Introduction

In the Western countries, the mortality from acute myocardial infarction (AMI) has decreased continuously¹ during the last decades. In particular, the incidence of ST-segment–elevation myocardial infarction (STEMI) decreased in the developed countries because of the effective primary and secondary prevention strategies and the treatment of risk factors/comorbidities.² However, globally, the incidence of both STEMI and non-STEMI is increasing, and severe complications, such as heart failure and cardiac arrhythmia, develop in survivors of myocardial infarction. Therefore, mechanisms that could be targeted in the acute phase after AMI or myocardial ischemia/reperfusion (I/R) to improve the long-term perspective of STEMI and non-STEMI patients especially with respect to heart failure and sudden cardiac death are of great clinical interest.^3,4

Meet the First Author, see p 1402

The ECM (extracellular matrix) provides a 3-dimensional scaffold to the heart that facilitates force transduction and maintenance of cellular architecture. Further, ECM forms a stable microenvironment facilitating the function of both cardiomyocytes and noncardiomyocytes.⁵ On ischemic injury and cardiomyocyte death, the ECM is acutely damaged by, for example, reactive oxygen species and activation of MMPs (matrix metalloproteinases). Subsequently, remodeling of the ECM is initiated involving de novo synthesis of ECM components not present in the healthy heart.⁵ Cardiac ECM remodeling, especially de novo formation and cross-linking of the collagenous matrix, stabilizes the infarct scar and is important for adaptation.⁶ However, continued remodeling that leads to interstitial fibrosis impairs contractile function and contributes to diastolic dysfunction.⁷

Instead of targeting the collagenous remodeling that occurs chronically after I/R, it is attractive to define mechanisms that are activated early after cardiac ischemia and set the stage for the chronic adverse remodeling. Among these mechanisms, the inflammatory response has been analyzed in great detail.⁸ Myocardial ischemia leads to cardiomyocyte death and endothelial leakage initiating an intense immune cell infiltration that aims to remove cellular debris and damaged ECM. Almost all immune cells of the innate and adaptive immune response are involved in this sterile inflammation, and monocyte/macrophages are thought to be the most important effector cells for infarct healing.⁹ This acute inflammatory phase is followed by a reparative, proliferative phase that involves both immune cells (reparative macrophages) and the activation of fibroblasts and angiogenic responses. However, our understanding of the fibroblast response is far from complete.^5,10 Similarly, the role of the complex noncollagenous matrix¹¹ that is synthesized de novo early after infarct is incompletely understood. However, this early provisional ECM likely is—in conjunction with growth factors, cytokines, and chemokines—critical for the phenotypes of immune cells, cardiac fibroblasts, pericytes, and endothelial cells in the acute phase after cardiac ischemia.

An interesting candidate ECM system in this regard is hyaluronan (HA). HA is a carbohydrate formed from repeating disaccharides (D-glucuronic acid and N-acetyl-D-glucosamine) linked by a glucuronidic β(1→3) bond. HA is synthesized by HAS (HA synthases) 1–3 at the plasma membrane and associates with different HA-binding proteins to form pericellular and extracellular HA matrix networks. Furthermore, HA signals through CD (cluster of differentiation) 44 and RHAMM (receptor of HA-mediated motility). From other pathologies such as neointimal hyperplasia,¹² atherosclerosis,¹³ and autoimmune¹⁴ diseases, it has been recognized that HA strongly affects immune responses.¹⁵ Additionally, HA drives volume expansion and activation of fibroblasts during disease-associated remodeling and embryonic development.^16,17 These roles of HA are, in part, mediated by HA receptors RHAMM and CD44 and, in part, by the fact that HA provides a very loose permissive matrix that facilitates cell movements. The abovementioned functions of HA contribute to wound healing and on the contrary in cardiovascular diseases also promote pathological hyperplastic responses including intimal hyperplasia during atherosclerosis.^15,18

In the adult heart, HA accumulates strongly in the perivascular matrix, but only very small amounts are present in the interstitial matrix of the myocardium. Interestingly, cardiac HA synthesis by HAS2 is critically required for cardiac development during embryogenesis. Specifically, chamber septation and heart valve formation are disturbed in Has2^−/− mice causing embryonic lethality.¹⁹ Deletion of the other 2 Has genes, Has1 and Has3, did not lead to a cardiac phenotype. Although we and others reported the accumulation of HA post-myocardial infarction, the function of HA post-myocardial infarction is yet unknown.^20,21

In the current study, we aimed to investigate the role of the early cardiac HA matrix after I/R injury of the heart. Furthermore, we attempted to develop a magnetic resonance imaging (MRI) approach for cardiac HA matrix to monitor HA accumulation in vivo and to provide a translational perspective for future studies.

Methods

The data that support the findings of this study are available from the corresponding author on reasonable request. A detailed description is provided in the Online Data Supplement.

Mice

Male 12- to 18-week-old C57BL/6J mice (Janvier Labs, Le Genest-Saint-Ile, France), Rosa26CreER^T2+/− mice (Taconic, Hudson, NY²²), and Rosa26CreER^T2+/−;Has2^flox/flox mice,²³ as well as Has1^−/− mice²⁴ and respective littermate controls, all on a C57BL/6J background, were used for experiments. Rosa26CreER^T2+/− mice and Rosa26CreER^T2+/−;Has2^flox/flox mice were treated intraperitoneally for 5 days with 1 mg per mouse per day tamoxifen (Sigma, Darmstadt, Germany) 3 weeks before induction of myocardial I/R injury. All mice underwent myocardial I/R injury as described previously.^25,26 All mice with a sufficient knockdown of Has2 that received a successful I/R injury as determined by visualization of ST-segment elevation and displayed no signs of infection after surgery were included in the analyses. Mice were excluded from the experiment when certain criteria of suffering were observed. These included weight loss >20% of body weight, cessation of food and water ingestion, or lack of voluntary movement.

Permission for animal experiments was granted by the Landesamt für Natur, Umwelt und Verbraucherschutz Nordrhein-Westfalen, Bezirksregierung Düsseldorf, Aktenzeichen 84-02.04.2012.A138 and Aktenzeichen 84-02.04.2015.A322.

Magnetic Resonance Imaging

Data were recorded at a Bruker AVANCE^III 9.4T wide bore NMR (nuclear magnetic resonance) spectrometer driven by ParaVision 5.1 (Bruker, Rheinstetten, Germany) and operating at frequencies of 400.21 MHz for ¹H and 376.54 MHz for ¹⁹F measurements. Images were acquired using a Bruker microimaging unit Micro 2.5 with actively shielded gradient sets (1.5 T/m) and a 25-mm birdcage resonator tunable to both ¹H and ¹⁹F as described previously.^27,28 Mice were anaesthetized with 1.5% isoflurane and kept at 37°C. The front paws and the left hind paw were attached to ECG electrodes (Klear-Trace; CAS Medical Systems, Branford), and respiration was monitored by means of a pneumatic pillow positioned at the animal’s back. Vital functions were acquired by a M1025 system (SA Instruments, Stony Brook, NY) and used to synchronize data acquisition with cardiac and respiratory motion.

For functional analysis, high-resolution images of mouse hearts were acquired in short-axis orientation using an ECG- and respiratory-gated segmented cine fast gradient echo cine sequence with steady-state precession essentially as described before.²⁹ A flip angle of 15°, echo time of 1.2 ms, and a repetition time of about 6 to 8 ms (depending on the heart rate) were used to acquire 16 frames per heart cycle. The pixel size after zero filling was 117×117 μm² (field of view, 30×30 mm²; acquisition time per slice for 1 cine sequence, ≈1 minute). Eight to 10 contiguous slices were acquired to cover the entire heart. Routinely, 8 to 10 short-axis slices were required for complete coverage of the left ventricle (LV). For evaluation of functional parameters (eg, end-diastolic volume, end-systolic volume, and ejection fraction), ventricular demarcations in end diastole and end systole were manually drawn with the ParaVision region-of-interest tool (Bruker, Rheinstetten, Germany). For late gadolinium enhancement, a bolus of Gd-DTPA (gadolinium-diethylenetriaminepentacetate; 0.2 mmol Gd-DTPA per kg body weight) was applied intraperitonally.

Results

Interference With Cardiac HA Matrix Formation Post-I/R Severely Impairs Hemodynamic Function

HA has been attributed important roles in proliferation of various cell types in wound healing¹⁶ and in inflammation.¹⁵ From morphological analyses in canine and murine infarcts,²⁰ it is known that HA affinity staining increases after I/R at 14 days in dogs and after 3 and 7 days in mice. Here, we demonstrate an even earlier induction of Has1 and Has2 mRNA in the murine I/R model at 6 hours post-I/R (Online Figure I) suggesting an acute onset of cardiac HA synthesis after I/R. Considering the rapid onset of Has1 and Has2 expression and the known functions of HA in other tissues and diseases, we tested the hypothesis that cardiac HA synthesis after I/R injury represents an endogenous repair mechanism.

To analyze the function of de novo HA synthesis after I/R, we utilized mice deficient for the 2 enzymes that were induced after I/R, Has2 and Has1, to determine the involvement of endogenous HA synthesis in an experimental model of I/R. Whereas Has1 deletion is constitutive, Has2 deletion was induced by application of tamoxifen in adult mice as described in the Online Data Supplement because of embryonic lethality of constitutive Has2 KOs (Has2 deficient).¹⁹ In addition, mice were treated with the pharmacological inhibitor of HA synthesis, 4-methylumbelliferone (4-MU).

Inhibition of HA synthesis by deletion of Has2 (Figure 1) and by 4-MU (Online Figure IIA) did not affect the initial ischemic area as determined by late gadolinium enhancement. Importantly, in the long-term follow-up deletion of Has2 (Figure 1B) and 4-MU treatment (Online Figure IIB) severely impaired hemodynamic function as indicated by decreased ejection fractions determined by MRI. However, whereas Has2 deletion was characterized by strongly increased end-systolic and end-diastolic volumes (Online Table III), 4-MU treated mice developed no changes end-systolic volume and end-diastolic volume but instead showed strongly decreased stroke volume and cardiac output. This difference to Has2 KO mice and the known off target effects of 4-MU such as biosynthesis inhibition of other polysaccharides containing D-glucuronic acid, the inhibition of MMPs, and metabolic effects in vivo led us to focus only on the genetic deletion of Has isoforms for mechanistic experiments in vivo. 4-MU was considered as additional proof of concept for the development of the chemical exchange saturation transfer (CEST) MRI (see below, Figure 2) in parallel with Has2- and Has1-deficient mice.

• Download figure
• Download PowerPoint

• Download figure
• Download PowerPoint

To focus on the long-term effects of Has2 deletion, the scar size was analyzed 3 weeks after I/R and was found to be increased in Has2 KO mice (Figure 1C). In contrast, deletion of Has1 had no significant effect on cardiac function and scar size (Online Figure III). Tension-to-rupture experiments were performed in LV stripes 3 weeks days after I/R to evaluate whether the stability of the scar tissue was affected as shown before in mice deficient in collagen modifying proteoglycans.³⁰ This analysis revealed only a trend toward reduced scar stability in Has2KO hearts (Online Figure IV) and thereby excludes major changes in the collagenous matrix in the chronic phase after I/R. Next, effects of HA on angiogenesis were considered. Twenty-one days after I/R, angiogenic responses in the remote and borderzone of Has2KO mice were unaffected (Online Figure V).

These data clearly indicate that endogenous HA synthesis post-I/R especially by HAS2 is an immediate, protective response that does not acutely affect infarct size but is instead required in the subacute phase post-I/R. This was confirmed in Langendorff experiments that revealed no acute functional differences after I/R between control and Has2 KO hearts (Online Figure VI) ex vivo.

HA Accumulates Early After I/R Injury and Can Be Imaged by CEST

To characterize the endogenous HA response in the ischemic hearts in more detail, we attempted to adjust MRI CEST techniques^31,32 to monitor cardiac matrix accumulation including HA after I/R injury as a function of time after injury. Although the CEST signal is not specific for a certain matrix component, the amount of exchangeable protons is especially high in hydroxylated sugar moieties of HA and glycosaminoglycans. As shown in Online Figure VIIA, HA gives rise to the strongest CEST signal followed by proteoglycans and heparin, whereas collagen (gelatin and fibrillar collagen), hydroxyproline, and matrigel result in much weaker signals in z-spectra acquired from individual phantoms. Similar z-spectra were obtained from the LV 24 hours after I/R in mice (Online Figure VIIB and VIIC). With respect to the location of HA, strong CEST signals were detected in the infarcted LV with lower intensity in the border zone and almost no signal in the remote myocardium (Online Figure VIIC) 24 hours post-I/R injury. To further enhance specificity and to avoid potential contaminations from CEST-active components in the blood (in particular sugars and glycosylated compounds), before induction of saturation, a black blood preparation was initiated for suppression of all signals from circulating blood (see the Online Data Supplement for more details, Online Figure VIII). In line with the CEST signal, affinity histochemistry detected HA accumulation already 24 hours post-I/R (Online Figure IX). In contrast, the staining of other ECM molecules associated with collagen scar formation revealed that these did not accumulate.²⁰ Specifically, collagen stained by Picrosirius Red and immunostaining of 2 collagen-binding proteoglycans that are known to be upregulated post-AMI (decorin and biglycan) showed no positive signal 24 hours post-I/R (Online Figure IX). This is in line with upregulation of these matrix compounds simultaneously or subsequently to fibroblast activation,^20,30 which is initiated around 3 days post-I/R. Together, these data show that cardiac HA accumulation occurs within 24 to 48 hours post-I/R and is detectable by CEST in vivo.

After establishment of the cardiac CEST MRI, the CEST approach was used (1) to monitor cardiac HA over time and under different experimental conditions and (2) to investigate mice treated with the pharmacological inhibitor of HA synthesis, 4-MU, and mice deficient of Has2 and Has1. As shown in Figure 2A, both 4-MU and Has2 KO led to a strongly reduced CEST signal 24 hours after I/R injury. The reduction in HA was more pronounced in the infarct region compared with the remote zone. In contrast, only a small decrease by trend occurred in Has1 KOs. Subsequently, the temporal development of the CEST signal was compared with the time course of morphological analysis of HA, collagen, biglycan, and decorin (Figure 2B through 2E). In the respective control groups, the CEST signal increases about 3-fold within 24 hours post-I/R (Figure 2B and 2D). Both 4-MU treatment and Has2 deficiency caused a strong reduction of cardiac CEST signals during the first 3 days post-AMI (Figure 2B and 2D). These data clearly indicate that cardiac CEST MRI allows to monitor HA accumulation post-I/R, especially in the first 1 to 3 days post-I/R. Interestingly, after 7 and 14 days, the CEST signals in 4-MU and Has2 KO mice approached those of controls most likely representing the increasing accumulation of the other ECM components such as proteoglycans, cellular fibronectin, thrombospondin, and subsequently, collagens. As mentioned above, the lack of diminished CEST signal in Has1 KO mice after I/R (Figure 2A) suggested that HAS1-derived HA is quantitatively less compared with the HAS2-derived HA. In line with this assumption is the finding that the hemodynamic function was not affected in Has1 KOs in contrast to Has2 KO and 4-MU–treated mice (Online Figure III; Online Table III).

HAS2 Is Required for Myofibroblast Responses

Next, it was considered that Has2 deletion affects either the fibroblast response or the inflammatory response post-I/R. First, fluorescent stainings of HA/α-SMA (α-smooth muscle actin) and HA/Mac-2 were performed in the infarct zone revealing that both cell types are in close contact with the cardiac HA matrix (Figure 3A and 3B). Isolated cardiac fibroblasts released substantial amounts of HA basally and in response to TGF (transforming growth factor)-β1 in vitro (Figure 3C). Of note, although about 10× less, monocytes isolated from the bone marrow released HA as well, and even bone marrow–derived macrophages secreted small amounts of HA. The main isoenzyme within fibroblasts was HAS2 (Figure 3D).The fibroblast/myofibroblast response was analyzed by immunostaining of vimentin that labels all cardiac cells with the exception of cardiac myocytes and α-SMA indicating myofibroblast activation (Figure 4A and 4B). Whereas vimentin was not affected, the amount of α-SMA–positive cells was decreased in Has2 KO versus control mice. Further analysis revealed that α-SMA–positive cells were reduced by trend in the infarct zone and significantly in the border zone but not in the remote myocardium (Figure 4C through 4E). The proliferation of myofibroblasts at 72 hours after I/R injury trended toward reduction in Has2 KO hearts as indicated by Ki67/α-SMA double stainings (Figure 4C and 4D). TGF-β1 is induced early after I/R and is an important growth factor for myofibroblast activation and reparative remodeling of ECM, as well as a modulator of the transition between the inflammatory phases after I/R.³³ To address possible changes in canonical and noncanonical TGF-β signaling, cardiac LV lysates were analyzed 72 hours after I/R with respect to phosphorylation of SMAD2 (mothers against decapentaplegic homolog), ERK (extracellular signal–regulated kinases) 1/2, and an AKT (serine/threonine protein kinase; Online Figure X) by immunoblotting. Interestingly, inhibition of AKT phosphorylation was evident in samples of Has2 KO mice. Furthermore, small trends toward reduced SMAD2 and ERK1/2 phosphorylation were observed.

• Download figure
• Download PowerPoint

• Download figure
• Download PowerPoint

For further mechanistic insight, fibroblasts were isolated from Has2 KO and control hearts and analyzed in vitro in the first passage. Has2 KO fibroblasts were less responsive to FCS with respect to proliferation (Figure 5A) underlining the trends toward less proliferation observed in vivo 72 hours post-I/R. As expected, Has2 expression (data not shown) and HA synthesis were reduced in Has2 KO cells (Figure 5B). Compensatory upregulation of Has1 and Has3 was not observed on deletion of Has2 (data not shown). The ability of cardiac fibroblasts to contract fibrillar collagen gels in vitro was used as a functional test of the myofibroblastic phenotype. Indeed both Has2 deletion and 4-MU (Figure 5C) strongly reduced the ability of cardiac fibroblasts to contract collagen gels. The myofibroblastic differentiation as detected by Acta2 mRNA expression in response to TGF-β1 was strongly reduced if HA synthesis was inhibited by 4-MU (Figure 5D) or a blocking antibody directed toward CD44’s HA-binding domain (Figure 5E) was applied. To further address the role of CD44 in activation of TGF-β1 signaling, SMAD2 phosphorylation was determined. As shown in Figure 5F, blocking CD44 indeed reduced TGF-β1–induced phosphorylation of SMAD2, whereas noncanonical TGF-β responses, such as AKT and ERK phosphorylation, were not affected (data not shown).

• Download figure
• Download PowerPoint

HA/HAS2 Is Needed for Cardiac Macrophage Responses Post-I/R

The uptake of perfluorocarbon particles was used to monitor the accumulation of phagocytic immune cells (mainly macrophages²⁷) post-I/R in Has2 KO mice. In turn, ¹H/¹⁹F MRI²⁷ revealed a strongly reduced ¹⁹F signal from perfluorocarbon-labeled immune cells in the infarcted LV indicative of anti-inflammatory effects in Has2 KO mice (Figure 6A). This was paralleled by changes in the concentration of circulating chemokines and cytokines, such as reductions of MCP-1/CCL2 (C-C chemokine receptor type 2), IL (interleukin)-13, and RANTES/CCL5 (CC-chemokine ligand 5; Online Figure XI). Subsequently, detailed flow cytometric measurements of the heart at different times post-I/R were performed. At 24 hours post-I/R, cardiac immune cells in Has2 KO hearts (Online Figure XII) and blood (Online Figure XIII) were comparable to control mice. To exclude possible earlier effects especially on invading neutrophils, hearts were also analyzed 12 hours post-I/R. However, no differences between Has2 KO mice and control mice were found (data not shown).

• Download figure
• Download PowerPoint

Subsequently, cardiac immune cells were investigated 72 hours post-I/R. Total cardiac monocytes (MHCII^lowCCR2⁺), neither Ly6C^high nor Ly6C^low, were not altered in Has2 KO hearts compared with control (Figure 6B through 6D). However, the amount of cardiac macrophages (F4/80⁺CD64⁺) and monocyte-derived macrophages (MHCII^highCCR2⁺) was decreased (Figure 6E and 6F). Resident MHCII^high CCR2⁻ macrophages were also reduced in Has2 KO mice in contrast to MHCII^lowCCR2⁻ macrophages (Figure 6G). A representative gating scheme is depicted in Online Figure XIV. Cardiac lymphocyte subsets were not affected (Online Figure XV). In addition, circulating immune cells (Online Figure XVI), splenic immune cells (Online Figure XVII), and immune cells in the bone marrow (Online Figure XVIII) were unaffected 72 hours after I/R.

Our data indicate that monocyte recruitment to the ischemic heart was not disturbed (as indicated by comparable numbers of systemic and local invading monocytes). Furthermore, we detected no differences between Has2 KO and control mice in mRNA expression of chemokines, cytokines, and cytokine receptors 72 hours post-I/R as determined by a polymerase chain reaction array from infarcted LV samples (Online Table V). This suggested that the expression of inflammatory genes by the majority of cardiac cell is, at least in average, unchanged. Instead, our FACS (fluorescence-activated cell sorting) data (Figure 6) suggest very specific effects on monocyte-derived macrophages and resident macrophages. Mechanistically, either effects on monocyte to macrophage differentiation, macrophage proliferation, or an increase in macrophage apoptosis were considered. Differences in macrophage proliferation were ruled out by Ki67/Mac-2 costaining in the total LV (Figure 7A), as well as in infarct, border, and remote zone (Online Figure XIX). Unchanged proliferation was confirmed by flow cytometric enumeration of BrdU (5-bromo-2’-deoxyuridine)-positive monocytes (Figure 7B), total BrdU⁺ macrophages (Figure 7C), and BrdU⁺ monocyte-derived macrophages (Figure 7D) in the infarcted heart. When investigating monocyte-to-macrophage transition in vitro, we observed no differences between monocytes isolated from control mice and Has2 KO mice in response to M-CSF (macrophage colony-stimulating factor; Online Figure XX). Importantly, we detected more apoptotic macrophages (Figure 7E and 7F) in hearts from Has2 KO mice compared with control mice 72 hours post-I/R. In line, increased apoptosis was also detected in cultures of bone marrow–derived macrophages in conditions with reduced HA synthesis or CD44 signaling thereby pointing toward antiapoptotic HA-mediated effects on macrophages (Online Figure XXIA and XXIB). Using a blocking antibody for the HA-binding domain of CD44 (Online Figure XXIC), these in vitro experiments suggested that effects on macrophage apoptosis are dependent on HA/CD44 signaling. Interestingly, no effects on apoptosis were observed when investigating annexin V binding in MEFSK4⁺ fibroblasts by flow cytometry 72 hours post-I/R (Online Figure XXII). Finally, the secretome of bone marrow–derived macrophages of Has2 KO mice was analyzed (Online Figure XXIII). These data indicate reduced secretion of cytokines such as CCL2 and CXCL1 (chemokine (C-X-C motif) ligand 1).

• Download figure
• Download PowerPoint

Has2 Deficiency Impairs Instructive Interactions of Monocyte/Macrophages and Fibroblasts

Finally, the interrelationship between fibroblasts and monocytes/macrophages was investigated. Coculture experiments showed a strong effect of monocytes on fibroblast activation and promotion of a myofibroblast phenotype as shown by α-SMA staining (Online Figure XXIVA) and increased Acta2 mRNA expression (Online Figure XXIVB). Importantly, cell activation was also detected vice versa in monocytes when cultured in the presence of fibroblasts: As shown in Online Figure XXIVC, transition from monocytes to macrophages was increased in the presence of fibroblasts to a similar extent as in the presence of M-CSF. Further, coculture of monocytes with fibroblasts protected from macrophage apoptosis (Online Figure XXIVD). However, this effect was independent of Has2 expression in fibroblasts.

In summary, the data obtained in Has2 KO mice suggest that decreased cardiac HA synthesis immediately post-I/R causes both impaired fibroblast to myofibroblast differentiation, as well as an increase in macrophage apoptosis, ultimately leading to impaired healing post-I/R. The main findings of the current investigations are summarized in Figure 8.

• Download figure
• Download PowerPoint

Discussion

The current results clearly show that murine hearts react strongly with increased HA synthesis and HA deposition to I/R injury. It is shown that inhibition of this HA response by genetic deletion of Has2 aggravates the hemodynamic dysfunction and increases scar size. Importantly, the HA-rich microenvironment critically affects 2 major cell types that ensure proper infarct healing: macrophages and cardiac fibroblasts. Thereby, the early HA response appears to be part of critical endogenous repair mechanisms after I/R.

The adult mouse myocardium contains large quantities of cardiac fibroblasts that maintain the cardiac ECM. On I/R, fibroblasts become activated and develop a secretory and proliferative phenotype. The concomitant release of fibroblast-derived cytokines and growth factors, as well as the de novo synthesis and deposition of ECM contribute to the reparative second phase after clearing of damaged cells and ECM. In addition to strongly decreased inflammation, Has2 KO hearts are characterized by reduced α-SMA–positive fibroblasts in the infarct and border zones. The current results suggest that this is not due to increased fibroblast apoptosis but instead due to decreased myofibroblastic differentiation and proliferation.

TGF-β1 is a well-recognized regulator of fibroblast and myofibroblast function, for example, during wound healing, and of the orchestration of ECM organization and homeostasis in noncardiac tissues. Less is known in the heart after AMI,³³ but TGF-β1 is likely crucial for the myofibroblast response after cardiac I/R. Of note, TGF-β1 can exert both beneficial and detrimental effects.³³ Whereas early inhibition of TGF-β1 impaired the outcome after experimental AMI,³⁴ inhibition in the subacute phase reduced adverse cardiac remodeling.³⁵ Here, we observed in cardiac fibroblasts after blocking of the HA receptor CD44 in vitro inhibition of TGF-β1–specific responses such as reduced SMAD2 phosphorylation and reduced α-SMA expression. These findings strongly suggest that HA-mediated signaling through CD44 promotes activation of TGF-β1 signal transduction in cardiac fibroblasts. SMAD2 phosphorylation has been used as readout for activation of TGF-β1 signaling. This is in line with reduced numbers of myofibroblasts after I/R in Has2 KO mice. It has been demonstrated that CD44 contributes to TGF-mediated SMAD phosphorylation in fibroblasts and cancer cells^36,37 and that CD44 cooperates with epidermal growth factor receptor in enhancing TGF-β1–mediated fibroblast activation.³⁸ TGF-β1 also activates noncanonical SMAD-independent signaling pathways including ERK and PI3K (phosphoinositide 3-kinase)³⁹ that were also addressed in the present study. Both have been shown to contribute to fibroblast activation.³⁹ The present data in Has2 KOs and experiments blocking of CD44, however, do not allow the conclusion that SMAD2 phosphorylation itself is critical for the observed phenotype. SMAD3 and noncanonical TGF-β1 signaling cascades may be considered in detail in future studies. SMAD3 is a promising candidate for further studies and has been attributed an important role in modulating the myofibroblast phenotype after I/R, especially regarding the control of fibroblast proliferation, scar remodeling, collagen synthesis, and cellular alignment in the infarct but not α-SMA expression⁴⁰ as altered in Has2 KOs.

TGF-β1 also regulates monocyte/macrophage-mediated inflammation. It activates monocytes and inhibits macrophages and thereby contributes to the temporal and spatial fine-tuning of inflammation and wound healing.³³ Evidence suggests that both initial activation of inflammatory response and subsequent suppression of inflammation and the stimulation of reparative programs by TGF-β1 are an important part in orchestration of the different phases of cardiac healing after I/R.^41,42 However, the precise functions of TGF-β1 in inflammation await further experimentation. Therefore, the observed interference between the early HA-rich microenvironment after I/R and the modulation of TGF-β1 responses by CD44 in fibroblasts could also extend to monocytes/macrophages. All considered, we hypothesize that the loss of HAS2 leads to reduced HA/CD44 signaling that in turn impairs myofibroblastic differentiation and function.

These results and conclusions are in line with data in Cd44 KO mice showing ventricular dilatation, decreased fibroblast response and decreased fibroblast proliferation, and collagen synthesis.⁴³ Moreover, it has been reported that the addition of HA promotes the contraction of collagen gels by myofibroblasts compared with collagen gels without HA.⁴⁴ In support of the assumption that CD44 fine-tunes the functional phenotype of cardiac fibroblasts, is our recent finding that IL-6 induces HA synthesis in cardiac fibroblasts. This in turn activates CD44 resulting in stimulation of cytokine release by fibroblasts.²¹ Furthermore, stimulation of cytokine release by HA/CD44 representing the proinflammatory fibroblast phenotype might connect the 2 observations made here that inhibition of HA synthesis in acute I/R causes both inhibition of macrophage-mediated inflammation and fibroblast activation.

It is shown here that in myocardial infarction, endogenous HA regulates the macrophage response. The impaired macrophage response likely contributes to the aggravated phenotype after I/R in Has2 KOs. There is ample evidence in the literature that HA modulates inflammatory responses by a variety of mechanisms including stimulation of TLRs (Toll-like receptors) and providing an adhesive matrix for macrophages.¹⁸ Another facet of HA in inflammation is its promoting effect on Th1 cell polarization that contributes to inflammation in autoimmunity and atherosclerosis.¹³ On the contrary, from studies in atherosclerosis, it is known that inhibition of HA synthesis impairs the endothelial glycocalyx, thereby facilitating adhesion and extravasation of monocytes and augmenting macrophage-driven inflammation and atheroprogression.⁴⁵ The observed phenotype in Has2 KO mice after I/R was very specific for cardiac immune cells because immune cells in all other compartments such as bone marrow, blood, and spleen were unaffected. The effect was also specific for cardiac macrophages because lymphocytes, neutrophils, and even monocytes were unchanged by Has2 deletion. The decrease in cardiac macrophages was driven by decreased numbers of MHCII^highCCR2⁺ monocyte-derived macrophages but also detected in resident MHCII^highCCR2⁻ macrophages. In general, MHCII-expressing macrophages (CCR2⁺ and CCR2⁻) are important coordinators of cardiac inflammation and crucial for antigen presentation.⁴⁶ Further, CCR2⁺ macrophages, not only monocyte-derived but also including resident macrophages, are known to be crucially involved in the immune response post-cardiac ischemia.^47,48 Resident macrophages contribute to the repair response post-myocardial infarction.⁴⁹ Particularly, the MHC^low subset has been described to be important for the removal of dead cells⁴⁶ but also to trigger the recruitment of other immune cells such as neutrophils and monocytes to the heart after injury.⁴⁸ Because of these complex and divergent functions of cardiac macrophage subsets, experimental approaches to target those cells resulted in contrary effects on the functional outcome dependent on which cell type was mainly affected: inhibition of monocyte influx and thereby subsequent decrease in cardiac CCR2⁺ macrophage content showed beneficial effects and reduced adverse remodeling and inhibited heart failure.^47,50 In contrast, broader approaches targeting all macrophages or also a reduction in resident macrophages showed opposite effects.⁵¹

Macrophage proliferation was not affected and the differentiation of monocytes into macrophages, based on in vitro results, was also independent of Has2 deletion. However, apoptosis of cardiac macrophages after I/R was increased in Has2 KO mice. Moreover, in vitro inhibition of HA synthesis and blocking CD44 also induced apoptosis in bone marrow–derived macrophages suggesting that the HA matrix present post-infarction augments the macrophage response by creating an antiapoptotic microenvironment. Antiapoptotic effects of CD44 have been reported in fibroblasts and malignant cells.⁵² Further, a recent study showed that CD44/HA binding is critical for the survival of monocyte-derived alveolar macrophages by promoting HA binding and thereby a protective HA coat.⁵³ Interestingly, phosphorylation of AKT was reduced in vivo in Has2 KO mice. It might, therefore, be considered that reduction of AKT signaling, which is in line with decreased macrophage apoptosis, might be caused by reduced CD44 signaling in macrophages. Indeed, CD44 has been shown to activate PI3K in macrophages, and phosphorylation of AKT inhibits macrophage apoptosis.⁵⁴ However, the role of HA/CD44 for macrophage apoptosis after cardiac I/R and the possible involvement of TGF-β1 signaling remains an important question for future studies.

In addition, the present results hint toward the contribution of a simple but possibly important mechanism that is based on instructive interactions between fibroblasts and monocyte/macrophages. It is shown here that monocytes/macrophages promote myofibroblastic differentiation in line with previous findings that cardiac fibroblasts support the transition of monocytes into macrophages and that fibroblasts protect macrophages from apoptosis. Thereby, the two effects namely reduction in macrophages and reduction in myofibroblasts might enforce each other and thereby aggravate the healing defect in Has2 KO hearts. The main findings and conclusions are schematically depicted in Figure 8.

Whereas determination of the properties of myocardial tissue and the extracellular volume by MRI using T1/T2 mapping and gadolinium enhancement is feasible and may be used for classification of disease stages,⁵⁵ it is, however, desirable to visualize and discriminate specific pathophysiologic processes after myocardial infarction. For this purpose, molecular imaging techniques that allow the visualization of specific molecular patterns have been developed in other preclinical studies.⁵⁶ To date, in cardiac imaging of the ECM post-myocardial infarction, it was attempted to visualize cardiac collagen or collagen degradation by MMPs. Cardiac collagen is associated with cardiac maladaptation and cardiac fibrosis leading to stiffening and hemodynamic dysfunction. Thereby, cardiac collagen is a marker of the relatively late phase in the (mal)adaptation to I/R. As shown here, HA matrix is likely the first cell-derived matrix that is elaborated post-I/R injury. Increased cellular fibronectin deposition has also been shown after 24 hours but continues to rise during a period of 7 days,²⁰ whereas Has1 and Has2 expression peak at 6 and 48 hours after I/R suggesting that the HA response is very concise (Online Figure I). To address the early remodeling, we present here a CEST MRI method to image the provisional HA matrix after I/R by combining the CEST approach with black blood adjustment. CEST MRI clearly visualized the HA response within the first 72 hours after I/R as proved by the suppression of the CEST signal by 4-MU and in Has2 KO mice. Introduction of a black blood preparation before the saturation transfer avoided potential contaminations from CEST-active components in the blood (in particular sugars and glycosylated proteins) and enhanced the specificity of the approach for imaging of the HA matrix. This allowed the sensitive detection of alterations in magnetization transfer ratio asymmetry images already 24 hours after I/R, which has not been reported before.

In conclusion, the current data show that cell-mediated HA synthesis by HAS2 is an acute response to cardiac ischemia, which is indispensable for proper macrophage and fibroblast responses and thereby crucial for infarct healing. Cardiac HA may be a promising prognostic marker or therapeutic target in the acute phase post-I/R because of the rapid onset of cardiac HA synthesis and its functional importance for fibroblast and immune responses.

Acknowledgments

We wish to acknowledge Petra Rompel and Kerstin Freidel for technical support.

Footnotes