Anaphylatoxin C5a creates a favorable microenvironment for lung cancer progression

Cell lines and primary cultures

Mouse Lewis lung carcinoma (3LL) and human lung cancer cell lines were obtained from the American Type Culture Collection (ATCC). Cells were grown in RPMI supplemented with 2 mM glutamine, 10% Fetalclone (Thermo), 100 U/ml penicillin, and 100 μg/ml streptomycin (Invitrogen). Immortalized normal human bronchial epithelial cells (HBEC) were established by introducing mouse Cdk4 and hTERT into normal HBECs (21, 22). The genetic alterations introduced in these cells are not sufficient to confer a full malignant phenotype. Two HBEC lines from two different donors were used: HBEC3-KT and HBEC10-KT. These cell lines were kindly provided by Dr JD Minna (University of Texas Southwestern Medical Center, Dallas). Cells were maintained in keratinocyte serum free medium supplemented with human recombinant EGF and bovine pituitary extract (Gibco). The immortalized BEAS-2B cells were obtained from ATCC and cultured in DMEM:Ham's F-12 (1:1) supplemented with human recombinant EGF, insulin, transferrin, epinephrine, triiodothyronine, retinoic acid, hydrocortisone, gentamicin sulfate/amphotericin-B and bovine pituitary extract (Lonza). Human umbilical vein endothelial cells (HUVEC) were obtained from Lonza and maintained in EGM-2MV: EBM-2 medium supplemented with human recombinant EGF, VEGF, FGF, R3-IGF-1, hydrocortisone, ascorbic acid, heparin, gentamicin sulfate/amphotericin-B and 2% fetal bovine serum (Lonza).

Patients

Plasma samples from fifty patients with advanced lung cancer (adenocarcinomas or squamous cell carcinomas at stages IIIB and IV) and from fifty healthy people (matched by age, sex and smoking exposure) were obtained from EDTA-treated blood at the Clínica Universidad de Navarra (Pamplona, Spain). The study protocol was approved by the Institutional Ethical Committee, and all patients gave informed consent. Lung tumors were classified according to the World Health Organization 2004 classification (23). C5a concentration was quantified in the plasma samples using the commercial kit Human Complement Component C5a (DuoSet, R&D), following the manufacturer's instructions.

Peptides

The cyclic hexapeptide AcF(OP(D)ChaWR) [acetylated phenylalanine–(ornithine-proline-(D)cyclohexylalanine-tryptophan-arginine)], a C5aR antagonist, was used in in vitro and in vivo experiments to block C5aR activity. The cyclic hexapeptide AcF(OP(D)ChaA(D)R) [acetylated phenylalanine–(ornithineproline-(D)cyclohexylalanine-alanine-(D)arginine] was used as control. Both peptides were derived from the C-terminal decapeptide sequence of C5a, which was modified to produce either antagonists or peptides with no activity (7). Both peptides were synthesized by PeptideSynthetics (Fareham, UK).

C5 deposition

Cells were detached from culture dishes with trypsin/EDTA, washed and resuspended in veronal buffer with magnesium (1.8 mM barbital, 3.1 mM barbituric acid, 141 mM NaCl, 0.5 mM MgCl₂, pH 7.4). Human NSCLC and immortalized bronchial cells (2×10⁵) were incubated with 20% normal human serum (NHS) or heat inactivated-NHS (HI-NHS) for 30 minutes at 37°C. Mouse 3LL cells were incubated with 40% normal mouse serum (NMS) or heat inactivated-NMS (HI-NMS). In human cells, C5 deposition was assessed using a polyclonal goat anti-human C5b (Quidel) in binding buffer and fluorescein isothiocyanate-conjugated rabbit anti-goat IgG (Serotec). In mouse cells, C5 deposition was assessed using a monoclonal antibody against mouse-C5 (clone Bb5.1) and Alexa Fluor-conjugated goat anti-mouse IgG (Invitrogen). After the addition of 7AAD (BD Biosciences), cells were washed and analyzed by flow cytometry in a FACScalibur with CellQuest software (BD Biosciences). Data were collected as mean fluorescence intensity.

C5a production in serum-free conditions

Cells were cultured for 24 hours in 100 mm² plates (2×10⁶ cells/10 ml) or 24-well plates (10⁵ cells/0.5 ml). Afterwards, cells were washed three times with PBS and 10 or 0.5 ml of RPMI 1640 basal medium was added, respectively. Conditioned media were collected after 48 hours. C5a concentration was assessed by ELISA using the commercial kit Human Complement Component C5a (DuoSet, R&D). Local C5a production by A549 cells was also evaluated in the presence of different inhibitors: compstatin, inhibitor of complement C3; antitrombin III (Sigma) plus heparin (Rovi), inhibitor of the coagulation system; lepirudin (Schering AG), inhibitor of thrombin; GM6001 (Millipore), metalloproteinase inhibitor; E-64 (Sigma), cysteine protease inhibitor; and serine protease inhibitors aprotinin (Trasylol, Bayer), soybean trypsin inhibitor (SBTI, Sigma) and TLCK (Sigma). In the case of compstatin, an improved 3 analog with the following sequence D-Tyr-Ile-[Cys-Sar-Val-Asp-Trp-Ala-His-Trp(1-Me)-Gln-Arg-Cys]-N-Me-Ile, and a corresponding control peptide D-Tyr-Ile-[Cys-Val-Trp(1-Me)-Gln-Asp-Trp-Sar-Ala-His-Arg-Cys]-N-Me-Ile were synthesized as described (24).

mRNA expression of complement proteins

Total RNA was purified using the RNeasy Kit (Qiagen), following the manufacturer's instructions. Reverse transcription was done using 2 μg of total RNA, SuperScript III (Invitrogen) and random hexamers (Applied Biosystems). Expression of human complement components C1q, C1r, C1s, C2, C4, C3, C5, factor B, factor D and properdin was assessed by real time PCR using commercial TaqMan probes (Applied Biosystems). PCR conditions were as follows: 2 minutes at 50°C, 10 minutes at 95°C and 40 cycles (15 seconds at 95°C and 1 minute at 60°C). Results were expressed for each gene as the relative gene expression normalized to the IPO8 mRNA expression (25).

Expression of C5aR in HUVECs and 3LL cells was studied by conventional PCR using human GAPDH and mouse HPRT as control genes, respectively. Primers used in conventional PCR are listed in Table I. PCR reactions were performed using 1 μL of cDNA, 10 μL of 2x Master Mix buffer (Promega) and 500 nM of sense and antisense primers (Sigma) in a total volume of 20 μL. For human C5aR and GAPDH, the reactions were performed according to the following conditions: 92°C for 2 minutes, followed by 30 cycles (30 seconds at 92°C, 30 seconds at 51°C, and 30 seconds at 72°C). For mouse C5aR and HPRT the following steps were performed: 92°C for 2 minutes, followed by 35 cycles (1 minute at 92°C, 1 minute at 61°C, 1 minute at 72°C). PCR products were separated on agarose gels and stained with ethidium bromide (Invitrogen).

Syngeneic mouse model of lung cancer

Female C57BL/6J mice of 8-12 weeks were used (Harlan Laboratories). The experimental protocols were reviewed and approved by the Institutional Animal Care and Use Committee of the University of Navarra. Murine 3LL cells (2.5×10⁴) were resuspended in 100 μl of PBS and mixed with 100 μl of growth factor reduced Matrigel (BD Biosciences). Cells were injected subcutaneously in the right flanks. For the pharmacological blockade of C5aR, mice received a subcutaneous injection of C5aR antagonist AcF(OP(D)ChaWR) every 3 days at a dose of 1 mg per kg body weight in 200 μl of PBS, starting 24 hours before the inoculation of tumor cells. The control peptide AcF(OP(D)ChaA(D)R) was used using the same administration regimen. For the blockade of C5 activity, mice received intraperitoneally 1 mg of the anti-mouse C5 monoclonal antibody Bb5.1 or an IgG control isotype in 100 μl of PBS every 3 days, starting 4 hours before the inoculation of cells. In all cases, tumors were measured every 3 days and the tumor volumes were calculated by the formula (L × W²)/2, where L is the length and W is the width.

Expression of C5aR by flow cytometry

C5aR expression on HUVEC and 3LL cells was detected by flow cytometry. Cells (2×10⁵) were incubated for 30 minutes at 4°C with rat anti-mouse C5aR (20/70, BioLegend) or mouse anti-human C5aR (S5/1, Serotec) in PBS 1% BSA 0.1% azide. After two washes, cells were incubated for 30 minutes at 4°C with the secondary antibodies: fluorescein isothiocyanate-conjugated rabbit anti-rat IgG (Serotec) or Alexa Fluor-conjugated goat anti-mouse IgG (Invitrogen). Finally, cells were washed twice and analyzed by flow cytometry on a FACSCalibur using CellQuest Pro software (BD Biosciences). Data were expressed as mean fluorescence intensity.

Cell proliferation

The MTT assay was used to study the proliferation of HUVEC and 3LL cell lines treated with human or murine C5a, respectively. For endothelial cells, 1.5×10³ HUVECs were seeded in assay medium (EGM-2MV without rhVEGF and rhFGF) in a 96-well plate. After 24 hours of incubation, dilutions of human C5a (Calbiochem) from 1 ng/ml to 2 μg/ml were added. Wells without cells were used as background. One hundred percent growth was established using cells grown in assay medium alone. bFGF at 10 ng/ml (Sigma) was used as positive control. All treatments were performed in sextuplicate. After 96 hours of incubation, 10 μL of 5 mg/ml MTT solution (Sigma) was added to each well. Plates were incubated for 4 hours at 37°C and then, 100 μL of solubilization buffer (10% SDS, 50% N-Ndimethylformamide, pH 4.7) was added. Plates were left on the cell incubator overnight. One day later, absorbance was measured at 540 nm with a reference wavelength of 690 nm in a Sunrise microplate reader (Tecan). For 3LL cells, 2×10³ cells were seeded in RPMI complemented with 0.5 % Fetalclone. Next day, dilutions of murine C5a (R&D) from 0.01 to 500 ng/ml were added in basal RPMI. Wells without cells were used as background. One hundred percent growth was assessed in wells with no C5a added. Cells were incubated for 48 hours at 37°C. The assay was developed as indicated above.

To measure the percentage of proliferation, the mean value in each condition was calculated by the formula:

HUVEC migration

For the evaluation of the chemotactic effect of C5a in HUVECs, 7.5×10⁴ cells were cultured in 24-well plates in EBM-2 basal medium on the upper chamber of transwells with 8 μm pore size (Corning-Costar). C5a dilutions ranging from 500 to 0.25 ng/ml were added. EBM-2 medium and EBM-2 medium with 10% fetal bovine serum were used as negative and positive controls, respectively. Cells were allowed to migrate for 48 hours in a cell culture incubator. Afterwards, cells from the upper chambers were removed by swiping with cotton swabs and cells adhered to the lower side of transwells were fixed with 4% formaldehyde (Panreac) and stained with 0.5% crystal-violet (Sigma). Migrated cells on the lower surface of the filter were visualized and photographed using an inverted microscope (Leyca DM ILM). At least five fields at 10x magnification were taken and counted using ImageJ software (NIH).

HUVEC tube formation

C5a ability to promote angiogenesis in vitro was tested using the tube formation assay. This method is widely utilized to assess the capacity of a molecule to inhibit or stimulate angiogenesis (26). Before experiments, cells were left in EGM-2MV/EBM-2 basal medium (1:1) for 8 hours. Afterwards, cells were washed and kept in EBM-2 basal medium overnight. Next day, 48-well plates were coated with 150 μl of growth factor reduced Matrigel (BD Biosciences) and allowed to gel for 2 hours at 37°C. Matrigel is a commercial basement membrane matrix derived from the Engelbreth-Holm-Swarm mouse sarcoma (27). Cells (2.5×10⁴) in EBM-2 basal medium were seeded on the gel and mixed with C5a at 10 ng/ml final dilution. Cells in EBM-2 medium alone were used as control. At least five repetitions were done per condition. Plates were incubated at 37°C for 5 hours. Photographs of five random optical fields in the central area of wells were taken at 10x magnification with a digital camera coupled to a phase microscope (Leica DM IL LED). Total tube length was calculated using ImageJ software (NIH).

In vivo matrigel PLUG assay

The effect of C5aR signaling in angiogenesis was determined in vivo by the Matrigel plug assay. Female mice at 10-12 weeks of age were subcutaneously injected with 0.5 mL of Matrigel containing 2.5×10⁴ 3LL cells and 500 ng/mL of C5aR antagonist [AcF(OP(D)ChaWR)] or control peptide [AcF(OP(D)ChaA(D)R)]. Peptides were also injected 24 hours before and after the inoculation of plugs, and one additional injection was performed 3 days later. Six days after inoculation, 25 mg/ml of fluorescein isothiocyanate-dextran (Sigma) in 200 μL of PBS was injected i.v. in the tail vein. Mice were sacrificed 15 minutes later. The plugs were extracted and fixed in 4% formaldehyde for 24 hours. Photographs of areas with vessels were taken using a Zeiss Axiovert confocal microscope.

Analysis of splenocyte subpopulations

Spleens from tumor-bearing mice were mechanically disaggregated in 5 ml of PBS using cell strainers with 70 μm pore size (BD Biosciences). Single-cell suspensions (approx. 5×10⁵ splenocytes) were preincubated with a monoclonal antibody to mouse CD16-CD32 (Fc block; 2.4G2; BD Pharmingen) and then incubated with a labeled primary antibody against mouse CD4 (RM4-5), CD8a (53-6.7), CD11b (M1/70) or Ly6C/Gr1 (AL-2) (all from BD Biosciences) diluted in FACS buffer (PBS, 0.24 μM EDTA and 5% Fetalclone). Staining of mouse regulatory T cells was performed using a kit from eBioscience according to the manufacturer's instructions.

Expression of immune molecules within the tumors

Portions of approx. 0.1 cm³ (a sphere with 0.6 cm diameter) were cut from the edge of tumors, frozen in dry ice, and maintained at -80°C until the extraction procedure. Total RNA was extracted and retrotranscribed to cDNA. Expression of molecules related to the immune response was performed by real-time PCR using specific primers for each gene (Table II) as previously described (28).

Statistical analysis

Significant differences were evaluated using the Student's t test, except in the case of the immune cells and molecules, for which the differences in expression were analyzed by the Mann-Whitney test. Data were analyzed using GraphPad Prism 5 (GradPad Software). p values < 0.05 were considered statistically significant.

Lung cancer cell lines produce C5a in vitro

The capacity of lung cancer cells to activate complement C5 was evaluated in seven human lung cancer cell lines and three non-fully transformed bronchial epithelial cell lines using normal human serum as the source of complement components. The amount of cell-derived C5a could not be directly quantified due to the presence of C5a in normal human serum. Alternatively, C5 deposition on cell membranes was evaluated by flow cytometry. C5 deposition was higher in lung cancer cell lines than in non-malignant bronchial epithelial cells (Fig. 1A). In fact, all lung cancer cell lines showed a significant increase in C5 deposition after incubation with normal human serum. However, no significant increase in C5 deposition was observed in the non-malignant bronchial epithelial cells.

Previous studies have demonstrated that some cell types are able to activate complement and release C5a in the absence of serum (29, 30). Therefore, the next objective was to evaluate the capacity of lung cancer cells to locally generate C5a in serum-free conditions. Previously, the expression of C5 mRNA in these cells was tested using real time PCR. All malignant and non-fully transformed pulmonary cell lines expressed C5 mRNA, although the levels were appreciably higher in some lung cancer cell lines (Supplementary Fig. S1). To analyze the production of C5a, lung malignant and non-malignant cells were cultured in serum-free medium for 48 hours. Conditioned media were collected and C5a levels were determined. All malignant cell lines, except H727, produced detectable C5a levels (Fig. 1B). Conversely, C5a was almost undetectable in non-malignant bronchial epithelial cells. For A549 and HBEC 3KT cells, a time-course was performed from 12 to 96 hours (Fig. 1C). In A549, C5a levels gradually increased from the beginning to the end of the experiment. However, in HBEC 3KT cells, the levels of C5a did not increased over time.

The alternative complement pathway has been associated with a local production of C5a in some cell types (29). The contribution of this pathway to the production of C5a by lung cancer cells in the absence of serum was investigated. First, the expression of the main components of the alternative pathway (C3, factor B, factor D and properdin) was examined by real time PCR (Supplementary Fig. S2). No apparent association was observed between the expression of these complement components and the levels of locally produced C5a. In fact, the expression of properdin was undetectable in most of the cell lines, and factor B was not expressed by any lung cancer cell line, except H727, which was precisely the lung cancer cell line that did not produce C5a. Next, the potential contribution of the classical complement pathway was assessed. Expression of the main classical complement factors (C1q, C1r, C1s, C2 and C4) was analyzed by real time PCR. No relation was observed between these elements of the classical pathway and the local production of C5a. In particular, C1q expression was not found in any of the cell lines analyzed (Supplementary Fig. S3). To definitively exclude the contribution of C3 convertases, the production of C5a was evaluated in the presence of compstatin, a molecule that prevents the activation of C3 by C3 convertases (31). No decrease in C5a levels was observed using compstatin up to 10 μM (Fig. 1D). C5a can be generated in a complement independent manner by different proteases (32). Under serum-free conditions, the production of C5a by A549 cells was significantly reduced in the presence of antithrombin III and heparin (Fig. 1D). The effect of other protease inhibitors was also studied (Fig. 1D). The production of C5a was not affected by the presence of the thrombin inhibitor lepirudin, the metalloproteinase inhibitor GM6001, the cysteine protease inhibitor E64, or the serine protease inhibitor aprotinin. On the other hand, SBTI and TLCK, inhibitors of trypsin-like serine proteases, significantly reduced the local production of C5a. At the concentrations used, these inhibitors did not affect cell proliferation, as determined by MMT assays (data not shown).

Anaphylatoxin C5a levels are increased in plasma of NSCLC patients

Generation of C5a by tumor cells may be followed by its systemic diffusion. Therefore, it could be speculated that C5a could be detected in peripheral blood from lung cancer patients. The concentration of C5a in plasma from fifty patients diagnosed with NSCLC was determined and compared with the concentration of this protein in fifty samples from control individuals matched by age, sex and smoking exposure (Fig. 2). Cancer patients showed significantly higher C5a plasma levels than healthy donors (27.76 ± 11.53 vs. 14.08 ± 8.24 ng/ml; p<0.001). No statistical differences were found between adenocarcinomas and squamous cell carcinomas, and between stage IIIB and IV (data not shown).

Blockade of C5aR slows tumor growth in vivo

The influence of C5a on lung tumor progression in vivo was studied using a syngeneic mouse model of lung cancer. In this model, 3LL cells were injected subcutaneously in the right flack of C57BL/6J mice. C5a activity was blocked by the administration of the cyclic hexapeptide AcF(OP(D)ChaWR), a C5aR antagonist. Prior to the use of this in vivo model, the capacity of 3LL cells to deposit C5 when incubated with mouse serum in vitro was confirmed (Fig. 3A). Then, 3LL cells were injected in the right flank of mice treated with the C5aR antagonist, or with the control cyclic hexapeptide AcF(OP(D)ChaA(D)R). Blockade of C5aR significantly decreased tumor growth (Fig. 3B). To further support this result, treatment with an anti-C5 monoclonal antibody (Bb5.1) also slowed down tumor growth, although in this case the results did not reach statistical significance (Supplementary Fig. S4). Finally, the addition of C5a to 3LL cells did not modify their in vitro proliferation (Fig. 3C), suggesting that the in vivo effect of C5a on tumor growth is mediated by other mechanisms.

C5a promotes angiogenesis

To elucidate the mechanisms by which C5a/C5aR supports tumor growth in vivo, the role of C5a in angiogenesis was studied. First, the effect of C5a in HUVEC proliferation and migration was evaluated. The expression of C5aR in HUVECs, both at the mRNA and protein levels, was verified (Fig. 4A-B). Proliferation assays were performed in HUVECs cultured for 96 hours in the presence of a wide range of C5a concentrations. No effect of C5a on HUVEC proliferation was observed (Fig. 4C). The capacity of C5a to induce HUVEC migration was investigated using the transwell chamber assay. HUVECs were plated on transwells and cultured during 48 hours with different concentrations of C5a. This factor induced HUVEC chemotaxis with a typical bell-shaped dose-response curve (Fig. 4D). A significant increase in migration, compared to the untreated cells, was observed within 0.125 and 5 ng/ml of C5a (p_0.125=0.015; p_0.25<0.001; p_0.5=0.012; p₅<0.001). The effect of C5a on the formation of tube-like stuctures on Matrigel was also tested. HUVECs were plated on gelled Matrigel in EBM-2 medium. After five hours of incubation, HUVECs treated with 10 ng/ml C5a showed more numerous tube-like structures than cells in basal medium (p=0.001) (Fig. 4E-F).

A Matrigel plug assay was performed to evaluate the effect of C5a on angiogenesis in vivo. Matrigel plugs mixed with 3LL cells and the C5aR antagonist or the control peptide were injected in the flanks of C57BL/6J mice. Peptides were injected throughout the experiment as explained in Materials and Methods. Six days after inoculation, FITC-conjugated dextran was injected i.v. and animals were sacrificed within 10-15 minutes. Mice treated with the C5aR antagonist showed less microvessels within their plugs than those treated with the control peptide (Fig. 5A).

In explanted tumors from the 3LL syngeneic mouse, the expression of the angiogenic factors VEGF and bFGF was analyzed (Fig. 5B). There were no differences in VEGF mRNA levels. However, there was a significantly lower expression of bFGF mRNA in tumors extracts from animals treated with the C5aR antagonist (p=0.023). On the other hand, there were no differences in tumor vascular density (CD31 staining) between mice treated with the C5a antagonist and those treated with the control peptide (data not shown). In tumors from animals treated with the anti-C5 Bb5.1 monoclonal antibody, a reduction in the expression of bFGF mRNA, close to statistical significance, was observed (p=0.089; Supplementary Fig. S4).

C5aR blockade reduces immunosuppression

C5a has been proved to be a key factor in the immune regulation of tumors, mainly by the recruitment of myeloid-derived suppressor cells (MDSCs) (11). The effect of C5aR blockade in the immune response to the lung tumors was assessed. For that purpose, splenocyte subpopulations in 3LL tumor-bearing mice treated with the C5aR antagonist or the control peptide were analyzed. There were no differences in the percentages of CD4⁺, CD8⁺ and regulatory T cells; however, total MDSCs were significantly reduced when C5aR was blocked (p=0.028; Fig. 6). Both monocytic and granulocytic MDSCs were reduced, although the differences were statistically significant only in the subpopulation of granulocytic MDSCs (p=0.095 and p=0.015, respectively). These results suggest that C5a is able to induce an immunosuppressive microenvironment that can be reversed when the C5a receptor is blocked. To confirm this observation, an evaluation of the expression of several immune-related molecules in the tumor microenvironment was carried out. Total RNA was extracted from explanted tumors of animals treated with the C5aR antagonist or the control peptide. Real time PCR was performed for twenty two molecules. Some of them could not be detected (IL2, IL4, FOXP3, IFN-γ, IL12p35 and OX40L), and others did not show any change between the two treatment groups, namely IDO, TNF-α, PD1, CCL17, CCL22, GITR, TGF-β, MCP1, granzyme A, granzyme B and perforin (data not shown). However, the expression of ARG1, CTLA4, IL6, IL10, LAG3 and PDL1 was significantly decreased in tumors from animals treated with the C5aR antagonist (Fig. 7). Most of these molecules are commonly associated with an immunosuppressive state. Blockade of C5 with the Bb5.1 monoclonal antibody confirmed the downregulation of some immune molecules within the tumors (Supplementary Fig. S4). These results confirm that C5a is implicated in the generation of an immunosuppressive tumor microenvironment.