Pathogenic de novo variants in PPP2R5C cause a neurodevelopmental disorder within the Houge-Janssens syndrome spectrum

Abstract

Pathogenic variants resulting in protein phosphatase 2A (PP2A) dysfunction result in mild to severe neurodevelopmental delay. PP2A is a trimer of a catalytic (C) subunit, scaffolding (A) subunit, and substrate binding/regulatory (B) subunit, encoded by 19 different genes. De novo missense variants in PPP2R5D (B56δ) or PPP2R1A (Aα) and de novo missense and loss-of-function variants in PPP2CA (Cα) lead to syndromes with overlapping phenotypic features, known as Houge-Janssens syndrome (HJS) types 1, 2, and 3, respectively. Here, we describe an additional condition in the HJS spectrum in 26 individuals with variants in PPP2R5C, encoding the regulatory B56γ subunit. Most changes were de novo and of the missense type. The clinical features were well within the HJS spectrum with strongest resemblance to HJS type 1, caused by B56δ variants. Common features were neurodevelopmental delay and hypotonia, with a high risk of epilepsy, behavioral problems, and mildly dysmorphic facial features. Head circumferences were above average or macrocephalic. The degree of intellectual disability was, on average, milder than in other HJS types. All variants affected either substrate binding (2/19), C-subunit binding (2/19), or both (15/19). Five variants were recurrent. Catalytic activity of the phosphatase was variably affected by the variants. Of note, PPP2R5C total loss-of-function variants could be inherited from a non-symptomatic parent. This implies that a dominant-negative mechanism on substrate dephosphorylation or general PP2A function is the most likely pathogenic mechanism.

Keywords: PP2A; PPP2R5C; PPP2R5D; autism; developmental delay; epilepsy; intellectual disability; macrocephaly; neurodevelopmental disorder; protein phosphatase 2A.

Figure 1

Facial photographs of ten individuals with de novo PPP2R5C missense variants Numbers of individuals and variants are as indicated in Table 1. Framed pictures are from the same individual, and in three individuals from different time points. P6: (I) and (J) are from age 20 months, (K) from age 7 years; P30: (L) is from age 7 years, (M) and (N) from age 14 years; P25: (O) is from age 8 years, and (P) and (Q) from age 14 years.

Figure 2

The representative face of ten PPP2R5C subjects generated by GAN inversion

Figure 3

Missense variant hotspots in PPP2R5C (A) Lollipop diagram showing the location of pathogenic (in red) and potentially pathogenic (loss-of-function) variants (in black) within B56γ1. (B) Structural representation of the PPP2R5C variants reported in this study. Ribbon diagram of PP2A B56γ1 (in gray) and C subunit (in sepia) based on the published crystal structure of the PP2A B56γ1 holoenzyme by Cho and Xu (PDB: 2IAE) were visualized in Molsoft MolBrowser v.3.9-2d software (ICM-Browser-Pro). Both B56γ1 and C lie on top of the horseshoe-shaped A subunit that was omitted from the structural representation to increase overall clarity. B56γ1 consists of 8 pseudo-HEAT repeats, with each pseudo-HEAT repeat harboring two anti-parallel α helices connected with an intra-repeat loop (IL). The large majority of pathogenic PPP2R5C variants affect amino acids residing in IL1, IL2 (acidic loop), IL3, IL7, and IL8 that are all facing the C subunit, and with the acidic loop (IL2) making direct contacts with the C subunit catalytic site. These intra-repeat loops are all highly conserved between B56γ and B56δ but are considerably differently affected in PPP2R5C- versus PPP2R5D-affected individuals (altered amino acids are highlighted in red; amino acids constituting the IL loops are underlined).

Figure 4

Functional characterization of PPP2R5C variants (A) Representative immunoblots assessing binding of liprin-α1, PP2A A subunit, and PP2A C subunit to B56γ or its variants. GFP-tagged WT or variant B56γ subunits were purified from transfected HEK293T cells by GFP pull-down, and coprecipitation of endogenous PP2A-A, PP2A-C, and liprin-α1 was determined by immunoblotting (Neg. Ctr., GFP-only transfected condition). (B–D) Quantification of binding abilities of GFP-tagged B56γ variants to the PP2A A subunit (B), C subunit (C), and the SLiM-containing PP2A substrate liprin-α1 (D). Immunoblot results were quantified and, for each individual variant, calculated relative to the immunoblot signals obtained for WT GFP-B56γ (present on the same blot and set at 100% in each independent experiment). Results represent the average values of at least four independent biological replicates (n ≥ 4) ± SD. A one-sample t test with FDR correction for multiple testing (compared to 100%, i.e., WT value, dotted line) was used to determine statistical significance of the observed binding differences between each individual variant and WT (^∗p ≤ 0.05, ^∗∗p ≤ 0.01, ^∗∗∗p ≤ 0.001, ^∗∗∗∗p ≤ 0.0001). The colors in these graphs correspond to the colors used to indicate the different loops and protein domains of B56γ in Figure 3.

Figure 5

Phosphatase activity of PPP2R5C variants Comparison of GFP-bound activities of PPP2R5C variants measured on a non-specific, commercial phospho-peptide (A) or on the indicated phospho-sites of established PP2A-B56γ substrates: AKT (B), ERK (C), and p70S6K (D) (phospho-peptide sequences: Table S3). PP2A activity was calculated relative to the GFP input of each variant (quantified by immunoblotting). The graphs represent the average PP2A activity ± SD for the PPP2R5C variants in comparison to activity of WT B56γ (set at 100% in each independent experiment, dotted line). Malachite green was used to quantify the released free phosphate. Results represent the average values of at least three independent biological replicates (n ≥ 3) ± SD. A one-sample t test with FDR correction for multiple testing (compared to 100%, i.e., WT value) was used to determine statistical significance of the observed differences between each variant and the WT value of 100% (^∗p ≤ 0.05, ^∗∗p ≤ 0.01, ^∗∗∗p ≤ 0.001, ^∗∗∗∗p ≤ 0.0001). The colors in these graphs correspond to the colors used to indicate the different loops and protein domains of B56γ in Figure 3.

Figure 6

The clinical spectrum in the individuals with pathogenic PPP2R5C variants About half of these individuals have mild intellectual disability/developmental delay; in the other half, the delay is moderate to severe (A). Growth parameters are within the normal range, except for an increased occipito-frontal head circumference (OFC) (B). Three-quarters of these individuals had neonatal concerns, such as feeding problems and hypotonia (C), and about half of them show behavior problems (D) or epilepsy (E).

Figure 7

Head circumferences (OFCs) in the individuals with pathogenic PPP2R5C variants Only one individual had microcephaly, while all the other head circumferences were from 0 to +6 SDs, and 14/26 OFCs were clearly in the macrocephalic range (>+2.5 SDs). For 3/5 variants with OFC observations at different ages (Glu122Lys, Ala130del, and Trp131Arg), there is a positive correlation with the age of the individuals, indicated by the three colored lines, one for each variant. The colors in this graph correspond to the colors used to indicate the different loops and protein domains of B56γ in Figure 3.