Tumor-derived protein tissue inhibitor of metalloproteinases-1 (TIMP1) correlates with poor prognosis in many cancers, including highly lethal pancreatic ductal adenocarcinoma (PDAC). The noncanonical signaling activity of TIMP1 is emerging as one basis for its contribution to cancer progression. However, TIMP1–triggered progression-related biological processes are largely unknown. Formation of neutrophil extracellular traps (NET) in the tumor microenvironment is known to drive progression of PDAC, but factors or molecular mechanisms initiating NET formation in PDAC remain elusive. In this study, gene-set enrichment analysis of a human PDAC proteome dataset revealed that TIMP1 protein expression most prominently correlates with neutrophil activation in patient-derived tumor tissues. TIMP1 directly triggered formation of NETs in primary human neutrophils, which was dependent on the interaction of TIMP1 with its receptor CD63 and subsequent ERK signaling. In genetically engineered PDAC-bearing mice, TIMP1 significantly contributed to NET formation in tumors, and abrogation of TIMP1 or NETs prolonged survival. In patient-derived PDAC tumors, NETs predominantly colocalized with areas of elevated TIMP1 expression. Furthermore, TIMP1 plasma levels correlated with DNA-bound myeloperoxidase, a NET marker, in the blood of patients with PDAC. A combination of plasma levels of TIMP1 and NETs with the clinically established marker CA19–9 allowed improved identification of prognostically distinct PDAC patient subgroups. These observations may have a broader impact, because elevated systemic levels of TIMP1 are associated with the progression of a wide range of neutrophil-involved inflammatory diseases.

Significance:

These findings highlight the prognostic relevance of TIMP1 and neutrophil extracellular traps in highly lethal pancreatic cancer, where a noncanonical TIMP1/CD63/ERK signaling axis induces NET formation.

Tissue inhibitor of metalloproteinases-1 (TIMP1) is a multifunctional protein (1), whose elevated levels in the blood consistently correlate with poor prognosis in virtually all cancer types (2), including pancreatic ductal adenocarcinoma (PDAC; refs. 3, 4), one of the most lethal common cancer types (5). The multifunctionality of TIMP1 is largely based on its two-domain structure and comprises the canonical antiproteolytic function mediated via the N-terminal domain (N-TIMP1; ref. 6) as well as a previously rather neglected noncanonical CD63-mediated signaling activity via the C-terminal domain (1, 4, 7). Although the clinical correlation between TIMP1 expression (4, 8, 9) and short survival of patients with PDAC suggest a contribution of TIMP1 to disease progression, the functional role of TIMP1 in the tumor microenvironment remains largely unknown.

Neutrophil granulocytes are one important component of the tumor microenvironment and their activation has been shown to mediate mostly pro-tumoral effects (10). In specific, neutrophils can form neutrophil extracellular traps (NET), which are released upon peptidyl arginine deiminase 4 (PAD4)-mediated citrullination of histones and resulting chromatin decondensation (11). Neutrophils lacking PAD4 are therefore deficient in NET formation (11). NETs consist of clusters of DNA fibers and contain citrullinated histones, as well as granule-derived proteins like myeloperoxidase (MPO) and neutrophil elastase (NE; ref. 12). Recently, NET formation has been shown to accelerate tumorigenesis, drive progression (13), as well as mediate therapy resistance (14) of PDAC. Therefore, NET formation is – in principle – discussed as a promising therapeutic target in PDAC (13). Although several studies have shown that tumor cell–derived factors are capable of triggering NET formation (14, 15), lack of knowledge about responsible factors and molecular mechanisms directly initiating NET formation in PDAC, impedes its specific therapeutic targeting.

In this study, we identified the signaling function of TIMP1 as trigger of NET formation and linked this discovery to PDAC. In view of the ubiquitous presence and multiple functions of neutrophils (16–19), as well as of TIMP1 (2, 20, 21) in cancer and inflammation, we foresee a crucial pathophysiologic relevance of TIMP1-triggered NET formation in inflammatory diseases as well as its clinical usability in many pathologic contexts.

Human samples

Human blood samples from patients with PDAC were collected in an EDTA tube, centrifuged, and plasma was snap frozen in liquid nitrogen and stored at −80°C until further analysis. Human PDAC tissue samples (formalin-fixed, paraffin embedded) of patients resected at Klinikum rechts der Isar, München, were retrieved from the tissue bank (MTBIO). Patients were free from bacterial infections at the timepoint of blood collection. Plasma TIMP1 as well as plasma NET levels did not differ significantly between treatment-naïve and pretreated patients (Supplementary Fig. S1A and S1B). Therefore, differences in both parameters that are detected in our cohort are not due to differences with regard to therapy. For more information on human samples, see Supplementary Material and Methods.

Animal models

The pancreatic cancer transgenic mouse model KPC (Pdx-1+/Cre; Kras+/LSL-G12D; Trp53+/LSL-R172H) used in this study was described in detail elsewhere (22). To study the role of TIMP1 and NETs in PDAC, Pdx-1+/Cre; Kras+/LSL-G12D; Trp53+/LSL-R172H mice were bred into TIMP1–deficient C57BL/6.129S4-TIMP1tm1Pds/J mice (23) (TIMP1−/−, obtained from The Jackson Laboratory; RRID:IMSR_JAX:006243) to obtain KPC-TIMP1 KO mice or KPC mice were bred into PAD4-deficient B6.Cg-Padi4tm1.1Kmow/J (C57BL/6-PAD4−/−) mice (PAD4−/−, obtained from The Jackson Laboratory; RRID:IMSR_JAX:030315; ref. 24) to obtain KPC-PAD4 KO mice. Classification and grading of the murine pancreatic neoplasia was performed according to most recent consensus classification (25) by an experienced comparative pathologist (to K. Steiger) blinded to sample identity. To enable the comparison of pancreatic tumor-bearing mice with histologically identical disease progression, only KPC mice with pancreatic cancer grades G2 and/or G3 (advanced-stage PDAC) were used for analysis. The median age of the analyzed mice was 13.5 weeks, with an interquartile range from 10.8 weeks to 17.8 weeks. For more information on mouse models and tissue staining see Supplementary Material and Methods.

Proteome analysis

Proteome analysis of stroma from pancreatic cancer tissue derived from human patients with PDAC was performed with recently published proteomics data (26). Abundance levels of all detected, annotated proteins (n = 4,089) from PDAC patient–derived tissues (n = 11) were correlated with TIMP1 protein levels by employing Spearman correlation. Proteins that showed significant correlation with TIMP1 protein levels (P < 0.05, n = 2,533) were ranked on the basis of their respective Spearman correlation coefficient and employed in gene set enrichment analysis (GSEA; RRID:SCR_003199; software version 4.1). GSEA was performed as previously described (ref. 27; Reference Gene Set: Human_GOBP_AllPathways_no_GO_iea_August_01_2020_symbol.gmt downloaded from http://download.baderlab.org/EM_Genesets/). Cytoscape (RRID:SCR_003032; software version 3.8) was employed for visualization of GSEA results as enrichment map (creation parameters: FDR q-value ≤ 0.05; normalized enrichment score > 2.0; overlap size > 100).

Isolation of primary human neutrophils and visualization and quantification of NETs

Primary human neutrophils were isolated from healthy individuals as described previously (28). For NET formation experiments, neutrophils were treated as indicated, stained, and images were acquired with a 20x objective with five pictures per well and with an exposure time of 40 ms. NET formation was quantified by automated calculation of SYTOX-positive areas employing CellProfiler (RRID:SCR_007358; ref. 29), applying the same threshold value to all pictures. Since we observed variations in background intensity in stimulations with human plasma, we calculated the threshold for these samples employing maximum correlation thresholding (MCT) as described before (30). Detection of cell-free DNA in the supernatant of neutrophils was performed as described previously (31). Briefly, neutrophils were treated as indicated for 4h and subsequently incubated with 1 U/ml micrococcal nuclease (MNase, #88216, Thermo Fisher Scientific Inc.) and 1 μmol/L Sytox Orange. Cells were pelleted and extracellular DNA content was determined in the cell-free supernatant by measuring fluorescence. For more information on NET staining, including employed antibodies, reagents, concentrations of stimulants and visualization and quantification of NETs see Supplementary Material and Methods.

Priming of neutrophils

After seeding of isolated neutrophils, media was replaced with priming media [RPMI-1640 (Biochrom) supplemented with 0.3% recombinant human serum albumin (cat. #A9731–1G, Sigma-Aldrich)], containing either 10 ng/mL recombinant human TNFα (cat. #570104, Biolegend; ref. 31), 250 units/mL IFNγ (cat. #570202, Biolegend; ref. 32), or 25 ng/mL GMCSF (cat. #130–093–868, Miltenyi Biotec; ref. 33). After incubation for 15 min at 37°C in a humidified 5% CO2 atmosphere, surface expression of CD63 was analyzed by flow cytometry or neutrophils were stimulated for 4h as indicated and extracellular DNA was quantified.

Flow cytometry

Surface expression of CD63 as well as TIMP1 binding to neutrophils was monitored by flow cytometry. Human recombinant human TIMP1 was fluorescently labelled using the Alexa Fluor 488 Microscale Protein Labeling Kit (Thermo Fisher Inc.) according to manufacturer's instructions. For detailed information on antibodies, reagents and evaluation of flow cytometry data, see Supplementary Material and Methods.

Western blot analysis

A total of 3 × 106 neutrophils were seeded and incubated for 1 hour at 37°C to allow adherence. Neutrophils were treated for 10 minutes as described and whole-cell extracts were isolated. Western blots were performed as described in Supplementary Material and Methods.

Immunofluorescence microscopy of human and murine tissue sections

Immunofluorescence microscopy was performed on paraffin-embedded sections of pancreatic tissue from human patients with PDAC or KPC mice. Tissue sections were prepared as described elsewhere (34). To quantify the surface area of NETs in regions expressing high or low levels of TIMP1, NETs were defined as colocalization of MPO, citH3 and DNA (35) and the surface area of NETs was calculated for regions with a TIMP1 signal above or below a certain threshold (TIMP1 high: above threshold = 20, TIMP1 low: below threshold = 20) employing Imaris 9.3.1 (RRID:SCR_007370). For detailed information on immunofluorescence stainings, see Supplementary Material and Methods.

MPO-DNA ELISA and TIMP1 ELISA

Quantification of NETs in human plasma by MPO-DNA ELISA was performed as described elsewhere (36). Plasma TIMP1 levels were determined in healthy human donors or patients with PDAC using the respective DuoSet ELISA kit (R&D Systems) according to manufacturer's instructions. For detailed information on ELISA measurements, see Supplementary Material and Methods.

Production and purification of recombinant human TIMP1 and N-TIMP1

Recombinant human (N-)TIMP1 was expressed in an endotoxin-free mammalian cell culture system of HEK293F cells (Thermo Fisher Scientific), and purified in a three-step protocol employing an Äktapure Fast Protein Liquid Chromatography (FPLC) system (Cytivia Europe; Supplementary Fig. S2A–S2C). Glycosylation (Supplementary Fig. S2D) and MMP-inhibitory activity (Supplementary Fig. S2E) were verified by PNGase F digest and an MMP-inhibition assay, respectively. For detailed information on the purification and biochemical characterization of (N-)TIMP1, see Supplementary Material and Methods.

Statistical analysis

Normal distribution was tested by Shapiro–Wilk tests. Associations between quantitative variables were tested by Spearman correlations due to non-normal distribution. Groups were compared using Student t test for independent samples in the case of normal distribution, or nonparametric Mann–Whitney test for independent variables in the absence of normal distribution. Univariate or multivariate (Supplementary Table S1A–S1C) Cox regression analysis was employed to calculate the HR including the corresponding 95.0% confidence interval. Flow cytometry data was analyzed for statistically significant differences between groups employing the Kruskal–Wallis test due to absence of equal variances. For more information on statistics, see Supplementary Material and Methods.

Study approval

This study was approved by the Ethics Committee of the Medical Faculty of the Technical University of Munich, Germany (#1946/07, #409/16S, #395/17S, #403/17S). The study population (Supplementary Table S2) comprised patients diagnosed with pancreatic cancer between 2011 and 2018 in the Department of Surgery, Klinikum rechts der Isar, Munich, who agreed to participate in the study by a written informed consent prior to inclusion in the study. The diagnosis was verified by postoperative definitive histologic examination, or, for patients without surgery, by cytology or clinical/radiologic information, to the best of state-of-the-art methodology. Analysis of patient data was conducted on a pseudonymized dataset. Animal experiments were performed in compliance with the Tierschutzgesetz des Freistaates Bayern and approved by the Regierung von Oberbayern (#ROB-55.2–2532.Vet_02–15–47).

Pathologic levels of TIMP1 induce NET formation

To get a first idea of which biological processes are associated with high TIMP1 levels in the tumor microenvironment, we employed a published proteome data set of tumor stroma derived from patients with PDAC (26). Gene-set enrichment analysis revealed neutrophil activation as biological process most prominently positively correlating with TIMP1 protein expression (Fig. 1A). Because protumorigenic effects of activated neutrophils on pancreatic cancer cells were shown to be mediated by NET formation (13), we investigated a potential functional link between NETs and TIMP1. For this, we exposed healthy donor-derived neutrophils to recombinant human TIMP1 at concentrations as detected in the plasma of PDAC patients (4, 8). We observed that TIMP1 triggered NET-release as demonstrated by extracellular colocalization of MPO, NE, and DNA (Fig. 1B) as well as citrullinated histone H3 (citH3), MPO and DNA (Supplementary Fig. S3A). The proportion of NET-forming neutrophils was significantly increased as compared to the untreated control (Supplementary Fig. S3B). This first observation was further quantified by two independent methods, namely automated detection and calculation of DNA-occupied areas (Supplementary Fig. S3C) as well as detection of NETs in the cell-free supernatant of TIMP1–stimulated neutrophils (Supplementary Fig. S3D), whereby neutrophils stimulated with the phorbol ester PMA served as positive control for these NET-formation assays (Supplementary Fig. S3E). In fact, the TIMP1-triggered increase in DNA-occupied areas and cell-free supernatants could be attributed to NET formation, since incubation of neutrophils with a NET-inhibitor targeting PAD4 (Cl-amidine) abolished TIMP1-induced NET formation (Supplementary Fig. S3F). Further analyzing TIMP1-mediated NET formation, we observed that physiological concentrations of 50 ng/mL TIMP1, as observed in the plasma of healthy donors (mean ± SEM measured as 44.4 ng/mL ± 15.6 ng/mL; n = 4), did not trigger NET formation (Fig. 1C). In contrast, pathophysiologic concentrations starting from 250 ng/mL TIMP1, as observed in the plasma of patients with PDAC (mean ± SEM, measured as 236.4 ng/mL ± 17.8 ng/ml; n = 68), significantly induced NET formation (Fig. 1C), suggesting a TIMP1 threshold of NET formation. In order to address the possibility that proteases or other factors may possibly interfere with the observed TIMP1 activity in complex human plasma, we next performed NET formation experiments in the presence of matched blood plasma obtained from the healthy donors who had provided the neutrophils. These plasma samples containing rather low TIMP1 levels were spiked with recombinant human TIMP1 to reach pathologic levels. In fact, such elevated, pathological concentrations of TIMP1 triggered NET formation also in presence of complex blood plasma (Fig. 1D). Of note, cell numbers per examined area were not influenced by TIMP1 (Supplementary Fig. S3G). Therefore, we concluded that the observed increase in DNA-occupied area was due to NET formation and not a result of alterations of cell numbers.

Figure 1.

Pathologic levels of TIMP1 induce NET formation. A, GSEA of published proteome data derived from human PDAC tissue. Illustrated are biological processes that significantly correlate with TIMP1 protein levels (creation parameters: FDR q-value ≤ 0.05; normalized enrichment score > 2.0; overlap size > 100). B, Representative confocal microscopy images of primary human neutrophils stimulated with 500 ng/mL TIMP1 or left untreated. Neutrophils were stained for DNA (blue), MPO (green), or NE (magenta). Scale bars in overview images, 50 μm. Scale bars in detail images, 20 μm. C, Quantification of DNA-occupied areas as read-out for NET formation after stimulation of primary human neutrophils with increasing concentrations of TIMP1 (c = 0 ng/mL, n = 42; c = 50 ng/mL, n = 12; 250 ng/mL, n = 12; 500 ng/mL, n = 21; or 2,000 ng/mL, n = 12). D, Quantification of DNA-occupied areas as read-out for NET-formation in the presence of healthy donor-matched plasma with (n = 15), or without (n = 15) supplementation of 500 ng/mL recombinant human (rh) TIMP1. Results are represented as the mean ± SEM. For statistical analyses, a two-sided paired t test in case of normal distribution or a nonparametric Mann–Whitney test in the absence of normal distribution was employed. **, P ≤ 0.01; ***, P ≤ 0.001; n.s., nonsignificant.

Figure 1.

Pathologic levels of TIMP1 induce NET formation. A, GSEA of published proteome data derived from human PDAC tissue. Illustrated are biological processes that significantly correlate with TIMP1 protein levels (creation parameters: FDR q-value ≤ 0.05; normalized enrichment score > 2.0; overlap size > 100). B, Representative confocal microscopy images of primary human neutrophils stimulated with 500 ng/mL TIMP1 or left untreated. Neutrophils were stained for DNA (blue), MPO (green), or NE (magenta). Scale bars in overview images, 50 μm. Scale bars in detail images, 20 μm. C, Quantification of DNA-occupied areas as read-out for NET formation after stimulation of primary human neutrophils with increasing concentrations of TIMP1 (c = 0 ng/mL, n = 42; c = 50 ng/mL, n = 12; 250 ng/mL, n = 12; 500 ng/mL, n = 21; or 2,000 ng/mL, n = 12). D, Quantification of DNA-occupied areas as read-out for NET-formation in the presence of healthy donor-matched plasma with (n = 15), or without (n = 15) supplementation of 500 ng/mL recombinant human (rh) TIMP1. Results are represented as the mean ± SEM. For statistical analyses, a two-sided paired t test in case of normal distribution or a nonparametric Mann–Whitney test in the absence of normal distribution was employed. **, P ≤ 0.01; ***, P ≤ 0.001; n.s., nonsignificant.

Close modal

TIMP1 mediates NET formation via interaction with CD63

Next, we studied the molecular mechanism of TIMP1-induced NET formation. Because TIMP1 is a multifunctional protein (1) harboring antiproteolytic activity in its N-terminal domain as well as signaling activity via interaction of its C-terminal domain with CD63, it was crucial to dissect these two features in NET formation assays at the molecular level. First, we employed a genetically engineered TIMP1 variant, called N-TIMP1. This variant lacks the entire C-terminal domain, necessary for binding of TIMP1 to CD63 and subsequent signaling (7), whereas its canonical anti-proteolytic function remains (6), as demonstrated by an MMP-9 inhibition assay (Supplementary Fig. S2E). Administration of equimolar concentrations of N-TIMP1 as compared to the full-length TIMP1 was assured. Deletion of the C-terminal domain of TIMP1 completely abolished NET formation (Fig. 2A; Supplementary Fig. S3H) demonstrating the necessity of the C-terminal domain for the NET-triggering activity of TIMP1. This finding pointed at the possibility that NET formation is mediated by interaction of TIMP1 with CD63. In fact, a rather small subpopulation of neutrophils exhibited high CD63 surface levels and it was this subpopulation, which showed most efficient TIMP1 binding (Fig. 2B). Examining the whole neutrophil population, we found that binding of TIMP1 positively correlated with the presence of CD63 (Fig. 2B; Supplementary Fig. S4A) and that TIMP1 directly bound neutrophil-derived CD63 (Supplementary Fig. S4B). These data are in line with our observation that only a part of neutrophils responded to TIMP1 with the formation of NETs (Supplementary Fig. S3B). Next, we tested the relevance of TIMP1/CD63 interaction for NET formation. Application of anti-CD63 antibodies, but not isotype control antibodies, led to a complete abrogation of TIMP1-induced NET formation (Fig. 2C; Supplementary Fig. S5A), demonstrating that NET formation indeed functionally depended on the interaction between TIMP1 and its receptor CD63. In a complementary approach corroborating this CD63-dependent mechanism of TIMP1–triggered NET formation, we scavenged TIMP1 with a synthetic peptide comprising 50 amino acids of the large extracellular loop of human CD63 (called LEL-CD63). This approach also abolished TIMP1–mediated NET formation (Fig. 2D). We next investigated whether priming of neutrophils towards NET formation, which is known to occur upon treatment with the cytokines TNFα, IFNγ, or GMCSF (31–33), might impact on their CD63 levels. Only TNFα increased CD63 levels on neutrophils (Fig. 2E) and further increased TIMP1–induced NET formation (Fig. 2F), whereas IFNγ and GMCSF had no impact on CD63 surface levels or NET formation (Fig. 2E and F). TNFα itself did not trigger formation of NETs (Supplementary Fig. S5B). CD63 surface levels (Supplementary Fig. S6A) or TIMP1–induced NET formation (Supplementary Fig. S6B) were the same independent of the employed media (HBSS or RPMI).

Figure 2.

TIMP1 mediates NET formation via interaction with CD63. A, C, and D, Quantification of DNA-occupied areas after stimulation of primary human neutrophils. A, Neutrophils were treated with 500 ng/mL recombinant human TIMP1 (n = 18), equimolar levels of recombinant N-TIMP1 (n = 18), or were left untreated (n = 24). B, Quantification of TIMP1 surface binding to different neutrophil populations. Neutrophils were treated with or without fluorescently labeled TIMP1 and analyzed for TIMP1 binding and CD63 surface expression using flow cytometry. Geometric means of TIMP1 Alexa Fluor 488 binding are represented as the mean ± SEM of three different donors. Statistical differences between groups were analyzed employing the Kruskal–Wallis test due to absence of equal variances (***, P ≤ 0.001). C, Neutrophils were treated with TIMP1 (n = 15) or were left untreated (n = 12). To block the TIMP1/CD63 interaction, neutrophils were preincubated with an antibody against CD63 before stimulation with TIMP1 (n = 6) or were only incubated with anti-CD63 antibodies (n = 6). D, Neutrophils were treated with TIMP1 (n = 24), were left untreated (n = 21), or were stimulated with TIMP1, which was preincubated with a peptide resembling amino acids 152 – 201 of the large extracellular loop of CD63 (n = 18), or were incubated with the peptide only (n = 18). Results are represented as the means ± SEM. E and F, Neutrophils were primed as indicated and analyzed for CD63 expression by FACS analysis (nonprimed, n = 4; TNFα-primed, n = 4; GMCSF-primed, n = 3; IFNγ-primed, n = 3; E), or were subsequently stimulated with TIMP1 (F). Cell-free DNA released by nonprimed neutrophils (n = 18), TNFα-primed (n = 21), GMCSF-primed (n = 6), or IFNγ-primed (n = 6) neutrophils was normalized to nonprimed neutrophils. For statistical analyses, a two-sided paired t test in case of normal distribution or nonparametric Mann–Whitney test in the absence of normal distribution was employed. *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; n.s., nonsignificant.

Figure 2.

TIMP1 mediates NET formation via interaction with CD63. A, C, and D, Quantification of DNA-occupied areas after stimulation of primary human neutrophils. A, Neutrophils were treated with 500 ng/mL recombinant human TIMP1 (n = 18), equimolar levels of recombinant N-TIMP1 (n = 18), or were left untreated (n = 24). B, Quantification of TIMP1 surface binding to different neutrophil populations. Neutrophils were treated with or without fluorescently labeled TIMP1 and analyzed for TIMP1 binding and CD63 surface expression using flow cytometry. Geometric means of TIMP1 Alexa Fluor 488 binding are represented as the mean ± SEM of three different donors. Statistical differences between groups were analyzed employing the Kruskal–Wallis test due to absence of equal variances (***, P ≤ 0.001). C, Neutrophils were treated with TIMP1 (n = 15) or were left untreated (n = 12). To block the TIMP1/CD63 interaction, neutrophils were preincubated with an antibody against CD63 before stimulation with TIMP1 (n = 6) or were only incubated with anti-CD63 antibodies (n = 6). D, Neutrophils were treated with TIMP1 (n = 24), were left untreated (n = 21), or were stimulated with TIMP1, which was preincubated with a peptide resembling amino acids 152 – 201 of the large extracellular loop of CD63 (n = 18), or were incubated with the peptide only (n = 18). Results are represented as the means ± SEM. E and F, Neutrophils were primed as indicated and analyzed for CD63 expression by FACS analysis (nonprimed, n = 4; TNFα-primed, n = 4; GMCSF-primed, n = 3; IFNγ-primed, n = 3; E), or were subsequently stimulated with TIMP1 (F). Cell-free DNA released by nonprimed neutrophils (n = 18), TNFα-primed (n = 21), GMCSF-primed (n = 6), or IFNγ-primed (n = 6) neutrophils was normalized to nonprimed neutrophils. For statistical analyses, a two-sided paired t test in case of normal distribution or nonparametric Mann–Whitney test in the absence of normal distribution was employed. *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; n.s., nonsignificant.

Close modal

TIMP1/CD63-triggered NET formation involves ERK activation

Because ERK activation is reported as an important hub of NET formation (37), and interaction of TIMP1 with its receptor CD63 has been described to induce ERK activation (38), we next tested the impact of TIMP1/CD63 interaction on ERK signaling in primary human neutrophils. ERK activation in neutrophils was induced by full-length TIMP1 (Fig. 3A), but not by N-TIMP1 (Fig. 3A), which cannot bind to CD63 (7). Furthermore, we were able to interfere with full-length TIMP1-triggered ERK phosphorylation by preincubation of neutrophils with anti-CD63 antibodies, but not with isotype control antibodies (Fig. 3B; Supplementary Fig. S6C) or preincubation of TIMP1 with LEL-CD63 (Fig. 3C). Finally, employment of U0126, a potent membrane-permeable inhibitor of MEK, the upstream kinase of ERK, prevented both, TIMP1–mediated ERK phosphorylation (Supplementary Fig. S6D) and NET formation (Fig. 3D; Supplementary Fig. S6E), demonstrating the necessity of the CD63–ERK signaling axis for TIMP1–induced NET formation.

Figure 3.

TIMP1/CD63-triggered NET formation involves ERK activation. A–C, Densitometric analysis of Western blots of cell lysates from neutrophils treated as indicated. A, Neutrophils were treated with 500 ng/mL recombinant wild-type TIMP1 (n = 3), equimolar levels of recombinant N-TIMP1 (n = 3), or were left untreated (n = 3). B, Neutrophils were treated with 500 ng/mL TIMP1 (n = 5), were preincubated with an antibody against CD63 before stimulation with TIMP1 (n = 3), or were left untreated (n = 5). C, Neutrophils were stimulated with 500 ng/mL TIMP1 (n = 4), were treated with TIMP1, which was preincubated with LEL-CD63 (n = 3), or were left untreated (n = 4). Bar graphs indicate the mean ± SEM of ERK phosphorylation normalized to GAPDH and related to the untreated control. For statistical analyses, a one-sample t test was employed. *, P ≤ 0.05. D, Quantification of DNA-occupied areas after stimulation of primary human neutrophils. Neutrophils were stimulated with 500 ng/mL TIMP1 (n = 15) or were left untreated (n = 14) after pretreatment with an inhibitor of ERK activation (U0126; n = 15) or DMSO. Results are represented as the means ± SEM. For statistical analyses, a nonparametric Mann–Whitney test due to absence of normal distribution was employed. *, P ≤ 0.05; n.s., nonsignificant.

Figure 3.

TIMP1/CD63-triggered NET formation involves ERK activation. A–C, Densitometric analysis of Western blots of cell lysates from neutrophils treated as indicated. A, Neutrophils were treated with 500 ng/mL recombinant wild-type TIMP1 (n = 3), equimolar levels of recombinant N-TIMP1 (n = 3), or were left untreated (n = 3). B, Neutrophils were treated with 500 ng/mL TIMP1 (n = 5), were preincubated with an antibody against CD63 before stimulation with TIMP1 (n = 3), or were left untreated (n = 5). C, Neutrophils were stimulated with 500 ng/mL TIMP1 (n = 4), were treated with TIMP1, which was preincubated with LEL-CD63 (n = 3), or were left untreated (n = 4). Bar graphs indicate the mean ± SEM of ERK phosphorylation normalized to GAPDH and related to the untreated control. For statistical analyses, a one-sample t test was employed. *, P ≤ 0.05. D, Quantification of DNA-occupied areas after stimulation of primary human neutrophils. Neutrophils were stimulated with 500 ng/mL TIMP1 (n = 15) or were left untreated (n = 14) after pretreatment with an inhibitor of ERK activation (U0126; n = 15) or DMSO. Results are represented as the means ± SEM. For statistical analyses, a nonparametric Mann–Whitney test due to absence of normal distribution was employed. *, P ≤ 0.05; n.s., nonsignificant.

Close modal

NET formation in PDAC is TIMP1-dependent and determines cancer survival

Next, we aimed to establish a causal relationship between TIMP1 and NET formation in the disease-context of pancreatic cancer with the help of genetically engineered mouse models. We here focused on investigating the formation of NETs in tumor-bearing pancreata of KPC mice, because TIMP1 is increasingly expressed in the primary tumor during PDAC progression (4). Of note, only mice with comparable, histologically verified invasive PDAC were employed for subsequent analyses (Supplementary Fig. S7A). Immunodetection of colocalized MPO, citrullinated histone H3, and DNA (Fig. 4A) in pancreata of KPC mice demonstrated a strong positive correlation between NET formation in tumors and plasma TIMP1 levels in PDAC-bearing mice (Fig. 4B). Of note, there was no correlation between plasma levels of TIMP1 and neutrophil numbers (Supplementary Fig. S7B), indicating an impact of TIMP1 on neutrophil activation, i.e., NET formation, but not on neutrophil recruitment. To distinguish the association of physiologic from pathologic TIMP1 levels with NET formation, we determined TIMP1 levels of healthy control mice (ranging from 0.78 to 2.19 ng/mL TIMP1; n = 10), and grouped the PDAC-bearing mice into a TIMP1low subpopulation (TIMP1 levels below 2.19 ng/mL), and a TIMP1high subpopulation (TIMP1 levels above 2.19 ng/mL). Of note, no difference in age between KPC TIMP1low and KPC TIMP1high mice was observed (Supplementary Fig. S7C). NET formation in pancreatic tumors increased from mice with physiologic to mice with pathological TIMP1 levels (Fig. 4B and C). In fact, complete abrogation of TIMP1 expression (Fig. 4C) resulted in a similar basal level of NETs in TIMP1 knockout PDAC-bearing mice as compared with PDAC-afflicted mice with physiological TIMP1 levels (Fig. 4C), while, in contrast, TIMP1high mice showed significantly increased NET formation (Fig. 4C). Genetic ablation of TIMP1 or depletion of NETs (knock out of PAD4; ref. 24) in KPC mice led to significantly improved survival, with no significant difference between TIMP1 KO mice and PAD4 KO mice (Fig. 4D).

Figure 4.

NET formation in PDAC is TIMP1–dependent and determines cancer survival. A, Isolation of murine tumors and NET staining. Immunofluorescence microscopy of pancreatic tissue from KPC mice with visualization of DNA (blue), MPO (green), and citH3 (magenta). Arrowheads, NET-forming neutrophils. Scale bars, 50 μm. B, Correlation between plasma levels of TIMP1 and NET-formation in murine pancreatic tumors (n = 9). C, Quantification of NET-forming neutrophils in TIMP1high (n = 6, plasma TIMP1 levels above 2.19 ng/mL), TIMP1low (n = 3, plasma TIMP1 levels below 2.19 ng/mL), and KPC-TIMP1 knockout (KO, n = 7) KPC mice. Neutrophils (MPO-positive) and NET-forming neutrophils (MPO and citH3 double-positive) were quantified manually. NET-positive neutrophils of each mouse are represented as the means ± SEM. For statistical analyses, a two-sided paired t test was employed. **, P ≤ 0.01; n.s., nonsignificant. D, Kaplan–Meier survival curves of KPC (n = 57), KPC-TIMP1 knockout (n = 18), and KPC-PAD4 knockout (n = 14) mice. Log-rank test (Mantel–Cox test) was used for statistical analyses.

Figure 4.

NET formation in PDAC is TIMP1–dependent and determines cancer survival. A, Isolation of murine tumors and NET staining. Immunofluorescence microscopy of pancreatic tissue from KPC mice with visualization of DNA (blue), MPO (green), and citH3 (magenta). Arrowheads, NET-forming neutrophils. Scale bars, 50 μm. B, Correlation between plasma levels of TIMP1 and NET-formation in murine pancreatic tumors (n = 9). C, Quantification of NET-forming neutrophils in TIMP1high (n = 6, plasma TIMP1 levels above 2.19 ng/mL), TIMP1low (n = 3, plasma TIMP1 levels below 2.19 ng/mL), and KPC-TIMP1 knockout (KO, n = 7) KPC mice. Neutrophils (MPO-positive) and NET-forming neutrophils (MPO and citH3 double-positive) were quantified manually. NET-positive neutrophils of each mouse are represented as the means ± SEM. For statistical analyses, a two-sided paired t test was employed. **, P ≤ 0.01; n.s., nonsignificant. D, Kaplan–Meier survival curves of KPC (n = 57), KPC-TIMP1 knockout (n = 18), and KPC-PAD4 knockout (n = 14) mice. Log-rank test (Mantel–Cox test) was used for statistical analyses.

Close modal

TIMP1 levels correlate with local and systemic NET formation in patients with PDAC

To translate the here-established preclinical findings of a connection between TIMP1 and NET formation into the clinically more relevant context of human PDAC, we investigated TIMP1–associated NET formation in patients with PDAC. Histologic analysis of PDAC patient–derived tumor tissues revealed colocalization of NETs with areas exhibiting high levels of TIMP1 (Fig. 5A), whereas NET-negative neutrophils were mainly observed in areas showing low levels of TIMP1 (Supplementary Fig. S8A). Towards evaluation whether TIMP1 expression is mainly observed in tumor or stromal areas within the human pancreatic primary tumor, we employed published transcriptomics data from patients with PDAC (39). This analysis revealed that high TIMP1 expression mainly corresponds to the activated stroma compartment of the primary tumor (Supplementary Fig. S8B). In addition, we observed that NET-associated factors, as previously identified in a proteomic approach (40), were also mainly attributed to the activated stroma (Supplementary Fig. S8B), further supporting our observation of a colocalization of TIMP1 and NETs. Next, we asked whether an association between TIMP1 and NETs may also be reflected in the plasma of patients with PDAC. Indeed, patients with PDAC with high levels of NET markers in the blood plasma (NEThigh) exhibited significantly higher plasma TIMP1 levels as compared with NETlow-patients (Fig. 5B). In addition, a positive correlation between plasma levels of TIMP1 and NET markers in patients with PDAC was observed (Fig. 5C).

Figure 5.

TIMP1 correlates with local and systemic NET formation in patients with PDAC. A, Representative confocal microscopy images of immunofluorescence staining of pancreatic tissue from a human PDAC patient with stainings for DNA (blue), MPO (green), citH3 (magenta), and TIMP1 (red). NETs were defined as colocalization of MPO, citH3, and DNA, and surface of NETs (yellow) was displayed employing Imaris software. Areas expressing high levels of TIMP1 are indicated in red and surrounded by dashed lines. Total surface of NETs was determined employing Imaris software and compared between areas expressing high levels of TIMP1 and areas expressing low levels of TIMP1. Results are represented as means, normalized to areas with low TIMP1 levels. Whiskers indicate min to max values of three different pictures. For statistical analyses, a one-sample t test was employed. *, P ≤ 0.05. Scale bars, 20 μm. B, Plasma TIMP1 levels from individual PDAC patients with low NETs [<0.1225 plasma NETs (corrected absorbance at 450 nm), n = 17] and patients with PDAC with high NETs [≥0.1225 plasma NETs (corrected absorbance at 450 nm), n = 51] are shown as dots; medians of both groups are indicated as crossbars; whiskers indicate min to max values. For statistical analyses, a nonparametric Mann–Whitney test due to absence of normal distribution was employed. *, P ≤ 0.05. C, Correlation of relative plasma NET levels with plasma TIMP1 levels from patients with PDAC (n = 68). For statistical analyses, a Spearman rank correlation due to absence of normal distribution was employed. Spearman rank correlation coefficient (R) and a linear trend line are indicated.

Figure 5.

TIMP1 correlates with local and systemic NET formation in patients with PDAC. A, Representative confocal microscopy images of immunofluorescence staining of pancreatic tissue from a human PDAC patient with stainings for DNA (blue), MPO (green), citH3 (magenta), and TIMP1 (red). NETs were defined as colocalization of MPO, citH3, and DNA, and surface of NETs (yellow) was displayed employing Imaris software. Areas expressing high levels of TIMP1 are indicated in red and surrounded by dashed lines. Total surface of NETs was determined employing Imaris software and compared between areas expressing high levels of TIMP1 and areas expressing low levels of TIMP1. Results are represented as means, normalized to areas with low TIMP1 levels. Whiskers indicate min to max values of three different pictures. For statistical analyses, a one-sample t test was employed. *, P ≤ 0.05. Scale bars, 20 μm. B, Plasma TIMP1 levels from individual PDAC patients with low NETs [<0.1225 plasma NETs (corrected absorbance at 450 nm), n = 17] and patients with PDAC with high NETs [≥0.1225 plasma NETs (corrected absorbance at 450 nm), n = 51] are shown as dots; medians of both groups are indicated as crossbars; whiskers indicate min to max values. For statistical analyses, a nonparametric Mann–Whitney test due to absence of normal distribution was employed. *, P ≤ 0.05. C, Correlation of relative plasma NET levels with plasma TIMP1 levels from patients with PDAC (n = 68). For statistical analyses, a Spearman rank correlation due to absence of normal distribution was employed. Spearman rank correlation coefficient (R) and a linear trend line are indicated.

Close modal

Plasma TIMP1/NET levels predict survival of patients with PDAC

Next, we tested the relevance and possible clinical applicability of our new findings of TIMP1 and NETs for the prediction of patient survival. High plasma levels of TIMP1 (TIMP1high; Fig. 6A) as well as plasma levels of NET markers (NEThigh; Fig. 6B) were individual indicators of an increased death-risk of patients with PDAC. Importantly, TIMP1 and NET levels were dependent predictors of survival (Supplementary Table S1C), and combination of both parameters markedly enhanced the prognostic power (Fig. 6C). Furthermore, we identified three distinct subpopulations with good (TIMP1lowNETlow patients, abbreviated as “TINElow”), intermediate [TIMP1lowNEThigh-patients and TIMP1highNETlow patients, both without significant difference in survival (Supplementary Fig. S9A), abbreviated as “TINEint”], or very poor (TIMP1highNEThigh-patients, abbreviated as “TINEhigh”) prognosis (Fig. 6C). Remarkably, TINEhigh-patients had an almost 16-fold (HR = 15.60) increased risk to die as compared with TINElow-patients (Fig. 6C). In comparison, the clinically applied biomarker CA19–9 separated patients with a HR of 4.76 (Fig. 6D). The prognostic values of TIMP1/NETs and CA19–9 were furthermore characterized employing ROC curve analysis. TIMP1, NET, or their combination showed a high sensitivity but a rather low specificity for survival-prediction. In contrast, CA19–9 showed a high specificity and a rather low sensitivity in this respect (Supplementary Fig. S9B). Importantly, combining TINE and CA19–9 resulted in a sensitive as well as specific prognostic indicator combination (abbreviated as “TINECA”; Supplementary Fig. S9B), exhibiting a superior prognostic value as compared with CA19–9 alone (Supplementary Fig. S9C). This combination allowed separation of four prognostically different subpopulations of patients with PDAC (Fig. 6E).

Figure 6.

Plasma TIMP1/NET levels predict survival of patients with PDAC. A–E, Kaplan–Meier survival curves of patients with PDAC grouped according to their plasma TIMP1 levels (A, C, and E), plasma NET levels (B, C, and E), or serum CA19–9 levels (D and E). Log-rank tests (Mantel–Cox test) were used for statistical analyses and Cox regression analysis was employed to calculate HRs, including the corresponding 95.0% confidence intervals, between indicated groups. Cut-off values, as determined by maximally selected log-rank statistics: plasma TIMP1, 174.42 ng/mL; plasma NETs, 0.1225 corrected absorbance at 450 nm; serum CA19–9, 1088 U/ml. One-year overall survival rates of distinct patient groups: TIMP1low (n = 29 patients), 86.1%; TIMP1high (n = 39 patients), 45.2%; NETlow (n = 17 patients), 83.9%; NEThigh (n = 51 patients), 53.9%; TIMP1lowNETlow (TINElow, n = 9 patients), 100.0%; TIMP1lowNEThigh/TIMP1highNETlow (TINEint, n = 26 patients), 69.3%; TIMP1highNEThigh (TINEhigh, n = 32 patients), 43.1%; CA19–9low (n = 54 patients), 63.8%; CA19–9high (n = 8 patients), not defined, patients of this group died within one year; TINECA 0 (no parameter above respective cut-off value, n = 10 patients): 16.1%; TINECA 1 (one parameter above respective cut-off value, n = 20 patients), 32.3%; TINECA 2 (two parameters above respective cut-off value, n = 25 patients), 40,3%; TINECA 3 (three parameters above respective cut-off value, n = 7 patients), 11.3%.

Figure 6.

Plasma TIMP1/NET levels predict survival of patients with PDAC. A–E, Kaplan–Meier survival curves of patients with PDAC grouped according to their plasma TIMP1 levels (A, C, and E), plasma NET levels (B, C, and E), or serum CA19–9 levels (D and E). Log-rank tests (Mantel–Cox test) were used for statistical analyses and Cox regression analysis was employed to calculate HRs, including the corresponding 95.0% confidence intervals, between indicated groups. Cut-off values, as determined by maximally selected log-rank statistics: plasma TIMP1, 174.42 ng/mL; plasma NETs, 0.1225 corrected absorbance at 450 nm; serum CA19–9, 1088 U/ml. One-year overall survival rates of distinct patient groups: TIMP1low (n = 29 patients), 86.1%; TIMP1high (n = 39 patients), 45.2%; NETlow (n = 17 patients), 83.9%; NEThigh (n = 51 patients), 53.9%; TIMP1lowNETlow (TINElow, n = 9 patients), 100.0%; TIMP1lowNEThigh/TIMP1highNETlow (TINEint, n = 26 patients), 69.3%; TIMP1highNEThigh (TINEhigh, n = 32 patients), 43.1%; CA19–9low (n = 54 patients), 63.8%; CA19–9high (n = 8 patients), not defined, patients of this group died within one year; TINECA 0 (no parameter above respective cut-off value, n = 10 patients): 16.1%; TINECA 1 (one parameter above respective cut-off value, n = 20 patients), 32.3%; TINECA 2 (two parameters above respective cut-off value, n = 25 patients), 40,3%; TINECA 3 (three parameters above respective cut-off value, n = 7 patients), 11.3%.

Close modal

In this study, we identified TIMP1 as NET-inducing factor in PDAC and elucidated the signaling mechanism leading to this NET-induction. Notably, this functional link of TIMP1 with NET formation led us to the discovery of the usefulness of determining TIMP1/NET levels as novel possible prognostic indicator for patients afflicted with PDAC. The here-described observations not only provide key insights for pancreatic cancer with infamous prognosis, they also constitute preambles to test this combination of plasma markers for other diseases, since high circulating levels of TIMP1 are observed in virtually all types of cancer (2) as well as in inflammatory diseases (20, 21).

At the molecular level, binding of TIMP1 to CD63 turned out as necessary starting point for TIMP1–mediated NET formation. A previous study reported on the importance of the binding of a ligand to CD63 for neutrophil activation, leading to increased adhesion of these cells (41). Indeed, an agonistic anti-CD63 antibody mediated this process as a surrogate ligand (41). In this study, we identified TIMP1 as a host-derived natural ligand for CD63 on neutrophils and unraveled its NET-regulating activity. Our observation, that TIMP1 triggered NETs only in a small proportion of neutrophils is in accordance with our observation that only few neutrophils express CD63 on their surface. This effect of TIMP1 on NET formation is comparable with other physiologic NET-inducing factors, including the cytokine midkine (42). Importantly, priming of neutrophils with TNFα, which is secreted by pancreatic tumor cells and promotes tumor progression (43), increased CD63 levels on neutrophils as well as TIMP1–mediated NET-formation. With regard to downstream mediators of TIMP1/CD63 signaling, we showed that ERK activation, a known hub in NET formation (37), is essential for the execution of TIMP1–induced NET formation. Importantly, we demonstrated that pathologic plasma TIMP1 protein concentrations, which are typically found in the circulation of patients suffering from diseases such as cancer (4) or sepsis (21), but not physiologic concentrations of TIMP1, are potent to induce NET formation in vitro and in vivo. These data suggest the existence of a TIMP1 threshold of neutrophil activation.

Several studies have linked NET formation to pancreatic cancer (13–15, 44). Although it has been shown that pancreatic cancer cells induce NET formation via the release of proteins (15) and that IL17 triggers a release of so-far unknown factors from cancer cells to induce NET formation in PDAC (14), so far, no cytokine had been identified to directly mediate NET formation in the context of PDAC. Therefore, our results provide first evidence of a PDAC-associated cytokine that directly triggers NET formation. This might have several pathologic consequences, depending on the localization of the formed NETs. Potential effects of NETs in the primary tumor range from induction of therapy resistance (14) and enhanced pancreatic tumor growth (13), to promotion of pancreatic cancer cell migration and invasion (15). Our finding that high TIMP1 expression, as well as of NET-associated factors, are mainly observed in the activated tumor stroma, is in line with previous observations that NETs are mainly detected in the intratumoral stroma of patients with PDAC (44). Interestingly, activated tumor stroma subtypes (39), and the local formation of NETs (44) have both been associated with poor prognosis of patients with PDAC. In addition to the local formation of NETs in the primary tumor, systemic formation of NETs has been closely linked to metastasis, with NETs attracting cancer cells to the target site of metastasis (45), and trapping circulating tumor cells in the liver (46). Because we have previously shown that TIMP1 is increasingly expressed in the primary tumor during PDAC progression and that tumor levels of TIMP1 correlate with plasma levels of TIMP1 in KPC mice (4), the here-identified NET-inducing mechanism might not only be relevant within the primary tumor but also occur in distant sites. This might be of special relevance during liver metastasis, because we have previously shown that TIMP1 creates a premetastatic niche in the liver during PDAC progression (4).

In view of these protumorigenic functions of NETs and the fact that TIMP1 is elevated already during early stages (3, 4) and correlates with poor prognosis (8) of PDAC, we investigated the link between TIMP1 and NET formation in a preclinical setting of genetically engineered PDAC-bearing mice (22). We found that NET formation was markedly reduced upon ablation of TIMP1, pointing at a key role of TIMP1 in this context. The clinical data underline the impact of TIMP1 on NET formation. Both, the fact that a minority of patients with PDAC with low TIMP1 levels (10%; 7 of 68) and TIMP1–ablated mice showed some NET formation indicates that also additional NET-inducing factors may play a role in this disease. Nevertheless, substantial reduction of NET formation in PDAC by genetic ablation of TIMP1 had a similar effect on prolongation of survival as observed in NET-ablated (PAD4 KO) PDAC-afflicted mice, providing a first hint for a critical role of TIMP1–induced NET formation in PDAC. Therefore, TIMP1–induced NET formation may be also of therapeutic interest and potential therapeutics targeting the TIMP1/CD63 interaction range from peptides to blocking antibodies targeting the C-terminal domain of TIMP1 (47). Our findings furthermore served as mechanistic and functional basis for the translation of these parameters into the clinic. At least, the fact that TIMP1 levels and NET levels are dependent predictors of survival is strongly in line with the main hypothesis of our manuscript, namely that TIMP1 triggers NET formation.

CA19–9, which is of diagnostic as well as prognostic value (48), is so-far the only broadly accepted biomarker in PDAC (48). ROC analyses suggest that the TIMP1/NET/CA19–9 marker combination possibly provides a sensitive and specific prognostic tool, which requires further validation. This would be a straight-forward approach, since all three markers are accessible in a liquid biopsy. In view of the fact that the identification of patient subgroups with an increased risk for disease progression is a major challenge for PDAC so far (49), the prognostic TIMP1/NET/CA19–9 combination may open a new avenue to monitor patients at-risk more precisely and to interfere with this highly lethal disease (5) as early as possible.

B. Schoeps reports grants from Deutsche Forschungsgemeinschaft and grants from Wilhelm-Sander-Stiftung during the conduct of the study. C. Eckfeld reports grants from Deutsche Forschungsgemeinschaft and grants from Wilhelm-Sander-Stiftung during the conduct of the study. O. Prokopchuk reports grants and personal fees from Department of Surgery, Klinikum rechts der Isar, School of Medicine, Technical University of Munich, Munich, Germany during the conduct of the study. D. Häußler reports grants from Deutsche Forschungsgemeinschaft and grants from Wilhelm-Sander-Stiftung during the conduct of the study. C.D. Hermann reports grants from Deutsche Forschungsgemeinschaft and grants from Wilhelm-Sander-Stiftung during the conduct of the study. A. Krüger reports grants from Deutsche Forschungsgemeinschaft and grants from Wilhelm-Sander-Stiftung during the conduct of the study. No disclosures were reported by the other authors.

B. Schoeps: Conceptualization, data curation, formal analysis, validation, investigation, visualization, methodology, writing–original draft, writing–review and editing. C. Eckfeld: Conceptualization, data curation, formal analysis, investigation, methodology, writing–review and editing. O. Prokopchuk: Resources, funding acquisition, writing–review and editing. J. Böttcher: Resources, visualization, methodology, writing–review and editing. D. Häußler: Resources, methodology. K. Steiger: Resources, formal analysis. I.E. Demir: Resources, writing–review and editing. P. Knolle: Resources, writing–review and editing. O. Soehnlein: Methodology, writing–review and editing. D.E. Jenne: Conceptualization, methodology, writing–review and editing. C.D. Hermann: Conceptualization, data curation, formal analysis, investigation, visualization, methodology, writing–original draft, writing–review and editing. A. Krüger: Conceptualization, resources, supervision, funding acquisition, investigation, writing–original draft, project administration, writing–review and editing.

This work was supported by grants to A. Krüger from the Deutsche Forschungsgemeinschaft, Bonn, Germany (KR2047/1-3 and KR2047/8-1) and the Wilhelm-Sander-Stiftung, Munich, Germany (2016.124.1 and 2016.124.2). O. Prokopchuk was supported by a Clinical Leave Stipend from the German Center of Infection Research (DZIF, grant TI07.001). We thank Gillian Murphy (Cambridge University, Cambridge, United Kingdom) and Hideaki Nagase (Oxford University, Oxford, United Kingdom) for their support in the production and purification of recombinant human TIMP1 and N-TIMP1. We also thank the Gewebebank des Klinikums Rechts der Isar und der TU München for providing patient tissues.

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.

1.
Grünwald
B
,
Schoeps
B
,
Krüger
A
. 
Recognizing the molecular multifunctionality and interactome of TIMP-1
.
Trends Cell Biol
2019
;
29
:
6
19
.
2.
Eckfeld
C
,
Häußler
D
,
Schoeps
B
,
Hermann
CD
,
Krüger
A
. 
Functional disparities within the TIMP family in cancer: hints from molecular divergence
.
Cancer Metastasis Rev
2019
;
38
:
469
81
.
3.
Yu
S
,
Li
Y
,
Liao
Z
,
Wang
Z
,
Wang
Z
,
Li
Y
, et al
Plasma extracellular vesicle long RNA profiling identifies a diagnostic signature for the detection of pancreatic ductal adenocarcinoma
.
Gut
2020
;
69
:
540
50
.
4.
Grünwald
B
,
Harant
V
,
Schaten
S
,
Fruhschutz
M
,
Spallek
R
,
Hochst
B
, et al
Pancreatic pre-malignant lesions secrete TIMP1, which activates hepatic stellate cells via CD63 Signaling to create a pre-metastatic niche in the liver
.
Gastroenterology
2016
;
151
:
1011
24
.
5.
Quaresma
M
,
Coleman
MP
,
Rachet
B
. 
40-year trends in an index of survival for all cancers combined and survival adjusted for age and sex for each cancer in England and Wales, 1971–2011: a population-based study
.
Lancet
2015
;
385
:
1206
18
.
6.
Murphy
G
,
Houbrechts
A
,
Cockett
MI
,
Williamson
RA
,
O'Shea
M
,
Docherty
AJP
. 
The N-terminal domain of tissue inhibitor of metalloproteinases retains metalloproteinase inhibitory activity
.
Biochemistry
1991
;
30
:
8097
102
.
7.
Jung
K-K
,
Liu
X-W
,
Chirco
R
,
Fridman
R
,
Kim
H-RC
. 
Identification of CD63 as a tissue inhibitor of metalloproteinase-1 interacting cell surface protein
.
EMBO J
2006
;
25
:
3934
42
.
8.
Prokopchuk
O
,
Grünwald
B
,
Nitsche
U
,
Jäger
C
,
Prokopchuk
OL
,
Schubert
EC
, et al
Elevated systemic levels of the matrix metalloproteinase inhibitor TIMP-1 correlate with clinical markers of cachexia in patients with chronic pancreatitis and pancreatic cancer
.
BMC Cancer
2018
;
18
:
128
.
9.
Lee
JW
,
Stone
ML
,
Porrett
PM
,
Thomas
SK
,
Komar
CA
,
Li
JH
, et al
Hepatocytes direct the formation of a pro-metastatic niche in the liver
.
Nature
2019
;
567
:
249
52
.
10.
Coffelt
SB
,
Wellenstein
MD
,
de Visser
KE
. 
Neutrophils in cancer: neutral no more
.
Nat Rev Cancer
2016
;
16
:
431
46
.
11.
Li
P
,
Li
M
,
Lindberg
MR
,
Kennett
MJ
,
Xiong
N
,
Wang
Y
. 
PAD4 is essential for antibacterial innate immunity mediated by neutrophil extracellular traps
.
J Exp Med
2010
;
207
:
1853
62
.
12.
Brinkmann
V
,
Reichard
U
,
Goosmann
C
,
Fauler
B
,
Uhlemann
Y
,
Weiss
DS
, et al
Neutrophil extracellular traps kill bacteria
.
Science
2004
;
303
:
1532
35
.
13.
Miller-Ocuin
JL
,
Liang
X
,
Boone
BA
,
Doerfler
WR
,
Singhi
AD
,
Tang
D
, et al
DNA released from neutrophil extracellular traps (NETs) activates pancreatic stellate cells and enhances pancreatic tumor growth
.
Oncoimmunology
2019
;
8
:
e1605822
.
14.
Zhang
Y
,
Chandra
V
,
Riquelme Sanchez
E
,
Dutta
P
,
Quesada
PR
,
Rakoski
A
, et al
Interleukin-17-induced neutrophil extracellular traps mediate resistance to checkpoint blockade in pancreatic cancer
.
J Exp Med
2020
;
217
:
e20190354
.
15.
Jung
HS
,
Gu
J
,
Kim
J-E
,
Nam
Y
,
Song
JW
,
Kim
HK
. 
Cancer cell-induced neutrophil extracellular traps promote both hypercoagulability and cancer progression
.
PLoS One
2019
;
14
:
e0216055
.
16.
Kessenbrock
K
,
Krumbholz
M
,
Schönermarck
U
,
Back
W
,
Gross
WL
,
Werb
Z
, et al
Netting neutrophils in autoimmune small-vessel vasculitis
.
Nat Med
2009
;
15
:
623
25
.
17.
Silvestre-Roig
C
,
Braster
Q
,
Wichapong
K
,
Lee
EY
,
Teulon
JM
,
Berrebeh
N
, et al
Externalized histone H4 orchestrates chronic inflammation by inducing lytic cell death
.
Nature
2019
;
569
:
236
40
.
18.
Jorch
SK
,
Kubes
P
. 
An emerging role for neutrophil extracellular traps in noninfectious disease
.
Nat Med
2017
;
23
:
279
87
.
19.
Yuzhalin
AE
. 
Citrullination in Cancer
.
Cancer Res
2019
;
79
:
1274
84
.
20.
Nukarinen
E
,
Lindström
O
,
Kuuliala
K
,
Kylänpää
L
,
Pettilä
V
,
Puolakkainen
P
, et al
Association of matrix metalloproteinases -7, -8 and -9 and TIMP -1 with disease severity in acute pancreatitis. a cohort study
.
PLoS One
2016
;
11
:
e0161480
.
21.
Lorente
L
,
Martín
MM
,
Solé-Violán
J
,
Blanquer
J
,
Labarta
L
,
Díaz
C
, et al
Association of sepsis-related mortality with early increase of TIMP-1/MMP-9 ratio
.
PLoS One
2014
;
9
:
e94318
.
22.
Hingorani
SR
,
Wang
L
,
Multani
AS
,
Combs
C
,
Deramaudt
TB
,
Hruban
RH
, et al
Trp53R172H and KrasG12D cooperate to promote chromosomal instability and widely metastatic pancreatic ductal adenocarcinoma in mice
.
Cancer Cell
2005
;
7
:
469
83
.
23.
Soloway
PD
,
Alexander
CM
,
Werb
Z
,
Jaenisch
R
. 
Targeted mutagenesis of Timp-1 reveals that lung tumor invasion is influenced by Timp-1 genotype of the tumor but not by that of the host
.
Oncogene
1996
;
13
:
2307
14
.
24.
Hemmers
S
,
Teijaro
JR
,
Arandjelovic
S
,
Mowen
KA
. 
PAD4-mediated neutrophil extracellular trap formation is not required for immunity against influenza infection
.
PLoS One
2011
;
6
:
e22043
.
25.
Hruban
RH
,
Adsay
NV
,
Albores-Saavedra
J
,
Anver
MR
,
Biankin
AV
,
Boivin
GP
, et al
Pathology of genetically engineered mouse models of pancreatic exocrine cancer: Consensus report and recommendations
.
Cancer Res
2006
;
66
:
95
106
.
26.
Tian
C
,
Clauser
KR
,
Öhlund
D
,
Rickelt
S
,
Huang
Y
,
Gupta
M
, et al
Proteomic analyses of ECM during pancreatic ductal adenocarcinoma progression reveal different contributions by tumor and stromal cells
.
Proc Natl Acad Sci U S A
2019
;
116
:
19609
18
.
27.
Reimand
J
,
Isserlin
R
,
Voisin
V
,
Kucera
M
,
Tannus-Lopes
C
,
Rostamianfar
A
, et al
Pathway enrichment analysis and visualization of omics data using g:Profiler, GSEA, Cytoscape and EnrichmentMap
.
Nat Protoc
2019
;
14
:
482
517
.
28.
Wilhelmsen
K
,
Farrar
K
,
Hellman
J
. 
Quantitative in vitro assay to measure neutrophil adhesion to activated primary human microvascular endothelial cells under static conditions
.
J Vis Exp
2013
:
e50677
.
29.
Carpenter
AE
,
Jones
TR
,
Lamprecht
MR
,
Clarke
C
,
Kang
IH
,
Friman
O
, et al
CellProfiler: image analysis software for identifying and quantifying cell phenotypes
.
Genome Biol
2006
;
7
:
R100
.
30.
Padmanabhan
K
,
Eddy
WF
,
Crowley
JC
. 
A novel algorithm for optimal image thresholding of biological data
.
J Neurosci Methods
2010
;
193
:
380
84
.
31.
Hazeldine
J
,
Harris
P
,
Chapple
IL
,
Grant
M
,
Greenwood
H
,
Livesey
A
, et al
Impaired neutrophil extracellular trap formation: a novel defect in the innate immune system of aged individuals
.
Aging Cell
2014
;
13
:
690
98
.
32.
Martinelli
S
,
Urosevic
M
,
Daryadel
A
,
Oberholzer
PA
,
Baumann
C
,
Fey
MF
, et al
Induction of genes mediating interferon-dependent extracellular trap formation during neutrophil differentiation
.
J Biol Chem
2004
;
279
:
44123
32
.
33.
Yousefi
S
,
Mihalache
C
,
Kozlowski
E
,
Schmid
I
,
Simon
HU
. 
Viable neutrophils release mitochondrial DNA to form neutrophil extracellular traps
.
Cell Death Differ
2009
;
16
:
1438
44
.
34.
Brinkmann
V
,
Abu Abed
U
,
Goosmann
C
,
Zychlinsky
A
. 
Immunodetection of NETs in paraffin-embedded tissue
.
Front Immunol
2016
;
7
:
513
.
35.
Albrengues
J
,
Shields
MA
,
Ng
D
,
Park
CG
,
Ambrico
A
,
Poindexter
ME
, et al
Neutrophil extracellular traps produced during inflammation awaken dormant cancer cells in mice
.
Science
2018
;
361
.
36.
Caudrillier
A
,
Kessenbrock
K
,
Gilliss
BM
,
Nguyen
JX
,
Marques
MB
,
Monestier
M
, et al
Platelets induce neutrophil extracellular traps in transfusion-related acute lung injury
.
J Clin Invest
2012
;
122
:
2661
71
.
37.
Hakkim
A
,
Fuchs
TA
,
Martinez
NE
,
Hess
S
,
Prinz
H
,
Zychlinsky
A
, et al
Activation of the Raf-MEK-ERK pathway is required for neutrophil extracellular trap formation
.
Nat Chem Biol
2011
;
7
:
75
77
.
38.
Xia
Y
,
Yeddula
N
,
Leblanc
M
,
Ke
E
,
Zhang
Y
,
Oldfield
E
, et al
Reduced cell proliferation by IKK2 depletion in a mouse lung-cancer model
.
Nat Cell Biol
2012
;
14
:
257
65
.
39.
Moffitt
RA
,
Marayati
R
,
Flate
EL
,
Volmar
KE
,
Loeza
SGH
,
Hoadley
KA
, et al
Virtual microdissection identifies distinct tumor- and stroma-specific subtypes of pancreatic ductal adenocarcinoma
.
Nat Genet
2015
;
47
:
1168
78
.
40.
Urban
CF
,
Ermert
D
,
Schmid
M
,
Abu-Abed
U
,
Goosmann
C
,
Nacken
W
, et al
Neutrophil extracellular traps contain calprotectin, a cytosolic protein complex involved in host defense against Candida albicans
.
PLoS Pathog
2009
;
5
:
e1000639
.
41.
Skubitz
KM
,
Campbell
KD
,
Iida
J
,
Skubitz
AP
. 
CD63 associates with tyrosine kinase activity and CD11/CD18, and transmits an activation signal in neutrophils
.
J Immunol
1996
;
157
:
3617
26
.
42.
Weckbach
LT
,
Grabmaier
U
,
Uhl
A
,
Gess
S
,
Boehm
F
,
Zehrer
A
, et al
Midkine drives cardiac inflammation by promoting neutrophil trafficking and NETosis in myocarditis
.
J Exp Med
2019
;
216
:
350
68
.
43.
Egberts
J-H
,
Cloosters
V
,
Noack
A
,
Schniewind
B
,
Thon
L
,
Klose
S
, et al
Anti-tumor necrosis factor therapy inhibits pancreatic tumor growth and metastasis
.
Cancer Res
2008
;
68
:
1443
50
.
44.
Jin
W
,
Xu
H-X
,
Zhang
S-R
,
Li
H
,
Wang
W-Q
,
Gao
H-L
, et al
Tumor-infiltrating NETs predict postsurgical survival in patients with pancreatic ductal adenocarcinoma
.
Ann Surg Oncol
2019
;
26
:
635
43
.
45.
Yang
L
,
Liu
Q
,
Zhang
X
,
Liu
X
,
Zhou
B
,
Chen
J
, et al
DNA of neutrophil extracellular traps promotes cancer metastasis via CCDC25
.
Nature
2020
;
583
:
133
38
.
46.
Cools-Lartigue
J
,
Spicer
J
,
McDonald
B
,
Gowing
S
,
Chow
S
,
Giannias
B
, et al
Neutrophil extracellular traps sequester circulating tumor cells and promote metastasis
.
J Clin Invest
2013
;
123
:
3446
58
.
47.
Warner
RB
,
Najy
AJ
,
Jung
YS
,
Fridman
R
,
Kim
S
,
Kim
H-RC
. 
Establishment of structure-function relationship of tissue inhibitor of Metalloproteinase-1 for its interaction with CD63: implication for cancer therapy
.
Sci Rep
2020
;
10
:
2099
.
48.
Khomiak
A
,
Brunner
M
,
Kordes
M
,
Lindblad
S
,
Miksch
RC
,
Öhlund
D
, et al
Recent discoveries of diagnostic, prognostic and predictive biomarkers for pancreatic cancer
.
Cancers
2020
;
12
.
49.
Neoptolemos
JP
,
Kleeff
J
,
Michl
P
,
Costello
E
,
Greenhalf
W
,
Palmer
DH
. 
Therapeutic developments in pancreatic cancer: current and future perspectives
.
Nat Rev Gastroenterol Hepatol
2018
;
15
:
333
48
.