PHACTR1 genetic variability is not critical in small vessel ischemic disease patients and PcomA recruitment in C57BL/6J mice

Recently, several genome-wide association studies identified PHACTR1 as key locus for five diverse vascular disorders: coronary artery disease, migraine, fibromuscular dysplasia, cervical artery dissection and hypertension. Although these represent significant risk factors or comorbidities for ischemic stroke, PHACTR1 role in brain small vessel ischemic disease and ischemic stroke most important survival mechanism, such as the recruitment of brain collateral arteries like posterior communicating arteries (PcomAs), remains unknown. Therefore, we applied exome and genome sequencing in a multi-ethnic cohort of 180 early-onset independent familial and apparently sporadic brain small vessel ischemic disease and CADASIL-like Caucasian patients from US, Portugal, Finland, Serbia and Turkey and in 2 C57BL/6J stroke mouse models (bilateral common carotid artery stenosis [BCCAS] and middle cerebral artery occlusion [MCAO]), characterized by different degrees of PcomAs patency. We report 3 very rare coding variants in the small vessel ischemic disease-CADASIL-like cohort (p.Glu198Gln, p.Arg204Gly, p.Val251Leu) and a stop-gain mutation (p.Gln273*) in one MCAO mouse. These coding variants do not cluster in PHACTR1 known pathogenic domains and are not likely to play a critical role in small vessel ischemic disease or brain collateral circulation. We also exclude the possibility that copy number variants (CNVs) or a variant enrichment in Phactr1 may be associated with PcomA recruitment in BCCAS mice or linked to diverse vascular traits (cerebral blood flow pre-surgery, PcomA size, leptomeningeal microcollateral length and junction density during brain hypoperfusion) in C57BL/6J mice, respectively. Genetic variability in PHACTR1 is not likely to be a common susceptibility factor influencing small vessel ischemic disease in patients and PcomA recruitment in C57BL/6J mice. Nonetheless, rare variants in PHACTR1 RPEL domains may influence the stroke outcome and are worth investigating in a larger cohort of small vessel ischemic disease patients, different ischemic stroke subtypes and with functional studies.

patency. We report 3 very rare coding variants in the small vessel ischemic disease-CADASIL-like cohort (p.Glu198Gln, p.Arg204Gly, p.Val251Leu) and a stop-gain mutation (p.Gln273*) in one MCAO mouse. These coding variants do not cluster in PHACTR1 known pathogenic domains and are not likely to play a critical role in small vessel ischemic disease or brain collateral circulation. We also exclude the possibility that copy number variants (CNVs) or a variant enrichment in Phactr1 may be associated with PcomA recruitment in BCCAS mice or linked to diverse vascular traits (cerebral blood flow pre-surgery, PcomA size, leptomeningeal microcollateral length and junction density during brain hypoperfusion) in C57BL/6J mice, respectively. Genetic variability in PHACTR1 is not likely to be a common susceptibility factor influencing small vessel ischemic disease in patients and PcomA recruitment in C57BL/6J mice. Nonetheless, rare variants in PHACTR1 RPEL domains may influence the stroke outcome and are worth investigating in a larger cohort of small vessel ischemic disease patients, different ischemic stroke subtypes and with functional studies.
Recently, genome-wide association studies (GWASs) identified Phosphatase and actin regulator 1 (PHACTR1) as a critical locus significantly associated to five different vascular disorders: coronary artery disease, migraine, fibromuscular dysplasia, cervical artery dissection and hypertension [1][2][3][4] , which represent frequent and significant comorbidities or risk factors linked to ischemic stroke, particularly with early-onset [5][6][7][8][9][10] . However, the potential pathogenic link between PHACTR1 genetic variability and other cerebrovascular disorders leading to ischemic stroke or critically influencing its outcome, such as small vessel ischemic disease (SVID) and brain collateral artery plasticity like posterior communicating arteries (PcomAs) recruitment, have not been investigated. Interestingly, a recent study reported 2 de novo PHACTR1 rare coding variants, mapping to the RPEL domains as causative factors for paediatric epileptic syndromes such as West syndrome and other neurodevelopmental disorders 11 . Therefore, suggesting that the PHACTR1 locus may harbour rare coding disease-modifying variant(s), that are likely to remain undetected in GWASs either because they are not targeted in the GWAS array or because, even when applying genotype imputation, the detection of very rare coding variants remains inaccurate.
To overcome this drawback, in the last 10 years, deep resequencing studies have been powerful strategies to effectively complement GWASs and unveil rare coding functional variants in the GWAS susceptibility loci [12][13][14][15] .
Finally, despite the critical importance of brain collateral arteries and, among these, PcomAs, their study in patients remains a major challenge, mostly due to their recruitment exclusively under moderate-severe acute hypoxic-ischemic conditions and their complex phenotype influenced by several additive factors (genetics, sex and aging) or comorbidities (diabetes and hypertension) and the absence of standardized methods for their accurate study [16][17][18] .
By contrast, C57BL/6J inbred mice, minimizing genetic and environmental variability, offer an ideal system to study PcomA plasticity, whose main differences have been already characterized [19][20][21] . Moreover, genome sequencing in different mouse strains shed light on the extensive similarities linking mouse and human genome and validated the importance of using mouse models to investigate the genetics of human diseases 22,23 . Therefore, to study the possible role of PHACTR1 genetic and mostly rare coding variability in SVID-CADASIL-like patients and PcomAs recruitment during acute and subacute ischemia in C57BL/6J mice, we performed exome sequencing in a multi-ethnic cohort of 180 early-onset independent familial and apparently sporadic SVID and CADASIL-like Caucasian patients from US, Portugal, Finland, Serbia and Turkey and used a combination of complementary techniques (genome and exome sequencing, T2 weighted magnetic resonance imaging [T2-MRI], arterial spin labelling cerebral blood flow [CBF], magnetic resonance angiography [MRA] and histology) in two experimental mouse models of cerebral ischemia (bilateral common carotid artery stenosis [BCCAS] and middle cerebral artery occlusion [MCAO]) (Fig. 1). The results from this approach do not support a critical role of PHACTR1 rare coding variants in SVID or CADASIL-like patients and PcomA patency during hypoxia-ischemia in mice.

Materials and methods
Patient cohort. The cohort was composed of 180 independent familial and early-onset apparently sporadic SVID and CADASIL-like Caucasian patients (96 and 84, respectively). Inclusion criteria for the SVID cohort included small vessel ischemic disease diagnosis based on TOAST classification, early age at onset (< 65 years [only two cases, whose age-at onset was 68 and 71 years old have been included in the study because familial and therefore enriched for genetic causative or risk factors]), absence of known pathogenic mutations in Mendelian small vessel disease genes (HTRA1, NOTCH3, ACTA2 and COL4A1) and no enrichment for vascular risk factors except for hypertension, which generally plays a critical role in elderly people 24 . The collection of samples included in this study has been approved by the ethics committee of the Faculty of Medicine, University of Coimbra, Coimbra, Portugal; the Ethics Committee of the School of Medicine, University of Belgrade (Serbia); the Ethics Committee of Istanbul Faculty of Medicine, Istanbul University and the Ethics Committee of the Hospital District of Southwest Finland. All NINDS Repository samples were collected only after an IRB-approved, signed informed consent was secured by the submitter. The diagnostic criteria for the CADASIL-like disease cohort were met by combining clinical symptoms, imaging data, and positive medical history in the absence of known NOTCH3 pathogenic mutations and based on the previous literature 25  www.nature.com/scientificreports/ in 12.2%, migraine in 10.0% and hypercholesterolemia in 7.2%. Given the prevalent role of hypertension and diabetes mellitus 2 in SVID in the elderly people 24 and the young age at onset of the cohort, these patients were considered enriched for genetic risk factors (Table 1). Finally, 478 controls > 60 years of age were selected from 'HEALTHY EXOMES' , HEX, a publicly available database, which collects exome sequencing data from elderly neuropathologically proven controls (https ://www.alzfo rum.org/exome s/hex) 26 .

BCCAS and MCAO mouse model, experimental design and exclusion criteria. Experiments
were approved by the Landesamt für Gesundheit und Soziales and conducted according to the German Animal Welfare Act and institutional guidelines. Thirty male C57BL/6J mice (purchased at 8 weeks of age, Charles River, Germany and 10 weeks of age Janvier France, respectively) were housed in a temperature (22 ± 2 °C), humidity (55 ± 10%), and light (12/12-h light/dark cycle) controlled environment. The animals underwent hypoperfusion between 9 and 13 weeks of age. Hypoperfusion was achieved by bilateral common carotid artery stenosis (BCCAS) (15 mice) or through 60 min transient blockage of left middle cerebral artery (MCA) and permanent occlusion of common carotid artery and external carotid artery (MCAO) (15 mice). Among these, 6 BCCAS and 9 MCAO mice with extreme phenotype were selected for the genetic study based on MRA, T2-MRI and CBF data 24 h post-surgery (Fig. 2).
The study was carried out in compliance with the ARRIVE guidelines.
Mouse cohort for the genetic screening. Our mouse cohort was composed of six BCCAS and nine MCAO mice with different PcomA patency phenotypes ( Table 2, Fig. 2). Selection of the mice was based on our previous and highly reproducible experiments in BCCAS and MCAO mice aimed at characterizing their vascular phenotype ( 27 and unpublished data). Inclusion criteria for this study were based on the brain acute hypoperfusion phase and particularly on MRA, T2-MRI scans and brain perfusion 24 h post-surgery (Fig. 2). The mouse phenotype was followed up and further characterized with additional MRI and MRA scans and histology at 1 week, 4 and 7 weeks.   28,29 and has been already described 27 . Briefly, this identifies 4 PcomA classes, based on the ratio between PcomA and basilar artery (BA) diameter: (1) PcomA < 10% of BA; (2) PcomA 11-20% of BA; (3) PcomA 21-30% of BA and (4) PcomA > 30% of BA. We identify class 1 and class 2 as 'non-patent' , class 3 as 'small' , class 4 as 'prominent' and included a fifth class, represented by PcomA > 60% of BA, described as 'very prominent' .
The diameters of the PcomAs were measured at the smallest point and the diameter of the BA was measured proximal to the superior cerebellar arteries both for the Evans Blue and fluorescent WGA stainings (MCAO mice)  www.nature.com/scientificreports/ or only with for Evans Blue staining (BCCAS mice) with ImageJ. The diameter of the PcomAs as a percentage of the diameter of the BA was calculated and used in the analysis as previously described 28,29 . In our mouse cohort, BCCAS mice with extreme dichotomous vascular phenotype were selected for genome sequencing: three BCCAS mice with two very prominent PcomAs (Fig. 2-BI) and 3 BCCAS mice with no patent PcomAs, which died between few hours and one day after surgery due to very severe hemispheric strokes (Fig. 2-BII). Three of these with the very effective vascular phenotype have been included in the exome sequencing cohort, together with nine MCAO mice, characterized by different left PcomA calibre: (a) three MCAO mice with prominent-very prominent left PcomA, that developed small ischemic lesions (≈ 5-10% of the left hemisphere), mostly affecting ventral areas (prefrontal cortex, striatum and ventral hippocampus), and were characterized by the most favourable stroke outcomes (Fig. 2-CI); (b) three MCAO mice with small PcomA, that survived the surgery but developed monolateral severe left strokes affecting up to 34% of the left hemisphere and including also dorsal areas (orbital cortex and cerebellum) (Fig. 2-CII) and (c) three MCAO mice with non-patent PcomA, which died between few hours and 2 days post surgery (Fig. 2-CIII) ( Table 2). Given the extreme inbreeding of the C57BL/6J strain, carefully inbred for over 70 years through more than 200 generations of brother-sister mating 30 , and the likely minimal influence of environmental factors, these mice were genetically considered as members of the same large multigenerational family coming from a small and isolated village. Moreover, the selection of extreme phenotypes (absent-small PcomA vs prominent-very prominent PcomA), allowed us to reach an high power for the detection of rare variants with large effect size, despite the small sample size 31,32 .

Statistical analysis and methods to prevent bias.
Power calculation was performed with R statmodpackage v1.4.32 for Fisher's exact test based on allelic association for the SVID cohort. The study had at least 80% power for the detection of common variants, MAF > 5%, with strong effect (OR < 0.6 or > 2), with a significance value of two-sided α = 0.05 (Fig. S1).
The joint effect of Phactr1 coding and non-coding variants on vascular traits was performed with two Sample t-test in R (version × 64 3.0.2, http://www.r-proje ct.org/). Low frequency and rare variants were defined as having a 1% < MAF < 5% and MAF < 1%, respectively, either in cases or controls. Minor allele frequency was based either on HEX database for elderly controls > 70 years of age or ExAC database version 0.3.1 database (http://exac.broad insti tute.org/).
Mice were randomized to receive hypoperfusion.
For the CNV analysis of the WGS data, Cnvkit in batch mode was used in a matched fashion as described in their manual for WGS data. www.nature.com/scientificreports/ Angiotool. MCAO vessel length and total vascular junctions were calculated for leptomeningeal microvessels, selecting always the same regions of interest (both in terms of brain area in both hemispheres and region of interest dimension) by using the software AngioTool v 0.6a as previously described 27 .
All methods were carried out in accordance with relevant guidelines and regulations.

Results
Exome sequencing in SVID-CADASIL-like patients. In our SVID-CADASIL like cohort 3/180 patients (1.7%) carried three rare heterozygous coding variants in PHACTR1 (p.Glu198Gln, p.Arg204Gly, p.Val251Leu) ( Table 3). Among these, one was novel (p.Arg204Gly) and two were very rare coding variants (MAF < 1 × 10e − 5) (p.Glu198Gln, p.Val251Leu). These have been predicted likely pathogenic (CADD > 15), cluster in well conserved protein domains (Fig. 3) and the carrier frequency in our cohort was higher, although not significantly (Fisher's exact test p-value < 0.05), when compared to 478 healthy neuropathologically proven controls (HEX database) (0.4%). However, these variants have been detected in exon 7 and 8 and, analogously to the PHACTR1 variants found in controls, between 2 different low complexity regions (https ://strin g-db.org/), outside the three RPEL regions or C terminal domain, actin and protein phosphatase 1 (PP1) binding sites, respectively, where all the PHACTR1 pathogenic mutations have been reported in West syndrome and other neurodevelopmental disorders 11,37 (Fig. 3). All the three PHACTR1 rare variant mutation carriers, one from US with SVID and two from Portugal with CADASIL-like disease displayed an early-age at onset (average age at onset = 56 range 52-59) ( Table 3). Moreover, none of the PHACTR1 missense mutation carriers presented coronary artery disease or migraine, two traits already significantly associated to PHACTR1 common non-coding variability 1,2 . One carrier from the US was affected by hypertension, nonetheless, given the high frequency of hypertensive patients in the US cohort (60.4%), we may probably exclude a direct or critical effect of PHACTR1 coding variability on hypertension.

Genome sequencing and exome sequencing in BCCAS and MCAO mice characterized by different PcomA patency. Phactr1 CNV in BCCAS.
A growing body of evidence reported copy number variants (CNVs) as main mechanism of genome evolution 38,39 . Considering that highly inbred strains like the C57BL/6J are not isogenic 30,40 and different vascular phenotypes such as the diverse degree of PcomA patency segregate within the strain, represent known phenotypic intrastrain differences 19,27 and may have been positively selected through the generations, we investigated the possibility that a CNV in Phactr1 may have determined the different vascular phenotype in C57BL/6J BCCAS mice. We recently described the main arterial collateral compensatory mechanisms in BCCAS 27 and these primarily involve the PcomA patency. Therefore, we selected BCCAS mice with opposite collateral plasticity phenotype: (1) very effective collateral vascularization: two very prominent PcomAs leading to no lesion (defined as gray or white matter hyperintensities) detectable on T2 weighted MRI either 2 days or 7 days post-surgery (three mice) (Fig. 2-BI) and (2) ineffective vascular phenotype: absence of PcomA collateral flow, leading to severe ischemic hemispheric lesions and to the death of mice between a few hours and 1 day post-surgery (three mice) (Fig. 2-BII) (Table 2).
We do not report any CNV in Phactr1, segregating with BCCAS with effective or not effective vascular traits (Fig. 4). Therefore, we may exclude that Phactr1 CNV may influence PcomAs patency. We next focused on point mutations as another likely genetic mechanism explaining intrastrain vascular differences and included in the study another stroke experimental mouse model (MCAO) with C57BL/6J background, characterized by left hemispheric strokes, whose severity is directly proportional to the left PcomA size 19,27 .
We performed exome sequencing in nine MCAO and three BCCAS mice with diverse PcomA calibre and investigated the hypothesis that PcomA spectrum size, ranging from no PcomA/non-patent PcomA to very prominent PcomAs may have been determined by coding missense mutations in Phactr1 (Fig. 2) (Table 4).

Single-variant study in BCCAS and MCAO mice.
In the 12 MCAO-BCCAS exomes, we identified 15 variants in Phactr1: 14 non-coding variants and one heterozygous stop-gain mutation (p.Gln273*). The most common noncoding variant detected was c.104-93T>C, carried by 8/12 (66.66%) mice. The maximum number of variants carried pro mouse was four non-coding variants, found in a BCCAS mouse (BCCAS_17) with a very effective vascular phenotype and, given the lack of genetic variants in Phactr1 in another BCCAS mouse (BCCAS_24) with prominent arterial collaterals, we may rule out a significant correlation between the enrichment for vari- Table 3. PHACTR1 coding variants detected in the SVID-CADASIL-like cohort and in the HEX database cohort of controls. CTRLS controls, Aa amino-acid, AAO age at onset, SVID small vessel ischemic disease, Neg. negative.    (Table 4). PHACTR1 p.Gln273* clusters in exon 5, outside the REPL regions or C terminus, reported as PHACTR1 functional domains 11,37 . Phactr1 p.Gln273* has been detected in one MCAO mouse characterized by a left non-patent PcomA, which developed a severe left hemispheric stroke and died few hours after the surgery (Fig. S2). This variant did not segregate with this specific vascular trait, as it was not detected in the other MCAO mice with non-patent/small left PacomA (2/3 mice, respectively) ( Table 4). Therefore, we may also rule out the possibility that coding mutations clustering outside Phactr1 functional domains may play a critical role in determining PcomAs recruitment or patency.

Phactr1 pooled variants in BCCAS and MCAO mice.
We then investigated the hypothesis that the synergic effect of Phactr1 coding and non-coding variants may have influenced the vascular phenotype in BCCAS-MCAO C57BL/6J mice. Particularly, we focused on a potential Phactr1 variant enrichment as determinant for different vascular traits such as (1) embryonic vasculogenesis and vessel density indirectly measured through CBF in naive mice pre-surgery 41 (Fig. 5A,B, Table S1); (2) PcomA size during brain ischemia (Fig. 5C, Table 4) and (3) leptomeningeal microcollateral length and junction density during subacute hypoperfusion (7d) (Fig. 5D-F, Tables S2 and S3). We do not report any significant association (t-test p-val < 0.05) between enrichment for Phactr1 variability in mice and severity of different vascular traits at different time points both when considering CBF pre-surgery in naive mice or macro and microcollaterals (PcomA size and leptomeningel microvessel average length and number of junctions) during brain hypoperfusion (Fig. 5).

Discussion
In this study we tested the hypothesis that genetic variability in PHACTR1 may have significantly influenced the risk for ischemic stroke, either increasing the susceptibility to small vessel ischemic and CADASIL-like diseases in patients or regulating the recruitment and plasticity of PcomAs in 2 C57BL/6J experimental stroke mouse models (BCCAS and MCAO).
In the SVID-CADASIL-like disease cohort, we report three very rare coding variants (p.Glu198Gln, p.Arg204Gly, p.Val251Leu) clustering outside the PHACTR1 known functional domains, RPEL and C-terminus domains, harbouring the binding sites for g-actin and PP1, respectively. These domains have been reported to be critical for PHACTR1 regulation of angiogenesis 42 and to harbour causative mutations for West syndrome and other neurodevelopmental disorders 11 . Mutations outside the RPEL and C-terminus domains have not been reported linked to any disorder or trait. Analogously to other proteins playing a key role in neurodegenerative disorders such as PSEN1, PSEN2, APP and CSF1R, whose pathogenic mutations cluster in definite functional domains (alpha-helix domain in the transmembrane domains, secretase domain and tyrosine kinase domains, respectively) 43,44 (www.molge ndata base), it is probable that only PHACTR1 rare coding variants in the RPEL or C-terminus domains may manifest phenotypically. www.nature.com/scientificreports/ Indeed, although PHACTR1 p.Glu198Gln, p.Arg204Gly and p.Val251Leu have been reported as likely pathogenic (CADD > 15) and are well conserved across different species, rare coding mutations in PHACTR1 low complexity regions have been detected also in elderly neuropathologically proven controls (Fig. 3, Table 3).
Moreover PHACTR1 p.Glu198Gln, p.Arg204Gly and p.Val251Leu carriers did not display migraine or coronary artery disease, both traits reported to be significantly associated to PHACTR1 1,2 .
Importantly, although we cannot exclude that rare coding variants in PHACTR1 functional domains may be causative or risk factors for ischemic stroke, cerebrovascular accidents have not been reported in West syndrome patients. However, given the rare incidence of the disorder (about one in 3000 children), the young age of the cases described (oldest patient reported = 28 years) 45 and the exclusive report of de novo mutations and not germline ones 11 , we may not rule out that (1) these patients affected by epileptic encephalopathy with very early-onset may not develop ischemic strokes later in life, (2) undiagnosed lacunar ischemic lesions may have triggered epileptic seizures in these patients and (3) unlike germline mutations, not reported in PHACTR1 functional domains, de novo mutations may not significantly compromise pre-existent effective arterial compensatory mechanisms and therefore lead to a less severe phenotype without cerebrovascular episodes.
The lack of a significant correlation between PHACTR1 genetic variability and ischemic stroke has been supported also by our experiments in brain ischemia mouse models (BCCAS and MCAO), where we do not report Phactr1 CNVs as responsible for PcomA patency. Analogously, we found a Phactr1 heterozygous stopgain mutation (p.Gln273*), clustering outside the RPEL or C-terminus domains in one MCAO mouse with left non-patent PcomA, developing severe hemispheric strokes and dying few hours after surgery. Importantly, no coding mutations in Phactr1 have been detected in other MCAO mice with similar vascular phenotype (left non-patent/small PcomA) and stroke outcome, suggesting that this variant does not segregate with a specific vascular trait and it is not likely to represent a pivotal factor determining PcomA patency.
We do not report any significant enrichment for Phactr1 coding and non-coding variants linked to particularly effective or non-effective vascular traits (vascular density pre-surgery, indirectly detected through CBF pre-surgery, PcomA size, microcollateral leptomeningeal density and average vessel length during subacute www.nature.com/scientificreports/ hypoperfusion in mice) (Fig. 5). Thus, implying that also the joint effect of Phactr1 variants may not result in a distinctive vascular phenotype. The strength of our study lies in the use of inbred mouse genetics to interpret and complement the more complex genetic variability of independent familial and apparently sporadic patients. The selection of inbred mice with opposite vascular phenotypes (3-6 mice with either prominent/very prominent PcomA or small/nonpatent PcomA), allowed us to reach, despite the small sample size, a relatively high power for the detection of rare coding functional variants. This enabled us to support the PHACTR1 findings in the SVID and CADASIL-like underpowered patient cohort.
In summary, our study shows that PHACTR1 rare coding variability outside the functional domains is unlikely to play a critical role in small vessel ischemic disease and CADASIL-like syndrome as well as brain collateral artery recruitment during brain hypoperfusion in mouse models. However, PHACTR1 coding variability in RPEL domains are worth investigating in a larger cohort of SVID patients, in different ischemic stroke subtypes and with functional studies.

Data availability
All data generated or analysed during this study are included in this published article (and its "Supplementary Information" files).