Crosstalk between circulating DPP3 and immune cells in the context of cardio-systemic stress

Summary

Dipeptidylpeptidase 3 (DPP3) is a ubiquitous intracellular peptidase recently identified in the circulation (cDPP3) as a biomarker and therapeutic target in circulatory failure. The source and mechanism of DPP3 release remain unclear. Based on increased DPP3 expression in the bone marrow compartment under stress conditions, we aimed to assess bone marrow’s contribution to DPP3 release. WT bone marrow was transplanted into irradiated DPP3-knockout (KO) mice, generating KO^{BM WT} mice, which recovered 44% of baseline cDPP3 activity. In stressed KO^{BM WT} mice, cDPP3 increase was comparable to stressed WT mice. DPP3 injection in WT mice induced cardiac neutrophil infiltration. Extracellular vesicles from plasma and bone marrow carried only a minimal fraction of DPP3, indicating mainly soluble release. These results reveal bone marrow-derived cells as a main driver of cDPP3 release under stress and provide a foundation to optimize its use as a biomarker and therapeutic target.

Graphical abstract

Highlights

• •Bone marrow-derived cells account for half of circulating DPP3 in normal state
• •In acute stress, bone marrow-derived cells are the main source of circulating DPP3
• •Elevated circulating DPP3 promotes cardiac neutrophil infiltration
• •DPP3 is released as a soluble factor, with minimal extracellular vesicle packaging

Cardiovascular medicine; biochemistry; immunology

Introduction

Dipeptidylpeptidase 3 (DPP3) is a zinc-dependent metallopeptidase that cleaves dipeptides from 4 to 10 amino acid residues, such as angiotensins, enkephalins, and endomorphins.^1,2,3 DPP3 is predominantly cytosolic,^4,5,6 where it participates in the regulation of oxidative stress through enhancing Nrf2 activity, thereby promoting the transcription of antioxidant genes.⁴ However, DPP3 is also detected circulating (cDPP3) in the plasma of healthy individuals, with substantially higher concentrations in critically ill patients.^7,8,9 Increased concentrations of cDPP3 at admission to the intensive care unit have been associated with a poorer prognosis and higher mortality in patients with various forms of circulatory failure, namely cardiogenic,^7,9 and septic^9,10 shock. In addition, the short-term evolution of cDPP3 concentrations also carries prognostic value.^7,10 Beyond its prognostic value, recent studies have identified cDPP3 as a cardiovascular mediator with myocardial depressant effects,¹¹ and a role in renal and systemic hemodynamic regulation,^11,12 likely via the cleavage of vasoactive peptides such as angiotensin II.^12,13 Conversely, cDPP3 inhibition by the neutralizing antibody procizumab (PCZ, also known as invobenitug (INN)) has shown promising results, improving cardiac function and kidney hemodynamics in acute heart failure and septic shock animal models,^11,14,15 underscoring the therapeutic potential of targeting cDPP3.

DPP3 is highly conserved and ubiquitously expressed,³ with high degree of sequence identity among the human, rat, and mouse proteins (>90%).¹⁶ Data on DPP3 expression are limited to physiological conditions,¹⁷ and cancer,^18,19 but are lacking in acute critical conditions. Moreover, the source of its release into circulation remains unknown. The fact that PCZ can inhibit the activity of cDPP3 implies that at least a fraction of cDPP3 exists in a soluble form; however, the possibility that cDPP3 may also be released in an insoluble (vesicle encapsulated) form has not been previously explored. Wattiaux et al. reported DPP3 release from T lymphocyte-derived Jurkat cells following in-vitro-induced cell death.²⁰ DPP3 lacks an N-terminal signal peptide²¹ and is therefore unable to follow the conventional protein secretory pathway.²² Among unconventional protein secretion pathways, extracellular vesicles (EVs) have recently gained increased interest.^23,24 Small (sEVs) and large (lEVs), previously known as exosomes and microvesicles, respectively, are released by all cell types and found in numerous bodily fluids.^25,26 EVs are implicated in several physiological and pathological processes such as intercellular communication, inflammation and immune response.²⁷ DPP3 release via EVs has never been studied.

Inflammation plays an important role in the pathophysiology of acute heart failure²⁸ and to a broader extent in circulatory failure.²⁹ In this context, the inflammatory response is triggered by recognition of a danger signal, such as hemodynamic stress, tissue injury, or pathogenic infiltration.³⁰ Neutrophils are among the first immune cell subsets to be activated and recruited in both sterile inflammation, as in cardiogenic shock,³¹ and septic inflammation.³² In septic shock, neutrophil recruitment is vital for pathogen clearance,³³ while they aid in clearing dead cell debris caused by cardiac injury in post-MI cardiogenic shock.^31,34 However, their overstimulation with the release of reactive oxygen species, inflammatory mediators and neutrophil extracellular traps (NETs) exacerbates the circulatory failure.^35,36,37 Recently, several dipeptidyl peptidases were shown to contribute to various aspects of immune function and inflammatory response,³⁸ some of which play a crucial role in the pathogenesis of septic shock.^39,40 For instance, DPP8 and DPP9 regulate immune cell activation and migration, particularly via inflammasome regulation.^41,42,43 DPP1 activates neutrophil serine proteases crucial for NETs formation and pathogen destruction.^44,45 However, DPP3 was not thoroughly explored in the context of inflammation.

In this study, we aim to comprehensively investigate the regulation and pathophysiological role of DPP3 during acute cardio-systemic stress. We begin by characterizing the modulation of DPP3 expression in clinically relevant models of acute cardiac and hemodynamic stress. We then seek to identify the source and form of cDPP3 release into the circulation. Finally, we investigate the contribution of DPP3 to inflammatory responses in the setting of acute cardio-systemic stress. A better understanding of cDPP3′ pathophysiological regulation and release into the bloodstream provides a more informed and effective clinical use as both a biomarker and a therapeutic target in critical illness.

Results

Profiling of DPP3 expression in acute cardio-systemic stress models

We used the isoproterenol-induced acute cardiac stress (ISO; Figure 1A) mouse model, in which the increase in cDPP3 was previously reported.¹¹ The model was validated by a significant decrease in left ventricular fractional shortening (LVFS) (≥10%) in ISO-treated compared to PBS-treated mice (Ctrl), (Figure S1). We observed a 1.8-fold increase in plasma cDPP3 activity in ISO-treated mice compared to controls (p = 0.018, Figure 1B).

Figure 1.

Profiling of DPP3 expression in acute cardiac stress

(A) Schematic representation of isoproterenol-induced cardiac dysfunction model in C57Bl6 mice.

(B) cDPP3 activity in plasma of ISO (n = 12) vs. Ctrl mice (n = 11).

(C) Quantitative real-time PCR (RT-qPCR) analysis of Dpp3 relative transcription expression, normalized to Gapdh in organs from ISO (n = 6) and Ctrl (n = 8) mice. Data are displayed on a log10-scaled y axis.

(D) Representative immunoblots for DPP3 and GAPDH, with (E) quantification of DPP3 band intensities normalized to GAPDH in organs from ISO (n = 7) and Ctrl (n = 8) mice. Data are displayed on a log10-scaled y axis.

(F) DPP3 activity, normalized to protein content, in BM SN of ISO (n = 7) and Ctrl mice (n = 8).

(G) DPP3 activity, normalized to cell count, in lysate from circulating immune BMDCs from ISO and Ctrl mice (n = 5 per group).

(H) RT-qPCR analysis of DPP3 relative mRNA expression normalized to GAPDH in biopsies from left ventricular (LV) of healthy individuals (Ctrl) (n = 14) and patients with ischemic cardiomyopathy (ICM) (n = 5).

(I) DPP3 protein expression, normalized to GAPDH, in tissue lysate from LV biopsies of healthy individuals (Ctrl) (n = 7) and patients with ischemic cardiomyopathy (ICM) (n = 11).

(J) DPP3 activity, normalized to total protein, in tissue lysate from LV biopsies of healthy individuals (Ctrl) (n = 8) and patients with ischemic cardiomyopathy (ICM) (n = 11). In all bars, data are presented as mean ± standard error of the mean (SEM). Comparisons were done by Mann-Whitney test in B, H, and I; by multiple Mann-Whitney tests in C, E, and G and by unpaired t test in F and J. Significance was presented as follows (∗p < 0.05, ∗∗p < 0.01) and non-significant (ns).

DPP3, dipeptidylpeptidase 3; cDPP3, circulating DPP3; ISO, isoproterenol; Ctrl, control; BMCs, bone marrow cells; BM SN, bone marrow supernatant; LT, lymphocytes T; NK, natural killers; LB, lymphocytes B; eosino, eosinophils; mono, monocytes; neutro, neutrophils; ICM, ischemic cardiomyopathy.

We next assessed DPP3 expression at both mRNA and protein levels in various organs, as well as in bone marrow (BM) cells and BM supernatant (BM SN). Dpp3 mRNA was upregulated only in the heart of ISO-treated mice compared to controls (1.54X, p = 0.01; Figure 1C). DPP3 protein expression was significantly downregulated, by a 3.05-fold, in the spleen of ISO mice (p = 0.002), while upregulated in BM cells (2.2-fold, p = 0.009; Figures 1D and 1E). Moreover, DPP3 activity in BM SN, normalized to protein content, was elevated in ISO mice compared to controls (1.23-fold, p = 0.01; Figure 1F). Interestingly, mean DPP3 activity was higher, by a 10.19-fold, in BM SN compared to plasma in control mice (comparing the y axis of Figures 1B and 1F); suggesting that DPP3 is highly present in the BM microenvironment even in physiological conditions. Given that the BM gives rise to all circulating blood cells, hereafter referred to as BM-derived cells (BMDCs), we assessed DPP3 activity in these cells, focusing on immune cell subsets following FACS sorting. DPP3 activity was significantly increased in neutrophils of ISO mice compared to controls (Figure 1G; p = 0.003).

We further studied DPP3 in a septic shock model induced by cecal ligation and puncture (CLP) in rats (Figure S2A). First, we confirmed the increase in cDPP3 activity in the plasma of CLP compared to sham animals (4.17-fold, p < 0.0001; Figure S2B). In line with the ISO mouse model, we showed higher DPP3 activity in BM SN compared to plasma in sham rats (3.78-fold, p = 0.0002), demonstrating the abundance of DPP3 in the BM. However, there was no significant modification of DPP3 activity in BM SN in CLP compared to sham rats (p = 0.7; Figure S2B).

Consistent with findings from the ISO model, Dpp3 mRNA expression was significantly upregulated (1.79-fold) in left ventricular (LV) biopsies from patients with ischemic cardiomyopathy (ICM) compared to healthy controls (Ctrl) (p = 0.028; Figure 1H). However, neither cardiac DPP3 protein expression nor cardiac activity, normalized to total protein content, differed between the groups (p = 0.47, Figure 1I; p = 0.13; Figure 1J) respectively.

Together, these findings demonstrate the upregulation of Dpp3 mRNA expression in the heart, and its protein expression in the BM and BM-derived cells (BMDCs) in acute cardiac stress. Based on these results, we focused our investigation on the heart and BMDCs as potential sources of cDPP3 release in acute cardio-systemic stress.

The heart represents only a minor source of cDPP3 release in cardiogenic shock

To investigate the role of the heart in cDPP3 release, we used a large animal model of cardiogenic shock (CS) induced by intracoronary ethanol injection in ewes (Figure S3A), mimicking a myocardial infarction-induced-CS. CS was confirmed by a 30% drop in mean arterial pressure (MAP) (Figure S3B) and/or in cardiac output (CO) (Figure S3C) associated with a lactate concentration >2.5 mmol/L (Figure S3D). At the end of the protocol (H+3), cDPP3 activity was significantly increased in CS compared to Sham animals, showing a 1.31-fold increase (p = 0.02; Figure 2A). Additionally, cardiac DPP3 protein expression was upregulated by 2.84-folds in CS animals compared to shams (p = 0.007; Figure 2B). We then assessed the trans-cardiac gradient of cDPP3 activity, defined as the difference in cDPP3 activity between the coronary sinus (cardiac venous blood) and the axillary artery (surrogate of cardiac left ventricular blood). First, we showed a significant increase in cDPP3 activity after CS induction in the coronary sinus (1.21-fold; p = 0.0015), and axillary artery (1.36-fold; p < 0.0001) compared to before CS (Figure S3E). Regarding the trans-cardiac gradient, our results show that before CS, the gradient was positive indicating a participation from the heart in cDPP3 release (Figure 2C). However, following CS induction, the gradient became negative (p = 0.0147). This result shows no substantial cardiac involvement in the release of cDPP3 after CS and suggests an extracardiac major source of cDPP3. In agreement with this, our data showed a lack of correlation between cDPP3 activity and high-sensitivity cardiac troponin T (hs-cTnT) levels at 3 h post-CS (H+3) (Spearman r = 0.28, p = 0.25; Figure 2D), suggesting that cDPP3 is not mainly released by cardiomyocyte necrosis. Finally, we assessed the relevance of our results in a cohort of patients who suffered myocardial infarction (MI) (PREGICA cohort).⁴⁶ In these patients, no correlation was found between cDPP3 concentration and high-sensitivity cardiac troponin I (hs-cTnI) neither at admission before revascularization (D0; n = 84, Spearman r = −0.15, p = 0.17; Figure 2E), nor four days after revascularization (D4; n = 303, Spearman r = 0.02, p = 0.69; Figure 2F), confirming that cardiomyocyte necrosis is not a main source of cDPP3 release in myocardial infarction and CS.

Figure 2.

The heart plays a minor role in the release of cDPP3

(A) cDPP3 activity in plasma of ewe with CS (n = 17) compared to sham (n = 5).

(B) Western-blot analysis of DPP3 protein expression normalized to GAPDH in the heart of animals with CS (n = 10) and sham animals (n = 3).

(C) Trans-cardiac gradient of cDPP3 activity, before induction of CS (before CS) (n = 10) and after CS (n = 10).

(D) Correlation between hs-cTnT and cDPP3 activity at the end of the protocol in sham (n = 4) and CS (n = 14) animals.

(E) Correlation between cDPP3 activity and hs-cTnI at admission before revascularization (D0) (n = 84) and (F) 4 days later (D4) (n = 303) in the PREGICA cohort of myocardial infarction (MI) patients. In all bars, data are presented as mean ± standard error of the mean (SEM). Comparison in A and B was made by unpaired t test. Comparison in C was made by Wilcoxon matched-pairs signed rank test. Correlations in D, E, and F were evaluated by Spearman test. Significance was presented as follows (∗p < 0.05, ∗∗p < 0.01) and non-significance (ns).

DPP3, dipeptidylpeptidase 3; CS, cardiogenic shock; MI, myocardial infarction; hs-cTnT, high sensitivity cardiac troponin T; hs-cTnI, high sensitivity cardiac troponin I.

In vitro, human ventricular cardiomyocytes (AC16) were subjected to various stresses, and cell mortality as well as cDPP3 activity in the cell culture supernatant were evaluated (Figure S4A). Compared to the low-serum starvation medium (Starv), cell mortality showed a trend toward an increase under hypoxic stress (Starv + hypoxia; p = 0.055), and a significant increase in a serum-free medium (FBS-free; p = 0.0007). Mortality was further increased in a serum-free medium under hypoxic stress (FBS-free + hypoxia; p < 0.0001; Figure S4A). However, cDPP3 activity, normalized to the number of cells, was not significantly modified in any of these stress conditions compared to Starv (Figure S4B). No correlation was found between cell mortality and cDPP3 activity in the cell culture supernatant (Spearman r = 0.378, p = 0.09; Figure S4C). Stress-induced cell death did not result in increased cDPP3 activity in the supernatant, suggesting that this form of cell death may not mediate cDPP3 release in disease, or that cDPP3 is released in an inactive form. Alternatively, mechanisms other than cell death may be involved. This finding is consistent with our in vivo and clinical data, suggesting that cardiomyocyte death is not the primary contributor to cDPP3 release in the used models. Interestingly, a stronger correlation was seen between cell death and cDPP3 release in other cell types such as the ovarian cancer cell line ID8 (Figure S4D), subjected to several stress conditions such as starvation and doxorubicin treatment. This suggests a cell-type- and stress-type-dependent mechanism of cDPP3 release.

BMDCs represent a major source of cDPP3 release into the circulation in physiological conditions and acute cardiac stress

We next investigated the role of BMDCs as a source of DPP3 release into the circulation. We thus used a model of total body irradiation (TBI), followed by heterologous BM transplantation (BMT) between WT and Dpp3-KO mice (Figure 3A).

Figure 3.

Bone marrow derived cells (BMDCs) represent a major source of cDPP3 release into the circulation in physiological conditions and acute cardiac stress

(A) Schematic representation of the experimental protocol of the heterologous bone marrow transplantation (BMT) between WT and Dpp3-KO mice followed be isoproterenol (ISO) treatment.

(B) cDPP3 activity in plasma of WT mice (n = 18) at baseline (one week before irradiation W-1), Dpp3-KO mice (KO) (n = 26) at baseline, WT ^{BM KO} (n = 19) at 4 weeks post BMT (W4) and KO^{BM WT} (n = 23) at 4 weeks post BMT.

(C) cDPP3 activity in plasma of ISO-treated WT^{BM KO} (n = 18) and KO^{BM WT} (n = 22) compared to PBS-treated WT^{BM KO} (n = 6) and KO^{BM WT} (n = 7).

(D) DPP3 activity, normalized to total protein content, in BM SN from ISO-treated WT^{BM KO} (n = 11) and KO^{BM WT} (n = 14) compared to PBS-treated WT^{BM KO} (n = 9) and KO^{BM WT}(n = 12). Dotted horizontal blue and red lines represent the mean of DPP3 activity in BM SN, normalized to protein content, from WT and ISO-treated WT mice, respectively.

(E) DPP3 activity, normalized to total protein content, in protein extracts from organs of ISO-treated KO^{BM WT} (n = 10) vs. PBS-treated KO^{BM WT} (n = 11).

(F) Quantification, by pixel count, of the macrophage marker F4/80 and (G) F4/80 colocalization with DPP3 in Dpp3-KO (n = 3), KO^{BM WT} + PBS (n = 3), and KO^{BM WT} + ISO (n = 2) mice.

(H) Representative immunofluorescence staining for the neutrophils marker NIMP-R14 in cardiac tissue sections from Dpp3-KO (top), KO^{BM WT} (middle), and ISO-treated KO^{BM WT} (KO^{BM WT} + ISO, bottom) mice, shown as individual channels and merged images (20x).

(I) Quantification of NIMP-R14-positive area (%), in Dpp3-KO, KO^{BM WT}, and ISO-treated KO^{BM WT} mice. Data represent 3 mice per group (n = 3), with 5 images analyzed per mouse.

Plasma concentrations of: (J) Interlukine-6 (IL-6), (K) programmed death ligand 2 (PD-L2), (L) colony stimulating factor 1 (CSF-1) and (M) C-C motif chemokine ligand 22 (CCL22) in ISO-treated KO^{BM WT} (KO^{BM WT} + ISO) vs. PBS-treated KO^{BM WT} (KO^{BM WT} + PBS) mice (n = 5) each. In all bars, data are presented as mean ± standard error of the mean (SEM). Comparisons in B, C, and D were made by two-way ANOVA. Comparison in E was made by Multiple Mann-Whitney tests. Comparisons in F, G, and I were made by Kruskal-Wallis test, and by Mann-Whitney test in J, K, L, and M. Significance was presented as follows (∗p < 0.05, ∗∗p < 0.01, ∗∗∗∗p < 0.0001) and non-significant (ns).

WT, wild-type; KO, Dpp3-knowkout; WT^{BM KO}, WT mice injected with bone marrow from Dpp3-KO mice; KO^{BM WT}, Dpp3-KO mice injected with bone marrow from WT mice.

WT and Dpp3-KO mice were irradiated together under the same conditions, and white blood cell ablation was confirmed one day after TBI (Figure S5A), achieving ∼83% reduction in total white blood cell count. BMT resulted in two groups of chimeric mice: irradiated WT mice that received BM from Dpp3-KO mice (WT^{BM KO}) and irradiated Dpp3-KO mice that received BM from WT mice (KO^{BM WT}). Four weeks after BMT, the success of the transplantation and the reconstitution of donor BM cells was reflected by the normalization of total white blood cell count in peripheral blood. White blood cell count was comparable to baseline, before irradiation, in KO^{BM WT} mice (p = 0.29) while lower than baseline in WT^{BM KO} (p = 0.01) (Figure S5B). The reconstitution was also validated by the presence of the donor’s genotype in the recipient’s hemopoietic compartments of both chimeric groups; BM cells (Figure S5C), BMDCs (Figure S5D), spleen (Figure S5E), but as expected, not the heart. Interestingly, in all hematopoietic compartments, WT^{BM KO} mice carried both WT and KO cells. However, KO^{BM WT} mice carried only the donor cells (WT) suggesting different capacity of repopulating between WT and Dpp3-KO cells.

Compared to Dpp3-KO mice, which had no detectable cDPP3 activity at baseline, KO^{BM WT} mice showed a cDPP3 activity mean value of 21.87 U/L (Figure 3B), which is equivalent to ∼44% of cDPP3 activity in WT mice at baseline.

These results demonstrate that BMDCs represent a major source of cDPP3 release into the circulation in physiological situations. Accordingly, WT^{BM KO} mice showed a significant decrease (39%) (p < 0.0001) in cDPP3 activity compared to their baseline activity (Figure 3B).

To further explore the role of BMDCs in the release of cDPP3 in acute cardiac stress, chimeric mice were subjected to ISO injections 6 weeks after BMT (Figure 3A). We were not able to detect the decrease in LVFS (Figure S6A), usually detected in ISO-treated mice, probably due to cardiac hypertrophy present in mice as a late consequence of irradiation. Echocardiography parameters, summarized in Table S1, demonstrate modulation in cardiac structural parameters pointing toward a hypertrophic pattern in mice that had irradiation and BMT (WT^{BM KO} and KO^{BM WT}) compared to their corresponding baselines before irradiation, WT and KO mice, respectively. These alterations were more pronounced in ISO-treated WT^{BM KO} and KO^{BM WT} mice. In line with these results, cardiac fibrosis quantified as a visual score of collagen staining by red Sirius tended to be higher in KO ^{BM WT} and WT ^{BM KO} mice compared to KO and WT mice, respectively, and the increase is more pronounced in ISO -treated KO ^BMWT and WT ^{BM KO} mice (Figure S6B).

Despite the absence of a clear cardiac phenotype, we observed a significant increase in cDPP3 activity in ISO-treated compared to PBS-treated KO^{BM WT} mice with a 2.15-fold increase (p = 0.006; Figure 3C), this fold-increase is comparable to what we observed in ISO-treated WT mice, indicating that BMDCs are responsible for the majority of increase in cDPP3 activity observed upon the induction of acute cardiac stress. An increase in cDPP3 activity in ISO vs. PBS-treated WT ^{BM KO} mice (p = 0.008) was also detected, possibly due to the regeneration of their original WT BM, or the possible participation of other organs in cDPP3 release which can’t be excluded. In BM SN, DPP3 activity, normalized to protein content, was increased in ISO-treated versus PBS-treated KO^{BM WT} mice with a 2.23-fold increase (p = 0.0013; Figure 3D), showing that BM cells release DPP3 in their surrounding microenvironment. No increase in DPP3 activity in BM SN of ISO-treated vs. PBS-treated WT ^{BM KO} mice was observed (p = 0.51).

To investigate whether DPP3 released from BMDCs under acute cardiac stress affects cardiac function, WT mice were injected, without pre-irradiation, with total BM, including cells and supernatant, from ISO-treated WT or ISO-treated Dpp3-KO mice, producing WT^{WT ISO} or WT^{KO ISO} mice, respectively, and the cardiac function was evaluated at baseline, 30- and 60-min post injection. As shown in Figure S6C, BM transplant from ISO-treated WT mice led to a mean decrease of 12.5% in LVFS at 60 min post injection compared to baseline, while a smaller decrease of 3.5% was seen with the injection of BM from ISO-treated Dpp3-KO mice. Moreover, only 25% of mice receiving BM of ISO-treated Dpp3-KO mice presented with acute cardiac dysfunction, and recovered normal cardiac function at 60 min post injection, while all mice that received BM cells from ISO-treated WT mice had a worsening cardiac function at 60 min. Although, the difference was not significant (p = 0.17), these results suggest that DPP3 specifically released from the BM participates in the myocardial depressing effect.

DPP3 role in immune cell infiltration and immune modulation

We measured DPP3 activity in organs of KO^{BM WT} mice, treated or not with ISO. DPP3 activity was detected across all organs of KO^{BM WT} mice. This activity was significantly increased in the heart of ISO-treated compared to PBS-treated KO^{BM WT} mice (2.56-fold, p = 0.01; Figure 3E). Furthermore, the cardiac expression of F4/80, a macrophage marker, showed a trend toward an increase in KO^{BM WT} mice, 6 weeks post BMT, compared to KO mice. F4/80 expression was further increased in ISO-treated KO^{BM WT} mice (Figure 3F), reflecting increased macrophage infiltration upon ISO treatment. Interestingly, a co-localization of DPP3 and F4/80 was detected in KO^{BM WT} and ISO-treated KO^{BM WT} (Figure 3G), demonstrating the infiltration of DPP3-expressing macrophages in the cardiac tissue of these mice. Similarly, ISO treatment induced neutrophil infiltration into the heart, as shown by immunofluorescence staining for the neutrophil marker NIMP-R14. ISO-treated KO^{BM WT} mice displayed a higher percentage of NIMP-R14-positive area compared with PBS-treated KO^{BM WT} mice (p = 0.0056) and KO mice (p < 0.0001; Figures 3H and 3I).

Consistent with these results, ISO treatment in KO^{BM WT} mice induced a shift in circulating cytokines toward a pro-inflammatory profile. We reported a significant increase in cytokines known to promote immune cell migration such as IL16 (1.69-fold increase, p = 0.004; Figure 3J), whereas levels of cytokines that support immunosuppressive and anti-inflammatory responses, such as PD-L2 (1.81-fold decrease, p = 0.004; Figure 3K), CSF-1 (1.46-fold decrease, p = 0.047; Figure 3L) and CCL22 (2.78-fold decrease, p = 0.0079; Figure 3M), were decreased in ISO-treated KO^{BM WT} mice compared to PBS-treated KO ^{BM WT} mice.

To explore whether DPP3 has an influence on immune cell migration and infiltration into the heart, C57BL6 mice were injected with either cDPP3 or PBS (Figure 4A). Compared to PBS-injected mice, DPP3-injected mice had a trend toward increased concentrations of all immune cells in the cardiac tissue, but only the increase in neutrophils concentration was significant with a 3.95-fold increase (p = 0.0016; Figure 4B). Gating strategy and corresponding dot plots are provided in Data S1 (Data S1).

Figure 4.

DPP3 induces immune cell infiltration and immune modulation

(A) Schematic representation of the experimental protocol of intravenous DPP3 injection in WT mice.

(B) Concentration of immune cells in the heart of mice injected with DPP3 or PBS (n = 4) each.

(C) Concentrations of immune cell subpopulations in blood, and (D) bone marrow (BM) in WT and Dpp3-KO (KO) mice (n = 10) each. Data are displayed on a log10-scaled y axis in C and D.

(E) Concentration of C-C motif chemokine ligand 5 (CCL5) in plasma and (F) BM of WT and KO mice (n = 5) each. In all bars, data are presented as mean ± standard error of the mean (SEM). Comparisons were made by two-way ANOVA in B, multiple t tests in C and D, and Mann-Whitney test in (E and F). Significance was presented as follows (∗p < 0.05, ∗∗p < 0.01) and non-significance (ns).

LT, lymphocytes T; NK, natural killers; LB, lymphocytes B; eosino, eosinophils; mono, monocytes; Neutro, neutrophils.

To further evaluate the possible effects of DPP3 on immune modulation and inflammatory profile, we analyzed blood and BM immune cell concentrations as well as cytokine profile in WT and Dpp3-KO mice. In the blood, no differences in cell concentration between WT and Dpp3-KO mice were detected (Figure 4C). However, in BM, there was a significant increase in the concentration of total lymphocyte T, including both CD4⁺ and CD8⁺ lymphocyte T, in addition to monocytes and neutrophils in Dpp3-KO mice compared to WT (Figure 4D). Gating strategy and corresponding dot plots are provided in Data S1 (Data S1).

CCL5 was significantly decreased in the plasma (1.62-fold decrease, p = 0.047; Figure 4E) with a non-significant trend toward an increase in BM SN (p = 0.8; Figure 4F) in Dpp3-KO mice compared to WT.

A minimal fraction of cDPP3 is released via EVs

To explore whether cDPP3 is released via EVs, we isolated large (lEVs) and small (sEVs) EVs from mouse plasma and BM SN. EVs were characterized using standard procedures⁴⁷ (Figure S7). Western blot confirmed EV markers including CD63, a transmembrane EV marker, and HSC70, a cytosolic EV marker and DPP3 was detected (83 KD) in both lEVs and sEVs (Figure 5A).

Figure 5.

A fraction of DPP3 is transported via EVs and upregulated during septic shock in humans

(A) Western blot qualitative analysis of cluster of differentiation 63 (CD63) at 26 KD (bottom), heat shock protein 70 (HSC70) at 73 KD (middle), and DPP3 at 83 KD (above), in protein extracts of EVs derived from bone marrow supernatant (BM SN) from Ctrl mice (left) and ISO mice (right). DPP3 activity (U/L) in large (lEVs) and small (sEVs) EVs derived from (B) plasma of control mice (Ctrl) (n = 6) and isoproterenol-treated mice (ISO) (n = 5), or from (C) BM SN of Ctrl (n = 6) and ISO mice (n = 6).

(D) DPP3 activity in lEVs and sEVs derived from BM SN of WT mice injected with bone marrow cells from Dpp3-KO (WT ^{BM KO}) and Dpp3-KO mice injected with bone marrow cells from WT mice (KO ^{BM WT}) (n = 5) for each.

(E) Quantification of Nano track analysis (NTA) results of plasma derived-lEVs and sEVs concentration (vesicle/ml) in healthy individuals (n = 7), septic shock (SS) (n = 6) and cardiogenic shock (CS) patients (n = 5).

(F) DPP3 activity (U/L) in lEVs and sEVs derived from plasma of healthy individuals (n = 8) and septic shock (SS) patients (n = 6).

(G) DPP3 activity (U/L) in lEVs and sEVs derived from plasma of healthy individuals (n = 8) and cardiogenic shock (CS) patients (n = 11). DPP3 activity (U/L) in lEVs and sEVs derived from cell culture supernatant of (H) human cardiomyocytes AC16 (n = 6) and (I) human monocytes THP-1 (n = 2). In all bars, data are presented as mean ± standard error of the mean (SEM). Comparisons were made by Multiple Mann-Whitney tests in B, C, F, and G, and by two-way ANOVA in D and E. Significance was presented as follows (∗p < 0.05, ∗∗p < 0.01, and ns for non-significant differences).

No difference in DPP3 activity was observed in plasma-derived lEVs or sEVs between ISO and control mice (Figure 5B), nor in their percentage to total plasma cDPP3 activity (Figure S8A). While BM SN-derived EVs showed higher DPP3 activity than plasma EVs, ISO treatment did not alter DPP3 activity in BM SN-derived lEVs (p = 0.9) or sEVs (p = 0.9; Figure 5C). Mean percentage of DPP3 activity in lEVs to total BM SN, was 8.04% in ISO vs. 10.63% in controls (p = 0.73); in sEVs 6.73% vs. 7.89% (p = 0.73; Figure S8B).

To confirm that DPP3-containing EVs in BM SN originate from BM cell, we analyzed EVs from KO^{BM WT} mice, at 6 weeks post-BMT. DPP3 activity was detected in both EVs subtypes, indicating BM origin (Figure 5D).

In human, EVs were isolated and characterized as in mice. Septic shock (SS) patients had a 3.33-fold increase in plasma sEVs compared to healthy controls (p = 0.003; Figure 5E), while no significant change was observed in lEVs (p = 0.72). In healthy individuals, DPP3 activity was low in lEVs and nearly absent in sEVs (Figure 5F). In SS patients, sEVs DPP3 activity increased 24.5-fold (p = 0.001; Figure 5F), likely due to the increase in sEVs concentration.

In cardiogenic shock (CS) patients, no change in EVs concentrations or EVs-DPP3 activity was detected compared to healthy controls (Figures 5E and 5G). In healthy individuals, the percentage of DPP3 activity in lEVs to total plasma had a mean of 7.9%, while sEVs-associated activity remained negligible (). No significant changes were found in the percentage of DPP3 activity in EVs to total plasma in CS patients compared to controls (p = 0.55 for both EVs subtypes; Figure S8C). Together these results show that a small, enzymatically active, fraction of DPP3 is released via lEVs and sEVs, both in plasma and BM SN, varying depending on the disease and the specific model used. DPP3 activity was also detected in EVs derived from AC16 human cardiomyocyte supernatant when cultured in serum-free medium (Figure 5H), supporting our earlier observation of the lack of correlation between cardiomyocyte cell death and DPP3 release. In human monocytes THP1, cDPP3 release into the supernatant was observed following lipopolysaccharide (LPS)-induced cell death, probably accompanied by immune activation (Figures S4E and S4F). Additionally, monocytes were also capable of releasing small amounts of DPP3 in lEVs and sEVs when cultured in serum-free medium (Figure 5I), warranting further study under additional stress or activation conditions, and in other immune cell types.

Discussion

In this study, we provide a comprehensive characterization of cDPP3 increase in acute cardio-systemic stress, including its expression patterns, cellular source, form of release into the circulation, and pathophysiological roles. Our findings identify BMDCs as a primary source of cDPP3, contributing to its upregulation both systemically and within injured cardiac tissue. We further show that cDPP3 promotes neutrophil cardiac infiltration and amplifies the pro-inflammatory response, likely exacerbating cardiac stress. Notably, a minor fraction of cDPP3 is released as a non-soluble form associated with EVs derived from BMDCs.

Spanning several species, we demonstrated upregulation of cDPP3 activity across three different models that represent cardio-systemic stress and hemodynamic instability including: acute cardiac stress (1.8-fold), cardiogenic (1.31-fold), and septic shock (4.17-fold). While cDPP3 increase has been previously reported in patients with cardiogenic^7,9 and septic^9,10 shock, as well as in models of acute cardiac stress¹¹ and septic shock,¹⁴ our study is the first to investigate it in a cardiogenic shock model. Currently, there is no available data on the source of cDPP3 release, other than speculations^10,48 suggesting its release from injured myocardium. Surprisingly, our trans-cardiac gradient results from the CS model challenge these speculations as does the lack of correlation between cDPP3 and cardiac troponins. This lack of correlation is coherent with what has been previously described in cohorts of acute coronary syndrome⁴⁸ and CS⁷ patients. Interestingly, we demonstrate an important source of cDPP3 release, BMDCs, suggesting a crosstalk between DPP3 and these cells extending from the BM to circulation and cardiac tissue. While BMDCs account for approximately 44% of total circulating DPP3 activity under physiological conditions, they appear to be the primary contributors to the increase observed during acute cardiac stress. In fact, the rise in cDPP3 activity in ISO-treated KO^{BM WT} mice was comparable to that seen in ISO-treated WT mice, relative to their respective controls. This suggests a more prominent role for BMDCs in cDPP3 release under stress conditions. It also raises the possibility that distinct mechanisms may be involved in regulating cDPP3 release during cardiac stress. The lack of correlation between Dpp3 mRNA and protein expression suggests regulation at post-transcriptional or post-translational levels, or may reflect the release of cDPP3. For example, in the spleen, the decrease in DPP3 protein under ISO, despite unchanged mRNA expression, may reflect the stress-induced mobilization of immune cells, such as monocytes, from the spleen into circulation and toward injury sites.⁴⁹ As immune cells express DPP3, their exit from the spleen could explain the observed reduction.

Building on our findings, a mechanistic trajectory of DPP3 during cardio-systemic stress can be drawn. Upon stress, BMDCs significantly increase the release of cDPP3. cDPP3 will exert its recently-described cardiovascular effects^11,12 through cleavage of angiotensin peptides^3,12 and possibly other unknown mechanisms. cDPP3 will further promote immune cell infiltration to the heart in parallel with the broader cytokine-driven pro-inflammatory response, typically observed during acute cardiac stress. By promoting immune cell recruitment and sustaining inflammation, cDPP3 likely exacerbates both myocardial injury and vascular dysfunction. Immune cells recruited to the heart also express and likely release DPP3 locally, suggesting a self-perpetuating pro-inflammatory loop whereby cDPP3 amplifies its own effects both systemically and locally within the heart, establishing a vicious cycle that exacerbates cardiac injury. One might expect a similar trajectory in circulatory failure.

Acute cardiac stress alters the BM in both DPP3-dependent and -independent ways. The greater impact of stressed-WT versus -Dpp3-KO BM on cardiac function suggests that elevated cDPP3 released from BMDCs contributes to immune-cardiovascular crosstalk and cardiac dysfunction.

Among the used models, cDPP3-fold increase was the highest in the CLP-induced septic shock model, probably due to the more important systemic inflammation and overstimulation of immune BMDCs in septic shock.^29,50 Surprisingly, the remarkable upregulation of cDPP3 in the plasma of animals with septic shock was not accompanied by an upregulation in the BM SN, which suggests that the increase of cDPP3 is more likely to be dependent on its release by circulating BMDCs rather than its increase in BM SN. Further experiments are needed to exclude a gradient of soluble DPP3 between BM SN and the blood.

Our results pave the way for a better understanding of cDPP3 in the pathophysiology of circulatory failure. Potential pathways governing DPP3 and BMDCs relationship are worth exploring. To start, the BM is a hypoxic microenvironment, and hypoxia is often associated with increased ROS production.^51,52 DPP3 is involved in the oxidative stress response via its influence on the Keap1-Nrf2/ARE signal.⁵³ A greater need to oxidative stress regulation in the BM compartment might trigger increased DPP3 expression and release from BM cells explaining its enrichment in this compartment across different species. Additionally, ROS fluctuations in the BM regulate the quiescent and cycling states of hematopoietic stem cells (HSCs) affecting their differentiation and motility.^54,55 The absence of DPP3 in Dpp3-KO mice may disturb ROS balance, influencing BMDCs behavior and sorting, thus explaining the increase in BM cellularity observed in our study in these mice. This increase is consistent with what Menale et al. have previously showed in Dpp3-KO mice.⁵⁶ DPP3’s role in promoting cell migration was previously shown in different types of cancer cells.^19,57 However, these effects have never been evaluated in immune cells. Our results showed increased neutrophil infiltration to the heart following DPP3 IV injection. This might be due to the myocardial depressant effect of cDPP3, which stresses the heart,¹¹ probably leading to this infiltration or to the concomitant catecholamine release upon DPP3 IV administration, as recently demonstrated.¹² Given the well-established role of catecholamines in promoting HSCs mobilization from the BM into circulation, as well as their tissue infiltration especially during stress,^58,59 this mechanism could explain the observed increase in neutrophil infiltration into the heart. The effect of DPP3 on BMDCs behavior was reflected also in the different genotype of hematopoietic compartments of WT^{BM KO} and KO^{BM WT} mice despite the identical irradiation and BMT conditions. One might argue that HSCs of WT mice are more robust, resistant to oxidative stress and more proliferative, compared to Dpp3-KO HSCs, giving them an advantage in repopulating the KO recipient’s hematopoietic system. Isoproterenol is known to influence immune cell function through agonism of β-adrenergic receptors (βARs) expressed on these cells.⁶⁰ Although the literature presents conflicting data on adrenergic modulation of neutrophil function, recent findings suggest that adrenergic signaling primarily inhibits ROS production in neutrophils.^61,62,63 We showed that isoproterenol induced an upregulation of DPP3 expression in circulating neutrophils, whether this upregulation contributes to the inhibition of ROS production remains to be further explored.

Being a cytosolic protein lacking a signal peptide, DPP3 was previously assumed to be released only by cell death as a soluble form.^4,21 We showed an unconventional secretion of DPP3 via EVs. Most plasma EVs originate from immune and non-immune blood cells.⁶⁴ Furthermore, the presence of DPP3 in EVs from KO^{BM WT} mice indicates their release from BMDCs. In-vitro, we showed that LPS induces cDPP3 release from human monocytes. Whether this release is induced only by cell death or also by the LPS-triggered monocyte activation needs further investigation. We showed also that monocytes are capable of releasing a minimal fraction of DPP3 via EVs. Further research is needed to fully characterize cDPP3 release from other BMDCs spanning different triggers and mechanisms.

Though DPP3’s EVs-fraction is minimal, the potential of EVs to deliver their cargo into recipient cells⁶⁵ may cf. greater pathophysiological importance than the soluble fraction. Whether DPP3 is involved in EV-mediated intercellular communication,⁶⁶ and what drives its encapsulation in EVs remain unclear. Our findings suggest EV release of DPP3 is model- or pathology-dependent: in the ISO model, increased cDPP3 is mainly in the soluble fraction, while in septic shock patients, sEVs partly account for the rise. Recent studies highlighted EVs’ altered concentration and cargo during sepsis indicating pathophysiological relevance.⁶⁷

Understanding the source of cDPP3 release paves the way for a better use of cDPP3 both as a biomarker and a therapeutic target. Evolution of cDPP3 concentrations within 24 h post-ICU admission has shown an important prognostic value in cardiogenic and septic shock.^7,10 Building on our results, strongly linking DPP3 to BMDCs, one might speculate that cDPP3 dynamics are linked to the severity, phase, and time course of systemic inflammation. Therefore, interpreting cDPP3 levels in the light of the patients’ inflammatory profiles may provide a better patient stratification. Immunomodulatory therapies have been extensively evaluated in cardiogenic and septic shock to restore immune homeostasis and preserve organ function.^68,69 Strategies have included, among others, direct targeting of pro-inflammatory cytokines such as with anti-TNF-α antibody,⁷⁰ as well as blockade of their receptors such as IL-1⁷¹ and IL-6⁷² receptors blockade. Broader approaches such as the use of corticosteroids have also been employed.^73,74 However, clinical efficacy has been limited, likely due to patient heterogeneity, the complexity of inflammatory pathways, and the context-dependent roles of these mediators. Inhibition of cDPP3 with its antibody PCZ has shown promising effects in restoring hemodynamic stability and improving cardiac function in models of circulatory failure.^11,14,15 Building on our findings, it becomes relevant to explore how PCZ may modulate the inflammatory response under these conditions. Notably, Garcia et al. have reported attenuation in myocardial inflammation, indicated by decreased IL-6 mRNA expression, following PCZ injection in a porcine model of septic shock.¹⁵ This was accompanied by lower levels of hs-cTnI in animals receiving PCZ, possibly reflecting attenuated inflammation-related cardiac injury, including impaired contractility, reduced perfusion, and increased cardiomyocyte death.

In conclusion, our findings reveal a crosstalk between DPP3 and BMDCs. Under acute cardio-systemic stress, BMDCs serve as a major source of cDPP3, which in turn promotes their infiltration into the heart. Once present, these cells further contribute to the cardiac upregulation of DPP3. cDPP3-induced cardiac dysfunction relies, at least partly, on its immune modulation effects. Although partially mediated by EVs, the precise mechanisms of DPP3 release remains to be elucidated.

Limitations of the study

Due to model-specific constraints, the preclinical models included only one sex, preventing the evaluation of sex-specific effects. Furthermore, demographic data, including gender and ethnicity, were not available for the human cohorts. We kept a general approach interpreting BMDCs results, considering the complexity of the BM microenvironment. Therefore, a cell-specific source investigation is needed in future studies. Furthermore, a detailed study of cDPP3 mechanisms of release from BMDCs including different types of cell death and other mechanisms should be considered in the future.

Resource availability

Lead contact

Requests for further information and resources should be directed to and will be fulfilled by the lead contact, Feriel Azibani (feriel.azibani@inserm.fr).

Materials availability

This study did not generate new unique reagents.

Data and code availability

• •The datasets generated and analyzed during this study are available from lead contact upon request.
• •This study did not generate original code.
• •Any additional information required to re-analyze data reported in this study is available upon request from lead contact.

Acknowledgments

This work was supported by Institut National de la Santé et de la Recherche Médicale, Fondation pour la Recherche Médicale, Université Paris Cité, Agence Nationale de la Recherche (ANR-24-CE17-5491) and Société Française d’Anesthésie et de Réanimation. The authors thank the IVETh Core Facility (NTA), the AMMICA/PFIC Platform at Institut Gustave Roussy (flow cytometry), V. Parietti Montcuquet (irradiation), Prof. J.-S. Silvestre (human monocytes), the CIRCULIS facility (cytokines concentration) and the Electron Microscopy Platform at the Jacques Monod Institute, UMR 7592 CNRS, for their technical support.

Author contributions

Conceptualization, N.A. and F.A.; investigation, N.A., H.N., A.N., C.O., G.M., L.D., and I.S.; technical support, M.T., M.S., E.P., M.Z., and A.B.; data interpretation; N.A., F.A., N.V., A.C., and A.P.; writing original draft, N.A. and F.A.; resources F.A., A.M., N.V., P.R., A.O., D.L., J.L., and K.S.; funding acquisition: F.A.and A.M.; supervision, F.A. and A.M.; review and editing: All authors.

Declaration of interests

K.S. is employed by 4TEEN4 Pharmaceuticals GmbH, which owns patent rights for the cDPP3 biomarker. F.A. received research grants from 4TEEN4. A.M. has received research grants from Roche Diagnostics, Abbott Laboratories, 4TEEN4 Pharmaceuticals GmbH, Windtree Therapeutics, and SphingoTec; honoraria for lectures from Merck, Novartis, Roche Diagnostics, and Bayer; is a consultant for Roche Diagnostics, Adrenomed, Corteria Pharmaceuticals, and FIRE-1; and is a co-inventor of a patent on combination therapy for patients with acute or persistent dyspnea, owned by S-FormPharma; member trial Committee for Secret-HF, sponsored by the French Government, for S-Form Pharma, and Implicity.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Anti- Ly6C (BV785, clone HK1.4)	Biolegend	Cat#128041; RRID: AB_2565852
Anti- CD4 (PE/Cy5, clone GK1.5)	Cytek	Cat#55-0041-U025
Anti- NK1.1 (PE/Cy7, clone PK136)	Biolegend	Cat#156513; RRID:AB_2888852
Anti- CD172a (Alexa Fluor 647, clone P84)	Biolegend	Cat#144028; RRID:AB_2721301
Anti- CD19 (APC/Cy7, clone 1D3)	Biolegend	Cat#152411; RRID:AB_2922473
Anti- CD3 (VioletFluor 450, clone 17A2)	Cytek	Cat#75-0032-U025
Anti- Ly6G (BV605, clone 1A8)	Biolegend	Cat#127639; RRID:AB_2565880
Anti- CD45 (PerCP, clone 30-F11)	Cytek	Cat#67-0451-U025
Anti- CD8 (RedFluor 710, clone 53–6.7)	Cytek	Cat#80-0081-U025
Anti- CD11b (VioletFluor 500, clone M1/70)	Cytek	Cat#85-0112-U025
Anti- Siglec-F (FITC, clone S17007L)	Biolegend	Cat#155504; RRID:AB_2750233
Anti-DPP3 (Humanized)	4TEEN4 Pharmaceuticals
Anti-GAPDH (Mouse)	Millipore	Cat#MAB374; RRID:AB_2107445
Anti-CD63 (Mouse)	Santacruz Biotechnology	Cat#sc-5275
Anti- HSC70 (Rat)	Enzo Lifesciences	Cat#ADI-SPA-815-F; RRID:AB_11181186
Anti- F4/80 (Rat)	Abcam	Cat#AB6640; RRID:AB_1140040
Anti- NIMP-R14 (Rat)	Abcam	Cat#AB2557; RRID:AB_303154
Anti-Human IgG	Sigma-Aldrich	Cat#A8792; RRID:AB_258414
Anti-mouse IgG	Sigma-Aldrich	Cat#A9044; RRID:AB_258431
Anti-rat IGg	Millipore.	Cat#AP183P; RRID:AB_92600
Alexa Fluor 488 Goat Anti-human IGg	Invitrogen	Cat#A11013
Alexa Fluor 594 Goat Anti-rat IGg	Abcam	Cat#AB150168;RRID:AB_3712336
TruStain FcXTM PLUS anti-mouse CD16/32	Biolegend	Cat#101319; RRID:AB_1574973
Chemicals, peptides, and recombinant proteins
DMEM/F12	Sigma	D6434
FBS	EMD Millipore	ES-009-B
L-Glutamine	EMD Millipore	TMS002-C
Penicillin-Streptomycin	EMD Millipore	TMS-AB2-C
DMEM	Sigma-Aldrich	D6429
Insulin-Transferrin-Selenium	Sigma-Aldrich	l1884-1VL
RPMI-1640	ATCC	30–2001
Lipopolysaccharide (LPS)	Sigma-Aldrich	L2880
Isoproterenol (ISO)	Sigma-Aldrich	I5627-5G
Oxytetracycline 10%	Centravet	OXY007
hDPP3	4Teen4
Critical commercial assays
BCA Protein Assay kit	Thermo Fischer
RNeasy® Kit	Qiagen	74106
QuantiTect® Reverse Transcription Kit	Qiagen™;	205313
FastStart Essential DNA Green Master® kit	Roche Diagnostics
Experimental models: Cell lines
Human ventricular cardiomyocytes (AC16)	Merck	SCC109
murine ovarian cancer cells (ID8)	Merck	SCC145
Human monocytes (THP-1)	Gift from PARC Center -Paris
Experimental models: Organisms/strains
C57BL6/J mice	Janvier Labs
Dpp3-KO mice	Gift from Graz University
Wistar rats	ENVIGO
Oligonucleotides
Dpp3-qPCR-Forward (5′ACTCTCAGAACCTCAGTGCA-3′)	Europhins genomics	5-4427-19/20
Dpp3-qPCR-Reverse (5′GGGTCTGTGTTGAGGACTGA-3′)	Europhins genomics	5-4427-20/20
Gapdh-qPCR-Forward (5′AACTTTGGCATTGTGGAAGG-3′)	Europhins genomics	3-5010-1/10
Gapdh-qPCR-Reverse (5′ACACATTGGGGGTAGGAACA-3′)	Europhins genomics	3-5010-2/10
Genotyping primer Dpp3-WT-Forward (5′-TTG CCT ACT TCC TGC CCA GA-3′)	Europhins genomics	5.3002-6/16
Genotyping primer Dpp3-WT-Reverse (5′-TTG GCT ATC CCT GTT GCA G-3′)	Europhins genomics	5.3002-6/16
Genotyping primer Dpp3-KO-R (Pair1, targets the LACZ) (5′-GTC GCT ACC ATT ACC AGT TG-3′)	Europhins genomics	5.3002-7/16
Software and algorithms
NanoSight NS-300	Malvern Panalytical Ltd., UK
NanoSight NTA software v3.3	Malvern Panalytical Ltd., UK
SpectroFlo Software	Cytek, CA, USA
MultiGauge V2.02 software	FUJIFILM Corporation, Japan
Vevo LAB V 5.8.0	FUJIFILM Visual sonics
ImageJ	National Institutes of Health, USA
GraphPad Prism software version 10.3.1	GraphPad, CA, USA
Other
qEVoriginal/70 nm Gen 2	IZON Science Ltd.	ICO-70

Experimental model and study participant details

Cell lines and in vitro experiments

Three cell lines were used: human ventricular cardiomyocytes (AC16, Merck SCC109), murine ovarian cancer cells (ID8, Merck SCC145), and human monocytes (THP-1, kindly provided by Prof. Jean-Sébastien Silvestre). All cell lines were tested for mycoplasma contamination and were negative. AC16 cell line was authenticated by Short Tandem Repeat analysis (STR) profiling (Merck) prior to use. AC16 cells were cultured in DMEM/F12 (Sigma D6434) with 12.5% FBS (EMD Millipore Cat. No. ES-009-B), 2 mM L-Glutamine (EMD Millipore Cat. No. TMS-002-C), and 1× Penicillin-Streptomycin (EMD Millipore Cat. No. TMS-AB2-C). ID8 cells were maintained in DMEM (Sigma D6429) with 4% FBS, 1% Pen-Strep, and 1% Insulin-Transferrin-Selenium (Sigma-Aldrich, ref: l1884-1VL). THP-1 cells were grown in RPMI-1640 (ATCC 30–2001) supplemented with 10% FBS and 0.05 mM 2-mercaptoethanol. All cultures were incubated at 37°C in 5% CO₂, 21% O₂, and 95% humidity. Once 80% confluency was reached, cells were transferred to 6-well plates and exposed to specific conditions. AC16 cells were incubated in low-serum (1% FBS) (starv) or serum-free media under normoxia or hypoxia (1% O₂). ID8 cells were subjected to low-serum (starv), serum-free media, and increasing doses (0.25, 0.5, 1 μM) of doxorubicin. THP-1 cells were treated with low-serum (starv), with or without lipopolysaccharide (LPS, 100 or 1000 ng/mL). After 24 h, supernatants were collected and cell viability was assessed by counting live cells using Malassez chambers.

Mouse models

A total of 186 mice was used in all experiments. Ten-week-old male C57BL6/J mice, weighing 25 g ± 2.5 were purchased from Janvier Labs. Dpp3-KO mice were kindly provided by Prof. Zimmerman from Graz University.³ Animal care and experimental procedures complied with the EU Directive 2017/32/EU, and French Departmental Direction of Animal Protection (APAFIS N° 04146.03 & 46116). The 3R rule was taken into consideration in the design and application of experimental protocols. Reporting of this work complies with ARRIVE 2.0 guidelines. All animals were housed in light- (on from 08:00–20:00 h) and temperature- (21–24 °C) controlled testing rooms, with food and water available ad libitum. Only male mice were used in this study due to the incompatibility of the isoproterenol-model with females.

Isoproterenol-induced cardiac stress model

Acute cardiac stress was induced in ten-week-old male C57BL6/J mice (n = 12) by subcutaneous injections of 300 mg/kg of isoproterenol (ISO) (I5627-5G, Sigma-Aldrich) in sterile phosphate-buffered saline (PBS) twice daily for two days as previously described.⁷⁵ Age matched control animals received PBS (n = 11). Heart function was monitored by echocardiography, at baseline and 12 h after the last ISO or PBS injection. Then, blood was sampled, mice were sacrificed and their organs and BM were collected for analysis. Bone marrow was collected as previously described.⁷⁶ Briefly, the right tibia and femur of each animal was extracted, cleaned from connective tissue and muscle, cut at metaphysis level on each side, and put in 500 μL Eppendorf tubes pierced with an 18G needle to create a small hole at the bottom. Each tube was put in 1.5 mL Eppendorf tubes containing 400 μL of PBS solution. After centrifugation at 10,000g for 1 min, a pellet of cells was found at the bottom of the 1.5 mL tube, containing BM cells, while the 400 μL PBS was now mixed with the BM cell stroma (therefore considered as BM SN). Both were cooled in liquid nitrogen and stored at −80°C until further analysis.

Total body irradiation and bone marrow transplantation

Ten weeks-old Dpp3 knockout mice (Dpp3-KO, n = 26) and wild-type littermates (WT, n = 18) were subjected to a single 9.5 Gy dose of total body irradiation (TBI) using an X-ray machine (X-RAD 320).⁷⁷ Mice were given oxytetracycline 10% (OXY007 Centravet) at 60 mg/L in drinking water starting from three days before irradiation up until 14 days post-irradiation. A complete blood count (CBC) was performed to validate white blood cell ablation (using MS9 – 5V automated full digital cell counter, Melet Schloesing Laboratories). Twenty-four hours post-irradiation, Dpp3-KO and WT mice were anesthetized by intraperitoneal injection of ketamine (80 mg/kg) and a drop of local anesthetic (Cebesine 0.4%) was applied to the eye of recipient mice, followed by retro-orbital injection of 4.5 × 10⁶ BM cells freshly isolated from age-matched WT and Dpp3-KO donor mice, respectively. A total of 21 donor mice was used (WT: n = 12, KO: n = 9). Bone marrow cells of each donor mice genotype were pooled before injecting the same amount to all receiving mice. Six weeks after BM transplantation (BMT), mice were randomized to receive isoproterenol or PBS injections as previously described. Blood sampling was done at baseline, four weeks post-BMT (W4) and at the end of the isoproterenol protocol. Finally, mice were sacrificed and organs and BM were collected, as previously described, for further analysis.

Bone marrow transplantation effects on cardiac function

First, BM was isolated from isoproterenol-treated WT (n = 3) and Dpp3-KO (n = 3) mice. Isolated total BM from each group was pooled and injected into pre-anesthetized WT recipient mice (n = 4 per group; 100 μL per mouse) that had received intraperitoneal ketamine and local ocular anesthesia as described above. Cardiac function was measured by echocardiography at baseline, 30 min, and 1 h post-injection. Mice with a baseline LVFS <50% were excluded from further analysis.

DPP3 injection in healthy mice

Ten-week-old C57BL6/J mice were injected either with human DPP3 (hDPP3), purified from human red blood cell lysate,⁷⁸ provided by 4TEEN4, as an intravenous bolus injection of 0.55 mg/kg or PBS (n = 4 per group). Fifteen minutes post-injection, mice were sacrificed.

Echocardiography

Cardiac function was evaluated by echocardiography (VEVO LT-200, Visual Sonics) using the M-mode long-axis parasternal cut method. Echocardiography was performed blindly in mice under ketamine anesthesia (80 mg/kg). Images were analyzed using Vevo LAB software. Cardiac dysfunction was defined as a decrease of at least 10% in LVFS compared to baseline or to matching controls.

Cecal ligation and puncture-induced septic shock (CLP) model in rats

Three-month-old male Wistar rats weighing 350–450 g (ENVIGO, Puteaux, France) were used. All experiments were conducted in accordance with the National and European Institutes of Health Guidelines for the use of laboratory rats and were approved by the French Departmental Direction of Animal Protection (Project APAFIS N° 51391) and complied to ARRIVE guidelines. Only males were used due to the incompatibility of the CLP model with females. Sepsis was initiated by cecal ligation and puncture. Shock happened 16 h post-surgery as previously described.¹⁴ Sham-operated animals were used as controls. Septic shock was defined as a mean arterial pressure (MBP) < 65 mmHg and cardiac dysfunction with an LVFS <45%. At the end of the protocol, blood was collected and animals were euthanized. Bone marrow cells and BM SN were collected following the same protocol used in mice BM isolation as previously described.⁷⁶

Cardiogenic shock model in Dorset Ewe

A total of twenty-three- 35 ± 6 months-old Dorset ewe was used in this study. Their mean body weight was 49.0 ± 1.5 kg. All animals received human care in compliance with “European Community Standards on the Care and Use of Laboratory Animals” published in 2010. The project was approved by the ethical committee N°050 of the French Ministry for Research and Innovation. The study employed only females, as males exhibit aggressive behavior when housed together. Cardiogenic shock (CS) was induced, as previously described,⁷⁹ by intra-coronary injection of ethanol (CS group, n = 17), while sham animals (n = 6) underwent interventricular coronary artery catheterization without ethanol injection. CS was defined as a 30% drop in cardiac output and/or MAP, associated with a lactate level >2.5 mmol/L. Blood sampling was done at three time points; just after general anesthesia and the installation of catheters (Baseline1-BL1), CS establishment (Baseline2-BL2) and 3 h after the induction of CS (H+3). For trans-cardiac gradient analysis, blood was collected from the coronary sinus through the insertion of a JR4 catheter (0.8 Fr, under fluoroscopic guidance) at the beginning of the procedure, as well as from an arterial catheter positioned in the axillary artery. Trans cardiac gradient was calculated as follows: cDPP3 activity in coronary sinus - cDPP3 activity in axillary artery. At the end of the protocol, animals were euthanized by injection of a lethal dose of sodium pentobarbital (≥100 mg/kg) before proceeding with sternotomy and heart extraction for downstream analysis.

Human cohorts

This study was carried out according to the current revision of the Declaration of Helsinki and approved by the Ethical Committees of all participating centers in PREGICA cohort, FROG-ICU cohort and End-stage heart failure biopsies collection. Written informed consent was obtained after oral and written information from all participants, their parents, or legal guardians. All human cohorts included both sexes. Cohorts were analyzed retrospectively and patient inclusion was based on the availability of sufficient residual plasma volume for the required analyses. Details on gender and ethnicity weren’t reported.

PREGICA (NCT01113268), is a prospective observational multicentric cohort that enrolled 410 patients admitted to the cardiac/emergency department with a first acute STEMI myocardial infarction with at least 3 akinetic left ventricular segments using cardio echography (Median age 57 years, 87% male), mostly devoid of any comorbidities (15% hypertension, 15% diabetes).⁴⁶ Blood samples were obtained from 304 patients before revascularization (D0, n = 84) and/or four days after revascularization (D4, n = 303). FROG-ICU (NCT01367093) is a prospective observational multicentric cohort that recruited patients at admission to the intensive care unit.⁸⁰ Blood samples were obtained at admission to and discharge from the ICU. Patients with adjudicated diagnosis of septic (n = 6, median age: 69, 50% male) or cardiogenic (n = 11, median age 69, 45% males) shock were used. Blood samples from healthy donors (n = 8, mean age 26 years, 62.5% male) were used as controls.

All blood samples were collected on EDTA Vacutainer tubes and immediately centrifuged at 2000g for 10 min at 4°C. Plasma samples were immediately stored at −80°C until use. The samples used were thawed once before the measurements.

End-stage heart failure left ventricular biopsies were obtained from a tissue collection from the Pitié Salpêtrière Hospital⁸¹ and included patients with ischemic cardiomyopathy (ICM) (n = 11, median age 45 years, 54% male), and healthy individuals (n = 14, median age 32 years, 62% males).

Method details

Extracellular vesicles (EVs)

EVs isolation

EVs were isolated from human and mice plasma and BM SN (starting volume: 500 μL), and from cell culture SN (2–4 mL). Mouse plasma samples were pooled as needed. To ensure higher purity, EVs were isolated using two sequential methods. First, size exclusion chromatography (SEC; IZON qEVoriginal/70 nm Gen 2.) to separate EVs from free proteins. EVs-containing Fractions were pooled and centrifuged at 16,000g for 20 min at 4 °c to pellet large EVs (lEVs). The supernatant was then ultracentrifuged (118,000g, 2.5 h, 4 °c; Optima XPN-90) to pellet small EVs (sEVs). Pellets were resuspended in PBS or extraction RIPA buffer depending on the downstream application.

Extracellular vesicles characterization

EVs were characterized following the MISEV 2023 guidelines.⁴⁷ Size distribution and concentration were analyzed by Nanotrack analysis (NTA) using a Nanosight NS-300 Malvern and NTA 3.3 software, as described.⁸² EVs structure was characterized by Transmission Electron microscopy (TEM) following a traditional negative staining protocol,⁸³ with minor modifications. Briefly, 4 μL of sample was applied to glow discharged formvar-coated grids, fixed with 2% paraformaldehyde, stained with 1% uranyl acetate and examined at 120 kV using TEM (Tecnai 12, Thermo Fischer Scientific) with a 4K CDD camera (Oneview, Gatan).

Flow cytometry

Flow cytometry was done on immune cells isolated from the heart, blood and BM. Cell suspension was prepared using gentleMACS Octo Dissociator with Heaters (Miltenyi Biotec 130-096-427). The suspension was then filtered, centrifuged and incubated with ACK 1X lysis buffer to deplete red blood cells. After counting, cells were incubated in ViaDyeTM Red Fixable Viability Dye Kit (Cytek, CA, USA) for 20 min at room temperature in the dark to mark dead cells. Cells were then washed with FACS buffer (PBS +0.5% BSA) and then incubated in TruStain FcXTM PLUS anti-mouse CD16/32 (Biolegend, San Diego, CA, USA) blocking reagent for 10 min at room temperature. This step prevents the binding of non-specific antibodies via the Fc receptors. Finally, the cells were incubated in an antibody cocktail, detailed in Table S2 for 30 min at room temperature in the dark. Flow cytometry was performed on an Aurora (Cytek, CA, USA). SpectroFlo Software (Cytek, CA, USA) was used to acquire data and analyze data.

Fluorescence-activated cell sorting

FACS was done on blood samples from the previously described ISO model (n = 5 per group). Following the same protocol described in the flow cytometry section, after incubation with the standard panel of immunophenotyping antibodies detailed in Table S2, cells were washed and resuspended in FACS buffers and sorted using ARIA-FUSION - UV-1 (BD Biosciences). FACS sorted cells were then lysed, as described in the protein preparation section, for further analysis.

Biochemical activity measurements

DPP3 activity was measured in the following samples: plasma and tissue lysates from heart, kidney, lung, spleen, and liver; cell lysates from BM cells and FACS-sorted blood cells; BM supernatant (BM SN); and EVs resuspended in PBS. DPP3 activity was measured by soluble activity assay using a synthetic fluorogenic substrate of DPP3, Arg2-β-naphthylamide, as previously described.⁸ Circulating DPP3, whether present in plasma or cell culture supernatant, will be referred to as (cDPP3). In contrast, intracellular DPP3 and DPP3 found in BM SN will be referred to as (DPP3). High-sensitivity Troponin I was measured using the Abbott ARCHITECT platform. High-sensitivity troponin T was measured on a Cobas E801 immunoassay analyzer (Roche Diagnostics).

Protein extraction and western blot

Organ tissues (heart, kidney, liver, lung, and spleen) from mouse models and heart tissue from the ewe CS model were homogenized by mechanical grinding using 10–15 mg of tissue. Homogenates were lysed in RIPA buffer (100 mM Tris-HCl, pH 7.5; 1 mM KCl; 1 mM EDTA; 0.2 mM EGTA; 1 mM β-glycerophosphate; 20 mM sodium orthovanadate) supplemented with protease and phosphatase inhibitors (Sigma-Aldrich, Saint Quentin Fallavier, France). Samples were centrifuged at 10,000×g for 10 min at 4 °C, and the resulting supernatants, containing cytosolic and nuclear proteins, were collected as tissue lysates. Cell lysates from BM cells, and FACS-sorted blood cells were prepared using the same lysis buffer followed by 2 min of sonication with 30 s on/off cycles. Protein concentrations in lysates were determined by spectrophotometry using the Pierce BCA Protein Assay Kit (Thermo Fisher Scientific, France). For Western blot analysis, equal amounts of protein (20 μg) were denatured at 95 °C for 10 min, separated by SDS–PAGE on 4–20% polyacrylamide gels (Bio-Rad, Hercules, USA), and transferred onto 0.2-μm nitrocellulose membranes. Membranes were blocked and incubated with primary and secondary antibodies (Tables S3 and S4). Chemiluminescent signals were detected using the Azure imaging system (Azure Biosystems) and quantified with MultiGauge v2.02 software (Fuji).

Nucleic acid extraction and amplification

Total mRNA was extracted with the RNeasy Kit (Qiagen, Hilden, Germany; ref. 74106) following the manufacturer’s protocol, eluted in 30 μL RNase-free water, and quantified by spectrometry using Nanodrop (Thermo Fisher Scientific, USA). Reverse transcription was performed on 0.5 μg RNA per reaction using the QuantiTect Reverse Transcription Kit (Qiagen; 205313). Quantification of the genes of interest was carried out by quantitative PCR (LightCycler 96, Roche Life Science) using the FastStart Essential DNA Green Master kit (Roche Diagnostics, 06924204001, France) according to the manufacturer’s instructions and the normalization was done to the expression of Gapdh, which was used as a housekeeping gene. Genomic DNA was extracted by incubating samples in 100 μL of 50 mM NaOH at 95 °C for 30 min, followed by the addition of 10 μL of 1 M Tris-HCl (pH = 8.0). PCR amplification was performed using Go Taq G2 kit (Promega, M7823) according to the manufacturer’s instructions with primers listed in Table S5.

Immunostaining

Double immunostaining was performed on cryostat heart tissue sections using anti-DPP3, F480 and NIMP-R14 antibodies. All antibodies are listed in Tables S3 and S4. Nuclei were counterstained with DAPI, sections mounted, and images were acquired by fluorescence microscopy. For each animal, five to ten images per slide were collected and analyzed using ImageJ (version 1.53t; NIH).

Olink cytokine profiling

Cytokine levels were measured in mouse plasma and BM SN using Olink Target 48 Cytokine panel as described previously.⁸⁴ Cytokines with a non-detection frequency greater than 75% were excluded from the final analysis. A list of cytokines is provided in Table S6.

Statistical analyses

All statistical analyses and graphs were made using GraphPad Prism software version 10.3.1. The normality of distributions was assessed using the Shapiro-Wilk test. Variables are expressed as mean ± SEM and distribution. For comparisons between two groups, two-tailed Student’s or Mann-Whitney tests were performed. For experiments involving more than two independent variables, a two-way analysis of variance (ANOVA) was employed to evaluate the interaction between factors, followed by the Tukey honestly significant difference test to identify intergroup differences. For experiments involving multiple comparisons, comparisons were performed using the one-way ANOVA or Kruskal-Wallis tests followed by the t-tests or Mann-Whitney tests adjusted for multiple comparisons (Holm-Sidak) method to control for Type I error. p-values less than 0.05 were considered statistically significant. All correlation tests were done using Spearman’s rank correlation method. Significance was presented as follows: ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001 and “ns” for non-significant differences.

Published: February 20, 2026