Characteristic perforin gene mutations of haemophagocytic lymphohistiocytosis patients in Japan


Ikuyo Ueda, MD, Department of Paediatrics, Kyoto Prefectural University of Medicine, Hirokoji-Kawaramachi, Kamigyo-ku, Kyoto, 602–8566, Japan. E-mail:


Summary. Perforin gene (PRF1) mutations appear to occur in about 30% of patients with haemophagocytic lymphohistiocytosis (HLH). We tested perforin expression and gene mutations in 14 HLH patients and six patients with Epstein–Barr virus-associated HLH (EBV-HLH) in Japan. Five of the 14 HLH patients had perforin abnormalities. The presence of PRF1 genetic abnormality correlated well with the lack of perforin expression as determined by flow cytometry. Sequencing showed that four patients had a compound heterozygous mutation while the fifth patient had a homozygous mutation. Three of the mutations we detected were novel. In contrast, none of the six EBV-HLH patients showed perforin abnormalities. Our data, combined with the PRF1 mutations in three previously reported Japanese patients, suggest that the 1090–1091delCT and 207delC mutations of the perforin gene are frequently present in Japanese HLH patients (62·5% and 37·5% respectively). Examination of the geographical origins of the ancestors in the perforin-mutant HLH patients revealed that they mostly came from the Western part of Japan, suggesting that the present-day cases may largely derive from a common ancestor.

Familial haemophagocytic lymphohistiocytosis {FHL [Mendelian Inheritance in Man (MIM) no. 267700]} is a rare autosomal recessive disorder that is characterized by fever, hepatosplenomegaly, pancytopenia, disseminated intravascular coagulation and central nervous system (CNS) disease (Janka, 1983; Arico et al, 2001a; Henter, 2002). Although the abnormality responsible for this disease is still unclear, it has long been noted that FHL patients are deficient in natural killer (NK) activity (Stark et al, 1987; Imashuku et al, 1998; Sullivan et al, 1998). Consequently, the diagnosis of FHL until recently was dependent on the following criteria: patients develop haemophagocytic lymphohistiocytosis (HLH) when they are under 2 years of age, other family members have had HLH, the patients lack NK activity and, finally, the patients have CNS signs and symptoms associated with characteristic magnetic resonance imaging (MRI) findings (Haddad et al, 1997; Henter & Nennesmo, 1997; Imashuku et al, 2002). Recently, however, genetic markers for FHL have been determined, which aids the diagnosis of this disease. Linkage analyses have enabled FHL patients to be classified as FHL1 [9q21.3–22 linked (MIM 603552)] (Ohadi et al, 1999), FHL2 [10q21–22 linked (MIM 603553)] (Dufourcq-Lagelouse et al, 1999) and another as yet undefined type (Graham et al, 2000). After this seminal discovery, it was found that the molecular cause of FHL2 is an aberration in the gene encoding the cytotoxic effector perforin (Stepp et al, 1999). Perforin gene [PRF1 (M31951)] mutations, as determined by sequencing, have since been found to occur in about 30% of HLH patients (Stepp et al, 1999; Clementi et al, 2001; Ericson et al, 2001; Suga et al, 2002). In addition, Kogawa et al (2001) demonstrated that lack of perforin expression, as determined by flow cytometry (FCM), correlated well with abnormalities in the perforin gene. In this study, we examined the perforin abnormality in 20 HLH patients by FCM and sequencing. In addition, Japanese patients were found to bear characteristic perforin gene mutations.

Patients and methods

Patients.  Twenty HLH patients consisting of 10 males and 10 females with a median age of 1·0 years (ranging from 1 week to 20 years old) participated in the study. Fourteen patients were diagnosed as having HLH based on the criteria described by Henter et al (1991). Six of them were diagnosed with EBV-HLH, according to the criteria described by Imashuku et al (2000). The EBV-HLH patients did not lack NK activity and did not have CNS disease. Of the 14 HLH patients, 12 were tested for NK activity as previously described with an effector:target (E:T) ratio of 20:1 (Imashuku et al, 1998, 2002), and six showed poor NK activity. Six of the 14 patients were positive for CNS disease, and this was confirmed with brain MRI. Only three of the 14 patients had a family history of HLH. Notably, one of these, who developed the disease when aged 11 years, had two siblings who had died in early infancy, indicating she is an atypical case of late-onset FHL (Table I). None of these HLH patients were reported by Suga et al (2002).

Table I.  Clinical and laboratory data in 20 HLH patients.
PatientOnset age of HLHFamily historyCNS disease (yes or no)NK activity* (%)%PerforinNucleotide change
  • *

    < 5% (defect), 5–18% (mild decrease), > 18% (normal).

  • Two of the 14 HLH patients were EBV related by retrospective analysis.

  • Pos, positive; neg, negative; na, not available; nd, not determined; ne, not evaluable.

111 yearsposyes2·0< 1·0< 1·0pos
2 1 monthnegyes2·0< 1·0< 1·0pos
3 2 monthsnegno1·0< 1·0< 1·0pos
4 2 monthsnegyes2·0ndndpos
5 3 monthsnegno1·0ndndpos
6 1 monthposno1·0ndndneg
7 1 years 7monthsnegyes16·819·731·0neg
8 2 yearsnegyes27·0ndndneg
9 2 monthsnegyes27·0ndndneg
10 5 monthsposno19·0< 1·0neneg
11 1 weeknegnona1·637·6nd
12 2 monthsnegno15·029·642·2nd
1310 monthsnegnona12·818·0nd
14 1 year 6 monthsnegno17·04·67·6nd
15 1 years 2monthsnegno26·0nd51·9nd
16 1 year 2 monthsnegno35·51·23·0nd
17 1 year 11 monthsnegnona1·22·1nd
18 4 yearsnegno57·012·117·7nd
19 5 yearsnegno33·021·09·1nd
2020 yearsnegnona39·645·3nd

Samples.  Peripheral blood (PB, n = 18) or bone marrow (BM, n = 1) mononuclear cells (MC) were obtained from the patients, then separated and promptly subjected to FCM. Peripheral blood was also obtained from the parents of the three patients with a family history of HLH. Aliquots of the blood cell preparations were frozen for subsequent DNA sequencing. Buccal swabs (n = 1) were obtained from a transplanted patient in whom PBMC and BMMC were not available. As controls, PBMC were obtained from age-matched normal infants and children (16 boys and 14 girls). All samples were obtained with the informed consent of the parents/guardians of the patients and control subjects.

FCM. PBMC and BMMC were first stained with phycoerythrin (PE)-conjugated anti-CD3 (Leu-4; Becton Dickinson, San Diego, CA, USA) or anti-CD8 (Beckman Coulter, Miami, FL, USA) for detection of surface membrane antigen, as previously described (Imashuku et al, 1998). Following fixation and permeabilization using Intrastain (Dako, Glostrup, Denmark) for cytoplasmic staining, the cells were then stained intracellularly with fluorescein isothiocyanate (FITC)-conjugated antiperforin antibody (Becton Dickinson). Five thousand cells in the lymphocyte gate were analysed using the EPICS XL-MCL system (Beckman Coulter). Intracellular perforin was considered deficient when less than 1·0% of either CD3+ or CD8+ cells expressed the antigen after subtracting negative controls stained by matched mouse isotype-irrelevant immunoglobulins.

Sequencing of the perforin gene.  Genomic DNA was extracted from PBMC, BMMC or buccal swab cells. The coding region of the perforin gene in exons 2 and 3 was amplified by a polymerase chain reaction (PCR). The primers used for the amplification have been described previously (Stepp et al, 1999). Direct DNA sequencing was performed in 10 patients using a BigDye Terminator cycle sequence kit and the ABI PRISM 377 Sequence Detection System (PE Biosystems, Foster City, CA, USA). In four of these 10 patients, the PCR products were subcloned into competent Escherichia coli cells (competent high DH5α, TOYOBO) and sequenced. The primers used for sequence were: for exon 2 (forward), 5′-GATAATCTGTGGCTGTGGGG-3′, 5′-TCCCAGTGGACACACAAAG-3′, (reverse), 5′-AGACCACCCAGAGTTTCCCG-3′, 5′-CCAGTCGTTGCGGATGCTAC-3′; and for exon 3 (forward), 5′-GAGGTCTCTCTCTTCTCGCA-3′, 5′-ACGGCAGCATCTCTGCCGAA-3′, 5′-GGAGGTGACCTTCATCCAAGC-3′ (reverse), 5′-CTCACTGTTCTCACCACACG-3′, 5′-GGGGTTGTTATTGTCCCACA-3′, 5′-CCCGAACAGCAGGTCGTTAAT-3′.

Statistical analysis.  Fisher's exact test was used to compare the two groups.


Perforin expression in lymphocytes by FCM

Perforin expression in the CD3+ and/or CD8+ cells obtained from 15 of the 20 patients (14 HLH, six EBV-HLH) and 30 age-matched control subjects are summarized in Table I and Fig 1. As shown, three HLH patients (patients 1, 2 and 3) completely lacked intracellular perforin expression (< 1·0%) in both their CD3+ and CD8+ cells. Another patient (patient 10) also lacked perforin expression in CD3+ cells, but his CD8+ cells were not evaluable. In contrast, perforin expression of CD3+ and CD8+ cells in control subjects were 8·5 ± 7·3% and 22·3 ± 17·5% respectively (age range, 1 month to 20 years old). Figure 2 shows the representative profiles obtained from patient 3 who showed a complete lack of perforin in CD8+ cells.

Figure 1.

Perforin expression by PBMC/BMMC. Open circles and closed diamonds indicate the HLH patients and control subjects respectively. Four patients with HLH (patients 1, 2, 3 and 10) lacked perforin expression in CD3+ cells, and three of them (patients 1, 2 and 3) also lacked perforin expression in CD8+ cells.

Figure 2.

Representative profiles by FCM of intracellular perforin-expressing CD8+ cells in patient 3 and a normal control subject.

Perforin DNA sequence

The perforin gene was sequenced in 10 HLH patients (Table I). Five HLH patients showed nucleotide changes in this gene. Three of these five patients were those shown to have deficient perforin expression by FCM (Fig 1). In the remaining nine HLH patients, perforin expression was either detectable by FCM or their perforin gene sequence was normal. Of the five HLH patients with perforin gene mutations, four shared a deletion of two base pairs at codon 1090–1091 (1090–1091delCT) (Fig 3, Table II). Four patients had a compound heterozygous mutation while one patient had a homozygous mutation. Three of the mutations were novel. Two of these were found in patient 1. Here, the compound heterozygous mutation consisted of the substitution of the adenine residue at position 1 with a guanine residue (1 A → G). Another mutation was the guanine to adenine conversion at nucleotide position 949 (949G → A). Patient 3 had a compound heterozygous mutation, with the cytosine being substituted by thymine at nucleotide position 1246 (1246C → T) in addition to the 1090–1091delCT mutation. In summary, one patient had a missense mutation while the remaining four patients had nonsense mutations.

Figure 3.

Comparison of the perforin gene sequence of patient 5 and a normal control subject. The deletion of two base pairs at codon 1090–1091 (CT) in the patient's perforin gene is shown by an arrow. Exon 3 of the perforin gene in the patient's PBMC was subjected to direct sequencing.

Table II.  Summary of perforin gene mutations detected in Japan.
 PatientNucleotide changePredicted effectMissense/Nonsense
  • *

    Heterozygous mutation;

  • **

    Homozygous mutation; P69fsht##, frameshift and termination at 106 AA; R361fsht###, frameshift and termination at 364 AA; L364fsht####, frameshift and termination at 456 AA; na, not available.

Present study11A → G*M1VMissense
949G → A*G317R 
1246C → T*Q416stop 
Suga et al (2002)11083delG*R361fsht###Nonsense
1491T → AC497stop 
1122G → A*W374stop 

Family tree of three HLH patients with perforin gene mutations

The PBMC of the parents of three perforin-mutant patients were available and thus the parental perforin genes could be sequenced. Unfortunately, samples were not available for the study from their grandparents. Figure 4 shows that two patients had a heterozygous mutation while one patient a homozygous mutation.

Figure 4.

Family trees of three HLH patients with perforin gene mutations.

Characteristic perforin gene mutations in Japan

As summarized in Table II, the 1090–1091delCT mutation was found in four of our five perforin-mutant HLH patients. Patients 2 and 4, who were unrelated, also had the same heterozygous mutations of 1090–1091delCT and 207delC. Suga et al (2002) previously reported the 1090–1091delCT mutation in one of their three HLH patients. Thus, the 1090–1091delCT mutation accounts for five of the eight HLH patients (62·5%) with perforin gene abnormalities who have been identified in Japan to date. We also detected the 207delC mutation in two of the five HLH patients in our study. Suga et al (2002) also observed this mutation in one patient. This suggests that the incidence of this deletion in Japan is also high (37·5%).

Ancestral origin of the patients with perforin gene mutations

No consanguinity was noted in our five HLH patients with perforin abnormalities. We determined the ancestral origin of these patients. Both the paternal and maternal ancestors in one patient came from the Western Kyusyu area (Saga, Nagasaki, Fukuoka). One of the three patients with 1090–1091delCT reported by Suga et al (2002) also had the same pattern of ancestral origin (N. Suga, personal communication). In another two of our patients, ancestors from one side of the family originated from the same area, and the other side of the family were from the Osaka/Yamaguchi and Chiba/Ibaraki areas. Another patient's ancestors came from the Kagawa/Ishikawa and Wakayama/Okayama. The only exception were ancestors of patient 1 (who carried the missense mutation), who were from the Chiba/Ibaraki areas. These observations indicate that the common mutations of 1090–1091delCT were restricted to people originating in the Western part of Japan (Kyusyu, Yamaguchi, Wakayama, etc.).

Correlation between phenotype and genotype

We compared various clinical parameters of the perforin-mutated HLH group (five patients) with those of the intact-perforin HLH group (nine patients) (Table I). The parameters examined were: age of HLH onset, presence of CNS disease and NK activity. Of the perforin-mutated HLH group, all four patients with nonsense mutations developed the disease at 1 to 3 months of age, while the single patient with the missense mutation developed HLH at 11 years of age. The median age of onset for patients with intact perforin was 5 months (ranging from 1 week to 2 years). CNS disease developed in three of the five perforin-mutated HLH patients and in three of the nine intact perforin patients (P = 0·58). At diagnosis, NK activity was deficient in all five perforin-mutated HLH patients, while only one of seven perforin-intact patients showed NK incompetency (P = 0·015).

Perforin gene mutations in HLH from a literature survey and present study

As summarized in Fig 5, various perforin gene mutations in HLH patients have been identified to date. The five mutations identified in our study are underlined. Three of these (1A → G, 949G → A and 1246C → T) have not been previously reported. It also appears that the 207delC and 1090–1091delCT may be characteristic of Japanese HLH patients.

Figure 5.

Perforin gene mutations in HLH patients from a literature survey and the present study. Summary of the data reported by Stepp et al (1999), Clementi et al (2001, 2002), Ericson et al (2001), Kogawa et al (2002), Feldmann et al (2002), Suga et al (2002) and the present study. The upper panel represents perforin gene mutations reported in this study and the lower panel lists those that have been previously described. L17fsht## = frameshift and termination at 51 AA, P69fsht## = frameshift and termination at 106 AA, R361fsht### = frameshift and termination at 364 AA, L364fsht#### = frameshift and termination at 456 AA.

Perforin expression in EBV-HLH patients

All six EBV-HLH patients showed detectable perforin expression, with 1·2–39·6% of the CD3+ cells and 2·1–51·9% of the CD8+ cells expressing perforin. In addition, the NK activity in four of the six patients tested was normal (range 26·0–57·0%).


Perforin is a prototype effector molecule of immune cells that mediate cytotoxicity. Perforin deficiency results in severe immune dysregulation and leads to the development of FHL2 (Moretta et al, 2000; Stepp et al, 2000). Perforin deficiencies can thus be used to diagnose patients with FHL. One means of determining perforin deficiency is to examine perforin expression by FCM. Rukavina et al (1998) analysed the peripheral blood of normal subjects of varying ages by FCM and found that the proportion of perforin+ lymphocytes increased after birth but declined rapidly after the age of 70 years. They also found that children (aged 5 years) differed from adults (20 years old or over) in that they have high levels of CD4+ perforin+ cells, which constituted 8·9% and approximately 1% in children and adults respectively. In contrast, Kogawa et al (2002), who used four-colour flow cytometric analysis, demonstrated that the perforin expression of CD8+ cells in children aged 0–1 years and 1–15 years constituted 0–3% and 2–11% of the total lymphocyte population respectively. They also found that perforin expression of CD56+ cells constituted 4–30% and 12–34% of all lymphocytes in those aged 0–1 years and 1–15 years. We used two-colour FCM to test the proportions of the CD3+ and CD8+ cells expressing perforin in 15 patients and 30 normal control subjects. Some HLH patients showed very low percentages (around 1–2%), but a complete lack of perforin expression was only noted in patients who were later found to have perforin gene mutations. Thus, low perforin expression correlates with the presence of perforin gene mutations, indicating that FCM of lymphocyte perforin levels is useful for the screening of FHL2.

We analysed perforin expression by FCM and sequenced exons 2 and 3 of the perforin gene in 14 HLH patients, and found five patients with mutations. Thus, mutations in the perforin gene or poor perforin protein expression were found in 35·7% of the HLH patients studied. This incidence is comparable to previous estimates of the prevalence of perforin abnormalities (Stepp et al, 1999; Clementi et al, 2001; Ericson et al, 2001; Kogawa et al, 2002; Suga et al, 2002). Two of the mutations occurred in more than one patient, namely, 1090–1091delCT and 207delC. The former mutation causes a frameshift after Leu364 and termination at 456 amino acid (AA), while the latter mutation causes a frameshift after Pro69 and termination at 106 AA. Suga et al (2002), who found perforin mutations in three Japanese HLH patients, also detected the 1090–1091delCT mutation in one patient and 207delC in another. Thus, 1090–1091delCT has been observed in five of the eight Japanese HLH patients with perforin mutations (62·5%), while 207delC has been detected in three patients (37·5%). This suggests that these particular perforin mutations may be characteristic of the Japanese population. This notion is supported by previous studies that found specific perforin mutations in other ethnic populations. For example, the Trp374stop appears to occur at high frequencies in Turkish families (zur Stadt et al, 2002), while Ala91Val may be an Italian mutation (Clementi et al, 2002). In addition, 50delT appears to be characteristic of African–American patients, while Gly149Ser occurs commonly in Hispanic families (Alexandra H. Filipovich, personal communication). These observations suggest that most present-day patients of FHL2 in most ethnic populations have received their perforin gene abnormality from a common ancestor. Supporting this notion is that although none of our perforin-mutated patients were directly related, the four patients with the 1090–1091delCT mutation all have ancestors who originated from the south-western part of Japan. As summarized in Fig 5, it was surprising that so many single nucleotide or two to three nucleotide deletions occur in exons 2 and 3 of the small gene of perforin, resulting in clinical features of HLH.

Feldmann et al (2002) have characterized the perforin genotype and phenotype of 14 unrelated families with perforin deficiencies. In all patients, perforin gene mutations led to undetectable intracellular perforin expression in cytotoxic cells, although some residual T-cell cytotoxic activity was associated with certain missense mutations. Clinical and biological analyses could not differentiate in general between the patients with nonsense and missense mutations. However, it was found that age at diagnosis, which tended to be similar for the members of the same family, was delayed in the patients from two families with missense mutations. These observations are supported by the fact that the single HLH patient we identified with a perforin gene missense mutation was diagnosed at 11 years of age. This is much older than the average age of FHL onset (less than 2 years old) (Janka, 1983; Henter et al, 1991).

A subpopulation of FHL patients with late-onset disease has been previously observed. For example, Allen et al (2001) described four late-onset HLH patients who were 12–24 years old when they were diagnosed. Three of these four patients were related and all patients had one or two affected siblings. The perforin gene was not sequenced but perforin expression was normal in two patients and not tested in the remaining two patients. Clementi et al (2002) also reported two siblings with perforin mutations who developed FHL at 21 and 22 years of age. Such atypical late-onset FHL patients make the differential diagnosis of HLH very difficult. Delineation of the specific perforin mutations that are associated with late-onset FHL will thus be very useful for diagnostic purposes, as well as providing clues about the biological mechanism that leads to this atypical FHL.

Our late-onset patient had two siblings who developed a fatal FHL in early infancy. Although a missense mutation might be responsible for the late onset of the disease, the same mutation has resulted in different age of onset within the same family. This finding underlies the influence of other genetic and environmental factors on the phenotype of FHL. However, whatever the onset age, the patient has a dismal outcome once the disease develops and is not successfully transplanted. Accordingly, it is important to measure perforin expression in all patients with HLH or haemophagocytic syndrome, including older children or young adults. This will improve the frequency of a correct diagnosis and ensure the use of the appropriate treatment needed to save these patients' lives.

When HLH is diagnosed, patients may have FHL or non-hereditary (secondary) disease. In Japan, a subset of patients develop typical EBV-HLH as a non-hereditary disease. In this study, we, therefore, defined 14 patients as HLH and six patients as EBV-HLH (Table I). Although EBV can act as a triggering factor in the development of FHL (Imashuku et al, 2002), the six patients of EBV-HLH in our study had normal perforin genes. To date, consistent genetic markers associated with the development of EBV-HLH have not been identified. A few EBV-HLH patients in Japan have been reported to have abnormalities in their SH2D1A gene, which has been identified as the causative gene of X-linked lymphoproliferative disease (XLP) (Sumazaki et al, 2001). But other studies have found that the majority of EBV-HLH patients are negative for this mutation (Arico et al, 2001b; Ma et al, 2001). Thus, it is likely that mutations in cytotoxic molecule(s) other than PRF1 or SH2D1A are involved in the highly prevalent EBV-HLH disease in Japan and other east Asian areas.

In summary, we found that lack of perforin expression correlates with the presence of mutations in the perforin gene, indicating that FCM can be gainfully employed to monitor perforin abnormalities in HLH patients. The precise genetic abnormalities causing HLH in the remaining patients are unknown. We also found two perforin mutations that frequently appeared in Japanese patients. Correct diagnosis is critical in improving the therapeutic outcome of patients with HLH. Consequently, it is necessary to further analyse the perforin gene mutations in HLH patients. This information will help improve the screening for perforin mutations. Furthermore, it may be of great clinical benefit if the other gene(s) responsible for other HLH patients can be identified.


We thank the many physicians who participated in this study by supplying the peripheral blood of the HLH patients. We also thank Yasuko Hashimoto for excellent secretarial assistance. This work was in part supported by the Grant-in-Aid for Scientific Research in Japan.