Fig 2: Enpp1 and NET in human TNBC. A,In silico analysis of the prognostic value of gene-expression levels using the Kaplan–Meier plotter (www.kmplot.com) of ENPP1 (Affy ID: 229088_at) and HP (Affy ID: 206697_s_at) in TNBC. Differences between groups were evaluated using the log-rank test. B, Left, quantification of ENPP1 immunostaining in a TMA of human breast cancer primary and relapsed breast tumors. Nonparametric Wilcoxon paired sample test was used for comparison. Right, representative images. C, Representative images of NET evaluation by multispectral imaging with an automated quantitative pathology imaging system (Vectra Polaris) in primary nontreated breast cancer tumors and their paired relapsed TNBC tumors from formalin-fixed, paraffin-embedded tissue sections of a TMA. Representative composite images immunostained with CD15 (clone Carb-3, blue), MPO (clone C8/144B, red), anticitrullinated histone (H3Cit; green, citrulline R2 + R8 + R17), cytokeratin (wide spectrum screening, cyan), and DAPI (gray). NET are costained with CD15, MPO, and H3Cit. NET are visualized as white structures in the full composite color images. Square is magnified in the bottom plots. Top: scale bar, 100 µm. Bottom: scale bar, 50 µm. D, Segmented fluorescence images for CD15, MPO, and H3Cit for magnified squares shown in C, showing coimmune localization of three markers, indicating bona fide NET. Scale bar, 50 µm. E, Quantification of the NET area of the multispectral images obtained. Nonparametric Wilcoxon paired sample test was used for comparison. F, Schematic of the mechanism of LRF at the cellular and molecular levels. Top, primary breast tumors release CTC. After surgery, the RTB constitutes a fertile soil for CTC. Enpp1hi CTC are more competent to engraft and efficiently colonize the RTB. Enpp1hi cells promote the intrusion of PMN-MDSC and strong immunosuppressed microenvironment in the recurrent tumor. PMN-MDSC create NET, which also abolish the immune attack. Bottom, at the molecular level, ENPP1 hydrolyzes ATP into AMP and pyrophosphate (PPi). Similarly, CD39 in the microenvironment hydrolyzes ATP to AMP. AMP is hydrolyzed to adenosine (Ado) by CD73 in the adenosinergic pathway (CD39–CD73). The accumulation of AMP in the TME induces the secretion of HP through P1 purinergic receptors. Hp and other components act as chemoattractant factors for PMN-MDSC that infiltrate and contribute to the incipient tumor colonization by an immunosuppressive TME from early stages of tumor formation. This also leads to immune landscape remodeling by impairing NK and CD8+ T-cell infiltration. Secretion of HP also induces NET formation, which further promotes an immunosuppressive microenvironment. Radiation obliterates the early engraftment of CTC. ENPP1 inhibition decreases the extracellular AMP and Ado levels and decreases HP with an enhancement of an immunocompetent microenvironment. ENPP1 blockade radiosensitizes the tumor, triggers the infiltration of CD8 and NK, and decreases the infiltration of Tregs and myeloid suppressor cells. In this scenario, tumor cells are less competent to colonize the RTB. Subsequent fractionated irradiation contributes to eradicate engrafted tumor cells, preventing locoregional failure, but does not prevent the subsequent homing of CTC post-IR.

Fig 3: Principal pathways of extracellular ATP hydrolysis. ATP is hydrolyzed to ADP and then ADP is hydrolyzed to AMP by ENTPD1, 2, 3, and 8. ATP can be degraded directly to AMP by alkaline phosphatase (ALPL/TNAP) and ENPP1 and 3. AMP, in turn, is degraded to ADO by NT5E and ALPL/TNAP.

Fig 4: CTC are enriched in Enpp1. A, Outline of the strategy used to enrich the LRF phenotype with and without IR (15 Gy) by isolation and expansion of recurrent cells and reinoculation in another cohort of mice. Cy1NoRx: cycle 1 with no radiation. Cy1Rx: cycle 1 with radiation. B, Left, relapse-free survival rate between ANV5 control and two different second cycle irradiated cell subpopulations (720Cy2Rx and 701Cy2Rx). Log-rank test was used. Right, tumor volume at resection. No differences in tumor margins between groups were detected (n = 20 mice in ANV5 and 720Cy2Rx and n = 15 mice in 701Cy2Rx). C, Hierarchical cluster diagram of top differentially expressed genes in control ANV5 cells and two second cycle IR cell subpopulations. D, Gene-expression levels of the indicated genes evaluated by RT-qPCR in cell lines derived from different in vivo cycles (Cy1, Cy2, and Cy3) in IR (Rx) and non-IR conditions (no Rx). E, Western blot analysis of Enpp1 levels in cells isolated in different cycles. Quantification of relative expression levels is shown. F, Scheme showing the CTC isolation in the blood, called “CTC-out,” derived from the primary tumor, whereas CTC reaching the RTB were called “CTC-in.” CTC-out cells isolated from blood were orthotopically implanted in another cohort of mice and reisolated to obtain CTC-out cells of cycle 2 (Cy2). To obtain CTC-in, second cycle labeled CTC-out were i.c. inoculated (50,000 cells) after tumor resection composed of unlabeled tumor cells. Different CTC-in cells were isolated for 4T1 cells. For ANV5, CTC-out were technically impossible to isolate in this model, and so we i.c. inoculated labeled ANV5-GFP+ cells (100,000 cells) into tumor-resected mice (originated from unlabeled ANV5 cells) to obtain CTC-in. G, In the 4T1 model, gene-expression levels of Enpp1 were evaluated by RT-qPCR in a panel of CTC isolated in several rounds of CTC-out and in CTC-in. In the ANV5 model, Enpp1 expression levels were evaluated in a panel of CTC-in. H, Western blot analysis of Enpp1 levels in CTC-in cells isolated in different cycles for ANV5 and CTC-in, and CTC-out for 4T1 cells. I, Left, scheme for the isolation of GFP+ cells either ANV5 (mock-transduced) or Enpp1-overexpressing cells (OE-Enpp1), i.c. inoculated one day post-resection of the GFP- tumor. The sixth day after i.c inoculation, mammary gland was disaggregated, minced to single-cell suspension, and cells were incubated in the presence of selection antibiotic puromycin for 2 weeks. Middle, quantification of all resistant single cell–derived colonies for each mouse. Mann–Whitney test was used for comparison. Bottom, representative image of single cell–derived colonies stained with crystal violet. Right, cell growth kinetics in vitro using MTS assay. Data were normalized with absorbance at day 0 and represent mean ± SD of six replicates. Experiment was performed three times with similar results. *, P < 0.05; **, P < 0.01.

Fig 5: Functional role of the ENPP1i in immune remodeling. A, Tumor volume at the day of resection after orthotopic inoculation of OE-ENPP1 cells. Mice (n = 8 mice/group) were treated daily with Enpp1i (CM3163) or vehicle. Flow-cytometric quantification of tumor-infiltrating immune subpopulations. B, LRF-free survival of two groups of mice (n = 25 mice/group) previously inoculated with 720Cy2Rx cells were treated daily (arrows) after tumor resection with the Enpp1i (CM3163) alone or with anti-CD8, anti-NK1.1–depleting antibodies (200 ug of antibodies 3 times per week), or vehicle. Right, tumors were resected at the indicated sizes. Tumor recurrence was detected by palpation and verified by histologic analysis. No differences in tumor margins between groups were detected. Bottom left, representative images of immunostaining of tumors (scale bar, 100 µm). Bottom right, quantification of IHC images of Gr1+ cells in all relapsed tumors using a blinded automated system based on ImageJ. Student t test. C, Tumor volume of tumor derived from orthotopic implantation of control (shGFP-720Cy2Rx) or shEnpp1a cells treated with the STING inhibitor (arrows; C-176, 5 mg/kg i.p. daily; n = 15 mice per group). Two-way ANOVA was applied. Mean ± SEM are represented. D, LRF-free survival after the surgical resection of tumors derived from shEnpp1 cells. After tumor resection, animals (n = 15 per group) were treated daily with the CD39 inhibitor (POM-1, 5 mg/kg) or vehicle control (arrows). Right, tumors were resected at the indicated sizes. E, LRF-free survival after tumor resection from OE-Enpp1–implanted cells. Mice (n = 15 mice per group) were treated with FD (4 doses of 6.2 Gy) irradiation on two consecutive days after surgery (sg), or in combination with the Enpp1i daily (arrows) at 50 mg/kg subcutaneously or sham-operated and treated with vehicle (control). Right, tumor volume at the day of surgery in each group. No differences in tumor margins between groups were detected. Log-rank test was used in Kaplan–Meier curves. *, P < 0.05; **, P < 0.01; ***, P < 0.0001.