Inhibiting DPP4 in a mouse model of HHT1 results in a shift towards regenerative macrophages and reduces fibrosis after myocardial infarction

Aims Hereditary Hemorrhagic Telangiectasia type-1 (HHT1) is a genetic vascular disorder caused by haploinsufficiency of the TGFβ co-receptor endoglin. Dysfunctional homing of HHT1 mononuclear cells (MNCs) towards the infarcted myocardium hampers cardiac recovery. HHT1-MNCs have elevated expression of dipeptidyl peptidase-4 (DPP4/CD26), which inhibits recruitment of CXCR4-expressing MNCs by inactivation of stromal cell-derived factor 1 (SDF1). We hypothesize that inhibiting DPP4 will restore homing of HHT1-MNCs to the infarcted heart and improve cardiac recovery. Methods and results After inducing myocardial infarction (MI), wild type (WT) and endoglin heterozygous (Eng+/-) mice were treated for 5 days with the DPP4 inhibitor Diprotin A (DipA). DipA increased the number of CXCR4+ MNCs residing in the infarcted Eng+/- hearts (Eng+/- 73.17±12.67 vs. Eng+/- treated 157.00±11.61, P = 0.0003) and significantly reduced infarct size (Eng+/- 46.60±9.33% vs. Eng+/- treated 27.02±3.04%, P = 0.03). Echocardiography demonstrated that DipA treatment slightly deteriorated heart function in Eng+/- mice. An increased number of capillaries (Eng+/- 61.63±1.43 vs. Eng+/- treated 74.30±1.74, P = 0.001) were detected in the infarct border zone whereas the number of arteries was reduced (Eng+/- 11.88±0.63 vs. Eng+/- treated 6.38±0.97, P = 0.003). Interestingly, while less M2 regenerative macrophages were present in Eng+/- hearts prior to DipA treatment, (WT 29.88±1.52% vs. Eng+/- 12.34±1.64%, P<0.0001), DPP4 inhibition restored the number of M2 macrophages to wild type levels. Conclusions In this study, we demonstrate that systemic DPP4 inhibition restores the impaired MNC homing in Eng+/- animals post-MI, and enhances cardiac repair, which might be explained by restoring the balance between the inflammatory and regenerative macrophages present in the heart.


Introduction
Hereditary Hemorrhagic Telangiectasia type 1 (HHT1) is a haploinsufficient genetic vascular disorder caused by mutations in the transforming growth factor beta (TGFβ) co-receptor endoglin (CD105). HHT1 is characterized by 'leaky' vessel formation due to impaired TGFβ signaling, demonstrated by internal organ bleeding, endothelial hyperplasia, arteriovenous malformations (AVMs) and recurrent epistaxis, up to 8 times a day [1,2]. Interestingly, endoglin heterozygous mice (Eng +/-) can develop the same clinical features as HHT1 patients [3,4]. In time, depending on the genetic background [5], Eng +/mice develop AVMs clearly visible in the ear, and even suffer from nosebleeds, making the Eng +/mouse a good experimental model to gain more insight in the etiology of HHT1.
The partial loss of endoglin results in reduced angiogenesis [6], and defective collateral artery formation after hind limb ischemia in Eng +/mice [7], emphasizing the necessity of sufficient endoglin expression for proper revascularization after tissue damage [8,9]. Although endoglin is mainly expressed by endothelial cells (ECs), several other cell types including smooth muscle cells (SMCs) and mononuclear cells (MNCs) have endoglin on their cell surface [10]. Due to the prominent role of endoglin in endothelial cell signaling and behavior, HHT1 is generally considered a result of endothelial dysfunction. However, it has become clear that the other cell types expressing endoglin, e.g. immune cells, may have a severe impact on angiogenesis and tissue repair as well [11][12][13].
Previously we showed that Eng +/mice display a diminished cardiac recovery after experimentally induced myocardial infarction (MI) compared to wild type littermates [14]. The reduced cardiac function is partially rescued when human control MNCs were injected into the tail vein of Eng +/mice. Interestingly, delivery of MNCs isolated from peripheral blood of HHT1 patients did not increase heart function post-MI. Furthermore, endoglin heterozygosity reduced the homing capacity of MNCs to the injured myocardium due to enhanced dipeptidyl peptidase-4 (DPP4, also known as CD26) expression levels [14,15]. This suggests that a defect in MNCs, and not only endothelial cells, may play a role in the pathology of HHT1.
During the cardiac post-injury response, homing of MNCs to the site of injury is regulated by the stromal cell-derived factor 1 (SDF1)-CXC chemokine receptor type 4 (CXCR4) axis. SDF1 levels are increased within the first 24 hours after MI [16]. MNCs expressing the SDF1 receptor CXCR4 respond to this gradient by homing to the site of injury. The SDF1-CXCR4 axis is tightly controlled by the negative regulator DPP4. The catalytic enzyme DPP4 inactivates SDF1 by cleaving off the first two of its amino-terminal peptides, thereby decreasing the ability to recruit CXCR4-expressing MNCs towards the SDF1 gradient [17].
Another important target of DPP4 is the glucagon-like peptide-1 (GLP1). Preserving GLP1 via DPP4 inhibition stimulates insulin secretion and has therefore been the recent focus for treatment of type 2 diabetes mellitus (T2DM). Interestingly, DPP4 inhibition was shown to have multiple off-target effects that are potentially beneficial in the treatment of cardiovascular disease [18,19]. These DPP4 inhibitory actions are GLP1 independent and can range from having an anti-inflammatory effect [20,21] to having a stimulating effect on MNC migration [15] and differentiation [22,23]. Pre-treating MNCs of HHT1 patients with the DPP4 inhibitor Diprotin A (DipA) before injection into the circulation restored their homing to the ischemic myocardium [15]. Unfortunately, this is not a suitable protocol for clinical practice to pretreat MNCs of HHT1 patients, as cells have to be isolated in sufficient numbers from the patients' peripheral blood, treated with a DPP4 inhibitor and re-injected. Therefore, the aim of this study is to determine whether systemic application of a DPP4 inhibitor is as effective in restoring homing of Eng +/-MNCs to the site of ischemic injury.

Animals and study design
Experiments and analyses were conducted on male endoglin wild type (Eng +/+ , or referred to as WT) and heterozygous (Eng +/-) transgenic mice and LysM-Cre-Eng fl/+ / LysM-Cre-Eng fl/fl (endoglin targeted recombination under regulation of the Lysosome M promoter) transgenic mice. All mouse strains were kept on a C57BL/6Jico background (Charles River). To obtain the Eng-conditional knockout mouse lines, endoglin floxed mice (Eng fl/fl ) were cross-bred with LysM-Cre [24,25] animals to create the LysM-Cre-Eng fl/+ and LysM-Cre-Eng fl/fl mice. All mouse experiments were approved by the regulatory authorities of Leiden University (the Netherlands) and were in compliance with the guidelines from Directive 2010/63/EU of the European Parliament on the protection of animals used for scientific purposes.
Humane endpoints were observed 5 days post-MI and onwards, as the following criteria and symptoms: when mice displayed reduced mobility, decreased grooming, and/or impaired reaction to external stimuli. In addition, for 3 days post-MI and onwards: when the wound area displayed bleeding, swelling, redness and/or discharge, the mice would be euthanized by carbon dioxide. The mice were weighed at the day of surgery and at the cardiac ultrasound time points and euthanized when more than 15% loss of weight occurred. All mice that died before meeting the criteria for euthanasia-just after myocardial infarction or within 10 days post-MI-died because of cardiac rupture due to the deterioration of cardiac tissue after ligation of the left anterior descending coronary artery. Animal health and behavior were monitored on a daily basis by the research and/or animal care staff, all trained in animal care and handling. Once animals reached endpoint criteria, the euthanasia was performed immediately or at the least the same day when reported.

Myocardial infarction in mice
Myocardial infarction (MI) was experimentally induced as described before [15]. The mice (n = 5-18 per group) were anesthetized with isoflurane (1.5-2.5%), intubated and ventilated, after which the left anterior descending (LAD) coronary artery was permanently ligated by placement of a suture. The mice were treated with the analgesic drug Temgesic, both pre-operative and 24 hrs post-operative to relieve pain. The mice were randomly allocated and treated i.p. with either 100 μl distilled water daily (Milli-Q ultrapure, sterile water = MQ treated or control group) or 100 μl DPP4 inhibitor (5 nMol, 55 μg/kg/day, Diprotin A, Sigma-Aldrich) for the first 5 or 14 days post-MI.

Cardiac function measurements
Mice were anesthetized with isoflurane (1.5-2.5%), after which cardiac ultrasound was performed and recorded with the Vevo 770 (VisualSonics, Inc., Toronto, CA) system, using a 30 MHz transducer (RMV707B). Imaging was performed on the longitudinal axis of the left ventricle using the EKV (Electrocardiography-based Kilohertz Visualization) in long axis view imaging mode. The percentage ejection fraction was determined by tracing of the volume of the left ventricle during the systolic and diastolic phase using the imaging software Vevo770 V3.0 (VisualSonics, Inc., Toronto, CA).

Cultured macrophages from mouse bone marrow
Monocytes were isolated from the mice femur and tibia and subsequently cultured in RPMI 1640 culture media (#11875093, Gibco, ThermoFisher Scientific), supplemented with 10% FBS (#10270, Fetal Bovine Serum, Gibco, ThermoFisher Scientific) and 1 ng/ml granulocyte-macrophage colony stimulating factor (GM-CSF, #315-03, Peprotech) to induce differentiation into macrophages. Macrophage cultures from 3 mice of each genotype were pooled to obtain sufficient protein to perform Western blot analysis.

Western blotting
Cultured macrophages were lysed on ice with cold radio immunoprecipitation assay (RIPA) lysis buffer (in house) supplemented with protease inhibitors (Complete protease inhibitor cocktail tablets, Roche Diagnostics, #11697498001) and protein concentration was measured using BCA protein assay (Pierce BCA Protein Assay Kit, #23225, ThermoFisher Scientific). Equal amounts of protein were loaded onto 10% SDS-polyacrylamide gel and transferred to an Immobilon-P transfer membrane (# IPVH00010, PVDF membrane, Millipore). The blots were blocked for 1 h using 10% milk in Tris-Buffered Saline and 0.1% Tween-20 solution and incubated O/N with goat anti-mouse endoglin (1:1000 dilution, BAF1097, R&D Systems) or mouse anti-β-Actin (1:10.000 dilution, A5441, Sigma-Aldrich). Blots were incubated for 1 h with horse radish peroxidase anti-goat (goat anti-mouse IgG Poly-HRP Secondary Antibody HRP conjugate, #32230, ThermoFisher Scientific) or anti-mouse (ECL mouse IgG, HRPlinked whole Ab #NA931, Sigma-Aldrich, GE Healthcare, UK). Blots were developed in a Kodak X-omat 1000 processor with Thermo Scientific SuperSignal West Dura (Extended Duration Substrate) or SuperSignal West Pico and exposed to Fuji SuperRX medical X-ray film. Analysis was performed using Image J (National Institute of Mental Health, Bethesda, Maryland, USA).
All acquisitions were performed on a LSRII flow cytometer (BD Biosciences) and analyzed by FACS Diva software (BD Biosciences) and Flowing software 2.5.1 (Cell Imaging Core, Turku Centre for Biotechnology, Finland). Gating strategies are provided in S7 Fig.

Morphometry
Infarct size was determined in Picrosirius Red staining images by calculating the percentage infarct area of the total left ventricular area. Cell infiltration was determined by quantification of 2 to 4 digital images per heart, at the border zone inside the infarcted area, taken at 40x magnification (CaseViewer 3D Histech). The same method was used for quantification of capillary and artery presence, except analysis was now at the outside borderzone of the infarct, where cardiomyocytes were still viable. Data were blinded to the investigator and quantified by using ImageJ v1.46r (NIH, USA).

Statistics
All results are expressed as mean ± standard error of the mean (SEM). Statistical significance was accepted at p<0.05. Statistical significance was evaluated using one-way ANOVA testing for difference between multiple groups. Adjustment for multiple comparisons with either Tukey's or Dunnett's testing and unpaired students T-testing for testing between two groups, using Graphpad Prism v6 for Windows. For data with groups where n = 3, non-parametric testing was performed with Kruskal-Wallis ANOVA and Dunn's multiple comparisons test. Significant differences between survival curves were tested with a log-rank Mantel-Cox test.

Involvement of macrophage-expressed endoglin during cardiac regeneration
We have previously shown that HHT1 MNCs are impaired in their homing capacity towards ischemic tissue [14]. To establish the endogenous contribution of MNCs to the Eng +/phenotype, a LysM-specific mouse model for targeting myeloid cells (monocytes, macrophages and granulocytes) was generated. Mice with either a heterozygous or homozygous deletion for LysM-specific endoglin were analyzed. To validate recombination, MNCs were allowed to differentiate towards macrophages in the presence of GM-CSF and western blot analysis showed reduced endoglin expression in the LysM-Cre-Eng fl/+ , and a near absence of endoglin in the LysM-Cre-Eng fl/fl macrophages (Fig 1A and 1B).
The deletion of endoglin in monocytes did not affect survival post-MI ( Fig 1C). Analysis of cardiac function showed a trend towards reduced ejection fraction (EF) in the LysM-Cre-Eng fl/fl mice ( Fig 1D, WT 34.07±3.65% vs LysM-Cre-Eng fl/fl 22.73±4.30%, P = 0.09), suggesting an involvement of endoglin in macrophage function during cardiac repair. However, since cardiac dysfunction of the LysM-Cre-Eng fl/fl phenotype post-MI is not as pronounced as the reduction in EF seen in the Eng +/mice after MI (Fig 1D, WT 32.45±4.06% vs. Eng +/-17.98 ±2.06%, P<0.05), we concluded that the injury response in HHT1 is likely to be an interplay between a multitude of cell types, only partially represented by the LysM population. We therefore continued by analyzing the effect of DPP4 inhibitor treatment in the Eng +/animals.

DPP4 inhibition restores the in vivo MNC homing capacity in Eng +/mice
Macrophage infiltration is important for proper repair of damaged tissue [12, 28,29]. Pretreatment of the HHT1-MNCs, expressing enhanced levels of DPP4, with a DPP4 inhibitor prior to intravenous injection after experimentally induced MI in the mouse restored homing of these cells to the infarct site. Therefore, we asked the question whether systemic administration of a DPP4 inhibitor would have the same effect, and treated mice with the DPP4 inhibitor Diprotin A (DipA) systemically via intraperitoneal (i.p.) injection from day 0 till day 5 post-MI. Flow cytometric analysis of blood samples at baseline revealed that the mice did not display any leukopenia because of endoglin heterozygosity (S1 Fig; gating  As observed for the CXCR4 expressing cells, DipA treatment did not affect the number of macrophages present in the infarcted hearts of wild type mice. Since there are two main populations of macrophages involved in tissue repair [12], we analyzed the effect of DPP4 inhibition on the inflammatory-like (M1) and regenerative-like (M2) macrophage subtypes in the infarct border zone (Fig 2D). Analysis of the percentage of M2 in the total macrophage population (represented by the percentage MAC3/CD206 positive cells) revealed a decrease in M2 in Eng +/mice compared to wild types (Fig 2E, WT 29.88 ±1.52% vs. Eng +/-12.34±1.64%, P<0.0001). After DipA treatment, there was a significant increase in the percentage of M2 macrophages within the total population of macrophages in both wild type and Eng +/mice, and consequently a decrease in M1 macrophages. Importantly, DipA was able to  restore the M1/M2 ratio in Eng +/mice to WT levels. We further corroborated our findings by flow cytometric analysis of the macrophage population in the infarct area. Inflammatory M1 (Ly6G -/CD11b + /Ly6C high ) were increased in the Eng +/at baseline compared to regenerative M2 macrophages (Ly6G -/CD11b + /Ly6C low ). DPP4 inhibition showed a trend towards a decrease in M1 macrophages (Fig 2F and gating strategy provided in S7D Fig). Thus, systemic DipA treatment of Eng +/mice restores homing of MNCs to, as well as the macrophage M1/ M2 balance in the injured heart 4 days post-MI.

DPP4 inhibition reduces infarct size
By treating the mice with DipA from day 0 to 5 post-MI, homing was stimulated during the peak of inflammatory cell influx (Fig 3A). To assess the effect of DipA treatment on infarct size we quantified the fibrotic area using Picrosirius red staining at day 14 post-MI (Fig 3B and  3C). As expected, control Eng +/mice showed an increase in infarct size compared to WT controls ( Fig 3C, WT control 24.30±2.12% vs. Eng +/control 46.60±9.33%, P = 0.009). Upon DipA treatment, no effect on infarct size was detected in wild type animals (Fig 3C, WT DipA 27.41 ±3.74%), however in Eng +/mice a significant decrease in infarct size was observed (Fig 3C, Eng +/control 46.60±9.33% vs. Eng +/-DipA 27.02±3.04%, P = 0.02), which resulted in similar infarct sizes as observed in the wild type mice.

Effect of DPP4 inhibition on cardiac function
Since DPP4 inhibition enhanced homing of circulating MNCs to the injured heart, reduced infarct size and increased the number of regenerative macrophages present in the heart, we investigated the effect of DipA treatment on cardiac function after MI using ultrasound at baseline, and day 7 and day 14 after MI. Baseline analysis showed that cardiac function at baseline was similar between wild type and Eng +/mice ( As we observed that endoglin heterozygosity skews the M1/M2 ratio towards a more inflammatory profile, both homing and/or differentiation of Eng +/-MNCs might be delayed. Therefore, we prolonged the daily DipA treatment from 5 till 14 days post-MI. Extended DPP4 inhibition resulted in a similar decrease in cardiac function in wild type mice compared to short term (5 days post-MI) treated mice (Fig 4C). While initially DPP4 inhibition in the Eng +/animals improves cardiac function, Eng +/mice showed a gradual reduction of cardiac function in the 6 month follow-up (ΔEF in Fig 4C and %EF is provided in S6 Fig). Thus, neither short term nor prolonged DipA treatment resulted in a long term beneficial effect on heart function in Eng +/mice.

DPP4 inhibition increases angiogenesis but decreases vessel maturation in Eng +/animals
Since neoangiogenesis is important for cardiac repair post-MI, we investigated the effect of DipA treatment on vascularization by staining the hearts for PECAM-1 and αSMA and analyzed the presence of capillaries and arteries in the infarct border zone. The number of capillaries in the Eng +/mice was increased compared to wild type animals (Fig 5A). In both wild type as well as Eng +/mice, DipA treatment further increased the number of capillaries present in the border zone (Fig 5A, Eng +/-61.63±1.43 vs. Eng +/treated 74.30±1.74, P = 0.001). Surprisingly, the number of arteries in the infarct border zone was significantly lower in the DPP4 inhibitor treated Eng +/mice (Eng +/-11.88±0.63 vs. Eng +/treated 6.38±0.97, P = 0.003), which may suggest that vessel maturation or arteriogenesis is impaired (Fig 5B).  tissue repair. Interestingly, a significantly higher number of macrophages was observed in the infarct border zone of the Eng +/mice compared to the wild type mice 14 days post-MI ( Fig  5C, WT 7.06±0.97 vs. Eng +/-25.13±2.76, P<0.0001). DipA treatment of Eng +/did not reduce the total number of macrophages to wild type levels.
Analyzing the macrophage phenotypes, we observed a disbalance in the M1/M2 ratio in the hearts of Eng +/mice 4 days post-MI. Therefore, macrophage quantity and subtype in the heart was also analyzed 14 days post-MI. Interestingly, Eng +/-DipA treated mice showed an increase in M2 levels reaching wild type levels (Fig 5D WT 51.7% vs. Eng +/treated 52.7%), while in wild type mice DipA treatment resulted in a significant drop in the percentage of M2 cells (Fig  5D WT 51.7% vs. WT treated 22.0%, P = 0.04). These results suggest that there is a prolonged inflammatory response and possibly an impaired resolution of inflammation in Eng +/mice, and the macrophage balance can be skewed towards the M2 subtype by DPP4 inhibition.

Discussion
In this study we demonstrate that after MI, the decrease in myocardial function observed in our HHT1 animal model is not solely caused by a disturbed endothelium response, but also in part by impaired MNC function. MNC homing and differentiation into inflammatory and regenerative macrophages is important for proper tissue repair. In our previous research, we found that MNC homing towards the ischemic myocardium is severely affected by endoglin heterozygosity, together with enhanced DPP4 levels [15]. Here, we demonstrate that homing of MNCs to the ischemic myocardium could be restored by systemic DPP4 inhibition. Although DipA treatment did reduce the infarct size in Eng +/mice to similar levels as in wild type animals, it did not result in long term improvement of EF. Detailed analysis of the infarcted ventricle demonstrated that neovascularization was increased in the Eng +/infarct border zone upon DPP4 inhibition, while arteriogenesis or vessel maturation was dramatically decreased. Furthermore, analyzing the macrophage subsets, a prolonged presence of inflammatory-like macrophages (M1) was observed. We therefore conclude that the inflammatory and regenerative function of Eng +/macrophages is affected. Interestingly, DPP4 inhibition could restore this M1/M2 imbalance to WT levels.
Complete endoglin deficiency in mice results in embryonic lethality around mid-gestation due to cardiovascular defects and the formation of enlarged, fragile vessels [4]. Eng +/mice or endothelial deletion of endoglin in adult mice results in a relatively mild phenotype. Only after a secondary hit like induction of tissue damage or inflammation, the impairments of endoglin deficiency become apparent in the form of reduced angiogenesis and the development of AVMs [30][31][32]. In the current study, we observed that the macrophage specific deletion of Eng did not result in similar levels of cardiac deterioration as found in Eng +/animals, implying a multifactorial nature of HHT1. This emphasizes the need for endoglin heterozygosity to be present in all cell types of the heart to recapitulate the HHT1 disease phenotype.
Furthermore, we show that the homing capacity of MNCs in the Eng +/mice could be restored by systemic DPP4 inhibitor treatment. In cardiovascular patients an association was found between increased DPP4 presence on MNCs and poor recovery after coronary intervention [33]. Together with our observation of decreased homing capacity of the MNCs, HHT1 now proves to affect other cell types and functions as well, in particular the lymphocytes and monocytes [31,32,34,35]. Macrophages are an important subfraction of the MNC population, as macrophages have both inflammatory and angiogenic capacity and are essential in tissue regeneration and remodeling [28]. The shift of the inflammatory-like macrophages (M1) to the regenerative-like macrophages (M2) is essential for optimal resolution of inflammation [36]. Immune function and the angiogenic capacity of macrophages are most likely impaired in Eng +/mice [37,38]. Recent studies suggested that there is an intrinsic inflammatory defect in HHT1 patients as well. Investigators report leukopenia, increased risk of infection [34] and impaired endoglin upregulation in activated monocytes of HHT1 patients [35]. We now observed an increased ratio of M1/M2 macrophages in Eng +/mice hearts post-MI, which may indicate that the increase in M1 macrophages is due to either impaired differentiation of the M1 towards M2 macrophages, or decreased homing capacity of M2 macrophages. In renal ischemia-reperfusion injury, the same disbalance in M1/M2 was also reported in Eng +/mice [39]. DPP4 inhibition has been shown to shift the M1/M2 balance towards more reparative macrophages [22,23]. We show that in Eng +/mice DipA treatment restored both the total number of macrophages as well as the M1/M2 ratio. This effect was maintained up to 10 days after cessation of DPP4 inhibitor treatment. The shift of the macrophage population towards M2 induced by DPP4 inhibition was also observed by Brenner et al. [22], when analyzing macrophage presence in the aortic walls of ApoE -/mice in a model for atherosclerosis. Likewise, a reduction of atherosclerotic lesions [20] and total macrophage presence was reported in ApoE -/mice treated with DPP4 inhibitor, together with a reduction in smooth muscle cell proliferation [40]. DPP4-null mice had an increased expression of M2 genes [41] and furthermore, DPP4 inhibition had anti-inflammatory effects on macrophages in general by decreasing pro-inflammatory NFκB signaling [42]. An increase in M2 macrophages was also observed upon DPP4 inhibitor treatment in another atherosclerosis model using LDLR -/mice [43], moreover, a recent meta-analysis showed a decreased infection rate in T2DM patients treated with a DPP4 inhibitor [44]. Here we show for the first time, to our knowledge, that DPP4 inhibition polarizes the M1/M2 ratio towards a more regenerative phenotype in a mouse model for HHT1. Furthermore, our data suggests that endoglin heterozygosity not only impairs homing of the MNCs, but affects proper differentiation and/or function of the macrophages, as we find more M1 at the infarct site and a decrease in resolution of immune cells 14 days post-MI in Eng +/animals.
In addition to improved homing of circulating (stem) cells, we and others show that DPP4 inhibition significantly stimulated vascularization and cardiac repair [45][46][47]. In conjunction, an increase in capillary number was observed at the infarct border zone, confirming an increase in endothelial regeneration, possibly by enhanced presence of regenerative M2 macrophages. Interestingly, the number of arteries present was found to be significantly lowered in the DPP4 inhibitor treated Eng +/mice. Vessel maturation is known to be affected in Eng +/mice, due to dysfunctional mural cells and pericytes [48,49]. Our results therefore suggest that DPP4 inhibitor treatment most likely further impairs vessel maturation in Eng +/mice, due to its involvement in vascular smooth muscle cell recruitment [50,51].
DPP4 inhibitors have been reported to increase cardiomyocyte survival, however the exact mechanism is not yet fully understood. DPP4 has many targets besides SDF1, like atrial natriuretic peptide (ANP)-a protein associated with cardiomyocyte survival [52]. DPP4 inhibition might therefore decrease the inactivation of ANP by DPP4, and contribute to cardiomyocyte survival/protection. Furthermore, when MI was performed in DPP4 -/mice, these mice show an increased survival compared to DPP4 +/+ mice [52]. DPP4 inhibitor treatment in previous reports already showed a modest, albeit not significant gain of cardiac function in wild type mice after MI [53]. Interestingly, at day 14 post-MI the infarct size in Eng +/mice treated with DipA was significantly smaller and similar in size compared to wild type control levels. Concordantly, we observed an initial increase in EF upon DipA treatment. However, improved cardiac function in Eng +/mice was not preserved at day 14 post-MI. These results suggest that DPP4 inhibition conveys a decrease of the fibrotic response in Eng +/animals, resulting in a reduced infarct size, though this is not reflected in a sustained improvement of heart function. This could be in part explained by the fact that although capillary presence is increased, in HHT1 endothelial functionality is impaired, forming leaky vessels. Secondly, the misbalanced MNC population present in HHT1 might result in an exaggerated and prolonged immune response, which is likely to disturb tissue repair.
We hypothesized that when homing of Eng +/-MNCs to the infarct site was restored, cardiac function would improve, as was observed in wild type animals [46,53]. Although increased homing and a decrease in infarct size were observed, this did not correlate with a long term positive effect on cardiac function. The Eng +/mice showed either a reduced differentiation towards, or presence of the regenerative macrophage subtype and interestingly this level could be corrected by DPP4 inhibition. However, although the fibrotic response was restored to wild type levels and MNC recruitment was re-established, the Eng +/mice still show a decrease in cardiac function. This suggests that the impaired function of the Eng +/-MNCs that homed to the infarct site still exists. Combining DPP4 inhibition with G-CSF treatment for enhanced MNC recruitment, Zaruba et al. showed that ameliorated MNC homing resulted in a significant increase in cardiac function in mice. However, testing this treatment in patients suffering from MI, the combination failed to show any beneficial effect on cardiac function [54,55].
How DPP4 inhibition affects MNC function is still unclear. DPP4 inhibition is already being tested in several clinical trials in cardiovascular disease (CVD), and meta-analyses of the risks involved have proven to be varied in outcome. Several meta-analyses have been performed in the field of T2DM. Upon DPP4 inhibition, no increase in CVD was observed in T2DM patients, but risk of infection was reduced [44]. In addition, although no effect was seen on cardiovascular mortality, short term DPP4 inhibition reduced chances of MI events, while long term treatment showed a possible increase in heart failure cases [56,57]. Other meta-analyses showed either no effect or a reduction of cardiovascular events, also with long term treatment [58][59][60][61].
As described in this study, DPP4 inhibition positively affects cardiac fibrosis and initially, also cardiac function. We showed by using a PSR staining that fibrosis is enhanced in Eng +/mice and decreases upon DPP4 inhibition. A relation between DPP4 inhibition and reduction in fibrosis was also demonstrated in a kidney model [62]. Furthermore, inhibition of DPP4-positive fibroblasts reduced scarring of murine dermal wounds [63] and DPP4 inhibition also mediated antifibrotic effects in dermal fibroblasts [64]. DPP4 inhibition is shown to be cardioprotective in several studies (reviewed in Grilo et al. [65]). A study overexpressing DPP4 resulted in promoted mammary tumorigenesis and transformation of epithelial cells [66], altogether indicating DPP4 inhibition is the way forward.
DPP4 and its inhibition has a multitude of functions and effects, many of which are still poorly understood. This is exemplified by the study from Zhu et al. [67], where they show that DPP4 is able to target neuropeptide Y and peptide YY in the cardiac nerves, proteins that can result in activation of cardiac fibroblasts. To the contrary, many anti-fibrotic aspects have been described, and by various mechanisms; DPP4 inhibitors were reported to lower active TGFβ (via reduction of DPP4 -CIM6PR membrane receptor interaction) and also decrease fibronectin expression [68][69][70]. Together, the beneficial effects of DPP4 inhibition on MNC homing, reduction in fibrosis and restoration of regenerative macrophage differentiation could provide a suitable treatment for general tissue repair in HHT1 patients. However, this study suggests that treating HHT1 patients with a DPP4 inhibitor post-MI should not be considered as monotherapy, but ought to be combined with additional MNC stimulating agents and/or arteriogenic stimulation. Although DPP4 inhibition is an accepted treatment in T2DM patients, a better understanding of its mode of action should first be gained before further conducting clinical trials for applying this treatment on any other disease or disorder like HHT1.

S1 Fig. Eng +/mice do not display leukopenia.
Flow cytometric analysis of the major leukocyte subsets in the circulation of the mice groups at day 0 (pre-MI and pre-DipA treatment). Leukocytes labeled for anti-mouse CD3, CD3 + /CD4 + , CD3 + /CD8 + and CD11b + /Ly6G -(n = 3-6, non-parametric ANOVA testing). Control = MQ treated, DipA = Diprotin A treated group. Data shown are mean ± SEM, Ã P<0.05. (B) Monocytes gating strategy in the blood. The first gating step is a gate for live cells using FSC and SSC. The second gating step is for monocytes, identified by CD11b positive and Ly6G negative labeling. (C) Granulocyte gating strategy in the blood. The first gating step is a gate for live cells using FSC and SSC, the granulocytes are subsequently identified by Ly6G. (D) Macrophage subsets gating strategy from MNCs isolated from the left ventricle. The first gating step is a gate for live cells using FSC and SSC, the monocytes are identified by CD11b positive and Ly6G negative labeling. The inflammatory-like M1 macrophages are then subsequently identified by Ly6C high and regenerative-like M2 macrophages identified by Ly6C low labeling. (TIF)