The Col4a2em1(IMPC)Wtsi mouse line: lessons from the Deciphering the Mechanisms of Developmental Disorders program

ABSTRACT The Deciphering the Mechanisms of Developmental Disorders (DMDD) program uses a systematic and standardised approach to characterise the phenotype of embryos stemming from mouse lines, which produce embryonically lethal offspring. Our study aims to provide detailed phenotype descriptions of homozygous Col4a2em1(IMPC)Wtsi mutants produced in DMDD and harvested at embryonic day 14.5. This shall provide new information on the role Col4a2 plays in organogenesis and demonstrate the capacity of the DMDD database for identifying models for researching inherited disorders. The DMDD Col4a2em1(IMPC)Wtsi mutants survived organogenesis and thus revealed the full spectrum of organs and tissues, the development of which depends on Col4a2 encoded proteins. They showed defects in the brain, cranial nerves, visual system, lungs, endocrine glands, skeleton, subepithelial tissues and mild to severe cardiovascular malformations. Together, this makes the DMDD Col4a2em1(IMPC)Wtsi line a useful model for identifying the spectrum of defects and for researching the mechanisms underlying autosomal dominant porencephaly 2 (OMIM # 614483), a rare human disease. Thus we demonstrate the general capacity of the DMDD approach and webpage as a valuable source for identifying mouse models for rare diseases.


INTRODUCTION
The Deciphering the Mechanisms of Developmental Disorders (DMDD) program is an international, systematic effort to identify and characterise the phenotypes of mouse embryos carrying embryonic or perinatal lethal mutations (Dickinson et al., 2016;Geyer et al., 2017c;Mohun et al., 2013;Weninger et al., 2014;Wilson et al., 2016). Comprehensive phenotype data of the embryos of 208 knockout lines are available online (https://dmdd.org.uk/), enabling individual researchers to identify lines relevant to their research.
The heart of DMDD is the characterisation of the phenotype of embryos harvested at embryonic day (E) 14.5. At this developmental stage, organogenesis is complete and comprehensive analysis of the role gene products play in organ formation is possible, even if defects resulting from gene mutation cause death in the foetal or perinatal period (Geyer et al., 2017b;Wilson et al., 2017).
In the DMDD pipeline, a Col4a2 mutant line [Col4a2 em1(IMPC)Wtsi ] was generated. Together with Col4a1, Col4a2 encodes extracellular matrix proteins, which participate in the macromolecular network of basement membranes (Hudson et al., 1993;Jeanne and Gould, 2017;Khoshnoodi et al., 2008;Sado et al., 1998). Basement membranes are thin, amorphous, specialised extracellular matrices, which function as an anchor for epithelial cells (Pozzi et al., 2017). Hence, they are essential components of all vessels and many organs and play a vital role in diverse biological events including embryogenesis, tissue remodelling, wound healing, protection of tissues and organs from exogenous factors, resistance to mechanical stress, and filtration of blood and air (Sado et al., 1998).
Consistent with the distribution of basement membranes and COL4A2 products, mutations of COL4A2 have been found to cause severe disorders with abnormalities of the nervous, vascular and visual system in humans (Jeanne and Gould, 2017;Meuwissen et al., 2015;Verbeek et al., 2012). Abnormalities of other organs are not described yet, except for the occurrence of a renal cyst associated with hematuria in a patient with heterozygote COL4A2 deletion (Gunda et al., 2014). However, systematic studies failed to detect COL4A2 mutations in families with hematuria and symptoms of thin basement membrane nephropathy (TBMN) (Zhang et al., 2007).
Autosomal dominant porencephaly 2 (OMIM #614483), a recently described and rarely diagnosed disease, can be caused either by segment deletions of chromosome 13, which carries the COL4A1 and COL4A2 genes, or by mutations of one of the COL4A2 alleles (Huang et al., 2012;Kirchhoff et al., 2009;McMahon et al., 2015;Meuwissen et al., 2015;Quélin et al., 2009;Vahedi et al., 2007;Yoneda et al., 2012). Heterozygous carriers of COL4A2 mutations chiefly suffer from symptoms caused by brain and eye defects. Brain abnormalities include cavities in the cerebral parenchyma, which result from developmental defects (Verbeek et al., 2012) or recurrent haemorrhagic strokes due to germinal matrix haemorrhage or bleeding from blood vessels with defective basement membranes (Meuwissen et al., 2015;Murray et al., 2014).
In contrast to patients with COL4A2 mutations, patients with COL4A1 mutations showin addition to brain and cardiovascular malformationsdefects of the kidneys and muscles (Jeanne and Gould, 2017;Meuwissen et al., 2015). Patients with segment deletions of chromosome 13, including deletion of both COL4A1 and COL4A2 as well as other genes, also suffer from a range of abnormalities, including brain, heart, lung, kidney and muscle defects, but also craniofacial and skeletal abnormalities (Huang et al., 2012;McMahon et al., 2015;Yang et al., 2013).
In the 1980s, first attempts were made to use the mouse as a model for researching the function of COL4A2, with studies mainly focusing on eye phenotypes (Favor, 1983;Favor, 1986;Favor and Neuhäuser-Klaus, 2000). In those, and a study in which embryos with point mutations were produced from random mutagenesis experiments [C3.Cg-Col4a2 ENU415 /Ieg, C3.D2(Cg)-Col4a2 ENU4003 / Ieg, C3.D2(Cg)-Col4a2 ENU4020 /Ieg], most homozygous embryos died around mid-gestation (E10.5-11.5). Only five Col4a2 ENU415 embryos survived organogenesis and were still alive at E15. These showed developmental delay, microphthalmia and extensive haemorrhages. In contrast, heterozygous mice survived the postnatal period, but suffered from eye defects, cutaneous haemorrhage and cerebral abnormalities (Favor et al., 2007). Lung malformations were found in heterozygous C3.D2(Cg)-Col4a2 ENU4003 /Ieg embryos at E18.5 (Loscertales et al., 2016). Strikingly, despite the important role of COL4A2 in the formation of basement membranes ubiquitously distributed in the mammal body, no defects in any other organ systems were identified in mouse embryos lacking both Col4a2 alleles.
In the scope of the DMDD program, we produced a Col4a2 mutant line, harvested homozygous mouse embryos at E14.5 and generated high-resolution digital volume data of these embryos with the highresolution episcopic microscopy (HREM) technique. This study aims to present detailed descriptions of the morphological phenotype of the Col4a2 mutants and examine the implications of the DMDD project in identifying and researching models of rare diseases by using autosomal dominant porencephaly type 2 as an example.

RESULTS
Comparisons of the three wild-type littermates collected from the Col4a2 em1(IMPC)Wtsi mutant strain with a reference collection of stage-matched controls (204 embryos in total) on the same genetic background did not reveal notable malformations. For the Col4a2 −/− mutants however, such comparisons revealed severe abnormalities in almost all organs (Table 1) and a delay in the development and growth of the individuals. Using the staging system of Geyer et al., complemented by crown-rump-length and Theiler staging in embryos in which the limbs were malformed (Geyer et al., 2017b), showed that all individuals appeared younger than expected for the time of harvesting.
All Col4a2 −/− mutants had a large number of defects of almost all organs, with the most complex malformations affecting the nervous, cardiovascular and skeletal systems.

Nervous system
All four mutants (100%) showed brain and cranial nerve abnormalities (see Movies 1-3). The most intriguing features were large numbers of very small tissue irregularities and tissue protrusion on the superolateral cerebral cortex resulting in a 'bumpy' appearance of the surface (Fig. 1A,B,I,J), and multiple small cyst-like structures beneath the lateral surfaces of the left and right telencephalic hemispheres (Fig. 1E,F). In addition, the basal face of the frontal lobe showed multiple solid and irregularly shaped, but voluminous tissue protrusions displacing the leptomeningeal connective tissues and olfactory fibres (Fig. 1M,N). The motoric portions of the trigeminal nerves were enlarged and haemorrhages were detected inside the trigeminal ganglia (Fig. 1Q,R). The oculomotor nerves had fewer but thicker roots and larger diameters than in controls (Fig. 1S,T).

Cardiovascular system
All mutants (100%) showed a broad spectrum of complex cardiovascular defects. This included serious heart malformations such as double-outlet right ventricle ( Fig. 2A), atrioventricular cushion abnormalities, muscular ventricular septal defect (Fig. 2C,D), bicuspid aortic valve and atrial septal abnormalities (Fig. 2F,G), but also abnormalities of the venous system, especially abnormal morphology or absence of the valve of the ductus venosus (Fig. 2Q).
In two of the mutants, these abnormalities were associated with malformations of the great intrathoracic arteries, such as right retroesophageal subclavian artery (Fig. 2H,J) and stenosis of the ascending aorta (Fig. 2J).

Skeleton
All mutants (100%) had defects of the axial and limb skeleton.
In the axial skeleton, all components were affected to various degrees, with all mutants (100%) showing rib abnormalities (Fig. 3A), such as abnormal cervical ribs, fused ribs, absent costovertebral joints and vertebra abnormalities, such as fused vertebral arches (Fig. 3C).  Other organs In addition to the defects in the skeleton and nervous and cardiovascular system, all mutants (100%) showed abnormalities of their lungs (cystic enlargements of their terminal airways; Fig. 4A), thymus, thyroid glands, and pancreas ( Fig. 4K,M) as well as an unusual, thin-walled cystic structure, which occupied quite a large space in the mesenchymal tissue along the midline between nasal septum and oral cavity (Fig. 4H). Additionally, each of the mutants showed at least one of the following abnormalities ( penetrance in parenthesis): a clearly discernible subcutaneous edema (3/4) (

DISCUSSION
Patients with single-allele mutations of COL4A2 suffer from autosomal dominant porencephaly type 2, cerebral small vessel disease, recurrent intracerebral haemorrhages (ICH), hydrocephalus, schizencephaly or severe eye defects. The chief symptoms of porencephaly type 2 are periventricular cystic lesions, hemiplegia, developmental delays, intellectual disability, microphthalmia, myopia and optic nerve hypoplasia.
Other types of porencephaly show a similar spectrum, but are caused by mutations of COL4A1 or segment deletions of chromosome 13, which affect both the COL4A2 and COL4A1 genes. For researching the role COL4A2 plays in normal development and in the genesis of porencephaly, knockout mouse models were engineered (Favor, 1983;Pöschl et al., 2004) and it quickly became evident that null mutants are prenatally lethal. Unfortunately, in most cases the homozygous individuals died before they had the chance to complete embryogenesis (Favor, 1983;Favor, 1986;Pöschl et al., 2004). This effectively prevented a comprehensive characterisation of the function of the COL4A2 products by analysing structural defects due to Col4a2 deletion.
In contrast, the DMDD program created a Col4a2 em1(IMPC)Wtsi mutant line, which produces homozygous individuals that survive organogenesis. In order to seek an explanation for this, we conducted gene expression analysis of three homozygous Col4a2 em1(IMPC)Wtsi embryos harvested at E9.5. These studies showed some residual Col4a2 gene expression (an 11× knockdown; 1/0.084), which most likely explains the longer survival times of homozygous embryos produced from the Col4a2 em1(IMPC)Wtsi line.
The advantage of this residual activity is that the DMDD line overcomes the limitations of most of the earlier models and permits identification of a large spectrum of important structural defects caused by Col4a2 deletion. This recommends the DMDD strain for studies aiming at researching the function of COL4A2 products in normal development and for developing new procedures for researching, diagnosing and treating autosomal dominant porencephaly type 2.
All examined Col4a2 mutant embryos (100%) have severe tissue abnormalities in the superficial layers of the cerebral cortex. Some lesions are cystic, others solid. This endorses the interpretation that Col4a2 products are essential for the formation of the membrana limitans gliae superficialis and its extension, the membrana limitans gliae perivascularis (Favor et al., 2007). We think that these membranes fail to properly organise the target area for migrating neural cells and, as a result, the migration and settlement of neural cells during cortex development is severely disturbed. Translated to humans, such abnormal cortical arrangement might cause symptoms such as reduced intellectual and motor abilities and epilepsy, which are indeed described to be associated with autosomal dominant porencephaly type 2 (Ha et al., 2016;Jeanne and Gould, 2017;McGovern et al., 2017;Meuwissen et al., 2015;Verbeek et al., 2012).
So far, an abnormal cortical tissue arrangement as observed in mice is not described in humans. However, it is highly likely that this might have escaped diagnosis, since the defects are small enough to go unnoticed in routine diagnostic procedures. In a comparable manner, this might as well be true for the abnormalities of the cranial nerves we observed in mice, but which were not yet described in humans. Building on our mouse data, we hypothesise that patients suffering from autosomal dominant porencephaly type 2 will benefit from checking for the existence of such small defects with the aid of specialised imaging techniques.
All the examined mouse mutants showed heart defects of a severity that is likely to explain prenatal death. But interestingly, the spectrum of the heart defects exceeded the spectrum expected from the distribution of gene expression data. According to earlier studies Col4a2 activity occurs in the atrioventricular valves, the great intrathoracic vessels, and the epicardium (Hanson et al., 2013;Klewer et al., 1998;Sado et al., 1998). Thus, it was unexpected that 100% of the mutants we examined had muscular septal defects. The absence of Col4A2 activity in the ventricular septum led us to the conclusion that the muscular septum defect is highly likely to result from abnormal biomechanical forces in the region of the muscular part of the septum, which are caused by the abnormally formed atrioventricular junction. This is an interesting hypothesis in itself, but also makes the Col4a2 mutant strain a promising model for researching the influence of biomechanical forces on cardiovascular development and remodelling.
COL4A2 products play an important role in the formation and function of all basement membranes (Sado et al., 1998). Such membranes are not restricted to the brain or heart, but exist in many   organs throughout the body . Consistent with this, we were able to show that in mouse embryos with deletion of both Col4a2 alleles, almost all organs are affected, but neither heart malformations nor malformations of the skeleton and many major organs have been described in previous gene ablation studies. One reason for this might be the early lethality of the homozygous embryos, which concealed organ defects and the differences in the manner of phenotyping (Favor et al., 2007;Kuo et al., 2014;Pöschl et al., 2004). Another might be that our phenotyping approach was based on 3D high-resolution digital HREM data analysed according to the standardised protocol and the guidelines developed for the DMDD program (Mohun et al., 2013;Weninger et al., 2014). However, we cannot rule out that the genetic background on which the mutations were studied has a significant influence on the penetrance and severity of pathologies.
Basement membranes also play a major role in airway development (Warburton et al., 1999) and are components of the skin and exocrine and endocrine glands (Sado et al., 1998). This explains why our Col4a2 mutant embryos show an abnormal branching pattern of the lower airways with reduced numbers of oversized alveoles and feature subcutaneous edema and cysts as well as defects of the pancreas and thyroid gland. The observed defects of the endocrine organs are structural, but they might result in functional defects as well. Whether this can be translated to humans is not clear. Nevertheless, it might be beneficial to test patients suffering from porencephaly for endocrine disorders, such as diabetes and hypothyroidism.
Defective function of either Col4a1 or Col4a2 products or deletion of both genes cause similar defects (Jeanne and Gould, 2017;Meuwissen et al., 2015). Furthermore, the Col4a1 and Col4a2 genes are located on the same segment of chromosome 13 and share a promoter region. Hence, we checked Col4a1 expression in three homozygous Col4a2 em1(IMPC)Wtsi embryos harvested at E9.5 in order to exclude a possible disruption of the gene in the Col4a2 em1(IMPC)Wtsi line created in the DMDD program. Since all examined embryos showed wild-type expression levels of Col4a1, we feel it safe to consider the phenotype abnormalities we observed to be a direct result of Col4a2 disruption.
Our study is based on phenotype information of only four embryos. Nevertheless, most features occurred in 100% of the examined embryos, wherefore the findings shed new light on the role Col4a2 plays in embryogenesis and tissue remodelling. In a translatory sense, it also enlarges the knowledge on the potential spectrum of defects associated with autosomal dominant porencephaly type 2 and might even be used as an argument for re-examining patients with high-resolution imaging techniques. However, the most important achievements are that our results stress the importance of DMDD for identifying mouse models for rare hereditary diseases and that they recommend the DMDD Col4a2 em1(IMPC)Wtsi mutant strain as a model for systematically examining autosomal dominant porencephaly type 2. Unfortunately, the line is no longer actively bred and thus no longer at our disposal. According to the regulations of DMDD, the line was archived after the initial phenotype analysis but can be re-derived from the EMMA repository (https://www.infrafrontier.eu/search?keyword=EM:11598) for systematic phenotype examinations.
The resulting genotype was verified by end-point PCR screening and direct sequencing of the deletion amplicon.   For more information about this mouse line, see http://www. mousephenotype.org/data/genes/MGI:88455. This mouse strain can be ordered from the EMMA repository, with EMMA ID: EM:11598 (https://www.infrafrontier.eu/search?keyword=Col4a2).
The HREM data were analysed at the Medical University in Vienna, employing the display, volume rendering, and virtual dissection tools of the 64 bit OsiriX (http://www.osirix-viewer.com/) and Amira (6.1, https://www.fei.com) software packages. Phenotype abnormalities were scored according to standardised protocols (Weninger et al., 2014;Wilson et al., 2016). They were executed on volume-rendered virtual 3D models with voxel sizes of 3×3×3 µm 3 and sequences of subsequent virtual, 3 µm thick section images cutting through the HREM volume data in various planes. For identifying abnormalities, we compared mutant data with the knowledge deduced from integrating information of data stemming from control embryos, which precisely matched the mutants in their developmental stages according to Geyer et al. (2017b). These reference data were created from systematic phenotype analysis of over 200 control embryos bred on the DMDD background and are already partly published (Geyer et al., 2017b;Geyer et al., 2017c). For documentation of abnormal morphology, we display images of sectioned volume models or 2D sections through the respective digital volume data of mutants and selected controls at section planes that are as closely comparable as possible (Fig. 5).