Mannan binding lectin-associated serine protease-2 (MASP-2) critically contributes to post-ischemic brain injury independent of MASP-1

Inflammatory mechanisms intrinsic to brain and blood-borne inflammatory mediators are among the major drivers of focal injury following cerebral ischemia. Among the inflammatory cascades, the complement system represents a powerful contributor to ischemic brain injury by several possible mechanisms including anaphylatoxin release, endothelial activation aiding leukocyte adhesion and recruitment, over-activation of the phagocytic system, and direct cellular lysis [1]. Activation of the complement can be achieved by three distinct pathways: the classical pathway (CP), the alternative pathway (AP), and the lectin pathway (LP). All three pathways encompass tightly regulated sequential activation cascades converging at the cleavage of the abundant complement component C3. Cleavage of C3 releases the small complement anaphylatoxin C3a and the large C3b fragment, which possess pro-inflammatory properties and promote opsonization and direct phagocytosis, respectively. In addition, C3b is also a constituent of the AP C3 convertase (C3bBb), and the deposition of C3b promotes the formation of more AP C3 convertases by binding to native factor B (fB) to form the AP zymogen complex C3bB [2]. C3b is also an essential component of both C5-cleaving convertase complexes (C3bBb(C3b)_n and C4b2a(C3b)_n, respectively). With the cleavage of C5, the last enzymatic step of complement activation is completed. C5 cleavage releases the potent complement anaphylatoxin C5a and the larger fragment C5b that initiates the formation of the membrane attack complex (MAC) by subsequent recruitment of the terminal cascade components C6-C9. This MAC inserts into cell membranes to form a pore that results in ion flux, causing cell lysis.

The critical contribution of the complement system to the pathophysiology of ischemia-reperfusion injury (IRI) has been demonstrated in several models of ischemia-reperfusion injury [3‐7]. It was initially hypothesized that the complement activation pathway responsible for IRI is the CP driven by natural antibodies binding to ischemia-damaged cells [8]. The involvement of the LP in IRI was first described by Collard et al. in 2000 [9], a finding that was later underlined by our own work reporting the observation that CP-deficient C1q^−/− mice were not protected in models of cerebral IRI [6]. The prominent role of the LP over the CP in mediating IRI was underlined by the protective phenotype of mannan-binding lectin (MBL) deficiency and the therapeutic effect of inhibitory molecules against MBL in various mouse models [7, 10‐14]. Deficiency or inhibition of MBL achieved long-lasting neuroprotection and improved functional outcome in mouse models of stroke with a wide window for therapeutic intervention (up to 24 h) [13]. The latter findings are supported by numerous recent publications, which underline the prominent role of the LP in the pathogenesis and progression of brain damage in a clinical context. In stroke patients, the LP was shown to be the relevant pathway in the progression of ischemic brain damage [12, 15‐17] with studies highlighting MBL [18, 19] and ficolin-3 [15], two different LP recognition molecules, as independent predictors of ischemic stroke outcome.

The initiation of the complement activation via the LP requires the binding of one or more of the five different human recognition subcomponents (i.e., MBL, collectin-11 (CL-11), ficolin-1, ficolin-2, or ficolin-3) or the five different mouse recognition components (i.e., MBL-A, MBL-C, CL-11, ficolin-A, or ficolin-B) to their cognate ligands on activating surfaces [20‐22]. The ligands that mediate the binding of LP recognition components on ischemic cells are presently unknown. LP activation complexes are formed when a multimolecular complex composed of oligomers of homotrimeric recognition subunits associated with LP-specific serine proteases, called mannan-binding lectin-associated serine proteases (MASPs), bind to cognate ligands on activator surfaces. Three MASP enzymes named MASP-1, MASP-2, and MASP-3 have been described and are encoded by two different genes. The MASP1 gene is located on human chromosome 3 (mouse chromosome 16) and encodes MASP-1 and MASP-3. The MASP2 gene is located on human chromosome 1 (mouse chromosome 4) and encodes the serine protease MASP-2 [23, 24].

Of the three different MASPs, only MASP-2 is able to cleave both C2 and C4 to form the LP C3 convertase C4bC2a and the C5 convertase C4bC2a(C3b)_n. Indeed, in the absence of MASP-2, but not of MASP-1/MASP-3, a complete inhibition of LP activation was observed [20, 25]. In addition, targeting MASP-2 by gene disruption or administration of antibodies that inhibit MASP-2 functional activity reduced IRI in models of myocardial, intestinal, or renal IRI [20, 26].

With respect to MASP-1, previous work suggested that, due to its ability to cleave C2 but not C4, it cannot drive LP activation in the absence of MASP-2 but may facilitate MASP-2-driven LP activation [20, 27‐29]. A recent study, however, proposed that MASP-1 has an essential role in driving MASP-2 and LP activation by being an exclusive activator of MASP-2 [30], analogous to the CP serine proteases where C1r is the exclusive activator of C1s [31]. Further studies attributed an additional role to MASP-1 and/or MASP-3 in driving AP activation by converting factor D (fD) and/or fB zymogen into their enzymatically active forms [32, 33].

As for the subsequent activation steps, the activation of complement C3, but not of complement C5, was shown to be instrumental in the development of cerebral IRI, as demonstrated in a mouse model of focal ischemia [34]. Thus, it appears that C5a is not required to mediate the hallmarks of post-ischemic inflammation such as endothelial cell activation, facilitation of leukocyte adhesion and recruitment, and activation of phagocytic cells.

This study reveals the involvement of MASPs in cerebral IRI by assessing the impact of MASP-2 and of combined MASP-1 and MASP-3 deficiency in gene-targeted mouse strains and in WT mice treated with a MASP-2-specific inhibitor on functional and histopathological damage following cerebral focal ischemia. We directly compared our findings against the phenotypes seen in fB-deficient and C4-deficient mice (fB^−/− and C4^−/−, respectively) analyzed in parallel.

This study provides a comparative analysis of the phenotypes of mouse lines with targeted deficiencies of the LP-specific serine proteases MASP-1, MASP-2, and MASP-3 in a mouse model of stroke. Our analysis includes functional outcomes based on behavior tests (focal deficit and general neurological deficits) and histopathological outcomes (infarct sizes) as well as immunohistochemistry for deposition of the C3 activation products as a parameter for complement activation events within the ischemic areas of the brain and for morphometric endpoints (morphology of CD11b-positive macrophages/microglial cells). An AP-deficient mouse line with a targeted deficiency of the essential AP zymogen fB and a CP-deficient mouse line with a targeted deficiency in C4 were also evaluated. When compared to the WT mice, a strong protective phenotype was observed in the MASP-2-deficient mice at the level of functional and histopathological outcomes. Absence of MASP-2 lead to reduced C3 deposition in the ischemic brain area and less prominent hypertrophic and amoeboid microglial morphology. Similarly, improved outcomes were observed in the WT mice treated with a MASP-2-inhibitory antibody, corroborating the findings observed in the MASP-2^−/− mice.

Overall, the CD11b morphological analysis revealed that, in the absence of MASP-2 functional activity, microglial cells within the ischemic areas primarily present as ramified microglia suggestive of an anti-inflammatory polarization state [41‐43] as opposed to the hypertrophic microglial morphology associated with phagocytic activity and pro-inflammatory state in the ischemic areas of WT brains. The complement system has a major role in the activation of microglia, which constitutively express receptors for C1q and for C3 cleavage products. As a consequence of complement component binding, microglia activate phagocytosis and cytokine production. Activated microglia in turn contribute to complement component production that feeds autocrine/paracrine signaling [41]. Since microglia act as a major contributor to post-injury inflammatory responses, it is plausible that complement activation through MASP-2-dependent processes could drive microglia from the ramified, surveying state towards the pro-inflammatory phagocytic phenotype, with visualized morphological changes of amoeboid cell shape, an enlarged cell soma, and retracted processes [40, 41, 44].

Since in vitro experiments suggested that MASP-1 fulfills a critical role in the activation of MASP-2 [45], we expected that the mice deficient in MASP-1 would also show protection from cerebral IRI following tMCAO. Surprisingly, the MASP-1/3^−/− mice presented with larger infarct volumes then their WT controls but showed a very similar degree of severity in neurological deficits. Likewise, the degree of C3 deposition within the ischemic areas in the brains of the MASP-1/3^−/− was similar to that seen in the WT mice. The morphometric analysis of microglia revealed no differences between the WT and MASP-1/3^−/− mice. The absence of any degree of protection of the MASP-1 and MASP-3 double-deficient mouse line indicates that neither MASP-1 nor MASP-3 are involved in the pathophysiological processes leading to cerebral IRI.

Several studies have indicated that MASP-1 acts as a rate-limiting protease in the activation of MASP-2 in serum. SFMI-1, a peptide that preferentially inhibits MASP-1, was reported to inhibit the LP to an extent similar to a MASP-2-specific peptide inhibitor, SFMI-2 [45]. Subsequent studies using second-generation peptide inhibitors of either MASP-1 or MASP-2 with greater specificity corroborated the initial results. These findings led to the hypothesis that, in the LP, MASP-1 is the initiating protease critically required to cleave MASP-2 zymogen into its enzymatically active form, analogous to the CP where C1r is the exclusive activator of C1s [30]. Separate in vitro studies using serum from a Malpuech–Michels–Mingarelli–Carnevale (3MC) patient lacking both MASP-1 and MASP-3 supported this hypothesis [46]. We previously reported that serum of mice deficient in both MASP-1 and MASP-3 maintains a reduced but clearly detectable LP functional activity [20, 27].

The in vivo data presented here demonstrate that targeting MASP-2 reduces cerebral IRI while the absence of MASP-1 and MASP-3 affords no protection. MASP-2 does not require MASP-1 to drive IRI. Thus, targeting MASP-1 is unlikely to disrupt the LP-mediated pathophysiological processes that lead to unfavorable outcomes.

It is undisputed that MASP-1 facilitates the conversion of zymogen MASP-2 into its enzymatically active form, most likely because the relatively low abundance of MASP-2 compared to MASP-1 is a limiting factor during the LP-specific trans-activation events involving the juxtaposition of activating complexes [47]. The non-protective phenotype of a combined MASP-1 and MASP-3 deficiency in a MASP-2-dependent pathophysiological process underlines that MASP-1 is not an essential activator of MASP-2 functional activity and highlights the fundamental differences between LP and CP activation events.

We demonstrate that targeting MASP-2 with a specific inhibitory antibody is very effective in limiting both the post-stroke neurological deficits as well as the associated ischemic lesion and in reducing the consequent activation of microglia towards the pro-inflammatory amoeboid phenotype.

This study complements our previous work in which we targeted the LP recognition subcomponents MBL-A and MBL-C by either MBL-specific mAbs or through the injection of an excess of a fluid-phase carbohydrate ligand orthologue [13]. There are advantages to targeting the single, low-abundance LP enzyme that critically drives the pathophysiology of cerebral IRI over targeting one or more of the five different LP recognition components. Mice, for example, do not express an orthologue for the human recognition subcomponent ficolin-3 (alias H-ficolin) [48]. Our own recent work has demonstrated that ficolin-3 drives LP activation in patients with subarachnoid hemorrhage [17]. In stroke patients, ficolin-1 (alias M-ficolin) serum levels are dramatically decreased at time point 6 h after the onset of symptoms, suggesting massive ficolin-1 consumption following cerebral ischemia. Ficolin-2 (alias L-ficolin) and ficolin-3 levels also decrease during the acute phase, while serum MBL levels remain unaffected [49]. These findings are in full agreement with a previously published clinical study reporting the consumption of ficolin-3 and ficolin-2 during the acute phase of stroke [15]. These reports strongly suggest that in man, ficolins are critically involved in triggering LP activation during the reperfusion phase and that blocking MBL in stroke patients might be therapeutically less effective than what we and others observed when targeting MBL in mouse models of cerebral ischemia. In addition to its low abundance, MASP-2 is exclusively synthesized in the liver [50], and the effectiveness of systemic MASP-2 inhibitory agents in target organs is not complicated by local biosynthesis.

The assessment of fB^−/− mice in our model of tMCAO suggests that the AP contributes to IRI following cerebral ischemia as indicated by the significant amelioration of general neurological deficits. An even stronger protection has previously been reported by Elvington et al. [51] in which a reduction in both neurological deficits and in ischemic volume were observed in fB^−/− mice. In our hands, the lack of AP activation does not affect the lesion size as assessed by cresyl violet staining following correction for oedema. The factor B data suggest a role for the AP in cerebral IRI by amplifying complement activation initiated by the LP. The predominant role of the MASP-2-dependent LP over the AP in the mediation of cerebral IRI pathology is underlined by the lack of a protective IRI phenotype in the MASP-1/-3 double-deficient mouse line, which is associated with low to undetectable AP functional activity [32].

The absence of a protective effect of C4 deficiency in cerebral IRI confirms the non-protective phenotype of C4 deficiency previously reported in models of myocardial and renal IRI and emphasizes the general importance of a MASP-2-dependent C4-bypass activation route [20, 22, 26].

This work demonstrates the pivotal role of the LP effector enzyme MASP-2 in driving tissue injury and unfavorable outcomes in mouse models of ischemic brain ischemia and the utility of inhibitory therapeutic antibodies against this enzyme in reducing IRI-mediated tissue and organ function loss.