C57BL/6 mice were obtained from Jackson Laboratory. Human EPO-overexpressing mice were generated as previously described. 14 TgN(PDGFBEPO)321Zbz consists of a transgenic mouse line, TgN(PDGFBEPO)321Zbz, expresses human EPO cDNA, and was initially reported by Ruschitzka et al, 14 subsequently named Tg(EPO). The expression is regulated by the platelet-derived growth factor promotor. We used the corresponding WT littermate controls named as WT. 27 Sex- and age-matched groups were used for the experiments with an age limit ranging between 12 and 29 weeks. The mice were housed in a specific-pathogen-free environment in our animal facility. General anesthesia was induced using a mixture of inhaled isoflurane, intravenous fentanyl, medetomidine, and midazolam. All procedures performed on mice were conducted in accordance with local legislation for the protection of animals (Regierung von Oberbayern, Munich) and were authorized accordingly.

The operation was carried out under general anesthesia. The procedure involved median laparotomy to access the inferior vena cava (IVC). A suture was placed around the IVC and ligated below the renal vein. To prevent complete stasis, a placeholder with a diameter of about 0.5 mm was inserted into the loop and subsequently removed after tightening. Side branches of the IVC were not ligated. As result, intravascular flow reduction occurred, ultimately leading to thrombus formation. Thrombus quantification was conducted by removing the IVC (segment between the renal vein and the confluence of the common iliac veins). The incidence and weight of the thrombus were documented.

To analyze the effect of short-term EPO administration on DVT formation, we subcutaneously (s.c.) injected 300 IU (10 IU/µL) EPO (Epoetin alfa HEXAL) three times a week into the gluteal region of C57Bl/6J mice purchased from the Jackson Laboratory. The injections were carried out for a duration of 2 weeks resulting in a total of six EPO treatments. After the completion of the 2-week EPO administration, there was a gap of 2 days before ligating the IVC. The ligation procedure was performed when the mice were 18 weeks old. The control group consisted of age- and sex-matched mice that received the same volume (30 µL) of a 0.9% NaCl solution per dosage.

The cardiac ultrasound analysis was conducted using a Vevo 2100 Imaging System (Visualsonics). To ensure sufficient tolerance during the investigation, short-term inhalation anesthesia (Isofluran CP, cp pharma) was administered during the investigation. Subsequently, the mice were then positioned on their back, and transthoracic echocardiography was performed.

Blood was collected from adequately anesthesized mice through cardiac puncture using a syringe containing citrate as an anticoagulant.

Blood cell counts were determined in citrated blood using an automated cell counter (ABX Micros ES60, Horiba ABX).

To remove the spleen, the mice were anesthetized as previously described. A lateral subcostal incision was made, followed by ligation and cutting of the splenic vessels. Subsequently, the spleen was removed and the surgical wound was closed with sutures. The organ removal procedure was performed 5 weeks prior to subsequent experiments, such as ligation of the IVC.

After harvesting thrombi, the organic material was embedded in OCT, rapidly frozen in liquid nitrogen, and stored at −80°C. Subsequently, 5 µm slides were sectioned using a cryotome (CryoStar NX70 Kryostat, Thermo Fisher Scientific). The staining procedure began with a fixation step using 4% ethanol-free formaldehyde (Thermo Fisher; #28908), followed by blocking with goat serum (Thermo Fisher; #50062Z). The following antibodies were used: CD41 (clone: MWReg30, BD Bioscience; #12-0411-83; isotype: rat IgG1), Fibrin(-ogen) (clone: polyclonal; DAKO; #A0080; isotype: rabbit IgG), Ly6G (clone: 1A8, Thermo Fisher; #12-9668-82; isotype: rat IgG2a), MPO (polyclonal, Dako; #A0398; isotype: rabbit IgG), TER119 (clone: TER-119; Thermo Fisher; #12-5921-83; isotype: rat IgG2b). Alexa-labeled secondary antibodies were used to induce fluorescence (Invitrogen; #A11007; #A11034). Nuclei were marked using Hoechst (ThermoFisher; #H3570). Image acquisition was performed on an AxioImager M2 (Carl Zeiss Microscopy) using corresponding AxioVision SE65 software. Near-field analysis of fibrin fibers was captured on an inverted Zeiss LSM 880 confocal microscope in AiryScan Super Resolution (SR) Mode (magnification, ×63 objective, with 5 to 6 random images acquired per thrombus). Further structural analysis of the fibrin fibers was conducted using Imaris (Oxford instruments). To quantify neutrophil extracellular traps, we identified DNA protrusions (Hoechst-positive) originating from Ly6G-positive cells and covered by MPO.

Statistical analysis was conducted using GraphPad Prism 5, employing a t -test. Based on clinical observations strongly suggesting an increase in thrombus formation in EPO overproducing mice, a one-sided t -test was performed. 8 9 The normal distribution of the data was confirmed using D'Agostino and Pearson omnibus normality testing. Thrombus incidences between groups were compared using the chi-square test.

To investigate the impact of chronic erythrocyte overproduction on DVT in mice, we analyzed EPO-overexpressing transgenic Tg(EPO) mice. As expected, this mouse strain exhibited a substantial increase in RBC count ( Fig. 1A ). Additionally, the RBC width coefficient and reticulocyte count were elevated, indicating enhanced RBC production ( Supplementary Fig. S1A, B [available in the online version]). In addition to influencing the RBC lineage, our analyses revealed a significant increase in white blood cell (WBC) count, primarily driven by elevated lymphocyte count ( Supplementary Fig. S1C, E [available in the online version]). However, neutrophils known as major contributors to venous thrombosis showed no significant changes in EPO transgenic mice, while platelet counts were significantly reduced ( Fig. 1B and Supplementary Fig. S1D [available in the online version]). Furthermore, autopsies of the animals confirmed the presence of splenomegaly ( Supplementary Fig. S1F [available in the online version]). 28 29

Based on clinical observations indicating a correlation between high EPO levels and increased incidence of DVT, we utilized an IVC stenosis model to evaluate venous thrombosis in EPO-overexpressing mice. 6 7 Our findings revealed a significant elevation in both the incidence and thrombus weight in Tg(EPO) mice compared to their WT littermates ( Fig. 1C, D ). To determine whether chronic EPO overproduction in transgenic mice affected cardiac function, we assessed parameters such as the left ventricular ejection fraction, fractional shortening, and heart rate, ruling out any alternations ( Supplementary Fig. S1G, H, J [available in the online version]), which aligns with previous publications. 30 Additionally, morphological parameters including left ventricular mass, left ventricular internal diameter end diastole, and inner ventricular end diastolic septum diameter were similar between Tg(EPO) and WT mice ( Supplementary Fig. S1I, K, L [available in the online version]).

Having observed a correlation between high EPO and hematocrit levels with increased thrombus formation, our aim was to investigate the factors involved in triggering thrombus development through histologic analysis of thrombus composition. In Tg(EPO) mice, the elevated hematocrit levels led to enhanced RBC accumulation within the thrombus, as indicated by the Ter119-covered area measurement ( Fig. 2A ). Given the interaction between RBCs and platelets, which can initiate coagulation activation, we examined the distribution of fibrinogen in relation to RBCs and platelets within the thrombi. 31 Our findings revealed a close association between the fibrinogen signal and RBCs, as well as between the platelet signal and RBCs, indicating interactions among these three factors ( Fig. 2E, F ). However, we observed significantly lower fibrinogen coverage in thrombi from EPO transgenic mice ( Fig. 2B ). Furthermore, the structure of the fibrin meshwork exhibited an overall “looser” morphology with significantly thinner fibrin fibers ( Fig. 3A-C ).

To quantify platelet accumulation in thrombi, we analyzed the CD41-covered area in thrombi of both mouse strains. Consistent with the reduced platelet count in peripheral blood, platelet accumulation was also decreased in thrombi from EPO transgenic mice ( Fig. 2C ).

As mentioned previously, inflammation plays a fundamental role in DVT formation. Therefore, we conducted an analysis to quantify the presence of leukocytes in the thrombus material. Our investigation focused specifically on neutrophils, as they represent the predominant leukocyte population in peripheral blood. Despite observing normal neutrophil counts, we identified a significant reduction in neutrophil recruitment within thrombi from EPO transgenic mice ( Fig. 2D ). In summary, our findings indicate an isolated increase in the number of RBCs within venous thrombi of EPO transgenic mice, while the levels of fibrinogen and platelets were decreased.

Due to the significant impact of chronic EPO overproduction in Tg(EPO) mice on peripheral blood count and its detrimental consequences on DVT formation, we proceeded to analyze the effects of 2-week periodic EPO injections on blood count and subsequent DVT formation in WT mice. Within just 2 weeks, a significant increase of RBC and reticulocyte count in peripheral blood was observed ( Fig. 4A and Supplementary Fig. S2A [available in the online version]). Conversely, platelet count exhibited a notable decrease in EPO-treated mice ( Fig. 4B ). Unlike EPO-overexpressing mice, the leukocyte counts and their differentiation into granulocytes, lymphocytes, and monocytes showed no differences between EPO-treated and nontreated mice ( Supplementary Fig. S2B–E [available in the online version]). Autopsy analyses further revealed a significant enlargement and weight increase of the spleen in EPO-treated mice ( Supplementary Fig. S2F, G [available in the online version]).

To further investigate the underlying cause of thrombocytopenia in EPO-treated mice, we examined the bone marrow composition. Previous studies by Shibata et al demonstrated a reduction in megakaryocytes in Tg(EPO) mice. 32 Therefore, we analyzed the bone marrow composition after 2 weeks of EPO treatment. However, we found no difference in the TER119-covered area, indicating no significant alternation ( Fig. 4C, D ). Similarly, the megakaryocyte count in the bone marrow showed no changes compared to the control group ( Fig. 4E, F ,). In terms of platelet morphology, we observed an increased platelet large cell ratio in the EPO-treated group ( Fig. 4G ). This suggests an elevated production potential of megakaryocytes, as immature platelets tend to have larger cell volumes compared to mature platelets. 33 These findings indicate that our EPO administration protocol enhanced the synthesis capacity of bone marrow stem cells, resulting in augmented erythropoiesis. However, the cellular composition of the bone marrow remained unchanged after 2 weeks of treatment.

To analyze the impact of 2-week EPO treatment on DVT formation, we utilized the IVC stenosis model. Despite similar changes in blood count in Tg(EPO) mice or WT mice after EPO administration, we observed comparable venous thrombus formation between mice treated with EPO for 2 weeks and the control group treated with NaCl ( Fig. 4H, I ). Since we previously observed that only long-term elevation of EPO levels with supraphysiologic hematocrit leads to increased thrombus formation, our focus shifted toward identifying the factors triggering thrombus formation. Therefore, we conducted a histological analysis of thrombus composition. Given the significantly thinner fibrin fibers observed in thrombi from Tg(EPO) mice, we investigated whether similar morphological changes occurred in mice treated with EPO for 2 weeks. Interestingly, the histological examination of the thrombi revealed a comparable thinning of fibrin fibers following EPO treatment ( Fig. 3D, E ).

In contrast to chronic EPO overproduction in Tg(EPO) mice, short-term administration of EPO does not increase the incidence of DVT, despite similar changes in blood cell counts. Therefore, the quantitative changes in blood count alone cannot explain the increased thrombosis observed in the presence of EPO overexpression in Tg(EPO) mice.

As the data suggested a qualitative change in RBCs in the context of EPO overproduction, we investigated whether splenic clearance of aged RBCs plays a critical role in the increased formation of DVT. In the spleen, aged and damaged RBCs are eliminated, ensuring the presence of young and flexible RBCs. 16 We examined the immediate impact of EPO on spleen morphology. Even a single injection of 300 IU EPO s.c. in mice resulted in a significant increase in spleen weight, despite no difference in blood count compared to the control group ( Supplementary Fig. S2F–H [available in the online version]). This striking phenotype was also observed in mice with chronic EPO overexpression ( Supplementary Fig. S1F [available in the online version]). 13

To investigate the role of splenic RBC clearance in DVT, we performed splenectomy 5 weeks prior to conducting the IVC stenosis model. Firstly, we analyzed the impact of splenectomy on blood cell counts in WT mice 5 weeks postsurgery. We observed an increase in granulocytes and lymphocytes after splenectomy ( Fig. 5A and Supplementary Fig. S3A, B [available in the online version]). Next, we examined the distribution of blood cells in response to DVT development. Similar to nonsplenectomized mice, we observed an increase in WBC count in the peripheral blood ( Fig. 5A ). Additionally, we noted a significant decrease in platelet count in splenectomized mice in response to thrombus development ( Fig. 5B ), which is consistent with the results obtained from nonsplenectomized mice.

Finally, we analyzed the impact of splenectomy on DVT formation in both WT mice and Tg(EPO) mice. Despite changes in blood cell counts and the effects on platelet removal, there was no difference in the incidence and thrombus weight in C57Bl/6 mice ( Fig. 5D, E ). Next, we examined EPO-overexpressing mice, which have been shown to have an increased risk of DVT formation. Despite significant splenomegaly, the incidence of DVT formation remained statistically unchanged after spleen removal ( Fig. 5F, G ). Therefore, splenectomy does not affect thrombus formation in the context of enhanced or normal erythropoiesis.