Mutations in RPSA and NKX2‐3 link development of the spleen and intestinal vasculature

Abstract Idiopathic intestinal varicosis is a developmental disorder defined by dilated and convoluted submucosal veins in the colon or small bowel. A limited number of families with idiopathic intestinal varices has been reported, but the genetic cause has not yet been identified. We performed whole‐exome and targeted Sanger sequencing of candidate genes in five intestinal varicosis families. In four families, mutations in the RPSA gene were found, a gene previously linked to congenital asplenia. Individuals in these pedigrees had intestinal varicose veins and angiodysplasia, often in combination with asplenia. In a further four‐generation pedigree that only showed intestinal varicosities, the RPSA gene was normal. Instead, a nonsense mutation in the homeobox gene NKX2‐3 was detected which cosegregated with the disease in this large family with a LOD (logarithm of the odds) score of 3.3. NKX2‐3 is a component of a molecular pathway underlying spleen and gut vasculature development in mice. Our results provide a molecular basis for familial idiopathic intestinal varices. We provide evidence for a relationship between the molecular pathways underlying the development of the spleen and intestinal mucosal vasculature that is conserved between humans and mice. We propose that clinical management of intestinal varices, should include assessment of a functional spleen.


| INTRODUCTION
The presence of dilated and convoluted submucosal veins in the colon or small bowel referred to as intestinal varices is a rare clinical entity with a poorly understood etiology (Speicher, Keegan, & Kirk, 2014).

| Whole-exome sequencing
After obtaining written informed consent, whole-exome sequencing was done using DNA isolated from blood, as described previously (Lelieveld et al., 2016). Briefly, exome capture was done using the Agilent SureSelect v4 Kit (Agilent, Santa Clara, CA). Exome libraries were sequenced on an Illumina HiSeq instrument (Illumina, San Diego, CA) with 101-bp paired-end reads at a median coverage of 75× and with >95% of exons having coverage >30×. Sequence reads were aligned to the hg19 reference genome using BWA version 0.5.9-r16. Variants were subsequently called by the GATK unified genotyper, version 3.2-2 and annotated using a custom diagnostic annotation pipeline. Variants were filtered for having less than 1% frequency in dbSNP, having <1% frequency in our in-house database and having <1% frequency in the "Exome Aggregation Consortium" (ExAC) database (www.exac.broadinstitute.org).

| Sanger sequencing
Sanger sequencing was done according to standard procedures, using M13 tailed forward and reverse primers for each exon of the RPSA gene or NKX2-3 gene. Primer sequences are given in Table S1.
Patients gave informed consent for the genetic studies, which were done in a routine diagnostic setting, and for inclusion of the data in this manuscript.

| Patients
The index patient in Family 1 presented with anemia from the age of 10, which required blood transfusions on several occasions. Colonoscopy demonstrated intestinal varices in the colon and to a lesser extent in the small bowel. At age 48, he developed bacterial meningitis.
Abdominal imaging showed an absence of the spleen. The index patient in Family 2 had been hospitalized with bacterial meningitis at the age of 3 years. She developed severe anemia from the age of 27 years. She was hospitalized repeatedly for recurrent gastrointestinal bleedings from varicose veins in the ascending and transverse colon. There was involvement of the small bowel with varicosities in the jejunum. On abdominal imaging, the spleen was described as "multiseptated" or "fragmented", possibly representing polysplenia.
Family 3 was reported previously as presenting with idiopathic congenital asplenia (Bolze et al., 2013). We re-evaluated the clinical data for some of the individuals from this family, who were known to have ectatic blood vessels in their intestines. The index patient presented at 17 years of age with severe fatigue and anaemia managed with repeated transfusions as no cause could be identified at that time. Asplenia was detected on an abdominal computed tomography scan. At age 24, an exploratory laparotomy identified abnormally dilated vessels in the wall of the distal duodenum. Since then, the patient has undergone repeated argon laser cauterization of duodenal blood vessels approximately every 6 months as they are inoperable. Family 4 has been reported previously (Wurfel et al., 2011). The index patient had a history of iron-deficiency anemia No variants in the NKX2-3 gene were found in these families. In  (Bolze et al., 2013). In Family 3, a c.538C>G (p.(Arg180Gly)) amino acid substitution was found, which was previously discovered and described in the context of a study on isolated congenital asplenia (Bolze et al., 2013). In Family 4, a F I G U R E 1 Pedigrees of Families 1-5 described in this study. The proband in each family is indicated by an arrow. Filled (black) symbols indicate intestinal varices. An asterisk indicates the presence of a heterozygous variant in either the RPSA or NKX2-3, that is, in Family 1, a c.223dup (p.(Ser75Lysfs*36)) variant in RPSA, in Family 2, a c.252G>C (p. (Gln84His) In Family 5 the whole-exome sequencing data of two distantly related affected relatives were compared (i.e., individuals IV:1 and III:9; see pedigree in Figure 1), using the filter settings detailed above.
Among the shared variants (see Table S2), a heterozygous c.268del

| Genotype-phenotype correlation
The families with RPSA variants (Families 1-4) (Bolze et al., 2013(Bolze et al., , 2018(Bolze et al., ) et al., 2014Bolze et al., 2018). Asplenia was also part of the extended phenotype in families described here. Both asplenia and intestinal varices may be occult for many years. Nonetheless, both disorders may have severe consequences, with asplenic patients developing meningitis or other forms of severe septic infections, and intestinal varices sometimes leading to severe bleeding necessitating hospitalizations, transfusions, and in some the removal of sections of the intestine. Several persons with RPSA mutation in these pedigrees developed bacterial meningitis or pneumonia, suggesting that persons with intestinal varices should be examined for the presence of a functional spleen.
A synthetic peptide recapitulating amino acids 161-180 of the RPSA-encoded laminin-receptor protein binds to laminin with high affinity (Castronovo, Taraboletti, & Sobel, 1991). Others have argued based on the crystal structure of the laminin-receptor protein that of this putative laminin-binding domain, only amino acid R180 is solventexposed, with a critical role for a binding face involving Phe-32, Glu-35, and Arg-155 (DiGiacomo & Meruelo, 2016;Jamieson, Hubbard, & Meruelo, 2011;Jamieson, Wu, Hubbard, & Meruelo, 2008). However, as far as we know, requirement specifically of the Arg-180 residue for laminin-binding properties of RPSA has not been experimentally proven, and Griffin et al. (2018) showed that this residue is required for pre-rRNA processing. It is, therefore, unclear at the moment which cellular process exerted by NKX2-3 are at the basis of asplenia pathogenesis. It is striking, however, that Arg-180 and the adjacent Gln-181 appear to constitute a hotspot for mutations leading to asplenia and intestinal varices (Bolze et al., 2018; Figure 3).
NKX2-3 functions in development and function of the intestinal lymphoid system and intestinal vasculature (Kellermayer et al., 2016;Yu et al., 2011). Nkx2-3 is embryonically and postnatally expressed in the midgut and hindgut of the mouse and the chicken (Pabst, Schneider, Brand, & Arnold, 1997;Wang et al., 2000). High human NKX2-3 messenger RNA (mRNA) expression is restricted to the colon, ileum, and spleen (gtexportal.org/home/gene/NKX2-3; Pabst et al., 1997). NKX2-3 has hitherto not been linked to a genetic disorder but the gene is a strong candidate gene for intestinal varices as it is expressed in human intestinal microvascular cells where it regulates VEGFA and MADCAM-1 signalling (Wang et al., 2000;Yu et al., 2011). Wang et al. (2000) replaced the NK-2-specific domain of were apparently normal. Pabst, Zweigerdt, and Arnold, (1999) studied Nkx2-3-null mice generated by targeted gene disruption. Homozygous Nkx2-3 −/− mice were growth-retarded, and the majority died before 3 weeks after birth. Reduced proliferation of the epithelium was shown in the fetal small intestine. In adult homozygous knockout mice, the small intestine showed altered villus morphology and increased villus size. Extensive extra vascularization of the small intestine was noted in these mice. Remarkably, the intestinal changes in NKX2-3 knockout mice were limited to the jejunum and ileum and absent in the colon, even though Nkx2-3 is also expressed in the hindgut. Outside of the digestive tract, splenic malformations were noted, with absent spleens in 20% of Nkx2-3 −/− mice and abnormalities in size or morphology of the spleen in the other mice (Pabst et al., 1999). Another study found ectopic vessel formation in the spleen of Nkx2-3 mutant mice, which were described as "high-endothelial venule-like" (Kellermayer et al., 2016).
Currently, no direct molecular connection between NKX2-3-and RPSA-related pathways is known, but data reviewed here provide hints that both genes may be involved in the same molecular processes in spleen and intestinal vasculature development from mesenchymal tissue during embryogenesis. An anatomical relationship may exist between asplenia and vascular malformation as the former is associated with anomalous venous drainage and possibly subsequent arteriovenous malformation (Arnautovic, Mazhar, Tereziu, & Gupta, 2017). Our finding that mutation of both RPSA and of NKX2-3 can cause intestinal varices complements previous studies showing that RPSA mutations affect spleen development in humans, and that knockout of NKX2-3 disrupts spleen development in mice. It is currently unclear why asplenia is not a feature associated with NKX2-3 mutations in humans.
Possibly, the expression of the spleen phenotype may be variably penetrant as has been described for RPSA mutations (Bolze et al., 2018) and the condition may be occult. The identification of more patients with NKX2-3 mutations in the future may define a broader clinical phenotypic spectrum linked to NKX2-3 mutations.
A possible molecular link between RPSA and NKX2-3 could be through NKX2-5, as NKX2-3 and NKX2-5 can heterodimerize (Kasahara et al., 2001) and NKX2-5 is part of a functional module that contributes to the development of the spleen in mouse (Burn et al., 2008;Czompoly et al., 2011;Koss et al., 2012). Moreover, depletion of RPSA in Xenopus causes severe reduction of NKX2-5 mRNA expression, that can be rescued by WT human RPSA mRNA but not by mutant p.Arg180Gly RPSA mRNA (Griffin et al., 2018).
In summary, we find that mutations in at least two genes can cause familial idiopathic intestinal varices, and hypothesize that there may be links between the molecular pathways involved in the development of the spleen and of the intestinal vasculature.

ACKNOWLEDGMENTS
We thank all the patients and families for their cooperation. We thank Kees van Roozendaal and Ragonda Souren for help with wholeexome sequencing and Sanger-sequencing analyses.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request.