CXCR4+ circulating tumor cells (CTCs) accumulate in subcutaneous, immune suppressive, CXCL12-loaded-hydrogel Authors: Luigi Portella, Dario Guido Di Febbraro, Dario Righelli, Giulia Bertolini, Giuseppe Guardascione, Giuseppina Rea, Crescenzo D'Alterio, Caterina Ieranò, Sara Santagata, Anna Maria Trotta, Marco Gentili, Roberta Sulsenti, Antonio Luciano, Giuseppe Palma, Francesca Bruzzese, Ilaria Assunta Parisi, Antonella De Luca, Luca Roz, Roberto Pacelli, Stefania Scala Published: 2026 ABSTRACT AbstractCirculating tumor cells (CTCs) sensing the chemokine CXCL12 and invading hyaluronic acid hydrogel (CXCL12 loaded hydrogel, CLG) might display metastatic attitude. CLG recovered-human lung (H460, A549), and ovarian cancer cells (IGROV-1) overexpressed the CXCL12 receptor, CXCR4, developed larger spheres and activate transcriptional programs of invasion and stemness. Interestingly, CLG-U87 glioblastoma cells highly expressed EpCAM and significantly upregulated the very specific NGFR, NTS, AQP1 and CMKLR1 genes, associated with cell proliferation, migration/invasion and metastasis. In a syngeneic model, subcutaneous CLG significantly attracted GFP-Lewis lung carcinoma (LLC) cells impairing lung colonization within 4 h from cell injection and up to twenty-one days. In lung, M1 macrophages rapidly increased post cancer cells injection while M2 prevailed after 10 days (T10). Neutrophils (Ly6Ghigh and Ly6Glow) infiltrated the lungs after ten days from cells injection with late expansion of immature Ly6Glow. According to lung niche, Empty gel (EG) and CLG were early infiltrated by macrophages and later by neutrophils. CLG generated an immunosuppressive environment defined by M1/M2/Ly6Glow able to capture and divert metastatic CTCs from the lung colonization.Keywords: CXCL12/CXCR4 axis, Hydrogel, Lung metastasis, Macrophages, Neutrophils Introduction The chemokine CXCL12 and the cognate Gi-coupled receptor CXCR4 are relevant to the metastatic process regulating cancer cell homing and tumor microenvironment composition [1]. Although 1 × 106 circulating tumor cells (CTCs)/gram of tumor tissue reach the circulation in a day [2], the majority of CTCs die in bloodstream due to oxidative stress, anoikis, immune surveillance and fluid shear stress [3,4]. Less than 0.01% of CTCs may eventually generate secondary tumors [3,5]. Fluid shear stress induces CXCR4 through active CXCR4/PI3K/AKT signaling on metastasis competent CTCs conferring survival advantage [3]. CTCs from breast, lung, pancreatic, hepatocellular, prostate and ovarian cancer patients express CXCR4 [[6], [7], [8], [9], [10], [11]], and CXCR4+CTCs possess poor prognostic significance [6,8,10,12]. Thus, CXCR4+CTCs may represents the most invasive cellular component within the heterogeneous CTCs population [3,13]. We previously demonstrated that subcutaneous hyaluronic gel loaded with CXCL12 (CXCL12-loaded hydrogel, CLG), diverts circulating murine B16-hCXCR4 cells from developing lung metastasis in immune competent C57Bl mice [14]. Moreover, a proof of principle clinical trial demonstrated that CLG isolated CTCs from ovarian, lung, endometrial, colorectal, renal cancers and glioblastoma. Although very unlikely, CTCs were reported in glioblastoma (GBM) patients [[15], [16], [17]]. Moreover, in glioblastoma hydrogels were utilized as localized carriers for therapeutic agents after GBM resection or to confine/focalize residual GBM cells to enhance the efficacy of subsequent localized interventions [18,19]. In ovarian, lung cancer and glioblastoma CLG was more accurate than the conventional, size dependent ScreenCell method [15]. To deeper investigate on the CLG mechanism of cancer cell trapping, CLG-recovered lung, ovarian and glioblastoma human cancer cells were analyzed to assess their metastasis-initiating capability. Moreover, to evaluate lung colonization, GFP- Lewis lung cells (GFP-LLC) were injected in the tail vein of immunocompetent mice carrying empty gel (EG) or CLG to assess gels interference in CXCR4+ CTCs lung colonization. Cancer and immune cells were longitudinally evaluated within lungs and gel niches. Results CLG recovered A549, H460 and IGROV-1 cells express high CXCR4 and acquire 3D growth capability CXCR4 and EpCAM were respectively overexpressed and decreased in CLG derived H460 and A549 as compared to parental cells (Fig. 1A). In CLG-IGROV-1 cells, both CXCR4 and EpCAM were overexpressed as reported to facilitate cell adhesion and proliferation [20], while CLG-U87 overexpressed EpCAM as compared to parental cells (Fig. 1A). U87 cells are reported not to express EpCAM [21,22] while EpCAM induction in glioma enhanced cell proliferation and self-renewal capability [23]. Recovered cells were also assessed for sphere forming capability. As shown in Fig. 1B, CLG derived H460-, A549-and U87- CLG developed larger spheres compared to parental while IGROV-1-CLG acquired sphere-forming capability in contrast with the parental cell line. CLG-derived cells express differential CXCR4/EpCAM and develop larger 3D spheres compared to parental. (A) CXCR4 and EpCAM flow cytometry on WT-, EG- and CLG-derived H460, A549, IGROV-1 and U87 cells. Data in bar graphs represent mean ± SEM of at least three independent experiments. A two-tailed t-test was used to assess significance. *P < 0·05, **P < 0·01 ***P < 0·001; (B). CLG-derived cells H460-, A549-, IGROV-1-, and U87-WT/EG/CLG were evaluated for sphere-forming capability for 72 h. Representative images of individual spheroids captured via optical inverted microscope at 10× objective on day 3. Data in graphs represent mean ± SD of at least two independent experiments. Two-tailed t-test was used to assess significance. *P < 0·05, **P < 0·01 ***P < 0·001. CLG recovered cells display transcriptional reprogramming toward proliferative and invasion pathways RNA Sequencing was performed on parental (WT), -empty gel (EG) and – CLG-H460, -IGROV-1 and -U87 cells (Fig. 2). Overall, CLG-derived cells were enriched for genes involved in cell invasion/migration, cell cycle promotion, apoptosis inhibition and stemness according to specific cell origin. H460-CLG overexpressed (>2.5 fold) ASPM, KIF-14 and CCNB1 genes respectively associated with stemness, migration and cell-cycle progression [[24], [25], [26]] and downregulated genes involved in cell cycle-inhibition and pro-apoptotic function (DLC1, ISG15, TP53INP1) [[27], [28], [29]]. IGROV-1-CLG upregulated genes related to stem cell functions, rearrangement of extracellular matrix and migration/invasion (POU5F1/OCT4, FN1 and CXCR4) [1,30,31] and downregulated CCN5 and PSMB9, which respectively suppress epithelial to mesenchymal transition and favor MHC-I CD8+ presentation [32,33]. U87-CLG glioblastoma cells significantly upregulated the very specific NGFR, NTS, AQP1 and CMKLR1 genes, associated with cell proliferation, migration/invasion and metastasis [[34], [35], [36], [37]] while MME, BASP1 and NERG1 genes, with tumor-suppressive function, were significantly downregulated [[38], [39], [40]]. In U87-CLG, GLUL is upregulated indicating “glutamine addiction” typical of advanced glioblastoma. Functional enrichment analyses indicated activation of pathways associated with cell cycle, p53 signaling, DNA repair, apoptosis, MAPK/TOR signaling and immune processes in H460 and IGROV-1 (Fig. 2D and E). In contrast, U87-CLG cell showed enrichment in glutamine metabolism, angiogenesis, transport/homeostasis and neuronal differentiation (Fig. 2D and E). No genes were commonly regulated across the three CLG-derived cell lines consistent with their different epithelial (H460 and IGROV-1) and glial/astrocytic (U87) origins [[41], [42], [43]]. Notably, 109 genes were shared between CLG-H460 and -IGROV-1, 5 between U87 and H460, and 7 between CLG-U87 and -IGROV-1 (Venn diagram, Fig. 3A). In CLG-IGROV-1 and -H460, genes associated with cell-cycle promotion, apoptosis inhibition and migration/invasion (ANLN, FOXM1, FAM83D, RAD54L, KIF20A and KIF14) were overexpressed (Fig. 2C and D and 3B). Thus, CLG enriches cell-line specific transcriptional programs that converge on proliferative and invasive pathways underling their higher metastatic potential. Differentially expressed genes (DEGs) and functional enrichment analysis in CLG-versus EG-derived cells. (A) Principal component (PC) analysis was performed using the normalized values of gene expression values. Clustering of WT- (blue dots), EG- (red dots) and CLG- (green dots) form H460, IGROV-1, and U87 cells was shown according to WT, EG or CLG status. (B) Volcano plots of differentially expressed genes in CLG- H460, -IGROV-1, and -U87 cells in pairwise comparisons versus EG-H460, -IGROV-1, and -U87 respectively, resulting in 569 for H460, 2481 for IGROV-1 and 55 for U87. Volcano plots show log2 fold change on x-axis and -log10 of p-values on y-axis, blue points highlight DEGs at adjusted p-value minor than 0.05. (C) Heatmaps of DEGs in CLG-versus EG-derived H460 (orange), IGROV-1 (purple), U87 (green), and in the genes shared by H460 and IGROV-1 (light blue). (D) GO Chord plots of selected DEGs in CLG-derived H460, IGROV-1 and U87 cells. The genes are linked to their assigned GO_BP pathway via colored ribbons. Observed logFC is displayed in descending intensity below to the selected genes from red (higher in CLG-derived cells) to blue (lower in CLG-derived cells). (E) Dotplots showing selected significant terms of KEGG (green) and Reactome (orange) pathway analysis displaying significantly altered expression patterns (p-adj < 0.05). The color of the dot designates the significance of the change in expression. The size of the dot signifies the number of enriched genes. The X-axis represents the gene ratio (the percentage of enriched genes among total genes in a pathway) and the Y-axis shows the name of the pathway. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.) Commonly regulated genes between CLG/EG cells. (A) Comparison of DEGs from EG/CLG-derived H460, IGROV-1 and U87 cells Venn diagrams. (B) Dotplot showing selected significant terms of KEGG (green) and Reactome (orange) pathway analysis derived from the 109 DEGs shared between H460 and IGROV-1 displaying significantly altered expression patterns (p-adj < 0.05). (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.) CLGs are infiltrated by immune and stromal cells in tumor-naïve, immunocompromised mice The effect of CLG/EG on the recruitment of innate immune and stromal cells, was evaluated in tumor naïve SCID mice in the absence of confounding interferences (neoplastic cells, adaptive immune cells). Two weeks post-injection, gels were recovered and analyzed for immune and stromal cell infiltration. CLGs were highly infiltrated by CD45+ immune cells (NK cells, macrophages and monocytes) as compared to EG (Fig. 4A–D). In addition, CLGs were infiltrated of stromal cells as fibroblast (CD90 expressing) and endothelial-associated/pericytes (PDGFRβ expressing, Fig. 4E and F). Thus, CLG is colonized by immune and stromal cells independently from cancer cells, setting a comfortable environment for circulating cancer cells. CLG in tumor naïve mice are infiltrated by immune and stromal cells. (A) CD45+ immune cells infiltrating EG (n = 2) and CLG (n = 3) two weeks after subcutaneous implantation in naïve SCID mice, flow cytometry. (B) CD45− IHC staining in the above evaluated EGs and CLGs. (C) Frequency of immune cell subsets within EG and CLG, expressed as percentage of gated CD45+ cells shown in (A). (D) Relative CXCR4 expression across the immune cell subsets identified in (B). (E) CD90 and PDGFRβ in cells infiltrating EG and CLG, flow cytometry. (F) PDGFRβ+-IHC in CLGs/EGs. CLGs reduce spontaneous lung metastatic spread in a NSCLC PDX-SCID mice A PDX model was generated from a NSCLC adenocarcinoma (LT710) subcutaneously expanded in SCID mice. When tumors reached an average volume of ∼400 mm3, EG or CLG were injected in the contralateral flank of the tumor. Gels and lungs were collected two weeks after injection, and human disseminated tumor cells (DTCs) and CTCs were identified by flow cytometry on dissociated lungs and gels respectively (DTC/CTCs: viable cells, negative for murine MHC-I). PDX-bearing mice lungs, analyzed through flow cytometry and IHC, displayed less DTC in CLG carrying mice as compared to control and EG-injected mice (Fig. 5A and B). In addition, CLG more efficiently recruited CTCs (Fig. 5 C). CLGs attract CD45+ immune cells more than EGs (Fig. 5D). In particular, we observed a trend toward high recruitment of neutrophils and monocytes, which may contribute to the establishment of a permissive niche for metastatic growth. These data indicate that CLG can effectively affect spontaneous metastatic dissemination. CLGs reduce spontaneous lung metastatic spread in a NSCLC PDX-SCID mice. (A) % of lung DTCs (viable cells, negative for murine MHC-I) in mice bearing LT710 PDX model w or w/o gel (empty gel or CLG). N = 2 independent exp, n = 6 mice/group, flow cytometry. (B) Metastatic foci in lungs analyzed in A, CK-IHC. (C) % of CLG/EG-CTCs trapped cells. Flow cytometry of cells defined as vital 7ADD- and mouse MHC negative cells. N = 2 independent exp, n = 6 replicates CLG and n = 4 replicates EG. (D) % of CLG/EG-CTCs trapped immune cells. % CD45+ immune cells within CD45+ gate. N = 2 independent exp, n = 6 replicates CLG and n = 3 replicates EG. CLGs reduce LLC-GFP lung cancer colonization in immunocompetent mice To evaluate the impact of CLG on lung colonization, EG or CLG were subcutaneously injected in C57BL6 mice and, 5 days later, GFP-Lewis lung cancer cells (GFP-LLC) were inoculated into the tail vein. Gels, lungs, spleen, liver and blood were recovered before cancer cell inoculation (T0), at 4 h (T4h), 10 (T10) and 21 days (T21) after LLC-GFP injection. Four hours post cell injection, LLC-GFP cells were detected in lungs of CTR (1.78 ± 0.4%), EG (0.44 ± 0.16%, p = 0.0274 vs CTR) and CLG (0.15 ± 0.01%, p = 0.0093 vs CTR) carrying mice (Fig. 6A). After 10 days, again lung GFP-LLC cells were more represented in CTR mice lungs (CTR 8.08 ± 0.58%, EG 4.8 ± 0.29%, CLG 3.81 ± 0.83%, Fig. 6A). After 21 days the GFP-LLC frequency decreased, although lungs from CTR group displayed the highest percentage of cells (CTR lungs 1.34 ± 0.36, CLG lungs 0.49 ± 0.05, p = 0.0309, Fig. 6A). Thus, as reported [14], CLG implantation reduces lung cancer cell seeding and further colonization. GFP-LLC cells infiltrated also EG and CLG gels. More cells were detected in CLG gels (T4h 0.41 ± 0.28%, T10 6.88 ± 1.42%, T21 0.31 ± 0.18%) as compared to EG gels (T4h 0.34 ± 0.31%, T10 1.88 ± 1.01%, T21 0.076 ± 0.007%, Fig. 6B). Cancer cells reduction at T21 may be ascribed to GFP immune activity as hypothesized by the higher frequency of CD8+ T-cells in lungs and gels [44]. As shown in Supplementary Fig. 1, the % of CD8+ T-cells increased in lungs derived from control, EG and CLG carrying mice peaking at T21. In EG-CLG gels, a main increase in % CD8+T-cells was shown between T4h and T10 and held up to day 21. CLGs reduce LLC-GFP lung cancer colonization in immunocompetent mice. Murine lungs, gels and blood were collected at T0, T4h, T10 and T21. (A) Murine lungs were processed and GFP-LLC cells (CD45-/CD31-/GFP+) identified through Flow Cytometry, (B) EG and CLG gels were HAse digested and GFP-LLC cells (CD45-/CD31-/GFP+) identified through Flow Cytometry. n = 2 EG at T10 time point due to sample technical loss. (C) Blood was collected by retro-orbital bleeding and 50 μl stained with APC anti-mouse CD45; red cells were lysed and samples analyzed in BD Trucount™ tubes allowing absolute count of blood GFP-LLC cells (CD45-/GFP + cells). (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.) Lastly, we looked at LLC-GFP cells in the blood (Fig. 6C). As previously showed [14], the number of circulating LLC-GFP cells was not significantly different in blood from CLG and CTR mice at the assessed time points. CLG/EGs impair lung niches development competing for CXCR4+ circulating tumor cells Immune cells characterizing lung niche were analyzed. Neutrophils (CD45+/Ly6G+) mature (Ly6Ghigh, immunoactive)/immature (Ly6Glow, immunosuppressive) and macrophages (CD45+/Ly6G-/F480+) M1 (CD86+/CD206-, immunoactive)/M2 (CD86-/CD206+, immunosuppressive) were evaluated in the lungs overtime. The gels injection did not affect macrophage and neutrophils infiltration of lung, liver and spleen at T0 (5 days post gels injection, Fig. 7 and Fig. S2). Lung niche composition is different in CLG- LLC mice. Lungs and gels were processed and evaluated for the infiltration of immune cells. Macrophages (CD45+/Ly6G-/F480+) total and M1- (Ly6G-/F480+/CD86+/CD206-) and M2- (Ly6G-/F480+/CD86-/CD206+) macrophages and Neutrophils (CD45+/Ly6G+) Mature- (Ly6Ghigh) and Immature- (Ly6Glow) were assessed overtime (A) TSNE dimensional reduction and pie charts are colored by cell types. Total Macrophages (yellow), M1-(violet), M2-(brown) macrophages, Ly6Ghigh (blue) and Ly6Glow (red)–neutrophils were shown. (B) Representative density plots showing total (left panel) and M1/M2 (right panel) macrophages at T4h (left side) and T10 days (right side) in lungs from CTR, EG and CLG mice. (C) Representative density plots showing total (left panel) and Ly6G high/Ly6G low (right panel) neutrophils at T10 (left side) and T21 (right side) days in lungs from CTR, EG and CLG mice. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.) Four hours post cancer cells inoculation, CTR lungs were infiltrated by macrophages, mainly M1, and neutrophils. After 10 days, neutrophils increased (Ly6Ghigh and Ly6Glow), while the total macrophages decreased, but polarized toward M2-macrophages. At 21 days, CTR lungs were predominantly infiltrated by neutrophils, mostly immature (Ly6Glow). Interestingly, in lungs from CTR mice wider variations in macrophages and neutrophils were registered (Fig. 7A–C). Finally, at 21 days, the immature neutrophils were significantly higher in CTR compared to CLG lungs (% Ly6Glow/total neutrophils in CTR: 92.76 ± 0.96 vs CLG: 70.76 ± 8.22 p = 0.024, Fig. 7A–C). In summary, lungs immune-microenvironment changed accordingly in CTR and EG/CLG with higher magnitude in CTR lungs. Macrophages increase was sustained by M1-macrophages while at ten days M1 decreased in favor of M2. Macrophages reduction at ten days was accompanied by increased neutrophils Ly6Ghigh/Ly6Glow. At twenty-one days, lungs were mostly infiltrated by neutrophils Ly6Glow. These difference may rely on the higher number of GFP-LLC cells in lungs (Fig. 6A). CLGs reproduce lung niche microenvironment CLG/EG cellular infiltration was evaluated. After 5 days of gel implantation (T0), EG and CLG gels appeared comparably infiltrated by neutrophils Ly6Ghigh, as lungs at T0, and macrophages (Fig. 8A). At 4 h, CLGs were infiltrated by macrophages, with M2 being prevalent over M1. At 10 days neutrophils expanded and macrophages reduced. At 21 days, immature neutrophils mainly infiltrated the gels. EG and CLG comparison showed difference in macrophages after 4 h and 10 days and neutrophils after 21 days. Specifically, after 4 h more macrophages were retrieved into CLG as compared to EG (Fig. 8A) with M1 in EG and M2 in CLG (Fig. 8A and B). Likely, CXCL12 released by CLG attracts CXCR4+ macrophages (Fig. 8B). At ten days macrophages reduced in both gels with CXCR4+ macrophages and M2 (% M2/total macrophages in EG: 9,94 ± 0.56 vs CLG: 16.53 ± 1.29 p = 0.024; Fig. 8A and B) higher in CLG compared to EG. Twenty-one days post-cell injection (T21), EG and CLG were predominantly infiltrated by neutrophils mainly Ly6Glow in CLG (% Ly6Glow/total neutrophils in EG: 53.4 ± 6.67 vs CLG: 72.5 ± 1.3, p = 0.041 Fig. 8A–C). CLGs reproduce immunosuppressive artificial microenvironment. EG and CLG gels were collected at T0, T4h, T10 and T21. (A) Flow cytometry analysis as above described for lungs. (B) Representative density plots showing total (left panel), CXCR4+ (middle) and M1/M2 (right panel) macrophages at T4h (left side) and T10 days (right side) in EG/CLG gels (C) Representative density plots showing total (left panel) and Ly6Ghigh/Ly6Glow (right panel) neutrophils at T10 (left side) and T21 (right side) days in EG/CLG gels. Thus, over time gels became increasingly infiltrated by neutrophils, along with captured LLC cells. Once GFP-LLC cancer cells are recruited within CLG, the CXCL12-loaded-hydrogel creates lung mimicking immune-suppressive microenvironment (high M2-macrophages and Ly6Glow neutrophils). GFP-LLC cells may attract CXCR4+macrophages then polarized toward a M2 phenotype, which in turn leads to dysfunctional neutrophils (prevalently Ly6Glow compared to EG). Discussion In this study, the CXCL12-hydrogel (CLG) trapped circulating CXCR4+ lung cancer cells reproducing an attracting environment for seeding. In vitro, CLG infiltrating lung and ovarian cancer cells express CXCR4 along with 3D proliferation capability while CLG glioblastoma cells express EpCAM, absent in the parental cells, as well as 3D proliferation capability. CLG derived lung and ovarian cancer cells whole transcriptome revealed overexpression of genes of migration, cell proliferation, stemness/mesenchymal transition (ASPM, KIF-14, CCNB1, POU5F1/OCT4, FN1, FOXM1, FAM83D and CXCR4) and reduction of genes related to cell death/apoptosis (DLC1, ISG15, TP53INP1, CCN5 and PSMB9) [1,[24], [25], [26], [27], [28], [29], [30], [31], [32], [33]]. Upregulated genes in lung and ovarian CLG-derived cells (ANLN, FOXM1, RAD54L and KIF20A) were recently associated to anoikis resistance in pancreatic CTCs [4]. U87-CLG significantly upregulated the very specific NGFR, NTS, AQP1 and CMKLR1 genes, associated with cell proliferation, migration/invasion and metastasis [[34], [35], [36], [37]] while MME, BASP1 and NERG1, associated with tumor-suppressive function, were significantly downregulated [[38], [39], [40]]. In addition, the advanced glioblastoma, “glutamine addiction” GLUL gene, is overexpressed in U87-CLG. Although glioblastoma shows mainly local recurrence [45] the interest for glioblastoma relied on the dismal prognosis but also on the possibility to utilize hydrogels [19]. CXCL12-hydrogels might carry therapeutic agents after GBM resection or might confine/focalize residual GBM cells [18,46]. CLG/EGs gels, subcutaneously injected in tumor naïve mice were infiltrated by immune and stromal cells with CLG being more infiltrated by NK cells, macrophages, monocytes and CD90+, PDGFRβ+ fibroblasts. In tumor carrying mice, subcutaneous EG/CLG gels attracted circulating lung cancer cells reducing lung colonization. In immunocompetent mice, the longitudinal kinetic approach across four time points (T0, T4h, T10, T21) showed that EG consistently entraps less cancer cells as compared to CLG. Unfortunately, the evaluation at ten days (T10) relied only on two evaluable EG gels due to a technical loss, which inherently constrains statistical power. Nevertheless, given the tight intra-group clustering (Fig. 6B), the T10 time point fits into a kinetic trend in which cell homing to CLG is inversely related to lung metastatic burden. Upon cancer cell seeding, lungs were early infiltrated by M1 macrophages and, after ten days, lungs displayed M2 macrophages and expansion of neutrophils. After 21 days, immature neutrophils were the most represented cell population in agreement with recently reported evidence [47]. Corresponding gels resembled lungs niche. EG/CLG niche were infiltrated by mature neutrophils before cancer cell injection, then M1 infiltrated EG and M2 infiltrated CLG, while the neutrophils were mainly unaffected. At 10 days, an increase in neutrophils and a reduction in macrophages was detected in EG/CLG gels with CLG showing a relative prevalence of M2 macrophages over M1. At 21days, immature neutrophils mostly infiltrated CLG resembling lung microenvironment. Macrophages are crucial in the metastatic niche as they interact with cancer cells promoting cells adhesion, extravasation and early colonization [48,49]. In absence of macrophages, cancer cells disseminated to the lung dropped significantly and tumor doubling time increased [50]. Murine lung cancer secretes locally SEMA7A that promote M2 polarization activating the integrinbeta1 receptor on macrophages [51]. Moreover, macrophages generate characteristic “microtracks” in the extracellular matrix through which cancer cells move supporting cancer cell migration/invasion [52]. Macrophages affect neutrophils function and phenotype, which in turn can actively contributed to immunesuppression. At micro-metastatic stage, macrophage-derived Il-1β drives pulmonary Ly6Ghigh Neutrophils Extracellular Trap-induced apoptosis (NETosis) of CD8+ T cells reducing them in the lung microenvironment and favoring breast cancer initial metastatic growth [47]. Moreover, a direct correlation described M2-macrophages polarizing neutrophils toward immaturity and inflammatory phenotypes (high Anxa1, Acvrl1, and Lta4h expression) through extracellular vesicles [53]. According to our and others finding, first mature Ly6Ghigh accumulate in the lung niche inducing CD8+ T cell apoptosis through NET formation, then Ly6Glow immature neutrophils accumulated in the later stage of lung metastasis. Ly6Glow cells display myeloid-derived suppressor cell (MDSC)-related gene expression signatures (as CD84 and CD274) and potent ability to suppress CD8+ T cell proliferation and function [47]. This evidence underscore the role of macrophages and neutrophils in lung and in gel describing CLG as comparable/alternative site for metastatic cancer cells. Although cancer cell traps were previously described [46,[54], [55], [56], [57], [58]] scaffold selection is critical. Natural scaffolds (e.g., polypeptides, poly-saccharides, and tissue extra cellular matrix components as hyaluronic acid) offer high biocompatibility, bioavailability and degrade overtime not requiring active removal [59]. Synthetic biomaterials, such as poly(ε-caprolactone) (PCL), polyglycolic acid (PGA) and polylactic acid (PLA) polyesters, are resistant to degradation ensuring long lasting cancer cell entrapment, tunable in term of mechanical strength, degradability and biocompatibility [60]. In melanoma, breast and colorectal syngeneic mice models, subcutaneous injection of cancer cells mixed to decellularized basement membrane of porcine urinary bladder (UBM) delayed the growth of subcutaneous tumor. Although macrophages from UBM mainly express M2 markers, deeper gene expression revealed M1 up-regulated genes suggesting that the binary M1/M2 model may not be sufficient to predict antitumoral immunity [54]. In accordance, we showed cancer cell reduction in CLG gels despite the M2 related immunosuppressive microenvironment. Previous cancer traps relayed on CXCR4/CXCL12 axis. In syngeneic breast cancer, 4T1 tumor was partially removed and synthetic hydrogel loaded with doxorubicine, CXCR4 antagonist and anti-PD-L1 was injected at surgical incision. The gel gradually degrades exerting local and abscopal effect on lung metastatic niche. Interestingly, the treatment generates long-term anti-tumor memory resistant to tumor rechallenge [58]. Both models required surgical implantation that limits the clinical translation. M-TRAP, a synthetic scaffold with type I collagen coating loaded with cancer-derived exosomes, was evaluated in ovarian cancer patients (MTRAP-2016-01). First, in vivo model of peritoneal implantation of the M-Trap generated an inflammatory reaction mimicking an artificial pre-metastatic niche with significantly improved mice survival [61]. In the MTRAP-2016-01 study, 23 stage IIIc/IV high-grade serous ovarian cancer patients underwent implantation of three M-Trap devices in the peritoneum of the abdominal cavity after debulking surgery and then removed by laparoscopy after 18-months or at disease progression. Due to surgical complexity, intraperitoneal adhesions developed and the following epithelialization limited the devices benefits [62]. Here we showed that CXCL12 loading in commercially available hyaluronic hydrogel recruits CXCR4+ circulating tumor cells, favored by CLG cellular infiltration. Specifically, CLG attracts tail vein injected CXCR4+LLC-GFP cells thanks to the development of a tumor friendly environment reducing lung colonization. M2 macrophages were rapidly recruited to CLG while neutrophils became later prominent, with the majority of immature neutrophils at 21 days post injection. Interestingly, CLG recovered lung cancer cells significantly overexpressed genes involved in innate immunity (IL1a, CXCL5, CD14 and ISG15) [28,[63], [64], [65]]. In conclusion, the CXCL12-loaded hydrogel is a biocompatible and biodegradable cancer trap that impairs lung cancer cells colonization through the development of a protumoral pseudo niche. This device represents a valid tool to investigate on the number and function of the CXCR4+ CTCs subset. CLG-CTCs detection is currently evaluated in lung, ovarian and breast cancer [15] (TRAP4MET-1). As CLG is physiologically digested over thirty days, it does not need to be removed and has no toxicity in vivo. Thus it is conceivable that subcutaneous/surgical site injection may affect local/distant recurrence. Materials and methods Cell lines Human ovarian IGROV-1 (RRID:CVCL_C8XN), were cultured in Roswell Park Memorial Institute 1640 (RPMI, Microgem), human lung A549 (RRID:CVCL_0023) and H460 (RRID:CVCL_0459), glioblastoma U87 (RRID:CVCL_3428) and murine Lewis lung (LLC) carcinoma cell lines were cultured in Dulbecco's Modified Eagle's Medium (DMEM, Microgem); RPMI and DMEM media were supplemented with 10% FBS, 1% penicillin/streptomycin and 1% L-Glutamine. Cells were maintained in 5% CO2 at 37°C and test proved mycoplasma-free. LLC cells were stably transduced to express Green-Fluorescent-Protein (GFP) using the pLenti-CMV-GFP-2A-Puro-Blank Vector (RRID:Addgene_102984). For GFP lentiviral particles production, the pLENTI-CMV-GFP-2A-Puro plasmid (RRID:Addgene_102984) was co-transfected with pCMVdR8.2 (RRID:Addgene_8455) and pCMV-VSV-G plasmids (RRID:Addgene_8454) into HEK293T cells (RRID:CVCL_0063) by Calcium phosphate. LLC cells were stable infected with lentiviral particles and 8 mg/mL polybrene (Sigma-Aldrich) for 48 h. GFP positive cells were sorted by FACS Aria III cell sorter (BD Bioscience) and characterized for GFP-, CXCR4 and EpCAM (Fig. S3). EG/CLG preparation and cells recovering The commercially available Hyaluronic Acid based gel Belotero Intense® (Merz Pharma) was loaded with CXCL12 (300 ng/ml, R&D System) gently mixed and immediately used. 25,000 cells A549, H460, IGROV-1 and U87 in 150 μl SF media were seeded on 150 μl empty gel (EG) or CXCL12-loaded gel (CLG) and allowed to migrate 16 h. Cells were recovered digesting the gel with 400 μl hyaluronidase (4 unit/ml HAase, Bioindustria) and grown in complete media for 6 days. Three-dimensional spheroids growth IGROV-1, A549, H460, U87 were evaluated for sphere-forming capability. Cancer cells were resuspended at 5 × 103 cells/ml in 5 ml complete media (RPMI supplemented with 10% FBS, 1% penicillin/streptomycin and 1% L-Glutamine for IGROV-1 and DMEM supplemented with 10% FBS, 1% penicillin/streptomycin and 1% L-Glutamine for A549, H460 and U87). A minimum of 20 drops (250 cells/50 μL) were plated on Petri dish lid spaced systematically to prevent drop coalescing and subsequently inverted in a single, fluid motion and placed over the bottom plate to promote hanging drops. To maintain adequate humidity and mitigate droplet evaporation during incubation, the bottom of the petri dishes was filled with 5 mL PBS [66]. Cells were placed in a humidified incubator with 5% CO2 at 37°C and growth for 72 h. The images of individual spheroids were captured via optical inverted microscope (using 10× objective, Carl Zeiss, Germany) on day 3. Spheroid images were analyzed with ImageJ software (RRID:SCR_003070). Flow cytometry All antibodies were incubated for 45 min at +4°C in the dark for surface markers staining. Flow cytometry on human cancer cells was performed using PE Anti-Human CXCR4 antibody clone 12G5 (R&D Systems Cat# FAB170P, RRID:AB_357076) and PE Mouse Anti-Human EpCAM clone HEA-125 (Miltenyi Biotec Cat# 130-091-253, RRID:AB_871665). Stained cells were evaluated with LSRFortessa (BD Biosciences) and data analyzed with FACS-Diva software 8.1 (BD Biosciences). For mice samples LIVE/DEAD™405 nm (ThermoFisher, cat. L34966) was used to define live cells. Macrophages and Neutrophils were stained as follow: LIVE/DEAD™405 nm (ThermoFisher, cat. L34966), APC rat anti-mouse-CD45 (BD Biosciences Cat# 559864, RRID:AB_398672), PE-Cy7 rat anti-mouse-Ly6G (clone 1A8, BD Biosciences Cat# 561236, RRID:AB_10611860), BrillantViolet 421 rat anti-mouse-F480 (clone BM8, BioLegend Cat# 123137, RRID:AB_2563102), FITC rat anti-mouse-CD11b (clone WT5, BD Biosciences Cat# 554982, RRID:AB_395627), PE rat anti-mouse-CXCR4 (clone L276F12, BioLegend Cat# 146506, RRID:AB_2562783), BV650 rat anti-mouse-CD86 (clone GL1, BD Biosciences Cat# 564200, RRID:AB_2738665), PerCP/Cy5.5 rat anti-mouse-CD206 (clone C068C2, BioLegend Cat# 141716, RRID:AB_2561992). Cancer cells and endothelia cells were stained with: LIVE/DEAD™405 nm (ThermoFisher, cat. L34966), APC rat anti-mouse-CD45 (clone 30-F11 (BD Biosciences Cat# 559864, RRID:AB_398672),), BV711 rat anti-mouse-CD31 (clone 390, BD Biosciences Cat# 740690, RRID:AB_2740374), BV786 rat anti-mouse-CD326 (clone G8.8, BD Biosciences Cat# 740958, RRID:AB_2740583). T effector cells: LIVE/DEAD™405 nm (ThermoFisher, cat. L34966), APC rat anti-mouse-CD45 (clone 30-F11, BD Bioscience, cat.5559864 RRID: AB_398672), PerCP/Cy5.5 rat anti-mouse-CD3 (clone 145-2C11, BioLegend Cat# 100328, RRID:AB_893318), BUV737 rat anti-mouse-CD8 (clone 53-6.7, BD Biosciences Cat# 612759, RRID:AB_2870090). All samples were evaluated at LSRFortessa (BD Biosciences) and all data were analyzed using FACS-Diva v8.1 (BD Biosciences) and FlowJo v10 (BD Bioscience, RRID:SCR_008520) software.” In vivo experiments Naïve 4-6 weeks old SCID mice (Charles River Laboratories) Mice were randomly assigned to be subcutaneously injected with EG and CLG. After 2 weeks post injection mice were sacrificed and gels were collected for flow cytometry and immunohistochemistry analysis. For establishment of the NSCLC xenograft models, small tumor fragments were obtained from patient surgical specimens after informed consent. Fragments were implanted s.c. By trocar gauge in one or both flanks of SCID mice, as previously described [67] and PDX were maintained by serial s.c. Passages of fragments from growing tumors into healthy mice. The PDX model used in the study, LT710, were of the adenocarcinoma subtype. To analyze the CTC/DTC in the lungs, single-cell suspensions from lungs were obtained after mechanical and enzymatical disaggregation as previously reported [13]. Single-cell suspensions were stained for 20 min at 4°C in sorting buffer with anti MHC class I-PerCP-eFluor, 710 (clone SF1-1.1, Ca# 46-5957-82, RRID:AB_10735380 eBioscience) to discriminate murine cells, and with the appropriate antibodies. Immunohistochemistry was conducted on Formalin-Fixed Paraffin-Embedded (FFPE) lung tissues were cut in 2.5/3 micron-thick section, dried, de-waxed, rehydrated and unmasked (with Dako PT-link, EnVision™ FLEX Target Retrieval Solution, High/Low pH – 96°C). Sections were incubated with primary antibody Anti-human Cytokeratin according to the manufacturer indications. The primary antibody detection was performed using Ultra Vision detection system-HPR polymer (Thermo Fisher Scientific) and DAB substrate chromogen (Dako). All slides were scanned and acquired with Aperio ScanscopeXT® (Leica Biosystems Aperio) obtaining a high-quality resolution digital image. For experimental metastasis assay forty-eight, six/eight-week-old, female C57BL/6 mice (RRID:IMSR_JAX:000664) were randomly divided into 3 groups: i. 16 CTR, ii. 16 EG and iii. 16 CLG. 100 μl EG or CLG (5 μg/ml CXCL12) gels were injected in the subcutaneous tissue of mice flank and five days later, 1 × 106 murine GFP-LLC cells were inoculated in the tail vein. Mice were sacrificed at T0, before cancer cell inoculation (n = 2 CTR, n = 3 EG and n = 3 CLG), after 4 h (T4h, n = 5 CTR, n = 4 EG and n = 4 CLG), 10 days (T10, n = 3 CTR, n = 3 EG and n = 3 CLG) and 21 days (T21, n = 6 CTR, n = 6 EG and n = 6 CLG) and gels, lungs, spleen, liver and blood were recovered. At T10 one EG gel matrix could not be fully recovered. At T0 gels, lungs, spleen, liver and blood were analyzed while at remaining time points we analyzed gels, lungs and blood while liver and spleen tissues were stored. Recovered gels were digested with hyaluronidase 4 Units/ml (HAase, Bioindustria) in PBS for 3 h at 5% CO2 at 37°C. Murine lungs, spleens and livers were cut into small fragments and minced using curved scissors. The homogenate was cleaned through a cell strainer (Corning cod. 431750), centrifuged twice at +4°C for 10 min and incubated with FcγR Block (Miltenyi Biotec, cod. 130-092-575) according to manufacture; gel recovered cells were then suspended in Stain Buffer (BD Bioscience, cod. 554656) for antibody staining. Additional information and antibodies staining procedure in Supplemental Methods. Flow cytometry-gating strategy in Fig. S4A–B. Before euthanasia, mice were anesthetized with pentobarbital (70 mg/kg) and 500 μl peripheral blood was collected by retro-orbital bleeding using 1 ml syringes with 10 μl fresh EDTA (Spinreact). 50 μl of collected blood samples were stained with APC rat anti-mouse CD45 (clone 30-F11, BD biosciences, cod. 559874) for 45 min in the dark at +4°C; red cells were lysed with BD Pharm Lyse™ (BD Bioscience, cod. 555899) according to manufactures instructions and analyzed in BD Trucount™ tubes (BD Bioscience, 663028) for absolute count of blood circulating LLC-GFP cells (identified as CD45-/GFP + cells). All analyses were performed blinded. All animal studies were performed in compliance with national regulations on the protection of animals used for scientific purposes (Italian decree n. 26 April 2014) and with the principle of the ‘3Rs’ (Replacement, Reduction and Refinement); Italian Ministry of Health permission n.448/2023-PR. Statistical analyses were performed using either a one-way ANOVA or a two-tailed Student's t-test. RNA extraction and real-time PCR RNA was extracted from cell lines with TRIzol Reagent (Invitrogen, Carlsbad, California, USA) following the manufacturer's instructions. Quantitative real-time PCR was performed using SYBR Green PCR Master Mix (Applied Biosystems, Foster, California, USA) and data were collected and quantitatively analyzed on QuantStudio 5 Real-Time PCR System with 2−ΔCt method. Relative mRNA expression was normalized toward GAPDH, GUSB, beta-actin and beta2-microglobulin gene expression. Primer sequences are reported in Supplemental Table S1. The primer pairs were subjected to a specificity checking process through the Primer-BLAST publicly available tool. RNA libraries preparation, sequencing procedures, and data analysis Total RNA was quantified using the Qubit fluorometer (Thermo Fisher) with Qubit RNA HS Assay Kit (Thermo Fisher). Polyadenylated RNA libraries were prepared using QIAseq stranded mRNA Select Kit (Qiagen). The mRNA Enrichment was obtained starting from 1 μg of total RNA using Oligo-dT probes attached to the surface of magnetic beads. Fragmentation was then performed at the reaction temperature of 95°C to obtain an insert size of ∼350 bp. Double-stranded cDNA was achieved by a reaction with RNase H-Reverse Transcriptase in combination with random primers followed by a second-strand synthesis, end repair, and A-addition step. The dual index technology was used to sequence multiple samples, thanks to the ligation of Illumina-compatible adapters. Finally, the last amplification step was performed on the purified libraries using the CleanStart PCR Mix, also able to degrade contaminating material. QIAseq Beads were used for reaction clean ups between protocol steps. The quality control and quantification of libraries were performed using the Agilent 2100 Bioanalyzer instrument and High Sensitivity DNA Kit (Agilent). Each QIAseq Stranded library was diluted to a concentration of 4 nM. Then, the libraries with different sample indexes were pooled in equimolar amounts. The pooled libraries at concentration of 1.6 pM were sequenced on Illumina NGS NextSeq 500 platform using 2×150 bp paired-end mid-output sequencing kit (Illumina). RNAseq FASTQ sample sequences were firstly trimmed with TrimGalore (v. 0.6.10, RRID:SCR_011847) in paired end, and then mapped on HG38 genome (December 2013 assembly) with STAR software (V. 2.7.3a RRID:SCR_004463) and default options [68]. The resulting BAM files were then processed with featureCounts function (RRID:SCR_012919) present into the Rsubread [69] R/Bioconductor package (V. 2.20.0) on Ensembl HG38 GTF (V.102 RRID:SCR_002344) to produce a count matrix of 64252 genes and 18 samples. To remove lowly expressed genes per each tissue, the filterByExpr function present in the edgeR package (v. 4.4.2) [70] resulting in a different amount of expressed genes per each cell line (see Supplemental Table 2). Samples were then divided for cell line and normalized with RUVs procedure present in the RUVSeq R/Bioconductor package (V.1.40.0, RRID:SCR_006263). A different ‘k’ parameter is chosen for each cell line on the best normalization afforded for each case (Cell Line H460 k = 2, IGROV-1 k = 1, U87 k = 1). The principal component analysis (PCA) showed H460, IGROV-1 and U87 cancer cells independent clustering with the first two components explaining ∼92% of variability, suggesting a marked difference in global gene expression profiles among cell lines (Fig. S5). After data normalization clusters are distinct as parental WT, EG and CLG with an explained variability of ∼85% for H460, ∼81% for IGROV-1 and ∼84% for U87. Then, for each cell line, differential expression analysis was performed using the glmQLFit followed by glmQLFTest from the edgeR R/Bioconductor package (V.4.4.2, RRID:SCR_012802). Most relevant genes were confirmed through quantitative Real Time PCR (Fig. S6). Functional enrichment analysis was then performed with gProfiler2 R package (V.0.2.3). The data discussed in this publication have been deposited in NCBI's Gene Expression Omnibus (GEO) (RRID:SCR_005012) [71] and are accessible through GEO Series identifier GSE317323. CRediT authorship contribution statement Luigi Portella: Writing – review & editing, Writing – original draft, Visualization, Methodology, Investigation, Funding acquisition, Formal analysis, Data curation, Conceptualization. Dario Guido Di Febbraro: Visualization, Project administration, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Dario Righelli: Writing – original draft, Software, Methodology, Investigation, Formal analysis, Data curation. Giulia Bertolini: Writing – original draft, Visualization, Validation, Resources, Methodology, Formal analysis, Data curation, Conceptualization. Giuseppe Guardascione: Visualization, Methodology, Investigation, Formal analysis, Data curation. Giuseppina Rea: Visualization, Methodology, Investigation, Formal analysis, Data curation. Crescenzo D'Alterio: Validation, Methodology, Formal analysis. Caterina Ieranò: Validation, Methodology, Investigation, Formal analysis. Sara Santagata: Validation, Methodology, Formal analysis. Anna Maria Trotta: Validation, Methodology, Formal analysis. Marco Gentili: Validation, Resources, Methodology, Formal analysis. Roberta Sulsenti: Validation, Resources, Methodology, Formal analysis. Antonio Luciano: Resources, Methodology. Giuseppe Palma: Resources, Methodology. Francesca Bruzzese: Supervision, Resources, Methodology. Ilaria Assunta Parisi: Resources, Methodology, Formal analysis. Antonella De Luca: Writing – review & editing, Supervision, Resources, Methodology. Luca Roz: Writing – review & editing, Supervision, Resources, Methodology, Conceptualization. Roberto Pacelli: Writing – review & editing, Supervision, Conceptualization. Stefania Scala: Writing – review & editing, Writing – original draft, Visualization, Supervision, Methodology, Investigation, Funding acquisition, Conceptualization. Funding This work was supported by Associazione Italiana per la Ricerca sul Cancro (AIRC IG24746 to SS) and Italian Ministry of Health (Ricerca Corrente 2025-2027 Funds L3/9_25, IRCCS-Pascale to LP). The funding sources had no role in the study design, collection of data, analysis, or interpretation of data, writing of the manuscript or the decision to submit for publication. Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Footnotes Supplementary data to this article can be found online at https://doi.org/10.1016/j.mtbio.2026.103648. Supplementary data The following are the supplementary data to this article: Multimedia component 1 Multimedia component 2 Data availability Data will be made available on request. References Associated Data Supplementary Materials Multimedia component 1 Multimedia component 2 Data Availability Statement Data will be made available on request.