Depletion of fibrinogen suppresses growth of primary tumors and metastasis of pancreatic ductal adenocarcinoma (PDAC)

Abstract

Background and Aims

Pancreatic ductal adenocarcinoma (PDAC) is an aggressive, highly metastatic disease that provokes dysregulation of the coagulation system. Patients exhibit significantly elevated circulating levels of blood clotting protein fibrin(ogen). Extravascular fibrin deposits contribute to the complex tumor microenvironment in PDAC.

Methods

We depleted fibrinogen in three PDAC patient-derived xenograft (PDX) models using technology platforms that are currently being tested clinically (antisense oligonucleotide or lipid nanoparticles containing siRNA) and monitored tumor growth and metastasis. Proteomics and spatial transcriptomics were used to interrogate the mechanisms behind the in vivo work.

Results

The role of fibrin on tumor progression was evaluated in vitro and in vivo and reduction of fibrin led to decreased tumor cell proliferation in vitro and significantly suppressed primary orthotopic tumor growth. Fibrin depletion provoked a significant shift in extracellular matrix (ECM)-associated proteins and serine protease inhibitors, suggesting a decrease in the activity of serine proteases known to be responsible for ECM remodeling and metastatic dissemination. Spatial transcriptomics revealed that tumors from fibrinogen-depleted mice exhibit significantly increased presence of stromal components, including tumor-restraining CAFs. Congruently, fibrinogen knockdown in a metastatic orthotopic model markedly impaired spontaneous metastasis to the liver. However, fibrinogen knockdown did not affect liver colonization in an intrasplenic injection model, which recapitulates the late stages of metastasis.

Conclusions

These data suggest that fibrin(ogen) reprograms the primary TME to support growth and promote early, but not late, metastatic steps. Our findings support prospective evaluation of a novel clinical approach involving the integration of fibrin(ogen)-targeting or depleting agents into chemotherapy regimens to control the spread of pancreatic cancer.

Graphical Abstract

Introduction

Pancreatic Ductal Adenocarcinoma (PDAC) has one of the highest mortality rates of all cancers, with a 5-year overall survival of just 13%, and a dismal 3% for patients with metastatic disease.¹ Surgical resection is the only curative option, however, a majority of patients present with unresectable, locally advanced or metastatic disease. In addition to its aggressive nature, PDAC patients often present with aberrantly activated coagulation, and 12-36% of PDAC patients experience venous thromboembolism (VTE), both of which have been linked with cancer progression and poor prognosis.²

The blood clotting system is activated following vascular injury and terminates in the generation of the active serine protease thrombin which then cleaves the provisional matrix protein fibrinogen into fibrin monomers that spontaneously assemble into insoluble fibrin matrix. Fibrin is the primary structural component of the blood clot, which is essential for hemostasis and tissue repair following injury. Fibrinogen is present in circulation at 2-4 mg/ml but is also an acute phase reactant in which levels can significantly increase (>8 mg/ml) during acute and chronic inflammation. Fibrinogen is synthesized by the liver being composed of six polypeptide chains: two Aα chains (encoded by Fga), two Bβ chains (encoded by Fgb), and two γ chains (encoded by Fgg). Complete assembly into a heterohexamer (Aα₂Bβ₂γ₂) is required for secretion of fibrinogen into circulation.³ Notably, plasma fibrinogen is elevated in PDAC, and high circulating fibrinogen concentrations have been positively associated with increased metastasis and decreased survival rates.⁴ Despite these intriguing correlations, precise mechanisms linking elevated fibrin(ogen) to PDAC pathogenesis remain undefined.

Two common mutations leading to the development of PDAC are activation of oncogenic KRAS and inactivation of TP53 (>90% and >70% respectively), and both mutations increase the level of tissue factor (TF), the initiator of the blood clotting system, which generates thrombin, and ultimately fibrin, in the PDAC tumor microenvironment (TME).^{5, 6} PDAC tumors contain significant extravascular fibrin deposits in the TME. The basis of fibrin deposits in the PDAC TME are linked to the “leaky” tumor vasculature and high TF expression by tumor and stromal cells including endothelial cells and tumor-associated macrophages.^7-10 In cancer, extravascular fibrin deposits contribute to tumor growth through multiple mechanisms.^{11, 12} Extracellular fibrin serves as a ligand for adhesion molecules and integrins to advance disease progression.¹³ Moreover, once tumor cells enter the vasculature they can attract fibrinogen and platelets to shield the tumor cells from anti-tumor immune cells.¹⁴ Several studies across different tumor types have demonstrated that reducing fibrin levels or enhancing fibrinolysis can improve drug delivery or tumor perfusion, highlighting the relevance of targeting fibrin.^15-17 Although fibrin is an important barrier in cancer, there are few studies directly targeting fibrin depletion in this disease context. Thus, although the concept of modulating fibrin has precedent in the field, our study is distinct in that it uncovers the molecular mechanisms linking fibrin to PDAC tumor biology, thereby providing novel insight into how fibrin depletion could be harnessed in future therapeutic strategies.

We investigate the ability of fibrinogen-lowering agents to suppress primary tumor growth and metastasis of PDAC using multiple orthotopic models and two independent methods of fibrinogen depletion. Both fibrinogen-specific antisense oligonucleotide (Fib ASOs) and hepatocyte-targeting lipid nanoparticles (LNP) carrying siRNAs against fibrinogen have significant clinical implications as both of these technology platforms are being tested clinically.^{18, 19} Proteomics and spatial transcriptomics were employed to unravel the impact of fibrinogen depletion on ECM composition and the overall architecture of the TME. Clinically, incorporation of such agents targeting fibrinogen synthesis may enhance the efficacy of existing standard of care therapies by potentially restricting tumor metastasis and ultimately lead to an improved prognosis for patients.

Materials and Methods

Orthotopic Patient-derived Xenografts (PDX) Models

Depletion of coagulation proteins prior to tumor implant

In vivo studies in mice were conducted in compliance with the National Institutes of Health guidelines and were approved by the Institutional Animal Care and Use Committee of Indiana University School of Medicine (IACUC Protocol #21165) as outlined in ²⁰.

All animal surgeries were performed by the Preclinical Modeling and Therapeutics Core at IUSCCC. PDX21 and PDX33 tumor engraftment models were generated from freshly resected tumors from PDAC patients undergoing surgery at IU Health (Table S1). Tumor tissues were surgically implanted orthotopically in the pancreas of immunocompromised NSG mice that were sex-matched with the patient.

In the PDX21 model, female mice were intraperitoneally treated with 7.5 mg/kg of Acetylgalactosamine (GalNAc) control antisense oligonucleotide (Control ASO, n=13) or Fibrinogen b-chain (Fgb)-targeted (Fib ASO; n=10) twice a week for 14 days to ensure fibrinogen depletion before tumor engraftment, and treatment was continued until the end of the study. ASOs were conjugated with the GalNAc ligand to improve targeting and uptake by hepatocytes.^21-23 In the PDX33 model, female mice received a single dose of 2 mg/kg control lipid nanoparticle (Control LNP, n=12) or Fga-targeting siRNA LNP (siFga LNP, n=10) for fibrinogen depletion, followed by pancreatic tumor engraftment, and treatment was continued for eight weeks.²⁴

In the metastatic Pa03C model, male NSG received Pa03C-GFP-Luc cell (1.3 × 10⁴) injection into the pancreas and treatment with Control (n=20) or Fib ASO (n=21) twice a week with 7.5 mg/kg from two weeks prior to tumor injection until necropsy at week 5.

Depletion of fibrinogen after orthotopic implant

One week after implant, mice were administered 2mg/kg of Control (n=11) or siFga LNP (n=11) intravenously every 10 days. Tumor, liver, and lung were harvested at necropsy and used in histological, RNA, or protein analysis.

Experimental Metastasis Models

For lung colonization studies, mice received tail vein injections of Pa03C-GFP-Luc (2x10⁵ cells) and were treated with ASOs twice a week from two weeks prior to tumor injection. For liver colonization studies, mice were pretreated either ASOs or LNPs, followed by a single intrasplenic injection of Pa03C-GFP-Luc (3.5x10⁴ cells).^{25, 26} Treatment was continued until necropsy at week 4. After 4 weeks, lung or liver was harvested and fixed for histology.

Mass Spectrometry (LC-MS/MS)

Pa03C tumors from mice treated with Control (n=8) or Fib ASO (n=8) were used for proteomic analysis. Sample preparation, mass spectrometry, analysis of raw files and peptide purification were performed in collaboration with IUSM Center for Proteome Analysis.²⁷

Histology

For the PDAC TMA, IUSCCC Biospecimen Collection and Banking Core provided tissues from PDAC patients (August 2010 to February 2021). Patient consent was obtained under institutional biobanking collection protocols for use of biological specimens in cancer research. To evaluate the association between tumor and normal adjacent tissue, a logistic regression model was used following fibrin staining in 412 samples. Tumor samples from the in vivo studies were analyzed using digital pathology.

Spatial Transcriptomics (ST)

Tumor tissue was harvested for ST four weeks after implant from mice that were treated with Control or Fib ASO and embedded in OCT for downstream analysis of human and mouse gene expression.

Flash frozen tissue sections were collected at necropsy, and quality of the tissues were confirmed using RNA integrity number (RIN) scores. Only tissues with RIN scores greater than 7 were used for this experiment. Tissue sections were first mounted on optimization slides to determine the optimal permeabilization time. After optimization, sections were mounted onto the capture areas of Visium Spatial Gene Expression slides (10x Genomics). The sections then underwent methanol fixation, Hematoxylin and Eosin (H&E) staining, and high-resolution imaging, following the Visium protocol (CG000160). Spatial gene expression libraries were prepared according to the Visium Spatial Gene Expression Reagent Kits User Guide (CG000239). Briefly, mRNA from permeabilized sections was captured by spatially barcoded primers located on the slide capture spots. Full-length cDNA was synthesized from polyadenylated mRNA and subsequently amplified.

Sequencing libraries were constructed from the amplified cDNA through enzymatic fragmentation, size selection, end repair, A-tailing, adaptor ligation, and PCR amplification. The resulting dual-indexed libraries were quantified using Qubit and assessed for quality using the Agilent Bioanalyzer 2100. Final libraries were sequenced on an Illumina NovaSeq X PLUS platform with input from The Center for Medical Genomics at IUSM.

Raw FASTQ files and accompanying H&E images were processed with SpaceRanger (10x Genomics). Reads were aligned to a dual-species reference (GRCh37 + mm10), after which unique molecular indices (UMIs) were assigned to individual Visium spots to generate spot-level gene-expression matrices.

Because a matched single-cell reference was unavailable, initial compartment identification used the unsupervised spatial deconvolution tool stDeconvolve. For each inferred “topic,” we inspected the top-ranked genes, classified them as human- or mouse-specific, and labelled the corresponding spots as cancer (human) or stromal (mouse). Topic-wise proportions were then aggregated to yield the total cancer and stromal fractions per slide.^{28, 29}

To resolve stromal heterogeneity—particularly the balance between restraining CAFs (restCAFs) and tumor-permissive CAFs (permCAFs)—we next applied Robust Cell Type Decomposition (RCTD) with a human PDAC single-cell RNA-seq reference containing cancer cells, restCAFs, permCAFs, macrophages, endothelial cells, and acinar cells, etc.³⁰ Given interspecies differences in the tumor microenvironment, absolute cell-type estimates were unreliable; therefore, we restricted interpretation to the relative abundance of restCAF signatures, which we believe that RCTD could provide a relative estimation of the signals.

Statistics

All statistical calculations were performed using GraphPad Prism 10. For comparison within the different groups Mann-Whitney or Welch’s t-test was used. For proteomics work, sample preparation, mass spectrometry, and differential expression of protein abundance was conducted using t test with p<0.05 as significance cutoff. Pathway enrichment analysis was conducted using a hypergeometric test against the MsigDB v6 human and mouse canonical pathway sets and Gene Ontology (GO) and p<0.05 was used as significant cutoff.

For additional details, see Supplemental Methods.

Results

Intensity of fibrin staining is higher in tumor tissue, correlates with OS, and fibrinogen depletion significantly impairs PDAC tumor growth.

A PDAC tissue microarray (TMA) revealed a clear distinction in fibrin deposition between tumor (n=165) and adjacent normal (n=103), indicating a robust and significant association between staining intensity and tumor status (Fig. 1A).²⁰ The association between intensity of staining and tumor status was evaluated using a logistic regression model which compared fibrin staining intensity to normal adjacent pancreatic tissues. From samples in the TMA, we further analyzed survival information from 74 of these PDAC patients which revealed that expression levels of fibrin directly and negatively correlated with overall survival (OS), confirming fibrin in the TME as a potential driver of PDAC disease severity (Fig. 1B).

Fig. 1: Fibrin staining is higher in tumor tissue, correlates with overall survival (OS), and fibrinogen depletion significantly impairs PDAC tumor growth in two patient-derived xenograft (PDX) models.

(A) Fibrin staining intensity and tumor status based on logistic regression analysis: n=165 PDAC, n=103 Normal. (B) Kaplan-Meier curves illustrating the association between fibrin staining and OS in PDAC. Representative sections of PDAC tissue and corresponding PDX tissue: H&E (C) and fibrin (D) Scale bars 300 μm. Experimental strategy for Control or Fib ASO (E); Fibrinogen was depleted before tumor engraftment. (F) Orthotopic tumor mass five weeks after implant. (G) Representative fibrin(ogen) stained histological sections (left) and quantitation (right) of PDX21 tumors after treatment. Scale bar 500 μm. (H) qRT-PCR analysis of fibrinogen expression in liver. (I) Plasma ELISA for fibrinogen at necropsy from PDX21 mice. Experimental strategy for Control or siFga LNP (J); Mice were treated with LNPs prior to tumor implant. (K) Orthotopic tumor mass 10 weeks after engraftment of PDX33. (L) Representative fibrin(ogen) stained histological sections (left) and quantitation (right). Scale bar 500 μm. (M) qRT-PCR analysis for fibrinogen expression in liver. (N) Plasma ELISA for fibrinogen at necropsy. Data are presented as mean ± SEM and analyzed using a Welch’s t-test with *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

To investigate the effects of fibrinogen depletion on PDAC growth, two PDX models were utilized. Histologic characterization of the surgically resected specimens in comparison to the PDX tissue confirmed histologic, morphologic, and fibrin deposition remained intact in the PDX models (Fig. 1C, D). These human tumors elicit the deposition of fibrin into the TME as observed in the microenvironment of human PDAC and validates the use of these mouse models to study fibrin-mediated pathways in disease progression.

To determine whether pancreatic tumor growth was dependent on fibrinogen, mice were pretreated with control or Fib ASO prior to tumor engraftment of PDX21 to the pancreas (Fig. 1E). Tumor mass was significantly lower in Fib ASO-treated mice compared to control group (Fig. 1F), and fibrin deposition was significantly reduced (3.6-fold) in the tumors of Fib ASO-treated mice (Fig. 1G). Expression and protein analyses revealed a 5-fold fibrinogen reduction in liver mRNA (Fig. 1H) and 7-fold reduction of circulating fibrinogen, confirming the Fga ASO was proficient at reducing fibrinogen levels (Fig. 1I).

To further confirm the contribution of fibrin to PDAC disease severity, an additional orthotopic model exhibiting moderate fibrin deposition and LNPs containing fibrinogen α-chain targeting siRNA (siFga) was used (Fig. 1J). Similar to Fib ASO-treated mice, PDX33 tumors showed significantly reduced tumor mass upon fibrinogen depletion (Fig. 1K). Fibrin deposition within tumors was 6-fold reduced in the siFga LNP group relative to controls (Fig. 1L), with a corresponding reduction in Fga mRNA in liver (6.6-fold, Fig. 1M), and reduction of fibrinogen protein in circulation (12-fold lower, Fig. 1N). Collectively, these data suggest fibrin deposition is an important component of the TME and impacts tumor growth.

Fibrinogen depletion significantly suppresses growth and liver metastasis of human Pa03C tumors in vivo.

PDX21 and PDX33 models did not show significant metastasis to the liver and lungs. Therefore, a third orthotopic model was employed to investigate the role of fibrinogen in growth, as well as metastasis, of pancreatic tumors. In the orthotopic model using Pa03C cells, tumor cells spontaneously metastasize to the liver and lungs recapitulating human PDAC.³¹ Using this model, circulating plasma fibrinogen was monitored for 3 weeks in non-tumor-bearing and tumor-bearing mice to assess tumor-stimulated changes in fibrinogen levels (Fig. 2A). As expected, the mass of the pancreas significantly increased over time in tumor-bearing mice (Fig. S1A, 2B). Fibrinogen levels in sham mice remained relatively constant with a slight drop between weeks 1 and 2 (Fig. S1B). In contrast, fibrinogen levels steadily increased over time with increasing pancreatic tumor mass, reaching levels ~2-fold higher in week 3 relative to week 1 (Fig. 2C), consistent with an ongoing, persistent inflammatory challenge. Tumor-bearing mice also had significantly higher fibrinogen levels compared to sham, aligning with clinical findings in PDAC patients, where plasma fibrinogen levels correlate with disease progression.⁴ Consistently we observed that in mice with depleted fibrinogen (Fig. S1C, S1D, 2D), tumor mass as well as primary tumor burden was significantly reduced as reflected by bioluminescence imaging (BLI, Fig. 2E, F, S1E). Fibrin deposition in tumors was also dramatically reduced in fibrinogen depleted mice (5-fold, Fig. 2G).

Fig. 2: Fibrinogen depletion significantly suppresses growth and liver metastasis of Pa03C tumors in vivo.

(A) Mice were given orthotopic injections of PBS (Sham) or Pa03C-GFP-Luc cells (Tumor). (B) Pancreas mass in tumor-bearing mice. (C) Plasma ELISA for fibrinogen from Pa03C model. Experimental strategy using ASOs (D) Pa03C-GFP-Luc cells were implanted into mice after fibrinogen depletion. (E) Orthotopic tumor mass of mice treated for 5 weeks. (F) BLI analysis of primary tumor burden using a piece-wise mixed-effect regression model. ROI are expressed as changes in Area under the curve (AUC) Total Flux. (G) Representative fibrin(ogen) stained (black arrows) sections (left) and quantitation (right) of tumors. Scale bar 1mm, 500 μm. (H) BLI analysis of liver. (I) Representative H&E-stained sections and quantitation of metastases (black arrows) within liver. Scale bar 2mm. Data are presented as mean ± SEM and analyzed using a Mann-Whitney or Welch’s t-test with *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

In addition to primary tumor growth, depletion of fibrinogen significantly reduced metastatic burden to the liver (Fig. 2H,I, S1F). Interestingly, there was no difference in lung metastases following fibrinogen depletion (Fig. S1G). Data from three orthotopic PDX models demonstrates that deposition of fibrin matrices in the microenvironment is a driver of tumor growth and liver metastasis in pancreatic cancer.

Fibrinogen depletion increases expression of matrix proteins and promotes remodeling of the matrix within the tumor and its TME.

In order to understand differentially expressed proteins in tumors from fibrin(ogen)-depleted mice compared to controls, tumors were analyzed using proteomics. Volcano plot comparing proteins differentially expressed revealed a decrease in 129 proteins, and an increase in 245 proteins upon depletion of fibrinogen (Fig. 3A). Multidimensional Scaling (MDS) analysis suggested a distinct profile of the abundance of ECM and stromal-related proteins in tumors from fibrin-depleted mice as compared to a less distinct difference between the two groups in the MDS plot of all proteins (Fig. 3B, S2A). The major difference in differentially expressed proteins in tumors from fibrinogen-depleted mice is in stromal and ECM-related proteins. Gene Set Enrichment Analysis (GSEA) demonstrated a significant differential expression of proteins in complement and coagulation cascades (p=6.83e-20), suggesting depletion of fibrin directly affects proteins within the coagulation system and inflammatory responses.

Fig. 3: Fibrinogen depletion increases the expression of matrix and matrix-associated proteins and promotes remodeling of the matrix within the tumor and its TME.

(A) Volcano plot illustrating protein abundance differences in Pa03C tumors from mice treated with Fib ASO relative to Control; (B) Multidimensional data analysis (MDA) plot showing an abundance of ECM proteins in tumors; Pathway enrichment analyses of (C) up-regulated and (E) down-regulated proteins of mouse (host) origin. GSEA analysis showing gene set enriched in the upregulated host matrisome (D) and downregulated proteins (F). Pathway enrichment analyses of (G) up-regulated and (I) down-regulated proteins of human (tumor) origin. GSEA analysis showing gene set enriched in matrisome or integrin (H) and down in DNA replication and DNA repair of tumor origin (J).

Changes in protein expression from the tumor (human) as well as host TME (mouse) can be observed. From the host TME, a remarkable upregulation in matrisome and ECM-related pathways is observed in tumors harvested from fibrinogen-depleted mice. These include matrisomal proteins (p=4.75e-35), ECM regulators (p=2.23e-16), integrins (p=2.85e-12), focal adhesion (p=8.73e-11), complement pathway (p=2.71e-13), and collagen formation (p=1.93e-7) proteins, further implying that depletion of fibrinogen triggered a cascade of ECM remodeling events in the TME (Fig. 3C,E). We also observed significant downregulation of proteins that enrich in pathways of mRNA processing and splicing (p=6.52e-7), transcription (p=6.02e-6), DNA repair (p=1e-4), and amino acid metabolism (p=0.003, Fig. 3D,F). Pathway enrichment analysis on human tumor proteins indicated upregulation of matrix-associated pathways in tumors from fibrinogen-depleted mice. Key observations include upregulation in human matrisomal proteins (p=2.9e-13), ECM regulators (p=1.99e-7), collagens (p=4.73e-5), and inflammatory proteins (p=0.0004), and downregulation in mRNA splicing (p=1.38e-20), DNA replication (p=5.37e-05), DNA repair (p=7.0e-5), cell cycle checkpoint (p=0.002), glycosphingolipid metabolism (p=0.0004), and glycosaminoglycan degradation (p=0.005) (Fig. 3G-J).

The decrease in proteins associated with DNA repair, cysteine and methionine metabolism, and glycosaminoglycan degradation are mainly found within tumor cells. While in the host TME, proteins within the pathways of sulfur amino acids metabolism, alanine, aspartate and glutamate metabolism are downregulated compared to control, suggesting different metabolic responses in cancer cells and host TME in response to fibrinogen depletion. Complete statistics of differentially expressed proteins and pathways are in Table S4-13. While certain ECM molecules are also expressed by cancer cells, extensive upregulation of matrisome and ECM-related proteins suggests fibrin is driving remodeling in the host TME. Consistent with this idea, several of upregulated proteins are collagens and other ECM molecules which are primarily synthesized by fibroblasts. There are a few reports that perhaps tumors can make fibrinogen, therefore expression of Fib chains α, β, and γ were evaluated in several PDAC samples as well as CAFs. In comparison to human liver, expression of all chains of fibrinogen were extremely low reflecting that the fibrinogen was indeed coming from the liver in our system (Fig. S2B). Removal of fibrin from the TME results in upregulation of proteins within pathways involved in ECM remodeling and is likely driven by crosstalk between the tumor and host stromal cells.

Depletion of fibrinogen mediates different signaling pathways in tumor cells compared to the host TME with TME changes consistent with remodeling of ECM.

Delving deeper into the differences between cancer cells and the host TME in response to fibrinogen depletion, pathways were ranked based on the differences of p values of their enrichment by human and mouse proteins. Distinct signaling pathways were more specifically enriched in human tumor cells vs host TME upon fibrinogen depletion. Within tumors, pathways of PD-1 signaling, cell-cell communication, fibrinolysis, integrin 5, apoptosis, Jak/STAT, and NFκB pathways are significantly enriched, while only cell-cell communication is slightly enriched within murine proteins (Fig. 4A). Strong activation of PD-1, Jak/STAT, NFκB and integrin signaling pathways suggests a role for fibrin in regulating the immune response via affecting signaling pathways in tumor cells. The models used above were all in immunodeficient mice therefore we also employed a subcutaneous flank model with KPC2 cells in Fibrinogen deficient mice and demonstrated that the tumors were indeed significantly smaller (Fig. 4B, 2.4-fold).

Fig. 4: Depletion of fibrinogen mediates different signaling pathways in tumor cells compared to the host TME with TME changes consistent with remodeling of the ECM.

(A) Pathways specific to human proteins (left) that are significantly upregulated with Fib depletion, in comparison to these pathways in the host TME (right). (B) Subcutaneous tumor mass of KPC2 in Fib -proficient or -deficient mice. (C) Pathways that are enriched in the host TME (left) in comparison to these pathways in human tumors (right). The red dotted line indicates p<0.05. Heatmaps showing the expression changes of matrisome genes involved in ECM remodeling derived from the tumor (D) or from the host TME (E) as well as Serpins with significant changes (F). (G) Western blot analyses and quantification of fibronectin and laminin from Pa03C orthotopic tumors. Each lane represents an individual tumor. (H) Representative sections (top) and quantitation (bottom) Emilin1 (black arrows), Scale bar 200μm. Rheometry of tumors following Fib depletion, elastic (I) and viscous (J) behavior. IHC quantitation of macrophage marker F4/80 (K) or platelet marker CD41 (L). Data are presented as mean ± SEM and analyzed using a Welch’s t-test with *p<0.05, **p<0.01.

Similarly, the most enriched pathways within the host TME were compared to those pathways within tumors. Although many of the pathways in Fig. 4 are shared, the degree of upregulation is much more dramatic in the mouse protein expression. One of the most upregulated pathways in both tumor and TME was the matrisome as shown in the heatmaps (Fig. 4D, E). Differential expression analysis revealed a significant increase in expression levels of structural proteins of the ECM that are involved in remodeling such as several subtypes of collagens, laminins, and glycoproteins in both human tumor cells and host stromal microenvironment (Fig. 4D,E). The most significantly upregulated pathway was the matrisome in both tumor and host TME, further indicating that fibrin is a modulator of ECM remodeling between tumor and host stromal microenvironment. Finally, alongside changes in ECM proteins, a concurrent and significant upregulation of protease inhibitors that prevent degradation of the ECM was observed.^{32, 33} Specifically, depletion of fibrinogen enhanced expression of the Serpin family in the TME and the tumor (Fig. 4F). These proteases contribute to PDAC pathogenesis, including MMPs as well as coagulation and fibrinolytic processes and may explain the increased ECM observed upon depletion.^{20, 34}

Next, we sought to validate the proteomics data as well as investigate activation of other pathways previously shown to be downstream of fibrin signaling. Western blot analysis revealed a significant upregulation of fibronectin and laminin in fibrin-depleted tumors (Fig. 4G). Likewise, the deposition of Emilin1, a tumor suppressor, was significantly higher (Fig. 4H).

Given the contribution of ECM proteins to structural integrity of the TME and their critical role in determining mechanical properties, we next evaluated stiffness of the Pa03C tumors. Storage modulus is used to evaluate the elastic behavior of the tumors, while loss modulus is an indicator of the viscous behavior of tumors. Both storage and loss modulus ratios were almost twice as high in tumors from fibrinogen-depleted mice, indicating loss of fibrin matrices generated stiffer, but smaller and less metastatic, tumors (Fig. 4I,J). However, this increased stiffness did not act as a barrier to infiltration of immune cells or tumor-associated platelets (Fig. 4K,L, S3A,B). Overall, loss of fibrin from the PDAC matrix changed the composition of the ECM and cumulated in impeding growth and progression of these PDAC tumors.

Loss of fibrin from the TME diminishes proliferation of pancreatic cancer cells and CAFs in vitro and promotes remodeling in vivo, increasing tumor-restraining CAF-rich stroma.

Crosslinking of fibrin and collagen promotes the formation of a fibrotic network within the stroma thereby enhancing mechanical properties of the ECM.³⁵ GSEA revealed increased collagen formation pathways in tumors from fibrinogen-depleted mice (Fig. S3C), confirmed by IHC with elevated collagen and collagen-producing αSMA+ myofibroblasts (Fig. S3D,E). In tumors collected when control mice were moribund, significant changes in proliferation, apoptosis, or expression of EMT markers were not observed, except for downregulation of Slug (Fig. S4).³⁶

To examine the impact of fibrin deposition on PDAC stromal composition, we investigated the in vitro proliferation of pancreatic cancer cells (PCCs) and CAFs using an engineered tumor-microenvironment-on-chip (T-MOC) platform.³⁷ PDAC stroma was reconstituted within the interstitial channel, where PCCs and CAFs were embedded in either a collagen-fibrin or collagen-only matrix. To mimic physiological coagulation, thrombin and cell culture medium were perfused through two side channels. Based on the in vivo models, Pa03C +/− CAF19 and Pa21 with their patient-matched CAFs were used in this 3D co-culture T-MOC system (Fig. 5A). In the co-culture, exclusion of fibrin from the matrix significantly diminished both tumor cell and CAF proliferation (Fig. 5B, S5). In order to further decipher the impact of fibrin on the stroma and tumor compartments, we performed spatial transcriptomics on Pa03C tumors in vivo (Fig. 5C-E). As a representative of a tumor cell marker, EPCAM expression is shown in Fig. 5D and TGFBR3 as a marker of restCAFs in Fig. 5E. By using STdeconvolve, we quantified the relative proportions of stromal and cancer components in each spatial spot. Consistent with proteomic findings, stromal components were significantly more abundant in tissue harvested from fibrinogen-depleted mice than in the control samples (p<10E-6, Fig. 5F). Conversely, the cancer component is significantly less in tissues harvested from fibrinogen-depleted mice compared to control samples (p<10E-6, Fig. 5F)). To further assess stromal heterogeneity, we applied DeCAF and RCTD³⁰, which revealed that tumors from fibrin-depleted mice harbored higher proportions of CAFs characteristic of tumor-restraining CAFs, rather than tumor-permissive (p<10E-6, Fig. 5G). These data reflect that removal of fibrin from the matrix alters the ECM and directly affects both CAFs and PCCs. These findings indicate that fibrin plays a crucial role in modulating cell growth and activity, reinforcing the notion that fibrin depletion can create an anti-tumor environment, as demonstrated in the in vivo studies.

Fig. 5: Loss of fibrin from the TME diminishes proliferation of pancreatic cancer cells and CAFs in vitro and promotes remodeling in vivo, increasing tumor-restraining CAF-rich stroma.

(A) Schematic of T-MOC system to recapitulate 3D PDAC Stroma. (B) Confocal images of T-MOC system. Fluorescent micrographs on Day 6 and cell growth quantification of Pa03C and CAF19, Scale bars 200μm. (C) H&E-stained tumor samples for spatial transcriptomics. Scale bar 400 μm. Cell marker gene expression level for tumor: EPCAM (D) and for restCAF: TGFBR3 (E); (F) Estimated host TME and human PDAC cell proportions across all samples by RTCD. (G) Proportion of restCAF in tumors from Control and Fib ASO treated mice. Significant difference between the two groups, with p<10E-6. Data are presented as mean ± SEM and analyzed using a Welch’s t-test with *p<0.05, **p<0.01.

Late stages of metastasis of Pa03C cells are not affected by fibrinogen depletion.

Disease progression through cancer cell metastasis involves tight regulation of a series of intricate steps that allow tumor cells to disseminate from the pancreas, followed by blood vessel intravasation, extravasation at a distant vascular bed, seeding, and finally, colonization into secondary sites. Due to the dramatic impact on metastasis in the spontaneous model, two experimental models of metastasis were employed: a tail vein injection model that results in lung metastasis and the intrasplenic injection model that results in liver metastasis. Using the tail vein model, we observed no differences in micrometastases or macrometastasis within the lungs of control or Fib ASO treated mice (Fig. S6A-F).

Similarly, depletion of fibrinogen with both ASOs and LNPs (Fig. 6A,B) followed by injection of tumor cells into the spleen did not significantly alter colonization of liver parenchyma (Fig. 6C, D). These findings are in contrast to our spontaneous metastasis model where fibrinogen depletion significantly limited liver metastasis (Fig. 2H, I). Plasma fibrinogen was significantly reduced (Fig. 6E, F). This implies that in this model of PDAC fibrin(ogen) establishes a tumor-promoting TME that supports primary tumor growth and enhances early metastatic steps for tumor cells to expand into the liver.

Fig. 6: Late stages of metastasis of Pa03C PCCs are not affected by fibrinogen depletion.

Pa03C cells were implanted into the spleen after pre-treatment with ASOs (A) or LNPs (B). Treatment was continued post-implant for 5 weeks until necropsy. Representative H&E-stained sections and quantitation of liver colonization (black arrows) from mice treated with ASO (C, Scale bar 5mm) or LNP (D, Scale bar 2mm). Plasma ELISA for fibrinogen at necropsy, ASOs (E) or LNPs (F). Data are presented as mean ± SEM and analyzed using Welch’s t-test with **p<0.01 and ***p<0.001.

Treatment with siFga LNP significantly impedes tumor progression of established human Pa03C tumors.

The data presented thus far was in the experimental models in which depletion of fibrinogen occurred prior to tumor implant. Figure 7 demonstrates the efficacy of fibrinogen reduction in a subject with an established tumor (Fig. 7A). Importantly for clinical translation, primary tumor mass and metastatic burden to liver was again significantly reduced even with depletion of fibrinogen following tumor implant (Fig. 7B,C, S7A), and with corresponding two-fold decrease in fibrin deposition within the TME (Fig. 7D) and significantly lower (4-9-fold) circulating plasma fibrinogen (Fig. S7B). Again, we observed no differences in metastatic lesions in lung (Fig. S7C).

Fig. 7: Treatment with siFga LNP significantly impedes tumor progression of established human Pa03C tumors.

(A) Experimental design to study efficacy of treatment with siFga LNP. One week after Pa03C orthotopic tumor growth, mice were administered Control or siFga LNPs. (B) Orthotopic tumor mass at week 5. (C) Representative H&E-stained sections (top) and quantitation (bottom) of Pa03C metastases (black arrows) within liver. Scale bar 300 μm (D,). Representative fibrin(ogen) stained sections (top) and quantitation (bottom) of Pa03C tumors. Scale bar 500 μm. Data are presented as mean ± SEM and analyzed using Welch’s t-test or Two-Way ANOVA with *p<0.05.

Discussion

The work presented here documents that deposition of fibrin within the complex PDAC TME plays a role in tumor growth and disease progression. Systemic fibrinogen depletion impaired primary tumor growth and liver metastasis in three independent, orthotopic PDAC models. This effect was achieved using two clinically translatable therapies to lower circulating levels of fibrinogen- ASOs and LNPs. Importantly, in the PDX model with pre-existing orthotopic tumors, the decrease in tumor growth and metastatic burden to liver was maintained upon fibrinogen depletion.

Targeting fibrinogen triggered a cascade of events leading to extensive ECM remodeling in primary Pa03C tumors accompanied by a significant increase in the stromal cell : tumor cell ratio and an increase in CAFs with a tumor-restraining phenotype. These findings underscore the influence of fibrin on both tumor cells and the surrounding stroma, including CAFs. Although tumors were smaller in the Fib depleted mice, increased ECM and matrisome suggest potential compensatory mechanisms within the TME that remain protective to the tumor. Therefore, we will continue exploring the most effective strategies to kill the tumor, combining fibrinogen depletion with chemotherapy and stromal-modulating agents. Finally, marked upregulation was observed for protease inhibitors that would further impede ECM degradation and remodeling necessary for tumor expansion and dissemination.

Increased ECM, collagen, and stiffness of tumors have been typically associated with disease progression. However, tumors from fibrin-depleted mice were smaller and less metastatic further supporting that precise nature of stromal changes can play a critical anti-tumor role. This effect appears to result from comprehensive ECM remodeling that includes an increase in the proportion of restCAFs, ultimately contributing to restrained tumor progression. Hence, as a field, dissection of pro- vs anti-tumor CAF phenotypes and ECM changes has become of upmost importance. Distinct collagen types and composition as well as cleaved collagen within the PDAC TME exert specific functions, influencing the aggressiveness of the disease.³⁸ We observed a dramatic upregulation of collagens 1A2, 2A1, 4A3, 5A2, and 7A1 supporting the concept that collagen deposition can also have tumor-restricting functions. Our findings are novel, yet corroborate existing studies that have reported tumor-suppressive roles for collagen in PDAC.^{39, 40} The source of the collagen that contributes to tumor progression is also important – is the tumor making it or the CAFs?^{40, 41} The biochemical makeup of the ECM changes as the tumor begins to produce ECM components instead of predominantly the fibroblasts. Together, our observed increase in ECM proteins and restCAF population upon fibrinogen depletion indicates changes in intrinsic tumor-CAF signaling which transforms ECM biomechanical behavior to limit growth and metastasis. More work is needed to understand whether loss of fibrin or upregulation of these proteins is driving this phenotype.

A potential complication to this approach is the risk of bleeding. However, PDAC patients tend to have excessively high fibrinogen levels,⁴² therefore this approach may bring the fibrinogen levels back in normal range and offer therapeutic benefit without undue risk. ASO or LNP treatments did not increase bleeding risks in our models, addressing one of the major safety concerns with targeting clotting factors. We previously documented that hypofibrinogenemia induced by siFga LNPs to levels similar to our studies was consistent with a preservation of hemostasis using multiple vessel injury models.⁴³ Selectively reducing fibrinogen may offer significant advantages for suppressing thromboinflammatory diseases (including cancer) over traditional anticoagulants that target the common pathway enzymes of the coagulation cascade as fibrinogen depletion would leave platelet function unperturbed.

By targeting fibrinogen production in the liver, our approach bypasses the stromal barrier and prevents fibrin‘s contribution to the PDAC TME. While a dramatic difference in metastatic burden to liver exists, similar effects were not observed in lung. These differences may be explained by fibrin’s role in reprogramming tumor cells such that they are less suited to grow in liver compared to lung, or simply inherent physiological variations in the microenvironment of liver versus lung tissues. Our studies were also conducted in immunodeficient mice and the presence of fibrin matrices in tumors can suppress immunity in cancer by serving as a protective scaffold, although studies here do not show differential recruitment of macrophages or platelets in the absence of fibrin(ogen). However, in the immunocompetent model, complete knockout of fibrinogen did not impair primary tumor growth of orthotopically implanted KPC cells, but did show a statistically significant diminution in a subcutaneous tumor growth model (Fig. 4B) and in metastasis to the lung following tail vein injection.²⁰

Future in vivo studies will investigate how tumors from mice depleted of fibrinogen will be respond to standard of care (SOC) such as FOLFIRINOX or Gemcitabine / Abraxane. PDAC patients could be a potential target population for advancing the use of siFga LNP as an anti-cancer agent. LNPs containing siRNAs are a well-established therapeutic platform with regulatory precedence, with infrastructure in place for large-scale manufacturing. A clinically relevant LNP formulation was used to deliver the siFga in these studies, comprised of lipids used in FDA-approved RNA drugs delivered using LNPs.^44-46 Findings presented here support the presence of fibrin(ogen) in the TME as a significant driver of disease progression and identify it as a therapeutic target for mitigating tumor growth and metastasis. Using multiple other modalities in combination with drugs such as Fib-ASO or siFga offers new translatable therapeutic strategies to extend overall survival in patients with PDAC. This new treatment approach can greatly affect how we manage PDAC clinically, providing new strategies to mitigate PDAC metastasis and improve surgery outcomes.

Data availability:

Raw and processed LC-MS/MS data have been uploaded to ProteomeXchange partner MassIVE with repository accession MSV000096002. Reviewer username and password are: MSV000096002_reviewer; password: Fibrinogen. The code used to analyze will be available in GitHub. The data related to the ST are available at DOI 10.5281/zenodo.17108058. All other data analysis and protein lists are provided in the Supplemental files.