Basement membrane and stroke

Abstract

Located at the interface of the circulation system and the CNS, the basement membrane (BM) is well positioned to regulate blood–brain barrier (BBB) integrity. Given the important roles of BBB in the development and progression of various neurological disorders, the BM has been hypothesized to contribute to the pathogenesis of these diseases. After stroke, a cerebrovascular disease caused by rupture (hemorrhagic) or occlusion (ischemic) of cerebral blood vessels, the BM undergoes constant remodeling to modulate disease progression. Although an association between BM dissolution and stroke is observed, how each individual BM component changes after stroke and how these components contribute to stroke pathogenesis are mostly unclear. In this review, I first briefly introduce the composition of the BM in the brain. Next, the functions of the BM and its major components in BBB maintenance under homeostatic conditions are summarized. Furthermore, the roles of the BM and its major components in the pathogenesis of hemorrhagic and ischemic stroke are discussed. Last, unsolved questions and potential future directions are described. This review aims to provide a comprehensive reference for future studies, stimulate the formation of new ideas, and promote the generation of new genetic tools in the field of BM/stroke research.

Introduction

The basement membrane (BM) is an amorphous structure located at the abluminal side of endothelial cells or basal side of epithelial cells.^1–5 It is composed of multiple highly organized extracellular matrix (ECM) proteins and has a thickness of 50–100 nm.^6–11 The four major types of ECM proteins found in the BM are collagen IV, laminin, nidogen, and heparan sulfate proteoglycans (HSPGs).^1,2,12,13 In addition, some other components, including fibulins, osteonectin and netrin-4, are also found in the BM.^3,14–20

In the brain, the BM is predominantly associated with blood vessels. Unlike peripheral organs, the brain has two different BMs: an endothelial BM made mainly by BMECs and a parenchymal BM produced primarily by astrocytes (Figure 1(a) and (b)).^21–23 Sandwiched between endothelial cells and astrocytic endfeet, pericytes also contribute to BM formation by synthesizing and depositing ECM proteins to endothelial and/or parenchymal BMs. These two BMs are separated by pericytes and appear as one in regions without pericytes in physiological conditions (Figure 1(a) and (b)).^21,22 During neuroinflammation, however, these two BMs can be separated at postcapillary venules by infiltrating leukocytes,^1,24 forming perivascular cuffs. Located at the interface of the circulation system and the CNS, the BM is well positioned to regulate blood–brain barrier (BBB) integrity under both homeostatic and pathological conditions.

Figure 1.

Anatomical location of the BMs at the BBB. (a) Diagram illustration of the endothelial and parenchymal BMs at the BBB. (b) An electron microscopy image showing endothelial and parenchymal BMs at the BBB. BM: basement membrane; BBB: blood–brain barrier. (c) The cellular source of different laminin isoforms and their distribution in the endothelial and parenchymal BMs. (d) Changes of the four major BM components after hemorrhagic and ischemic stroke.

Here in this review, I first briefly introduce the composition of brain BMs. Next, the functions of the BM and its major components in BBB maintenance and stroke pathogenesis are discussed and summarized. Furthermore, unsolved questions, current challenges, and potential future directions are described. This review aims to provide a comprehensive reference for future studies, stimulate the formation of new ideas, and promote the generation of new genetic tools in the field of BM/stroke research.

BM composition

Brain BMs mainly consist of collagen IV, laminin, nidogen, and HSPGs.^1,2,12,13 The structure of each component and their expression in cerebral blood vessels are discussed below.

Collagen IV

Collagen IV is the most abundant component of the BMs. It is a trimeric protein containing three α chains. Currently, 6 collagen IV α chains (Col4a1-6) have been identified.^25–30 Each α chain contains an N-terminal 7S domain, a triple-helical domain, and a C-terminal globular non-collagenous-1 (NC1) domain (Figure 2(a)). During collagen IV network assembly, three α chains bind to each other forming a trimer, called protomer. These protomers then dimerize via their NC1 domains and tetramerize through their 7S domains, forming sheet-like suprastructures (Figure 2(a)).^2,31,32 In the cerebrovasculature, collagen IV is predominantly produced by endothelial cells,^33–35 astrocytes,^33,36 and pericytes.³⁷

Figure 2.

Structures of major BM components. (a) Structures of collagen IV monomer (α chain), trimer (protomer), and network. (b) Structure of a fully assembled laminin molecule. (c) Structure of nidogen-1/2. (d) Structure of HSPG2. BM: basement membrane; HSPG2: heparan sulfate proteoglycan-2.

Laminin

Laminin is a large family of cross-shaped heterotrimeric proteins composed of α, β, and γ subunits.^22,38–41 Each laminin subunit contains various globular (LN, L4, and LF) domains, rod-like (LE) repeats, and a coiled-coil domain (Figure 2(b)).^41,42 Unique to all α subunits are five globular (LG) domains at the C-terminus (Figure 2(b)).^41,42 These subunits bind to each other through their C-terminal coiled-coil domains, forming a large laminin molecule with one long arm and two or three short arms (Figure 2(b)).⁴¹ So far, there are five α, four β, and three γ variants.⁴¹ Various combinations of these subunits generate a large number of different laminin isoforms.^43,44 It is worth noting that not all possible combinations have been experimentally confirmed. In mammals, 20 laminin isoforms have been identified or proposed based on in vivo and in vitro studies (Table 1).⁴⁴ Like collagen IV, laminins are able to form sheet-like suprastructures via intermolecular self-assembly at their N-terminal domains.⁴⁵

Table 1.

Laminin isoforms.

Original name	Alternative name	Abbreviation
Laminin-1	Laminin-α1β1γ1	Laminin-111
Laminin-2	Laminin-α2β1γ1	Laminin-211
Laminin-3	Laminin-α1β2γ1	Laminin-121
Laminin-4	Laminin-α2β2γ1	Laminin-221
Laminin-5(A)	Laminin-α3(A)β3γ2	Laminin-3(A)32
Laminin-5B	Laminin-α3Bβ3γ2	Laminin-3B32
Laminin-6	Laminin-α3β1γ1	Laminin-311
Laminin-7	Laminin-α3β2γ1	Laminin-321
Laminin-8	Laminin-α4β1γ1	Laminin-411
Laminin-9	Laminin-α4β2γ1	Laminin-421
Laminin-10	Laminin-α5β1γ1	Laminin-511
Laminin-11	Laminin-α5β2γ1	Laminin-521
Laminin-12	Laminin-α2β1γ3	Laminin-213
Laminin-13	Laminin-α3β2γ3	Laminin-323
Laminin-14	Laminin-α4β2γ3	Laminin-423
Laminin-15	Laminin-α5β2γ3	Laminin-523
–	Laminin-α2β1γ2	Laminin-212
–	Laminin-α2β2γ2	Laminin-222
–	Laminin-α3β3γ3	Laminin-333
–	Laminin-α5β2γ2	Laminin-522

Two nomenclature systems exist for laminins. The original system names laminins using numbers based on the order of their discovery, such as laminin-1, laminin-2, and so on.⁴⁶ The alternative system names laminins using both Greek letters and numbers, which indicate the subunits and genetic variants, respectively.⁴³ For example, laminin-1 is written as laminin-α1β1γ1, which can be abbreviated as laminin-111. Compared to the original system, the alternative system provides information regarding the trimeric composition of laminins, and thus is more widely used nowadays. A comparison of these two nomenclature systems is illustrated in Table 1.

In the cerebrovasculature, laminins are made by endothelial cells, astrocytes, and pericytes.^22,47,48 Interestingly, these cells synthesize different laminin isoforms. Specifically, endothelial cells mainly make laminin-411 and -511,^21,49,50 whereas astrocytes predominantly produce laminin-211.^21,51 Our unpublished data show that pericytes mainly synthesize γ1- and α4/α5-containing laminins (most likely laminin-421 and -521). This cell-specific expression pattern allows differential distribution of distinct laminin isoforms in the endothelial and parenchymal BMs: the former is rich in laminin-411 and -511, while the latter mainly contains laminin-211 (Figure 1(c)).²²

Nidogen

Nidogen, also known as entactin, is a glycoprotein containing three globular (G1-3) and multiple rod-like (Link, EFG, and Thyroglobulin) domains (Figure 2(c)). In mammals, two genetic variants of nidogen (nidogen-1/2) have been identified. Unlike collagen IV and laminin, nidogens are unable to self-assemble or form sheet-like supramolecules. They can, however, interact with collagen IV and laminin via different domains, suggesting that they may function as a cross-linker to connect/stabilize collagen IV and laminin networks. Nidogens are mainly generated by endothelial cells,³⁵ astrocytes,⁵² and pericytes^35,53 in the cerebrovascular system.

HSPG

One important HSPG is heparan sulfate proteoglycan-2 (HSPG2), also called perlecan. It is a large multi-function protein composed of five domains (domain I-V) (Figure 2(d)).^54–57 Domain I has three glycosaminoglycans and a sea urchin sperm protein-enterokinase-argin (SEA) motif; Domain II contains four low-density lipoprotein (LDL) receptor class A repeats and an immunoglobulin (Ig) domain; Domain III has multiple laminin B and laminin EGF domains; Domain IV consists of a series of Ig domains; and Domain V, also known as endorepellin, contains many laminin G and EGF domains. Like nidogen, HSPG2 cannot form supramolecular sheet-like structures.⁵⁴ Instead, it is able to interact with many molecules,^54,57–65 including other BM components (ECM proteins) and heparin-binding growth factors. Similarly, HSPG2 expression is detected in endothelial cells,^66–68 astrocytes,^69,70 and pericytes³⁵ in cerebral blood vessels.

BM and BBB

In physiological conditions, the BMs play a variety of important roles, ranging from providing structural support to modulating molecular signaling.^22,71 In addition, accumulating evidence suggests that the BMs also contribute to vascular integrity. For example, leukocytes are more frequently observed between endothelial cells and the underlying BMs than in the process of passing between endothelial cells, in electron microscopy studies.^72–74 In addition, it has been reported that: (1) leukocytes take ∼5 min to cross the endothelial layer, but 20–30 min to penetrate the underlying BMs^75–78; and (2) T lymphocytes spend 9-10 min crossing the endothelium, but are trapped in the outer surface of blood vessels for ∼30 min.⁷⁹ These results strongly indicate that crossing the BMs is the rate-limiting step in leukocyte infiltration, highlighting a critical role of the BMs in the maintenance of vascular integrity.

In the brain, the BMs are located at the interface of the circulation system and the CNS (Figure 1).^80,81 This unique anatomical structure/location, together with that the BM is a non-cellular component of the BBB, suggests that the BMs also participate in the regulation of BBB integrity. The BMs may exert this important function directly by serving as a physical barrier at the BBB or indirectly by signaling to endothelial cells, astrocytes, and/or pericytes. In support of the latter possibility, receptors for BM components (ECM proteins), including integrins and dystroglycan, have been identified in these cells.^40,47 More importantly, genetic ablation of these receptors and their respective ligands usually results in similar phenotypes,^12,40,41,48 strongly suggesting that ECM proteins exert functions by binding to their receptors. Below, we discuss the function of each major BM component in BBB maintenance.

Collagen IV

Col4a1/2, two of the most frequently used α chains, make the major collagen IV isoform α1(IV)₂α2(IV). Genetic ablation of Col4a1/2 leads to embryonic lethality at embryonic day (E) 10.5-11.5.⁸² Interestingly, BM defects were detected in these mutants at E10.5–11.5 but not before E9.5,⁸² suggesting that collagen IV is dispensable for initial BM assembly but essential for BM maintenance and function. To overcome the embryonic lethality, a series of less severe mutations in Col4a1/2 have been generated in mice.^83,84 These mutants develop CNS phenotypes, including compromised vascular integrity, stress-induced hemorrhage and adult-onset stroke, to various degrees (see “Collagen IV changes in hemorrhagic stroke” section for details),^83–85 suggesting an indispensable role of collagen IV in the maintenance of cerebrovascular integrity.

Laminin

Since different cells synthesize distinct laminin isoforms, laminin’s functions are discussed in a cell-specific manner. Endothelial cells mainly make laminin-411 and -511.^21,49,50 It has been shown that ablation of laminin α4 leads to disrupted vascular integrity and hemorrhage at embryonic and perinatal stages, but not in adulthood.⁸⁶ This age-dependent phenotype is due to compensatory and ubiquitous expression of laminin α5 in the vascular tree at postnatal stage.⁸⁷ Unlike laminin α4, laminin α5,⁸⁸ β1,^89,90 or γ1,^91–92 null mice are embryonic lethal, preventing investigation of their functions in BBB integrity. To overcome this early lethality and enable such investigation, laminin α5 endothelium-specific (Tie2-Cre) conditional knockout (α5-TKO) mice were generated. Using two different laminin α5-floxed lines, our group (unpublished data) and others^93,94 independently demonstrate that α5-TKO mice develop normally without BBB defects under homeostatic conditions, suggesting a dispensable role of endothelial laminin α5 in BBB maintenance. Due to potential compensation between laminin-411 and -511, it remains unclear how endothelial laminin regulates BBB integrity under physiological conditions. We are currently generating transgenic mice lacking both laminin isoforms in endothelial cells. These mutants will enable us to answer this important question.

Unlike endothelial cells, astrocytes predominantly produce laminin-211.^21,51 To investigate the role of astrocytic laminin in BBB integrity, we generated neural progenitor- and neuron-specific laminin γ1 conditional knockouts by crossing the laminin γ1^flox/flox mice⁹⁵ with nestin-Cre and CaMK2a-Cre transgenic lines, respectively. The former mutants (termed NKO hereafter) have laminin deficiency in both neurons and glial cells, while the latter mutants (termed CKO hereafter) have laminin deficiency in neurons only. NKO but not CKO mice display BBB disruption⁹⁶ and age-dependent hemorrhagic stroke,⁹⁷ suggesting that astrocytic rather than neuronal laminin is required for BBB maintenance. Consistent with this finding, laminin α2 null mice develop severe BBB breakdown,⁹⁸ again suggesting an indispensable role of astrocytic laminin in BBB maintenance.

In addition to endothelial cells and astrocytes, pericytes also synthesize and deposit laminins into the BMs. Although laminin α2 mRNA was found in pericytes in next-generation sequencing,⁹⁹ our unpublished data show that pericytes mainly produce γ1 - and α4/α5-containing laminins. To investigate the function of pericytic laminin, pericytic laminin conditional knockouts (PKO) were generated by crossing the laminin γ1^flox/flox mice with the Pdgfrβ-Cre line. It should be noted that there is no pericyte-specific marker available currently,¹⁰⁰ and that PDGFRβ marks both pericytes and vascular smooth muscle cells (vSMCs).^100,101 Thus, laminin is abrogated in both pericytes and vSMCs in PKO mice. To distinguish pericyte- and vSMC-derived laminins, vSMC-specific conditional knockouts (SKO) were generated by crossing the laminin γ1^flox/flox mice with the SM22α-Cre line. In the C57Bl6-FVB mixed background, the PKO mice usually die within four months and show hydrocephalus & BBB breakdown/micro-hemorrhages with incomplete penetrance.¹⁰¹ These changes, however, are absent in SKO mice,^101,102 suggesting a critical role of pericyte- rather than vSMC-derived laminins in BBB maintenance. Since hydrocephalus itself can compromise BBB integrity, it is not clear if the BBB breakdown phenotype is caused by loss of pericytic laminin or secondary to hydrocephalus. Based on that hydrocephalus is highly genetic background dependent,^103–106 we crossed the PKO mice into various genetic backgrounds, aiming to reduce the incidence of hydrocephalus. Our unpublished data show that, in the C57Bl6 background, the PKO mice develop mild and age-dependent BBB disruption without hydrocephalus or micro-hemorrhages, strongly indicating that pericytic laminin also contributes to BBB maintenance under homeostatic conditions, although to a lesser extent compared to astrocytic laminin. The cellular source of different laminin isoforms and their distribution in endothelial and parenchymal BMs are summarized in Figure 1(c). The major phenotypes of these laminin mutant mice are summarized in Table 2.

Table 2.

Laminin mutants and their phenotypes.

Laminin mutations	Type of mutation	Cre lines	Target cells	CNS phenotypes	References
Laminin α2−/−	Global knockout	–	All cells	BBB breakdown	98
Laminin α4−/−	Global knockout	–	All cells	Hemorrhage at embryonic and perinatal stages	86
Laminin α5−/−	Global knockout	–	All cells	Embryonic lethality around E17 and exencephaly	88
Laminin β1−/−	Global knockout	–	All cells	Embryonic lethality at E5.5 and lack of BMs	89,90
Laminin γ1−/−	Global knockout	–	All cells	Embryonic lethality at E5.5 and lack of BMs	89–92
Laminin α5 -ΤKO	Conditional knockout	Tie2-Cre	Endothelial cells	Grossly normal under homeostatic conditions	93,94 Unpublished data
Laminin γ1-NKO	Conditional knockout	Nestin-Cre	Neural progenitor cells	Severe BBB disruption and age-dependent spontaneous hemorrhagic stroke	96,97
Laminin γ1-CKO	Conditional knockout	CaMK2a-Cre	Forebrain neurons	Grossly normal under homeostatic conditions	96,97
Laminin γ1-PKO	Conditional knockout	Pdgfrβ-Cre	Pericytes and vSMCs	In the C57Bl6-FVB mixed background: hydrocephalus, BBB compromise and micro-hemorrhages (incomplete penetrance)	101
Laminin γ1-PKO	Conditional knockout	Pdgfrβ-Cre	Pericytes and vSMCs	In the C57Bl6 background: age-related mild BBB compromise	Unpublished data
Laminin γ1-SKO	Conditional knockout	SM22α-Cre	vSMCs	Grossly normal under homeostatic conditions	101,102

CNS: central nervous system; vSMCs: vascular smooth muscle cells.

Nidogen

Genetic abrogation of either nidogen-1^107–110 or nidogen-2¹¹¹ results in a grossly normal phenotype, suggesting that they may be able to compensate for each other’s loss in these single mutants. Consistent with this hypothesis, nidogen-1/2 double knockout mice die shortly after birth and show severe BM defects in multiple organs,^112,113 suggesting an important role of nidogen-1/2 in BM maintenance. Based on the observed BM defects in the brain, we expect that BBB integrity is compromised in these double mutants. This hypothesis needs further validation/investigation in future studies.

HSPG

Global knockout of HSPG2 leads to severe developmental defects in multiple organs.^114,115 The mutants die either around E11 or at the perinatal stage due to defective cephalic, myocardial, and skeletal development.^114,115 Although BM formation is not affected, BMs deteriorate in areas with increased mechanical stress in these mutants,¹¹⁴ indicating an essential role of HSPG2 in BM maintenance rather than assembly. Consistent with BM changes, a hemorrhagic phenotype in the brain, skin and lung is observed in these mutants,¹¹⁴ suggesting that HSPG2 is required for the maintenance of vascular integrity.

BM and stroke

Stroke is a clinical emergency caused by rupture (hemorrhagic) or occlusion (ischemia) of cerebral blood vessels. It is the fifth leading cause of death and the leading cause of serious long-term disability in the United States.¹¹⁶ One key pathological change after stroke is inflammatory cell infiltration across the compromised BBB, which profoundly influences disease progression. Due to their important roles in BBB maintenance and vascular integrity, the BMs have been hypothesized to affect inflammatory cell extravasation and stroke pathogenesis. Previous studies have demonstrated that immune cell infiltration predominantly occurs at specific low expression regions characterized by reduced expression of certain BM constituents.^87,117–121 For example, neutrophils^117–120 and monocytes¹¹⁷ preferentially penetrate the BM in regions with reduced levels of laminin α5 and collagen IV in peripheral tissues. In addition, in both lymph nodes and brains, T lymphocytes mainly extravasate across blood vessels through laminin α4^high and laminin α5^low regions.^87,121 It is speculated that inflammatory cells use these low expression regions to infiltrate the brain after stroke. However, due to the lack of knowledge on the alterations of specific laminin isoforms after stroke and the existence of controversial results on how collagen IV/laminin change after ischemic stroke (Figure 1(d)), the function of the BM in inflammatory cell extravasation after stroke remains largely unknown and needs further investigation. Here, we discuss BM changes and functions in both hemorrhagic and ischemic stroke.

BM and hemorrhagic stroke

Multiple studies have shown that the BM negatively correlates with hemorrhagic stroke. For example, BM degradation has been reported in rats at 24–72 h after subarachnoid hemorrhage¹²² and in mice with age-dependent spontaneous intracerebral hemorrhage (ICH).⁹⁷ In addition, loss of BM components and/or their mutations usually lead to BM defects and hemorrhagic stroke (see below for details). BM loss during hemorrhagic stroke may be caused by diminished ECM protein synthesis or increased ECM protein degradation. We lean towards the second possibility. This is because: (1) ECM proteins usually have a long turnover rate,¹²³ and (2) various ECM-degrading proteases, including matrix metalloproteinases (MMPs) and thrombin, are dramatically induced/activated after hemorrhagic stroke.^124–132 Changes of major BM components after hemorrhagic stroke are discussed below and summarized in Figure 1(d).

Based on these findings, it is logical to hypothesize that inhibiting BM-degrading enzymes and/or supplementing BM components may have a therapeutic potential in hemorrhagic stroke. Consistent with this speculation, various MMP inhibitors, including minocycline, GM6001 and BB-1101, have been found beneficial in hemorrhagic stroke.^124,133,134 In contrast to these results, BB-94, another broad-spectrum MMP inhibitor, has been reported to increase hematoma size and apoptosis after hemorrhagic stroke.¹³⁵ In addition, minocycline failed to affect clinical and radiological outcomes in patients with hemorrhagic stroke in a pilot study.¹³⁶ This inconsistency may be due to different experimental models and/or low specificity of these MMP inhibitors. We are currently investigating the therapeutic potential of exogenous BM components in hemorrhagic stroke.

Collagen IV changes in hemorrhagic stroke

Accumulating evidence suggests that collagen IV plays a pivotal role in the pathogenesis of hemorrhagic stroke. First, collagen IV is negatively associated with hemorrhage in the brain. Reduced collagen IV expression is found in mice lacking endothelial expression of serum response factor, which develop ICH.¹³⁷ Similarly, dramatically lower levels of collagen IV are detected in the hemorrhagic regions in ischemic brain.¹³⁸ Second, loss of collagen IV plays a causative role in hemorrhagic stroke. For example, collagenase, which degrades collagen IV, is widely used to induce ICH in rodents.^139–141 Third, loss-of-function mutations in Col4a1/2 cause ICH to various degrees. It has been shown that mice with a splice-site mutation that skips Col4a1 exon 41 (Col4a1^Δex/+) display perinatal cerebral hemorrhage associated with birth trauma.^85,142 Although surgically delivered mutants do not show cerebral hemorrhage at the perinatal stage, they develop ICH in adulthood.^85,143 Cellular studies demonstrate that this mutation leads to synthesis of mutant collagen IV protein, which cannot be secreted and thus is accumulated inside cells.^83,142,144 To further investigate the molecular mechanism underlying the hemorrhagic phenotype, Col4a1 mutants containing loxP flanked exon 41 (Col4a1^Flex41/+) were generated.⁸⁵ By crossing this line with an inducible ubiquitous Cre line (R26-Cre^ER) and/or various cell-specific Cre lines (Tie2-Cre, Pdgfrβ-Cre, Gfap-Cre), this mutation was induced at different developmental stages and/or in different cell types. It has been shown that induction of such mutation: (1) at embryonic and early postnatal stages, but not in adulthood, causes age-related ICH⁸⁶; and (2) in endothelial cells and/or pericytes but not astrocytes leads to ICH.⁸⁵ These findings suggest that endothelium- and/or pericyte-derived Col4a1 plays an important role in cerebrovascular development rather than maintenance. Additionally, although not as severe as Col4a1^Δex41/+ mutants, mice with various point-mutations (G394V, G658D, G912V, G1038S, G1180D, G1344D, and S1582P) in Col4a1 develop ICH to various degrees.^84,85 Similarly, a point mutation (G646D) in Col4a2 also results in mild ICH.^84,85,144 The severity of ICH in these Col4a1/2 mutants is summarized in Table 3. Consistent with these animal studies, mutations in Col4a1/2 have also been identified in human patients with hemorrhagic stroke.^84,143–158 Together, these results suggest a crucial role of collagen IV in the pathogenesis of hemorrhagic stroke.

Table 3.

Collagen IV mutants and their phenotypes.

Collagen IV mutations	Type of mutation	Cre lines	Target cells	CNS phenotypes	Severity of ICH	References
Col4a1/2−/−	Global knockout	–	All cells	Embryonic lethality at E11, BM defects	–	82
Col4a1Δex41/+	Skipping exon 41	–	All cells	ICH, porencephaly	Severe	85,142,143
Col4a1Flex41/+	Inducible conditional knockout	R26-CreER	All cells with tamoxifen	ICH and macroangiopathy	Depending on induction time	85
	Conditional knockout	Tie2-Cre	Endothelial cells	ICH, porencephaly and macroangiopathy	Moderate	85
	Conditional knockout	Pdgfrβ-Cre	Pericytes and vSMCs	ICH, porencephaly and macroangiopathy	Moderate	85
	Conditional knockout	Gfap-Cre	Astrocytes	ICH	Very mild	85
Col4a1G394V/+	Point mutation	–	All cells	ICH	Mild	84,85
Col4a1G658D/+	Point mutation	–	All cells	ICH	Mild	84,85
Col4a1G912V/+	Point mutation	–	All cells	ICH	Mild-moderate	84,85
Col4a1G1038S/+	Point mutation	–	All cells	ICH	Moderate	84,85
Col4a1G1180D/+	Point mutation	–	All cells	ICH	Moderate	84,85
Col4a1G1344D/+	Point mutation	–	All cells	ICH	Moderate	84,85
Col4a1S1582P/+	Point mutation	–	All cells	No ICH	-	84,85
Col4a2G646D/+	Point mutation	–	All cells	ICH	Mild	84,85,144

ICH: intracerebral hemorrhage; CNS: central nervous system.

Laminin changes in hemorrhagic stroke

Due to the substantial up-regulation and activation of various BM-degrading proteases after hemorrhagic stroke, laminin levels are expected to decrease in hemorrhagic brains. Consistent with this hypothesis, significantly lower levels of laminin were found in hemorrhagic areas compared to non-hemorrhagic areas in ischemic brain.¹⁵⁹ Interestingly, loss of laminin is associated with hemorrhagic stroke in an isoform-specific manner. First, a previous study from our laboratory showed that NKO but not CKO mice developed age-dependent spontaneous ICH,⁹⁷ suggesting a causative role of loss of astrocytic laminin (laminin-211) in hemorrhagic stroke. Next, although micro-hemorrhages and hydrocephalus occur hand-in-hand in PKO mice in the C57Bl6-FVB mixed background, micro-hemorrhages are absent in these mutants in the C57Bl6 background when the incidence of hydrocephalus is reduced to a negligible level. These findings suggest that micro-hemorrhages in PKO mice in the mixed background are probably secondary to hydrocephalus and that loss of pericytic laminin does not cause hemorrhagic stroke. Additionally, due to potential compensation between laminin-411 and -511, whether loss of endothelial laminin results in hemorrhagic stroke remains unknown. Mice with endothelium-specific ablation of both laminin-411 and -511 will enable us to answer this important question.

Nidogen changes in hemorrhagic stroke

How nidogen levels change in hemorrhagic stroke remains largely unknown. One in vitro study showed that nidogen fragmentations were found in astrocyte medium after exposure to plasma kallikrein,¹⁶⁰ suggesting that nidogen may be degraded after hemorrhagic stroke. This result needs to be validated in vivo using hemorrhagic models. Due to the grossly normal phenotype of nidogen-1/2 single knockouts and the perinatal lethality of the double knockouts, the role of nidogen in the pathogenesis of hemorrhagic stroke remains unclear. Mice with simultaneous deletion of nidogen-1/2 in different cell types (conditional nidogen-1/2 double knockouts) will provide insights into the function of nidogen in hemorrhagic stroke.

HSPG changes in hemorrhagic stroke

Like nidogen, how hemorrhagic stroke affects HSPG2 level is unclear. In addition, due to early embryonic lethality, the function of HSPG2 in the development and progression of hemorrhagic stroke is unknown. Similarly, HSPG2 conditional and/or inducible knockout mice will enable us to answer these important questions.

BM and ischemic stroke

Similar to hemorrhagic stroke, BM dissolution is associated with ischemic stroke. For example, loss of BM is found soon after the onset of ischemia.¹⁶¹ In addition, it has been reported that BM degradation occurs as early as 10 min after reperfusion in the middle cerebral artery occlusion (MCAO) model,¹⁶² and that loss of BM can be detected as early as 1–3 h after ischemia.^161,162 Ultrastructurally, the well-defined electron-dense BMs become diffused and faint after ischemia.^163,164 Biochemically, many BM components are degraded and reduced after ischemic stroke.^161,165 Like in hemorrhagic stroke, BM loss in ischemic stroke is mainly due to enhanced protein degradation rather than reduced protein synthesis. This is supported by the following observations. First, MMP expression and their activities are markedly up-regulated after ischemic stroke.^166–171 Second, the expression levels and enzymatic activities of cathepsin B and L, which primarily degrade HSPG2 and other cellular components,¹⁷² are dramatically enhanced after ischemia.^172,173 The changes of major BM components after ischemic stroke are discussed below and summarized in Figure 1(d).

Recombinant tissue-type plasminogen activator (tPA) is the only FDA-approved therapy for ischemic stroke. tPA exerts its protective effects by inducing plasmin-dependent fibrinolysis inside blood vessels.¹⁷⁴ However, with the progression of disease, tPA leaks out of blood vessels through compromised BBB. Outside blood vessels, tPA substantially enhances MMP-9 levels and exaggerates MMP-mediated BM degradation.^175–177 This deleterious effect outside blood vessels may explain the short (4.5 h) therapeutic window of exogenous tPA in ischemic stroke. Based on these results, we hypothesize that inhibiting MMPs and/or supplementing BM components may be beneficial in ischemic stroke. In agreement with this hypothesis, many broad-spectrum MMP inhibitors, including minocycline, doxycycline, GM6001 and BB-94, displayed protective effects in ischemic stroke,^178–181 although one study showed that intravenous injection of minocycline was safe but not effective.¹⁸² SB-3CT, an MMP-2/9 specific inhibitor, restored laminin degradation, reduced infarct volume, and improved neurological function in a mouse model of ischemic stroke.^183,184 In addition, many studies showed that MMP inhibitor-tPA combined therapies significantly reduced hemorrhagic complications of tPA.^{178,185–189} We are currently investigating the therapeutic effects of BM components in ischemic stroke.

Collagen IV changes in ischemic stroke

Unlike in hemorrhagic stroke, controversial results exist on how collagen IV level changes in ischemic stroke. Most studies support that collagen IV is reduced after ischemic stroke. For instance, reduced collagen IV levels were found in rats with thromboembolic¹⁹⁰ or suture^132,191 model of MCAO. In addition, substantial loss of collagen IV was also detected in non-human primate baboons (Papio anubis/cynocephalus) after ischemia-reperfusion injury.^172,192 Furthermore, an association between ischemic stroke and Col4a1 mutations was found in humans,^150,193,194 suggesting that collagen IV may be involved in the pathogenesis of ischemic stroke. In contrast to these results, one study found increased collagen IV expression in the spinal cord 24 h after ischemia–reperfusion injury.¹⁹⁵ In addition, it was also reported that MCAO alone without systemic inflammation failed to affect collagen IV level in the brain.¹⁹⁶ This discrepancy may be due to different organs examined (brain vs. spinal cord). Whether ischemic stroke affects collagen IV level in an organ-specific manner needs further investigation.

Laminin changes in ischemic stroke

Like collagen IV, laminin changes after ischemic stroke have been controversial. On one hand, there is evidence demonstrating that laminin is reduced after ischemic stroke. It has been shown that ischemic stroke increases the activity of MMP-9, which degrades laminin in mouse brain.¹⁸⁴ In addition, laminin expression is substantially decreased in Mongolian gerbils (Meriones unguiculatus) up to 72 h of reperfusion after ischemia.¹⁹⁷ Similarly, ischemia-reperfusion injury induces a gradual and continuous reduction of both laminin expression and laminin-positive vessel number in baboons.^172,192 Furthermore, diminished laminin level is also found in patients with ischemic stroke.¹⁹⁸ Interestingly, loss of laminin correlates well with the increase/activation of various proteases,^172,192,197 again suggesting that proteolytic degradation is responsible for the loss of laminin in ischemic stroke.

On the other hand, there is also evidence showing that laminin expression is unaffected or even enhanced after ischemic stroke. For example, laminin level was found unchanged in the brain after MCAO without systemic inflammation.¹⁹⁶ Laminin was also reported to be up-regulated in the ischemic core within 24 h after injury.^199,200 Additionally, astrocytes have been shown to express high levels of laminin two to three days after ischemia, which forms a barrier and separates the ischemic regions from healthy tissue.^199,200 Furthermore, it has been demonstrated that ischemia induces a transient up-regulation of laminin in the brain in a COX-2 dependent manner.²⁰¹

These paradoxical results may be explained by different animal models used and time points examined. In addition, distinct fixation methods may also contribute to this discrepancy. It has been shown that heavy formaldehyde fixation masks laminin antigen in the BM, while mild-moderate formaldehyde fixation reveals it.²⁰² Furthermore, it is worth mentioning that most of the above-mentioned studies used a pan-laminin antibody to examine laminin expression. It remains unclear whether different laminin isoforms have distinct susceptibility to ischemia and how these laminin isoforms change after ischemia. Future research should address these important questions.

Nidogen changes in ischemic stroke

Like in hemorrhagic stroke, nidogen changes in ischemic stroke are largely unknown. One in vitro study showed that oxidative stress, one pathological alteration found in ischemic stroke, enhanced nidogen expression in human BMECs,²⁰³ suggesting that nidogen may be up-regulated after ischemic stroke. Whether ischemic stroke increases nidogen expression in vivo and how nidogen affects the pathogenesis of ischemic stroke need further investigation.

HSPG changes in ischemic stroke

There is evidence showing that ischemic stroke decreases HSPG2 levels. One study reported a 43–63% reduction of HSPG2 within a few hours after ischemic injury in baboons,¹⁷² suggesting that HSPG2 is one of the most sensitive ECM proteins to proteolysis in ischemic brain.¹⁷² Unlike other ECM proteins, which get degraded, HSPG2 is cleaved into small fragments after stroke. For instance, high levels of HSPG2 domain V (endorepellin)^204,205 and its C-terminal fragment²⁰⁶ are detected in rodents in multiple ischemic stroke models. Consist with these reports, a similar result was found in humans with ischemic stroke.²⁰⁷ Further studies demonstrate that these HSPG2 fragments are bioactive and exert a beneficial role in ischemic stroke. It has been reported that these HSPG2 fragments are able to diminish ischemic volume, reduce neuronal death, modulate astrogliosis, enhance angiogenesis, and improve motor function.^204,206,208

Future directions

Due to its intrinsic complexity and technical difficulties in its research, the BM has been understudied. With advances in biochemistry and genetics, many valuable research tools (e.g. transgenic mouse lines) have been generated and substantial progress has been made in this field. Many important questions, however, remain to be answered. These questions include: when and where is each BM component expressed? What is the relative contribution of these BM components from each cell type? Is there functional compensation among different isoforms of ECM proteins? How do distinct laminin isoforms change after stroke? How do nidogen and HSPG2 levels change in hemorrhagic and ischemic stroke? How does each BM component affect the pathogenesis of hemorrhagic and ischemic stroke? Answers to these questions rely on the generation of cell-specific conditional knockouts, compound mutants (e.g. double knockouts), and/or inducible knockouts. Future studies should focus on generating these mutants and answering the above-mentioned questions.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was partially supported by AHA Scientist Development Grant (16SDG29320001).

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.