- Alessandra Santoro1,2,
- Sonia Cannella2,
- Antonino Trizzino2,
- Giuseppa Bruno2,
- Carmen De Fusco3,
- Luigi D. Notarangelo4,
- Daniela Pende5,
- Gillian M. Griffiths6 and
- Maurizio Aricò2,7⇓
- 1 Divisione di Ematologia I, A.O. V. Cervello, Palermo, Italy
- 2 Onco Ematologia Pediatrica, Ospedale dei Bambini “G. Di Cristina”, Palermo, Italy
- 3 Onco Ematologia Pediatrica, Ospedale Pausilipon, Napoli, Italy
- 4 Division of Immunology, Children’s Hospital, Harvard Medical School, Boston, USA
- 5 Istituto Nazionale per la Ricerca sul Cancro, Genoa, Italy
- 6 Cambridge Institute for Medical Research, Addenbrooke’s Hospital, Cambridge, UK
- 7 Oncoematologia Pediatrica, A.O.U. Meyer, Firenze, Italy
- Correspondence: Maurizio Aricò, U.O. Oncoematologia Pediatrica, Azienda Ospedaliero-Universitaria Meyer, viale Pieraccini, 24, 50139 Firenze, Italy. E-mail:
Mutations of UNC13D have been described in patients affected by familial hemophagocytic lymphohistiocytosis (FHL3). The Munc13-4 protein contributes to the priming of the secretory granules. Mutation in this gene results in defective cellular cytotoxicity and the familial hemophagocytic lymphohistiocytosis clinical picture. Among reported mutations, few are predicted to impair splicing. Yet, functional impact of these mutations has not been addressed. We identified 18 out of 31 familial hemophagocytic lymphohistiocytosis families showing at least one mutation responsible for splicing error. We identified some known and three novel splicing mutations: one falls at the acceptor site of exon 11 and 2 are deep intronic mutations in IVS1 and in IVS30. We demonstrated that these deep intronic mutations affect regulatory sequences causing aberrant splicing. We report that UNC13D mutations leading to splicing errors represent the majority of mutations observed in familial hemophagocytic lymphohistiocytosis. This finding has implications for designing strategies for analysis of the families with suspected familial hemophagocytic lymphohistiocytosis.
Hemophagocytic lymphohistiocytosis (HLH) is a genetically heterogeneous disorder characterized by a hyperinflammatory syndrome with fever, hepatosplenomegaly, cytopenia and sometimes central nervous system involvement.1 The clinical picture may resemble that of leukemia, so bone marrow aspiration is usually performed early during the diagnostic work-up, allowing for hemophagocytosis by activated macrophages.
In most cases the natural course of HLH is rapidly fatal within a few weeks unless appropriate treatment, including corticosteroids, cyclosporine, etoposide, anti-thymocite globuline, can obtain transient disease control.2–4 So far, only patients who underwent hematopoietic stem cell transplantation have been cured.5–7
Differential diagnosis of HLH may be difficult.8 To this purpose, diagnostic guidelines for HLH have been established by the Histiocyte Society.9,10 In particular, demonstration of frequent association with common pathogens, together with evidence of impaired natural killer cytotoxic activity, provided the rationale for considering HLH as a selective immune deficiency.11–13 Starting from the original report by Farquhar et al. in 1952, autosomal recessive inheritance was proposed as the basis for the familial form of HLH (FHLH or FHL).
Since 2003, mutations of UNC13D (that encodes for the Munc13-4 protein) have been associated with FHL (FHL3, MIM 608898).15 Munc13-4 is a critical effector of the exocytosis of cytotoxic granules priming cytotoxic granule fusion. Mutations in this gene impair the delivery of the effector proteins, perforin and granzymes, into the target cells resulting in defective cellular cytotoxicity and a clinical picture which appears very similar to that of FHL2 (MIM 603553), associated with PRF1 mutations.16–19
Recently, we screened the patient population for mutations of UNC13D and found that FHL3 accounts for a significant proportion of cases of FHL in Italy.21 Among the observed mutations, scattered over the entire gene sequence, a few were reported to be responsible for impaired splicing of the gene products.15,21–23 Six splice mutations are located in the exon/intron boundaries (322-1G>A, 753+1G>T, 754-1G>C, 1389+1G>A,1545-1G>T,1596+1G>C), 2 mutations (610A>G and 1847A>G) in the U1 exon recognition site, and only one (569+5G>A) in the intronic region. However, the functional implication of such mutations has not been widely addressed. Recently, an increasing number of disease-related mutations have been reported to be responsible for aberrant splice process,22 increasing the focus of our research on mutations which might affect the splicing machinery. Notably, 15% of pathogenic mutations in mammalian genes affect mRNA splicing but, for some genes such as NF1 and ATM, nucleotide substitutions affecting the splicing process may account for up to 50% of all pathogenic mutations.
The splicing process involves many different proteins, including those of the spliceosome, which usually binds to the classical splice sequences located at the exon-intron borders (donor and acceptor sites). However, auxiliary sequences, scattered all over the gene, may serve as exonic and intronic splicing enhancers (ESE and ISE) that help in the recognition of exons.
In this report we show that UNC13D mutations leading to splicing errors are frequent in FHL3. Their detection and identification may be instrumental in the diagnosis of this disease.
Results and Discussion
We identified 31 families with FHL3 due to biallelic UNC13D mutations. Eighteen (58%) had at least one mutation predicted to be responsible for a splicing error (Table 1). A total of 7 mutations affecting splicing were identified. Only 2 of them, 753+1G>T and 1389+1G>A, had been already reported by other groups. The remaining 5 have only been observed in our patients: the 610A>G (M204V) and 1847A>G (E616G) (included in our recent report),20 the previously unreported 952-1G>A, and in two families we found an allele containing deep (>10nt from the exon/intron boundary) intronic mutations (one had IVS1 +59C>T, +394C>T, +525G>T, and the other had IVS30+431delA).
This mutation, which falls at the donor site of exon 9, was found in ten families (UPN 6, 173, 180, 182, 196, 211, 225, 285, 363, 419). In five families it was at the homozygous state, while five were compound heterozygous for: 1847A>G (E616G) (n=2), 2346delGGAG (782fsX12), 1387C>T (Q463X), or 2570T>G (F857C).
A study of cDNA of 2 homozygous patients (UPN 285 and 419) showed that the only detected specific product was smaller than normal, indicating probable exon skipping. The amplification product was, therefore, sequenced and the whole exon 9 was found to be missing. In 3 patients, 2 homozygous (UPN 180 and 285), and one compound heterozygous (UPN 225), Western Blot analysis confrmed the absence of the Munc13-4 protein (data not shown).
This mutation was found in five families (UPN 198, 211, 225, 237, and 249). Two were homozygous, and three compound heterozygous: two for 753+1G>T, and one for 2782C>T(R928C). It falls at two nucleotides from the end of exon 20, where it is predicted to disrupt U1 splice site recognition. To verify this, we cloned and sequenced cDNA from a homozygous patient, showing that 1847A>G disrupts splicing and results in several RNA products. Abnormal splicing results were found in 6 out of the 8 clones screened: 3 clones showed abnormal splicing of exon 20, 2 clones showed intronic exonization of 130 bp, and one clone showed an in-frame insertion of 54 intronic nucleotides between exons 1 and 2.
In 2 patients, one homozygous (UPN 249) and one compound heterozygous (UPN 237), Western Blot analysis confirmed the absence of the Munc13-4 protein (data not shown).
Mutations 952-1G>A and 1389+1G>A
These were found in one family each. The 952-1G>A, which was found in one homozygous patient (UPN 390), falls at the acceptor site of exon 11. The 1389+1G>A, which was found in combination with 2346delGGAG in one patient (UPN 350), falls at the donor splice site of exon 15. Unfortunately, no material from these cases was available for protein study.
Mutation 610A>G (M204V)
Mutation 610A>G, (M204V), found in association with 2650C>T(Q884X) and 3226insG (1076fs) in one patient (UPN 289), is located at five nucleotides from the end of exon 7. The reported mutations make the sequence more closely complementary with U1snRNA, and might be expected to affect splicing, although there is no confirmed evidence of this.
Western Blot analysis documented the absence of the Munc13-4 protein (data not shown).
Deep IVS1 mutations
In one patient (UPN 383), the maternal allele carried 2346delGGAG, and the paternal three deep intronic mutations: IVS1+59C>T, +394C>T and +525G>T. All 3 involve putative intronic splicing enhancer (ISE) motifs: 2 of them created a new ISE (R0952 and R0987) and the other (+525G>T) disrupted a pre-existing ISE (R0952) (Figure 1). Due to the uncertain role of deep IVS1 mutations, we amplified the messenger RNA. Together with a normal-sized band we found a larger band suggesting an insertion; both bands were present in the father. Therefore, to characterize the splice error, the larger PCR product was cloned. By screening and sequencing ten colonies, we identified an intronic insertion of 130bp between exons 1 and 2 in five clones, and the retention of the intron between exons 3 and 4 in another clone (Figure 1). Western Blot analysis confirmed the absence of the Munc13-4 protein (Figure 1C).
Deep IVS30 mutations
We identified a patient (UPN 356) carrying a mono-allelic mutation in the coding region 2346delGGAG; the patient showed the skipping of exon 31 at cDNA level (Figure2). To search for the genomic mutation responsible for exon 31 skipping in the mRNA, we sequenced the associated intron, and found a deep nucleotide deletion in intron 30: IVS30+431delA. Interestingly, this mutation occurs in a putative intronic splicing silencer (ISS, R1018), a cis-element that promotes exclusion of a specific exon from the mature mRNA.
Western Blot analysis confirmed the absence of the Munc13-4 protein (Figure 2).
We found the insertion of the same 130bp cryptic-exon in two unrelated families with FHL3 (UPN 249 and 383). By a BLAST search of GeneBank, the inserted sequence was identified as part of intron 1. This inserted sequence was identical in the 2 patients. The novel cryptic-exon is delimited at both 5′ and 3′ ends by ISE sequence; it includes the IVS1+394 nucleotide and a stop codon, located 7 codons after the inserted sequence. Cryptic splice sites used by the pseudoexon were predicted by NNSPLICE, assigning at both donor and acceptor sites a high score of 0.93 and 0.85 respectively.
UNC13D cDNA sequence in control subjects
To investigate whether the aberrant splicing found in our series represented normal alternative transcripts, we amplified cDNA from 10 healthy controls and assessed the correct UNC13D RNA splicing. No bands of unexpected size were found in any control cDNA. Furthermore, the entire correct cDNA sequence was obtained from 3 of these healthy subjects.
The diagnosis of HLH may be difficult. In particular, the clinical syndrome defined as HLH may be shared by patients with infection-associated, transient disease, usually cured by etiological anti-infectious therapy, sometimes associated to short-term immuno-suppressive therapy.4,23 On the contrary, patients with FHL not only deserve a prompt, intensive treatment with chemo-immunotherapy, but can only be cured by hematopoietic stem cell transplantation (HSCT).4,5–7 Therefore, the genetic study of patients with HLH, which at present remains the gold standard for discrimination between the secondary and the genetic form, is urgently needed for the families involved and the attending physicians.8,16,24
PRF1 mutations, associated with FHL2, account for only 40% of the population in our experience (Aricò and Santoro, unpublished data). Based on our preliminary studies, an additional 40% of our patients have FHL3 (Aricò and Santoro, unpublished data). The observation of some families with a single pathogenic mutation prompted us to investigate alternative defects of Munc13−4.
The splicing machinery may be affected by the size of the gene, with genes with more exons being more prone to splicing errors. The complexity of the UNC13D splicing process may suggest that, as already reported for NF1, ATM, CHRNE or COL4A5, it may be more vulnerable to pathogenic splicing errors.
The 753+1G>T is the most frequently observed mutation in our population, and this has also been reported by other groups.14,21 It falls at the consensus site and causes the skipping of the adjacent exon. Yet, splice errors may also result from deep intronic mutations involving intronic splicing enhancer (ISE) motifs which promote exon recognition. By studying patient UPN 383, we documented that deep intronic mutations might either create a new ISE (IVS1+59C>T and +394C>T) or disrupt a pre-existing one (IVS1+525G>T). In particular, mutation IVS1+394C>T seems to promote the activation of the nearby splicing and the insertion of a cryptic exon. This finding is in keeping with observations in Alport syndrome and in the familial form of retinoblastoma in which the intronic exonization was caused by a deep intronic mutations in the COL4A5 and RB1 genes respectively.
Another unexpected finding was the identification of the same 130bp cryptic-exon activation in two unrelated families. This insertion is part of intron 1, is delimited by two ISE motifs at both 5′ and 3′ ends, it used strong cryptic splice sites (predicted by in silico analysis) and contains a stop codon leading to a truncated protein that will be degraded by nonsense-mediated decay activity.
In conclusion, we report that UNC13D mutations responsible for splicing errors are a very frequent cause of FHL3. In fact, over one half of the total pathogenic mutations arise from an UNC13D splice error. Although a few disease-causing splicing mutations at the canonical splice site sequences have been previously described, we describe the importance of deep intronic mutations which cause aberrant splicing in contributing to this disease. This is important when designing strategies for mutation analysis of UNC13D in patients with FHL. Although patients with a single pathogenic mutation identified by exon-by-exon approach are the most obvious targets for splicing studies, in these cases, the presence of two splicing, hidden mutations cannot be ruled out. Evidence of impaired degranulation as detected by the CD107a expression flowcytometry assay,24 or defective protein at immunoblotting, may herald FHL3 and therefore strongly support the need for such an extended UNC13D analysis. Identification of UNC13D mutations allows genetic diagnosis of FHL3 in children and young adults, with immediate therapeutic implications, including indication for HSCT, selection of familial donor and prenatal diagnosis.
Authorship and Disclosures:
AS: study design, supervisor of molecular study, analysis of the data, manuscript writing; SC: molecular study; AT: clinical data collection; GB: mutation analysis; CDF: clinical data collection; LDN: analysis of the results, manuscript writing; DP: cellular cytotoxicity study; GMG: protein and confocal study; MA: PI, study design, analysis of the clinical data, writing the manuscript, financial support. The authors reported no potential conflicts of interest.
Funding: partly supported by grants from the Italian Ministry of Health (Ricerca Finalizzata 2004 “Istiocitosi e Tumori”), Associazione Italiana Ricerca sul Cancro (AIRC), Associazione Ricerca Sindrome Emofagocitica (ARSE), and the Histiocytosis Association of America (HAA). GMG is supported by the Wellcome Trust.
The online version of this article contains a supplementary appendix.
- Received December 14, 2007.
- Revision received January 25, 2008.
- Accepted February 21, 2008.
- Copyright© Ferrata Storti Foundation