Browse the corpus

The Genetic Architecture of Congenital Diarrhea and Enteropathy. BACKGROUND: Next-generation sequencing has enabled precision therapeutic approaches that have improved the lives of children with rare diseases. Congenital diarrhea and enteropathies (CODEs) are associated with high morbidity and mortality. Although treatment of these disorders is largely supportive, emerging targeted therapies based on genetic diagnoses include specific diets, pharmacologic treatments, and surgical interventions. METHODS: We analyzed the exomes or genomes of infants with suspected monogenic congenital diarrheal disorders. Using cell and zebrafish models, we tested the effects of variants in newly implicated genes. RESULTS: In our case series of 129 infant probands with suspected monogenic congenital diarrheal disorders, we identified causal variants, including a new founder NEUROG3 variant, in 62 infants (48%). Using cell and zebrafish models, we also uncovered and functionally characterized three novel genes associated with CODEs: GRWD1, MYO1A, and MON1A. CONCLUSIONS: We have characterized the broad genetic architecture of CODE disorders in a large case series of patients and identified three novel genes associated with CODEs. (Funded by the National Institutes of Health and others.).

Infants had a clinical diagnosis of congenital diarrhea (defined as chronic diarrhea lasting more than 2 weeks and starting in infancy with no attributed anatomical, infectious or allergic cause. (See ref. 1 for specific types of diarrhea based on stool osmolality including osmotic, secretory and mixed.) They were treated locally at the Hospital for Sick Children (HSC) in Toronto, Canada, Vanderbilt University Medical Center (VUMC), Boston Children’s Hospital (BCH), and the University of California Los Angeles (UCLA), and referred from across the United States to VUMC, BCH, and UCLA, and from several international institutions to BCH and UCLA. Families who previously consented to genetic testing and further research studies were analyzed as part of the PediCODE consortium (Pediatric Congenital Diarrhea and Enteropathies; www.PediCoDE.org) using a single NIH protocol IRB-P00027983 for UCLA, BCH, and VUMC in the United States and REB 1000072542 for the HSC in Canada. All families provided written consent to participate in the study; at least one parent or guardian provided written consent for the participation of their children. Demographic characteristics of the study population are provided in Table S1. Individual CODE patients or families were previously sequenced, and raw reads were aligned as described in Supplemental Material. Principal component analysis of the case series demonstrated a well-distributed ancestral diversity (Figure S1). Quality-control measures are outlined in the Supplementary Materials and Tables S2–6.

al CODE patients or families were previously sequenced, and raw reads were aligned as described in Supplemental Material. Principal component analysis of the case series demonstrated a well-distributed ancestral diversity (Figure S1). Quality-control measures are outlined in the Supplementary Materials and Tables S2–6. An overview of the genetic analysis of the case series and identification of monogenic CODE genes is provided in Figure 1A. Briefly, GEMINI (short for “genome mining”) was used to identify candidate variants based on known inheritance models of previously identified genes that, when variant, are known to cause CODE and are listed in OMIM14 (Table S7). We first searched exome data for rare (gnomAD15 allele frequency < 0.01) and damaging (Combined Annotation Dependent Depletion (CADD) Score16 > 20) variants in previously published CODE genes. We then carried out manual filtering based on confirmatory inheritance pattern, segregation, previous annotation in ClinVar, concurrence with clinical features associated with phenotypes of known genetic disease, and overall pathogenicity based on the ACMG/AMP classification17. An additional manual screen of the remaining cases was carried out to determine prior ClinVar-annotated pathogenesis and to search for novel CODE genes. Cases were considered solved when an ACMG/AMP pathogenic or likely pathogenic classification was identified in an OMIM CODE gene that fit the case description and expected inheritance pattern.

en of the remaining cases was carried out to determine prior ClinVar-annotated pathogenesis and to search for novel CODE genes. Cases were considered solved when an ACMG/AMP pathogenic or likely pathogenic classification was identified in an OMIM CODE gene that fit the case description and expected inheritance pattern. Functional studies to investigate novel candidate genes included the use of proximity-dependent biotin identification (BioID), cell-based assays, and the creation of novel zebrafish models (see methods in the Supplementary Appendix). Briefly, BioID was used to identify alterations in interactions between proteins encoded by candidate CODE gene variants, as compared to their wild-type counterparts with known binding-partner proteins. High-confidence interactors were defined as those with a Bayesian false discovery rate (BFDR) ≤0.01. Zebrafish studies were carried out at the Zebrafish Genetics and Disease Models Facility at The Hospital for Sick Children, Toronto, Canada, with approval from the Animal Care Committee, Animal Use Protocol #65759. Candidate genes in zebrafish were knocked out using a CRISPR-Cas9 system (Figure S2 and S3). The Student’s t-test was used to compare the means of two groups of zebrafish. RNA was extracted and sequenced from pooled wildtype or variant zebrafish and differential expression analysis was carried out using the DESeq2 package in R. Several cell lines, including CACO-2BBE (monolayers), Hela, MDCK-FCRN, and HT-29 cells were used for functional studies; statistical analyses of these studies are described in the figure legends and Supplementary Appendix.

We analyzed NGS data obtained from the PediCODE consortium sites (PediCODE.org). Overall, 139 infants (including 10 sibling pairs) who presented with CODE, and 182 parents and siblings without CODE were analyzed (Table S1). Of the 129 families analyzed, 98 had at least one family member (other than the proband) exome sequenced to assist with making a genetic diagnosis (Tables S2). Fourteen probands (1 sibling pairs) originated from consanguineous matings, as suggested by a relationship coefficient exceeding 0.1 (Table S3). In 62 of 129 probands, we identified causal variants in one of 24 known monogenic CODE genes (Figure 1B, Table S8). Of these 62 diagnosed probands, 58 (94%) had autosomal recessive (AR) and 4 (6%) had X-linked disorders (Table S8). Cases in nearly half the probands were linked to genes involved in epithelial trafficking and polarity, including EPCAM (12 cases), MYO5B (8 cases), SKIV2L (3 cases), and TTC7A (4 cases). Other CODE gene variants detected in this analysis included those in SLC9A3 (5 cases), DGAT1 (5 cases), XIAP (4 cases), and NEUROG3 (3 cases). Eight of the 10 sibling pairs who presented with CODE harbored the same pathogenic variant as the proband (Figure 1B and Table S8). Collectively, variants in these 8 genes caused disease in 52 (74)% of the identified cases. Seven infants from 6 unrelated families with Tufting Enteropathy carried a previously reported founder variant identified in the Arabic Gulf, EPCAM Q167Pfs*2118. Four infants with enteric anendocrinosis from 3 unrelated Bedouin families (each with a WES calculated relationship coefficient < 0.1) had a novel variant in NEUROG3 variant (Q137R).

elated families with Tufting Enteropathy carried a previously reported founder variant identified in the Arabic Gulf, EPCAM Q167Pfs*2118. Four infants with enteric anendocrinosis from 3 unrelated Bedouin families (each with a WES calculated relationship coefficient < 0.1) had a novel variant in NEUROG3 variant (Q137R). Three probands had variants that were not discoverable through exome sequencing; genome sequencing or Sanger sequencing was required to identify the causal variant in these probands, including Proband 68, who had an XIAP deletion. WGS also identified a PERCC1 variant in Proband 57, a recently identified gene required for enteroendocrine cell function that was previously unannotated and, therefore, not captured by WES19. Proband 5 had an intronic splice donor site variant in SLC9A3 that was missed due to poor coverage of Exon 8 (Figure S4). Furthermore, examination of median sequencing depth showed that several genes had poor coverage of exon 1, as has been noted in previous studies. Notably, several exons in MYO5B, the second most common cause (when pathogenically variant) of CODE in this study, had poor sequence, suggesting that deeper sequencing of MYO5B should be considered if there is a high suspicion for microvillus inclusion disease and negative results from MYO5B exon sequencing (Figure S4). Further interrogation of the exome data from patients with negative WES analysis revealed three novel candidate CODE genes based on genetic heritability, population frequency, CADD score, known gene function, or available animal models (detailed in Table S9).

Three probands had variants that were not discoverable through exome sequencing; genome sequencing or Sanger sequencing was required to identify the causal variant in these probands, including Proband 68, who had an XIAP deletion. WGS also identified a PERCC1 variant in Proband 57, a recently identified gene required for enteroendocrine cell function that was previously unannotated and, therefore, not captured by WES19. Proband 5 had an intronic splice donor site variant in SLC9A3 that was missed due to poor coverage of Exon 8 (Figure S4). Furthermore, examination of median sequencing depth showed that several genes had poor coverage of exon 1, as has been noted in previous studies. Notably, several exons in MYO5B, the second most common cause (when pathogenically variant) of CODE in this study, had poor sequence, suggesting that deeper sequencing of MYO5B should be considered if there is a high suspicion for microvillus inclusion disease and negative results from MYO5B exon sequencing (Figure S4). Further interrogation of the exome data from patients with negative WES analysis revealed three novel candidate CODE genes based on genetic heritability, population frequency, CADD score, known gene function, or available animal models (detailed in Table S9). The first candidate gene, GRWD1, was identified in a sib pair (comprising a male and female) who presented with congenital diarrhea requiring parenteral nutrition and diffuse arterial hypoplasia. Exome sequence analysis revealed rare and damaging maternally inherited 19:48451128 A/G (p.H307R; CADD score 25) and paternally inherited 19:48452786 G/T (p.V368F; CADD score 25) GRWD1 variants.

ing a male and female) who presented with congenital diarrhea requiring parenteral nutrition and diffuse arterial hypoplasia. Exome sequence analysis revealed rare and damaging maternally inherited 19:48451128 A/G (p.H307R; CADD score 25) and paternally inherited 19:48452786 G/T (p.V368F; CADD score 25) GRWD1 variants. Glutamate-rich WD40 repeat containing 1, GRWD1, is a highly conserved member of the WD40 protein family (Figure 2A). It is required for late steps of large ribosomal subunit assembly20 and is a regulator of p5321. As knockout of grwd1 was lethal in zebrafish, we generated first-generation (F0) mosaic grwd1 mutant zebrafish using CRISPR-Cas9 editing (crispant grwd1) that survived until 15 days post fertilization (dpf; Figure S2). Though crispant grwd1 larvae at 8 dpf had normal body length, they displayed significantly reduced gut length (Figure 2B) with disrupted intestinal architecture including enlarged, rounded goblet cells, disorganized enterocytes, and irregular gut lumen surface compared to the intestines of wildtype control fish (Figure 2C). The goblet cell area was greater in crispant grwd1 zebrafish than in control fish (Figure 2C). Differential expression analysis and pathway analysis of whole 8 dpf grwd1 crispant larvae revealed perturbation in several ribosomal and p53 regulated genes (Figure 2D).

o the intestines of wildtype control fish (Figure 2C). The goblet cell area was greater in crispant grwd1 zebrafish than in control fish (Figure 2C). Differential expression analysis and pathway analysis of whole 8 dpf grwd1 crispant larvae revealed perturbation in several ribosomal and p53 regulated genes (Figure 2D). Ribosome biogenesis dysfunction triggers nucleolar surveillance22, a regulatory loop in which unassembled ribosomal proteins forms a complex with the ubiquitin E3 ligase Hdm2, which in turn leads to p53 stabilization and increased expression of the target genes of p5323. Consistent with the activation of nucleolar surveillance, differential RNA expression analysis of whole 8 dpf grwd1 crispant larvae revealed perturbation in several ribosomal and p53 regulated genes (Figure 2D). To characterize the GRWD1 CODE protein variants, we conducted proximity-dependent biotin identification (BioID; Figure 2E and Table S10)24 on the GRWD1 protein and the missense variants H307R and V368F in human HEK293 Flp-In T-REx cells. BioID of GRWD1 identified a high-confidence interaction with a single ribosomal protein, RPL3, and with multiple ribosome biogenesis factors important for nucleolar steps of large subunit assembly25. Both H307R and V368F variant proteins displayed reduced interactions with RPL3 and ribosome assembly factors and increased interactions with prefoldin and CCT family protein folding chaperones. In budding yeast and C. elegans, the GRWD1 orthologues rrb1 (yeast) and GRWD-1 (C. elegans) are reported to act as chaperones for rpl326,27, and thus critical to the formation and function of the ribosomal peptidyl transferase center28. Decreased RPL3 binding for the GRWD1 CODE variants was confirmed in an orthogonal affinity purification assay (Figure 2F). Finally, Flag-tagged versions of the H307R and V368F GRWD1 variants were predominantly localized to the cytoplasm and the wildtype protein was primarily localized to the nucleus in transfected HeLa cells (Figure 2G). The GRWD1-variant proteins were mislocalized and had defective RPL3 chaperone activity, resulting in defective ribosomal biogenesis; and our grwd1 crispant zebrafish model demonstrated intestinal goblet cell abnormalities. We conclude that GRWD1 deficiency results in defective ribosomal biogenesis, and specifically in the gut, goblet cell dysfunction resulting in a novel CODE.

ive RPL3 chaperone activity, resulting in defective ribosomal biogenesis; and our grwd1 crispant zebrafish model demonstrated intestinal goblet cell abnormalities. We conclude that GRWD1 deficiency results in defective ribosomal biogenesis, and specifically in the gut, goblet cell dysfunction resulting in a novel CODE. The second candidate gene, MYO1A, was identified in a male infant who presented at two months of age with an abrupt onset of diarrhea peaking at 150 ml/kg/day and a 15% weight loss. With total parenteral nutrition the diarrhea continued. At 12 months, his stool remained liquid, averaging 20–50 ml/kg/day. Rare bi-allelic variants in MYO1A were identified, including a paternally inherited 12:57037571 T/A (p.I678F; CADD score 25), and a maternally inherited 12:57044132 C/T (p.D240N; CADD score 24; Figure 3A). Myosin-1A, MYO1A, is an actin-based monomeric motor protein whose expression is limited to the intestinal tract, where it localizes almost exclusively to, and is one of the most abundant proteins in, the brush border. It is critical for normal enterocyte brush-border function29,30. Myo1a knockout mice show multiple defects in the intestinal brush border, including extensive membrane herniations and fused microvilli31 similar to alterations observed in the proband’s duodenal biopsy at four months of age, which showed: (i) closely packed enterocytes, some of which had a rounded teardrop shape (Figure 3B) and (ii) a reduction of MYO1A at the brush border (Figure 3C). Similarly, when overexpressed in CACO-2BBE cells, the MYO1A I678F and D240N variants both exhibited aberrant localization when compared to a wild-type construct, which demonstrated strong enrichment in apical microvilli as expected (Figure 3D–F).

Figure 3B) and (ii) a reduction of MYO1A at the brush border (Figure 3C). Similarly, when overexpressed in CACO-2BBE cells, the MYO1A I678F and D240N variants both exhibited aberrant localization when compared to a wild-type construct, which demonstrated strong enrichment in apical microvilli as expected (Figure 3D–F). The proband had severe diarrhea that resolved by his 2nd birthday, consistent with the phenotype observed in Myo1a KO mice that survive as a result of redistribution to the brush border of other, compensatory class I myosins, MYO1C and MYO1D31. It would seem that the patient’s MYO1A variants result in the mislocalization from the microvilli of MYO1A.

that resolved by his 2nd birthday, consistent with the phenotype observed in Myo1a KO mice that survive as a result of redistribution to the brush border of other, compensatory class I myosins, MYO1C and MYO1D31. It would seem that the patient’s MYO1A variants result in the mislocalization from the microvilli of MYO1A. The third CODE gene candidate, MON1A, was identified in a female newborn who presented with diarrhea on day three of life that developed into profound diarrhea with vomiting and failure to thrive before 1 year of age. Exome sequence analysis of this consanguineous family identified a homozygous 3:49911685 G/A MON1A variant (p.R249C; CADD score 27; Figure 4A). The MON1 homolog A is expressed by the cell’s secretory trafficking apparatus and is a binding partner of CCZ132, together comprising a guanine nucleotide exchange factor (GEF) complex33 that supports RAB7A function in late endocytic trafficking34. As expected, BioID of MON1A identified CCZ1B, along with several proteins identified in a previous RAB7A BioID analysis35. While maintaining the interaction with CCZ1B, the MON1A R249C variant displayed notably reduced interactions with several RAB7A endocytic trafficking regulators (Figure 4B, Table S11). Consistent with a defect in endosome sorting caused by the disruption of MON1A/CCZ1 complex function32, immunohistochemistry of a duodenal biopsy from the proband demonstrated that NHE3 and EZRIN were mislocalized from the apical brush border and displayed significantly fewer RAB7-positive vesicles compared to those observed in biopsies from a healthy control (Figure 4C). Overexpression of the R249C MON1A variant resulted in reduced RAB7+ vesicle size and number, effects that could be rescued by complementation with MON1A (Figure 4D). Functional studies demonstrated that MON1A is necessary for endo-lysosome formation and acidification (Figure 4E) and similarly, polarized trafficking was altered by R249C MON1A with reduced bidirectional epithelial transcytosis of IgG in MDCK-FcRN cells (Figure 4F). Finally, mon1a knockout (−/−) zebrafish (Figure S3) did not show any obvious developmental defects but displayed multiple intestinal abnormalities, including morphologically atypical goblet cells with higher mucin-secretion and reduced activity of lysosome-rich enterocytes (Figure 4G–H). The R249C MON1A variant thus appears to be defective in its roles in endosomal sorting and Rab7-dependent endosome maturation in the gut, leading to a novel enteropathy.

The first candidate gene, GRWD1, was identified in a sib pair (comprising a male and female) who presented with congenital diarrhea requiring parenteral nutrition and diffuse arterial hypoplasia. Exome sequence analysis revealed rare and damaging maternally inherited 19:48451128 A/G (p.H307R; CADD score 25) and paternally inherited 19:48452786 G/T (p.V368F; CADD score 25) GRWD1 variants. Glutamate-rich WD40 repeat containing 1, GRWD1, is a highly conserved member of the WD40 protein family (Figure 2A). It is required for late steps of large ribosomal subunit assembly20 and is a regulator of p5321. As knockout of grwd1 was lethal in zebrafish, we generated first-generation (F0) mosaic grwd1 mutant zebrafish using CRISPR-Cas9 editing (crispant grwd1) that survived until 15 days post fertilization (dpf; Figure S2). Though crispant grwd1 larvae at 8 dpf had normal body length, they displayed significantly reduced gut length (Figure 2B) with disrupted intestinal architecture including enlarged, rounded goblet cells, disorganized enterocytes, and irregular gut lumen surface compared to the intestines of wildtype control fish (Figure 2C). The goblet cell area was greater in crispant grwd1 zebrafish than in control fish (Figure 2C). Differential expression analysis and pathway analysis of whole 8 dpf grwd1 crispant larvae revealed perturbation in several ribosomal and p53 regulated genes (Figure 2D).

The second candidate gene, MYO1A, was identified in a male infant who presented at two months of age with an abrupt onset of diarrhea peaking at 150 ml/kg/day and a 15% weight loss. With total parenteral nutrition the diarrhea continued. At 12 months, his stool remained liquid, averaging 20–50 ml/kg/day. Rare bi-allelic variants in MYO1A were identified, including a paternally inherited 12:57037571 T/A (p.I678F; CADD score 25), and a maternally inherited 12:57044132 C/T (p.D240N; CADD score 24; Figure 3A). Myosin-1A, MYO1A, is an actin-based monomeric motor protein whose expression is limited to the intestinal tract, where it localizes almost exclusively to, and is one of the most abundant proteins in, the brush border. It is critical for normal enterocyte brush-border function29,30. Myo1a knockout mice show multiple defects in the intestinal brush border, including extensive membrane herniations and fused microvilli31 similar to alterations observed in the proband’s duodenal biopsy at four months of age, which showed: (i) closely packed enterocytes, some of which had a rounded teardrop shape (Figure 3B) and (ii) a reduction of MYO1A at the brush border (Figure 3C). Similarly, when overexpressed in CACO-2BBE cells, the MYO1A I678F and D240N variants both exhibited aberrant localization when compared to a wild-type construct, which demonstrated strong enrichment in apical microvilli as expected (Figure 3D–F).

The third CODE gene candidate, MON1A, was identified in a female newborn who presented with diarrhea on day three of life that developed into profound diarrhea with vomiting and failure to thrive before 1 year of age. Exome sequence analysis of this consanguineous family identified a homozygous 3:49911685 G/A MON1A variant (p.R249C; CADD score 27; Figure 4A). The MON1 homolog A is expressed by the cell’s secretory trafficking apparatus and is a binding partner of CCZ132, together comprising a guanine nucleotide exchange factor (GEF) complex33 that supports RAB7A function in late endocytic trafficking34. As expected, BioID of MON1A identified CCZ1B, along with several proteins identified in a previous RAB7A BioID analysis35. While maintaining the interaction with CCZ1B, the MON1A R249C variant displayed notably reduced interactions with several RAB7A endocytic trafficking regulators (Figure 4B, Table S11). Consistent with a defect in endosome sorting caused by the disruption of MON1A/CCZ1 complex function32, immunohistochemistry of a duodenal biopsy from the proband demonstrated that NHE3 and EZRIN were mislocalized from the apical brush border and displayed significantly fewer RAB7-positive vesicles compared to those observed in biopsies from a healthy control (Figure 4C). Overexpression of the R249C MON1A variant resulted in reduced RAB7+ vesicle size and number, effects that could be rescued by complementation with MON1A (Figure 4D). Functional studies demonstrated that MON1A is necessary for endo-lysosome formation and acidification (Figure 4E) and similarly, polarized trafficking was altered by R249C MON1A with reduced bidirectional epithelial transcytosis of IgG in MDCK-FcRN cells (Figure 4F). Finally, mon1a knockout (−/−) zebrafish (Figure S3) did not show any obvious developmental defects but displayed multiple intestinal abnormalities, including morphologically atypical goblet cells with higher mucin-secretion and reduced activity of lysosome-rich enterocytes (Figure 4G–H). The R249C MON1A variant thus appears to be defective in its roles in endosomal sorting and Rab7-dependent endosome maturation in the gut, leading to a novel enteropathy.

We have characterized the broad genetic architecture of CODE disorders, highlighting the genes most commonly responsible and establishing the landscape of variants and heritability, including a potentially new Bedouin founder NEUROG3 variant. The diagnostic yield of ~50% was higher than the 4% observed in monogenic IBD11 and other monogenic disorders36, and similar to some high-yield neuromuscular disorder studies37. We identified causal variants in half of the CODE cases diagnosed in tertiary hospitals. It is possible that some variants in CODE genes escaped identification by exome sequence analysis. This seems particularly likely in infants with autosomal recessive disease, in whom only one variant was identified. The “missing variant” might have been detectable through genome sequencing, or obtaining better sequence coverage when exome sequencing.

s in CODE genes escaped identification by exome sequence analysis. This seems particularly likely in infants with autosomal recessive disease, in whom only one variant was identified. The “missing variant” might have been detectable through genome sequencing, or obtaining better sequence coverage when exome sequencing. It is likely that many individuals have rare or private genetic variants in novel genes. We also uncovered three novel CODE candidate genes and provided functional validation showing the variants alter protein function in relevant intestinal models that meet the criteria for reporting38. We used proximity-dependent biotinylation (BioID) to further support pathogenic classification of these three novel CODE genes by identifying dysregulated protein interactions in relevant intestinal pathways that are disrupted with candidate CODE variants. However, additional cases are required to further understand the pathogenesis of CODE and define genotype–phenotype correlations. Overall, we demonstrate the utility of next generation sequencing as a powerful tool to identify known and novel pathogenic variants that cause congenital diarrhea.