Hypomorphic mutations in PRF1, MUNC13-4, and STXBP2 are associated with adult-onset familial HLH

Familial hemophagocytic lymphohistiocytosis (familial HLH; MIM 267700)¹  is an inherited immune deficiency characterized by the overactivation and excessive proliferation of macrophages and T lymphocytes. This leads to infiltration and damage of organs, including the liver and the CNS. Classic clinical features include prolonged fevers, cytopenias, hepatosplenomegaly, and signs of immune activation. Hemophagocytosis is the hallmark of HLH in general, but it may not present at early stages of the disease. The functions of natural killer (NK) cells and cytotoxic T lymphocytes are reduced or absent. familial HLH in children is rapidly fatal without the appropriate immunosuppressive treatment followed by hematopoietic stem cell transplantation.

To date, defects in at least 7 genes (ie, PRF1, MUNC13-4, STX11, RAB27A, STXBP2, SH2D1A, and BIRC4) are known to be associated with familial HLH.^1-7  The protein products of the 5 autosomal-recessive (AR) genes PRF1, MUNC13-4, STX11, RAB27A, and STXBP2 are all involved in the PRF1-dependent cytotoxic pathway, so affected patients often demonstrate abnormal function of cytotoxic lymphocytes. Perforin is constitutively contained within the secretory granules of all cytotoxic lymphocytes, facilitating the entry of granule contents into target cells that are infected, spent, or dangerous to the immunologic well-being of the organism, which results in the cytotoxic response. Mutations in the other AR genes interfere with the delivery of the cytotoxic granules to the contact surface with target cells and their extrusion of contents into the contact zone. Most studies on the AR forms of familial HLH focus on patients who present with symptoms of the disease within the first years of life.^8-11  Only a few cases with later onset have been reported.^12-15 

In this study, we describe the genetic and immunologic findings of adult patients in North America who have been referred to the Diagnostic Center for Heritable Immunodeficiency at Cincinnati Children's Hospital Medical Center (CCHMC) for testing.

This study was approved by the CCHMC institutional review board. During the 5-year study period, 1531 patients were referred for genetic testing due to a suspected diagnosis of HLH and/or associated conditions, as determined by their referring physicians. Clinical and demographic features were collected based on the information provided by ordering physicians on the test requisition form.

Genomic DNA was obtained from blood, BM, B-lymphoblastoid cell lines, and autopsy tissues using a standard procedure. All coding exons and at least 50 base pairs of the adjacent intronic region of the PRF1, MUNC13-4, and STXBP2 genes were amplified by PCR, followed by bidirectional sequencing. Detailed methodologies have either been published previously^2-5  or available upon request. Mutation nomenclature is based on the recommendations of the American College of Medical Genetics and the Human Genome Variation Society.

Several in silico analysis methods for the prediction of functional consequences of sequence variants were used to provide initial classification. In particular, we used SIFT (http://blocks.fhcrc.org/sift/SIFT.html), which uses evolutionary information from homologous proteins,¹⁶  and PolyPhen (http://www.bork.embl-heidelberg.de/PolyPhen/), which incorporates structural information into classification rules.¹⁷  The Grantham Scale¹⁸  was also used to evaluate the significance of amino acid substitutions.

Minor allele frequency was found in the National Center for Biotechnology Information single-nucleotide polymorphism database and also identified through testing 50 unrelated individuals from a southern Ohio control population (race: 83.4% white, 11.8% black, 1.6% Asian, 1.6% Hispanic, and 0.2% Native American). NK-cell cytotoxicity and perforin expression were analyzed as described previously at the Diagnostic Immunology Laboratory at Cincinnati Children's Hospital.^19-21  The results were compared with the normal ranges for age-matched controls that have been obtained in our laboratory.

In total, 1531 patients were referred for genetic testing due to a suspected diagnosis of HLH and/or associated conditions as determined by their referring physicians. Of these patients, 175 were 18 years of age or older. Twenty-five of the adult patients (14%) possessed mutations or sequence variants in PRF1, MUNC13-4, or STXBP2 (Table 1). All of the mutations were either missense base substitutions or splicing site variations. In comparison, in patients with age of onset below 18 years, we found mutation(s) in PRF1, MUNC13-4, and STXBP2 in ∼ 30% of the patients, which included nonsense, missense, splice site, and other type of deleterious mutations. Table 2 shows the number of sequence variants found in different age groups in patients with clinical suspected HLH. In some of these younger patients, mutations and sequence variants were found in > 1 gene.

Twelve missense mutations and sequence variants were identified in PRF1 in 18 patients. The A91V-PRF1 genotype was found in 12 patients in both heterozygous and homozygous states. Eight sequence variants in MUNC13-4 were found in 7 patients; 2 were splice-site changes and 5 were missense mutations. Two splice-site sequence variants in STXBP2 were identified in 2 patients. Two patients (P18 and P21) were double heterozygous, with the A91V-PRF1 genotype in the PRF1 locus and a second mutation in either STXBP2 or MUNC13-4. More than 1/2 of the patients (13 of 25) carried only one mutation in one of these genes, which are inherited in an autosomal recessive fashion. It is possible that other types of mutations (eg, gross deletions, insertions, or complex rearrangements) were not able to be detected by the methodology used in this study or that these patients possess mutations in additional genes that have not yet been discovered to be associated with HLH.

Eleven novel sequence variants were identified in this study. In silico analyses of the predicted structural effects of the missense variants are summarized in Table 3.^13,22-24  A91V, P188L, R356W, and T450M in PRF1 and Y61C in MUNC13-4 are consistently predicted to be pathogenic, with no presence in general healthy populations except A91V (4%-7% by multiple studies).^20,24  The sequence variant A91V-PRF1 and its relationship with familial HLH have been studied extensively in recent years.^25,26  Numerous clinical studies^13-15,24  have documented variability in cytotoxic function in individuals who are heterozygous or homozygous for the A91V substitution. Most of the evidence suggests that although it is a “milder” or hypomorphic mutation, A91V is not clinically neutral.^27,28  The predictions for the rest of the sequence variants are largely inconclusive.

Perforin expression was measured by flow cytometry in 7 of the patients with PRF1 sequence variants. The level of perforin correlated well with the genetic findings in adult patients. Perforin was absent or severely low in patients with biallelic mutations in PRF1, whereas slightly decreased or at the lower normal level in patients with 1 mutation (Table 1). Results of NK-cell functional testing were available for 14 patients. NK-cell function was typically observed to be low in these adult patients with HLH, but was not closely correlated with the number of mutations identified in PRF1, STXBP2, or MUNC13-4.

In young children, the presenting symptomatology of familial HLH has been well described^9,29  and forms the basis for the diagnostic criteria. In older children and adults, a broader spectrum of clinical phenotypes (eg, encephalitis, autoimmune lymphoproliferative disease, acute lymphoblastic leukemia, aplastic anemia, and systemic onset juvenile idiopathic arthritis)^{11,14,15,30-35}  have been reported. In some of these reported cases, patients were cleared from infections before the overt diseases, and therefore a negative clinical history of HLH does not rule out this syndrome. In patients with familial HLH who carry hypomorphic genetic defects, residual NK- and T-cell function, although decreased, may be sufficient to prevent the development of clinical HLH for many years. In the patient cohort of the present study, 12% of patients were 18 years and older at the time of diagnosis. Two patients apparently developed the first symptoms of HLH in their mid-seventies. The observations of double heterozygosity in 2 adult patients (P18 and P21), with the A91V-PRF1 genotype in the PRF1 locus and a second mutation in either STXBP2 or MUNC13-4, are very interesting. Although there is no known direct interaction among these 3 proteins, additive effects are possible given that they all play critical roles in the perforin-dependent cytotoxic pathway. Additional research is required to further define the functional effects of these genetic alterations and to determine the optimal treatment for adult patients with HLH. Anecdotal information from physicians who submitted samples for testing indicated that a range of treatments have been used, including steroid therapy alone, HLH-94–type therapy, and, in a young adult with recurrent disease, allogeneic hematopoietic stem cell transplantation.

With increasing awareness about HLH, more adult patients and/or patients with milder and/or relapsing clinical courses are being identified. In this study, we found that almost all of the sequence variations in PRF1, MUNC13-4, and STXBP2 carried by adult patients were either missense or splice-site changes representing hypomorphic mutations that played a contributing role in the development of late-onset HLH when patients were challenged by viral infection and other types of environmental stresses. Genetic and immunologic diagnostic testing and timely HLH-directed treatment should be considered for adult familial HLH patients.

The authors thank the staff of the CCHMC diagnostic immunology and molecular genetics laboratories, a FOCIS Center of Excellence.

This work was supported by grants from the Histiocytosis Association of America, the Jeffrey Modell Foundation, the National Institutes of Health (HL091769), and the Doris Duke Charitable Foundation.

National Institutes of Health

Contribution: K.Z. and A.H.F. designed the research, collected, analyzed, and interpreted the data, and wrote the manuscript; K.A.R. analyzed the results and wrote the manuscript; R.A.M., P.S.K., and M.B.J. collected the data and reviewed the manuscript; D.K., J.A.J., and J.V. performed the research and collected the data; J.M. analyzed the data and wrote the manuscript; and Q.W. collected the data and created the tables.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Kejian Zhang, MD, Associate Professor of Clinical Pediatrics, Division of Human Genetics, MLC 4006, Cincinnati Children's Hospital Medical Center, 3333 Burnet Ave, ML 4006, Cincinnati, OH 45229; e-mail: kejian.zhang@cchmc.org.