could be focused on immunotherapy combination because of the alterations found in the AP-1 complex of TAT-Cx43266-283-treated mice tumor tissue, whose function is tightly related to the immune response, especially when mediated by the MAPK/ERK pathway.66
One of the limitations of our in vivo study is that tumor cells were intracranially implanted in the brain, and therefore, the effects on all the metastatic steps prior to the formation of brain tumors were not explored. However, our findings provide a proof of concept for the potential use of TAT-Cx43266-283 in treating established KRAS-mutant NSCLC-derived brain metastases, either as a monotherapy or in combination with other treatments. Future research using patient-derived tumor fragments to confirm KRAS mutation as biomarker of TAT-Cx43266-283 response in NSCLC, as well as to evaluate the efficacy of its combination with emerging therapies, would help moving TAT-Cx43266-283 towards its clinical application. In addition, these results open new avenues for the study of TAT-Cx43266-283 in other tumor types driven by these molecular alterations.

Supplementary Material
vdag001_Supplementary_Data