SHOC2 is a novel cause of central conducting lymphatic anomaly

Abstract

Central conducting lymphatic anomaly (CCLA) is a devastating disorder where precision medicine can be lifesaving. A patient with complex congenital heart disease presented with post-operative refractory chylothorax. Lymphatic imaging demonstrated abnormal intercostal lymphatic perfusion consistent with CCLA and suggestive of a RASopathy, a group of genetic disorders characterized by dysregulated RAS-MAPK signaling. Genetic testing showed a variant of uncertain significance (p.(Ala308Val)) in SHOC2, a scaffolding protein that regulates RAS-MAPK signaling but has not yet been described in CCLA. Trametinib, a MEK1/2 inhibitor, was trialed without improvement. A conversion/extension assay was used to evaluate pathogenicity of the VUS compared to a known pathogenic variant. Transient transgenic zebrafish expressing the wild type human SHOC2 protein or SHOC2 p.(Ala308Val) protein in the developing lymphatic and venous endothelium were used to evaluate the effect on the vasculature. Western blot was used to explore the underlying molecular mechanisms. Major to minor axis ratio was significantly increased in developing embryos injected with SHOC2 r.(Ala308Val) mRNA, indicating pathogenicity through dysregulation of MAPK/mTOR signaling. Transient transgenic zebrafish expressing SHOC2 p.(Ala308Val) in the lymphovenous endothelium developed pericardial edema and cystic malformations of the caudal plexus and dilation of the thoracic duct. In concordance with the lack of clinical response to trametinib, immunoblot results indicate that SHOC2 p.(Ala308Val) may be inducing lymphatic malformations through non-canonical mTOR signaling rather than dysregulated RAS-MAPK signaling. SHOC2 is a novel cause of central conducting lymphatic anomaly. Future research should investigate the cellular and molecular mechanisms as well as therapeutics.

INTRODUCTION

Central conducting lymphatic anomaly (CCLA, MIM #620014) is a complex condition characterized by abnormalities of the central lymphatics that typically arise during embryonic development. CCLA can manifest as chylothorax, lymphangiectasia, nonimmune fetal hydrops, and edema.¹ Since CCLA can encompass any dysfunction of the central lymphatics, a great degree of phenotypic heterogeneity is observed among individuals with this condition.² Recent advances in genetic testing have identified pathogenic variants in genes regulating rat activated sarcoma/mitogen-activated protein kinase (RAS/MAPK) signaling as the most common causes of CCLA and demonstrated that MEK inhibitors, such as trametinib, are efficacious in single patient studies.^2-6

Here, we present an individual with complex single-ventricle congenital heart disease who developed severe chylothorax following cardiothoracic surgery. A RASopathy gene panel showed a germline heterozygous variant of uncertain significance (VUS) in (SHOC2) NM_007373.4 c.923C>T p.(Ala308Val). Dynamic contrast magnetic resonance lymphangiography (DCMRL) showed significant abnormal intercostal lymphatic perfusion which has been previously reported in RASopathy patients with CCLA.^2,7-9 The patient’s phenotype and clinical presentation led us to hypothesize that this SHOC2 VUS is pathogenic.

Noonan syndrome-like disorder with loose anagen hair is caused by monoallelic pathogenic variants in SHOC2 (HGNC:15454; NSLH, MIM #607721), one of several RASopathies with features which overlap with classical Noonan syndrome.¹⁰ SHOC2 is a highly conserved leucine rich repeat (LRR) scaffold protein that forms a holophosphatase complex by binding the catalytic subunit of protein phosphatase 1 (PP1c) and GTP-loaded M-RAS.¹¹ This complex regulates RAS/RAF/MAPK signaling by dephosphorylating an inhibitory serine residue on RAF kinases, releasing it from its autoinhibitory state, potentiating MAPK signaling.^12,13 Variants in SHOC2 have not previously been reported as a cause for central conducting lymphatic anomaly.

We hypothesized that mutant SHOC2 p.(Ala308Val) was responsible for the lymphatic disease observed in this patient. Zebrafish have proven to be a robust model for examining the pathogenicity of gene variants in lymphatic disease due to their optical clarity, genome conservation with humans, and fluorescent marker lines aiding in the visualization of the developing lymphatics.^3-5,14-17 Here, we utilized an in vivo zebrafish model to provide novel insights into SHOC2 as a cause of central conducting lymphatic anomaly as well as begin to elucidate the underlying molecular mechanisms which may inform future therapeutic avenues for patients with SHOC2-related lymphatic disease.

MATERIALS AND METHODS

Editorial Policies and Ethical Considerations

Informed written consent for publication was obtained from the proband’s legal guardians.

Study Participant

Clinical genetic testing of genomic DNA isolated from whole blood was performed using the Invitae RASopathies and Noonan Spectrum Disorders Panel, the Invitae Congenital Heart Disease Panel, and the Children’s Hospital of Los Angeles (CHLA) Center for Personalized Medicine Vascular Anomalies and Mosaic Disorders Panel (VMD4Kids), covering a total of 305 genes (Table S1). Invitae’s variant classification system “Sherloc” is modified from American College of Medical Genetics and Genomics/Association for Molecular Pathology (ACMG/AMP) criteria.¹⁸ CHLA’s variant analysis was performed according to the ACMG/AMP 2015 guidelines.¹⁹

Dynamic contrast-enhanced magnetic resonance lymphangiogram (DCMRL) was performed using serial T1-weighted imaging, with gadolinium-based contrast injected into bilateral inguinal lymph nodes and the hepatic (periportal) lymphatics.^20,21 Patency of the thoracic duct was assessed via intranodal injection of an ultrasound enhancing agent (sulfur hexafluoride lipid-type A microspheres) and ultrasonography of the subclavian veins.²² Selective lymphatic imaging and interventions were performed under fluoroscopic guidance with Iopamidol contrast.

Zebrafish

All zebrafish used in this study were handled according to current Animal Care and Use Committee protocols in AAALAC accredited facilities. Developing larvae from 0-7 days post-fertilization (dpf) were kept in incubators at 28.5°C and adult fish were kept in 1.8L tanks on a re-circulating system with daily feedings of Gemma 300 Micro (Skretting).

The Tg(mrc1a:EGFP)^y251 and Tg(kdrl:GFP)^la116 zebrafish used in this study were a gift from the laboratory of Dr. Brant Weinstein at NICHD.¹⁴ The Tg(mrc1a:mtagBFP2-CAAX)^y722; Tg(kdrl:eGFP)^la116 zebrafish were generated for this study as described below.

To generate the Tg(mrc1a:mTagBFP2-CAAX)^y722 transgenic line, the mrc1a promoter from the Tg(mrc1a:EGFP)^y251 transgenic line was synthesized into a Tol2 site-containing vector along with the mTagBFP2 sequence²³ and a 3’ CAAX motif²⁴ using VectorBuilder. This plasmid was injected at a dose of 25 pg with 25 pg of transposase mRNA into single-cell zebrafish embryos from wild-type EK adults, screened for transgene insertion at 2 dpf, and then allowed to mature until adulthood. A founder was identified and outcrossed to homozygous nacre mutant adults to maintain them in a background line without pigmentation. A single adult F1 founder was outcrossed again to a nacre mutant to ensure single insertion of the mrc1a:mtagBFP-CAAX gene (50/50 mendelian ratio of BFP/WT offspring). These adults were in-crossed to ensure there was no lethal or harmful phenotype in the homozygous BFP offspring. They were then crossed with Tg(kdrl:GFP)^la116 zebrafish²⁵ to generate the double transgenic line Tg(mrc1a:mtagBFP2-CAAX)^y722; Tg(kdrl:eGFP)^la116 utilized in this study.

Plasmids and transposase mRNA

The human SHOC2 WT and p.(Ala308Val) sequences (GenBank: NM_007373.4 and NM_007373.4:c.923C>T) were synthesized into the pzTol2 expression vector backbone by VectorBuilder.²⁶ Transgene expression was driven by the mrc1a zebrafish promoter¹⁴ and was linked to an mScarlet3 fluorescent protein²⁷ with a P2A-T2A²⁸ ribosomal skipping peptide. Transposase mRNA was synthesized in an SP6 reaction using Not-I linearized pCS2+ vector containing the transposase mRNA sequence with the mMESSAGE mMACHINE SP6 kit following the manufacturers specifications (Invitrogen #AM1340).²⁶

Zebrafish convergence and extension assay

For convergence extension assays, the human SHOC2 WT and p.(Ala308Val) sequences (GenBank: NM_007373.4 and NM_007373.4:c.923C>T) were cloned from the pzTol2 vectors into the pCSDest backbone (Addgene# 22423).²⁹ The positive control, SHOC2 p.(Ser2Gly) (NM_007373.4:c.4A>G) variant was generated via site directed mutagenesis using the SHOC2 WT plasmid as the template and below primers:

	Primer Sequence (5’ – 3’) c.4A>G SNV represented in red
Forward	GGGGACAAGTTTGTACAAAAAAGCAGGCTTCGCCACCATGGGCGGCTCCCTGGGAAAG
Reverse	GGGGACCACTTTGTACAAGAAAGCTGGGTACACCATAGCCCTGTATGGTCCCT

The PCR fragment was then cloned into the pCSDest backbone using the Invitrogen Gateway Cloning system (Invitrogen# 12537100).

mRNA was synthesized as above and single-cell stage embryos were injected with 83 pg/nL of corresponding mRNAs using a PLI-90 microinjector (Harvard Apparatus, Cambridge, MA). Following microinjections, embryos were maintained at 25°C and imaged at the 6-8 somites stage using a 2X water immersion objective (0.06NA) and a Nikon Eclipse Ni-E microscope equipped with a DS-Ri2 camera. Yolk measurements of the major to minor axis ratio were performed using the Nikon NIS Elements Analysis software. Two-tailed, unpaired t-tests were performed using Prism 10. The individual measurements are available in Source Data Figure 3B.

Zebrafish Tol2 transient transgenic model

Injections into zebrafish embryos for Tol2 transgene expression were performed at the one-cell stage. 12.5 pg of corresponding plasmid DNA was co-injected with 12.5 pg of transposase mRNA in a solution with 0.2 M KCl and 0.01% phenol red (Sigma-Aldrich #P0290). Injections were performed using the Nanoject III (Drummond Scientific, Broomall, PA) injector with a Leica M80 Stereo Microscope equipped with a 1X objective.

At 2dpf, larvae were manually de-chorionated and visually screened for mScarlet3 expression via Leica M80 fluorescence stereoscope microscopy with 1X objective as a marker for transgene insertion. Each larva with positive mScarlet3 expression was retrieved from the plate and pooled into a 100 mm culture dish with fresh embryo media containing 0.00005% of the anti-fungal methylene blue (J.T. Baker #Q473-03) in MilliQ Water with 0.06 mg/mL Instant Ocean Sea Salt (Instant Ocean #SS6-25). Survival, pericardial edema, cystic formations and circulation were monitored daily through 5dpf (Source Data Figure S1.xlsx). Counts for each dependent variable were averaged across three independent injection experiments and compared using Student’s t-test in Prism 10 (GraphPad, Boston MA).

From 2-5dpf, a selection of representative embryos with transgenesis, without edema and with normal circulation was chosen for confocal imaging. For consistency, the same larvae were imaged on 2-3 dpf, 3-4 dpf, and 4-5 dpf. The larvae were anesthetized and mounted in 1.5% low-melt agarose for confocal imaging on a Nikon Ti2 inverted microscope equipped with an AXR resonance confocal scanner. Z-stack images were obtained of whole embryos with a step-size of 2 uM using a 20X water immersive objective (0.95NA). Maximum intensity projections and AI-denoising were performed during post-acquisition using the NIS-Elements AR 5.41.02 software (Nikon, Melville NY).

Measurements of the lateral facial lymphatic were obtained by taking five sequential diameter measurements of the facial lymphatic from the major branch off of the jugular and otolithic lymphatic vessels to the tip of the vessel just ventral to the eye. Measurements of the caudal aorta and caudal vein plexus were obtained by taking 16 sequential diameter measurements after the first intersomitic vessels (ISVs) directly following yolk sac, with approximately 2 measurements taken between each ISV. The thoracic duct measurements were acquired using a similar approach, with a total of 10 sequential diameter measurements. Posterior cardinal vein measurements were acquired using a similar approach but starting at the most cranial portion of the vessel over the yolk sac, with a total of 10 sequential diameter measurements. All measurements were averaged for each fish and compared using ANOVA in Prism 10 (GraphPad, Boston MA). Measurements are available in Source Data Figure 4.xlsx.

Western blot

Fifteen zebrafish embryos at either the 6-8 somite stage or larvae at 5dpf (dependent on assay) were collected and manually lysed in RIPA buffer (ThermoFisher # 89900) supplemented with 1X HALT protease and phosphatase inhibitor cocktail (ThermoFisher #78442) and incubated on ice for 30 minutes. Following a second manual homogenization, samples were centrifuged at 4°C to clear the lysate. The lysate was transferred to a new tube and quantified using a Qubit 4 Fluorometer (ThermoFisher #Q33238) with the Broad Range Protein Detection Assay (ThermoFisher #A50668). The cleared lysate was then denatured and reduced in 1X NuPAGE LDS sample buffer (ThermoFisher #NP0007) with 1X NuPAGE Sample Reducing Agent (ThermoFisher #NP0004) by boiling for 5 minutes at 95°C. Twenty micrograms of each sample was separated on a NuPAGE 4-12% Bis-Tris gel (ThermoFisher #NP0322) via SDS-PAGE gel electrophoresis for 1 hour at 200V. A sample of PageRuler Pre-stained Protein Ladder was also run in tandem for kDa reference (ThermoFisher #26616). Separated samples were then transferred to a nitrocellulose membrane using the Invitrogen iBlot2 at 20-25V over 7 minutes (Invitrogen #IB21001).

For target detection, membranes were blocked with 3% BSA (Avantor #0332) in TBS (Crystalgen #302-003-1000) with 0.1% Triton-X (Sigma-Aldrich #X100) for one hour at room temperature and probed with the appropriate primary and secondary antibody (Table S2). Blots were incubated with SuperSignal West Pico Plus ECL (ThermoFisher #34580) and imaged using the BioRad ChemiDoc Imaging System (BioRad #12003153). Western blot quantitation was performed with the ImageLab software from BioRad. The raw blots from Figure 3 are available in Source Data Figure 3G bactin.pdf and Source Data Figure 3G pERK ERK.pdf. The raw blots from Figure S1 are available in Source Data Figure S1.pdf. The densities are available in Source Data Figure 3H Densities.xlsx and Source Data Figure S1 Densities.xlsx.

Figure 3: Expression of SHOC2 p.(Ala308Val) in zebrafish embryos disrupts early conversion and extension movements.

A) Representative images and B) quantification from CE assay demonstrating changes in the yolk major to minor ratio (n=25 per cohort). C-F) Whole-mount zebrafish embryo immunostaining for mCherry to measure relative expression levels of the injected SHOC2 WT or VUS transgene. G) Western blot analysis of phosphorylated-ERK, total-ERK, and beta actin in pooled samples from the above experiment with H) quantification of pERK normalized to input. All statistics calculated by two-tailed, unpaired, T-test. Each dot represents the measurement from a single embryo. p-values are represented on each graph.

Zebrafish embryo immunofluorescence

Following imaging for the CE assay, embryos were fixed with 4% paraformaldehyde (PFA) in phosphate buffered saline (PBS) for 2 hours at room temperature, washed with PBS and blocked with 5% Goat serum and 0.25% Triton X-100 in PBS for 1 hour at room temperature. Embryos were then incubated with the appropriate primary and secondary antibodies overnight at 4°C or at room temperature for 1 hour, respectively (Table S2). Images were captured with a Marianas 3I Spinning disk microscope using a 5X Zeiss objective and a sCMOS (Hamamatsu ORCA Flash4) camera. Images were pseudo-colored in red using SlideBook software.

Statistics

Statistics were performed in Prism 10 (GraphPad, Boston MA) and are specified in the legend.

RESULTS

Case report

The proband is a term female with a prenatal diagnosis of double inlet left ventricle with transposition of the great arteries (situs solitus with left ventricular loop and left transposition [S,L,L]), anterior/leftward aorta, restrictive ventricular septal defect, severely hypoplastic left-sided morphologic right ventricular outflow chamber, aortic coarctation, dysplastic and redundant left atrioventricular valve with stenosis, and restrictive atrial septal defect. The pregnancy was conceived by in vitro fertilization and complicated by diet-controlled gestational diabetes. There was no known family history of genetic conditions or congenital heart disease. Her growth parameters at birth were within normal limits. On day of life 2 at a weight of 3.5 kg, she underwent the Norwood procedure (aortic arch reconstruction, atrial septectomy, and modified Blalock-Taussig-Thomas shunt).

Her postoperative course was complicated by high volume chylous chest tube output, averaging approximately 120 mL/kg/day by postoperative day 7. Echocardiography and cardiac catheterization did not show any hemodynamic or anatomic driver of the chylothorax. There was normal left ventricular systolic function, no significant atrioventricular valve dysfunction, no aortic arch obstruction, atrial pressure of 5-7 mmHg (normal estimated less than 7), pulmonary vascular resistance of 1.7 indexed Wood Units (normal less than 2 iWU), and no venous thrombosis or obstruction. She remained on room air and hemodynamically stable, but severe chylothorax persisted despite standard medical management with chest tube placement, low-fat diet with Monogen and Enfaport formulas, and octreotide (two trials of 10 mcg/kg/hr for 10-12 days).³⁰ On postoperative day 35, DCMRL demonstrated an intact, hypoplastic, left-sided thoracic duct with diffuse retrograde perfusion of posterior intercostal and retroperitoneal lymphatics, similar to the central lymphatic phenotype of RASopathies.² Abnormal perfusion of bilateral pleural spaces was also observed, but no discrete source could be identified, consistent with CCLA (Figure 1A-B, Supplemental Movie 1). Contrast-enhanced ultrasound lymphangiography demonstrated patency of the thoracic duct to the left subclavian vein.

Figure 1: DCMRL demonstrates CCLA as the cause of refractory chylothorax.

A) Inguinal lymph node dynamic contrast magnetic resonance lymphangiogram (DCMRL) of the proband with chylothorax following neonatal cardiothoracic surgery. Arrows indicate regions of abnormal lymphatic perfusion into intercostal and retroperitoneal spaces. B) Selective fluoroscopy angiogram of the left-sided thoracic duct (arrows) draining superiorly to the left venous angle. C) Fluoroscopy following percutaneous lymphatic intervention. Arrows indicate sodium tetradecyl sulfate mixture within retroperitoneal and posterior intercostal lymphatic channels. Arrowhead indicates the tip of the thoracic duct catheter. D) Follow-up intrahepatic DCMRL at 8 months of age, showing progression of posterior intercostal lymphatic collaterals (arrows) particularly to the left, retrograde perfusion of retroperitoneal and mesenteric lymphatics (arrowhead), and late filling of the thoracic duct.

Clinical genetic testing identified two germline heterozygous variants of uncertain significance (VUS): SHOC2 c.923C>T (p.(Ala308Val)) and ZFPM2 c.2336G>A (p.(Gly779Glu)). Noonan syndrome-like disorder with loose anagen hair is caused by monoallelic pathogenic variants in SHOC2 (NSLH, MIM #607721). Notably, on physical examination she did not have the typical features such as facial gestalt, short stature, pectus excavatum, webbed neck, or easily pluckable hair. Parental testing for the SHOC2 variant demonstrated that it was maternally inherited, however the mother of the patient denies a personal history of any cardiac, neurodevelopmental, or growth differences outside of scoliosis. ZFPM2 (HGNC:16700) encodes the multitype 2 zinc finger protein and has been associated with congenital heart disease including Tetralogy of Fallot and double outlet right ventricle.³¹

On postoperative day 45, at age 6 weeks and 5 days, she underwent percutaneous transabdominal access of the thoracic duct under fluoroscopic guidance for further diagnostic imaging and potential intervention. Selective thoracic duct lymphangiography demonstrated a diffuse network of abnormal channels arising from the low thoracic duct, without a discrete target(s) for a selective embolization. There was concern that selective lymphatic embolization(s) or total thoracic duct embolization would unmask lymphatic dysfunction in other compartments. With her central lymphatic phenotype suggestive of a RASopathy, the decision was made to first trial trametinib, a MEK1/2 inhibitor, off-label.³² The patient was initially started on trametinib 0.025mg/kg PO once daily and after one week was increased to 0.04 mg/kg PO once daily. After four weeks of trametinib therapy there was no improvement in chylothorax, with bilateral chest tube output averaging 130 mL/kg/day. The gradual accrual of chylothorax-associated morbidity despite trametinib therapy precluded further attempts at medical management.

Percutaneous embolization of the abnormal pulmonary perfusion channels from the low thoracic duct and cisterna chyli was performed using a slurry-mixture of 1 mL 3% sodium tetradecyl sulfate, 0.5 mL lipiodol, and 1 mL air (Figure 1C).³³ By post-procedure day 6 there was resolution of chylothorax. Her care was gradually de-escalated, and she was discharged at 5 months of age on a low-fat diet and high dose diuretic regimen. She experienced recurrence of pleural effusions with any weaning of diuretic medication.

At three months of age, further genetic testing was performed on peripheral blood (CHLA Center for Personalized Medicine: Vascular Anomalies and Mosaic Disorders gene panel) which identified a likely pathogenic variant in GLMN c.844_847del (p.(Leu282GlnfsTer10)), VAF = 39.5%, likely pathogenic, heterozygous, germline. Autosomal dominant variants in GLMN (HGNC:14373) cause glomuvenous malformations (GVMs, MIM #138000) when a somatic “second-hit” pathogenic variant is also present.³⁴ The patient did not display any skin lesions characteristic of glomuvenous malformations and neither of the proband’s parents reported a history of skin lesions but were not specifically examined by our team.

Follow-up intra-hepatic DCMRL at 8 months of age demonstrated significant enlargement and progression of posterior intercostal lymphatic collaterals, retrograde perfusion of retroperitoneal and mesenteric lymphatics, perfusion of small subclinical left chylothorax, and no filling of the thoracic duct (Figure 1D).

Variant analysis of SHOC2 p.(Ala308Val)

Given that the central lymphatic phenotype was similar to what is seen in other patients with RASopathies, we hypothesized that the p.(Ala308Val) variant in SHOC2 was pathogenic. Canonically, in HEK293T and human colon carcinoma cell lines, SHOC2 activates ERK signaling by forming a holophosphatase complex with PP1C and MRAS (Figure 2A) which dephosphorylates an inhibitory phosphate on RAF.¹³ The p.(Ala308Val) variant is present at the initiation of the tenth LRR domain which spans from residue 308 to 329 (Figure 2B-C red box, 2D). Prior cryo-electron microscopy analysis showed that the most extensive SHOC2-PP1C interactions were present in the ascending loops of LRRs 2-5 and 7-11.¹¹

Figure 2: SHOC2 c.923C>T p.(Ala308Val) is a variant of uncertain significance.

A) Schematic of SHOC2’s interaction with RAS and mTOR signaling ⁴⁴. B-C) AlphaFold predicted structures of SHOC2 wild-type and SHOC2 p.(Ala308Val) with corresponding amino acid sequence. D) SHOC2 protein sequence alignment was performed using DNASTAR MegAlign Pro with the following GenBank sequences: human: NP_031399.2, zebrafish: NP_001038251.1, mouse: NP_062632.2, rat: NP_001013173.1, chicken: NP_001384212.1, xenopus tropicalis: XP_031761665.1, and pig: XP_005671515.1. SHOC2 is 92% conserved between humans and zebrafish. E) Schematic demonstrating the two binding domains of SHOC2, the location within those domains of this VUS and a known gain of function pathogenic variant in SHOC2, p.(Ser2Gly).

Various in silico analyses present conflicting predictions of pathogenicity (Table S3). The SHOC2 c.923C>T p.(Ala308Val) variant has been reported to NCBI ClinVar as a VUS on three separate occasions (ID# 2079012) in a RASopathy patient, a patient with an unspecified phenotype, and a patient with an unspecified “cardiovascular phenotype.” The reporting lab classified this variant as a VUS based on the current ACMG/AMP criteria.¹⁹

SHOC2 p.(Ala308Val) disrupts early convergence and extension during embryonic development

Zebrafish are an excellent model system for studying RASopathies and their associated phenotypes. SHOC2 is highly conserved across species including the zebrafish (92% conservation, Figure 2D). Activation of PI3K and ERK signaling is a critical regulator of cell movement during zebrafish gastrulation^35-36, thus a convergence extension (CE) assay was used to assess the pathogenicity of the p.(Ala308Val) variant compared to a negative control (SHOC2 WT) and a positive control (SHOC2 p.(Ser2Gly), Figure 2E). This technique has been previously used to validate the pathogenicity of RASopathy-causing single nucleotide variants.³⁷ Expression of other NSLH-associated SHOC2 variants result in oblong shaped embryos quantified by measuring the major to minor axis ratio.¹⁷ Global expression of WT SHOC2 did not affect embryonic development (Figure 3A), whereas expression of a known pathogenic variant (p.(Ser2Gly)) or the proband’s VUS p.(Ala308Val) significantly (p = 0.0015, p = 0.0018, respectively) increased the major to minor axis ratio (Figure 3B). Immunofluorescence staining of mCherry expression confirmed that expression levels were similar across all conditions (Figure 3C-F). Western blot analysis of ERK activation in these embryos revealed that activation of ERK was decreased in the SHOC2 p.(Ser2Gly) larvae and slightly decreased in the SHOC2 p.(A308V) embryos. This led us to suspect that disruption of PI3K/mTOR signaling rather than MAPK/ERK may be the driving force of the early disruptions of conversion and extension.

SHOC2 p.(Ala308Val) zebrafish exhibit edema, decreased venous flow, and body axis defects

To investigate the effect of this variant on lymphatic development more specifically, we used Tol2 transgenesis and the zebrafish mrc1a promoter to drive expression of human SHOC2 p.(Ala308Val) linked to an mScarlet3 fluorescent protein with a ribosomal skipping 2A protein in the lymphovenous endothelium to facilitate downstream live imaging. A human SHOC2 WT plasmid was used under the same conditions to ensure that any observed phenotype was not due to an overexpression of wild-type SHOC2 protein (Figure S1A).

Initial experiments were performed with grossly phenotypic larvae. Injections of the various constructs were performed into Tg(mrc1a:EGFP)^y251 zebrafish, in which the lymphatic and venous endothelium are labeled with EGFP, allowing for vessel-specific phenotypes to be observed. No significant differences in survival rate were observed between uninjected, SHOC2 WT, and SHOC2 p.(Ala308Val) expressing embryos (Figure S1B). At 5 dpf, uninjected and SHOC2 WT expressing larvae exhibited pericardial edema at 5.6% (n=216) and 8.5% (n=80), respectively (Figure S1C, black arrow, Figure S1D). However, SHOC2 p.(Ala308Val) expressing larvae showed a significant increase in pericardial edema compared to wild-type controls (23%, n=76, p=0.0079, Figure S1D). The percent of larvae with trunk edema (presence of more extensive edema around the trunk) also increased from 2.5% (n=216) in uninjected control larvae and 2% (n=80) in SHOC2 WT larvae to 15.5% in SHOC2 p.(Ala308Val) larvae (n=76, p=0.0031, Figure S1E). From 2dpf to 5dpf, SHOC2 p.(Ala308Val) expressing larvae exhibited higher rates of body axis defects than uninjected and WT siblings (Figure S1I). By 5dpf, 21% of SHOC2 p.(Ala308Val) expressing larvae had body axis defects, compared to approximately 6% of SHOC2 WT larvae and 3% of uninjected larvae (n=76, p=0.0124, Figure S1F). There was no significant change in pericardial edema or body axis defects over time (Figure S1H-I). Finally, to quantify circulation, the presence or absence of red blood cell movement through the dorsal aorta and caudal vein plexus was observed. In all cases, pulsatile movement of red blood cells was observed in the blood vasculature when flow was absent, evidence of retained cardiac muscle function. However, at 5dpf, 15% (n=76) of SHOC2 p.(Ala308Val) expressing larvae had a cessation of blood flow. This was a significant increase compared to 3% (n=216) and 4% (n=80) observed in uninjected and SHOC2 WT expressing larvae (p=0.0068, Figure S1G).

SHOC2 p.(Ala308Val) causes dilations of the thoracic duct

To more specifically evaluate both blood and lymphatic vascular phenotypes, the experiment was repeated in a transgenic line expressing a blue fluorescent protein in the lymphovenous vessels and a green fluorescent protein in the arterial and venous vasculature, Tg(mrc1a:mtagBFP2-CAAX)^y722; Tg(kdrl:eGFP)^la116. To ensure that general toxicity of the variant was not a factor, that phenotypes were not secondary to edema,³⁸ and that maldevelopment of early vascular structures was not influencing later lymphatic development, larvae without edema, body axis defects, and normal circulation were selected at 2 dpf for further phenotyping.

There was no significant difference in the diameter of the lateral facial lymphatic between the SHOC2 WT expressing larvae and the SHOC2 p.(Ala308Val) expressing larvae at either 3 dpf or 5 dpf (Figure 4A, C, E, F). The development of the parachordal line appeared typical (Figure 4B). There was a significant increase in the diameter of the thoracic duct in the SHOC2 p.(Ala308Val) expressing larvae, compared to SHOC2 WT expressing larvae (p=0.0498, Figure 4D,G). There was no significant difference in the caudal plexus or the caudal aorta (Figure 4H,I). There was a significant decrease in the diameter of the posterior cardinal vein at 5 dpf (Figure 4J).

Figure 4: Expression of SHOC2 p.(Ala308Val) in zebrafish larvae causes dilation of the thoracic duct.

A-D) Representative images and quantification of the lateral facial lymphatic for each cohort at 3 dpf (A, E) or 5 dpf (C, F) where N equals the number of fish measured. B) Representative images of the trunk and parachordal line at 3 dpf. Representative imaging (D) and quantification of the of the diameter of the thoracic duct (G), caudal vein plexus (H), caudal aorta (I), and posterior cardinal vein at 5 dpf. White line outlines thoracic duct. All statistics calculated by ANOVA with multiple comparisons. Each dot represents the average of the measurements from a single embryo. p-values are represented on each graph.

SHOC2 p.(Ala308Val) may activate PI3K/mTOR pathway through non-canonical signaling

Due to the patient’s lack of response to trametinib, we hypothesized that dysregulated PI3K/mTOR signaling could be driving the patient’s phenotype rather than ERK. To investigate this, western blots were performed from lysates of 5 dpf SHOC2 WT or SHOC2 p.(Ala308Val) expressing larvae that had a grossly observable phenotype beginning at 2 dpf. Relative basal levels of phosphorylated-ERK in the SHOC2 p.(Ala308Val) larvae was not significantly different compared to SHOC2 WT larvae (0.89-fold change, ns) (Figure S1J, K). However, there was an increase in both phosphorylated AKT (3.27-fold change, p<0.0001) and phosphorylated-S6 (8.34-fold change, p=0.0045), both of which are downstream effectors of PI3K/mTOR signaling in Ala308Val larvae compared to WT (Figure S1J, L-M).

DISCUSSION

Here we report a patient with complex single-ventricle congenital heart disease and CCLA who failed to respond to traditional trametinib therapy with genetic testing identifying a variant of uncertain significance in SHOC2 (c.923C>T p.(Ala308Val)). Lymphatic imaging findings were consistent with those of other patients with RASopathies, thus the SHOC2 VUS was suspected to be pathogenic. To confirm this, the variant was globally expressed in the developing embryo and was found to disrupt conversion/extension movements similar to a known pathogenic variant, which is consistent with mTOR/MAPK-driven pathogenicity as previously described and suggesting pathogenicity of this SHOC2 variant.^35-37

To more specifically define lymphatic malformations related to the SHOC2 p.(Ala308Val) protein, we expressed it in the developing lymphovenous endothelium of zebrafish. Initially, we observed a heterogeneous and severe phenotype resulting in pericardial edema, body axis defects, dilation of the lateral facial lymphatic and malformations of the thoracic duct. To ensure that general toxicity or edema of the variant was not driving the phenotype, “well-appearing” larvae were selected for analysis and demonstrated a significantly increased diameter of the thoracic duct at 5 dpf. This confirmed that the SHOC2 p.(Ala308Val) protein was indeed inducing a lymphatic phenotype and is likely contributing to the proband’s phenotype. A significant decrease in the posterior cardinal vein was also observed. It could be that there are fewer cells contributing to the vein and more to the thoracic duct or that there is a difference in proliferation or cell size. Based on the experiments performed, it cannot be excluded that this may be due to the venous progenitors that contribute to the thoracic duct. Further work, including nuclei counting experiments combined with time lapse imaging, are needed to determine the cellular mechanisms of how different mutant SHOC2 proteins affect vascular development. A limitation of this model is that the mutant protein was expressed in both lymphatic and venous tissue. Therefore, future work could be performed to isolate the effect of the mutant protein solely on one tissue type or the other using different zebrafish promoters such as dab2 or cdh6.^39-40

Notably, this is the first reported case of a pathogenic variant in SHOC2 causing CCLA. Symptoms associated with lymphatic defects have been reported before. One previous study described a patient with increased nuchal translucency while another described a patient with severe fetal hydrops both due to SHOC2 p.(Ser2Gly).^41-42 Additionally, a previous study in zebrafish showed that CRISPR-mediated knock-out of SHOC2 resulted in larvae with ocular and trunk edema.⁴³ A recent pre-print also supports our conclusions by showing that SHOC2 is necessary for lymphangiogenesis in a loss of function in vivo zebrafish model.⁴⁴

While there is functional evidence that this SHOC2 variant affects lymphatic development, the patient nor the patient’s mother have the classical features associated with NSLH (though the mother has not been formally evaluated). This could be due to variable expressivity as is observed in other RASopathies, however this has not yet been reported in association with SHOC2. Future work could express this SHOC2 variant under a cardiac promoter to examine its role in potentiating the cardiac phenotype observed in this proband. It is known, however that individuals with other variants in SHOC2 can exhibit atrial and ventricular septal defects similar to the patient presented in this report.¹⁰ It is possible that the SHOC2 variant may impact lymphatic development in a genotype-phenotype specific manner.

Additionally, the contribution of the pathogenic variant in GLMN or the VUS in ZFPM2 in lymphatic disease is unclear. Given the mild lymphatic phenotype in our model, it is possible these variants contribute to this patient’s disease. Pathogenic variants in GLMN cause glomuvenous malformations with 90% penetrance, though these were not seen in the patient or affected parent.⁴⁵ A previous case report also discussed two patients with GVMs and severe chylothorax, however genetic testing to confirm a GLMN variant was not performed.⁴⁶ Glmn knockout mouse embryos develop pericardial effusions at an increased rate.⁴⁷ Pathogenic variants in ZFPM2 are known to cause congenital heart disease including double inlet right ventricle and Tetralogy of Fallot.³¹ The VUS in ZFPM2 has not previously been reported in ClinVar or the literature, and other variants present at this and surrounding residues in gnomAD are predicted to be benign. Further studies to evaluate the pathogenicity of the ZFPM2 VUS variant or the role of GLMN in lymphatics may be helpful to understand if there is a polygenic effect on the lymphatics in this patient.

Our data suggest that mutant SHOC2 may be acting non-canonically through PI3K/AKT signaling, which is also supported by the lack of response to trametinib. Previous work has introduced the possibility of a non-canonical role of SHOC2 in PI3K/mTOR signaling through direct binding of p110α in HEK293T cells.⁴⁸ It is possible that this mutant SHOC2 protein may be disrupting its ability to interact with PP1c and therefore preventing the initiation of RAS signaling through dephosphorylation of RAF. As a result, mTOR signaling may be compensating for lost RAS signaling leading to the results we found in our model. The role of SHOC2 in regulating PI3K/mTOR signaling has not been well characterized and should be further explored. These results suggest that additional therapeutics such as the mTOR inhibitor sirolimus or the PI3Kα inhibitor alpelisib could be considered. Sirolimus has also been reported to improve cardiac remodeling outcomes in a patient with Noonan syndrome with multiple lentigines (MIM #151100).⁴⁹ Additionally, a recent preprint demonstrated that mTOR signaling may be downstream of SHOC2 signaling in human lymphatic endothelial cells.⁴⁴

In conclusion, SHOC2 is a novel cause of CCLA. The SHOC2 p.(Ala308Val) mutant protein may act non-canonically through mTOR signaling to drive the lymphatic malformation pathogenesis, which has significant implications for therapy. Future investigation is needed to elucidate the molecular mechanisms by which the more common SHOC2 p.(Ser2Gly) variant may contribute to the lymphatic malformations previously reported and investigate the response to therapeutics for both variants to inform treatment for individuals with SHOC2-related diseases.

Supplementary Material

Supplemental information contains one table containing the gene lists for the genetic testing panels (Table S1), one table with the antibodies used in this study (Table S2), one table with in silico prediction tool results (Table S3), and one figure (Figure S1) containing initial phenotyping and western blot data from the more severe phenotype larvae. There is also one supplemental movie from the proband’s DCMRL.

Data Availability:

All experimental and clinical data that support the above findings are discussed in the main text and/or supplemental information sections of this manuscript. Additional clinical information is not available due to participant privacy protections. Please address clinical correspondence to Dr. Gregory Adamson.