Phenotypical heterogeneity in RAG‐deficient patients from a highly consanguineous population

Summary

Mutations affecting recombination activation genes RAG1 and RAG2 are associated with variable phenotypes, depending on the residual recombinase activity. The aim of this study is to describe a variety of clinical phenotypes in RAG‐deficient patients from the highly consanguineous Egyptian population. Thirty‐one patients with RAG mutations (from 28 families) were included from 2013 to 2017. On the basis of clinical, immunological and genetic data, patients were subdivided into three groups; classical T^–B^– severe combined immunodeficiency (SCID), Omenn syndrome (OS) and atypical SCID. Nineteen patients presented with typical T^–B^–SCID; among these, five patients carried a homozygous RAG2 mutation G35V and five others carried two homozygous RAG2 mutations (T215I and R229Q) that were detected together. Four novel mutations were reported in the T^–B^–SCID group; three in RAG1 (A565P, N591Pfs*14 and K621E) and one in RAG2 (F29S). Seven patients presented with OS and a novel RAG2 mutation (C419W) was documented in one patient. The atypical SCID group comprised five patients. Two had normal B cell counts; one had a previously undescribed RAG2 mutation (V327D). The other three patients presented with autoimmune cytopaenias and features of combined immunodeficiency and were diagnosed at a relatively late age and with a substantial diagnostic delay; one patient had a novel RAG1 mutation (C335R). PID disorders are frequent among Egyptian children because of the high consanguinity. RAG mutations stand behind several variable phenotypes, including classical SCID, OS, atypical SCID with autoimmunity and T^–B⁺ CID.

Introduction

Defects in V(D)J recombination affect the formation of immunoglobulins (Igs) and T cell receptors (TCRs), leading to a block in differentiation of B and T cells and thus a combined immunodeficiency (CID) 1. The clinical presentation and immunological phenotypes varies depending on the type of genetic defect and whether it is a null mutation or a hypomorphic mutation with residual V(D)J recombination activity 1.

Null mutations cause severe combined immunodeficiency (SCID), with the absence of both B and T cells and preserved natural killer (NK) cells (T^–B^–NK⁺). Hypomorphic mutations enable synthesis of RAG proteins with a low residual activity; this partial defect is associated with a variety of phenotypes, including Omenn syndrome (OS) and CID 2, 3. Depending on the ethnicity, it is estimated that RAG1/2 mutations accounts for nearly 50% of patients with T^–B^–NK⁺ SCID patients 4.

It has been shown that some patients with hypomorphic RAG1/2 mutations are free of skin symptoms and have normal or slightly low B cell counts and Ig levels 3.

Infants with null mutations usually present early in life with a range of infections; if not treated, the condition is usually fatal within the first 2 years of life. Clinical presentations include oral candidiasis, persistent diarrhoea with growth impairment and/or interstitial pneumonitis 5, 6. These patients require immediate immune reconstitution. Haematopoietic stem cell transplantation (HSCT) in the first few months of life (without pre‐transplant conditioning) may result in successful T cell reconstitution 4, 7.

The clinical spectrum of RAG deficiency extends beyond those of classic SCID and OS; it includes milder, non‐classical CID phenotypes, such as recombination activation gene (RAG) deficiency with expansion of γδ T cells, RAG deficiency characterized by granulomatous lesions, early‐onset autoimmunity and isolated CD4⁺ lymphopenia. These forms are often referred to as ‘leaky’ or ‘atypical’ SCID 4, 8. In a study by Felgentreff et al. 9, ‘atypical’ SCID was defined as an immunodeficiency disease resulting from mutations in SCID‐causing genes in patients having a presentation other than typical SCID or OS and with a T cell count above 100/μl. Another retrospective study performed on a large number of patients by Shearer et al. 10 defined leaky SCID patients as those with a reduced number of CD3 T cells < 1000/µl (for age up to 2 years), < 800/µl (for age between 2 and 4 years) and < 600/µl (for age above 4 years), with absence of maternal engraftment and T cell function less than 30% of the normal.

Epigenetics and other factors should be considered when predicting the clinical phenotype 11, 12. Even within the same family, a certain genotype may result in different phenotypes presenting either as classical SCID/OS or SCID with expansion of γδ T cells 4. OS occurs in patients with SCID‐causing gene mutations and presents with erythroderma, lymphoproliferation and αβ T cell expansion in combination with severe infections and/or failure to thrive in the first year of life 4.

Based on the data presented in the literature, the spectrum of clinical presentations in atypical SCID includes 9:

• Increased susceptibility to severe infectious diseases: e.g. pneumonia, Pneumocystis jirovecii pneumonia, severe courses of viral infections, persistent mucosal candidiasis, skin or organ abscesses, bacterial sepsis and/or meningitis.

• Immune dysregulation: e.g. autoimmune cytopenia, autoantibody‐mediated diseases, lymphoproliferation (hepatosplenomegaly and lymphadenopathy), colitis and/or protracted diarrhea without reported pathogen, granulomata and skin rash.

• Lymphoma.

Due to the poorly defined clinical and immunological phenotype of atypical SCID these patients are sometimes diagnosed in adulthood, very late in life 9.

The present report describes the clinical characteristics, molecular diagnoses and outcomes in a series of infants with confirmed RAG1/2 mutations in Egypt during a 5‐year period and focuses on the different clinical phenotypes.

Patients and methods

A total of 31 patients from 28 different families were assessed between January 2013 and December 2017 in the Primary Immunodeficiency Clinic at Cairo University Children’s Hospital (Cairo, Egypt). Five patients (2, 3, 4, 17 and 25) were previously reported 13. Based on the clinical presentations and the immunological data, the 31 patients were divided into three subgroups: 19 typical T^–B^–NK⁺ SCID, seven OS and five atypical SCID/CID. Informed consent was obtained from the legal guardians of the participating patients and the study was approved by the local Institutional Review Board (IRB). The study was conducted in accordance with the principles of the Declaration of Helsinki.

Methods

SCID patients were diagnosed according to the criteria issued by the International Union of Immunological Societies (IUIS) 14 and the European Society of Immunodeficiency Disorders (ESID) 15. All patients with typical SCID had lymphopenia and early onset of infections, whereas the clinical diagnosis of OS was based on the presence of atopic or seborrheic dermatitis and/or erythroderma in the absence of maternal engraftment. Immunodeficient patients with RAG1/2 mutations associated with atypical clinical and/or laboratory data were included in the ‘atypical SCID’ group.

Immunophenotyping and Ig quantification

A differential complete blood count was performed. Immunophenotyping of blood lymphocytes and NK cells was performed using flow cytometry and specific monoclonal antibodies for T cells (CD3, CD4 and CD8), NK cells (CD56) and B cells (CD19). Antibodies were purchased from Beckman Coulter (Villepinte, France). Data were acquired on a CYTOMICS FC 500 flow cytometer and CXP software (version 2.2, Beckman Coulter). Serum concentrations of IgG, IgA and IgM were determined by nephelometry on a Nephstar protein analyser (Goldsite Diagnostic Inc., Shenzhen, China), whereas the serum IgE concentration was measured by enzyme‐linked immunosorbent assay (ELISA) (General Biologicals Corporation, Hsinchu, Taiwan; cat. no. S00062).

Molecular diagnosis

Genomic DNA was extracted using a QIAamp DNA blood minikit, according to the manufacturer’s instructions. RAG1/2 genes were sequenced in all patients as previously described 13. Polymerase chain reaction (PCR) products were sequenced on a 310 Genetic Analyzer (Applied Biosystems, Grand Island, NY, USA) using the same primers that were used to amplify the PCR fragments. Sequences were compared with the reference sequence published by the National Centre for Biotechnology Information (NM_000448.2 for RAG1 and NM_000536.3 for RAG2) and analysed using the Basic Local Alignment Search Tool (blast). Prediction of functional effects of amino acid substitutions was performed using the Polymorphism Phenotyping version 2 software tool (PolyPhen‐2) and sift software. Parents of affected children were sequenced in order to confirm heterozygosity for the detected mutations.

Whole‐exome sequencing (WES) for patients 27 and 28 was performed with Agilent’s SureSelect all Exon V6+UTR kit on a HiSeq4000 system (Illumina, San Diego, CA, USA) and the detected mutations were confirmed by Sanger sequencing.

The raw reads were first cleaned by removing adapter sequences, trimming low‐quality ends, and filtering reads with low quality (phred quality < 20). Sequence alignment of the resulting high‐quality reads to the human genome (GRCh37) was carried out using Bowtie2 (version 2.3.2).

Subsequently, SNVs and InDels were detected by GATK (version 3.7) according to the human best practices guidelines (June 2016 version). We used SnpEff (version 4.2) to annotate the variants.

Genetic counselling was offered to families with affected children. Nine families came for prenatal diagnosis, which was performed in 12 pregnancies by analysing fetal DNA extracted from chorionic villous sampling or amniotic fluid samples 16.

Statistical methods

Data were described statistically in terms of mean ± standard deviation (± s.d.), median and range or frequencies (number of cases) and percentages when appropriate.

Comparison between the three groups was performed using the one‐way analysis of variance (anova) test with Bonferroni two‐group comparisons. P‐values less than 0·05 were considered statistically significant. All statistical calculations were performed using computer program spss (Statistical Package for the Social Sciences, SPSS Inc., Chicago, IL, USA) release 15 for Microsoft Windows (2006).

Results

Typical SCID

Nineteen patients, 11 males and eight females, presented with classical T^–B^–SCID (patients 8–26). The mean age at onset was 1·9 months ± 1·5 s.d. and the mean age at diagnosis was 4·7 months ± 3·5 s.d. All but two patients (18 and 24) were born to consanguineous families.

The main presenting symptoms were recurrent pneumonia and persistent diarrhoea. All the patients in this group had marked lymphopenia, low serum Ig levels and very low or absent T and B cell counts. The clinical and laboratory data are summarized in Table 1.

Table 1.

Patients’ clinical and laboratory data

Pheno‐type	Patient No.	Gender	No. of Affected siblings	Age at		Presentation	Outcome	Lymphocytes Cells/µl (%)	CD3 Cells/µl (%)	CD4 Cells/µl (%)	CD19 Cells/µl (%)	CD56 Cells/µl (%)	Eosinophil%	IgE IU/ml	IgG mg/dl	IgM mg/dl	Mutant gene	Mutation
Reference 0‐12m				onset	Diagnosis			3400–9000/ (11–70%)	1900–5900 (49–76%)	1400–4300 (31–64%)	300–3000 (6–41%)	160–1100 (3–18%)	0–6%	< 150	286–1680	21–192
OS (n=7)	1	♀	2	2 m	6 m	Exfoliative erythroderma, pneumonia	No HLA matching performed, died at 9 m	1712 (16%)	1273 74·4%	841 49·1%	38 2·2%	207 12·1%	1	n.d.	671	252.9	RAG1	a,bp.R142X
	2	♂	0	40 d	4 m	Exfoliative erythroderma, alopecia, diarrhoea, oral thrush, hepatomegaly, lymphadenopathy, GVHD following non‐irradiated blood transfusion	No HLA matching performed, died at 4 m	3800 (10%)	3420 90%	760 20%	0 0%	106 2·8%	4	n.d.	< 149	23.8	RAG1	a,bp.E425del
	3	♀	1	40 d	40 d	Pneumonia, marked exfoliative erythroderma, oral thrush, hepatomegaly, lymphadenopathy	No matched related donor, died at 8 m	3770 (26%)	1662 44%	1304 34·6 %	8 0·2%	1139 30·2%	43	8	118	18	RAG1	a,bp.Q812X
	4	♀	0	At birth	33 m	Marked exfoliative erythroderma (itchy, patchy hypopigmentation), diarrhoea, oral thrush, pneumonia, draining ears, hepatomegaly, sepsis	No HLA matching performed, died at 33 m	1947 (33%)	531 27·3%	52 2·7%	77 4%	609 31·3%	33	98	819	292	RAG1	a‐bp.R975Q
	5	♂	3	1 m	1 m	Exfoliative erythroderma, diarrhea, sepsis, lymphadenopathy	Died	14 112 (48%)	12 446 88·2%	11 572 82%	28 0·2%	875 6·2%	0	14 700	508	129	RAG1	a,bp.R559S
	6#	♂	0	2 m	6 m	Exfoliative erythroderma, alopecia, diarrhoea, pneumonia, hepatomegaly, lymphadenopathy	Died	3772 (41%)	1777 47·1%	1728 45·8%	82 2·2%	460 12·2%	20	11·6	177	24	RAG2	a,bp.R148X
	7	♀	0	1 m	4 m	Exfoliative erythroderma, pneumonia, hepatomegaly	Died	3550 (40%)	1714 48·3%	887 25·5%	0 0%	71 2%	15	n.d.	n.d.	n.d.	RAG2	a,bp.C419W
Typical T‐B‐SCID (n=19)	8	♀	2	3 m	7 m	Pneumonia, diarrhoea, BCGitis	Haploidentical with parents	2400 (31%)	276 11·5%	214 8·9%	10 0·4%	1404 58·5%	1	n.d.	< 149	< 9·8	RAG2	a,bp.G35V
	9	♀	4	3 m	6 m	Pneumonia, oral thrush	Died (multi‐drug‐resistant Klebsiella pneumonia)	1750 (35%)	35 2%	28 1·6%	9 0·5%	368 21%	0	n.d.	n.d.	< 9·8	RAG2	a,bp.G35V
	10	♂	1	Screened at 11 d		Pneumonia, diarrhoea	Living	1400 (19%)	0 0%	0 0%	0 0%	862 61·6%	0	n.d.	139	< 9·8	RAG2	a,bp.G35V
	11	♂	2	5 m	7 m	Axillary LN, oral thrush, diaper rash	Died from sepsis and pneumonia	672 (16%)	2 0·2%	1 0·1%	5 0·7%	468 69·7%	22	n.d.	61·5	< 9·8	RAG2	a,bp.G35V
	12	♂	0	2 m	8 m	Pneumonia, diarrhoea, oral thrush, diaper rash	Living, haploidentical with his siblings	2436 (28%)	134 5·5%	109 4·5%	34 1·4%	1754 72%	2	n.d.	< 149	< 9·8	RAG2	a,bp.G35V
	13	♀	0	5 m	7 m	Pneumonia, diarrhoea	Prepared for matched sibling donor, HSCT	658 (14%)	35 5·4%	8 1·2%	6 0·9%	356 54·1%	2	n.d.	< 149	< 9·8	RAG2	a,bp.G95R
	14	♂	2	4 m	14 m	Pneumonia, chronic diarrhoea, diaper rash	Died at 15 m	1036 (28%)	238 23%	17 1·7%	1 0·1%	777 75%	1	n.d.	n.d.	124·1	RAG2	a,bp.T215I a,bp.R229Q
	15	♀	1	2 m	4 m	Chronic diarrhoea, oral thrush, diaper rash	Died at 3 m	945 (45%)	27 2·8%	22 2·3%	8 0·8%	161 17%	1	25	140	12·1	RAG2	a,bp.T215I a,bp.R229Q
	16^	♂	1	1 m	4 m	Pneumonia, diarrhoea	Died at 4 m from pneumonia	137 (11%)	0 0%	0 0%	3 2%	123 90%	1	n.d.	340	6	RAG2	a,bp.T215I a,bp.R229Q
	17^	♂	2	Screened at 15 d		Pneumonia, oral ulcers	Haploidentical HSCT, died at 15 m (severe GVHD, post‐transplant CMV infection)	684 (12%)	5 0·7%	3 0·4%	8 1·1%	382 55·8%	0	n.d.	n.d.	n.d.	RAG2	a,bp.T215I a,bp.R229Q
	18	♀	1	2 m	4 m	Pneumonia, oral thrush, bilateral draining ear	Died within few days of diagnosis	2640 (40%)	246 9·3%	185 7%	34 1·3%	1584 60%	0	n.d.	166·8	30·4	RAG2	a,bp.T215I a,bp.R229Q
	19	♀	0	1 m	1 m	Pneumonia, oral thrush, sepsis	Died	770/ (10%)	8 1%	4 0·5%	0 0%	92·5 12%	0	n.d.	415	< 9·8	RAG2	a,b p.F29S
	20#	♀	1	2 m	3 m	Pneumonia	Matched with the living brother, parents refused BMT	704 (11%)	3 0·5%	3 0·5%	0 0%	338 48%	3	n.d.	234	< 9·8	RAG2	a,bp.R148X
	21	♂	3	At birth	3 m	Diarrhoea, oral thrush, buttock large ulcer and excoriation	Haploidentical with parents, non‐identical with sister, died	736 (29%)	2 0·2%	1 0·1%	30 4·1%	453 61·6%	2	0·4	610	< 9·8	RAG1	a,bp.Q407H
	22ʺ	♂	0	3 m	4·5 m	Pneumonia, vomiting, fever	Died	370 (8·2%)	148 40%	85 23·2%	3 0·9%	19 5%	0	n.d.	n.d.	n.d.	RAG1	a,bp. N591Pfs*14
	23ʺ	♂	1	Screened at 11 d		GVHD after non‐irradiated exchange transfusion performed for neonatal jaundice	Died at 18 d	2555 (35%)	12 0·5%	3 0·1%	15 0·6%	1073 42%	0	n.d.	n.d.	n.d.	RAG1	a,bp. N591Pfs*14
	24	♂	0	40 d	8 m	Pneumonia, diarrhoea, oral thrush	Haploidentical with parents and sister, died during conditioning	840 (30%)	345 41·1%	270 32·1%	5 0·6%	271 32·3%	0	16·8	300	30	RAG1	ap.K621E bp.R829S fs*
	25	♀	1	10 d	1 m	Pneumonia, meningitis, BCGitis, oral thrush, enlarged inguinal and axillary LNs	Died (meningitis)	440 (22%)	44 10%	16 3·6%	2 0·5%	83 18·8%	14	n.d.	665	< 9·8	RAG1	a,bp.R829S fs*
	26	♂	1	0·5 m	8 m	Neonatal sepsis, pneumonia, diarrhoea, oral thrush	Living	1173 (17%)	1 0·1%	0 0%	1 0·1%	574 49%	2	n.d.	< 149	< 9·8	RAG1	a,bp.A565P
T‐B+ SCID (n=2)	27	♀	1	5 m	10 m	Pneumonia, diarrhoea, oral thrush, skin rash, hepatomegaly	Died	1722 (7%)	254 14·8%	182 10·6 %	430 25%	654 38%	69	n.d.	n.d.	n.d.	RAG1	a,bp.R841W
	28	♀	2	2 m	4 m	Vomiting, diarrhoea, oral thrush, diaper rash, draining ears	Died at 5 m	3800 (47%)	786 2%	775 20·4%	1387 36·5%	342 9%	2	13·4	110	< 9·8	RAG2	a,bp.V327D
Atypical SCID /CID (n=3)	29	♀	0	12 m	4 years	Repeated episodes of AIHA, exfoliative skin rash, diarrhoea, pneumonia, destructive granuloma in the tongue and mouth ulcers, emaciation	Living, mother pregnant in a healthy baby for cord blood transplantation	2190 (30%)	1292 59%	48 2·2%	98 4·5%	503 23%	2	8·5	> 2863	329·7	RAG1	ap.C335R bp. Q812X
	30	♂	0	8 m	2·5 years	Pneumonia, perianal cellulitis, thigh abscess, oral thrush, napkin dermatitis, repeated episodes of AIHA	Transplanted from his father (haploidentical HSCT)	1300 (26%)	299 23%	125 9·6%	291 22·4%	199 15·3%	1	18 100	1060	6·8	RAG1	a,bp. R973H
	31	♀	1	9 m	3 years	Pneumonia, repeated episodes of AIHA, exfoliative and itchy skin rash	Living, trying to find a suitable donor	5772 (37%)	110 1·9%	58 1%	58 1%	5195 90%	22	2·2	524·7	< 9·8	RAG2	a,bp.K127X

M = months; d =  days; CD =  cluster of differentiation; RAG =  recombination activating gene; BMT =  bone marrow transplantation; HSCT =  haematopoietic stem cell transplantation; AIHA =  autoimmune haemolytic anaemia; GVHD =  graft‐versus‐host disease; HLA =  human leucocyte antigen; LNs =  lymph nodes; CMV =  cytomegalovirus; #,^ = indicates siblings; a  =  one allele , b =  the other allele.

RAG1 mutations were detected in six patients (five had a homozygous mutation and one had a compound heterozygous mutation), while RAG2 homozygous mutations were detected in 13 patients. Seven mutations were previously reported in SCID patients, while four novel mutations were reported among this group of patients; three in RAG1 (p.N591Pfs*14, p.K621E, p.A565P; patients 22, 24 and 26, respectively) and one in RAG2 (p.F29S; patient 19).

Two homozygous missense mutations in RAG2 were observed to occur concomitantly in five patients (p.T215I and R229Q), whereas the same missense RAG2 mutation (p.G35V) was identified in five other patients. All mutations and their phenotypical presentation are summarized in Table 2.

Table 2.

Phenotypical presentation of different mutations

	Mutation
Gene	c.	p.	Patient no.	Patient phenotype	Previously reported
RAG1	424C>T	R142X	P1	OS	Atypical SCID/leaky SCID 17
					OS 17, 18
	1277_1279delAAG	E425del	P2	OS	13
	2434C > T	Q812X	P3	OS	13
			P29	Atypical SCID/CID
	2924G > A	R975Q	P4	OS	13
					OS 11
					CID‐G/A 19, 20
	1677G>C	R559S	P5	OS	OS 17, 21
	1221G>C	Q407H	P21	SCID	16
	1768_1769insTACC	N591Pfs*14	P22,P23	SCID	(Novel)
	1861A>G	K621E	P24	SCID	(Novel)
	2487_2488 delGA insTT	R829S fs*	P24,P25	SCID	13
	1693G>C	A565P	P26	SCID	(Novel)
	2521C>T	R841W	P27	T–B+SCID	Hypomorphic SCID with CMV infection 20
					Atypical SCID/OS 11
					T‐B+SCID 2
	1003T>C	C335R	P29	Atypical SCID/CID	(Novel)
	2918G>A	R973H	P30	Atypical SCID/CID	OS 22
RAG2	442C>T	R148X	P6	OS	OS 11
			P20	SCID
	1275C>G	C419W	P7	OS	(Novel)
	104G>T	G35V	P8,9,10,11,12	SCID	SCID 18, 23
					OS 23, 24
	283G>A	G95R	P13	SCID	OS 23
	644C>T	T215I	P14,15,16,17,18	SCID	OS 23
	686G>A	R229Q	P14,15,16,17,18	SCID	SCID 25
					OS 11, 22, 26, 27
	86T>C	F29S	P19	SCID	(Novel)
	980T>A	V327D	P28	T–B+SCID	(Novel)
	379A>T	K127X	P31	Atypical SCID/CID	SCID 5, 35

SCID = severe combined immunodeficiency; OS = Omenn syndrome = CID‐G/A: combined immunodeficiency with granuloma/autoimmunity; RAG = recombination activation gene.

The OS group

Seven patients, three males and four females (patients 1–7) with a diagnosis of OS were included. The mean age at onset among this group was 1·2 months ± 0·6 s.d. and the mean age at diagnosis was 7·9 months ± 11·2 s.d. Except for patient three, all patients in this group were born to consanguineous parents. All seven patients suffered from exfoliative dermatitis and pneumonia. Patient 2 suffered graft‐versus‐host following transfusion with non‐irradiated blood. Six of the seven patients died in the first few months of life, whereas patient 4 died at the age of 2·7 years. Five patients (2, 3, 5, 6 and 7) had normal lymphocyte counts, while two patients (1 and 4) presented with lymphopenia. All patients with OS had very low to undetectable B cell counts. Five patients had homozygous mutations in RAG1 and two had homozygous mutations in RAG2. All the mutations detected in this group were previously reported in OS patients, except for a novel missense mutation in RAG2 (p.C419W) detected in patient 7.

Atypical cases

This group included five patients (27–31). The mean age of onset was 7·2 months ± 3·8 s.d., the mean age at diagnosis was 25·6 months ± 18·3 s.d. and the mean diagnostic delay was 18·4 months.

Patients with RAG mutation and preserved B cells

Two female patients (27 and 28) born to two unrelated consanguineous families presented with diarrhoea, skin rash and pneumonia at an early age. Both had normal B cell counts and frequencies, although patient 28 had very low serum immunoglobulins. Both patients had homozygous missense mutations; one previously reported in RAG1 (patient 27) and the other, a novel mutation, was RAG2 (p.V327D; patient 28).

Atypical SCID/CID

The main presentation in these patients (29–31) was autoimmune haemolytic anaemia (AIHA), requiring repeated blood transfusions. AIHA was occasionally triggered by the administration of intravenous immunoglobulin infusions (in patient 30, whose blood group was AB+). All three patients had normal or high serum IgG levels; in addition, patient 30 also had a very high serum IgE level. The three patients displayed marked CD4⁺ cell lymphopenia and low to borderline B cell counts. In all three patients, the RAG1/2 mutations were identified relatively late (at age 2·5–4 years). Despite the severe clinical courses, these three patients are alive (aged 5, 4 and 4·2 years for patients 27–29, respectively). Two are being prepared for HSCT while patient 30 was transplanted from his haploidentical father. Patient 29 developed a necrotizing granuloma involving the tongue leading to partial sloughing; the symptoms were dramatically reduced by treatment with prednisone (Fig. 1). A novel missense mutation was detected in RAG1 (p.C335R) in patient 29.

Figure 1.

Necrotizing granuloma affecting the tongue and buccal mucosa in patient 29.

Comparing the different groups

There was a statically significant difference between the three groups regarding the age of onset (P = 0·0001) as well as the age of diagnosis (P = 0·0001), with the atypical SCID group presenting and diagnosed at an older age.

Discussion

Mutations in the RAG1 and RAG2 genes result in extremely variable phenotypes ranging from classical SCID 17 and OS 11, with the presence of oligoclonal T lymphocytes, leaky/atypical SCID, with varying T and B cell counts and the expansion of γδ T lymphocytes 20, often with cytomegalovirus infection, to delayed‐onset combined immunodeficiency with granuloma and/or autoimmunity (CID‐G/A) 28 and common variable immunodeficiency (CVID) 29.

During a period of 5 years, we were able to diagnose 31 patients with RAG mutations at the PID centre at Cairo University Children’s Hospital. The patients were subdivided into three groups according to their presentation. The age of onset of the manifestations was significantly higher in the atypical SCID group. There was a delay in diagnosis in all three groups, although this delay was significantly longer in the atypical SCID group. Despite a better awareness of the possibility of genetic diagnosis of PIDs in Egypt, the delay in diagnosis reported herein probably reflects that many cases go undiagnosed.

Although RAG mutations have been extensively described in different populations, they present differently among Egyptian children. For RAG1 (ENST00000299440.5), 1044 variants have been reported with approximately 300 variants documented to be pathogenic according to their SIFT and PolyPhen‐2 scores, while for RAG2 (ENST00000618712.4), 529 variants have been reported with approximately 120 variants considered pathogenic 30. In general, RAG1 mutations have been reported more frequently than RAG2 mutations in different studies from different populations, including distant and nearby countries: Saudi Arabia, China, Japan, Turkey, Serbia, East Slovakia and throughout Europe 3, 5, 6, 11, 12, 31, 32. In our cohort, RAG2 mutations were detected in 54·8% of the patients (17 of 31), who presented mainly with typical SCID with a distinctive pattern of inheritance. Two missense mutations in RAG2 (T215I and R229Q) were detected together in a homozygous form in five patients (14–18) from four presumably unrelated families that lived in areas distant from each other. The five patients presented with similar symptoms at a very early age. The T215I and R229Q mutations have been described previously, but occurring in isolation; the T215I mutation has been reported only once before in a Jewish patient presenting with OS, whereas the R229Q mutation has been reported in five patients of Jewish or Arab origin and in OS patients of various ethnic origins 22, 23. A study by Tirosh et al. 33 that measured the recombinase activity of RAG2 mutant proteins, revealed that the T215I variant had 67·2% of wild‐type recombinase activity, while the R229Q had only 8·9% residual activity. The T215I variant has a relatively high CADD‐PHRED score of 21·4 and is reported in the gnomAD database with an MAF of 0·002984, although it was as high as 0·02527 in South Asians, including 11 homozygous subjects. Thus, according to the American College of Medical Genetics and Genomics criteria, the T215I variant is scored as a variant of unknown significance and additional studies are needed to resolve its pathogenicity 33.

In a previous study 13, we reported that the healthy parents and both grandmothers of patient 17 carried both mutations in a heterozygous form. The occurrence of these two mutations on the same allele suggests a founder effect. This finding might be explained by the deep‐rooted inbreeding and consanguinity documented in Egyptian history 16. The RAG2 mutation G35V, was found in five unrelated patients in our typical SCID group (patients 8–12); interestingly, all of them came from the same governorate, which is located in Egypt’s Delta. This mutation has been reported before by Tabori et al. in three unrelated patients: one presenting with OS phenotype and two as classical SCID 23. The G35V mutation has been shown to affect the binding of RAG2 to RAG1’s ZnC2 and RNase H (RNH) domains and impair recombination activity 34, which was 0·4% of the wild‐type, as estimated by Tirosh et al. 33

In the present study we noted four patients with homozygous mutations (patients 3, 18, 30 and 31), although their parents were not consanguineous; this observation reflects the prevalence of RAG1/2 mutations in our population. We also observed a broad spectrum of clinical presentations among patients with RAG1/2 mutations. Some mutations resulted in different clinical phenotypes, even within the same family. For example, the RAG2 mutation (R148X) was reported in two siblings; a brother (patient 6), who presented with Omenn phenotype, and a sister (patient 20) presenting with typical SCID. The two siblings shared the same environmental factors that may be implicated in the phenotype; however, the only difference was that the brother had the bacille Calmette–Guérin (BCG) vaccine, whereas his sister did not receive the BCG vaccine or any other live attenuated vaccines. With follow‐up of patient 20, at age 7 months she began to suffer from recurrent infections; concomitantly, erythroderma began to appear, reflecting the possibility that the patient is shifting to the OS phenotype later in life. This has been hypothesized previously by Niehues et al. 4 as they suggested that early exposure to pathogens may induce T cell expansion that contribute to the phenotype. In the present study we observed some previously reported mutations which presented differently among our cohort; for example, the R229Q mutation in RAG2, that was considered a severe mutation with residual recombination activity of 8·9% 33, was found in five of our SCID patients and has been previously reported in patients with the OS phenotype 11, 22, 26, 27. Similarly, the G35V that had 0·4% recombination activity 33 presented with the classical SCID phenotype in five of our patients and had been previously reported in two OS patients 23, 24. Also, the R841W mutation in RAG1 was found in patient 27 with the T^–B⁺SCID phenotype, and had been previously reported in three studies of patients with OS 11: a patient with hypomorphic SCID and CMV infection 20 and a RAG‐deficient patient with residual B cell count 2. Although the patients reported in these three studies had different presentations, all of them had residual B cell counts; this observation serves to illustrate how the same mutation may give rise to different phenotypes 2, 11, 20. The RAG1 mutation detected in patient 30 (R973H) is another example of how the same mutation can be associated with different phenotypes. It has already been reported in patients with OS 22; however, patient 30 presented with features of CID and autoimmunity.

The K127X RAG2 mutation previously described in patients with classic T^–B^–SCID in Saudi Arabia and Israel 5, 35 presented differently in patient 31. Although this mutation was shown to be severe, with only 0·1% of the wild‐type recombinase activity 33, our patient had normal lymphocyte counts, with marked CD4 lymphopenia, and her RAG2 mutation was only detected at the age of 3 years. Nevertheless, one of the patient’s siblings died within the first few months of life displaying typical manifestations of SCID. Thus, the genotype–phenotype correlation which exists for this disease as previously reported is not absolute, and other environmental, genetic and epigenetic factors most probably contribute to determine the clinical phenotype 4, 33.

We characterized five RAG‐deficient patients with atypical presentations (patients 27–31). Patients 27 and 28 presented with classical SCID symptoms but had normal B cell frequencies and counts. The RAG1/2 mutations in these two patients were detected by whole exome sequencing, as the presence of B cells meant that a RAG defect was a remote possibility. The mutation R841W detected in patient 27 resulted in the substitution of an arginine by a tryptophan at position 841.

Among the 31 patients included in this study, seven novel mutations were reported; one patient presented with OS (patient 7), four with classical SCID (patients 19, 22, 24 and 26) and two with atypical SCID phenotype (patients 28 and 29). The novel missense RAG2 mutation (C419W) detected in patient 7 with OS, where cysteine was replaced by tryptophan, affects the RAG2 plant homeodomain that specifically recognizes histone H3 trimethylated at lysine 4 (H3K4me3). Mutations that abrogate RAG2’s recognition of H3K4me3 severely impair V(D)J recombination in vivo 36.

The novel RAG1 mutation (c.1768_1769insTACC, p. N591Pfs*14) detected in two typical SCID brothers (patients 22 and 23)] is predicted to encode if translated for a truncated protein that lacks more than 30% of the polypeptide chain; however, such a product may be unstable. The missing segment encompasses a large part of RAG1’s catalytic core, including part of the heptamer‐binding region, the zinc finger B and the C terminus 22. Another previously undescribed RAG1 mutation (K621E) was detected in a heterozygous form (in patient 24, who had compound heterozygous mutations in RAG1) where the highly conserved positively charged lysine at position 621 had been replaced by a negatively charged glutamic acid. The difference in physicochemical properties probably affects the structural integrity of RAG1’s highly conserved RNH domain and thus binding to RAG2. Kim et al.’s 34 assessment of the crystal structures of RAG1 and RAG2 highlighted a very close RAG1 mutation (A622P) that affected RAG2 binding to the RNH domain of RAG1.

The previously undescribed mutation documented in patient 19 (F29S) is also a missense mutation in which the hydrophobic residue phenylalanine located at position 29 is replaced by the polar residue serine. This affects the propeller region of RAG2’s core domain. This area is highly conserved and is close to the RAG1–RAG2 interface.

Patient 26 had a novel homozygous mutation (A565P) in RAG1, where the amino acid alanine was replaced by proline in the highly conserved Pre‐RNase H (PreRNH) domain at the RAG1/RAG2 interface.

We detected another previously unreported mutation (V327D) in patient 28 that substitutes a hydrophobic valine at position 327 with a polar aspartic acid residue; the ensuing difference in physicochemical properties is likely to have affected RAG2 core protein.

Patient 29 had compound heterozygous mutations in RAG1; one of them has not been reported previously (C335R). In this missense mutation, the cysteine at position 335 of RAG1’s RING finger domain is replaced by an arginine. The RING finger and the adjacent zinc finger motif form a single domain that co‐ordinates zinc ions and has histone H3 ubiquitin ligase activity, which is required for a normal level of chromosomal V(D)J recombination 32. In addition to the autoimmune manifestations, patient 29 developed a necrotizing granuloma involving the tongue leading to partial sloughing; however, the condition responded well to prednisone.

In general, comparing the different phenotypes documented in this report with those reported in other studies, as illustrated in Table 2, clarifies that one genotype may result in different phenotypes depending on many factors, such as exposure to infections, receiving live attenuated vaccines, environmental factors, epigenetics and more, to be elucidated.

Conclusion

Studying RAG mutations among highly consanguineous populations such as that in Egypt extends the spectrum of the clinical presentation of bi‐allelic mutations in RAG genes and leads to a clearer understanding of the effect of these mutations on enzyme activity. However, our study revealed that the genotype–phenotype correlation previously demonstrated in RAG deficient patients is not absolute. Factors other than the type of the mutations are clearly involved in determination of the patients’ phenotypes. Studying such factors is important, as they affect disease severity and outcome.

Disclosures

All authors have no conflicts of interest.

Author contributions

M. S. and R. El H. wrote the manuscript and performed the molecular diagnosis, A. E. performed the flow cytometry analysis, R. El K. and S. Lotfy collected the clinical data, D. S. Abd El A. and J. Boutros followed‐up the patients and contributed to the diagnosis, N. G. and A. E. revised the manuscript and followed‐up the patients. L. O., A. M. and J. P. S. performed whole‐exome sequencing for patients 27 and 28, analysed the data and revised the manuscript. M. v.d. B. performed Sanger sequencing for patients 6, 18 and 31 and revised the manuscript. J. C. and R. G. performed next‐generation sequencing for patient 30 and revised the manuscript.