Targeting AXL inhibits the growth and metastasis of prostate cancer in bone

Abstract

Purpose:

After failing primary and secondary hormonal therapy, castration-resistant and neuroendocrine prostate cancer metastatic to the bone is invariably lethal, although treatment with docetaxel and carboplatin can modestly improve survival. Therefore, agents targeting biologically relevant pathways in PCa and potentially synergizing with docetaxel and carboplatin in inhibiting bone metastasis growth are urgently needed.

Experimental Design:

Phosphorylated (activated) AXL expression in human prostate cancer bone metastases was assessed by immunohistochemical staining. We evaluated the effects of a novel soluble AXL signaling inhibitor, sAXL (batiraxcept or AVB-S6-500), on the tumor growth and lung metastases in PCa patient-derived xenograft models (PDX) that implanted intratibally. After injection of LuCaP cells into the tibiae, tumors were treated with batiraxcept and docetaxel or carboplatin alone or in combination, and tumor growth was monitored by serum PSA or bioluminescence. Tumor burden was quantified by human-specific Ku70 staining, and metastasis to the lung was determined using qPCR. Transcriptomic profiling, western blotting and immunohistochemistry were performed to identify treatment-regulated gene and protein profile changes.

Results:

High AXL phosphorylation in human PCa bone metastases correlated with shortened survival. Batiraxcept alone or in combination with docetaxel or carboplatin significantly suppressed intratibial tumor growth and suppressed metastasis to the lung through multiple mechanisms, including repression of cancer stemness genes (CD44, ALDH1A1, TACSTD2, ATXN1) and the PI3K, JAK, MAPK, and E2F1/NUSAP1 signaling pathways.

Conclusions:

Our study provides a robust preclinical rationale and mechanisms of action for using batiraxcept as a single agent or in combination with docetaxel or carboplatin to treat lethal mPCa.

Introduction

An estimated 35,250 prostate cancer (PCa) patients will succumb to the disease in the United States in 2024, and the five-year overall survival (OS) rate in metastatic prostate cancer (mPCa) is ≤ 34% [1]. In advanced PCa, bone metastasis occurs commonly [2] and produces considerable morbidity due to fractures, intolerable bone pain, and spinal cord compression [3]. Androgen deprivation therapy (ADT) has been the standard of care for high-risk non-metastatic PCa and metastatic hormone-sensitive PCa [4]. Although nearly all prostate cancers respond to ADT, nearly all patients progress to metastatic castration-resistant adenocarcinoma (mCRPC AC) within months to years [5], for which the median OS time is < 2 years [2, 6]. Treatment of mCRPC AC with second-generation anti-androgens such as abiraterone and enzalutamide frequently induces transdifferentiation of adenocarcinomas to the more aggressive and lethal variant neuroendocrine PCa (NEPC) [7]. In addition, responses to docetaxel as the first-line chemotherapy for mCRPC AC and platinum-based therapy such as carboplatin for NEPC are short-lived and resistance occurs quickly [8]. Moreover, the efficacy of FDA-approved bone-targeting therapies for mCRPC, such as bisphosphonates, denosumab, radium-223, and Lutetium-177 PSMA, is limited [9]. Therefore, effective bone metastasis-targeting therapies for mCRPC patients are urgently needed.

Dysregulation of the AXL signaling pathway can enhance tumor growth and metastasis [10]. In PCa, AXL signaling has been shown to drive tumor growth [11], progression [12], bone metastasis [13], resistance to docetaxel [14] and ADT [15]. AXL is overexpressed in primary and metastatic PCa tissues compared to non-cancerous prostate tissues [11, 16] and correlates with tumor grade [16]. Genetic inhibition of AXL in PCa cells significantly prolonged OS in a preclinical intracardiac injection model [17], supporting the importance of AXL in disease progression. Notably, a recent study revealed that AXL signaling is a key driver of the CRPC-stem cell-like (CRPC-SCL) subtype, which is less responsive to ADT [18]. In the bone marrow niche, AXL signaling promotes PCa invasion, proliferation, survival [16, 19], and establishment of PCa stem cells (CSCs) [20]. Therefore, AXL signaling is a potential therapeutic target for bone metastases in mPCa [21].

We investigated the clinical relevance of AXL protein phosphorylation, which activates AXL signaling, in bone specimens from mCRPC patients. We also evaluated the effects of batiraxcept (AVB-S6-500) [22], a novel ultra-high-affinity soluble AXL (sAXL) decoy receptor capable of sequestering growth arrest-specific 6 (GAS6), the ligand of AXL receptor, as a single agent or in combination with docetaxel or carboplatin, standard-of-care chemotherapeutic agents for mCRPC and NEPC, respectively, on bone tumor growth and lung metastases using intratibial patient-derived xenograft models. Finally, we investigated the molecular mechanisms of action of batiraxcept by identification of treatment-regulated gene and protein profile changes.

Materials and Methods

Reagents

All antibodies used in this study are listed in Supplementary Table S1.

Cell culture

LuCaP 147, 147CR, 35 (RRID:CVCL_4853), and 35CR xenografts were obtained from Department of Urology, University of Washington and spheroids were generated and cultured in StemPro hESC SFM (Invitrogen, Carlsbad, CA) supplemented with 10 nM of R1881 and 2 μM of Y-27632 as described previously [23, 24], and incubated at 37°C with 5% CO₂. The LuCaP spheroids were tested by short-tandem repeat analysis and proved to be unique and of human male origin [23], and mycoplasma testing was conducted routinely.

Generation of luciferase-expressing LuCaP 49 spheroids

The stable LuCaP 49 (RRID:CVCL_4750) luciferase-expressing cells (LuCaP 49-luc) were generated with a pLenti CMV V5-LUC Blast vector (RRID:Addgene_21474) and selected in StemPro hESC SFM containing 2 μg/ml blasticidin as described [25]. The Dual-Luciferase Reporter Assay System (Promega, Madison, WI) was used to confirm the luciferase expression.

Patient tumor specimens and immunohistochemical (IHC) staining

Human prostate cancer bone biopsy specimens were obtained from the Stand Up 2 Cancer/ Prostate Cancer Foundation-funded West Coast Prostate Cancer Dream Team project under an approved protocol overseen by the UCSF Institutional Review Board and conducted in accordance with the Belmont Report implemented through the Common Rule. All individuals provided written informed consent [7, 26].

The formalin-fixed and paraffin-embedded specimens (n=31) were used to perform IHC staining using anti-phospho-AXL as described previously [25, 27, 28] using a protocol supplied by the manufacturer (Cell Signaling, Danvers, MA). Immunostaining was carried out using the VECTASTAIN^® ABC system, according to the manufacturer's instructions (Vector Laboratories, Burlingame, CA). A four-point staining intensity scoring system was applied by two independent pathologists to quantify the relative expression of phospho-AXL in cancer components of the specimens [29, 30].

Animal studies

All procedures complied with the Stanford University Institutional Animal Care and Use Committee and NIH guidelines. LuCaP 147, 147CR, 35, 35CR, 49, or LuCaP 49-luc spheroids were cultured in ultralow attachment plates (Corning, Corning, NY) [25]. Variable responses to therapeutic agents in LuCaP PDXs have been reported [31]. Assuming the variation in tumor growth among mice carrying the same LuCaP PDX line is 20%, 9 mice per group were estimated to be necessary to detect a 35% inhibition in tumor growth with 80% power and an alpha value of 0.05.

Spheroids were dissociated enzymatically with ACCUTASE^™ (Invitrogen, Carlsbad, CA) and mechanically using a fire-polished Pasteur pipette. The cell mixture was passed through a 40-μm sieve and disaggregation into single cells was confirmed microscopically. HEPES-buffered saline (HBS) (20 μl) containing 2X10⁵ single cells was injected into the right tibiae of six- to eight-week-old male RAG2^−/−γC^−/− mice as described previously [25]. For LuCaP 35CR and 147CR lines, androgen deprivation was accomplished by surgical castration of the mice. Serum prostate-specific antigen (PSA) levels were determined in mice bearing LuCaP 147, 147CR, 35, and 35CR at day 14 and 28 using a human PSA-total ELISA kit (Sigma-Aldrich, St. Louis, MO) according to manufacturer’s instructions. Mice were randomized into four groups, and the levels of murine serum GAS6 were determined for all treatment groups for the LuCaP 147 PDXs before treatment initiation using a mouse GAS6 ELISA kit (Sigma-Aldrich, St. Louis, MO). All groups were treated for 30 days by intraperitoneal injection: phosphate-buffered saline (Vehicle), batiraxcept (20 mg/kg/every other day (QOD)), docetaxel (10 mg/kg/once a week (QW)), and batiraxcept (20 mg/kg/QOD) + docetaxel (10 mg/kg/QW). LuCaP 49-luc cell growth was monitored by whole-body bioluminescence intensity (BLI) imaging using a Lago optical imaging system (Bruker, Tucson, Arizona) on day 14 and 28. Mice were then randomized into four groups for 30 days of treatment by intraperitoneal injection: phosphate-buffered saline (Vehicle), batiraxcept (20 mg/kg/QOD), carboplatin (5 mg/kg/QW), and batiraxcept (20 mg/kg/QOD) + carboplatin (5 mg/kg/QW). MicroCT was performed using a Bruker Skyscan 1276 (Bruker, Billerica, MA) at the Stanford Center for Innovation in In-Vivo Imaging. Image analysis and 3D representations were conducted with GEHC Micro View Version MicroView Analysis (G.E. Healthcare, Wauwatosa, WI). After sacrifice, the tumor-bearing legs and organs were harvested. The contralateral hind legs of the animals were harvested as controls for histomorphometric analysis.

Mouse specimens and quantification of IHC signals

Intratibial PDX and lung lobes were fixed in 10% buffered formalin overnight [25]. Mice tibiae were fixed in 10% buffered formalin for two days and incubated in 14% EDTA (Sigma-Aldrich, Burlington, MA) to decalcify for 15 days. After embedding in paraffin, 5 μm slices were sectioned for Masson-Goldner staining and IHC using a human Ku70 antibody to evaluate the treatment response [32, 33]. IHC for lung lobes and tibiae of mice was performed as described previously [25, 27, 29, 30, 34]. The IHC signals from murine tibial sections were analyzed and quantified by ImageJ software version 1.54f as described [34]. We used one slide from a matching area of every mouse tibia in each treatment arm (8-9 mice per arm). Therefore, 32-36 slides were used for each LuCaP PDX line for Ku70 quantification.

Immunofluorescence

Mouse tibial sections were de-paraffinized and rehydrated before antigen retrieval (0.1 % trypsin for 5 min), followed by incubation with blocking serum and appropriate antibodies overnight at 4°C, followed by 1 hour incubation with secondary antibodies, Alexa 488 (Abcam Cat# ab150113, RRID:AB_2576208, Abcam Cat# ab150077, RRID:AB_2630356,) or Alexa 555 (Thermo Fisher Scientific Cat# A-21422, RRID:AB_2535844), at room temperature. Slides were mounted with DAPI Fluoromount-G (DAPI, 4,6diamidino-2-phenylindole) (Southern Biotech, Birmingham, AL, USA) and imaged using Leica Application Suite X software on a Leica CTR 6500 (Leica, Wetzlar, Germany) [35].

RNA extraction and RT-qPCR

Total RNA from mouse tissues was purified using the Qiagen RNeasy Mini Kit (Qiagen, Valencia, CA) and subject to DNase treatment using the RNase-Free DNase Set (Qiagen) or TURBO DNA-free Kit (Life Technologies). Relative gene expression was determined by RT-qPCR with the ΔCT method using human-specific GAPDH (Forward: 5’-AGATCCCTCCAAAATCAAGTG-3'; Reverse: 5’-CAAAGTTGTCATGGATGACC-3') and universal GAPDH (Forward: 5’-CCATGGAGAAGGCTGGGG-3'; Reverse: 5’-CAAAGTTGTCATGGATGACC-3') reference genes [36].

RNA-Seq and expression analysis

RNA-Seq was performed by Novogene using the Illumina NovaSeq PE150 platform, a paired-end sequencing technology with a 150-bp read length [37]. We aligned the reads to both human and mouse genomes (hg38 and mm10), imported these aligned files into R, and used the R package ‘XenofilteR’ v. 1.6 to remove all mouse reads from the dataset. The reference genome and gene model annotation files were downloaded from genome website directly. Hisat2 v2.0.5 was used to build the index of the reference genome and paired-end clean reads were aligned to the reference genome. The read numbers mapped to each gene were counted by featureCounts v1.5.0-p3. The FPKM value of each gene was determined by the length of the gene and read counts mapped to the gene. Raw data is available at Gene Expression Omnibus (GSE281461). Differential expression analysis of two conditions (with two biological replicates per condition) was conducted using the DESeq2 R package (version 1.20.0). Adjusted p-values were obtained using Benjamini and Hochberg's method to control the false discovery rate. Genes with an adjusted P-value ≤ 0.05, as determined by DESeq2, were designated as differentially expressed. Additionally, Gene Set Enrichment Analysis (GSEA) was performed using GSEA version 3.0. Adjusted p-values < 0.05 and FDR < 0.25 were considered statistically significant. Heatmaps were generated using genes from the GSEA software, with batiraxcept vs. vehicle gene expression as the phenotype label.

Protein extraction and western blots

Protein extraction and western blots were performed as described previously [34, 35]. Blot visualization was achieved using SuperSignal West Femto Maximum Sensitivity Substrate (Thermo Scientific), Amersham ECL Western Blotting Detection Reagent (G.E.), or Clarity ECL Western Blotting Substrate (Bio-Rad), followed by imaging using Image Lab software on a ChemiDoc XRS System (Bio-Rad).

Statistical analysis

Univariable and multivariable Cox proportional hazards models were utilized to assess whether p-AXL IHC staining was associated with survival time, without and with adjustments of ECOG scores, PSA, LDH, ALP, and Hemoglobin levels at the time of biopsy. P-values < 0.05 were considered statistically significant. For the Kaplan-Meier analysis, the log-rank test was performed to assess significance. These analyses were performed using SAS statistical software (version 9.34, SAS Institute, Cary, NC). Additional statistical analyses were performed using GraphPad Prism 10.2.3 (GraphPad Software, Boston, MA). A one-way ANOVA test followed by Tukey's test was used to correct the multiple comparisons. An unpaired t-test (parametric) or two-tailed Mann–Whitney test (nonparametric) was used to pairwise comparison. All error bars represent the mean ± SEM. Notably, p-values are indicated as *, p < 0.05; **, p < 0.01; ***, p < 0.001;NS, p > 0.05 not significant.

Data availability statement

RNA-seq data from LuCaP PDX xenografts is available at Gene Expression Omnibus (GSE281461). Other raw data in this study are available upon request to the corresponding author.

Results

High AXL phosphorylation correlated with poor clinical outcome in mCRPC patients

To determine the correlation of phosphorylated AXL (p-AXL) protein level with clinical outcomes in human bone metastases, we evaluated the p-AXL expression in 31 mCRPC bone biopsies by IHC using a four-point IHC scoring system (Supplementary Table S2) [29]. Twelve patients showed low p-AXL expression with staining observed in 0–1% of the cells (staining intensity score = 0) (Fig. 1A) or in less than 10% of the cells (staining intensity score =+1) (Fig. 1B), while 19 patients showed high p-AXL expression with staining present in 10%-50% of the cells (staining intensity score = +2) (Fig. 1C), or more than 50% of the cells (staining intensity score = +3) (Fig. 1D). Kaplan-Meier analysis revealed an inverse correlation of p-AXL protein expression levels in bone metastases with OS in this patient cohort (Fig. 1E). Both univariable and multivariable logistic regression analyses (Fig. 1F) showed that high p-AXL protein expression was significantly associated with survival outcome. Overall, these results indicate that high p-AXL protein expression in bone metastases is associated with shortened survival in mCRPC patients.

Figure 1.

High p-AXL level is correlated with shorter survival in mCRPC. A-D, Representative images of p-AXL expression in mCRPC bone metastases by IHC. Scale bars = 50 μm. Negative staining (A, score=0). Weak staining (B, score=+1). Moderate staining (C, score=+2). Intense staining (D, score=+3). E, Kaplan-Meier analysis of mCRPC patients with low and high p-AXL expression in bone metastases. Log-rank (Mantel-Cox) test, p<0.0001. F, Univariable and multivariable regression analysis of p-AXL, ECOG PS, PSA, LDH, ALP, and Hemoglobin predictive of survival outcome. ECOG PS: Eastern Cooperative Oncology Group Performance Status. PSA, LDH, and Hemoglobin were detected at biopsy.

Batiraxcept as a single agent or in combination with docetaxel significantly inhibited growth and metastasis of LuCaP mPCa AC PDX in bone tissues

To evaluate the effects of AXL inhibition in the bone microenvironment, we administered sAXL receptor, batiraxcept (AVB-S6-500) that sequesters the AXL ligand GAS6 (Fig. 2A), either alone or in combination with docetaxel or carboplatin, using an intratibial injection model of PCa bone metastases (Fig. 2B). To represent the heterogeneity of mPCa, we implanted cells from five LuCaP PDXs, including four adenocarcinomas (AC), namely LuCaP 35, 147, and their derived castration-resistant (CR) lines, 35CR and 147CR, and one NEPC, LuCaP 49. Analysis of previously published RNAseq data [31] demonstrated that AXL mRNA levels were similar in LuCaP 35, 35CR, 147, and 49 (Fig. 2C), while GAS6 levels were higher in 35 and 35CR than 147 and 49 (Fig. 2D). In addition, 35/35CR and 147/147CR have been shown previously to display differences in serum PSA levels and in responsiveness to docetaxel (Fig. 2E) [31]. Therefore, LuCaP 35, 35CR, 147, 147CR, and 49 appeared to be appropriate PDX models to evaluate the treatment effects of batiraxcept, docetaxel/carboplatin, and batiraxcept + docetaxel/carboplatin combination therapy.

Figure 2.

Schematic diagram illustrating experimental design to evaluate AXL inhibition in the growth and metastasis of tumors in bone using PCa PDX. A, Mechanism of action of the sAXL decoy receptor (batiraxcept/AVB-S6-500). B, Timeline of tumor generation in bone and treatments. C and D, AXL (C) and GAS6 (D) gene expression in LuCaP 35, 35CR, 147, and 49 PDX cell lines from GSE66187. E, Characteristics of LuCaP PDX used in this study [31]. CR, castration-resistant; AC, adenocarcinoma; NE, neuroendocrine; LN, lymph node. PSA in serum: +, 5-100 ng/ml; +/−, 0.1-4.9 ng/ml, −, not detected. Response to Castration and Docetaxel in mouse subcutaneous xenograft model : +++, tumor volume decreases significantly for a prolong period; ++, tumors progress but slower than control tumors; +, negligible response; −, no response; a, treatment increased body weight loss. Figures 2A and 2B were created in BioRender. Chiu, C. (2025) https://BioRender.com/w55j627 and r82y980.

A previous study [38] had demonstrated that murine GAS6 can regulate the proliferation of human PC3 cells. We therefore measured levels of mouse serum GAS6 in all arms of the LuCAP147 mice before treatment by ELISA and found no significant differences in murine serum GAS6 levels between groups (Supplementary Fig. S1). The growth of tumors in bone was evaluated by IHC staining for human-specific Ku70 antigen (Fig. 3A and B; Supplementary Fig. S2A and S2B) as described previously [25, 32, 33]. Both batiraxcept and docetaxel as single agents significantly reduced the growth of all four LuCaP mPCa AC PDX (Fig. 3C and D; Supplementary Fig. S2C and S2D). In addition, batiraxcept outperformed docetaxel in the two mCRPC AC PDX, 147CR and 35CR, but not in their hormone-sensitive counterparts, 147 and 35, suggesting batiraxcept was more effective than docetaxel in suppressing tumor growth of mCRPC AC in bone. Moreover, the combination of batiraxcept and docetaxel led to the greatest suppression (75%-95%) of tumor growth in the bone of all four mPCa AC PDX. Interestingly, 147CR and 35CR (Fig. 3A and B) were more resistant to docetaxel than 147 and 35 (Supplementary Fig. S2A and S2B) in the intratibial xenograft models, suggesting that progression to androgen independence promotes resistance to docetaxel in mPCa AC in the bone microenvironment. In contrast, 147CR and 35CR were more sensitive to batiraxcept than 147 and 35, indicating a greater dependency of androgen-independent mPCa AC on AXL signaling compared to their parental lines. Therefore, batiraxcept as a single agent appeared to be more effective than docetaxel, particularly for androgen-independent bone metastases. Furthermore, batiraxcept further inhibited tumor growth in bone when combined with docetaxel regardless of androgen dependency. Batiraxcept and docetaxel as single agents or in combination reduced mouse serum PSA (Supplementary Fig S3) and increased bone mineral density (BMD) of the tibiae to a similar extent as determined by microCT (Supplementary Fig S4). None of the treatment groups displayed significant differences in body weight, indicating overall low toxicity of batiraxcept, docetaxel and combination therapy at the doses administered (Supplementary Fig S5).

Figure 3.

Batiraxcept and docetaxel alone or in combination significantly inhibited bone tumor growth and metastasis in LuCaP mCRPCa AC PDX models. A-B, Representative images of Masson-Goldner staining and Ku70 IHC of PDX cells in the tibiae of LuCaP 147CR (A), and LuCaP 35CR (B). Human CRPCa cells marked by Ku70 were significantly decreased after 30 days of treatment by intraperitoneal injection with batiraxcept (20 mg/kg/QOD, n=8-9, docetaxel (10 mg/kg/QW, n=9) or the combination of batiraxcept (20 mg/kg/QOD, n=9) and docetaxel (10 mg/kg/QW, n=9) compared to vehicle controls (n=8-9). C-D, Quantification of the IHC Ku70-positive areas in mouse tibia specimens treated with vehicle, batiraxcept, docetaxel, and batiraxcept + docetaxel in A-B. Fold change: combination batiraxcept and docetaxel (B+D) group set as 1. E, Inhibition of LuCaP 35CR tumor cells metastasis to mouse lung lobes by batiraxcept, docetaxel, and batiraxcept + docetaxel combination therapy. The arrowheads indicate Ku70-positive metastatic tumor nodules in the lung, which metastasized from the mice tibiae.F, Quantification of fold changes (docetaxel group set as 1) of human-specific GAPDH level normalized against universal GAPDH level in different treatment groups for LuCaP 35CR. *, p < 0.05; **, p < 0.01; ***, p < 0.001; NS, p > 0.05 not significant. All scale bar = 50 μm.

In lung tissues from vehicle-treated mice carrying LuCaP 147CR and 35CR PDX intratibial tumors, we observed large Ku70- and AR-positive tumor nodules while small clusters of tumor cells were found in docetaxel-treated mice (Fig. 3E; Supplementary Fig S2E). In contrast, Ku70-positive cells were rarely present in lung tissues from batiraxcept-treated mice, suggesting AXL inhibition significantly decreased metastasis from bone to the lung.These tumor nodules/clusters were negative for neuroendocrine marker SYP, indicating that they were AC and not NEPC (Fig. 3E; Supplementary Fig S2E). To quantify the amount of PCa cells in the lung, we performed qRT-PCR using human-specific and universal GAPDH primers in LuCaP 147CR and 35CR PDX lung tissues. Consistent with IHC staining, the percentages of tumor cells were extremely low in lungs of batiraxcept-treated mice compared to vehicle-treated mice, while approximately one-fourth of metastatic tumor cells were observed in lungs of docetaxel-treated mice compared to control (Fig. 3F; Supplementary Fig S2F), demonstrating that batiraxcept as a single agent or in combination with docetaxel can robustly inhibit CRPC metastasis from the mouse tibiae to the lung.

Batiraxcept as a single agent or in combination with carboplatin significantly inhibited NEPC intratibial tumor growth and metastasis to the lung

LuCaP 49 was derived from a patient with poorly differentiated NEPC (Fig. 2E). We generated stable luciferase-expressing cells by lentiviral transduction and implanted them intratibially. Bioluminescence was measured at two and four weeks after injection and 35 mice were randomized into vehicle (PBS, n=8), batiraxcept (20 mg/kg/QOD, n=9), carboplatin (5 mg/kg/QW, n=9), and combination therapy of batiraxcept (20 mg/kg/QOD, n=9) and carboplatin (5 mg/kg/QW, n=9) (Fig. 4A) treatment groups. Batiraxcept as a single agent outperformed carboplatin in inhibiting intratibial tumor growth as determined by bioluminescence imaging (Fig. 4B) and Ku70 IHC staining (Fig. 4C and D). Combination therapy significantly reduced tumor growth compared to carboplatin alone based on bioluminescence imaging. However, combination therapy compared to batiraxcept alone did not show statistically significant differences in suppression of tumor growth based on bioluminescence imaging (Fig. 4A and B). In contrast, analysis of tibiae by Ku70 IHC staining demonstrated that batiraxcept + carboplatin decreased tumor growth to a greater degree than batiraxcept alone and carboplatin alone (Fig. 4C and D). As observed in the LuCaP AC PDX models, batiraxcept outperformed carboplatin in inhibiting NEPC LuCaP 49 metastases to the lung. Specifically, No Ku70- or SYP-positive tumor cells were observed in batiraxcept-treated compared to vehicle-treated tissues, while fewer Ku70- and SYP-positive tumor cell clusters were observed in carboplatin-treated tissues compared to vehicle-treated tissues. These cell clusters were AR-negative, consistent with their NEPC derivation (Fig. 4E). qRT-PCR using human-specific GAPDH revealed that the percentages of LuCaP 49 cells in the lungs of batiraxcept monotherapy-treated mice were <5% of vehicle controls, while human LuCaP 49 cells in carboplatin-treated mice were reduced only 33% compared to controls (Fig. 4F). The combination of batiraxcept and carboplatin further decreased the spreading of LuCaP 49 metastatic cells in lung compared to single agent treatments (Fig. 4F). Therefore, batiraxcept alone or in combination with carboplatin can strongly suppress NEPC growth in the bone and suppress metastasis to the lung.

Figure 4.

Batiraxcept alone and in combination with carboplatin inhibited tumor growth and metastasis in the LuCaP 49 NEPC PDX model. Tumor-bearing mice (n=35) were injected intraperitoneally for 30 days with vehicle (PBS, n=8), batiraxcept (20 mg/kg/QOD, n=9), carboplatin (5 mg/kg/QW, n=9), or batiraxcept (20 mg/kg/QOD, n=9) + carboplatin (5 mg/kg/QW, n=9). A, Representative images of tumor bioluminescence. B, Quantification of whole-body bioluminescence on day 30 after intratibial injection of 1 × 10⁵ luciferase-labeled LuCaP 49 cells into the right tibiae of male RAG2^−/−γC^−/− immunodeficient mice (n = 8-9 per arm). C, Ku70 expression of LuCaP 49 cells in mouse tibiae. D, Quantification of intratibial LuCaP 49 cells in control and treatment groups. E, Representative lung metastases (arrows) stained with anti-Ku70, SYP, and AR antibodies. F, Quantification of human-specific GAPDH fold-changes (carboplatin group set as 1) by RT-qPCR in mouse lungs from each of the treatment groups *, p < 0.05; **, p < 0.01; ***, p < 0.001; NS, p > 0.05 not significant. All scale bars = 50 μm.

Batiraxcept suppresses cancer stemness in CRPC

To understand the mechanisms by which AXL inhibition suppressed mPCa AC in the bone microenvironment, we profiled the transcriptomes of LuCaP 147 and 147CR intratibial tumors with or without batiraxcept treatment by RNA-Seq. LuCaP 147CR tumors showed high mRNA levels of GAS6 and AXL as well as known downstream targets of the AXL signaling pathway, including AKT (PIK3CB, PIK3AP1, PIK3R5), JAK (JAK1), and MAPK (MAP3K13). Interestingly, transcript levels of genes associated with cancer stemness (CD44, TACSTD2, ATXN1, ALDH1A1) were higher in 147CR tumors compared to 147 tumors (Fig. 5A), consistent with previous findings that stemness plays a role in castration-resistance [18]. Single-agent batiraxcept treatment significantly reduced the expression of GAS6, PIK3R5, JAK1, CD44, TACSTD2, and ATXN1 in LuCaP 147CR tumors compared to vehicle-treated tumors, suggesting that AXL inhibition suppressed stemness in these tumors. In contrast, no significant differences in mRNA expression of GAS6, AXL, PIK3CB, PIK3AP1, PIK3R5, JAK1, MAP3K13, CD44, TACSTD2 or ATXN1 were observed in docetaxel-treated tumors compared to vehicle-treated controls (Fig. 5B). Consistently, TACSTD2 and ATXN1 protein levels were significantly reduced in batiraxcept treated tumors as a single agent or in combination with docetaxel compared to controls as determined by IHC (Fig. 5C), while no changes in TACSTD2 protein levels were observed in docetaxel-treated tumors compared to controls (Fig. 5D). In LuCaP 35 tumors, RNA-Seq revealed that batiraxcept as a single agent induced expression of necroptosis-associated genes such as HIST2H2AA3, CHMP5, IFNB1, STAT1 and ZBP1 (Fig. 5E), and Gene Set Enrichment Analysis (GSEA) confirmed enrichment of the necroptosis pathway in batiraxcept-treated tumors compared to controls (Fig. 5F). Finally, combination therapy of batiraxcept and docetaxel decreased the expression of stemness markers CD44 and ALDH1A1 as evaluated by IHC and immunofluence staining (Fig. 5G-H), which was consistent the observation of lower ALDH1A1 mRNA levels by RNA-Seq in tumors treated with combination therapy compared to controls (Fig. 5I). These results suggest that AXL inhibition suppresses tumor growth in the bone microenvironment through multiple mechanisms, including decreasing cancer stemness and induction of necroptosis.

Figure 5.

The mechanisms of action of batiraxcept in CRPC cells. A, LuCaP 147CR PDX intratibial tumors expressed higher mRNA of AXL (GAS6, AXL), AKT (PIK3CB, PIK3AP1, PIK3R5), JAK (JAK1), MAPK (MAP3K13) signaling and cancer stemness-associated genes (CD44, TACSTD2, ATXN1, ALDH1A1) compared to parental LuCaP 147 tumors. B, Batiraxcept treatment significantly inhibited mRNA expression of GAS6, PIK3R5, CD44, TACSTD2, and ATXN1 genes in LuCaP 147CR PDX intratibial tumors compared to the vehicle-treated controls. C and D, Representative images (C) and quantification of IHC staining of TACSTD2 (D) and ATXN1 antibodies on intratibial LuCaP 147CR PDX tumors in vehicle, batiraxcept, docetaxel, and combination therapy groups. E, Heatmap shows the relative expression of necroptosis (HSA04217) genes between batiraxcept monotherapy and vehicle controls. F, GSEA demonstrating that batiraxcept treatment correlated with necroptosis in intratibial LuCaP 35 PDX tumors. G, Combination therapy of batiraxcept and docetaxel inhibited the expression of ALDH1A1 and CD44 proteins in LuCaP 35 PDX bone tumors. H, Representative immunofluorescence images of Ku70 and ALDH1A1 showed fewer ALDH1A1-positive stem-like cells with combination therapy compared to controls. I, Intratibial LuCaP 35 tumors treated with combination therapy show significantly downregulation of ALDH1A1 mRNA. All scale bar = 10 um.

Identification of pathways affected by batiraxcept treatment of PCa cells in the bone microenvironment

To characterize the pathways targeted by the single agent batiraxcept and combination therapy with docetaxel, we performed IHC staining and western blotting to assess the AXL signaling pathway in intratibial LuCaP 147CR and LuCaP 35 tumors (Fig. 6A and B; Supplementary Fig S6) compared to controls. Total AXL protein levels were similar in both groups. However, IHC staining for phospho-AXL, indicating activation of AXL signaling, was significantly lower in batiraxcept-treated tumors. Likewise, AKT, immediately downstream of AXL [39-41], showed no change in total protein levels, but phospho-AKT was significantly decreased in batiraxcept-treated tumors compared to controls (Fig. 6A; Supplementary Fig S6). Since AXL can also modulate MAPK signaling [39], we evaluated ERK and found that total protein levels of ERK did not change (or increased slightly on western blot), while phospho-ERK decreased significantly after AXL inhibition alone or in combination with docetaxel (Fig. 6A and B; Supplementary Fig S6). Interestingly, treatment with a single agent of batiraxcept or combination therapy of batiraxcept significantly decreased protein levels of transcription factor E2F1 and its downstream target NUSAP1 [42] (Fig. 6A and B). Therefore, inhibition of AXL signaling with batiraxcept or batiraxcept combined with docetaxel acts directly to suppress several downstream signaling pathways in the bone microenvironment (Fig. 6C). Based on these findings, combination therapy of batiraxcept and docetaxel or carboplatin could provide a novel therapeutic strategy to treat prostate cancer metastatic to the bone that has failed hormonal therapy (Fig. 6D).

Figure 6.

The molecular mechanisms of batiraxcept in prostate cancer cells in the bone microenvironment. A, IHC staining demonstrates batiraxcept alone or in combination with docetaxel reduced p-AXL, E2F1, NUSAP1, and p-AKT protein levels compared to vehicle-treated controls in intratibial LuCaP 147CR PDX model. Scale bar = 10 μm. B, Western blots demonstrated batiraxcept treatment downregulated p-AXL, E2F1, NUSAP1, p-AKT, and p-ERK1/2 in intratibial LuCaP 35 PDX tumors. C, Summary of AXL signaling pathways of PCa cells in the bone microenvironment. D, Proposed mechanism of action for batiraxcept and docetaxel or carboplatin combination therapy to inhibit tumor growth and metastasis as well as overcome chemoresistance in PCa metastatic to the bone. Figures 6C and 6D were created in BioRender. Chiu, C. (2025) https://BioRender.com/f51l010, w94d739, and g56q608.

Discussion

In bone biopsies from patients with metastatic PCa, we demonstrated that p-AXL expression was associated with poor clinical outcomes. In LuCaP mPCa AC and NEPC PDX implanted intratibially, AXL inhibition by batiraxcept suppressed tumor growth and metastasis to the lung more effectively than docetaxel or carboplatin in androgen-independent tumors. Moreover, combination therapy with batiraxcept and docetaxel and/or carboplatin was more effective than either alone in inhibiting tumor growth in both CRPC and NEPC. Finally, batiraxcept as a single agent blocked metastasis of intratibial tumors to the lung more effectively than docetaxel or carboplatin alone. These findings provide a strong rationale for clinical testing of batiraxcept as a single agent or in combination with docetaxel or carboplatin to treat patients with PCa metastatic to the bone [16, 17].

Batiraxcept has demonstrated significant therapeutic effects in preclinical models of multiple cancers [40, 43] and is being evaluated in clinical trials in several cancer types. When combined with paclitaxel, batiraxcept elicited an overall response rate of 34.8% with two complete responses in a phase 1b trial in ovarian cancer (NCT03639246). In a phase III study, the combination did not significantly improve progression-free survival (PFS) compared to paclitaxel alone (NCT04729608). However, in a secondary analysis focusing on patients whose ovarian cancers expressed high levels of AXL (20% of total patient population), median PFS was 5.78 months in the batiraxcept + paclitaxel arm and 3.71 months in the control paclitaxel arm, HR 0.55 (CI, 0.31 to 0.98; p=0.042), and median OS was 17.8 months in the batiraxcept + paclitaxel cohort and 8.11 months in the paclitaxel cohort, HR 0.32 (CI, 0.14 to 0.73; p=0.006). Therefore in patients selected for high tumor expression of AXL, batiraxcept improves both PFS and OS in ovarian cancer patients with high AXL expression when combined with paclitaxel compared to paclitaxel alone (NCT04729608) [44]. In the mCRPC bone metastasis samples examined in our study, 61% expressed high levels of AXL protein. Thus, high tumor AXL levels could be used to identify the subset of patients with CRPC who could benefit from batiraxcept in combination with taxanes. Currently, batiraxcept is being tested in clinical trials of clear cell renal cell carcinoma (NCT04300140), pancreatic cancer (NCT04983407) [10], urothelial carcinoma (NCT04004442), uterine cancer (NCT05826015), and platinum-resistant or recurrent ovarian, fallopian tube, or primary peritoneal cancer (NCT04019288).

Proteomic and transcriptomic analysis of treated PCa PDX tumors revealed several molecular mechanisms through which batiraxcept inhibits mPCa bone tumor growth and metastasis. In LuCaP PCa PDX, batiraxcept inhibited the AKT signaling pathway in other cancer types [40, 41]. Interestingly, batiraxcept suppressed cancer stemness genes (CD44, TACSTD2, ATXN1, and ALDH1A1) at both transcript and protein levels. AXL is a direct downstream target of YAP/TAZ signaling in the CRPC-SCL (stem cell-like) subtype identified recently in organoids derived from human tissues [18]. Since the CRPC-SCL comprised a significant fraction of CRPC in human datasets, AXL inhibition could be used to target this subtype specifically. In addition, GSEA showed that batiraxcept induces necroptosis in intratibial mPCa PDX tumors, consistent with a previous report in which a necroptosis sensitivity screen in 941 human cancer cell lines identified AXL as a mediator of necroptosis because its inhibition can result in restoring sensitivity to necroptosis [45]. The mechanisms of action in batiraxcept were confirmed by IHC and western blotting showing that batiraxcept suppressed AXL, AKT, MAPK, E2F1, and NUSAP1 signaling pathways at the protein level while docetaxel treatment did not. Therefore, batiraxcept and docetaxel differentially affect these essential signaling pathways in cancer progression; Batiraxcept directly targets stem cell driver pathways in PCa, suppressing growth and metastases, while docetaxel does not.

In our PDX models, batiraxcept significantly reduced tumor growth when combined with docetaxel and carboplatin compared with either agent alone. Genetic or pharmacologic inhibition of AXL has been shown to enhance the response of various cancers to taxanes, including mesenchymal, uterine, ovarian, and breast [46, 47] and to platinum-based drugs [48, 49], likely through the inhibition of PI3K/AKT, MAPK/ERK, NF-κB, and c-ABL signaling [39, 50, 51]. In the presence of GAS6, the viability of PCa cells treated with docetaxel is significantly increased [16], and forced AXL overexpression in PCa cells is sufficient to induce resistance to docetaxel [14]. Moreover, inhibition of AXL suppressed the growth of docetaxel-resistant PCa cell line-derived subcutaneous xenografts, and these effects were significantly augmented when AXL inhibition was combined with docetaxel treatment [14]. In addition to the molecular mechanisms employed by batiraxcept to inhibit mPCa progression that we identified, the treatment effects may be explained by a recent report that GAS6 secreted by osteoblasts as well as PCa cells is present at high levels in the bone microenvironment thereby activating the AXL signaling pathway specifically in PCa bone metastases [16, 19, 20], which in turn induces dormancy of PCa cells in the bone and protects them from therapies targeting proliferation such as docetaxel.

Broadly speaking, the addition of targeted therapies to docetaxel regimens has failed to significantly improve outcomes for patients with mCRPC [52]. Although several tyrosine kinase inhibitors that block AXL have been tested in mCRPC and NEPC, including dasatinib, axitinib, tivozanib, crizotinib and cabozantinib, none of these TKIs have shown improvements in survival, either alone or in combination with docetaxel. The best studied of these, cabozantinib, is a potent inhibitor of VEGFR, MET and several tyrosine kinases including AXL [53]. Cabozantinib failed to improve OS in COMET-1, a Phase III prospective randomized trial of heavily treated men with mCRPC [54]. However, these results must be placed in context before dismissing consideration of batiraxcept as a candidate therapy. Compared to cabozantinib and other TKIs, batiraxcept inhibits AXL signaling with high specificity at clinically achievable dosing. In cell-based assays, cabozantinib inhibited phosphorylation of AXL at relatively high dosage levels (IC₅₀: 42 μmol/L) [53, 55], while batiraxcept suppresses serum GAS6 (the ligand for AXL) to levels below the limits of detection by a validated ELISA in healthy volunteers at a dose of only 1 mg/kg [56]. Therefore, batiraxcept eliminates AXL signaling with high specificity and without toxicity since AXL is not required for normal tissue function [57]. On the other hand, Cabozantinib given at doses necessary to achieve clinical responses shows significant toxicity, particularly hepatotoxicity, due to the important roles of VEGFR2 and MET in normal development and tissue repair [58, 59].

While COMET-1 did not demonstrate improvement in OS, a pooled analysis of 1147 patients with mCRPC from COMET-1 and COMET-2 trials demonstrated improved OS (HR 0.80, 95% CI 0.67–0.95; p = 0.012) for cabozantinib after adjusting for other prognostic factors [60]. In a recent phase II study comparing cabozantinib plus docetaxel/prednisone vs. docetaxel/prednisone alone, the median time to progression (TTP) and OS time favored the addition of cabozantinib (21.0 vs 6.6 months; P = 0.035 and 23.8 vs 15.6 months; P = 0.072, respectively) [61]. Analysis of secondary endpoints in COMET-1 demonstrated that cabozantinib significantly improved radiographic PFS (median 5.6 vs 2.8 mo; p < 0.001) [54]. In the COMET-2 trial, there was a trend for better OS (9 vs 7.9 mo; HR 0.71, 95% CI 0.45–1.12; p = 0.121), and significantly higher bone scan response with cabozantinib (31% vs 5.2%; p < 0.001) [62]. Moreover, in a phase II nonrandomized expansion study of pre-treated mCRPC patients with bone metastases, cabozantinib improved bone scans, pain scores, analgesic use, measurable soft tissue disease, circulating tumor cells, and bone biomarkers [63, 64]. A recent phase II trial of treatment-naive mCRPC patients with bone metastases showed that cabozantinib induced significant improvements on bone scans in one-third of patients with higher BMP–2 levels at baseline [65]. Since one-third of CRPC have been characterized as stem cell-like and driven by AXL expression [18], it is possible that these cancers account for bone scan responses seen with AXL inhibition. In ovarian and kidney cancers, high serum sAXL/GAS6 ratios are associated with improved overall response rates, suggesting that proper patient selection could be used to identify patients likely to benefit from batiraxcept [41, 66, 67]. Given the absence of side effects at the highest doses of batiraxcept administered in a Phase I clinical trial (NCT03401528) and the effectiveness of batiraxcept across diverse prostate cancer bone PDXs, clinical testing of batiraxcept could be warranted, particularly in tumors displaying high AXL pathway activity.

Over the past several years, treatment of advanced prostate cancer has shifted towards intensive therapy for metastatic hormone sensitive PCa, including dual inhibition of androgen signaling, docetaxel, and radiation therapy [68]. While these approaches have significantly improved survival, nearly all patients progress. While it is unknown whether AXL inhibition is effective in castration sensitive prostate cancer, batiraxcept demonstrated significant activity in prostate cancer models derived from heavily treated mCRPC similar to these patients. Given the limited treatment options for these patients, clinical testing of batiraxcept in these patients could be warranted.

In conclusion, batiraxcept as a single agent or in combination with docetaxel or carboplatin significantly reduces PCa growth in the bone microenvironment and inhibits metastasis through multiple mechanisms of action. Therefore, tackling lethal mPCa in the bone microenvironment by batiraxcept, particularly in combination therapy with docetaxel or carboplatin, can potentially overcome chemoresistance and target intractable cancer stem cells in the bone marrow niche. These results nominate batiraxcept as a novel candidate for AXL-targeting therapy for patients with mCRPC or NEPC that is metastatic to the bone.

Translational Relevance.

We evaluated the efficacy of soluble AXL (sAXL or batiraxcept), a decoy receptor that can potently inhibit AXL signaling, as a single agent or in combination with docetaxel or carboplatin to treat prostate cancer (PCa) bone metastases. We used intratibial injection of multiple patient-derived xenografts with different characteristics, reflecting a traditional phase II clinical trial without pre-selection for a particular tumor characteristic. Inhibition of tyrosine kinase receptor AXL was highly effective as a single agent and further reduced tumor growth when combined with docetaxel or carboplatin in suppressing PCa tumor growth in the bone and metastasis to the lung. AXL inhibition downregulated the expression of critical cancer stem cell-related genes, suppressed E2F1/NUSAP1 signaling pathways and significantly decreased tumor growth and metastasis. Our findings provide compelling preclinical data for testing batiraxcept in patients with prostate cancer with bone metastases.