Hypomorphic mutations of SEC23B gene account for mild phenotypes of congenital dyserythropoietic anemia type II.

Congenital dyserythropoietic anemia type II, a recessive disorder of erythroid differentiation, is due to mutations in SEC23B, a component of the core trafficking machinery COPII. In no case homozygosity or compound heterozygosity for nonsense mutation(s) was found. This study represents the first description of molecular mechanisms underlying SEC23B hypomorphic genotypes by the analysis of five novel mutations. Our findings suggest that reduction of SEC23B gene expression is not associated with CDA II severe clinical presentation; conversely, the combination of a hypomorphic allele with one functionally altered results in more severe phenotypes. We propose a mechanism of compensation SEC23A-mediated which justifies these observations.

Introduction

Congenital dyserythropoietic anemia type II (CDA II, OMIM 224100) is a genetic hyporegenerative anemia characterized by ineffective erythropoiesis and distinct morphological abnormalities of the erythroblasts in the bone marrow (BM). Anemia of variable degree, jaundice and splenomegaly are common clinical findings [1]. This condition belongs to COPII-related human genetic disorders [2]. It is due to mutations in SEC23B (chr 20p11.23), a component of COPII complex, the core trafficking machinery of the endoplasmic reticulum-Golgi [3]. Approximately 60 different causative mutations have been described, localized along the entire coding sequence of the gene [1,4-6]. The most frequent are nucleotide substitutions (75% missense/nonsense), whereas frameshift and splicing mutations were observed in 15% and 10% respectively. The vast majority of patients have two mutations (in the homozygous or compound heterozygous state), according to the pattern of autosomal recessive inheritance. In no case homozygosity or compound heterozygosity for two nonsense mutations was found, a situation likely to be lethal. However, few cases with two hypomorphic mutations have been described so far [4,5].

Here we characterize three novel CDA II cases, two of them with fully hypomorphic genotype. We demonstrated a compensatory mechanism SEC23A-mediated of SEC23B hypo-expressed alleles.

Material and methods

Patients and mutational screening

Diagnosis of CDA II was based on history, clinical findings, laboratory data, morphological analysis of aspirated bone marrow and whenever possible on evidence of hypoglycosylated band 3 by SDS-PAGE. Samples were obtained after informed consent for the studies, according to the Declaration of Helsinki. Whenever possible, relatives were investigated. Genomic DNA and mutational screening were performed as previously described [4].

In silico and ex vivo analyses on mRNA

Prediction analyses for splice site mutations were performed by web server tools, splice site prediction by neural network (http://www.fruitfly.org/seq_tools/splice.html) and human splicing finder (http://www.umd.be/HSF/) (Table 2).

Table 2

In silico analysis of intronic mutations.

Nucleotide changea	Exon intron	Splice site type	Predicted effect on mRNA
			BDGP		HSF
			Scoreb	Δ (%)	Scoreb	Δ (%)
c.834 + 3A>C	7–8	Donor	0.93 > null	− 100	8.88 > 3.31	− 62.7
c.221 + 163A>G	2–3	Cryptic donor	Null > 0.99	+ 100	1.79 > 8.95	+ 400
c.1404 + 5G>A	12–13	Donor	0.99 > 0.66	− 33.3	10.67 > 6.6	− 38.1
c.221G>A	2	Donor	0.42 > null	− 100	5.53 > − 1.37	− 124.8

BDGP, Berkeley Drosophila Genome Project (http://www.fruitfly.org/seq_tools/splice.html).

HSF, Human Splicing Finder (http://www.umd.be/HSF/).

Nucleotides are numbered from the A of the ATG initiation codon (ENST00000377475).

WT sequence score > mutated sequence score.

RNA isolation from peripheral blood mononuclear cells (PBMCs), cDNA preparation and quantitative real-time (qRT)-PCR were performed as described [7]. Relative gene expression was calculated by using the 2− ΔCt method, while the mean fold change = 2− (average ΔΔCt) was assessed using the mean difference in the ∆Ct between the gene and the internal control [8].

SEC23B coding sequence was covered by five overlapping PCR fragments and amplified by KAPA2G Robust HotStart ReadyMix (Kapa Biosystems, Cape Town, South Africa). Sequence primers are available on request (achille.iolascon@unina.it).

Protein analysis

Protein extraction from PBMCs and western blotting (WB) were performed as previously described [7,9]. A specific rabbit anti-SEC23B antibody (1:500) (BioLegend, San Diego, CA) was used. Mouse anti-β-actin antibody (1:5000) (Sigma-Aldrich, Milan, Italy) was used as the control for equal loading. WB analysis on gradient was performed by precast gel (Biorad, Milan, Italy). Particularly, 60 μg of total extract proteins was loaded into each lane and was separated by gradient 4–15% SDS PAGE bisacrylamide gel, followed by transfer to PVDF membranes (Biorad, Milan, Italy).

Results

Clinical and genetic features

The clinical features of the three probands are presented in Table 1. Onset symptoms (anemia, jaundice) were in the first decade of life. At diagnosis, they exhibited a normocytic anemia with a reticulocytosis not corresponding to the degree of anemia. Patient B-II.1 was firstly diagnosed with hereditary spherocytosis. She subsequently underwent splenectomy with a slight improvement of anemia. BM examination of patients A-II.1 and C-II.1 showed erythroid hyperplasia, with bi- and tri-nucleated erythroblasts (Fig. 1s). Patients A-II.1 and B-II.1 exhibited a milder phenotype than patient C-II.1, with a higher absolute reticulocyte count (Table 1).

Table 1

Clinical data of CDA II patients.

	A-II.1	B-II.1a	C-II.1
Gender (male/female)	F	F	M
Age at diagnosis (years)	11	35	9
Onset of symptoms (years)	8	At birth	3
Consanguinity (yes/no)	n.a.	No	n.a.
Geographical origin	Sweden	Italy	USA
RBC (× 106/μL)	3.1	3.3	3.0
Hb (g/dL)	8.8	10.5	7.2
Ht (%)	27.0	29.0	23.0
MCV (fL)	87.0	90.0	77.0
MCH (pg)	n.a.	32.0	24.2
MCHC (g/dL)	33.2	35.5	31.4
RDW (%)	n.a.	17.3	20.2
Retics abs count (× 103/μL)	103	111	34
PLT (× 103/μL)	204	311	334
Ferritin (ng/mL)	45	168	82
Transfusion N/year	1/1	n.a.	–
Splenomegaly (yes/no)	Yes	Yes	Yes

n.a., not available.

Data before splenectomy.

We found five novel nucleotide replacements in SEC23B: three intronic mutations (c.834 + 3A>C; c.221 + 163A>G; c.1404 + 5G>A), one nucleotide insertion (c.1419_1423insC, p.I473Ifs*47) and one G>A transition (c.221G>A, p.C74Y). None of these mutations is present in the 1000 Genome project. Accordingly to recessive inheritance pattern, the patients were compound heterozygotes for two mutations (Fig. 1A).

Fig. 1

SEC23B novel mutations and analysis of their effect on gene and protein expressions.

Panel A. Inheritance pattern of SEC23B mutations in three unrelated patients were showed. All mutations are novel, except the nucleotide replacement c.1489C>T (p.R497C).

Panel B. SEC23B relative expression respect to the reference gene, β-actin, is shown. All patients exhibited a reduced SEC23B expression compared to control group (CTRs, 4.88 ± 0.37; A-II.1, 0.16 ± 0.01, p < 0.001; B-II.1, 2.59 ± 0.09, p = 0.006; C-II.1, 3.49 ± 0.23, p < 0.0001). Data are presented as mean ± SE. p Value has been calculated by Student t test.

Panel C. WB analysis showed a reduced expression of SEC23B in CDA II patients of comparable extent to that observed by qRT-PCR. The histograms show the densitometric quantification. Sizes (in kDa) are on the left.

Role of the intronic mutations on SEC23B gene and protein expression

In the first case A-II.1, the association of two splice site mutations led to a marked reduction of SEC23B expression at mRNA and protein levels (Figs. 1B–C). Particularly, the c.834 + 3A>C mutation is predicted to abolish the intron 7–8 donor splice site, while the c.221 + 163A>G to create a cryptic donor site (Table 2). Accordingly, we found an RNA decay of the first allele in sequenced cDNA (Fig. 2A), and a reduced expression of the second one (Fig. 2s).

Fig. 2

gDNA and cDNA sequencing analysis.

Panel A. Analysis of both SNPs rs41309927 G>A and rs2273526 C>G highlighted the presence of a double peak (heterozygous state) in the electropherogram from gDNA of proband A-II.1. Conversely, cDNA sequencing just showed the presence of the minor allele nucleotides, suggesting an RNA decay mechanism for the allele c.834 + 3A>C, which is in cis with major alleles of both SNPs (see Fig. 1A).

Panel B. Analysis of frameshift mutation c.1419_1423insC, p.I473Ifs*47 in proband B-II.1 highlighted the insertion of the base C and the consequent frameshift in genomic DNA (gDNA); on the contrary, the mutation was not found in sequenced cDNA, suggesting an RNA decay mechanism for this allele. Sequencing analysis of SNP rs2273526 C>G in both gDNA and cDNA confirmed this suggestion (see Fig. 1A).

Panel C. Analysis of the mutation c.221G>A, p.C74Y in CDA II patient C-II.1 highlighted the presence of a double peak (heterozygous state) in the electropherogram from gDNA; conversely, the mutation was not found in sequenced cDNA, suggesting an RNA decay mechanism for this mutated allele. Sequencing analysis of SNP rs17807673 C>T in both gDNA and cDNA confirmed this suggestion (see Fig. 1A).

Conversely, patient B-II.1, compound heterozygous for the splice site (c.1404 + 5G>A) and the frameshift (c.1419_1423insC) mutations, exhibited a mild reduction of mRNA expression compared to healthy subjects (approximately 50%) (Fig. 1B). WB analysis showed comparable results (Fig. 1C). However no protein product of lower molecular weight was found as an effect of frameshift mutation, which could lead to the formation of a truncated protein of 519 amino acids (predicted molecular weight: 57.8 KDa) (data not shown), leading to the hypothesis of an RNA decay of this allele. Accordingly, we found the selective expression of the wild type allele in sequenced cDNA (Fig. 2B).

Patient C-II.1 is a compound heterozygous for two missense variations: c.1489C>T, p.R497C, already described as CDA II causative mutation [9]; c.221G>A transition, which resulted in the aminoacidic substitution C74Y. In this case, we suspected SEC23B expression levels similar to those observed in the control group. However, we found a reduction of SEC23B gene and protein expression of approximately 30% (Figs. 1B–C). Since c.221G>A transition occurs at the last nucleotide of the exon 2 donor splice site, we hypothesized that this mutation would also be predicted to affect splicing of SEC23B (Table 2). When we amplified the cDNA of the patient, no additional band on agarose gel was highlighted, leading to the supposition of a complete RNA decay of this mutated allele. Hence, we found the selective expression of the c.1489C>T allele in sequenced cDNA (Fig. 2C).

Molecular mechanism of SEC23B hypomorphic genotypes

In order to explain the slight reduction of SEC23B expression in the patient B-II.1, we studied the effect of c.1404 + 5G>A mutation on mRNA processing. Amplification of the specific exon regions, encompassing the mutation, of SEC23B cDNA from normal whole blood mRNA produced a single transcript of the expected size (560 bp). By contrast, cDNA of the patient highlighted the presence of two bands on agarose gel, one corresponding to the expected size fragment and an additional 90-bp shorter transcript, due to the skipping of exon 12, as confirmed by sequencing analysis of aberrant cDNA product (Fig. 3A). QRT-PCR analysis by specific primers showed a very low level of exon-12 skipped transcript expression when compared to SEC23B full transcript both in the proband (11%) and in the father (2%) (Fig. 3B). Accordingly, in silico analysis predicted a slight reduction of the score between wild type and mutated donor site sequence (Table 2). This incorrectly spliced RNA, however, retained the correct reading frame and encoded a SEC23B protein lacked 30 amino acids, with a predicted molecular weight of approximately 83 KDa (Fig. 3C).

Fig. 3

Molecular characterization of the c.1404 + 5G>A mutation and analysis of SEC23A compensative mechanism.

Panel A. Amplification of the exon region 8–13 of SEC23B cDNA from patient B-II.1 and control subjects. After the amplification by PCR, the products were separated on 1.5% agarose gel. The top band measures 560 bp, while the bottom band in lane 1 measures approximately 470 bp (exon-12 skipped transcript). On the left, sequencing analysis of the 470-bp PCR product is shown.

Panel B. SEC23B full transcript and exon-12 skipped isoform expression respect to the reference gene, β-actin, is shown. Patient B-II.1 exhibits a reduced expression of SEC23B full length transcript compared to 3 healthy subjects (B-II.1, 2.50 ± 0.11; CTRs, 4.50 ± 0.42), while father's proband shows an intermediate level of expression (B-I.1, 3.13 ± 0.08). Exon-12 skipped transcript is expressed at 11% in the proband (0.27 ± 0.05) and at 2% in her father (0.07 ± 0.01). No aberrant transcript was found in the controls (0%). Data are presented as mean ± SE.

Panel C. The upper panel shows the schematic representation of the 30 amino acid loss in SEC23B protein isoform derived from exon-12 skipped transcript. Below, detection of SEC23B protein is shown. Sizes (in kDa) are on the right. * SEC23B protein product from exon-12 skipped transcript.

Panel D. SEC23A relative expression respect to SEC23B gene is shown. Patient A-II.1 exhibited a strong upregulation in respect with the paralog SEC23B (fold change 4.55 ± 0.39). Similarly, patient B-II.1 exhibited an upregulation of approximately 3 fold in respect with the paralog SEC23B (fold change 2.85 ± 0.19). No difference in gene expression between the two paralogs has been found in patient C-II.1 (fold change 1.07 ± 0.01). Data are presented as mean ± SE.

When we analyzed SEC23A expression in all three patients, we found an upregulation of approximately 4 and 3 fold in respect with the paralog SEC23B in patients A-II.1 and B-II.1, respectively. Conversely, no compensatory effect of SEC23A expression has been found neither in C-II.1 patient nor in control subjects (Fig. 3D).

Discussion

This study represents the first description of molecular mechanisms underlying SEC23B hypomorphic genotypes. The inheritance pattern of the mutations here described confirms the allelic heterogeneity of this condition, as the most of causative variations are inherited as private mutations. Our analyses suggested that the association of two hypomorphic alleles led to a strong reduction of SEC23B expression, without generating severe clinical presentation. Indeed, patients A-II.1 and B-II.1 exhibited a milder phenotype compared to patient C-II.1. Of note, they share a clinical presentation comparable with a previously described CDA II case, characterized by a similar genotype [4]. On the other side, clinical presentation of patient C-II.1 fully matched with CDA II cases with one missense and one nonsense mutation, according to previous genotype–phenotype correlation study [10]. Moreover, the molecular mechanism of this patient could explain the severe phenotype of some patients with two missense mutations [10], since other exonic mutations could impair splice sites.

We proposed a compensatory mechanism SEC23A-mediated which could balance the reduced expression of its paralog in those patients with low SEC23B expression. Taken together, our observations explain how a fully hypomorphic genotype could result in a milder CDA II phenotype. Moreover, they confirm the hypothesis that the total absence of SEC23 proteins is supposed to be lethal. This is in agreement with studies on zebrafish morphants which showed that both Sec23 genes carry specific but partially redundant roles, at least in craniofacial cartilage maturation [11]. However, it seems that COPII-related disorders could be also due to the defective transport of special tissue-specific cargoes beyond to the differential, tissue-specific expression of COPII paralogs [12]. Understanding of the role of SEC23A–B paralogs in humans may provide a means of therapeutic intervention by modulating their expression.

Authorship

RR and AI designed and conducted the study, and prepared the manuscript; CL performed cDNA and qRT-PCR analyses; MRE performed western blotting analysis and sequencing analysis; AG and FrV collected clinical data; TE and EY cared for the patients; FV did the routine laboratory tests.

Disclosure

The authors declare no conflict of interest.