Rare Variant Association Analysis Uncovers Involvement ofVNN2in Stroke Outcome

A stroke’s functional outcome presents vast variability among patients, which is influenced by age, sex, characteristics of the lesion, and genetic factors. However, there is very little knowledge about stroke recovery genetics. R ecently, some GWAS (Genome -Wide Association Studies) have . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 19, 2024. ; https://doi.org/10.1101/2024.09.18.24313937doi: medRxiv preprint highlighted the involvement of common or low -frequency variants near or within PATJ, PPP1R21, PTCH1, NTN4 and TEK genes , whereas the role of rare variants is still unclear. This study aims to identify the genetic contributions to differences in stroke outcome analyzing the effect of rare variants.

We performed a pilot study analyzing 90 exomes of extreme good or bad recovery (modified Rankin Scale (mRS) at 3 months 0 -1 vs 4-5) to select target genes involved in stroke recovery. To expand this study, 702 additional samples were sequenced by Targeted Next -Generation Sequencing capturing loci selected from the pilot study, GWAS studies and literature input. Here, we performed continuous (mRS 0-6) and dichotomous (mRS 0 -1 vs 3-6) analyses, yielding one candidate gene. Protein structure and stability analysis were performed on selected variants.

Our work identified rare coding variants in VNN2 associated with patients with a better stroke recovery (∆DIC > 10, equivalent to p -value 1.6 kcal/mol), meanwhile, another variant, located in the active site, could affect the electrostatic surface.

VNN2 could play a role in post -stroke inflammation altering the cell adhesion and migration of neutrophils during recovery. KEY WORDS : ischemic stroke, genetics, rare variants, VNN2, inflammation, neutrophils, transendothelial migration. Nonstandard abbreviations and Acronyms - Damage-associated molecular patterns (DAMPs) . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 19, 2024. ; https://doi.org/10.1101/2024.09.18.24313937doi: medRxiv preprint - Matrix metallopeptidases (MMPs) - Blood brain barrier (BBB) - Ischemic stroke (IS) - Genome-wide Association Study (GWAS) - Next-Generation Sequencing (NGS) - Rare variant association tests (RVAS) - Whole exome sequencing (WES) - Cardioembolism (CE) - Large‐artery atherosclerosis (LAA) - Other determined (OD) - Genome Analysis Toolkit (GATK) - QualbyDepth (QD) - Quality (QUAL) - StrandOddsRatio (SOR) - FisherStrand (FS) - RMSMappingQuality (MQ) - MappingQualityRankSumTest (MQRankSum) - ReadPosRankSumTest (ReadPosRankSum) - Principal component analysis (PCA) - Glycosylphosphatidylinositol (GPI) 3. Introduction Worldwide, stroke is the second leading cause of death and adult disability. Approximately 1.1 million people in Europe suffer a stroke every year, and its incidence and prevalence are expected to increase along with the aging of the population1. The outcome of a stroke is affected by many factors including gender, age, stroke severity, size, and location. However, even adjusting for these factors, significant clinical heterogeneity remains. This heterogeneity can be partially explained by genetic vari ation affecting proteins involved in the stroke recovery processhttps://paperpile.com/c/bPJy0R/jEWX2. . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 19, 2024. ; https://doi.org/10.1101/2024.09.18.24313937doi: medRxiv preprint Stroke recovery is a complex process that involves neuronal, vascular and immune responses3. Initially, after the ischemic event, in response to the injury, necrotic and dying neurons release damage -associated molecular patterns (DAMPs). These activate an innate immune response, including g lial activation and infiltration of blood -borne immune cells into the brain. Microglia secrete matrix metallopeptidases (MMPs) that disrupt the integrity of the blood brain barrier (BBB), facilitating the invasion of macrophages and neutrophils. While this immune response is beneficial at stroke onset, if this pro -inflammatory response is extended, as it happens in aged individuals, it contributes to the participation of T -cells and a magnification of the immune response, and a worse stroke outcome 4. Once this first response has taken place, poststroke recovery processes lead to restoration or compensation of function. These are based on the induction of key biological processes such as angiogenesis, neurogenesis, a xonal sprouting, dendritic branching, synaptogenesis, and oligodendrogenesis3. An avenue to understand the relevance of all these processes is to identify key genes involved in stroke outcome, and the biological pathways that mediate the identified allele-outcome correlations. Only a few studies have investigated the role of genome -wide genetic variation in ischemic stroke (IS) outcome, focusing on the common (>1%) genetic variation captured by classic GWAS (i.e. genotyping arrays) 5-8. Studies looking at mid -term outcome (60 to 190 days) have identified two genome -wide significant associations, with variants near PATJ5, a gene related to tight junction formation and maintenance 9, and variants regulating PPP1R216, a gene involved in learning, memory and neuronal plasticity 10-12. They also found suggestive association with PTCH1, TEK, and NTN46. In addition, the analysis . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 19, 2024. ; https://doi.org/10.1101/2024.09.18.24313937doi: medRxiv preprint of the global effect of copy number variation, or genomic imbalance, on stroke outcome, yielded association of increased genomic imbalance with poorer stroke outcome 7. Finally, another GWAS has uncovered associations of excitotoxicity related genes with short-term (24h) stroke outcome8, which is also correlated with 90 -days outcome 13. Meanwhile, the potential effect of rare variants, tackled by exome or whole genome sequencing, remains unexplored2. Importantly, there are few to none widely accepted neuroprotective or neuroreparative drugs, nor p ersonalized approaches to guide therapies to mitigate ischemic brain injury or enhance recovery 2. Therefore, advances in the identification of novel outcome related variants and genes, coupled with the understanding of the pathways through which these genes affect stroke outcome, should provide critical knowledge for drug selection and allow for personalized therapy approaches. The improvement of Next -Generation Sequencing (NGS) methods has allowed the study of both common and rare variants in complex diseases. Some studies have shown that rare variants could have higher impact on the structure, stability, or function of proteins than common variants 14,15, explaining part of the heritability of complex traits 16. However, association tests of individual rare variants require very large sample sizes, which are difficult to obtain for very specific phenotypes, such as a stroke’s functional outcome. To overcome this limitation, various rare variant association studies (RVAS) aggregate the effects of variants affecting the same biological entity (e.g. genes), and test for association of the aggregated variants with the phenotype. Other RVAS go further and consider more complex scenarios, such as heterogeneity of the . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 19, 2024. ; https://doi.org/10.1101/2024.09.18.24313937doi: medRxiv preprint variant’s effects (SKAT, SKAT -O), or variant -specific characteristics (BATI), providing increased statistical power17-19. In the present analysis we explore the effect of rare variants on stroke outcome by applying BATI, which also integrates patients - and variant -specific characteristics as covariates. We per formed a two -phase study, including exome sequencing in a small (n=90) but carefully curated cohort of extreme outcome phenotypes to select genomic regions that were further explored in a targeted sequencing analysis in 702 additional stroke cases. One gen e, VNN2, harbored an excess of rare variants that potentially affect protein function in individuals with better outcome scores. 4. Methods Study Design The study was divided into two phases: a first approach through a whole exome sequencing (WES) pilot study , and a follow -up by targeted resequencing analysis. The WES pilot study was performed on a selection of 90 cases matched by age, gender, and stroke type and location, divided into good (mRS 0-1) or poor (mRS 3 -5) functional outcome. A second phase involve d targeted resequencing analysis of 702 cases with a wider range of outcomes (mRS 0 -6). Selected regions for resequencing included the top genes identified in the pilot WES study together with genes and regions around identified GWAS hits, as well as a few candidate genes from the literature (Supplementary bed file). Study Subjects European ancestry patients with a diagnosis of IS according to World Health Organization criteria and available DNA samples were selected from the . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 19, 2024. ; https://doi.org/10.1101/2024.09.18.24313937doi: medRxiv preprint Spanish Stroke Genetics Conso rtium (GeneStroke) cohorts. The Pilot study included 69 patients from the BASICMAR study 20 and 21 patients recruited at hospital Vall d’Hebron for the GRECOS 21 and Geno -tPA22 studies. These patients were distributed in two subgroups based on their mRS sco res at three months (0-1, “good outcome”, 49 patients vs 3-4-5 “poor outcome”, 41 patients), which were matched by age, gender, stroke location and type (Table 1). Inclusion and exclusion criteria are detailed in Supplementary methods. The targeted resequencing phase included 702 patients from the same cohorts (441 samples from BASICMAR study and 261 samples from GRECOS and Geno - tPA studies; Table 2 and Figure S1), fulfilling the same inclusion and exclusion criteria, except for the patients with unusual st roke (“other determined etiology” in TOAST). Exome sequencing data from the pilot stud y is deposited at the European Genome Archive (EGA dataset ID: EGAD00001004808). Targeted resequencing data is available upon request. The study was approved by the Institutional Review Boards of the participant hospitals (CEIm -PSMAR (2008/3083/I); IRB00002850 (PR(AG)157/2011)) and the CEIC of Fundació Sant Joan de Déu (C.I: PIC -82-17), and all participants provided written informed consent to participate: The research was conducted in accordance with the Helsinki declaration. Sequencing For the pilot study, DNA obtained from peripheral blood lymphocytes was fragmented using a covaris system . Libraries were generated with TruSeqTM DNA Sample Preparation v2 Kits (Illumina, Inc., San Diego, CA, USA) followed by exome capture with NimbleGen SeqCap EZ Library SR v3.0 (Roche, Inc., Madison, WI, USA) in pools of 5 samples, which were sequenced to a 30-40x . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 19, 2024. ; https://doi.org/10.1101/2024.09.18.24313937doi: medRxiv preprint coverage on an Illumina HiSeq2500 at the Spanish national center for genomic analysis (CNAG). For the targeted phase study, libraries were prepared from DNA from peripheral blood lymphocytes and captured with a custom agilent sureselect capture kit , using an in -house sample preparation system. Libraries were pooled and sequenced on Illumina Hiseq 2500 at the CRG -CNAG, to a targeted coverage of 150x. Bioinformatic Analysis FASTQ raw sequences were processed following an in -house pipeline (Figure S2) including quality control with FASTQC, alignment with BWA -mem (http://bio-bwa.sourceforge.net/), duplicate marking with Picard (https://broadinstitute.github.io/picard/), samtools ( http://www.htslib.org/) processing, local realignment base recalibration and variant calling with the Genome Analysis Toolkit (GATK; https://software.broadinstitute.org/gatk/). Variants were called by HaplotypeCaller using standard parameters except for min-pruning, which was set at 5 in the targeted analysis. Genome build 37 (GRCh37/hg19) was used as the reference genom e. For exome sequencing, variants were further filtered with VQSR, following standard recommendations, while for the targeted resequencing, the resulting variants were hard -filtered following GATK recommendations (https://gatk.broadinstitute.org/) [QualbyDepth (QD) < 2.0; Quality (QUAL) 3.0; FisherStrand (FS) > 60.0; RMSMappingQuality (MQ) < 40.0; MappingQualityRankSumTest (MQRankSum) < -12.5; ReadPosRankSumTest (ReadPosRankSum) < -8.0]. Multiallelic variants were filtered ou t with BCFtools . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 19, 2024. ; https://doi.org/10.1101/2024.09.18.24313937doi: medRxiv preprint (http://www.htslib.org/). The filtered VCF files were annotated with ANNOVAR (https://annovar.openbioinformatics.org/). Quality Control In the pilot phase, 91 samples passed all qua lity criteria. In the targeted phase, three samples were removed based on inclusion and exclusion criteria, and another five were discarded because of missing information. Quality control on the pilot phase samples and the remaining 694 samples of the targ eted phase was performed on the rvGWAS framework (available at https://github.com/hanasusak/rvGWAS). In both phases, variants with a genotyping call rate 95% called variants, and outlier samples based on principal component analysis (PCA) and the number of called variants (Figure S3). This resulted in the removal of 13 samples in the targeted analysis, leaving a total of 681 samples for analysis. Finally, vari ants in top selected genes were manually curated by revision of variant calling files in IGV v. 2.8.0. rvGWAS framework For downstream statistical analysis, only rare (European Allele Frequency 20) exonic and splice-site variants were considered in both the pilot and the follow -up phase. In the pilot phase, association with the 3 -month mRS was considered under a dichotomous model only (outcome mRS 3m 0-1 vs 3-5, analyzed with BATI and SKAT-O). In the follo w-up, we considered two models, a continuous model (outcome mRS 3m 0 -6, all 681 samples) and a dichotomous model (outcome mRS 3m 0-1, 219 samples, vs 3-6, 336 samples, totaling 555 samples). In both . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 19, 2024. ; https://doi.org/10.1101/2024.09.18.24313937doi: medRxiv preprint cases, analysis was performed with the Burden test, SKAT-O, KBAC, and BATI. Analyses were adjusted for age, sex, TOAST, and initial NIHSS. For the BATI test, variant effect (classified as loss-of-function or missense) was included as a covariate. An empirical significance threshold for the follow -up test (ΔDIC) was obtained performing 1000 simulations randomizing the 3 -month mRS score. In each simulation, genes were ranked by their ΔDIC values, and the highest score was extracted, selecting the threshold at 0.1% significance level (ΔDIC0.001= 7.268). The default threshold for 0.1% of significance is ΔDICdefault > 1019. Protein Structure and Stability Analysis Structural prediction for VNN2 was obtained from AlphaFold2 protein structure database prediction (European Molecular Biology Laboratory –European Bioinformatics Institute [EMBL -EBI], Hinxton, UK; https://www.alphafold.ebi.ac.uk/)24. The University of California, San Francisco (UCSF) Chimera program 25 and the back-bone dependent rotamer library we re used for structural interpretation and visualization 26. Other tools used include TOPCONS27, PROSITE, InterPro and PFAM (see supplementary material). The effect of variants on protein stability was calculated using FoldX (http://foldxsuite.crg.eu/)28. The structures were optimized to the FoldX force field command from the structure of VNN2 protein. The ∆∆G values were estimated as the difference between the energy of the wild -type and protein mutation (five replicates for each point variation). Values above 1,6 kcal/mol (twice the standard devi ation) were considered to significantly destabilize the protein29,30. . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 19, 2024. ; https://doi.org/10.1101/2024.09.18.24313937doi: medRxiv preprint 5. Results Pilot study The objective of the pilot study was to select approximately 100 candidate genes, potentially e nriched in rare variants in either dependent or independent patients, to be followed up in a larger cohort by targeted resequencing. The final list of targeted regions which included regions selected based on the BATI and SKAT-O results and additional regions based on the literature, are provided in Table S1. Targeted NGS study highlights VNN2 as a candidate gene The analysis under the continuous outcome model did not yield any gene with significant enrichment of rare coding variants (Table S 2 and S 3). However, TMPRSS7, the top gene in the BATI analysis, showed a ΔDIC value of 6.9, close to the 1% empirical significance threshold (ΔDICempirical = 7.268). In the analysis under the dichotomous model, we compared patients with a 3 - month mRS of 0 or 1 wit h those with a 3 -month mRS score over 3, using the following test and corresponding thresholds: Burden, SKAT-O, KBAC (p-value 10) and BATI (ΔDIC 0.01 < 7.268). Analysis with BATI showed one gene, VNN2, to be significantly enric hed (ΔDICVNN2 < 10.87; ΔDICempirical = 7.268; ΔDIC default = 10) for rare variants in patients with better outcome (Table 3). While not significant, VNN2 was also among the top genes identified in all other tests (Tables S4, S5 and S6). . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 19, 2024. ; https://doi.org/10.1101/2024.09.18.24313937doi: medRxiv preprint VNN2 variants identified in IS patients affect protein stability Six VNN2 variants were identified in the follow -up cohort, five of which were identified in six patients with good recovery, and one in one case with poor recovery. In addition, in the pilot study we had identified an additional two variants in cases with good rec overy, while no variants were present in patients with poor recovery. We explored the potential effect of the identified VNN2 variants in silico (Table 4), and all of them were predicted to have a significant effect on the protein, either by affecting its stability, by altering its electrostatic surface or by truncating the protein. Six out of eight variants are located in the CN hydrolase domain (Figure 1A and 1B). Two variants, p.(Ser46Phe) and p.(Leu53Pro), are in the first alpha helix. Both substitutions are expected to have a clear structural impact on the helix. In one case, by replacing a small proline for a bulky, aliphatic leucine; in the other, a very large, aromatic, non -polar amino acid (Phe) for a tiny, polar one (Ser). This substitution (p.(Ser46Phe)) is predicted to create a steric effect that would compromise its interaction with the Aspartic residue in position 49 (Figure 1C). On the other hand, variants p.(Val174Met) and p.(Arg205Ser), located in the beta-pleated sheets A and B, are predict ed to affect the interaction of their corresponding residues with the threonine residue at 198 and the phenylalanine at 199, respectively, destabilizing the region between these beta sheets (Figure 1D). Variant p.(Ala265Thr) is also predicted to cause a st eric shift destabilizing the protein (Figure 1F), as it causes the substitution of a larger uncharged polar residue (Thr) for a small nonpolar aliphatic residue (Ala) in an internal . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 19, 2024. ; https://doi.org/10.1101/2024.09.18.24313937doi: medRxiv preprint hydrophobic region of the protein close to the active site. Finally, p.(Gly284Cys), the last variant in the CN -hydrolase domain, located at the end of the beta B fold sheet, would cause an effect on the adjacent loop affecting its mobility (Figure 1G). The remaining two variants affect the same amino acid in the Vanin C domain. These variants are located in the exon 5 and may trigger nonsense -mediated decay (NMD). Even if NMD does not occur, the variant would still remove two thirds of this domain as well as the propeptide, including the loss of the GPI anchor. Moreover, the vari ant p.(Arg393Gln) is predicted to cause a change in the surface electrostatic charge of the protein. It is located at the access of the substrate to the active site, in a stretch of 4 consecutive arginines, conserved between species, that confer a positive charge to this region, which is affected by the substitution of an uncharged glutamine for the electropositive arginine (Figure 1E). 6. Discussion After a stroke, the interaction of different environmental and genetic factors may define the functional outcome of the cerebrovascular accident. Although GWAS studies have successfully identified common variants involved in stroke recovery, rare variants have not been explored yet. Here, for the first time, we performed an association study for rare variants involved in stroke recovery. This analysis involved a targeted analysis of 100 genomic regions in 702 patients, considering both a continuous (3 -month mRS 0-6) and dichotomous (3 -month mRS 0 -1 vs 4 -5) outcome variable model. Analysis was performed using SKAT -O and the BATI rare variant association . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 19, 2024. ; https://doi.org/10.1101/2024.09.18.24313937doi: medRxiv preprint test, which allows the integration of patient - and variant -specific features as covariates 19, and was proposed to have an improved power for the identification of the risk genes, especially in architectures with high genetic heterogeneity. BATI analysis yielded one gene, VNN2, significantly (ΔDIC > 10, equivalent to p -value < 0.001) associated with functional independence at 3 months. VNN2 (Vanin-2)31 encodes a glycosylphosphatidylino sitol (GPI) -anchored extracellular protein also known as GPI -8032. VNN2 has been identified to play a role in the cellular adhesion and transmigration processes of human neutrophil extravasation 32. While the p recise role of VNN2 in this multi -step process remains to be defined, it seems to take place in the transition from rolling to firm adhesion32-34. On the other hand, VNN2, together with VNN1 and VNN3, belong to the vanin protein family, characterized by their pantetheinase activity. Pantetheinase hydrolyzes pantetheine to pantothenic acid (vitamin B 5) and cysteamine, activating the stress pathway and inflammation. While thi s activity is stronger in VNN1, VNN2 also has this capability through its CN-Hydrolase domain35,36. We have identified eight rare variants in VNN2, seven of them present in patients with a good stroke outcome, while the remaining variant was present in a patient with poor outcome. In silico protein structure analysis predicted an effect in protein stability for the six missense variants located in the CN - Hydrolase domain (Table 4), indicating a potential effect of these variants through the modification of VNN2’s enzymatic activity. While th e remaining missense variant, (p.(Arg393Gln)), is located in the vanin domain, and was not . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 19, 2024. ; https://doi.org/10.1101/2024.09.18.24313937doi: medRxiv preprint predicted to affect protein stability, it was predicted to affect the electrostatic surface charge, potentially limiting the access to the nearby active site. The remaining variant, p.(R393*), would lead to a truncated protein sequence lacking the propeptide sequence and the GPI anchor site. However, given the location of the nonsense mutation, it would be expected to lead to NMD and absence of protein. Therefore, all the heterozygous identified variants could lead to a reduction in VNN2 function, which could be associated with a lower inflammatory response. The role of inflammation and immune mechanisms in the stroke recovery process is being increasingly recognized. Neutrophil extravasation and infiltration during inflammation are major contributors to poor ischemic outcomes37. Neutrophil interaction with endothelial adhesion molecules, a process in which VNN2 could be involved, together with the ischemic environment, is reported to shift the neutrophil phenotype from the protective N2 to more damaging N1 phenotype37. It is possible that the rare VNN2 variants identified in these individuals could affect its function in the neutrophil extravasation process, or even the shift from N2 to N1 neutrophils and t herefore reduce the infiltration of neutrophils into the brain parenchyma 37. Alternatively, the variants could lead to a reduced pantetheinase activity, conferring higher resistance to oxidative stress and leading to reduced levels of inflammation. To summarize, we have identified an excess of rare variants, predicted to affect protein stability or function, in stroke patients with a low mRs score at the 3 - month evaluation, classified as a good functional outcome. We hypothesize that . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 19, 2024. ; https://doi.org/10.1101/2024.09.18.24313937doi: medRxiv preprint the loss or reduction of VNN2 activity co uld lead to lower inflammation levels. It has been shown that while this inflammation process is a necessary event, excessive inflammation can be detrimental in the recovery process 4. A lower neutrophil extravasat ion or protection against oxidative stress caused by reduced VNN2 activity could lead to a reduced inflammatory response and allow for a better outcome. In conclusion, we present the first study of rare variants involved in stroke recovery, highlighting a possible relationship between stroke outcome and rare variants in VNN2. This protein could act on oxidative stress response, or cell adhesion and migration of neutrophils contributing to a good outcome after stroke. Therefore, VNN2 might be a novel therapeutic target for stroke recovery. Nonetheless, to expand our knowledge of the rare variant architecture of IS recovery, a cohort amplification is required to improve the detection of rare variants, the identification of novel genes, as well as increasing c aptured regions with WES data. These insights would help to improve statistical power in both the continuous and dichotomous models. 7. Acknowledgments We thank all the patients who suffered an ischemic stroke and allowed us to use their data in the study. 8. Sources of Funding This research was supported by Fundació Marató TV3 through the GENIUS study (307/U/2017), the GODs study (776/C/2011) and the Epigenesis study (188/C/2017); by the Fondo de Investigaciones Sanitarias -Instituto de Salud Carlos III grant PI21/00890 and RICORS Ictus network (RD21/0006/0004, . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 19, 2024. ; https://doi.org/10.1101/2024.09.18.24313937doi: medRxiv preprint RD21/0006/0007) supported with Next Gene ration funds; and by grant PID2022-141461OB-I00 funded by Spanish Ministerio de Ciencia, Innovación y Universidades MICIU/AEI/10.13039/501100011033 and by ERDF/EU, and by AGAUR-2021SGR-1093 and AGAUR -2021SGR-0656 from Generalitat de Catalunya. EA -C is supported by gra nt FPU2021/ 01324 from MICIU/AEI/10.13039/501100011033. 9. Disclosures None. 10. References 1. Bejot Y, Bailly H, Durier J, Giroud M. Epidemiology of stroke in Europe and trends for the 21st century. Presse Med . 2016;45:e391 -e398. doi: 10.1016/j.lpm.2016.10.003 2. Lee JM, Fernandez -Cadenas I, Lindgren AG. Using Human Genetics to Understand Mechanisms in Ischemic Stroke Outcome: From Early Brain Injury to Long -Term Recovery. Stroke. 2021;52:3013 -3024. doi: 10.1161/STROKEAHA.121.032622 3. Ma Y, Yang S, He Q, Zhang D, Chang J. The Role of Immune Cells in Post-Stroke Angiogenesis and Neuronal Remodeling: The Known and the Unknown. Front Immunol . 2021;12:784098. doi: 10.3389/fimmu.2021.784098 4. Finger CE, Moreno -Gonzalez I, Gutierrez A, Moruno -Manchon JF, McCullough LD. Age -related immune alterations and cerebrovascular inflammation. Mol Psychiatry . 2022; 27:803-818. doi: 10.1038/s41380 - 021-01361-1 . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 19, 2024. ; https://doi.org/10.1101/2024.09.18.24313937doi: medRxiv preprint 5. Mola-Caminal M, Carrera C, Soriano -Tarraga C, Giralt -Steinhauer E, Diaz-Navarro RM, Tur S, Jimenez C, Medina -Dols A, Cullell N, Torres - Aguila NP, et al. PATJ Low Frequency Variants Are Associated With Worse Is chemic Stroke Functional Outcome. Circ Res . 2019;124:114 - 120. doi: 10.1161/CIRCRESAHA.118.313533 6. Soderholm M, Pedersen A, Lorentzen E, Stanne TM, Bevan S, Olsson M, Cole JW, Fernandez -Cadenas I, Hankey GJ, Jimenez -Conde J, et al. Genome-wide association meta-analysis of functional outcome after ischemic stroke. Neurology. 2019;92:e1271 -e1283. doi: 10.1212/WNL.0000000000007138 7. Pfeiffer D, Chen B, Schlicht K, Ginsbach P, Abboud S, Bersano A, Bevan S, Brandt T, Caso V, Debette S, et al. Genetic Imbalance Is Associated With Functional Outcome After Ischemic Stroke. Stroke. 2019;50:298 - 304. doi: 10.1161/STROKEAHA.118.021856 8. Ibanez L, Heitsch L, Carrera C, Farias FHG, Del Aguila JL, Dhar R, Budde J, Bergmann K, Bradley J, Harari O, et al. Multi -ancestry GWAS reveals excitotoxicity associated with outcome after ischaemic stroke. Brain. 2022;145:2394-2406. doi: 10.1093/brain/awac080 9. Medina-Dols A, Canellas G, Capo T, Sole M, Mola -Caminal M, Cullell N, Jaume M, Nadal -Salas L, Llinas J, Gomez L, et al. Role of PATJ in stroke prognosis by modulating endothelial to mesenchymal transition through the Hippo/Notch/PI3K axis. Cell Death Discov . 2024;10:85. doi: 10.1038/s41420-024-01857-z 10. Giese KP, Mizuno K. The roles of protein kinases in learning and memory. Learn Mem. 2013;20:540-552. doi: 10.1101/lm.028449.112 . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 19, 2024. ; https://doi.org/10.1101/2024.09.18.24313937doi: medRxiv preprint 11. Shioda N, Fukunaga K. Physiological and Pathological Roles of CaMKII - PP1 Signaling in the Brain. Int J Mol Sci . 2017;19. doi: 10.3390/ijms19010020 12. Heroes E, Lesage B, Gornemann J, Beullens M, V an Meervelt L, Bollen M. The PP1 binding code: a molecular -lego strategy that governs specificity. FEBS J . 2013;280:584 -595. doi: 10.1111/j.1742 - 4658.2012.08547.x 13. Heitsch L, Ibanez L, Carrera C, Binkley MM, Strbian D, Tatlisumak T, Bustamante A, Ribo M, Molina C, Davalos A, et al. Early Neurological Change After Ischemic Stroke Is Associated With 90 -Day Outcome. Stroke. 2021;52:132-141. doi: 10.1161/STROKEAHA.119.028687 14. Tennessen JA, Bigham AW, O'Connor TD, Fu W, Kenny EE, Gravel S, McGee S, Do R, Liu X, Jun G, et al. Evolution and functional impact of rare coding variation from deep sequencing of human exomes. Science. 2012;337:64-69. doi: 10.1126/science.1219240 15. Nelson MR, Wegmann D, Ehm MG, Kessner D, St Jean P, Verzilli C, Shen J, Tang Z, Bacanu SA, Fraser D, et al. An abundance of rare functional variants in 202 drug target genes sequenced in 14,002 people. Science. 2012;337:100-104. doi: 10.1126/science.1217876 16. Wainschtein P, Jain D, Zheng Z, Group TOAW, Consortium NT -OfPM, Cupples LA, Shadyab AH, McKnight B, Shoemaker BM, Mitchell BD, et al. Assessing the contribution of rare variants to complex trait heritability from whole -genome sequence data. Nat Genet. 2022;54:263 -273. doi: 10.1038/s41588-021-00997-7 . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 19, 2024. ; https://doi.org/10.1101/2024.09.18.24313937doi: medRxiv preprint 17. Wu MC, Lee S, Cai T, Li Y, Boehnke M, Lin X. Rare -variant association testing for sequencing data with the sequence kernel association test. Am J Hum Genet. 2011;89:82-93. doi: 10.1016/j.ajhg.2011.05.029 18. Lee S, Wu MC, Lin X. Optimal tests for rare variant effects in sequencing association studies. Biostatistics. 2012;13:762 -775. doi: 10.1093/biostatistics/kxs014 19. Susak H, Serra-Saurina L, Demidov G, Rabionet R, Domenech L, Bosio M, Muyas F, Estivill X, Escaramis G, Ossowski S. Efficient and flexible Integration of variant characteristics in rare variant association studies using integrated nested Laplace approximation. PLoS Comput Biol . 2021;17:e1007784. doi: 10.1371/journal.pcbi.1007784 20. Roquer J, Rodriguez -Campello A, Gomis M, Jimenez -Conde J, Cuadrado-Godia E, Vivanco R, Giralt E, Sepulveda M, Pont -Sunyer C, Cucurella G, et al. Acute stroke unit care and early neurological deterioration in ischemic stroke. J Neurol . 20 08;255:1012-1017. doi: 10.1007/s00415-008-0820-z 21. Fernandez-Cadenas I, Mendioroz M, Giralt D, Nafria C, Garcia E, Carrera C, Gallego -Fabrega C, Domingues -Montanari S, Delgado P, Ribo M, et al. GRECOS Project (Genotyping Recurrence Risk of Stroke): The U se of Genetics to Predict the Vascular Recurrence After Stroke. Stroke. 2017;48:1147-1153. doi: 10.1161/STROKEAHA.116.014322 22. Carrera C, Carcel -Marquez J, Cullell N, Torres -Aguila N, Muino E, Castillo J, Sobrino T, Campos F, Rodriguez -Castro E, Llucia -Carol L, et al. Single nucleotide variations in ZBTB46 are associated with post - . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 19, 2024. ; https://doi.org/10.1101/2024.09.18.24313937doi: medRxiv preprint thrombolytic parenchymal haematoma. Brain. 2021;144:2416-2426. doi: 10.1093/brain/awab090 23. Rentzsch P, Witten D, Cooper GM, Shendure J, Kircher M. CADD: predicting the deleteriousness of variants throughout the human genome. Nucleic Acids Res. 2019;47:D886-D894. doi: 10.1093/nar/gky1016 24. Jumper J, Evans R, Pritzel A, Green T, Figurnov M, Ronneberger O, Tunyasuvunakool K, Bates R, Zidek A, Potapenko A, et al. Highly accurate protein structure prediction with AlphaFold. Nature. 2021;596:583-589. doi: 10.1038/s41586-021-03819-2 25. Pettersen EF, Goddard TD, Huang CC, Couch GS, Greenblatt DM, Meng EC, Ferrin TE. UCSF Chimera --a visualization system for exploratory research and a nalysis. J Comput Chem . 2004;25:1605 -1612. doi: 10.1002/jcc.20084 26. Shapovalov MV, Dunbrack RL, Jr. A smoothed backbone -dependent rotamer library for proteins derived from adaptive kernel density estimates and regressions. Structure. 2011;19:844 -858. doi : 10.1016/j.str.2011.03.019 27. Tsirigos KD, Peters C, Shu N, Kall L, Elofsson A. The TOPCONS web server for consensus prediction of membrane protein topology and signal peptides. Nucleic Acids Res . 2015;43:W401 -407. doi: 10.1093/nar/gkv485 28. Schymkowitz J, Borg J, Stricher F, Nys R, Rousseau F, Serrano L. The FoldX web server: an online force field. Nucleic Acids Res . 2005;33:W382-388. doi: 10.1093/nar/gki387 . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 19, 2024. ; https://doi.org/10.1101/2024.09.18.24313937doi: medRxiv preprint 29. Bruque CD, Delea M, Fernandez CS, Orza JV, Taboas M, Buzzalino N, Espeche LD, Solari A, Lucc erini V, Alba L, et al. Structure-based activity prediction of CYP21A2 stability variants: A survey of available gene variations. Sci Rep. 2016;6:39082. doi: 10.1038/srep39082 30. Buonfiglio PI, Bruque CD, Lotersztein V, Luce L, Giliberto F, Menazzi S, Francipane L, Paoli B, Goldschmidt E, Elgoyhen AB, et al. Predicting pathogenicity for novel hearing loss mutations based on genetic and protein structure approaches. Sci Rep . 2022;12:301. doi: 10.1038/s41598-021-04081-2 31. Galland F, Malergue F, Bazin H, Mattei MG, Aurrand-Lions M, Theillet C, Naquet P. Two human genes related to murine vanin -1 are located on the long arm of human chromosome 6. Genomics. 1998;53:203-213. doi: 10.1006/geno.1998.5481 32. Suzuki K, Watanabe T, Sakurai S, Ohtake K, Kinoshita T, Araki A, Fujita T, Takei H, Takeda Y, Sato Y, et al. A novel glycosylphosphatidyl inositol-anchored protein on human leukocytes: a possible role for regulation of neutrophil adherence and migration. J Immunol. 1999;162:4277-4284. 33. Watanabe T, Sendo F. Physical association of beta 2 integrin with GPI - 80, a novel glycosylphosphatidylinositol -anchored protein with potential for regulating adhesion and migration. Biochem Biophys Res Commun . 2002;294:692-694. doi: 10.1016/S0006-291X(02)00538-7 34. Yoshitake H, Takeda Y, Nitto T, Sendo F. Cross -linking of GPI -80, a possible regulatory molecule of cell adhesion, induces up -regulation of . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 19, 2024. ; https://doi.org/10.1101/2024.09.18.24313937doi: medRxiv preprint CD11b/CD18 expression on neutrophil surfaces and shedding of L - selectin. J Leukoc Biol. 2002;71:205-211. 35. Martin F, Malergue F, Pitari G, Philippe JM, Philips S, Chabret C, Granjeaud S, Mattei MG, Mungall AJ, Naquet P, et al. Vanin genes are clustered (human 6q22 -24 and mouse 10A2B1) and encode isoforms of pantetheinase ectoenzymes. Immunogenetics. 2001;53:296 -306. doi: 10.1007/s002510100327 36. Nitto T, Onodera K. Linkage between coenzyme a metabolism and inflammation: roles of pantetheinase. J Pharmacol Sci . 2013;123:1 -8. doi: 10.1254/jphs.13r01cp 37. Bui TA, Jickling G C, Winship IR. Neutrophil dynamics and inflammaging in acute ischemic stroke: A transcriptomic review. Front Aging Neurosci . 2022;14:1041333. doi: 10.3389/fnagi.2022.1041333 . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 19, 2024. ; https://doi.org/10.1101/2024.09.18.24313937doi: medRxiv preprint 11. Tables Table 1. Characteristics of the pilot phase samples. mRS (3 months) Sex (female/male) Age (range) Age (mean) LAA CE OD Undetermined Incomplete 0/1 24/25 53-87 73.26 12 22 0 13 2 3/4/5 20/21 51-90 74.4 11 20 0 6 3 LAA, large ‐artery atherosclerosis; CE, cardioembolism; OD, other determined; Undetermined, two or more causes; Incomplete, incomplete evaluation. . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 19, 2024. ; https://doi.org/10.1101/2024.09.18.24313937doi: medRxiv preprint Table 2. Characteristics of the samples from the follow-up cohort. Characteristics mRs at 3 months 0 1 2 3 4 5 6 Sex (female/male) 41/47 61/70 57/69 63/58 49/43 20/5 54/44 Age (range) 28-92 39-94 40-98 26-100 36-93 41-93 31-99 Age (mean) 70 72 74 76 77 77 81 LAA 13 23 29 23 18 4 9 CE 42 65 73 70 48 15 70 OD 18 31 19 17 11 2 8 Undetermined 3 4 2 5 9 1 10 Incomplete 10 7 3 6 6 2 1 NA 2 1 0 0 0 1 0 Total 88 131 126 121 92 25 98 LAA, large ‐artery atherosclerosis; CE, cardioembolism; OD, other determined; Undetermined, two or more causes; Incomplete, incomplete evaluation ; NA, not available. . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 19, 2024. ; https://doi.org/10.1101/2024.09.18.24313937doi: medRxiv preprint Table 3. Top genes in the rare variant association study (dichotomous model) according to BATI . The results were performed in a dichotomous model using mRs at 3 months 0-1 vs 3-6. Gene Total variants Variants in poor Variants in good Carriers in poor Carriers in good DIC VNN2 6 1 5 1 6 10.87* RASAL1 9 6 5 6 7 2.888 ECE2;EEF1AK MT4-ECE2 11 4 7 4 6 2.196 FGD4 8 2 6 2 6 2.156 NEK10 9 7 2 10 2 1.981 *ΔDICdefault > 10 equivalent to pval 7.268 equivalent to pval < 0.001. Table 4. VNN2 protein evaluation. CADD 1.3 30 Protein Variation Domain Protein Effect Protein Stability ΔΔG values CADD (v 1.3) p.(S46F) CN hydrolase Stability 1.99 ± 0.09 23.6 p.(L53P) CN hydrolase Stability 3.65 ± 0.03 25.8 p.(V174M) CN hydrolase Stability 1.7 ± 0.27 27 p.(R205S) CN hydrolase Stability 2.72 ± 0.09 22.7 p.(A265T) CN hydrolase Stability 5.9 ± 0.09 25 p.(G284C) CN hydrolase Stability 1.96 ± 0.30 23 p.(R393Q) Vanin C Electrostatic Surface -0.18 ± 0.06 20.8 p.(R393*) Vanin C NMD; GPI Anchor - 34 . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 19, 2024. ; https://doi.org/10.1101/2024.09.18.24313937doi: medRxiv preprint 12. Figure Legends Figure 1. VNN2 protein structure and rare variants identified by WES and Targeted NGS analysis. (A) Lineal protein domains of VNN2. (B) Structural analysis of variants in VNN2 protein . Human protein structure prediction with AlphaFold2. In the upper panel, a zoom of the active site of VNN2 is shown. The lower panel shows the structure of VNN2 protein. In red, the CN Hidrolase Domain. In light blue, the Vanin C Domain. In pink, the propeptide/transmembrane region. Black arrows indicate the variants analyzed. (C, D, E, F, G) Zoomed regions highlighting conformational changes due to the amino acid variants. . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 19, 2024. ; https://doi.org/10.1101/2024.09.18.24313937doi: medRxiv preprint . CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 19, 2024. ; https://doi.org/10.1101/2024.09.18.24313937doi: medRxiv preprint