Cancer produces a variety of collateral effects in patients beyond the malignancy itself, including threats to distal organ functions. However, the basis for such effects, associated with either primary or metastatic tumors, are generally poorly understood. In this study, we show how heart and kidney vascular function is impaired by neutrophils that accumulate in those tissues as a result of tumor formation in two different transgenic mouse models of cancer (RIP1-Tag2 model of insulinoma and MMTV-PyMT model of breast cancer). Neutrophil depletion by systemic administration of an anti-Gr1 antibody improved vascular perfusion and prevented vascular leakage in kidney vessels. We also observed the accumulation of platelet–neutrophil complexes, a signature of neutrophil extracellular traps (NET), in the kidneys of tumor-bearing mice that were completely absent from healthy nontumor-bearing littermates. NET accumulation in the vasculature was associated with upregulation of the proinflammatory adhesion molecules ICAM-1, VCAM-1, and E-selectin, as well as the proinflammatory cytokines IL1β, IL6, and the chemokine CXCL1. Administering DNase I to dissolve NETs, which have a high DNA content, restored perfusion in the kidney and heart to levels seen in nontumor-bearing mice, and also prevented vessel leakage in the blood vasculature of these organs. Taken together, our findings strongly suggest that NETs mediate the negative collateral effects of tumors on distal organs, acting to impair vascular function, and to heighten inflammation at these sites. Cancer Res; 75(13); 2653–62. ©2015 AACR.
Cancer evolves from a local disease where one or a few cells have circumvented the normal checkpoint controls and gained potential to expand in an uncontrolled manner, to a disease with systemic effects such as thrombosis, metastasis, and organ failure in the affected individual. The underlying mechanisms behind these systemic, and often fatal, consequences following growth of a primary tumor are still under intense investigation. It is clear that tumors “hijack” normal cellular functions and physiologic processes that provide a growth or survival advantage. Both the coagulation and innate immune systems can contribute to malignant development and the examples are numerous. Tissue factor (TF) is normally expressed in subendothelial spaces, and therefore only exposed to circulating blood cells after injury to the blood vessel wall. However, both tumor cells and the tumor endothelium can start to express TF, leading to thrombin generation, and hence platelet activation and coagulation in the tumor microenvironment (1). Recruitment and activation of platelets in the tumor microenvironment can promote tumor angiogenesis, growth, and dissemination by several distinct mechanisms (2, 3). Tumor-promoting properties of the M2 subtype of macrophages have been documented. In contrast with the M1, or tumor-killing macrophage, the M2 macrophages represent a major component of the tumor stroma and can promote tumorigenesis via their immunosuppressive and angiogenesis stimulating capacity (4). Similar to macrophages, subpopulations of neutrophils with either tumor-promoting or tumor-inhibitory properties have been identified (5).
Another interesting observation in this context, connecting the coagulation and innate immune systems, is the fact that neutrophil extracellular traps (NET) have been found to play a role in malignant progression (6, 7). NETs were identified in 2004 (8) and are extracellular networks that primarily consist of DNA released from neutrophils together with antimicrobial peptides and proteases derived from neutrophil granules. These NETs trap and kill bacteria and were identified as a novel mechanism by which the innate immune system efficiently protects us from infections (8). Formation of NETs, or NETosis, is critically dependent on interaction between platelets and neutrophils (9, 10). In addition to their role in microbial defense, it has during the past years become clear that NETs also form in other situations than bacterial infection, for instance, during inflammatory disease (11, 12). It was recently shown that NETs promote thrombus formation, thereby providing a link between inflammation and thrombosis (13). NETs have also been implicated in cancer-associated thrombosis and it has been shown that G-CSF secreted from the primary tumor can predispose neutrophils to NETosis (14). NET-like structures have also been found inside tumors (15) and hypothesized to play a tumor-promoting role. In a recent study by Cools-Lartigue and colleagues (16), it was described that NETs can sequester circulating tumor cells and promote metastasis (16), highlighting infection-induced NETosis as a risk factor for dissemination of a preexisting tumor.
Studies of tumor-induced effects on distant organs have primarily focused on tissues that represent metastatic sites. Interestingly, tumor-driven formation of both pro- and antimetastatic niches has been reported, which commonly involve recruitment of various types of leukocytes or bone marrow–derived cells to the metastatic site (17–20). For example, VEGFR1-positive bone marrow–derived hematopoietic progenitor cells were reported to prepare a premetastatic niche before the arrival of metastasizing tumor cells (18, 21), thus exhibiting a prometastatic function. Neutrophils were also recently shown to accumulate in the lungs before the arrival of metastatic cells in a mouse model of breast cancer. In contrast with the progenitor cells reported to promote metastasis, these so called tumor-entrained neutrophils instead counteract metastatic seeding by generating H2O2 and CCL2 (20). Surprisingly little is, however, known about the situation in organs that are not targets for metastasis or affected by the primary tumor. A large proportion of cancer-related deaths are, however, caused by thrombosis and general organ failure (22–24). As an example, up to 50% of multiple myeloma patients have renal failure at diagnosis (25), and acute renal failure (ARF) constitutes a major source of morbidity and mortality in cancer patients (26). For this reason, we set out to analyze tumor-induced systemic effects on distant organs, which are not directly affected by growing tumor cells. For obvious reasons human biopsy material from such tissues are rare and mouse models, therefore, become important tools for such investigations. Using two different orthotopic and spontaneously metastasizing tumor models, we analyzed the presence of blood-derived cells in organs distant from the primary tumor that do not represent metastatic sites. We found a significant increase in the number of neutrophils in heart and kidneys of tumor-bearing mice compared with healthy individuals. Moreover, vascular function was significantly impaired in these organs, a situation that was normalized upon neutrophil depletion. Platelet–neutrophil complexes, indicative of NETs, were found in kidneys from individuals with cancer. DNase treatment, which dissolves NETs due to their content of extracellular DNA, or anti–G-CSF treatment, completely restored kidney vascular function to levels seen in healthy mice. Collectively, our data strongly suggest that the impaired peripheral vessel function in individuals with cancer is due to NETosis and may contribute to cancer-induced organ failure.
Materials and Methods
Animal work was approved by the local ethics committee (C315/10, C127/13, and C77/13) and performed according to the United Kingdom Coordinating Committee on Cancer Research guidelines for the welfare of animals in experimental neoplasia. RIP1-Tag2 (genetic background C57Bl6) and MMTV-PyMT mice (genetic background FVB/n) were used for the study: Littermates lacking the transgenes, that is, healthy C57Bl6 and FVB/n mice, were used as controls. RIP1-Tag2–positive mice received drinking water supplied with 5% sucrose from ten weeks of age, to relieve hypoglycemia. For euthanization, mice were either anesthetized by i.p. injection of 2% avertin and perfused with 10 mL of 1× PBS (pH 7.4) and 10 mL of 2% paraformaldehyde (for analysis by immunostaining), or sacrificed by cervical dislocation (for RNA expression analysis). RIP1-Tag2/C57BL6 mice were analyzed at 14 weeks of age, and MMTV-PyMT mice were analyzed at 13 weeks of age. For evaluation of tumor burden, RIP1-Tag2 tumor volume was calculated using the formula [(π/6) × width2 × length]. Mammary tumors from MMTV-PyMT mice were weighed immediately after dissection.
Neutrophil depletion was performed by i.p. injection of an anti-Ly6G antibody (Bio-X-Cell, clone RB6-8C5). Treatment was initiated in 12-week-old MMTV-PyMT mice and 13-week-old RIP1-Tag2 mice, with totally four injections of 100 μg antibody/mouse given every second day. The mice were sacrificed, as described above, one day after the last injection.
DNase1 (10 U in 100 μL 0.9% NaCl; Fermentas, EN0521) was administered by i.p. injection once daily during 3 consecutive days. Treatment was initiated in 12.5 weeks old MMTV-PyMT mice and 13.5 weeks old RIP1-Tag2 mice. The mice were sacrificed, as described above, 4 to 5 hours after the last injection. Tumors and kidneys were dissected and prepared for analysis as described above.
G-CSF was inhibited in vivo by i.p. injection of anti–G-CSF antibody (R&D Systems; MAB414). Treatment was initiated in 12-week-old MMTV-PyMT mice and 13-week-old RIP1-Tag2 mice. The mice were injected daily with 10 μg antibody/mouse, with a total of seven injections, and sacrificed the day after the last injection.
Perfusion of the peripheral vasculature was analyzed by lectin injection (27, 28). Briefly, FITC-conjugated lectin (Vector Laboratories; Lycopersicon Esculentum, FL-1171) was administered by retroorbital injection of 150 μL solution (0.5 mg/mL) into anesthetized mice. The solution was allowed to circulate for 2 to 3 minutes before the mice were sacrificed as described above. Tissue cryosections (kidney and heart) from mice injected with FITC-conjugated lectin were immunostained for CD31 to quantify the total vessel area. The degree of perfusion was determined by the ratio FITC-lectin–positive area/CD31-positive area, analyzed with ImageJ, revealing the proportion of vessels with functional blood flow.
Cryosections fixed in ice-cold methanol were used for all immunostainings. Sections (5 μm thick) were used for all immunostainings except for those analyzed using confocal microscopy (60–80-μm thick sections). Unspecific antibody binding was blocked by 1 hour incubation with 3% BSA in 1× PBS. Primary antibody incubation was performed in room temperature for 2 hours or at 4°C overnight, followed by 45 minutes incubation with secondary antibody. Double staining for CD41 and Gr1, with both primary antibodies derived from the same species, was performed by initial staining for CD41 according to standard procedure, followed by overnight incubation at 4°C with the APC-conjugated Gr1 antibody to prevent overlapping immunoreactivity. Tissue sections were thereafter counter stained with Hoechst (Molecular Probes, H3570) and mounted with Fluoromount-G (Southern Biotech, 0100-01).
The following antibodies and dilutions were used in the study: anti-CD31 (BD Biosciences; 557355, 1:1,000), anti-CD41 (BD Biosciences; 553847, 1:300), anti-fibrinogen (DakoCytomation A0080; 1:200), anti-Gr1 (BD Biosciences; 553123, 1:30), APC-conjugated anti-CD41 (BD Biosciences; 561083, 1:100), anti-CD68 (Serotec; MCA1957, 1:300), anti–ICAM-1 (R&D Systems; AF796, 1:250), and anti–VCAM-1 (R&D Systems; AF643, 1:250). Secondary antibodies: anti-goat Alexa 555 (Molecular Probes; A21431, 1:1000), anti-rabbit Alexa 586 (Molecular probes; A11011, 1:1,000), anti-rat Alexa 586 (Molecular Probes; A11077, 1:1,000), anti-rat Alexa 647 (Jackson Immuno Research; 712-605-150, 1:1,000).
Imaging was done on a Nikon Eclipse 90i microscope using the NIS Elements 3.2 software. Images were analyzed using the ImageJ 1.44 software (NIH). Confocal imaging was done on a Zeiss LSM 710 microscope, using the ZEN 2012 software.
Quantification of neutrophils with extracellular DNA tails
Blood cells were isolated from whole blood by centrifugation at 150 × g for 15 minutes. Cells were thereafter plated on slides using cytospin (centrifugation 500 rpm, 6 minutes), fixed in ice-cold methanol and immunostained according to the protocol described above. Proportion of neutrophils undergoing NETosis was determined by staining for DNA (Hoechst) and neutrophils (Gr1), and quantification of neutrophils with secreted DNA tails related to total neutrophil count.
Quantification of DNA in plasma
DNA in plasma was quantified using the Quant-iT Picogreen dsDNA Assay Kit (Life Technologies, P7589), according to the manufacturer's instructions. Blood samples were collected by heart puncture, and plasma was isolated by centrifugation at 1,100 × g for 10 minutes.
RNA extraction and qRT-PCR
The expression of proinflammatory mediators and endothelial activation markers was analyzed by qPCR on RNA extracted from kidneys. qPCR was also used to determine expression of G-CSF in primary tumors from MMTV-PyMT mice, RIP1-Tag2 mice, and C57BL6 mice with subcutaneously grafted B16 melanoma. RNA was extracted from kidney and tumor tissue using the RNeasy Midi Kit (Qiagen; 75142). Tissues were homogenized in lysis buffer and RNA was isolated and purified according to the procedure recommended by the manufacturer. RT-PCR was performed using the SuperScript III Reverse Transcriptase (Invitrogen, 18080-044). SYBR Green (Life Technologies, 4309155) was used for the qPCR reaction. The genes and associated primers used for qPCR are listed in Supplementary Table S1.
Statistical analysis was performed using Prism (GraphPad Software Inc.). The nonparametric two-tailed Mann–Whitney analysis was used to compare data in various groups. Correlation analysis was performed using the Spearman correlation test; *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001.
Mice with tumors display impaired vascular function in peripheral tissue
Systemic vascular effects of cancer were analyzed using two well-established transgenic mouse models; the MMTV-PyMT mouse model developing mammary carcinomas with subsequent metastatic spread primarily to the lung, and the RIP1-Tag2 mouse developing carcinomas in the insulin-producing cells of the pancreas, with metastatic spread to lung and liver. Analysis of renal tissue from these tumor-bearing mice demonstrated a significantly impaired vascular perfusion, measured by injection of FITC-conjugated lectin, compared with healthy control mice without tumors (MMTV-PyMT: P = 0.0358; RIP1-Tag2: P = 0.0002; Fig. 1A). Vascular function was further analyzed by immunostaining for extravasated fibrinogen, allowing evaluation of the degree of leakage from the renal vasculature. Both MMTV-PyMT and RIP1-Tag2 mice displayed increased vascular leakage compared with healthy mice lacking tumors (MMTV-PyMT: P = 0.0417; RIP1-Tag2: P = 0.0004; Fig. 1B). These results clearly demonstrate that tumor-bearing individuals display a remarkably impaired function of the vasculature in a peripheral tissue not directly affected by the primary tumor or metastases.
Peripheral vascular perfusion correlates with primary tumor size
As described above, perfusion of the renal vasculature was found to be significantly impaired in two different mouse models of cancer. However, although the MMTV-PyMT mice all displayed a low perfusion, RIP1-Tag2 mice clearly formed two separate groups; one group of mice with extremely poor perfusion (5%–30%), and one group with perfusion corresponding well with the control mice lacking tumors (around 80%; Fig. 1A, bottom right). Interestingly, a correlation between perfusion of the renal vasculature and size of the primary tumors was found. Mice with primary tumor volumes larger than 100 mm3 had significantly impaired perfusion compared with mice with tumor volumes below 100 mm3 (P = 0.0355; Fig. 1C). This suggests that primary tumor burden is an important factor for the systemic vascular effects seen in tumor-bearing mice.
Increased neutrophil infiltration in kidneys from tumor-bearing mice
To determine whether the reduced peripheral vessel function in mice with cancer was mediated via altered infiltration of immune cells, kidney sections from MMTV-PyMT mice were immunostained for detection of macrophages (CD68), neutrophils (Gr1), and platelets (CD41). Neutrophil infiltration was significantly increased in renal tissue from tumor-bearing mice compared with tumor-free control mice (MMTV-PyMT: P = 0.0159; RIP1-Tag2: P = 0.0085; Fig. 2A). Platelet numbers in the kidneys were not significantly altered, although there was a trend toward increased platelet numbers in kidneys from MMTV-PyMT mice (Fig. 2B). A similar result was observed for infiltrating macrophages, where MMTV-PyMT mice demonstrated a slight but not significant increase (MMTV-PyMT: P = 0.1714; RIP1-Tag2; P = 0.9714; Fig. 2C). Interestingly, a significant correlation was found between primary tumor weight and the number of neutrophils infiltrating the kidney glomeruli of MMTV-PyMT mice, suggesting that the primary tumor may promote neutrophil infiltration in a distant organ (P = 0.0108; Fig. 2D).
Neutrophil depletion improves function of the renal vasculature in tumor-bearing mice
The results described above show that the decrease in renal vascular function is paralleled with increased neutrophil infiltration in mice with cancer. To investigate whether the increased neutrophil infiltration was linked to the impaired vascular function, MMTV-PyMT and RIP1-Tag2 mice were neutrophil depleted by injection of an antibody against the neutrophil cell surface marker Gr1. Successful neutrophil depletion was verified by immunostaining (Supplementary Fig. S1A). Neutrophil depletion significantly improved function of the renal vasculature, as illustrated by increased FITC-lectin perfusion (MMTV-PyMT: P = 0.0399; RIP1-Tag2: P = 0.0465; Fig. 3A and Supplementary Fig. S2A) and reduced vascular leakage (MMTV-PyMT: P = 0.0295; RIP1-Tag2: P = 0.0052; Fig. 3B and Supplementary Fig. S2B). Altogether, this indicates that neutrophils are important as mediators of the tumor-induced effects on peripheral vessel function.
NETs are responsible for the tumor-induced negative effects on peripheral vascular function
As previously described, platelet numbers were not significantly increased in kidneys from tumor-bearing mice. However, doublestaining for neutrophils and platelets demonstrate neutrophil–platelet complex formation in renal tissue sections from both RIP1-Tag2 and MMTV-PyMT mice (Fig. 4A). This phenomenon could not be observed in any of the analyzed healthy control mice. Neutrophils and platelets are known to colocalize in NETs; web-like structures composed by neutrophil-secreted extracellular DNA and neutrophil-derived proteases (10). NET formation was originally described as a microbial defense mechanism, but recently cancer was also described as a NET-inducing factor (14). To investigate whether the neutrophil–platelet aggregates detected in renal tissue from tumor-bearing mice could be NETs, MMTV-PyMT mice were treated with DNase, a well-established method to dissolve NETs, for 3 days. The DNase treatment resulted in a significant reduction of the number neutrophil–platelet complexes in the kidneys of MMTV-PyMT mice (P = 0.0049; Fig. 4A), suggesting that these complexes are indeed NETs. Analysis of kidney sections with confocal microscopy revealed that the NETs were almost exclusively localized within the renal vasculature (Fig. 4B). The intravascular location of the neutrophil–platelet complexes were also asessed in optical sections from three different angles, further supporting this finding (Fig. 4C). To further support that NETs are present in the circulation of tumor-bearing mice, neutrophils from peripheral blood were stained for Gr1 and DNA, and the proportion of neutrophils with extracellular DNA tails was determined. The results revealed a significantly higher number of neutrophils with DNA tails in tumor-bearing mice compared with healthy mice (P = 0.0294; Fig. 4D). In line with this, elevated levels of circulating DNA was found in plasma from tumor-bearing mice (P = 0.0489; Fig. 4E). Perfusion and vascular leakage in kidneys from DNase treated tumor-bearing mice was further analyzed to elucidate whether NETs may be responsible for the negative effects on peripheral vessel function. Indeed, perfusion of the renal vasculature was restored in these mice to levels comparable with tumor-free mice (MMTV-PyMT: P = 0.0159; RIP1-Tag2: P = 0.0026; Fig. 5A and Supplementary Fig. S3A). In parallel, the increased fibrinogen leakage in tumor-bearing mice was completely abolished after DNase treatment (MMTV-PyMT, P = 0.0079; RIP1-Tag2, P = 0.0022; Fig. 5B and Supplementary Fig. S3B). DNase treatment did not suppress the total number of neutrophils in the kidney (Fig. 5C), but decreased the proportion of neutrophils aggregating with platelets (Fig. 4A). These results demonstrate that neutrophils impair vascular function in the kidneys of tumor-bearing mice by forming NETs. The intravascular localization of the complexes indicates that NETs prevent efficient perfusion by vessel occlusion.
To explore whether our observations were specific for kidney or also applied to other organs in an individual with cancer, we analyzed FITC-lectin perfusion of heart, because this tissue is not a metastatic site in the MMTV-PyMT model. Again, we found a significantly reduced vascular perfusion in tumor-bearing mice compared with healthy littermates (Fig. 6A) and an elevated number of neutrophils in the heart (Fig. 6B). Neutrophil depletion or DNase treatment resulted in normalized perfusion, indistinguishable from healthy individuals (Fig. 6A). Neutrophil numbers were not affected by the DNase treatment (Fig. 6B). These data strongly suggest that cancer-induced systemic and intravascular NET formation reduces vascular function in organs not directly affected by the primary tumor or metastases.
A potentially important tumor-derived factor in this context is G-CSF, which was recently reported to predispose neutrophils for NET formation (14). C57BL6 mice with subcutaneously grafted B16 melanoma tumors did not display decreased perfusion or increased leakage of the renal vasculature and only a slight, but nonsignificant increase in neutrophil infiltration compared with healthy control mice (Supplementary Fig. S4A–S4C). Quantification of neutrophil–platelet complexes revealed that some complexes were formed in mice with B16 tumors, but the proportion of neutrophils aggregating with CD41 was significantly lower compared with the proportion in MMTV-PyMT mice (P = 0.0061; Supplementary Fig. S4D). Interestingly, qPCR for G-CSF revealed a strong expression of G-CSF in MMTV-PyMT tumors, whereas expression in B16 tumors was negligible in comparison (Supplementary Fig. S4E). RIP1-Tag2 tumors expressed approximately 10 times more G-CSF than B16 melanoma. These results support the hypothesis that the increased NET formation is driven by G-CSF produced by the primary tumor. To test this hypothesis directly, we treated MMTV-PyMT mice with an anti–G-CSF antibody daily during a week before analyzing renal vascular perfusion and neutrophil–platelet complex formation. Anti–G-CSF treatment restored vascular perfusion of the kidney vasculature and reduced complex formation between platelets and neutrophils (Supplementary Fig. S5A and S5B), further supporting a role for G-CSF in tumor-induced impairment of peripheral vessel function.
Increased endothelial activation and expression of proinflammatory cytokines in kidneys from mice with cancer
Our results demonstrate that mice with tumors have strikingly poor vascular perfusion of the kidneys. Maintenance of a continuous blood flow is essential for regulating endothelial activity. This suggests that kidneys from mice with tumors could have an altered activation status of their vascular endothelium, which in turn may promote inflammation. To determine whether endothelial activation is increased in renal vasculature from MMTV-PyMT mice, RNA expression levels for the endothelial adhesion molecules ICAM-1, VCAM-1 and E-selectin were analyzed in kidneys by qPCR. The results show significantly increased RNA expression for ICAM-1 (P = 0.0159), VCAM-1 (P = 0.0317) and E-selectin (P = 0.0079; Fig. 7A). To further assess the inflammatory status in kidneys from tumor-bearing mice, qPCR analysis of a panel of 15 cytokines and chemokines (Supplementary Table S1) involved in neutrophil recruitment and inflammatory responses, was performed. We found a significant upregulation of the proinflammatory cytokines IL1β (P = 0.0079; Fig. 7B), IL6 (P = 0.0079; Fig. 7C), and CXCL1 (P = 0.0317; Fig. 7D) in kidneys from MMTV-PyMT mice, as compared with kidneys from healthy mice. Upregulation of these soluble factors and adhesion molecules likely provides an explanation for the increased presence of neutrophils in the kidneys of RIP1-Tag2 and MMTV-PyMT mice. DNase treatment suppressed the expression of E-selectin, IL1β, IL6, and CXCL1 to levels comparable with those of healthy littermates (Fig. 7A–D). In line with this, some of the DNase-treated mice showed a marked reduction in expression of ICAM-1 and VCAM-1, and the significant upregulation detected in tumor-bearing mice was lost (Fig. 7A). Together these molecular data clearly show that an inflammatory response can be induced in a “healthy” organ in an individual with cancer, as a result of tumor-induced systemic effects.
Cancer mortality is mainly connected to systemic effects induced by the primary tumor such as metastasis, deep vein thrombosis, and a general organ failure. Because of these tumor-induced effects on the systemic environment in an individual with cancer, we aimed to investigate potential changes with respect to vascular function and inflammation in organs that are not targets for metastasis or affected by the primary tumor. Using the two orthotopic and metastasizing models RIP1-Tag2 for insulinoma and MMTV-PyMT for breast cancer, both believed to closely resemble disease development in the human situation, we found a dramatic reduction in the vascular function in kidney and heart compared with healthy littermates of the same genetic background (FVB/n and C57BL6). In some individuals, renal perfusion was as low as 5% to 30%, but with an average 50% reduction compared with mice without tumors. Moreover, there was a 5- to 10-fold increase in the amount of extravasated fibrinogen in the kidney, indicative of vascular leakage. We found these data unexpected, but looking into the situation in humans it is, however, clear that there is a high incidence of renal complications as a consequence of cancer or its treatment (29, 30). ARF is a cause of substantial morbidity in cancer patients and is indeed characterized by hypoperfusion of the kidney vasculature (25).
In addition to the reduced vascular functionality, we found a significantly elevated number of Gr1+ neutrophils in heart and kidney tissue from mice with cancer. These neutrophils were strongly associated to the observed vascular effects, because depletion of Gr1+ cells normalized kidney vascular function in both RIP1-Tag2 and MMTV-PyMT mice. Even more striking was the effect of DNase treatment, which completely restored perfusion and prevented leakage in kidney and heart vasculature, resulting in a perfusion indistinguishable from the healthy individual. DNase treatment did not significantly alter the total number of neutrophils in the kidney or heart of tumor-bearing mice but reduced the number of neutrophil–platelet complexes, suggesting that NETs are the cause of the impaired vascular function in individuals with cancer. G-CSF has been reported to predispose neutrophils to form NETs (31) and tumor-derived G-CSF has been implicated as a possible cause of cancer-related thrombosis by priming circulating neutrophils to undergo NETosis (14). In agreement, we found no impairment of peripheral vessel function or increased neutrophil infiltration in kidneys from mice with subcutaneous B16 melanoma, a tumor type that expressed very low levels of G-CSF compared with MMTV-PyMT tumors (220-fold difference) and RIP1-Tag2 tumors (10-fold difference). Moreover, administration of an anti–G-CSF antibody during one week restored kidney vascular perfusion of MMTV-PyMT mice. These data further support NETosis as a cause of the impaired vascular function in distant organs in mice with cancer.
By which mechanisms could NETs cause impaired function of the vasculature? NETs are reported to be cytotoxic for endothelium and to directly induce endothelial cell damage (9, 32). These harmful effects on endothelium by NETs are likely caused by the release of proteases such as elastase and myeloperoxidase. Impaired perfusion could also create an ischemic condition in small vessels, thereby causing endothelial cell damage. In addition, histones, which are present in the NETs, have been demonstrated to exert endothelial cell cytotoxicity (33). We could however not detect any increased apoptosis in kidney endothelium from mice with tumors compared with healthy littermates (data not shown), but this does not exclude an elevated endothelial cell death due to other mechanisms than apoptosis. It is, however, possible that the decreased blood perfusion we detected in distant organs in mice with cancer is simply a consequence of small occlusions in the microvessels, generated by NETs. The latter suggestion is supported by the fact that DNase treatment restored vascular function and reverted the inflammatory phenotype in kidneys of tumor-bearing mice.
A change in blood flow can affect the activation state of the endothelium and induce an upregulated expression of adhesion molecules, promoting influx of inflammatory cells. Indeed, we could detect elevated expression levels of ICAM-1, VCAM-1, and E-selectin in the kidneys of mice with tumors. Moreover, we found increased transcriptional levels of the inflammatory cytokines IL1β and IL6, as well as the chemokine CXCL1. These molecules are important mediators of inflammatory responses and are expressed by leukocytes and platelets, but also epithelial cells and tumor cells in the case of CXCL1 (34–36). CXCL1 possesses strong neutrophil chemoattractant capacity (36). Both IL1β and IL6 are also expressed by tubular epithelial cells in diseased kidneys (37, 38). In line with this, IL1β and IL6 are overexpressed in kidneys from mice with ARF induced by ischemia or cisplatin treatment (39–41). The importance of IL6 in ARF has been demonstrated in IL6-deficient mice (39). Mice lacking IL6 display reduced endothelial ICAM-1 expression and neutrophil infiltration in renal tissues, paralleled with less histologic signs of tubular injury, upon experimentally induced ARF (39). Increased expression of CXCL1, IL1β, and IL6 in the kidney of tumor-bearing mice indicates an ongoing inflammatory response, with a molecular signature resembling ARF. Interestingly, DNase treatment reverted this inflammatory phenotype in the kidneys from tumor-bearing mice, indicating that NETs are causing these changes. The fact that the proinflammatory state was reversible, is encouraging from a therapeutic perspective and suggests that DNase could potentially be a useful drug to prevent tumor-induced systemic inflammation and possibly organ failure. DNase is already in clinical use for the management of cystic fibrosis, where DNA containing debris is collected in the lungs, demonstrating its safety as a drug (42).
On the basis of our data, we believe that the sequence of events leading to reduced perfusion of peripheral vessels in individuals with cancer are initiated by tumor secreted factors (likely G-CSF but additional soluble molecules may also play a role), which promote NET formation in the periphery, causing a reduced blood perfusion in distant organs and as a consequence upregulation of leukocyte adhesion markers in the vasculature. This would further add to the inflammatory state by recruiting leukocytes to the affected tissue. It is, therefore, possible that we look at two different populations of neutrophils in kidney and heart tissue from tumor-bearing mice; (i) neutrophils engaged in intravascular NETosis and (ii) neutrophils that have extravasated and infiltrated the kidney tissue as a result of upregulation of vascular adhesion molecules.
Tumor vessels have been demonstrated to exhibit a reduced expression of endothelial adhesion molecules, mainly caused by an increased exposure to proangiogenic molecules such as VEGF and bFGF in the tumor microenvironment (43). Dirkx and colleagues (44) also report on tumor-induced systemic suppression of leukocyte–vessel wall interaction, although less pronounced compared with the situation within the tumor. This systemic effect was also assumed to be caused by circulating proangiogenic factors, but not directly addressed. Although the finding in Dirkx and colleagues may seem in contrast with our finding of a tumor-induced upregulation of endothelial adhesion markers, there are several possible explanations for this discrepancy. First of all, tumor-induced NETosis and subsequent impairment of peripheral vessel function does not occur in all tumor types, as exemplified in our study by the subcutaneous B16 melanoma model. Mice with B16 tumors display very limited platelet–neutrophil complexes in the kidney and no impairment of the renal vascular perfusion. Moreover, B16 tumors do not express G-CSF, a molecule that promotes NET formation. Clearly, there is a dependency on tumor type for systemic NETosis and the subsequent peripheral vascular effects we find. In the study by Dirkx and colleagues, B16 melanoma was used in addition to the human colon carcinoma cell line LS174T in immunocompromised mice. We have no information on G-CSF expression in LS174T, and it is uncertain if human G-CSF would be equally potent as mouse G-CSF in eliciting host NETosis. Second, the study by Dirkx and colleagues analyzed total leukocyte–vessel wall interaction within the ear vasculature, whereas we measure expression of adhesion molecules, as well as neutrophil infiltration, in the kidney. To address a potentially upregulated VEGFR2 signaling in kidneys of tumor-bearing mice, derived from tumor-secreted VEGF in the circulation, we performed imunostainings with an antibody recognizing the complex of VEGFA and VEGFR2 (GV39M), but did not find an elevated signal compared with kidneys from healthy mice (data not shown). A recent publication also nicely demonstrate that due to polarization of VEGFR1/2 expression—VEGFR1 on the luminal side and VEGFR2 on the abluminal side of blood vessels (45), VEGF in the circulation is less likely to affect the endothelium compared with VEGF coming from the tumor tissue (abluminal side).
In conclusion, our data strongly suggest that the impaired peripheral vessel function we detect in individuals with cancer is due to formation of intravascular NETs. It is possible that tumor-induced NETosis is a contributing factor to general organ failure commonly seen in cancer patients. This process may be more or less pronounced depending on tumor type–specific expression of soluble factors promoting NET formation. It is clear that tumor-induced impacts on distant organs are not limited to metastatic sites. One could speculate that the tumor-induced effects on peripheral vasculature we observe does play a role in metastasis, but that the actual growth and survival of the metastatic cells in an organ is critically dependent on the correct microenvironment. Considering that cancer mortality is mainly due to systemic effects induced by the primary tumor, including dissemination to distant organs, thrombotic events, and organ failure, increased knowledge of these processes are warranted to improve current therapeutic strategies.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Conception and design: J. Cedervall, A. Dimberg, A.-K. Olsson
Development of methodology: J. Cedervall, Y. Zhang, H. Huang, A.-K. Olsson
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): J. Cedervall, Y. Zhang, H. Huang, L. Zhang, J. Femel, A.-K. Olsson
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J. Cedervall, Y. Zhang, A.-K. Olsson
Writing, review, and/or revision of the manuscript: J. Cedervall, J. Femel, A. Dimberg, A.-K. Olsson
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): J. Cedervall, Y. Zhang
Study supervision: A. Dimberg, A.-K. Olsson
This study was supported by The Swedish Cancer Society #11 0653 and the Swedish Research Council #2010-6903-75363-44 (A.-K. Olsson) and #2012-77PK-22157-01-2 (J. Cedervall).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The authors thank the kind supply of tumor tissue from Dr. Elise Langenkamp, Uppsala University, and Prof. Kristian Pietras, Lund University, Sweden.
Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/).
- Received November 10, 2014.
- Revision received March 2, 2015.
- Accepted March 30, 2015.
- ©2015 American Association for Cancer Research.