Membrane-derived particles shed by PSMA-positive cells function as pro-angiogenic stimuli in tumors

Abstract

Cell membrane-derived particles (Mp) are rounded membrane-enclosed particles that are shed from tumor cells. Mp are formed from tumor membranes and are capable of tumor targeting and immunotherapeutic agents because they share membrane homology with parental cells; thus, they are under consideration as a drug delivery vehicle. Prostate-specific membrane antigen (PSMA), a transmembrane glycoprotein with enzymatic functionality, is highly expressed in Mp and extracellular vesicles (EV) from prostate cancer (PCa) with poor clinical prognosis. Although PSMA expression was previously shown in EV and Mp isolated from cell lines and from the blood of patients with high-grade PCa, no pathophysiological effects have been linked to PCa-derived Mp. Here, we compared Mp from PSMA-expressing (PSMA-Mp) and PSMA-non-expressing (WT-Mp) cells side by side in vitro and in vivo. PSMA-Mp can transfer PSMA and new phenotypic characteristics to the tumor microenvironment. The consequence of PSMA transfer to cells and increased secretion of vascular endothelial growth factor-A (VEGF-A), pro-angiogenic and pro-lymphangiogenic mediators, with increased 4E binding protein 1 (4EBP-1) phosphorylation.

Graphical Abstract

Comparative impact of PSMA-Mp and WT-Mp on tumor microenvironment: PSMA-expressing membrane particles transfer psma, enhancing angiogenesis and 4EBP-1 phosphorylation in PSMA-negative tumors (Figure made with Biorender.com software).

Introduction

The field of Extracellular Vesicles (EVs), particles, and cell membrane-derived particles (Mps) focuses on studying small membrane pockets released by cells in response to various stimuli. However, our understanding of these vesicles and particles is still evolving (1). They are subcategorized according to their origin and size (1,2). Mp have a median size average in diameter of 200 nm and can be manufactured (3) by treating cancer cells with drugs, buffers, or physical techniques like some EV subtypes (2,4). Apoptotic bodies are an EV subtype originating from membrane disintegration after injury and cell death, producing vesicles 1–5 μm in diameter. Microvesicles are produced from viable cells by outward bulging of the plasma membrane and range from 50 to 1,000 nm in diameter (5). Exosomes are single-membrane units shed by cells and loaded with intracellular contents ranging from 20 to 160 nm (1,5). Here, we compared Mp naturally shed by overexpressing Prostate Cancer (PCa) cells (PSMA-Mp) to PSMA-non-expressing (WT-Mp) cells to understand the contribution of Mp to PSMA-driven effects in creating a pro-tumoral microenvironment.

Prostate-specific membrane antigen (PSMA), a transmembrane glycoprotein, is present in most primary prostate cancer, recurrent tumors, and regional to distant lymph node (LN) metastasis. Therefore, PSMA has been used as a diagnostic biomarker (4). Thus far, researchers have explored vesicles derived from cells as a tool in PCa diagnosis. PSMA-positive exosomes are described in plasma of patients with increasing Gleason scores. Padda et al. (6) and Krishn et al. (7) characterized exosomes and microvesicles from patient plasma as a subtype with a size ranging from 100 nm to 1 μm and enriched in PSMA, CD9-tetraspanin, and β1-integrin (8). A subsequent study (9) isolated a mixed population of exosomes with a mean size of 200 nm from patients with metastatic PCa. Most research into PSMA in membranes and extracellular vesicles (EV) has evaluated biomarker for liquid biopsy diagnosis, without further investigation of their biological role (7,9) and in cell-to-cell signaling transference.

EV from melanoma cells have been shown to modify the tumor microenvironment (TME) to foster a pre-metastatic niche (6) and LN metastases (10). Angiogenesis in the TME depends on factors secreted by the tumor and stromal cells to recruit and develop new blood vessels (11). Vascular endothelial growth factor-A (VEGF-A) and angiogenin are prime pro-angiogenic factors that are also induced in the TME (12) and sites of inflammation (13). Increased microvessel density in primary cancer samples is associated with poor prognosis (14,15). High VEGF-A levels in human and murine PCa tumors are associated with poor survival and recurrence (12). VEGF-A is a secondary messenger that recruits tumor-associated macrophages (TAM) to solid tumors, which can modify the TME in a pro-tumoral manner (13). We have previously shown a positive correlation between PSMA expression and poor prognosis of PCa in tissues, with increased detection of phosphorylation-4EBP-1 (p-4EBP-1) (16). p-4EBP-1 is a major downstream effector of the AKT pathway (16,17) and one of its functions is upregulation of VEGF translation (18), which is downregulated by AKT inhibitors (17). PCa patients with high phosphorylation on 4EBP-1 correlated with LN metastasis, or the 4EBP-1 protein intensity expressed in the cytoplasm had 2.5 times higher mortality rate than those with reduced protein expression (19). We showed that circulating PSMA-Mp can transfer PSMA to parental PSMA-negative cells, altering their phenotype and inducing them to secrete VEGF, angiogenin, and pro-lymphangiogenic factors. These factors enhance microvascular density and initiate the lymphangiogenic switch. Our data indicate that PSMA-Mp can directly influence the TME through biologically active components, with potentially far-reaching consequences for tumor development, detection, and therapy.

Methods

The Supporting Information section contains detailed information about methods and supplementary figures.

Cell culturing for mice models, Mp analysis, and liposome confection

All cells were cultivated in fully supplemented media, as recommended by ATCC, as specified in the supplementary section. Every time we performed Mp isolation, we removed FBS from the medium. In addition, every Mp isolation batch and liposome characterization were conducted in the exchange buffer to filter PBS, quantified, and analyzed by DLS and NTA, as described below.

Mp and tissue protein content analysis in vitro and in vivo

The homogenization of Mp, cells, cell media without FBS or VEGF, or tissue samples were extracted (when applicable) and quantified as described in the supplementary data section. To summarize, the technologies included flow cytometry, ELISA, angiogenesis, and AKT-phosphorylated protein arrays. Moreover, to access VEGF, PSMA, and vesicles within tissues, we first labeled them, followed by molecular imaging of the mice. All procedures performed to collect this data are described in the Supporting Information section.

Tumor mouse models and Mp injections to mimic tumor Mp-shedding

The IACUC of Memorial Sloan Kettering Cancer Center approved the animal protocols and followed the Guide for the Care and Use of Laboratory Animals by the National Institutes of Health-NIH). All animal procedures were performed under anesthesia with a 1–4% isoflurane-oxygen mixture. Mp biodistribution in healthy mice was assessed in six-week-old male Balb/c nude outbred mice using BODIPY-CER- Mp s injected daily through the tail vein. These were followed by images, as described below. In addition, the healthy mice were subjected to PET imaging before raster-scan optoacoustic mesoscopy (RSOM) and IVIS, as described below and in the Supporting information section. Balb/c nude outbred male and six-week-old mice were purchased from Taconic Laboratories for all experiments. The tumor-bearing mouse models included groups with one subcutaneous injection of 2×10⁶ PC3WT cells on one side of the lower flank (right), or on both sides (PC3WT-L and PC3WT-R) with 2×10⁶ cells each, or with two implants of PC3WT (L) and PC3PSMA (R). Additionally, mice without tumors were used to monitor WT-Mp, PSMA-Mp, or PBS injections for 25 days. The measurement of the length and depth of the tumor with a digital caliper determined the tumor volume based on an approximation of an ellipse volume. After 33 days, all mice that developed measurable tumors were imaged for baseline, and time point values by RSOM or IVIS as described in the legends. The mice were assigned to specific groups and injected daily, as described by other authors. We created two groups: (1) received PSMA-Mp daily injections and (2) WT-Mp (2×10¹⁰ vesicles/injection/day/mouse) with “n”= up to 4 animals per group due to the excessive work of Mp fresh isolation, quantitation and labeling with fluorescent dyes. At the end of the experiment, constant CO₂ airflow inside the chamber induced euthanasia in mice. Tissue samples were also collected for analysis.

Angiogenesis morphology and quantitation assays

RSOM was performed with a scanner to follow vessels growing in real time for up to 25 days in a row, as previously described (20). To validate the quantitation in segmented images, we used the immunofluorescence of vessels, TME, and leakiness by PSMA-fluorescent probe extravasation from the vessels. After collecting data from the experiments, we performed immunofluorescence in these tissues, acquired images using confocal microscopy, and used image J for quantification. All the procedures are described in detail in the Supporting information section.

RNA extraction from Mp for RT-qPCR reaction

Mp were isolated from wild type (WT) and PSMA-overexpressing PC3 cells as described above. First, RNA was collected from both Mp using an RNA isolation kit from Exiqon, following the manufacturer’s instructions. Then, cDNA was synthesized from 1 μg of RNA using High-Capacity and real-time qPCR using SYBR Green (SYBR Green Master Mix) chemistry for the PSMA gene (forward:5’-TTG GAA TCT TCC TGG AGG TG-3’; reverse:5’-CTG CTA TCT GGT GGT GCT GA-3’). We used Flotilin-1 as an endogenous control (forward, 5’-GAA GAC GGA GGC TGA GAT TG-3’; reverse, 5’-ATC CGT GCA TCT TTT TGG AC-3’). We used independent Mp isolation (batches) from PC3 WT or PSMA-overexpressing cells for each reaction. The 2^delta-delta Ct method was used to determine the relative gene expression (fold change), using the WT-Mp as the reference and PSMA-Mp as the target sample for fold-increase calculations by ABI 7500 Real-Time PCR.

Gene expression and molecular signature analyses

We downloaded a dataset for prostate cancer patient samples (Firehose Broad GDAC portal, http://firebrowse.org/, The cancer genome atlas-TCGA, GEO # 104419). First, we defined “High-PSMA” tumors as the ones with expression levels > 8 and “Low-PSMA” tumors as those that fell below 8 (RSEM normalized RNA-Seq, gene/GAPDH). We then used Spearman’s correlation coefficient for standard co-expression analysis to assess the correlation between the two gene expression levels using R (version 3.3.2). Finally, we calculated the disease-free survival from the gene expression profiling interactive analysis (GEPIA) webserver (21) and “High-PSMA” vs. “Low-PSMA” tumors (n=320) plus VEGFA, Placental Growth Factor (PlGF), PDGF-BB, and ANGPT2 genes signatures identified as 5-signatures group.

Statistical analysis

We used the GraphPad Prism version 9.0 for Windows^® for statistical tests indicated in each figure legend, with alpha and “P.” The Microsoft Office^® Excel software 2007 helped organize the data.

Results

PSMA-Mp transfers PSMA to recipient cells.

To investigate whether PSMA cargo in isolated Mp could trigger pro-angiogenic factors in PCa, we utilized the PSMA-negative human PCa cell line PC3 (PC3WT) as a model of PSMA-negative PCa and a recipient of Mp from PSMA-expressing cells. We characterized the isolated Mp by determining their marker expression and assaying their potential to transfer detectable PSMA to recipient cells. In addition, we observed nuanced differences in the size distribution among Mp isolated from different PSMA-PCa donor cell lines. The supplementary data show the characterization of isolated Mp and in vitro functional assays of uptake and PSMA transfer to recipient cells (Supplementary Fig. 1 and 2).

PSMA overexpression in engineered PC3 cells (PC3PSMA) did not affect the size of the shed Mp (PSMA-Mp) (Fig. 1A), consistent with previously published results for Mp isolations (22). When comparing PSMA-Mp or WT-Mp, the MPs originating from other PSMA-expressing cells exhibited significant size variations (Supplementary Fig. 1A–D). The isolated WT-Mp and PSMA-Mp sizes across multiple batches showed non-significant variations (Fig. 1B, Supplementary Fig. 1D). The Mp isolation method established here (Supplementary Fig. 1A) was derived from centrifugation methods (22) and is explained in detail in the Supporting information section. Immunoblotting of the total extracted Mp protein showed (1) biomarkers flotilin-1, CD9, β1-integrin, and β-actin (Fig. 1C). Immunoblotting revealed that PSMA-Mp had enhanced levels of PSMA (Fig.1C) compared to other PSMA positive vesicles (see Supplementary Fig. 1F). Flow cytometry analysis demonstrated that approximately 10% of Mp express Annexin-V, a biomarker shared with microvesicles (see Supplementary Fig. 1G). Confocal and fluorescence microscopy revealed comparable Mp uptake by PC3WT cells, despite our initial exploration of potential differences in uptake due to varying PSMA expression in Mps sourced from diverse cells (Supplementary Fig. 1H – M). However, only PSMA-Mp retained detectable PSMA proteins in PC3WT recipient cells (Supplementary Fig. 1N – O).

Figure 1: Characterization of Mp by size, PSMA expression, and PSMA-transference to recipient cells.

(A) WT-Mp and PSMA-Mp size characterization from different Mp extractions (n=12) measured by dynamic light scattering analysis showed an (B) equivalent size distribution range. (C) Immunoblot of total Mp protein content, separated by native gel electrophoresis. Immunoblotting of the Mp markers flotillin-1, CD9-tetraspanin, and beta-actin was performed. PSMA was detected in both dimer (arrowhead) and monomer (blue arrow) forms via immunoblotting in total protein extraction from Mp. (D) Representative fluorescence microscopy micrographs of PC3WT cells after a 24 h incubation period with fluorescently labeled Mp (membrane-label green, PKH67-labeled) followed by staining with a primary PSMA antibody and secondary Cy3-labeled antibody (red) along with DAPI nuclear staining (blue) compared to PC3PSMA cells (as positive control) or negative control (PC3WT cells plus LNCaP-Mp followed by staining with the secondary Cy3-antibody and DAPI). Scale bar = 10 μm.

This highlights that although LNCaP-Mp and 22RV-Mp are taken up by PC3WT cells, they lack a sustained presence of PSMA, as indicated by its absence in immunoblotting analysis. This implies a scenario where the PSMA protein undergoes degradation within the cells after acquiring PSMA from LNCaP or 22RV. A similar uptake and PSMA blotting pattern was also observed when PSMA-Mp was transferred to endothelial and macrophage cells (Supplementary Fig. 2A–D).

Having established that PSMA-Mp could transfer PSMA protein to recipient cells, we investigated the biodistribution and in vivo effects of both WT- and PSMA-Mp in healthy adult nude mice (Supplementary Fig.3A–E). Consistent with the in vitro results, we detected no signs of toxicity in mice, such as no changes in body weight, ulceration, bleeding, or labored breathing for up to 25 days of daily tail vein injections of isolated Mp. WT-Mp and PSMA-Mp were distributed in a similar pattern in all organs (Supplementary Fig.3A–C), although PSMA-Mp induced greater ¹⁸F-FDG uptake in Axillary and Sciatic lymph nodes (A.A.LN or S.LN) (Supplementary Fig. 3D). These mice free of tumors, when exposed to PSMA-Mp tail injections resulted in up to twofold increases in signal from PSMA and anti-VEGF fluorescent probes in comparison to the mice receiving WT-Mp or PBS tail injections (Supplementary Fig. 3F–G). These results support the Mp circulation, as the A.A.LN returns excess interstitial fluid back to the systemic circulation and it is an immune defense sensor against foreign particles within the body’s circulation. Physiologically, cells and particles from subcutaneous tumor implants can potentially be transported via lymph vessels into the thoracic duct and subsequently gathered by either the right or left axillary lymph nodes. This phenomenon has been described by other authors in relation to skin diseases and malignancies (23). In a mice model with dual tumors implantation, after injecting PC3WT cells on the left flank and PC3PSMA cells on the right flank, we noticed a twofold increase in PSMA uptake in the A.A.LN on the right side, as well as higher levels of PSMA compared to animals that received PC3WT cells on both sides. (Supplementary Fig. 3H and I). These studies established PSMA-Mp as vehicles for PSMA transfer to recipient cells both in vitro and in vivo.

PSMA-Mp modify cell types present in the TME to increase secretion of pro-angiogenic factors

Previous studies have demonstrated the expression of PSMA in tumor neovasculature and retinal revascularization models (24), suggesting a pro-angiogenic function of PSMA. Shapiro et al. showed that PCa tumors in PSMA-knockout mice were less vascularized (24). Therefore, we sought to determine whether Mp carrying (and transferring) PSMA could contribute directly to the release of vascular factors in the TME. To elucidate the role of Mp in tumor angiogenesis, we first investigated whether Mp could carry pro-angiogenic factors such as protein cargo. An array of human pro-angiogenic factors within the Mp was quantified (Table 1). No notable differences were seen in angiogenic factor encapsulation between WT-Mp and PSMA-Mp in successive batches. External PSMA incubation with PC3WT cells improved pro-angiogenic factor encapsulation by WT-Mp. The absence of statistically significant differences in pro-angiogenic factor encapsulation between WT-Mp and PSMA-Mp rules out the direct transfer of factors as the cause of the heightened pro-angiogenic environment. We also explored the mRNA content of Mp, evaluated by qPCR, and demonstrated an increased level of PSMA-encoding mRNA in Mp originating from PC3-PSMA cells. This increase in PSMA reflects the increased intracellular PSMA transcript levels in the cells and their shed Mp (Supplementary Fig. 4B). Elevated mRNA levels detected inside PSMA-Mp showed that Mp can transfer mRNA to recipient cells in addition to protein. The mRNA transfer could further contribute to PSMA levels in recipient cells (25,26), potentially modifying the TME.

Table 1:

Summary of pro-angiogenic factors analyzed through quantibody protein array.

Factor	Abbreviations
Angiogenin	-
Vascular Endothelial Growth Factor A	VEGF-A
Angiopoietin-2	ANG-2
Epidermal Growth Factor	EGF
basic Fibroblast Growth Factor	bFGF
Heparin-binding EGF-like growth factor	HB-EGF
Hepatocyte Growth Factor	HGF
Leptin	-
Platelet derived Growth Factor BB	PDGF-BB
Placental Growth Factor	PlGF

We utilized a cytokine array to investigate the potential biological effects of exposing tumor, endothelial, and macrophage cells to PSMA-Mp (Fig. 2 and Supplementary Fig. 5A–F). Notably, PSMA-Mp exposure led to enhanced secretion of VEGF-A and angiogenin in all cell types (Fig. 2A–B, Supplementary Fig. 5A–B). Furthermore, endothelial cells (EC) exposed to PSMA-Mp exhibited increased expression of VEGF-A and angiogenin (Fig. 2B, Supplementary Fig. 5B) compared to cells incubated with WT-Mp or liposomes. However, no significant difference was detected in any of the eight tested factors among batches (Supplementary Fig. 5D). Similarly, macrophages displayed a consistent response to both WT and PSMA-Mp exposures, resulting in similar elevations in VEGF and angiogenin secretions (Supplementary Fig. 5C). Only PSMA-Mp induced the upregulation of ANG-2, HGF, and placental growth factor (PlGF; Fig. 2C and Supplementary Fig. 5C). PlGF and VEGF-A may also contribute to the polarization toward a tumor-promoting, immunosuppressive M2 macrophage phenotype (27).

Figure 2: VEGF-A and Angiogenin secretion into cellular media after recipient cell’s exposure to Mp.

In vitro exposure of (A) tumor parenchyma (PC3WT) or (B) Endothelial cells and (C) macrophages (THP-1) with Mp for up to 24 hours, measured by protein array. Graphs show the average concentration of VEGF and Angiogenin. N=3 independent experiments for each (2-way ANOVA was used to compare all data, with WT-Mp acting as a control; ****P <0.0001 and *P=0.0472).

To demonstrate the role of PSMA in VEGF-A expression, knockdown of PSMA in PSMA-expressing PCa cells (LNCaP) reduced VEGF-A secretion (Supplementary Fig. 5H). PC3-PSMA cells and CT26 cells overexpressing mouse PSMA (CT26mPSMA) both exhibited increased VEGF-A secretion (Supplementary Fig. 5I and 5 J, respectively). Because pro-inflammatory cytokines can also induce pro-angiogenic activity, we examined cytokines and molecules involved in macrophage migration, proliferation, chemoattraction, or activation (Supplementary Fig. 5L and 5M). However, our experiments demonstrated no statistically significant differences in the secretion of proinflammatory cytokines by tumor cells exposed to WT-Mp, PSMA-Mp, or liposomes. When compared to WT-Mp or liposomes, only MPs carrying PSMA demonstrated an elevation in the levels of proliferation-inducing factors in both endothelial cells (EC) and macrophages (Supplementary Fig. 5M). Our data showed that PSMA-Mp increased pro-angiogenic signals in stromal cells prevalent in the TME (Supplementary Fig. 5D and 5G) via increased expression of VEGF in tumors and angiogenin in EC. In addition to its impact on tumor microenvironment alterations, PSMA-Mp can promote the proliferation of macrophage cells within the bone marrow through the augmentation of G-CSF or GM-CSF, which is facilitated by endothelial cell (EC) secretion.

PSMA-Mp increased PSMA expression in PCa tumor xenografts

Following the observation of increased pro-angiogenic factors in the TME after exposure to PSMA-Mp in vitro, we next xenografted mice subcutaneously with PC3WT tumors to investigate the effects of PSMA-Mp in vivo on primary tumors. Once the tumors reached a volume of ~200 mm³, mice received a daily tail vein injection of either WT-Mp or PSMA-Mp.

Mice from both groups showed similar Mp-fluorescent (NIR-Labeled-Mp) distribution in tumors and organs (Fig. 3A), corroborated by ex vivo imaging (Supplementary Fig. 6A). PSMA-Mp injections increased PSMA in tumors by 4.6-fold compared to WT-Mp at day-25. PSMA expression levels were detected using the PSMA-targeting fluorescent agent YC27-Cy5.5(28) (Fig. 3B), which binds and enters PSMA-positive cells by endocytosis. Our ex-vivo study showed no significant difference in Mp organ accumulation either on tumor-bearing or healthy mice (Supplementary Fig. 6A–B). These results align with previous reports showing equal PC3 cell-derived-vesicles accumulation without signs of toxicity in mice bearing PC3WT tumors 24h after i.v. injection. Ex vivo tumor images revealed increased PSMA levels (increase in YC-27-Cy5.5 fluorescence, Supplementary Fig. 6B, insert) in the PSMA-Mp-exposed group. Fluorescence (Fig. 3C) and whole-tumor confocal microscopy (Supplementary Fig. 6D) confirmed enhanced PSMA-positive areas in tumors and spleens (Fig. 3D). Fluorescence microscopy showed a 6.3-fold increase in tumors and 5.6-fold fluorescence in the spleen of the PSMA-Mp compared to the WT-Mp group. Tissue autofluorescence background subtraction was performed on immunofluorescence microscopy images and in vivo imagery, as described in Supplementary Fig. 6C.

Figure 3: Mp distribution in tumors from mice receiving either PSMA-Mp or WT-Mp.

Nude mice bearing PSMA-negative PC3 tumors received daily intravenous injections (tail vein) of WT-Mp or PSMA-Mp for 25 days. Multispectral fluorescent imaging (MFI) acquisition followed by automated spectral unmixing images highlights each specific fluorophore probe. Representative optical images (A-B) showing BODIPY-CER-Mp (NIR-labeled-Mp) and PSMA (YC27-Cy5.5) distribution and quantitation. Immunofluorescence microscopy of sectioned tumor (C) and spleen tissue (D), PSMA-positive areas shown by immunofluorescence (green). (The results represent the analysis of twenty different fields (20x magnification) and quantification from 2 independent experiments. Graphs represent the average and SD of n = 4 animals each. The unpaired t-test was used to evaluate for significance where *P≤0.04and ***P≤0.004).

PSMA-Mp induces tumors to recruit more blood vessels and secrete pro-angiogenic factors in vivo

We next explored the effects of PSMA-Mps on the recruitment and morphology of tumor vessels in vivo by comparing baseline vasculature (day 1) to the 9- and 25-day time points course of daily Mp injections. We evaluated changes in subdermal tumor vasculature recruitment using RSOM. RSOM uses pulsed laser light to induce thermoelastic expansion of hemoglobin molecules, resulting in detectable ultrasound emissions that allow longitudinal noninvasive visualization of microvessel architecture with high spatial resolution (20,26).

On day 25, the density of smaller vessels (indicated by the green color-coded vessels in Fig. 4A) and the vascular branching ratio (Fig. 4B) increased significantly in mice receiving PSMA-Mp compared to baseline vasculature. The tumor tissue was cleared (27) and stained for endothelial marker CD31. By stacking a series of confocal microscopic image sections in adjacent layers of the tumor tissue, we also constructed a 300 μm 3D confocal image (with Z-, Y-, and X-axis spatial resolution). Microscopy confirmed more CD31 positive vessels within the tumors in mice exposed to PSMA-Mp than to WT-Mp (Supplementary Fig. 6D). In addition, correlation analysis confirmed that PSMA-Mp, but not WT-Mp, induced increased PSMA levels in the stroma and vasculature.

Figure 4: Peripheral and intratumoral microvascular density revealed by RSOM.

(A) Green and red color-coded vessels formations inside PC3WT tumor xenografts in vivo during Mp stimuli. Note visual tumoral neoangiogenesis morphologies, such as glomeruloid microvascular proliferation (GMP) and vascular malformations (VM). (B) The representative graphic shows the number of branches inside the tumors at 9 and 25 days after the first intravenous injections to the initial number of branches (****P=0.00001; 2-way ANOVA with Bonferroni correction). Average of three independent experiments.

After confirming that PSMA levels increased on the 25^th day in the tumors of animals from the PSMA-Mp group, we quantified VEGF-A in vivo and ex-vivo. The anti-VEGF-A-Cy7 fluorescent antibody showed an increased accumulation of VEGF-A in tumors over time. This experiment revealed significantly higher VEGF-A levels in animals that received PSMA-Mp (Fig. 5A–C) than in WT-Mp at 9 and 25 days (1.7-, 2.3-fold respectively; normalized to the fluorescence before the first Mp i.v. injection). Proteins extracted from tumors corroborated these images in the ex vivo analysis (Fig. 5D). The Cy7 quantitation showed no VEGF-A changes in organs from healthy or tumor-bearing animals (Supplementary Fig. 7A–D).

Figure 5: Intratumoral assessment of VEGF-A, pro-angiogenic factors, and microvascular branch length in PC3WT tumor xenografts and spleen in mice receiving daily injections of WT-Mp or PSMA-Mp.

(A) Representative in vivo epifluorescence imaging of PC3WT tumor-bearing mice injected daily with Mp and injected with (B) Cy7-labeled anti-VEGF-A antibody, observed in tumors (24-hours and 96-hours (*) after antibody biodistribution). The images and the quantification are representative of 2 independent experiments (n = 4 mice). All statistical analysis was conducted using 2-way ANOVA (*P≤0.05) (C) Ex-vivo tumor analysis of VEGF-A-Cy7 fluorescent signals (*P≤0.01; unpaired test t) (D) A proteomic array assay of 10 pro-angiogenic factors was performed to quantify pro-angiogenic protein levels in tumors. Whole protein extracts were normalized for respective tumor weight. (*P≤0.01) Each color code in the heatmap columns represents the average of 4 animals at either 9 or 25 days. (E) Representative CD31 immunostaining of tumor tissues on day 9 (left) and day 25 (right) along with (F) Representative spleen microvasculature images from mice exposed to Mp. (G & H) Quantification of vasculature branch length per area (μm²) at days nine & 25 for WT- & PSMA-Mp exposed PC3T tumors. (I) Quantification of spleen CD31 expression levels from PC3T xenografted mice exposed to WT- or PSMA-Mp. Images and the quantification are representative of 2 independent experiments (n = 4 mice, *P≤0.01; ***P=0.001). All statistical analysis was conducted via an unpaired t-test. GC, germinal center; MZ, marginal zone; CA, central artery and RP, red pulp.

The immunospot array showed a significant increase in VEGF-A (2-fold) in tumors from the PSMA-Mp group on day 9 after the Mp injection, which increased at day 25 to 100-fold more VEGF-A and angiogenin compared to the WT-Mp group. In addition, a significant increase in six out of ten evaluated pro-angiogenic factors was observed in tumors from the PSMA-Mp group (Fig. 5D, right column) over a period of 25 days. This upregulation of pro-angiogenic factors in mice receiving PSMA-Mp injections was accompanied by changes in microvessel density (MVD, CD31⁺ analysis, number of branches) in the tumors (Fig. 5E), as observed by RSOM (Fig. 4A–B). The quantification showed increased branching in tumors from the PSMA-Mp group at 9- and 25-days after Mp injections (Fig. 5G and H). Due to the spleen’s pivotal function in filtering antigens from the bloodstream, we investigated morphophysiological alterations in spleen from groups of mice exposed to WT-Mp or PSMA-Mp. Similar enhancement of MVD-CD31+ was observed in the spleens of mice from the PSMA-Mp group (Fig. 5F and I). In mice, injecting PC3WT cells on the left side and PC3PSMA cells on the right side resulted in three times higher anti-VEGF concentration only in the axillary lymph nodes compared to mice that received PC3 cells on both sides (Supplementary Fig. 7E ). However, the tumors in mice that received PC3WT cells on the right flank and PC3PSMA cells on the left flank exhibited a slight increase in VEGF levels, although not statistically significant, when compared to tumors that received PC3WT cells on both sides (Supplementary Fig. 7F).

Our data showed that exposure of mice to PSMA-Mp (up to 25 days) induced morphological vessel changes in tumors accompanied by increased levels of VEGF-A in comparison to WT-Mp. PSMA-Mp successive injections also increased factors that act in loop and concomitantly with VEGF-A (bFGF, HB-EGF, PDGF-BB, and PlGF)(31) in a pro-tumoral manner (28). These biochemical and morphological changes show that PSMA-Mp induces more pro-tumoral alterations in the TME than WT-Mp.

To establish whether PSMA may be driving a similar set of pro-angiogenic mediators in patient tumors, we explored The Cancer Genome Atlas (TCGA) data to perform Gene Set Enrichment Analysis (GSEA) and found specific genetical signatures linked to high PSMA expression. The random data comprised 492 clinical samples, allowing us to compare mRNA levels from PSMA-high versus PSMA-low patient tumor samples (29). These gene expression profiles revealed an upregulated pro-angiogenic molecular signature concomitantly in PSMA-high-expression samples. This GSEA molecular signature on this PSMA-plus-scenario is like our in vitro and in vivo results (Supplementary Fig. 8A–F), that is, with an increased co-expression of VEGF-A, PlGF, PDGF-BB, and ANG-2 (Supplementary Fig. 8A–D). The same gene expression data also showed that these combined PSMA-gene-set co-expressions are unfavorable to overall survival rates compared to the lower gene expression (Supplementary Fig. 8F).

Macrophages are recruited to tumors and tissues that express VEGF-A (30) and differentiate to M2-phenotype in wound healing (31,32) and they secrete PDGF-BB to the TME (28). CD68 is used as a generic marker for tumor-associated macrophages (TAM) in PCa (33) and is frequently associated with tumor progression and poor prognosis (34). We aimed to investigate whether PSMA-Mp and VEGF-A could similarly prime the PCa TME (10).

Tumors and spleens from the PSMA-Mp group harbored 2.5-fold more CD68⁺ cells than those from the WT-Mp group after 25 days Mp injections (Fig. 6A and B). The increased number of CD68⁺ cells coincided with increased fluorescence from fluorescently labelled anti-VEGF antibodies and increased VEGF-A protein levels. The liver and lung tissues showed no differences in the number of CD68⁺ cells (Supplementary Fig. 9A).

Figure 6: Quantification of CD68⁺ and p4EBP-1⁺ cell levels inside tissues at day 25 post Mp exposure.

Immunofluorescence images (left columns) and quantification (right columns, graphs) of tissues after daily Mp injections into mice showing tumor (A) and spleen (B) CD68⁺ cell recruitment. GC=germinal center, RP=red pulp. (C) p4EBP-1⁺ cytoplasmic and nucleus staining in tumor cells. The images represent tumors from each group ranging from 2× to 40× magnification; quantification results represent repeated independent experiments (n = 3). The data represent the average and SD of animals each. All statistical analysis was conducted using Student’s unpaired t-test, and the results were considered significant *P ≤0.05, ***P≤0.001).

On day 25, tumors from the PSMA-Mp group demonstrated also increased expression of PDPN (2-fold) and PDGF receptor β (PDGFRβ), multifunctional markers associated with tumor angiogenesis, lymphangiogenesis, and LN metastasis (35–37) (Supplementary Fig. 9B and D). PDGFRβ is the receptor for PDGF-BB, a dual pro-angiogenic and pro-lymphatics factor detected enhanced in protein extracts from tumors exposed to PSMA-Mp (Fig. 5D). These findings show that PSMA-Mp induce an increase in tumor angiogenesis by VEGF-A that physiologically recruit more macrophages to the TME, followed by pro-angiogenic and pro-lymph angiogenesis cell recruitment.

We have previously shown that 4EBP-1 phosphorylation from AKT signaling increases in human PCa tissue samples with higher PSMA expression(16); here, we investigated whether PSMA-Mp could have the same effect on tumors in mice. To this end, we evaluated the phosphorylation status of 18 proteins associated with the AKT signaling pathway. Tumors exposed to PSMA-Mp demonstrated significant upregulation of phosphorylated 4EBP-1, GSK3b, RSK1, and RSK2 25 days after daily PSMA-Mp injections (Supplementary Fig. 10A and B) and PDKK1 phosphorylation on both groups. Immunofluorescence images confirmed more 4EBP-1 phosphorylation in tumors from the PSMA-Mp group than in the WT-Mp group (Fig. 6C).

Because AKT/mTOR activation can also induce cell growth, we further questioned whether PSMA-Mps could result in proliferation stimuli via MAP kinase (24) through p38 mitogen-activated protein kinase phosphorylation (phospho-p38-MAPK) detection. The differences in p38 mitogen-activated protein kinase phosphorylation (phospho-p38-MAPK) between the groups became apparent only after 25 days. Mice injected with PSMA-Mp (Supplementary Fig. 10C and D) had slightly more phosphorylated p38 in protein tumor extracts than tumors from mice receiving WT-Mp, even before volume differences were measured using a caliper (Supplementary Fig. 10E).

Taken together, our results show that PSMA-Mps alter the TME by increasing pro-angiogenic and lymphatic factors, driven by the secretion of VEGF-A and angiogenin. Furthermore, these TME modifications prime tumor tissue for lymphangiogenesis. Notably, the activation of 4EBP-1 via phosphorylation (AKT-pathway end mediators) activates a cancer-promoting factor, VEGF, without enhancing cell turnover (38,39).

Discussion

EV have emerged as a method for PCa detection; however, few studies have explored EV and Mp pathophysiology(1). Mp are natural nanomaterials that deliver proteins and mRNA to the TME, thereby modulating signaling to and from recipient cells (1). They are naturally occurring liposome-like nanocarriers derived from the cells of the host organism. It has been shown that cell-derived vesicles are able to transfer proteins, miRNA, and drugs to recipient cells in vivo (5). However, the biological consequences of their presence have not been fully elucidated. For example, Mp originating from PCa lesions may carry certain proteins such as PSMA. What is the biological consequence of transferring this protein? PSMA expression and angiogenesis are correlated in experimental and clinical studies, without fully comprehensive functional studies. Our previous work described PSMA and folic acid as upstream initiators of the AKT pathway in PCa, culminating in 4EBP-1-phosphorylation (16). In this study, we investigated whether Mp derived from PSMA-PCa cells could effectively transfer functional PSMA to TME recipient cells, resulting in tissue alterations.

Functional transfer of PSMA from expressing to non-expressing cells was achieved through isolated Mp from PSMA-overexpressing PCa cells, inducing increased expression of VEGF-A and angiogenin in tumor cells and other cells of the TME, both in vitro and in vivo. The transfer of PSMA by Mp induces additional cellular changes associated with a PSMA-positive phenotype. Previously, we demonstrated that PSMA activates an AKT signaling cascade through its carboxypeptidase enzymatic function, releasing glutamate to stimulate metabotropic glutamate receptors functioning as G protein-coupled receptors (16). PSMA from Mp was integrated without signs of protein degradation and was able to trigger AKT signaling, as described previously (16). Because we also detected PSMA mRNA in Mp, it is reasonable to conclude that recipient cells may translate PSMA from the delivered mRNA payload. A functional PSMA molecule could originate in the recipient cell by two different mechanisms: directly as an intact protein by membrane fusion, and/or as a source of PSMA mRNA for translation by the recipient cell. Mp-mediated conferral of a pro-angiogenic phenotype could be an essential biological function of PSMA as a promoter of angiogenesis.

PSMA is expressed in the tumor neo vasculature (24); however, its role in promoting pro-angiogenic factors in vivo is a novel and previously unknown aspect of PSMA’s biological function, significantly expanding our understanding of this protein’s function. We are continuing our research into PSMA molecular pathways to enhance modifications in the TME, with the hope that our future work will unveil the complete molecular pathway of PSMA in TME angiogenesis. It is plausible that PSMA’s natural(24) actions over new vessels are amplified in the dysregulated cancer milieu.

Sustained exposure of tumors to PMSA-Mps triggers changes in tumor vascular density and structure, as shown via high-resolution optoacoustic imaging. RSOM does not require an invasive window chamber such as intravital microscopy, allowing transdermal observations of non-inflamed tissue over time. RSOM can produce higher-resolution images of perfused vessels inside tumors than other non-invasive techniques, such as micro-CT and micro-MRI blood volume measurements (20,26). Tumors in mice that received PSMA-Mps exhibited progressive time-dependent changes in tumor vessel formation, including increased endothelial cell branching and vascular malformations, as confirmed by microscopic analysis of tumor tissue. In vivo, epifluorescence showed a ten-fold increase in intratumoral VEGF-A and pro-angiogenic factors at 9 and 25 days in the PSMA-Mp-exposed groups. Upon injection with a PSMA-binding probe (40) (YC-27-Cy5.5), PSMA-Mp mice also demonstrated brighter fluorescent signals within tumor tissue than WT-Mp mice, indicating increased amounts of PSMA within these tumors. These results suggest that PSMA-mediated priming of the PC3 TME with released pro-angiogenic factors is affected by Mp and PSMA delivery.

The upregulation of VEGF-A is widely recognized for its role in facilitating macrophage infiltration into tumors, leading to tumor angiogenesis (31) and unfavorable prognostic outcomes for patients (13). On day 25, both tumors and spleens of PSMA-Mp-exposed mice exhibited pronounced CD68+ staining, indicating ongoing communication between tumor cells and various cellular components of the tumor microenvironment (TME), including macrophages and endothelial cells. This communication actively promotes and sustains the processes of angiogenesis and lymph angiogenesis (31) through the induced secretion of VEGF-A (35). Furthermore, melanomas release extracellular vesicles (EVs) containing elevated levels of nerve growth factor receptors, facilitating their dissemination through the lymphatic system and subsequent uptake by lymphatic endothelial cells, contributing to the metastasis of cancer cells to lymph nodes (10). Furthermore, VEGF-A increased 100-fold at day 25 compared to the levels at 9 days after PSMA-Mp injection, highlighting the capability of PSMA-Mp to shape a microenvironment favorable to angiogenesis and alter the local immune environment through an influx of macrophages (Fig. 7). We also detected increased expression of the PDGF-BB receptor, PDGFRβ, within tumor tissue from mice treated with PSMA-Mp, which can strongly activate pericytes and lymphatic endothelial cells (LEC) (41). Similarly, in our experiments involving the implantation of both PC3WT and PC3PSMA cells, we observed elevated levels of PSMA and VEGF in the lymph nodes, indicating a consistent association between PSMA-Mp and enhanced VEGF expression in vivo.

Figure 7: PSMA in Mp stacks biological activity in recipient cells.

PSMA-Mp activates a switch in the tumor microenvironment that drives prostate cancer angiogenesis at early time points in PCa. This tissue switch is activated by increased secretion of pro-angiogenic factors to TME cells, leading to increased 4EBP-1-phosphorylation and then boosting the TME to produce pro-angiogenic and pro-lymph-angiogenesis secondary mediators (Figure made with Biorender.com software).

We observed a novel effect of PSMA, wherein PSMA transferred via Mp creates a pro-angiogenic TME. This is mediated by VEGF-A secretion by tumors and endothelial cells. However, our results did not show statistically significant differences in tumor size (volume) for up to 25 days, nor changes in survival p38-marker. In addition, short-term daily exposure to Mp (9 days) did not increase the phosphorylation of proliferation mediators in tumor cells inside TME (such as p38, p53, PTEN, and others), regardless of whether PSMA was present in the Mp or if VEGF-A secretion into the TME was increased. Our findings show that PSMA-Mp triggers PCa to produce VEGF and angiogenin, modifying TME cells even before triggering tumor cell proliferation through MAPK activation, as other authors showed in breast cancer cells (42). Hypoxic regions within tumors can contribute to the secretion of VEGF-A. However, in the model utilized for this study, the tumors displayed a slow growth pattern with consistent diameter and volume sizes across all groups. As a result, the investigation did not focus on studying the effects of hypoxia at this stage. Consequently, the detection of PSMA-Mp tumor and lymph nodes alone cannot be relied upon to determine the prognosis of prostate cancer, as previously indicated (6,7).

Our results have a novel implication because PSMA-Mp triggered VEGF-A, which is also implicated in lymph angiogenesis (28,31,43) which may facilitate metastasis to the LN (42). LN metastasis is an important prognostic indicator (44). The normal prostate epithelium has few lymphocytes within the tissue, but this number increases 100-fold (45) in inflammatory prostatitis, preconditioning to PCa, and further changes to LN activation to predict adverse outcomes (46) and stage (47). A recent study by García-Silva et al.(10) demonstrated that Mp derived from melanomas delivers the nerve growth factor receptor (NGFR) to promote LN metastasis. The authors showed that melanoma models acutely exposed to melanoma-derived Mps elicited Lyve-1 (a dual EC and LEC marker), PDGF-BB, and PDGFRβ increase, resulting in metastasis. In our study, we observed that PSMA-Mp administration led to a notable increase in ¹⁸F-FDG uptake in the lymph nodes of mice over a period of 25 days. Moreover, the increased presence of macrophages, PDPN, PDGF-BB, and PDGFRβ was found to facilitate the enhancement of microvessel density. These factors also play a role in activating pericytes and have been associated with the stimulation of lymph/angiogenesis (10). As we demonstrated by implanting dual PC3WT and PC3PSMA cells in the same mouse, both VEGF and PSMA exhibited enhanced levels in the auxiliary lymph nodes compared to mice with dual PC3WT implants. These findings indicate the coexistence of these factors in tissues and suggest a potential association with the enhancement of lymphangiogenesis. Lymphatic vessels are frequently observed around tumor tissues co-existing with areas of neo angiogenesis, they can originate from endothelial cells from pre-existing vessels and could be the starting point for PCa lymphatic metastasis (43,48). The PSMA protein (and PSMA mRNA) is delivered via Mp and results in AKT activation and 4EBP-1-phosphorylation (16) in tumor cells, increasing microvessel density through secretion of pro-angiogenic and pro-lymphangiogenic factors. AKT activation followed by 4EBP-1-phosphorylation in PCa is correlated with tissue invasiveness and lymph node metastasis (16,48–50) and has a higher disease progression (19). Phosphorylated-phosphoinositide‐dependent kinase‐1 (p-PDPK1) typically follows AKT activation as a stimulus for PCa proliferation. In our study, we observed an enhancement of pPDK1 from day 9 to day 25. Importantly, this enhancement was independent of PSMA carried by Mp in vivo, a finding consistent with results reported by other researchers to tumor cells (51).

In conclusion, we showed that within a time limit of 25 days in mice, PSMA-Mp derived from PSMA-positive tumor cells induced more pro-angiogenic and pro-lymphangiogenic factors in PCa by transferring PSMA and combining a PSMA-positive phenotype in both tumor and stromal cells. As a result, enhanced vessel recruitment and angiogenic factors are sustained by increased cell recruitment to the TME, and 4EBP-1-phosphorylation further promotes ongoing translation. While we have not measured PSMA-Mp in patient sera, their presence has been described (6,7) and our data increases in biomarkers associated to metastasis via angio- and lymphangiogenesis. The data presented here opens the discussion that PSMA-Mp, VEGF, and 4EBP-1 phosphorylation may not only serve as diagnostic factors but could also contribute to changes in the TME, potentially furthering disease progression through the broad enhancement of mRNA decapping and enhanced proteins transductions. Therefore, further evaluation of clinical Mp data is warranted. This work bridges an experimental gap from previous studies by showing that the TME is modulated by PSMA-Mp beneficial to PCa around a 25-days window in already formed tumor masses, exceeding the observation time limit of prior studies in vivo.

Highlights.

• PSMA-Mp transfer PSMA protein to recipient cells, inducing pro-angiogenic changes in the tumor microenvironment (TME).

• PSMA-Mp delivery of PSMA promotes VEGF-A and angiogenin secretion in TME cells, enhancing angiogenesis and lymphangiogenesis.

• In vivo experiments show sustained exposure to PSMA-Mp leads to increased microvessel density, altered vascular architecture, and elevated CD68+ cell infiltration in tumors.

• PSMA-Mp-mediated AKT activation and 4EBP-1 phosphorylation promote a pro-angiogenic and pro-lymphangiogenic TME, without significant changes in tumor size.

• This study suggests that PSMA-Mp may contribute to disease progression in prostate cancer by modulating the TME through enhanced angiogenesis and lymphangiogenesis.

Data Access Statement:

All raw and processed data published here are available upon e-mail to corresponding authors.