Blocking Interleukin-1β transiently limits left ventricular dilation and reduces cardiac lymphangiogenesis in a mouse pressure-overload model

Blocking inflammatory pathways, e.g. the inflammasome or interleukin (IL)-1β, is a promising therapeutic approach in heart failure (HF). We hypothesized that IL-1β may stimulate cardiac lymphangiogenesis in response to left ventricular (LV) dilation following pressure-overload. Consequently, blocking IL-1β may reduce lymphangiogenesis, delaying resolution of myocardial edema and inflammation, aggravating cardiac fibrosis, and accelerating HF development. Here we investigated the effects of anti-IL-1β treatment during HF development following pressure-overload, induced by transaortic constriction (TAC) in BALB/c mice prone to LV dilation. We examined the molecular mechanisms of IL-1β in macrophages and lymphatic endothelial cells, and assessed links between perivascular fibrosis and lymphatics in HF patients with ischemic or dilated cardiomyopathy (DCM). We found that early anti-IL-1β treatment transiently delayed LV dilation, but did not alter cardiac lymphangiogenesis, hypertrophy, or dysfunction at 8 weeks post-TAC. In contrast, late anti-IL-1β treatment reduced cardiac lymphangiogenesis in response to LV dilation. This was linked to a cell non-autonomous role of IL-1β in promoting cardiac lymphangiogenesis through stimulation of macrophage production and maturation of VEGF-C. Surprisingly, despite reduced lymphatic density in late anti-IL-1β-treated mice, cardiac inflammation, interstitial fibrosis, and dysfunction were not aggravated. Further, we found that perivascular lymphatic density, unaltered by IL-1β, was negatively associated with perivascular fibrosis in HF patients and our TAC model. In conclusion, IL-1β blockage elicited transient functional cardiac benefit when initiated before onset of LV dilation post-TAC in mice. In contrast, late treatment reduced cardiac lymphangiogenesis, but did not significantly accelerate HF development, likely reflecting lymphatic transport dysfunction post-TAC. Our data suggest that the therapeutic window for anti-IL-1β treatment is crucial, as initiation of treatment during the lymphangiogenic response, induced by LV dilation, may diminish the potential cardiac benefit in HF patients. Finally, our data indicate that IL-1β-independent perivascular lymphangiogenesis may limit perivascular fibrosis.


Introduction
Interleukin-1β (IL-1β) signaling has emerged as a promising therapeutic target to limit inflammation in cardiovascular diseases, as highlighted in the CANTOS study, where Canakinumab reduced major adverse cardiovascular events in atherosclerotic patients. 1Currently, other IL-1β blocking agents, such as Anakinra, are being evaluated in the setting of ischemic heart failure (HF) 2 .While IL-1β protein levels also are increased in non-ischemic dilated cardiomyopathy (DCM) HF patients 3 , fewer experimental studies have investigated the cardiac impact of IL-1β-targeting in nonischemic HF, such as following pressure-overload induced by transverse aortic constriction (TAC).IL-1β not only regulates inflammation and fibrosis (in part by stimulating IL-6), but also impacts cardiac function directly due to negative inotropic effects 4 .In addition, IL-1β may stimulate cardiomyocyte hypertrophy via upregulation of insulin-like growth factor (Igf1) 5 .The potential of IL-1β as a therapeutic target to limit HF development in nonischemic cardiomyopathy currently remains uncertain.Most recently, CC chemokine receptor-2 (CCR2)-expressing myeloid cells have been suggested as a key source of IL-1β in failing human hearts, and found to be essential for driving cell-fate commitment of cardiac fibroblasts into periostin (Postn)-expressing matrifibroblasts mediating interstitial fibrosis after pressure-overload in mice 6,7 .Promisingly, reduction of cardiac IL-1β production during pressure-overload, by early treatment with a selective pharmacological inflammasome (NLRP3) inhibitor, reduced both cardiac hypertrophy and fibrosis at 5 weeks post-TAC in male C57Bl/6J mice. 8However, another study, employing an IL-1β-blocking antibody, also post-TAC in male C57Bl/6J mice, demonstrated that the reduction in left ventricular (LV) hypertrophy led to accelerated LV dilation and HF development 9 , potentially linked to reduced IL-1β-induced VEGF-A-mediated cardiac angiogenesis in response to cardiac hypertrophy.Similarly, a recent study, based on Nlrp3 gene deletion in male C57Bl/6J mice, revealed that reduction of early compensatory cardiac hypertrophy following TAC led to accelerated LV dilation and aggravated cardiac dysfunction 10 .
Recently, we demonstrated that while pressure-overload leads to compensated concentric hypertrophy in female C57Bl/6J mice, in BALB/c mice, characterized by elevated cardiac Il1b gene expression post-TAC, it leads to eccentric hypertrophy with severe LV dilation. 11Moreover, we demonstrated that LV dilation-induced wall stress, but not LV hypertrophy per se, triggers massive cardiac lymphangiogenesis, which however fails to limit cardiac inflammation and edema. 119][20] Interestingly, we previously observed a striking correlation between cardiac Il1b and Vegfc expression 11 , in line with previous reports demonstrating both direct and indirect pro-lymphangiogenic effects of IL-1β in other organs.This includes stimulation of VEGF-C-mediated proliferation of lymphatic endothelial cells (LEC) via an autocrine loop 18 , as well as IL-1β-mediated activation of VEGF-C-release by macrophages to promote lymphangiogenesis during chronic inflammation in other tissues 21 .In contrast, local inflammation induces lymphatic dysfunction, as pro-inflammatory mediators, including IL-1β, act on lymphatic smooth muscle cells to reduce lymphatic drainage. 22ditionally, in the heart, where lymphatic drainage depends directly on the force of cardiac contraction 23 , the negative inotropic effects of IL-1β may contribute to cardiac lymphatic dysfunction in cardiovascular diseases.
Here, we investigated the impact of anti-IL-1β treatment on cardiac lymphatics and HF development post-TAC in BALB/c mice prone to LV dilation.We hypothesized that elevated cardiac Il1b may provide a molecular link between increased wall stress and cardiac lymphangiogenesis in response to pressure-overload.Specifically, we argued that the striking lymphangiogenesis observed in our TAC model may be induced by elevated cardiac IL-1β levels.This would mechanistically couple cardiac wall stress (induced by LV dilation) to local VEGF-C production by cardiac cells, as cardiac fibroblast IL-1β secretion is induced by mechanical stress 5 .We expected that interfering with IL-1β signaling post-TAC may have deleterious effects including i) reduction of compensatory cardiac hypertrophy leading to accelerated LV dilation; and ii) reduction of cardiac lymphangiogenesis leading to aggravation of inflammation and fibrosis.To investigate the cardiac functional and lymphatic impact of IL-1β in different stages of pressure-overload-induced HF, we applied an anti-IL-1β antibody (Gevokizumab) before and after development of LV dilation in BALB/c mice (early and late treatment, Fig. 1a).Further, to dissect the mechanisms involved in IL-1β-induced lymphangiogenesis, we investigated the molecular effects of cytokine stimulation in vitro in macrophage and LEC cultures.

Results
Early, but not late, anti-IL-1β treatment delays LV dilation and dysfunction but does not limit HF development post-TAC.
Indeed, the reduction in LV hypertrophy index, observed in TAC controls starting from 3 weeks, was delayed in early-treated mice (Fig. 1e).However, at the end of the study, this beneficial functional cardiac effect was lost, and treated animals displayed similar reduction of fractional shortening (FS) as compared to control TAC mice (Fig. 1d).
In contrast, late anti-IL-β treatment tended to increase LV dilation (Fig. 1b, Table 1), although it did not significantly aggravate cardiac dysfunction, hypertrophy, or pulmonary edema at 8 weeks (Fig. 1d-g).Moreover, cardiac gene expression of Nppa and Nppb (encoding ANP and BNP respectively) were similarly increased, as compared to TAC controls, at 8 weeks post-TAC in both early and late treated mice (Fig. 1h).Evaluation of cardiomyocyte sizes revealed moderate cardiac hypertrophy at 8 weeks post-TAC, which was unaltered by anti-IL-β treatment (Fig. 1j).

Only early anti-IL-1β treatment moderately affects cardiac inflammation post-TAC
We and others have previously shown biphasic upregulation of cardiac IL-1β during experimental pressure-overload, with a rapid increase (from 3 days post-TAC 21 ), followed by a second progressive increase (from 3-8 weeks post-TAC). 11,24Our cardiac immunohistochemical analyses revealed increased cardiac IL-1β levels, notably in areas of macrophage accumulation and interstitial fibrosis (Suppl.Fig. 2a).Pressureoverload not only activates inflammatory pathways in the heart, but also systemic innate immune cell activation, which in the chronic setting leads to splenomegaly. 24,25 found that both early and late anti-IL-1β treatment significantly reduced spleen weight at 8 weeks post-TAC (Fig. 2a), reflecting systemic anti-inflammatory effects of the treatment, as we previously described in rats. 26Only late anti-IL-1β treatment led to a compensatory increase in circulating IL-1β plasma levels, as evaluated by ELISA (Fig. 2b).Of note, rather than neutralizing the cytokine, Gevokizumab acts by reducing IL-1β binding to the IL-1R receptor. 26In contrast, neither early nor late treatment altered IL-1β cardiac protein or gene expression levels post-TAC (Fig. 2c, d; Suppl.Fig. 2a).Unexpectedly, anti-IL-1β treatment did not reduce cardiac gene expression of other proinflammatory cytokines, including Il6 and Il33, evaluated by qPCR, as compared to control TAC mice (Fig. 2e).Of note, all three cytokines, and notably Il6, correlated with the degree of LV dilation at 8 weeks post-TAC (Suppl.Fig. 2b-d).
Taken together, the observed limited benefit of early anti-IL-1β treatment on cardiac inflammation is in line with the transient and moderate improvement of cardiac dysfunction post-TAC seen in our study.In contrast, our finding that late anti-IL-1β treatment did not suffice to reduce cardiac inflammation following pressure-overload indicates that other pro-inflammatory cytokines induced by cardiac wall stretch, such as IL-6 4 , may counteract reduced IL-1β signaling, notably in the decompensated phase post-TAC.

Anti-IL-1β treatments alters cardiac lymphangiogenesis post-TAC.
We previously demonstrated that cardiac Vegfc gene expression is increased starting from 3-4 weeks post-TAC in BALB/c mice, stimulating cardiac lymphangiogenesis during the subsequent rapid LV dilation occurring in the decompensation phase (6-8 weeks post-TAC). 11Western blot analysis of cardiac samples at 8 weeks post-TAC revealed that VEGF-C protein levels, specifically the ratio between mature and immature forms of the growth factor, were significantly increased post-TAC as compared to sham mice (Fig. 3a), while VEGF-D protein levels were not altered (Suppl.Fig. 3a).These results are in line with our previous report that cardiac lymphangiogenesis post-TAC in BALB/c mice is essentially mediated by VEGF-C. 11urther, our new data suggest that enhanced maturation of cardiac VEGF-C may occur post-TAC in BALB/c mice.Interestingly, we observed subpopulation-dependent correlation of cardiac macrophage densities and lymphatic densities post-TAC, with infiltrating macrophages potentially exerting deleterious effects, while tissue-resident macrophages were linked to enhanced lymphangiogenesis (Suppl.Fig. 3b, c).
Similarly, tissue-resident macrophages have recently been proposed as a preferential source of VEGF-C in the heart. 27xt, to investigate the impact of IL-1β on macrophage production of VEGF-C, we stimulated bone marrow-derived macrophages with recombinant cytokine for 4h or 24h and analyzed gene expression levels of Vegfc and associated genes involved in VEGF-C maturation, such as Adamts3 that cleaves VEGF-C N-terminal propeptide under guidance of Ccbe1 28 .We found that IL-1β induced transient upregulation of Ccbe1, while both Vegfc and Adamts3 levels were significantly increased at both 4h and 24h of stimulation (Table 2).Further, whereas IL-1β did not impact expression of Cathepsin D (Ctsd) or proprotein convertases (Psck5 or Psck7), three additional proteases known to modulate VEGF-C maturation 28 , it did stimulate expression of Furin, an intracellular proprotein convertase that matures pro-VEGF-C by cleaving its C-terminal propeptide, after 24h of IL-1β stimulation.In agreement, western blot analyses of macrophage-conditioned media revealed an increase in mature VEGF-C levels following IL-1β stimulation (Suppl.Fig. 4).In contrast, in murine lymphatic endothelial cells (LECs), which expressed low, but detectable levels of Il1r1, IL-1β did not upregulate Ctsd or Furin (Suppl.Table 2).Further, the levels of expression in murine LECs of other proteases involved in VEGF-C maturation were either very low (Pcsk7), or undetectable (Adamts3, Adamts14, Pcsk5, Ccebe1), both at baseline and after IL-1β.In addition, whereas Vegfc was not readily quantifiable in LECs, Vegfd expression was not influenced by IL-1β.In contrast, Flt4 levels were increased (Suppl. Table 2).These findings thus support a majoritarian cell non-autonomous role of IL-1β in stimulation of lymphangiogenesis in the heart.Further, our data are in line with previous reports demonstrating a key role of macrophages in regulation of cardiac lymphangiogenesis post-TAC 27 as well as after MI 29 .
As mentioned above, early anti-IL-1β treatment reduced infiltrating macrophage levels post-TAC, and we found that this potentiated lymphatic capillary expansion at 8 weeks (Fig. 3b, Suppl.Fig. 5).However, the treatment did not modify cardiac VEGF-C levels (Fig. 3a), open vessel densities, or sizes at 8 weeks (Fig. 3c, d).Moreover, cardiac gene expression analysis revealed that early anti-IL-1β treatment did not alter levels of key lymphatic markers and regulators (Pdpn, Ccl21, Lyve1, Vegfc, and Vegfd) as compared to control TAC mice (Suppl.Table 3).
In contrast, in agreement with our in vitro data in macrophages supporting our hypothesis of a role of IL-1β in linking LV dilation to stimulation of lymphangiogenesis, we found that late anti-IL-1β, initiated at 4 weeks post-TAC, before LV dilation, potently reduced cardiac levels of mature VEGF-C (Fig. 3a), leading to reduced lymphangiogenesis at 8 weeks (Fig. 3b, c, Suppl.Fig. 5) despite pronounced LV dilation in both late-treated and TAC control groups (Fig. 1b, c).Interestingly, while lymphatic capillary and precollector (open vessel) densities were reduced, lymphatic vessel sizes were increased (Fig. 3d).This enlargement of lymphatics ("profile switch" from 5-20 µm size range to 20-50 µm vessel diameters (Suppl.Fig. 3d)) was sufficient to partially compensate for the reduction in lymphangiogenesis following anti-IL-1β treatment, and total cardiac lymphatic areas remained unchanged, as compared to TAC controls (Fig. 3e).In contrast, coronary perivascular lymphatic density, increased post-TAC, was not altered by either early or late anti-IL-1β treatment (Fig. 3f).

Although cardiac expression analyses indicated that late anti-IL-1β treatment reduced
Vegfd levels (Suppl.Table 3), western blot analyses of VEGF-D did not confirm this (Suppl.Fig. 3a).Further, despite the reduced lymphatic density, we found that late anti-IL-1β treatment increased cardiac Lyve1 and Ccl21 gene expression, as compared to control TAC mice (Suppl.Table 2).However, immunohistochemical analyses revealed that CCL21 lymphatic gradients were not increased, but paradoxically decreased in late-treated mice (Fig. 3g, h).Notably, reduced CCL21 gradients may limit lymphatic drainage of CCR7 + immune cells, including macrophages.However, as noted previously, we did not observe any increase in cardiac macrophage levels at 8 weeks in late anti-IL-1β-treated mice as compared to TAC controls.Thus, the mechanisms, and functional relevance, of reduced cardiac lymphatic CCL21 gradients in late anti-IL-1β-treated mice remain to be investigated.Moreover, given these discrepancies noted between cardiac gene and protein expression levels in late anti-IL-1β-treated mice, it is possible that inhibition of IL-1β may have influenced cardiac proteases also related to CCL21 maturation 30 .

Early, but not late, anti-IL-1β treatment modulates cardiac fibrosis post-TAC
Cardiac inflammation and edema, if unresolved by lymphatic drainage from the cardiac interstitium, may aggravate cardiac fibrosis. 23Our recent study highlighted that while interstitial fibrosis occurred at 8 weeks post-TAC in BALB/c mice, they were protected against perivascular fibrosis, as compared with C57Bl/6J mice. 11We proposed that this may be due to expansion of perivascular lymphatics noted only in the former.In support, we found negative correlation (p=0.04) between the extent of coronary perivascular fibrosis and perivascular lymphatic densities (Fig. 4a).This suggests that IL-1β-independent perivascular lymphangiogenesis during pressure-overload may prevent perivascular fibrosis locally, despite generalized lymphatic dysfunction in the heart contributing to unresolved inflammation and edema driving interstitial fibrosis post-TAC.We previously reported increased lymphatic density around coronaries in clinical biopsies from HF patients, especially in the setting of DCM. 11Histological analyses revealed that this was associated with lower levels of perivascular fibrosis in DCM as compared to ischemic HF patients (Fig. 4b, c), as evaluated in small arterioles (Fig. 4d, e).Interestingly, while our study is the first to show a negative correlation between lymphatics and fibrosis in the perivascular niche, previous studies have also reported low levels of perivascular fibrosis in DCM patients. 31 our experimental study, we found that while early anti-IL-1β treatment slightly reduced cardiac-infiltrating macrophages, it unexpectedly increased interstitial fibrosis as compared to control TAC mice (Fig 4f).This would suggest potential beneficial immune effects of IL-1β in the semi-acute phase (1-5 weeks) post-TAC, contributing to the delay in TAC-induced LV dilation.However, this transient prevention of LV dilation led to accelerated changes in LV wall remodeling between 6-8 weeks post-TAC.Thus, chronic phase hyper-activation of mechanical-stress responses in LV matrifibroblasts may have contributed to aggravate interstitial fibrosis at 8 weeks in early anti-IL-1β treated mice.In contrast, neither early nor late anti-IL-1β treatments altered TAC-induced development of replacement fibrosis (micro-scars) (Fig. 4g) or perivascular fibrosis (Fig. 4h), as evaluated in 16-60 µm in diameter coronary segments (Fig. 4i).

Discussion
In this study, we demonstrated that anti-IL-1β treatment with Gevokuzimab during pressure-overload did not significantly alter chronic cardiac inflammation, nor was it sufficient to durably prevent deleterious cardiac remodeling, fibrosis, or HF development post-TAC in BALB/c mice.In contrast, in line with our hypothesis, we found that IL-1β appears to play a key role in the cardiac lymphangiogenic response induced by LV dilation.This lymphatic expansion is triggered in the chronic stage following pressure-overload, between 4-8 weeks post-TAC in BALB/c mice.Our data revealed that late anti-IL-1β treatment reduced TAC-induced cardiac lymphangiogenesis, without altering cardiac macrophage levels or Vegfc gene expression in the heart.Instead, our data showed reduced cardiac mature VEGF-C protein levels in late-treated TAC mice.Our in vitro macrophage study supports a direct role of IL-1β in promoting both Vegf-c expression and VEGF-C maturation, linked to upregulation of Furin, Ccbe1, and Adamts3.In contrast, early anti-IL-1β treatment, discontinued week-5 post-TAC, did not reduce cardiac lymphangiogenesis.Indeed, surprisingly, it slightly increased lymphatic capillary, but not precollector, densities.
Potentially, this reflects a deleterious impact of infiltrating macrophages, reduced by early anti-IL-1β treatment, on cardiac lymphangiogenesis during chronic pressureoverload.Alternatively, as LV dilation was transiently delayed up to 6 weeks post-TAC, it is possible that the subsequent accelerated increase in cardiac wall stress upon termination of treatment may have led to increased Vegfc expression in early treated mice.However, at 8 weeks, we did not find any differences in cardiac lymphangiogenic growth factor gene or protein levels in early-treated mice versus TAC controls.Another potential mechanism is that cardiac IL-1β may exert deleterious effects on lymphatic function post-TAC before the onset of lymphangiogenesis.Early anti-IL-1β treatment could then modify lymphatic capillary reactivity to lymphangiogenic factors, despite similar lymphangiogenic stimuli in the decompensating phase post-TAC.Further studies are needed to investigate this possibility.
The pro-lymphangiogenic effects of IL-1β appear to be mainly indirect, similar as previously described in mice with tracheal Il1b-overexpression. 21 In this setting, blockage of VEGF-C/-D signaling, with soluble VEGFR-3, sufficed to inhibit the prolymphangiogenic effects of IL-1β, demonstrating its dependency on the VEGFR-3 axis.
The authors further demonstrated that IL-1β stimulated tissue-recruitment of macrophages, and increased their release of VEGF-C.In contrast, we found a negative correlation between infiltrating cardiac macrophages and lymphatic densities post-TAC.Thus, in our study, cardiac IL-1β did not stimulate lymphangiogenesis post-TAC by promoting cardiac recruitment of macrophages.Instead, in line with a previous report in C57Bl/6 mice 27 , we found weak, but positive, correlation between densities of tissue-resident cardiac macrophages and lymphatics.However, IL-1β does not appear as a major regulator of LYVE1 + tissue-resident cardiac macrophage expansion, as neither early nor late phase blockage of IL-1β post-TAC altered densities of this subpopulation.Instead, our in vitro data indicated that IL-1β may influence macrophage activity rather than numbers, by promoting VEGF-C maturation.We demonstrate that IL-1β stimulation of macrophages, but not LECs, led to upregulation of proteases, including Furin and Adamts3, which play key roles in VEGF-C maturation.Of note, recent cardiac single cell RNA sequencing (scRNAseq) analyses suggest that IL-1β may also regulate extracellular matrix remodeling by increasing cardiac fibroblast expression of endogenous protease inhibitors, such as TIMPs 6,7 .
Conversely, inhibition of IL-1β signaling in the clinic has been found to reduce tissue plasminogen activator plasma levels. 32Another potential mechanism of IL-1β is direct activation of LECs via IL-1R.This would enhance LEC NFκB signaling, promoting increased Flt4 expression thus sensitizing LECs to VEGF-C produced in autocrine or paracrine manners 33 .However, our unpublished molecular analyses revealed that only 1-2% of cardiac LECs express IL-1R1, and this was not significantly altered after TAC.
Our in vitro studies in murine LECs confirm low Il1r1 expression, which nevertheless was sufficient for IL-1β to upregulate Flt4 expression (1.5 fold-change) in LECs.
In addition to stimulating macrophage-dependent VEGF-C-mediated lymphangiogenesis post-TAC, our ex vivo data indicate that IL-1β may influence lymphatic immune cell recruitment by stimulating CCL21 release in LECs.Interestingly, while cardiac Ccl21 gene expression increased in late anti-IL-1β treated mice, we showed that lymphatic CCL21 chemokine gradients were reduced.This is expected to limit immune cell drainage from the heart, slowing inflammatory resolution.Indeed, CCL21 is the main chemokine released by LECs, and its receptor, CCR7, is expressed in both lymphoid and myeloid cells [34][35] .Our in silico analysis of recently published scRNAseq data on cardiac-infiltrating immune cells from C57Bl/6J mice post-TAC 36 or post-MI 37 indicate that in the heart, both B and T cells have elevated Ccr7 expression.We previously showed, by flow cytometry in BALB/c mice at 8 weeks post-TAC, that cardiac B cell levels are low, but there is retention post-TAC of macrophages and T cells, especially CD4 + T cells. 11Thus, reduced lymphatic CCL21 levels, together with reduced lymphatic density following inhibition of IL-1β, could potentially increase cardiac T cell retention post-TAC.Intriguingly, our preliminary data indicate a decrease, rather than an increase, in cardiac T cells in anti-IL-1β-treated TAC mice.This may reflect reduction of local inflammatory processes in the heart by anti-IL-1β treatment, leading to reduced T cell recruitment or expansion.We also expected that the treatment would limit cardiac pro-inflammatory macrophage infiltration post-TAC, and potentially increase tissue-resident macrophage proliferation, known to be cardioprotective post-TAC. 27However, we found no differences in cardiac-resident CD68 + /Lyve-1 + macrophage levels in either early or late-treated groups.Further, only early anti-IL-1β treatment was associated with a slight decrease in cardiac-infiltrating CD68 + /Lyve-1 -macrophage levels.Importantly, in late-treated mice, where IL-1βmediated-lymphangiogenesis was inhibited, the absence of significant effects on cardiac macrophage levels suggests generalized dysfunction of cardiac lymphatics post-TAC.Indeed, we previously showed that the extensive cardiac lymphatic network post-TAC in BALB/c mice is inefficient to resolve cardiac edema, inflammation, or interstitial fibrosis, and thus fails to prevent LV dilation and HF development. 11Our current study further indirectly demonstrates that cardiac lymphatic function post-TAC is not improved by late anti-IL-1β treatment.
We were surprised by the absence of functional cardiac effects of anti-IL-1β treatment at 8 weeks post-TAC, although early anti-IL-1β treatment transiently delayed LV dilation, with beneficial functional effects persisting up to 6 weeks post-TAC.These results contrast with previous studies demonstrating that Nlrp3 gene deletion, which reduced IL-1β production, or early anti-IL-1β antibody therapy, accelerated cardiac dysfunction post-TAC in C57Bl/6J mice, despite reducing cardiac hypertrophy, macrophage infiltration, and fibrosis. 9,10It appears that prevention of adaptive hypertrophy, mediated by low levels of IL-1β in C57Bl/6J mice, may promote subsequent development of LV dilation and dysfunction post-TAC, potentially due to increased wall stress.Additionally, the authors suggested insufficient angiogenesis matching cardiac hypertrophy as a driver of accelerated cardiac dysfunction in Nlrp3deficient or anti-IL-1β treated mice 9,10 .In contrast, in BALB/c mice, pressure-overload does not lead to extensive hypertrophy, thus lessening the role of cardiac angiogenesis and the functional impact of potential anti-hypertrophic effects of anti-IL-1β treatment.Indeed, our previous studies have revealed maintenance, but not expansion, of blood vascular networks post-TAC in BALB/c mice. 11Of note, Gevokizumab decreases the binding affinity of IL-1β to IL-1R, resulting in a 30-fold reduction of IL-1β signaling, but it does not completely block IL-1β signaling, different from Canakinumab and Anakinra. 38We expect that in our study early anti-IL-1β treatment may have reduced, but not completely blocked, cardiac IL-1β signaling, thus maintaining weak but beneficial compensatory hypertrophy 5 , while transiently blocking the detrimental direct cardiac effects caused by elevated levels of IL-1β.It is possible that early and prolonged anti-IL-1β treatment would have been required in our study to reach durable benefit on cardiac function.However, long-term treatment, including during LV dilation, would have reduced cardiac lymphangiogenesis, thus potentially confounding beneficial cardiac effects of IL-1β blockage.Of note, the risk for an immune response directed against a therapeutic antibody prevents long-term treatment with Gevokizumab in mice.
Other cytokines, including IL-6, may be involved in the pathophysiology of the DCMlike phenotype post-TAC in BALB/c mice.IL-6 has been shown to drive both cardiac fibrosis and pathological hypertrophy following pressure-overload in C57Bl/6 mice. 39 found that anti-IL-1β treatment did not reduce cardiac Il6 expression, which may have contributed to the lack of beneficial effects on cardiac fibrosis and HF development found in both early-and late-treated mice.Importantly, the lack of cardiac benefit of late anti-IL-1β treatment in our study suggests that in patients with chronic pressure-overload IL-1β-blocking therapy may be inefficient, and potentially deleterious, by reducing cardiac lymphatic density, if initiated during the process of LV dilation.Intriguingly, the beneficial perivascular lymphangiogenesis occurring post-TAC, which seems to protect against perivascular fibrosis, was not influenced by anti-IL-1β treatment.In agreement, recent scRNAseq analyses coupled to spatial transcriptomics revealed that perivascular fibroblasts, different from Postn + matrifibroblasts, have little interaction with cardiac-infiltrating IL-1β-producing CCR2 + myeloid cells 6 .Further studies are thus needed to identify the triggers of perivascular lymphangiogenesis in the heart and to unravel how these lymphatics act to potentially prevent perivascular fibrosis during HF development.Notably, a histological marker of edema would be required to investigate if these expanded lymphatics surrounding coronary arterioles suffice to locally restore perivascular pressure-gradients caused by excessive blood vascular leakage, which occurs during pressure-overload 11 .
Taken together, our study provides, for the first time, a molecular mechanism linking LV dilation to lymphatic remodeling following pressure-overload (Fig. 5).Moreover, our data suggest that IL-1β may influence not only lymphatic expansion, but also lymphatic expression profiles or secretomes.Intriguingly, the absence of deleterious cardiac effects following inhibition of lymphangiogenesis by late anti-IL-1β treatment is in support of our previous finding that the expanded lymphatic network in BALB/c mice post-TAC appears dysfunctional and/or immature 11 .In contrast, we and other have demonstrated that inhibition of lymphangiogenesis during pressure-overload in C57Bl/6J mice and in Wistar rats, by Flt4 deletion or anti-VEGFR-3 treatment, aggravates cardiac inflammation and dysfunction. 11,17,40,41Further investigations are needed to determine why IL-1β-mediated lymphangiogenesis leads to aberrant hypersprouting and dysfunctional cardiac lymphatics in BALB/c mice.Molecular analyses are underway to determine how expanded cardiac lymphatics post-TAC differ from healthy lymphatics in BALB/c mice, in order to identify new targets to treat lymphatic dysfunction in HF.
As for the clinical outlook, a recent case-report in a stage III HF patient, secondary to idiopathic inflammatory DCM, demonstrated rapid improvement of LV function following Anakinra treatment, associated with reduced IL-6 and BNP levels. 42We speculate that this cardiac benefit of IL-1β inhibition in an already severely dilated heart may also be related to reduction of negative inotropic effects of the cytokine.In our study, we cannot exclude that the observed inhibition of lymphangiogenesis may have contribute to the lack of functional cardiac benefit of anti-IL-1β in late-treated mice.
Thus, anti-IL-1β may have a dual therapeutic window where targeting of IL-1β in the early stage, before LV dilation, or alternatively at later stages of the disease, when LV dilation and cardiac lymphangiogenesis are well-established, may be required to obtain clinical benefit in HF patients.

Study approval
Animal experiments performed in this study were approved by the regional Normandy ethics review board in line with E.U and French legislation (APAFIS #23175-2019112214599474 v6, #32433-2022070715208369 v2).A total of 88 BALB/c female mice, surviving TAC or sham-surgery were included in this study.Anonymized human heart samples evaluated in this study were collected with informed consent by the Bichat hospital biobank (U1148 BB-0033-00029/ BBMRI, coordinator JB Michel) authorized for tissue collection by the Inserm institutional review board as previously reported 11 .
Experimental mouse model, macrophage cell culture, and human samples TAC surgery was performed in adult female BALB/c mice (Janvier Laboratories, France).Briefly, a minimally invasive method was used to constrict the aortic arch, using a 26G needle.Double-banding of the aorta was applied to prevent suture internalization and model variability, as described. 11Anti-IL-1β treatment consisted of repeated intraperitoneal injections of a blocking monoclonal antibody against IL-1β, (Gevokizumab/XOMA-052 provided by Servier France), as previously described 26 , using a dose of 200 µg/mouse (10 mg/kg -1 ) three times per week starting from week-1 until week-5 post-TAC (early treatment) or starting from week-4 until week-8 post-TAC (late treatment).
Cardiac samples from end-stage HF patients (recipients of cardiac transplants at the Bichat Hospital in Paris, France) were examined by histology (Sirius Red staining).
Discarded cardiac autopsy samples, obtained from age-matched donors without cardiovascular disease, were used as healthy controls, as previously reported 11 .

Cardiac functional, cellular, and molecular analyses
Cardiac function was evaluated by echocardiography as described. 11Cardiac sections were analyzed by histology and immunohistochemistry, as described, 11 to determine lymphatic vessel densities and sizes, immune cell infiltration (macrophages), cardiomyocyte hypertrophy, and fibrosis (Sirius Red staining).Perivascular fibrosis was defined as the area of perivascular fibrosis surrounding the media in similar-sized arterioles, as described. 31

Figure 5 Schematic view of cardiac impact of IL-1β during pressure-overload
In response to increased wall stress (a), cardiac cells, including cardiomyocytes, fibroblasts and CCR2 + recruited macrophages, release IL-1β.This stimulates both production and maturation of VEGF-C in resident cardiac macrophages.In parallel, IL-1β acts on LECs to increase VEGFR3 (Flt4) expression.Together, these changes lead to stimulation of cardiac lymphangiogenesis.In cardiac pathology (b), LV dilation triggers IL-1β-dependent macrophage-mediated stimulation of cardiac lymphangiogenesis.However, in the setting of cardiac inflammation, the expanded lymphatic network remains immature and/or dysfunctional, and fails to limit accumulation of immune cells, edema, and interstitial fibrosis.In contrast, perivascular lymphangiogenesis, triggered in an IL-1β-independent manner during LV dilation, limits perivascular fibrosis.Antibody-mediated blockage of IL-1β in early phase post-TAC in BALB/c mice, before the onset of lymphangiogenesis, transiently limits LV dilation.In contrast, late anti-IL-1β treatment, initiated during LV dilation, blocks cardiac lymphangiogenesis, but not perivascular lymphangiogenesis, and does not alter cardiac remodelling or dysfunction.Created with BioRender.com.

Suppl. Methods & Materials
Human samples Five micrometer thick sections were prepared from paraffin-embedded human septal samples.Standard immunohistological protocols, including citrate antigen retrieval, were used to reveal podoplanin-positive lymphatic vessels using a mouse anti-human podoplanin/D2-40 antibody validated for clinical practice (Dako, # M3619, diluted 1:50).Signal was revealed following incubation with HRP-conjugated donkey anti-mouse secondary, followed by DAB detection kit (Vector Laboratories, Burlingame, CA).All patient slides were processed together for uniformity.Lymphatics were examined using a Zeiss axiovision light microscope equipped with CCD cameras at x10.Lymphatic vessels, identified as Podoplanin + vascular structures in the subendocardium, were counted by an observer blinded to the patient category.Total lymphatic density, open lymphatic density (lumenized vessels), lymphatic vessel diameter, and density of perivascular lymphatics (defined as lymphatics within 200 µm distance of a large blood vessel) were determined.On average for each patient 1.2 ±0.2 mm² of the septal sample was analyzed to determine lymphatic vessel sizes and densities.

Experimental Model
Female BALB/c mice (22-24 g) were obtained from Janvier.Animal housing and experiments were in accordance with National Institutes of Health guidelines, and the study was ethically approved by the Normandy University regional review board according to French and EU legislation (APAFIS #23175-2019112214599474 v6; APAFIS #32433-2022070712508369 v2).Minimally-invasive transversal aortic banding constriction (TAC) was performed on 8 week-old mice, as previously described. 1Mice were anaesthetized by intraperitoneal injection of ketamine (100 mg kg -1 Imalgene ® ) and xylazine (10 mg kg -1 Rompun ® 2%, Bayer Health Care) and placed on mechanical ventilation.The operator performed a minimal thoracotomy with an incision at the level of the first intercostal space.The aortic arch was visualized under low-power magnification.A snare, made of 7-0 polypropylene suture, was passed under the aorta between the origin of the right innominate and left common carotid arteries.Two suture bands were placed side-by-side to create an elongated stenosis and prevent internalization of the suture, as described. 2A bent 26-gauge needle was placed next to the aortic arch, and the sutures were snugly tied around the needle and the aorta.After banding, the needle was quickly removed.The skin was closed, and mice recovered on a warming pad until fully awake.The sham procedure was identical except that the aorta was not banded.Buprenorphine (50 µg/kg, Buprecare ® , Axcience) was injected subcutaneously 6 hours after surgery and twice per day until 3 days post-operation.Anti-IL-1β treatment consisted of a blocking monoclonal antibody against IL-1β, (Gevokizumab/XOMA-052 provided by Servier France), as previously described. 3The antibody was administered by repeated intraperitoneal injections of 200 µg/mouse (10 mg/kg -1 ) three times per week starting from week-1 until week-5 post-TAC (early treatment) or starting from week-4 until week-8 post-TAC (late treatment).
Lyve1 expression.These macrophages, also differing in size and morphology from the elongated cell body and nucleus typical of lymphatic endothelial cells, were readily excluded from lymphatic counts.Photos were captured at x20.
Due to the absence of conclusive markers to distinguish precollectors from capillaries in murine cardiac lymphatics, we have made use of the fact that precollectors preferentially run in a baseto-apex fashion, whereas lymphatic capillaries lack such consistent spatial organization.Hence, by evaluating vessels with a lumen parallel to horizontal cardiac sections, the precollector population of vessels is more frequently detected as vessels running perpendicular to the section (= open lumen vessels).To assess the size and frequency of open lymphatic vessels (diameter > 5 µm; a population enriched in precollector vessels), between 10-20 images from the LV free-wall were collected and analyzed for each mouse.The density (open vessels/mm²) and lymphatic lumen sizes were measured and used to calculate mean vessel diameter.The parameter % open lymphatic area was calculated as the sum total of all measured lymphatic diameters divided by the total cardiac area analyzed for each mouse heart.On average for each mouse 2.2±0.3 mm² of the LV was analyzed to determine lymphatic vessel sizes and % open lymphatic area.
Cardiomyocyte sizes were evaluated in sections double stained for CD31 and WGA imaged at x40.Cardiac macrophages were defined as CD68 positive cells.Infiltrating macrophages were defined as LYVE1 -cells, and cardiac-resident macrophages as LYVE1 + cells.Images were captured at x20.

Histology in mouse or human cardiac sections
Cardiac cryosections (8 μm) of mouse samples, or paraffin sections (5 μm) of human samples, were processed for Sirius Red staining, as described. 1Cardiac interstitial collagen density was evaluated in mouse cardiac sections imaged on a light microscope (Zeiss) equipped with an x40 objective.Replacement fibrosis (micro-scars) was evaluated in these images as total number/cardiac section of fibrotic hotspots, identified as images containing >11% fibrotic area per image, analyzing on average 14±0.5 images/sample (average total examined area 0.5 mm² of each LV cross-section).Perivascular fibrosis was evaluated in images taken at x20, focused on arterioles with a lumen diameter of 16-60 μm in both mice and men.On average 5 vessels/cardiac section were used to determine perivascular fibrotic area (μm²), as previously described. 4Images were analyzed by an operator blinded to treatment groups using Fiji imaging software (NIH).

Real-Time Polymerase Chain Reaction
Murine cardiac samples were collected at 8-weeks post-TAC.RNA extraction was performed with Trizol method.Briefly, samples were homogenized (45 sec.) in TRIZOL using Precellys tubes with beads.RNA was obtained after incubation, successively, in chloroform and isopropanol and three washes in ethanol.RNA quality was analyzed by Nanodrop 2000 (Thermo Fisher Scientific).cDNAs were generated using a reverse transcriptase after DNAse treatment.Real-time PCR was performed on a LightCycler 480 (Roche Molecular Biochemicals) with a commercially available mix (FastStart DNA Master SYBR Green I kit; Roche) on white 96-well plates.The reactions were performed in duplicate for each sample.Concentrations were calculated using a standard curve, obtained by serial dilutions of samples from an organ with high expression of the gene of interest.To normalize gene expression, values are expressed as ratios of a reference gene (E2F).Healthy sham animal ratios are then set as 100%, and values in TAC-operated animals are expressed relative to their respective sham controls.The sequences of the specific primers are detailed in Suppl.Table 5.
Suppl.Western blot analysis in cardiac samples Cardiac protein was extracted using Cell lysis buffer (#895347, RnD Systems) in microbead tissue dissociation tubes (Precellys #P000933-LYSK0-A -0.5 mL).Twenty-five micrograms of denatured cardiac protein from healthy sham-operated or TAC-operated mice at 8 weeks post-TAC treated or not with anti-IL-1β antibodies, were separated on stain-free 4-20% agarose gels (Criterion TGX Precast Gel, Biorad), followed by blotting with a rabbit anti-VEGFC antibody (#ab9546, Abcam; 1:1000), or a rabbit anti-VEGFD antibody (#ab155288, RnD systems; 1:500), and finally a mouse anti-actin antibody (#15597191, Invitrogen; 1:10000) diluted in PBS/BSA 5%.After repeated washing, HRP-conjugated secondary antibodies (goat anti-rabbit #P0448, Dako; 1:5000, donkey anti-mouse #A16011, Invitrogen; 1:5000) were used.Target proteins were visualized using ECL chemiluminescence kit (#1705061, BioRad).Ratios of specific bands to actin signal per well was analyzed in ImageJ.For VEGF-C and VEGF-D multiple bands were present, representing the various cleaved forms of the growth factors.For VEGF-C, the ratio of mature to immature forms was calculated as the ratio of the densities of cleaved, mature monomeric VEGF-C (21 kDa bands) to the densities of full-length immature monomeric pro-VEGF-C (31 kDa bands) 5 .For VEGF-D only the full-length immature pro-VEGF-D form (42 kDa dimer) was visible (cleaved, mature form at 32 kDa too weak for analysis).VEGF-C and VEGF-D protein ratios were compared between sham and TAC groups using two-way ANOVA followed by Sidak's posthoc test.
Cardiac gene expression was analyzed by RT-qPCR, and Protein levels of VEGF-C and VEGF-D were assessed by Western Blot.Cardiac and plasma levels of IL-1β were evaluated by ELISA (RnD Systems: ELISA DuoSet anti-IL-1β /IL-1F2 #DY401-05).For details see Suppl.Methods.StatisticsData are presented as mean ± s.e.m.Comparisons were selected to determine: 1) impact of pathology (healthy sham vs. TAC); 2) effect of treatment (anti-IL-1β-treated vs. TAC controls): 3) effect of IL-1β treatment vs. control untreated macrophages.Statistical analyses for comparisons of two independent groups were performed using either Student's two-tailed t-test for groups with normal distribution, or alternatively by Mann Whitney U test for samples where normality could not be ascertained based on D'Agostino & Pearson omnibus normality test.For comparisons of three groups or more either one-way ANOVA followed by Sidak's multiple comparison posthoc (for parameters with n>7 with normal distribution), or alternatively Kruskal-Wallis nonparametric analysis followed by Dunn's posthoc for multiple comparison (for parameters with non-Gaussian distribution) were performed.Longitudinal echocardiography studies were analyzed by paired two-way ANOVA followed by Dunnett's posthoc, while morphometric data and gene expression data were analyzed by two-way ANOVA followed by Sidak's posthoc for multiple pair-wise comparisons or Dunnett's posthoc to compare 3 or more groups.Non-parametric Spearman rank order tests were used for evaluating correlations.Outlier samples were identified as an individual value exceeding group mean±4SD in groups with n≥7.All analyses were performed using GraphPad Prism software.carried out by C.H and T.L; M.B and A.Z designed and carried out in vitro macrophage assays and expression analyses; M.V carried out surgical mouse model; C.H carriedout cardiac gene expression analyses; O.L and C.V carried out in vitro LEC assays and expression analyses; C.H and C.V performed protein analysis by ELISA and Western Blot; JB.M managed the biobank at Bichat Hospital and contributed clinical data for this study; P.M, A.Z, and V.T participated in study design; A.Z contributed critical comments on the manuscript draft; C.H. and E.B designed the study, analyzed results, and prepared the manuscript draft.All authors approved of the final version of the manuscript.

Figure 4
Figure 4 Cardiac perivascular fibrosis post-TAC in mice and HF patients linked to perivascular lymphatics, independently of IL-1β Correlation analysis of perivascular fibrosis (µm²) and perivascular lymphatic density (vessels/mm²) around coronary arteries in mice at 8 weeks post-TAC (a).Quantification of perivascular fibrosis (µm²) in healthy controls (n=9, white bar), ischemic HF patients (n=8, grey bar), and DCM patients (n=9, green bar) (B).Arteriolar sizes for evaluation of perivascular fibrosis in patients (c).Quantification of perivascular lymphatic densities in patients (d, adapted from Heron et al. 11 ).Examples (e) of cardiac perivascular lymphatics, stained for PDPN, and perivascular fibrosis in Sirius Red-stained clinical HF samples; scalebar 20 µm.Evaluation of interstitial fibrosis (f) at 8 weeks post-TAC in sham (n=7; grey bars w. white circles), control TAC (n=19; yellow bars w. black circles), and early (n=6; red bars w. black circles) vs. late (n=10; red bars w. black squares) anti-IL-1β-treated TAC mice.Quantification of microscars (g) and perivascular fibrosis (h) at 8 weeks post-TAC.Arteriolar sizes for evaluation of perivascular fibrosis in mice (i).Mean ± s.e.m are shown (f, g, h), or violin or box plot with median indicated (b, c, d, i).Groups compared by one-way ANOVA followed by Sidak's multiple comparison test (c, i) or by non-parametric Kruskal Wallis followed by Dunn's posthoc test (b, d, f, g, h).* p<0.05; ** p<0.01 versus sham; # p<0.05 versus TAC control or DCM patients.Non-parametric Spearman rank order test was used for correlation (a).

Table 5
-list of primers used in mouse cardiac samples for qPCR