Mutations in Col4a1 Cause Perinatal Cerebral Hemorrhage and Porencephaly

See allHide authors and affiliations

Science  20 May 2005:
Vol. 308, Issue 5725, pp. 1167-1171
DOI: 10.1126/science.1109418


Porencephaly is a rare neurological disease, typically manifest in infants, which is characterized by the existence of degenerative cavities in the brain. To investigate the molecular pathogenesis of porencephaly, we studied a mouse mutant that develops porencephaly secondary to focal disruptions of vascular basement membranes. Half of the mutant mice died with cerebral hemorrhage within a day of birth, and ∼18% of survivors had porencephaly. We show that vascular defects are caused by a semidominant mutation in the procollagen type IV α 1 gene (Col4a1) in mice, which inhibits the secretion of mutant and normal type IV collagen. We also show that COL4A1 mutations segregate with porencephaly in human families. Because not all mutant mice develop porencephaly, we propose that Col4a1 mutations conspire with environmental trauma in causing the disease.

Porencephaly [Online Mendelian Inheritance in Man (OMIM) record 175780] is a rare central nervous system disease usually diagnosed in infants. Type I or encephaloclastic porencephaly is characterized by cerebral white-matter lesions and degenerative cavities. Severe cases have drastic consequences, including profound disability and death. Infants who survive are often diagnosed with poor or absent speech development, epilepsy, hydrocephalus, seizures, mental retardation, and cerebral palsy. It has been suggested that porencephalic cavities in humans result from focal cerebral degeneration involving hemorrhages (1). Association studies suggest that clotting-factor genes may contribute to genetic susceptibility by predisposing to thrombophilia (2). Despite these associations, the genetic and environmental etiology of familial cases is not established (310), and it seems reasonable that a distinct mechanism involving primary defects of vasculature could predispose to hemorrhage and porencephaly.

To advance the understanding of porencephaly, we have identified and characterized a new mouse mutant (generated by random mutagenesis) with severe perinatal cerebral hemorrhage. In addition to cerebral hemorrhage, mutant mice are smaller than control littermates and can have multiple pleiotropic phenotypes (including ocular abnormalities, mild renal abnormalities, and reduced fertility) that appear to be influenced by genetic context (11). Homozygous mutant mice are not viable after mid-embryogenesis, and about 50% of heterozygous mice die within a day of birth (Fig. 1A). Reduced viability may be explained by variability in the severity and/or location of cerebral hemorrhages that are externally visible in most mutant pups. Detailed analysis of a subset of postnatal day 0 (P0) pups identified cerebral hemorrhages in 12 out of 12 mutant pups but in 0 out of 9 littermate controls (Fig. 1, B and C). About 18% of adult heterozygous mutant mice had obvious porencephalic lesions (6 out of 33) that were not observed in wild-type controls (0 out of 17, Fig. 1, D to G).

Fig. 1.

Mutant mice have cerebral hemorrhage, reduced viability, and porencephaly. (A) Mutant mice (Col4a1+/Δex40, see Fig. 2) have reduced perinatal viability. Col4a1+/Δex40 mice were bred to wild-type C57BL/6J mice, and the expected proportion of Col4a1+/Δex40 progeny was 50%. Col4a1+/Δex40 pups were present at the expected frequency on the day of birth (P0). By weaning age (P21), over 50% of the mutants were missing (P < 0.001). Almost all of the missing pups died within a day of birth, suggesting that stress associated with parturition may be an important factor in their deaths. n, number of mice. (B and C) After birth, cerebral hemorrhages are evident in mutant [(C), arrow] but not control (B) pups. (D to G) Porencephalic cavities [(E), arrow] and (G) were observed in adult mutant, but not age-matched control [(D) and (F)], mice. (G) is from the same brain shown in (E) and demonstrates the absence of the left cerebral cortex (22). All scale bars, 1 mm.

We mapped the causative gene to a 226-kilobase region on chromosome 8 that contains a single gene encoding procollagen type IV α 1 (Col4a1) (Fig. 2A and fig. S1). It has previously been suggested that a large undefined region including this locus contains a gene involved in cerebral hemorrhage in mice (12). Control and mutant embryos expressed Col4a1 transcripts of distinct size (Fig. 2B). In mutant transcripts (hereafter called Col4a1Δex40), exon 39 was spliced directly to exon 41 because of a mutation in the splice-acceptor site of exon 40 (Fig. 2C).

Fig. 2.

The mutant mice have a mutation in the Col4a1 gene. (A) Map of critical interval. More than 2000 mice were analyzed, and key recombinant chromosomes are shown (22). The thin lines indicate the chromosomal region excluded by recombination. Numbers of mice with each type of chromosome are indicated on the left. Phenotypes of critical recombinants were confirmed by progeny testing. Physical locations for all markers are in megabases (Mb) from the centromere [National Center for Biotechnology Information (NCBI) mouse build 33]. (B) Reverse-transcription polymerase chain reaction of exons 35 to 41 of Col4a1 revealed a smaller amplicon (transcript) in mutant embryos as compared with controls. The mutant transcript lacks exon 40. (C) Sequence analysis of genomic DNA revealed a G-to-A transition in the splice acceptor site of exon 40. The intron sequence is shown in lowercase and the exon 40 sequence is shown in uppercase.

Type IV collagens are basement membrane (BM) proteins (1315) that are expressed in all tissues, including the vasculature (16). COL4A1 and COL4A2 are the most abundant type IV collagens. COL4A1 and COL4A2 form heterotrimers with a 2:1 stoichiometry, respectively (17, 18). Assembly of the heterotrimers is initiated by the C-terminal non-collagenous domains, and the heterotrimer forms a triple helix along the length of the collagenous domains (fig. S2B). The skipped exon in Col4a1Δex40 mice is in the triple helix–forming domain and codes for exactly 17 amino acids. One mechanistic hypothesis is that after normal heterotrimer initiation, the mutant proteins alter triple helix formation or structure and thus inhibit heterotrimer secretion into the BM. The triple helix domain consists of long stretches of Gly-Xaa-Yaa repeats (where Xaa and Yaa are amino acids), with frequent interruptions in the repeats. These interruptions are thought to be important for the flexibility of BM collagens (1921). Because the skipped exon contains a repeat imperfection whose position is highly conserved across species (19, 20), an alternative hypothesis is that the mutant protein is assembled and secreted, but its structural properties in the vascular BMs are altered.

To distinguish between these hypotheses, we assessed the effect of the mutation on the secretion of COL4A1 and COL4A2 into BMs. Because Col4a1Δex40/Δex40 embryos are not viable after mid-gestation, we assayed for COL4A1 and COL4A2 expression in the Reichert's membrane (RM, a BM of embryonic origin) by immunolabeling embryonic day 9.5 (E9.5) embryos. Col4a1+/+ embryos showed robust labeling of COL4A1 and COL4A2 in the RM (Fig. 3B). Similarly, both secreted and intracellular COL4A1 and COL4A2 were present in Col4a1+/Δex40 embryos (Fig. 3, E, J, and M). In contrast, although the RM of Col4a1Δex40/Δex40 embryos labeled robustly for laminin (Fig. 3G), there was no evidence of COL4A1/A2 heterotrimer secretion into this BM (Fig. 3, H, L, and O). Instead, the proteins appeared to accumulate within the parietal endoderm cells. These findings are consistent with the hypothesis that heterotrimers form but are not secreted (22). Similar to our results, a mutation of the nematode Caenorhabditis elegans ortholog of Col4a1 (emb-9) results in the impaired secretion and intracellular accumulation of the products of both emb-9 and the Col4a2 ortholog let-2 (23).

Fig. 3.

The COL4A1 mutant inhibits collagen secretion into the BMs. (A to C) Col4a1+/+, (D to F, J, and M) Col4a1+/Δex40, and (G to I, K, L, N, and O) Col4a1Δex40/Δex40 mice are shown. In (A), (D), and (G), immunolabeling for laminin demonstrates the presence of RMs in embryos of each genotype (22). In (B) and (E), an antibody that recognizes both COL4A1 and COL4A2 shows the presence of these proteins in RMs of Col4a1+/+ (B) and Col4a1+/Δex40 (E) embryos, which is confirmed by colocalization with laminin, shown in (C) and (F), respectively. (G) to (I) show that in contrast, Col4a1Δex40/Δex40 embryos do not have collagen staining in the RM (arrow) and instead show nonsecreted mutant protein in the parietal endoderm cells (arrowheads). In (J) and (M), specific monoclonal antibodies show that COL4A1 (J) and COL4A2 (M) are present in both the parietal endoderm cells and the RM of Col4a1+/Δex40 embryos. (L) and (O) show that in contrast, no COL4A1 or COL4A2 is present in the RM of Col4a1Δex40/Δex40 embryos (arrows). Differential interference contrast (DIC) images showing RMs are shown in (K) and (N). Scale bar, 0.2 μm.

In further support for a dominant role of the mutant collagen in pathogenesis, mice heterozygous for null alleles of Col4a1 and Col4a2 are normal (24). Presumably, these heterozygous mice produce fewer COL4A1/2 heterotrimers than do control mice, but all of the heterotrimers should have normal structure and be secreted. Although Col4a1+/Δex40 mice may secrete even less normal collagen [possibly only 25% (fig. S2)], the absence of a phenotype in mice with a null allele is consistent with a requirement for mutant proteins to induce disease. Analysis of mutations in the C. elegans Col4a2 ortholog let-2 supports this notion (25). Consistent with a potential pathogenic role of nonsecreted proteins, we observed swollen vascular endothelial cells with prominent vesicles in Col4a1+/Δex40 mice (Fig. 4).

Fig. 4.

Col4a1+/Δex40 mutant mice have structural defects in the cerebral vasculature. Electron micrographs of cerebral vessels of mice of the indicated genotypes are shown (22). (A) The vascular BMs of all analyzed Col4a1+/+ mice had well-defined edges, uniform density, and consistent thickness (between arrows). (B) In contrast, the vascular BMs of Col4a1+/Δex40 mice often had uneven edges, inconsistent density, and highly variable thickness (between arrows). Focal disruption (asterisks) and herniation (H) of BMs were specific to mutant mice and probably represent damage due to a weaker BM. Disruptions were observed in about 20% of vessels, whereas qualitative differences such as variable thickness and inconsistent and rough edges existed in all mutant vascular BMs analyzed. Mutant endothelial cells appear enlarged with an accumulation of vesicles, a characteristic of deficient secretion (arrowheads). Further analysis of phenotypes observed in noncerebral tissue reveals that BM defects are not restricted to the brain (11). L indicates vessel lumen. Scale bar, 0.5 μm.

COL4A1 and COL4A2 give strength to BMs (24, 26). Mouse embryos homozygous for null alleles of both Col4a1 and Col4a2 (24), or for null alleles of type IV collagen–processing enzymes (26, 27), die around mid-gestation with disrupted embryonic BMs. Similarly, BMs are weak in C. elegans type IV collagen mutants (23, 28). To investigate whether the inhibition of heterotrimer secretion in Col4a1+/Δex40 mice compromises the structural integrity of the vascular BMs, we assessed the BMs of cerebral vessels by electron microscopy. Compared with controls, Col4a1+/Δex40 mice had uneven BMs with inconsistent density and focal disruptions (Fig. 4). Although detailed studies are needed, we found that BMs in other tissues also were affected. However, the major site of hemorrhage was the brain. This may be explained by tissue-specific compositional differences in the vascular BMs or the vascular wall (16). Alternatively or additionally, the mutant BMs may greatly predispose to hemorrhage at times of stress. Substantial stress on the head during birth could explain the substantial cerebral hemorrhage.

To determine whether COL4A1 mutations in humans also cause porencephaly, we assessed two families with autosomal dominant porencephaly (6, 10) for mutations in COL4A1 (GenBank accession number 000013) on chromosome 13q34 (Fig. 5). The first family (6) has a Gly1236 to Arg1236 (G1236R) mutation that segregates with the porencephaly but was not present in 192 ethnically and geographically matched Dutch control chromosomes. The second family (10) has a Gly749 to Ser749 (G749S) mutation that also segregates with the phenotype but was not present in 192 ethnically and geographically matched Italian control chromosomes. Both mutations change conserved Gly residues within Gly-Xaa-Yaa repeats in the triple helix domain (fig. S2). Glycine has a single hydrogen-atom side chain, and there is little tolerance for amino acids with larger side chains that likely disrupt the triple helix during collagen assembly (29). Gly-to-Arg and Gly-to-Ser mutations have been shown to be pathogenic in the C. elegans Col4a1 ortholog (23, 25) and in human COL4A5, respectively (30).

Fig. 5.

COL4A1 mutations in two human families with porencephaly. (A and C) Sequence chromatograms for unaffected members from family 1 and family 2, respectively. (B) Chromatogram for an affected patient from family 1, showing the G3706A transition mutation. (D) Chromatogram for an affected patient from family 2, showing the G2245A transition mutation.

We have shown that mutations in Col4a1 can lead to perinatal cerebral hemorrhage and can predispose to porencephaly. Because not all mice of a defined genetic background develop porencephaly, the COL4A1 mutations in humans may conspire with environmental factors (for example, birth trauma) to cause the disease. In addition to having porencephaly, one adult patient from family 1 suffered from recurrent hemorrhagic strokes, which were caused by his weakened vessels (fig. S3). Thus, it is possible that mutations in COL4A1 could contribute to hemorrhagic stroke in the absence of porencephaly. In addition to COL4A1, alleles of COL4A2 or other genes encoding BM or BM-associated proteins (including laminins, entactin/nidogen, perlecan, and integrins) also may be important predisposing factors for cerebral hemorrhage with or without porencephaly.

Our findings may have important implications for disease prevention in families with porencephaly resulting from vascular defects caused by mutations in COL4A1 or other BM genes. For at-risk individuals, preemptive measures could be taken to reduce stress on the abnormal cerebral vessels that may in turn reduce the neurological deficits. For example, cesarean delivery of at-risk babies may increase the likelihood of survival and the possibility of a healthy life by decreasing the severity of cerebral hemorrhage and its sequelae.

Supporting Online Material

Materials and Methods

SOM Text

Figs. S1 to S3

Table S1

References and Notes

References and Notes

View Abstract

Navigate This Article