Expression of ID4 protein in breast cancer cells induces reprogramming of tumour-associated macrophages

The SKBR3, MDA-MB-468, HL60 and U937 cell lines were grown at 37 °C with 5% CO₂ and maintained in RPMI medium containing 10% heat-inactivated FBS and penicillin/streptomycin. HL60 and U937 cells were differentiated by treatment with 1,25-dihydroxyvitamin D₃ (VitD3) (Sigma-Aldrich, St. Louis, MO, USA) at a concentration of 250 ng/ml. Monocytic differentiation was assessed by fluorescence-activated cell sorting (FACS) as previously reported [33] using allophycocyanin (APC) anti-human CD11b (BD Biosciences, San Jose, CA, USA), PerCP-Cy5.5 (peridinin chlorophyll protein complex-cyanine 5.5) anti-human CD14 (BD Biosciences) and phycoerythrin-immunoglobulin G1 (PE-IgG1) isotype control (eBioscience Inc., San Diego, CA, USA) antibodies for the evaluation of CD11b-CD14 co-expression as a marker of monocytic differentiation. A minimum of 10,000 events were collected for each sample with a flow cytometer (CyAN ADP; Beckman Coulter Life Sciences, Brea, CA, USA) using Summit 4.3 software (Beckman Coulter Life Sciences) for data acquisition and analysis.

An expression vector containing a hemagglutinin (HA)-tagged ID4 coding sequence [28] or control empty vector was transfected in cancer cells using Lipofectamine 2000 reagent (Thermo Fisher Scientific, Waltham, MA, USA) in ID4 overexpression experiments. RNAiMAX reagent (Thermo Fisher Scientific, Waltham, MA, USA) was used to transfect small interfering RNAs (siRNAs) in BC cells. Sequences of siRNAs directed to ID4 were previously reported [30]. Monocytic cell lines were transfected with plasmids, mimic and locked nucleic acid (LNA) oligonucleotides (Dharmacon, Lafayette, CO, USA) using the TransIT-X2® Dynamic Delivery System (Mirus Bio LLC, Madison, WI, USA) following the manufacturer’s instructions. Full-length cDNA (including 5′-UTR and 3′-UTR) of human GRN (NM_002087.2), cloned in the pCMV6-XL5 plasmid vector, was generously provided by Dr. Peter Nelson.

Mouse bone marrow-derived macrophage precursors were obtained from rodents by flushing the femurs and tibias with 2% FBS in PBS. Differentiation was induced by culturing precursors in colony-stimulating factor 1 (CSF1)-rich conditioned media (CM) derived from L929 fibroblast cell culture. Differentiation was evaluated by FACS analysis using the following antibodies: anti-mouse F4/80 antigen APC (17-4801; eBioscience), Ly-6G (Gr-1) APC (17-5931; eBioscience, San Diego, CA, USA), CD14 PE (12-0141; eBioscience, San Diego, CA, USA) and CD107b (Mac-3) PE (12-5989; eBioscience, San Diego, CA, USA).

Human peripheral blood-derived monocytes were isolated from blood donors using Lymphoprep solution (Axis-Shield, Dundee, UK) followed by isolation of CD14⁺ cells with the Monocyte Isolation Kit II (Miltenyi Biotec, Bergisch Gladbach, Germany). Differentiation was achieved through 1-week culturing in RPMI medium containing recombinant CSF1 (human macrophage colony-stimulating factor, catalogue number 8929SC; Cell Signaling Technology, Danvers, MA, USA).

CM from BC cells were prepared by culturing cells for 24 hours in serum-free RPMI medium. CM were centrifuged to eliminate cell residues before preparation of aliquots and storage at − 80 °C. When si-ID4 BC cells were used to prepare CM, we always collected CM before 48 hours from transfection because of proliferation of cells being delayed after this time point in the si-ID4 condition (Additional file 1: Figure S3).

Migration of mouse bone marrow-derived macrophages in response to SKBR3 cells was evaluated using 3-μm-pore Boyden chambers (Corning Inc., Corning, NY, USA). Infiltration of F4/80⁺ macrophages in Matrigel plugs containing CM from BC MDA-MB-468 cells was evaluated by subcutaneous inoculation of a solution composed of 500 μl of Matrigel (BD Biosciences) and 50 μl of a 10 × concentration of CM. In the negative control, the CM was replaced with serum-free medium. Plugs were recovered at day 7, fixed for 18–24 hours in 4% (vol/vol) buffered formaldehyde, and then processed with paraffin wax. IHC was performed using F4/80 antibody (MA5-16363; Pierce Biotechnology, Rockford, IL, USA). All procedures involving animals and their care were conducted in conformity with institutional guidelines, which are in compliance with national and international standards.

Tumours from 62 patients included in this study were previously described in a study by Novelli et al. [34], which was reviewed and approved by the ethics committee of the Regina Elena National Cancer Institute and contained data for which written informed consent was obtained from all patients. Characteristics of these patients are included in Additional file 2: Table S1. BC specimens for IHC analysis were fixed for 18–24 hours in 4% (vol/vol) buffered formaldehyde and then processed with paraffin wax. Anti-ID4 (MAB4393; EMD Millipore, Billerica, MA, USA), anti-oestrogen receptor (clone 6F11; Novocastra, Florence, Italy), anti-progesterone receptor (anti-PgR, clone 1A6; Novocastra), and anti-HER2 (A0485; Dako, Milan, Italy) were evaluated by IHC in 5-μm-thick paraffin-embedded tissues. Monoclonal antibodies (mAb) directed against ID4 were incubated at a dilution of 1:200 overnight at 4 °C, and anti-ER and anti-PgR mAb and the polyclonal antibody anti-HER2 were incubated for 60 minutes at room temperature. Immunoreactions were revealed by a streptavidin-biotin enhanced immunoperoxidase technique (Super Sensitive MultiLink; BioGenex, Fremont, CA, USA) in an autostainer (Bond III; Leica Biosystems, Wetzlar, Germany). Diaminobenzidine (DAB) was used as a chromogenic substrate. Evaluation of the IHC data was performed independently and in a blinded manner by two investigators (EG and EM).

For immunocytochemistry assay, cells were seeded onto glass coverslips (Paul Marienfeld, Lauda-Königshofen, Germany) in 6-well dishes (Corning Inc.) at 4 × 10⁴ cells/well, cultured with RPMI or CM, and fixed with 4% formaldehyde in PBS for 15 minutes at room temperature. Cells were permeabilized with 0.25% Triton X-100 in PBS for 10 minutes. After washing with PBS, the coverslips were incubated with anti-ID4 antibody diluted in 5% bovine serum albumin (BSA)/PBS for 2 hours at room temperature. Cells were incubated with peroxidase inhibitor before primary antibody incubation. Protein staining was revealed through DAB enzymatic reaction, and nuclei were counterstained with haematoxylin.

For immunofluorescence, cells grown in the presence of RPMI or CM (48 hours), as well as cells transfected with mimic oligonucleotides (48 hours), were concentrated onto microscope slides using cytospin and fixed and permeabilized as already described. Slides were blocked for 30 minutes in 5% BSA/PBS at room temperature and then incubated with an anti-HIF-1A antibody (A300-286A; Bethyl Laboratories, Montgomery, TX, USA) diluted in 5% BSA/PBS for 2 hours at room temperature. Cells were incubated with secondary antibody Alexa Fluor 594 (1:500; Thermo Fisher Scientific) for 45 minutes. Nuclei were stained with DAPI (Thermo Fisher Scientific).

For the Western blot analysis, cells were lysed in radioimmunoprecipitation assay buffer or 8 M urea. The protein concentration was measured using a Bio-Rad protein assay kit (Bio-Rad Laboratories, Hercules, CA, USA). The lysate was mixed with 4 × Laemmli buffer. Total protein extracts were resolved on polyacrylamide gel and then transferred onto nitrocellulose membrane. The following primary antibodies were used: Gapdh (sc-32,233), ID4 (H70) sc-13047, ID4 (B5) sc-365656, HA (12CA5) sc-57592 (Santa Cruz Biotechnology, Dallas, TX, USA); HIF-1A (A300-286A; Bethyl Laboratories); GRN (PA5-29909), EphB2 (PA5-14607), and Mdk (PA5-30601; Thermo Fisher Scientific). Secondary antibody fused with horseradish peroxidase was used for chemiluminescence detection on a UVITEC instrument (Uvitec, Cambridge, UK). VEGFA blocking antibody (AF-293-NA; R&D Systems, Minneapolis, MN, USA) was added to CM and incubated for 30 minutes at room temperature before being used to culture macrophages, following the manufacturer’s instructions.

Four microlitres of CM was mixed with 4 μl of Growth Factor Reduced Matrigel (BD Biosciences) and 0.5 μl of phenol red. The mixture of CM and Matrigel was injected into the perivitelline space of Tg(fli:EGFP) casper zebrafish embryos at 48 hours post-fertilisation. The injection was performed using glass micropipettes with capillaries of 0.75-mm internal diameter. The following parameters were used for the micropipette puller (P-1000; Sutter Instruments, Novato, CA, USA): heat 510, pull 100, velocity 200, time 40, and pressure 500. The parameters of the PicoPump injector (World Precision Instruments, Sarasota, FL, USA) were set to inject 1 nl of CM. Within 24 hours after injection, the neovascular response originating from the developing subintestinal vessels was observed on a fluorescence stereoscope.

Differentiated U937 cells were transfected with siRNAs directed to GRN mRNA or control siRNAs for 8 hours and subsequently cultured with CM from MDA-MB-468 cells. After 72 hours of culture, CM was collected and used to perform tube-formation assays as described by Pruszko et al. [30].

Viability of U937 cells was assessed using the ATPlite assay (PerkinElmer, Waltham, MA, USA) at the indicated time point and according to the manufacturer’s instructions. Differentiated U937 cells (1 × 10⁵ cells), previously transfected with GRN expression vector, were seeded into 96-well plates and cultured for 48 hours in CM from MDA-MB-468 cells. Luminescence was read by using the EnSpire® Multimode Plate Reader (PerkinElmer).

We previously demonstrated that ID4 protein expression is associated with high microvessel density in BC. Mechanistically, ID4 promotes the production of pro-angiogenic cytokines in BC cells, leading to enhanced endothelial cell proliferation and migration [28, 30]. Because the onset of angiogenic switch, identified as the formation of a high-density vessel network, is closely associated with the transition to malignancy and is regulated by infiltrating macrophages in primary mammary tumours [5], we investigated whether ID4 promotes angiogenesis by influencing the behaviour of macrophages. We first evaluated whether any association existed between ID4 protein expression and infiltrating TAMs in human BC by staining a series of 62 TNBCs for ID4 protein and for the widely used macrophage marker CD68 [15, 35]. The choice of TNBC was based on evidence that increased ID4 expression is specific to this subtype, characterised by the absence of oestrogen receptor, PgR and HER2 receptors, and mostly attributable to the BLBC molecular subtype, as reviewed by Baker et al. [23]. Expression levels of ID4 in representative TNBC and BLBC cohorts are shown in Additional file 3: Figure S1. Pathological characteristics of the 62 analysed TNBC cases are included in Additional file 2: Table S1.

In agreement with the literature [18, 28], we observed that ID4 protein was detectable in 75% of the analysed specimens. On the basis of protein expression, we divided the analysed tumours into low expressers (comprising negative tumours and tumours scored as 1+) and high expressers (comprising tumours scored as 2+ and 3+). We observed that high CD68 protein expression was significantly associated with the ID4 high expresser group (P = 0.028) (Fig. 1a). Representative images of TNBC showing high or low protein levels of ID4 and CD68 are shown in Fig. 1b. ID4 and CD68 proteins were not associated with other pathological characteristics in this group of patients.

High levels of ID4 expression have been correlated to decreased survival in TNBC and BLBC [17, 20, 21]. Macrophage infiltration has been correlated to angiogenesis in BC, but the study of its prognostic significance has led to contradictory results, probably because of the existence of various intratumoural macrophage populations with different properties [12].

To evaluate ID4 prognostic power in relation to macrophage infiltration, we interrogated the Kaplan-Meier Plotter database (www.​kmplot.​com) [36], which contains a compendium of studies with gene expression and relative survival data for BLBCs. Interestingly, we observed that high ID4 expression was strongly associated with low probability of DMFS (n = 232) and OS (n = 241), specifically in the group of tumours characterised by high expression of CD68 (and therefore highly infiltrated by macrophages) (Fig. 1c and Additional file 4: Table S2), whereas no association of ID4 with survival was present in the low-CD68 group (Fig. 1d and Additional file 4: Table S2). A similar result was obtained when a macrophage signature comprising a subset of eight widely used markers (CD14, CD105, CD11b, CD68, CD93, CD33, IL-4R and CD163) for the mononuclear phagocyte system [37] was used to identify tumours highly infiltrated by macrophages (Fig. 1e and f and Additional file 5: Table S3). Analysis of gene expression data from The Cancer Genome Atlas (TCGA) cohort of BLBC confirmed that high ID4 expression is associated to low probability of overall survival specifically in the CD68-high and macrophage signature (MacSig)-high groups (Additional file 6: Figure S2a–d). The TCGA cohort allowed us also to assess that ID4 and CD68 do not associate with the clinical variables T, N and G (as observed in the TNBC cohort analysed by IHC and described in the previous paragraph), whereas ID4 significantly associates with mutated TP53 status (Additional file 6: Figure S2e). Moreover, because none of the considered patients from the TCGA cohort received neoadjuvant treatment, we can assert that the observed associations are independent of particular treatment regimens. These results indicated that the combination of ID4 and macrophage markers represents a powerful predictive indicator in BLBC.

On the basis of the observed association between ID4 protein expression and TAMs, we wondered whether ID4 expression in BC cells influences macrophage recruitment. To address this, CD34⁺ progenitors from mouse bone marrow were isolated, differentiated in vitro to macrophages (Fig. 2a), and evaluated for their migratory capacity in response to BC cells with ID4 expression depleted or not (Fig. 2b–c). As shown in Fig. 2c, a lower number of macrophages migrated towards ID4-depleted (si-ID4) BC cells than that for control (si-SCR) cells.

To evaluate if ID4 expression in BC cells influences the recruitment of macrophages in vivo, we performed Matrigel assays. Briefly, Matrigel plugs containing CM from MDA-MB-468 BC cells, transfected with an expression vector for HA-tagged ID4 or an empty vector (Fig. 2d and e), were inoculated subcutaneously in mouse flanks and recovered after 7 days. According to previous reports [38, 39], IHC staining of Matrigel plugs with mouse monocyte/macrophage marker F4/80 showed the presence of F4/80⁺ cells within regions of massive cellular infiltration inside the Matrigel. A higher number of F4/80⁺ cells was observed in plugs containing CM from ID4-overexpressing cells than that in control plugs (Fig. 2f–g).

Because one of the major activities exerted by TAMs is the promotion of angiogenesis, we next analysed whether ID4 expression in BC cells affects the expression of angiogenic genes in macrophages. To this end, we took advantage of a TLDA containing probes for a panel of 94 angiogenesis-related genes. Macrophages obtained from differentiation of HL60 cells [40, 41], cultured with CM from MDA-MB-468 cells transfected with an ID4 expression vector (ID4) or an empty vector (EV), were evaluated along with control macrophages cultured in RPMI medium (Fig. 3a and b). In this experimental setting, we detected 36 expressed genes, 11 of which were modulated in an ID4-dependent manner (1 downregulated and 10 upregulated genes) (Additional file 4: Table S3). The ID4-dependent paracrine induction in macrophages of a subset of these genes, comprising ephrin B2 (EPHB2), midkine (MDK), EDIL3 and GRN, was validated by RT-qPCR (Additional file 7: Figure S4a) and Western blotting (Additional file 7: Figure S4b). We verified that ID4 overexpression did not affect the expression of these genes in MDA-MB-468 cells (Additional file 7: Figure S4a, right panel).

Moreover, using an additional macrophage cell line (U937), we observed that the expression of selected ID4-dependent angiogenesis-related genes (EPHB2, GRN and NRP2) was induced in macrophages cultivated in CM compared with RPMI medium (Fig. 3c); as expected, this induction was impaired when CM was derived from si-ID4 BC cells (Fig. 3c–f). Interestingly, analysis of HIF-1A, a master regulator of angiogenesis, revealed that the expression of this transcription factor in macrophages depends on the level of ID4 expression in BC cells (Fig. 3c, g and h and Additional file 7: Figure S4c). Altogether, these results showed that high ID4 expression in BC cells is associated with the activation of a pro-angiogenic programme in macrophages.

Because the expression of the angiogenesis-related genes in macrophages depends on the expression of ID4 in BC cells, we reasoned that a soluble factor, secreted in an ID4-dependent manner from BC cells, is probably responsible for the observed gene expression reprogramming of macrophages. In this regard, we recently reported that ID4 protein promotes the synthesis of pro-angiogenic isoforms of VEGFA at the expense of the anti-angiogenic ones in BC cells [30]. We then explored whether VEGFA was responsible for the observed effects. We first cultured differentiated U937 cells in CM from VEGFA-depleted (si-VEGFA) or control (si-SCR) BC cells. Analysis of a panel of angiogenesis-related factors evidenced a partial decrease of their expression after VEGFA depletion (Fig. 3I and j). Next, we observed that the addition of VEGFA blocking antibody to the CM from BC cells subsequently used to culture U937 cells partially impaired the induction of this panel of angiogenesis-related factors (Fig. 3k). These results indicate that ID4-dependent gene expression modulation in macrophages is at least in part under the control of VEGFA signalling.

It has been extensively reported that the angiogenic programme is tightly controlled also at the post-transcriptional level by miRNAs in cancer. To explore whether the ID4-dependent reprogramming of macrophages also involved miRNAs, we evaluated the expression of members of the miR-15/107 group, which were previously correlated to angiogenesis in vertebrates and reported to target GRN and HIF-1B [42‐47].

We observed that miR-107, miR-15b and miR-195 are downregulated in macrophages cultured with CM from ID4-overexpressing BC cells (CM ID4) compared with macrophages cultured with CM from BC cells with control empty vector (CM EV) (Additional file 5: Figure S5a). On the contrary, expression of these miRNAs was recovered in the presence of CM from si-ID4 BC cells in two macrophage cell lines (Fig. 4a and b and Additional file 8: Figure S5b–e). We evaluated the expression of miR-96, which exhibits oncogenic activity in BC [48], as a control, and we observed that it shows a trend opposite to that of miR-107 (Fig. 4c). Recovery of miR-107, miR-15b and miR-195 expression was also observed in U937 cells cultured in the presence of CM from VEGFA-depleted BC cells (Additional file 8: Figure S5f), indicating that VEGFA signalling also controls, at least in part, miRNA expression in TAMs.

Time-course analysis of macrophages cultured with CM from BC cells revealed downregulation of these miRNAs (Fig. 4d and e and Additional file 8: Figure S5f). Analysis of pre-miR-107 expression in the same conditions highlighted that decrease of mature miR-107 was accompanied by an accumulation of its precursor (Fig. 4f), suggesting an inhibition of the processing of this miRNA in the presence of CM from BC cells. Altogether, these results indicated that expression of ID4 in BC cells leads to a paracrine downregulation of miR-107, miR-15b and miR-195 in macrophages.

Next, we focused on miR-107, which shows the strongest ID4-dependent paracrine downregulation in macrophages, and evaluated whether it affects the expression of GRN and HIF-1B, two well-established targets [44, 49]. To this end, we inhibited miR-107 in U937 cells by transfecting an LNA oligonucleotide (Fig. 4g). As shown in Fig. 4h, miR-107 inhibition recovered GRN and HIF-1B protein expression, mimicking the effect of si-ID4 BC-derived CM. We also observed induced protein expression of EphB2 and HIF-1A (Fig. 4h), which, as the majority of the angiogenesis-related factors that are activated in an ID4-dependent paracrine manner in macrophages, are predicted to be targeted by the miR-15/107 group members (Additional file 5: Table S3).

To further investigate the relevance of miR-107 downregulation associated with CM, we overexpressed miR-107 using mimic oligonucleotides in macrophages cultured with CM from MDA-MB-468 BC cells (Fig. 4i). As shown in Fig. 4j and k, the forced expression of miR-107 led to decreased GRN mRNA and protein levels. Similar results were observed for HIF-1A (Additional file 8: Figure S5g and h). Our results indicated that the expression of angiogenesis-related genes is strictly controlled by the activity of the ID4-dependent miR-107 in macrophages.

Among the ID4-dependent angiogenesis-related genes upregulated in macrophages, GRN particularly attracted our attention, because this growth factor is specifically expressed in TNBC and BLBC [50] and has recently been correlated to tumour angiogenesis in mesothelioma [51]. In macrophages, GRN has been reported to control cytokine production [32], but its effect on the angiogenic potential of these cells has not been explored yet.

To evaluate the ability of GRN to confer angiogenic potential to macrophages, we performed in vivo angiogenic assays. To this end, a full-length GRN expression vector, containing 5′- and 3′-UTRs, or control EV was transfected in U937-derived macrophages, which were then cultured with RPMI or CM from MDA-MB-468 cells. As shown in Fig. 5a and b, although GRN mRNA expression levels were comparable between RPMI and CM conditions, GRN protein overexpression was observed only in macrophages cultured with CM. This result further underlined that GRN expression in macrophages is strictly controlled at the translational level and that its protein expression is obtained only in the presence of CM, possibly as a consequence of miR-107 downregulation (as shown in Fig. 4d and e).

We then evaluated the angiogenic potential of macrophages transfected with GRN or EV and cultured in CM from MDA-MB-468 cells using transgenic zebrafish embryos expressing enhanced green fluorescent protein in the entire vasculature. Specifically, Matrigel plugs containing CM from each experimental condition were injected into the perivitelline space of zebrafish embryos, and neovascular response originating from the developing subintestinal plexus was evaluated. Injection of Matrigel plugs containing PBS alone or PBS supplemented with recombinant VEGFA (rhVEGFA) was used as a negative and positive control, respectively. As shown in Fig. 5c and d, we observed a greater number of embryos presenting two or more spikes sprouting from subintestinal plexus in the GRN overexpression condition compared with that in the EV condition. Accordingly, in the GRN overexpression condition, we also observed a reduced number of embryos showing one or no spikes (Fig. 5c and d). No effects on macrophage viability and differentiation were observed in the presence of GRN overexpression (Additional file 9: Figure S6).

Next, we evaluated the effect of GRN depletion on macrophage angiogenic potential. To this end, we transfected siRNAs directed to GRN or control siRNAs (si-SCR) in U937-derived macrophages, which were then cultured with CM from MDA-MB-468 cells (Fig. 5e). CM from each experimental condition was then evaluated in tube-formation assays involving the growth of endothelial cells. Conditions with RPMI medium supplemented or not with rhVEGFA were included as positive and negative controls, respectively. As shown in Fig. 5f and g, GRN depletion led to the significant decrease of tube-formation potential.