Factors involved in the generation of memory CD8+ T cells in patients with X-linked lymphoproliferative disease (XLP)


  • L. Belmonte,

    1. Instituto de Investigaciones Hematológicas, Academia Nacional de Medicina, Buenos Aires, Argentina, and
    Search for more papers by this author
    • These authors contributed equally to this study.

  • C. Parodi,

    1. Instituto de Investigaciones Hematológicas, Academia Nacional de Medicina, Buenos Aires, Argentina, and
    Search for more papers by this author
    • These authors contributed equally to this study.

  • P. Baré,

    1. Instituto de Investigaciones Hematológicas, Academia Nacional de Medicina, Buenos Aires, Argentina, and
    Search for more papers by this author
  • A. Malbrán,

    1. Departamento de Alergia e Inmunología, Hospital Británico, Buenos Aires, Argentina
    Search for more papers by this author
  • B. Ruibal-Ares,

    1. Instituto de Investigaciones Hematológicas, Academia Nacional de Medicina, Buenos Aires, Argentina, and
    Search for more papers by this author
  • María M. de E. De Bracco

    Corresponding author
    1. Instituto de Investigaciones Hematológicas, Academia Nacional de Medicina, Buenos Aires, Argentina, and
      Dr María M. E. de Bracco, Inmunología, Instituto de Investigaciones, Hematológicas (IIHEMA), Academia Nacional de Medicina, P. de Melo 3081, 1425, Buenos Aires, Argentina.
      E-mail: mebracco@hematologia.anm.edu.ar
    Search for more papers by this author

Dr María M. E. de Bracco, Inmunología, Instituto de Investigaciones, Hematológicas (IIHEMA), Academia Nacional de Medicina, P. de Melo 3081, 1425, Buenos Aires, Argentina.
E-mail: mebracco@hematologia.anm.edu.ar


We have analysed the phenotype of T lymphocytes in two X-linked lymphoproliferative disease (XLP) patients with the same SH2D1A mutation differing in initial exposure to Epstein–Barr virus (EBV) and treatment. While memory T lymphocytes (with low CCR7 and CD62L expression) prevailed in both XLP patients, in patient 9, who developed acute infectious mononucleosis (AIM) and received B cell ablative treatment, the predominant phenotype was that of late effector CD8 T cells (CD27, CD28, CCR7, CD62L, CD45 RA+, perforin+), while in patient 4 (who did not suffer AIM) the prevalent phenotype of CD8 T lymphocytes was similar to that of normal controls (N) or to that of adult individuals who recovered from AIM: CD27+ , CD28+, CCR7, CD62L, CD45 RO+ and perforin. CD57 expression (related to senescence) was also higher in CD8 T cells from patient 9 than in patient 4, AIM or N. Persistently high EBV viral load was observed in patient 9. The results obtained from this limited number of XLP patients suggest that events related to the initial EBV encounter (antigen load, treatment, cytokine environment) may have more weight than lack of SH2D1A in determining the long-term differentiation pattern of CD8 memory T cells.


X-linked lymphoproliferative disease (XLP) is a human immune dysfunction caused by inactivating mutations in the SAP/SH2D1A/DSHP gene [1,2]. It is characterized principally by an inappropriate response to Epstein–Barr virus (EBV) infection. Most affected boys develop fulminant infectious mononucleosis and haemophagocytic syndrome, leading to death in 50% of the cases. Surviving patients develop dysgammaglobulinaemia or hypogammaglobulinaemia and/or lymphoproliferative diseases (malignant lymphoma, lymphoid vasculitis). The immune response of XLP patients against EBV is not efficient for the control of EBV expansion. Contrary to the co-ordinated immune response to EBV seen in normal individuals, XLP patients exhibit a dysregulated response characterized by an excessive accumulation of activated CD8+ T cells, natural killer (NK) cells and macrophages and by the inability to control acute EBV infection. Abnormalities in the balance of T helper 1 (Th1)/Th2 responses and increased production of Th1 cytokines have been described in XLP patients and mouse XLP models [3]. The increased occurrence of EBV-associated malignancy in immunocompromised individuals [4], as well as clinical improvement upon adoptive transfer of EBV-specific T lymphocytes, indicates that, as for other viruses, cellular immunity is important for EBV control [5]. Previous work [6] has demonstrated that, in normal individuals, EBV-specific CD8+ T cells accumulate within the CD27+ CD28+ (most of them CCR7+, early differentiated cells) and CD27+ CD28 (intermediate) compartment during primary EBV infection and remain enriched within these compartments during the lifelong persistence of EBV infection (Table 1). In contrast, in the chronic phase of cytomegalovirus (CMV) infection, CD8+ T cells are enriched in differentiated CD27 CD28 cells (late effector phenotype) (Table 1) [6]. The observed differences in the phenotype and function of CD8 memory cells specific for HIV, hepatitis C virus (HCV), EBV and CMV can be attributed to antigen concentration, repetitive stimulation and environmental cytokines occurring in each patient [7,8]. In a mouse experimental system, CD27 engagement by CD70 present on B cells during antigen presentation has also been shown to influence CD27 persistence in memory effector cells [9] and this, in turn, could be important to determine the relative ability of the infected host to control viral infection.

Table 1.  Schematics of naive and effector memory cell phenotype.

In this study, we focused on the quality of CD8+ memory cells in two XLP patients with identical SH2D1A mutations, but different initial exposure to EBV [10] and different treatment [11]. Studies were performed after more than 2 years of recovery from the acute initial episode. This represents an ideal opportunity to examine the influence of SH2D1A deficiency, treatment and initial EBV exposure in the differentiation of CD8+ T cells into memory cells. We will show that in both siblings most CD8 T cells lack CCR7 and CD62L, suggesting a T effector memory (TEM) phenotype [12]. However, one of the patients (no. 9), whose initial EBV encounter as a young adult led to acute infectious mononucleosis (AIM) involving high expansion of EBV and B lymphocyte-ablating treatment [11], developed a CD8 memory profile with a predominance of CD8+ memory cells corresponding to the late memory effector phenotype (CD27 CD28) [6]. In contrast, his brother (patient no. 4), who did not have overt infectious mononucleosis as the XLP disease marker and did not receive B cell ablative treatment, had effector memory CD8+ cells with the CD27+, CD28+ phenotype (early/intermediate effector memory differentiation pattern) as most normal (N) individuals [6], suggesting that the magnitude of initial antigen exposure, inflammatory environment and/or treatment may have defined the quality of CD8+ memory cells present during the chronic phase.

The intracellular interferon (IFN)-γ response after stimulation with EBV lysates was lower and the leucocyte–EBV load remained persistently higher in patient 9 than in patient 4.

The predominance of CD8+ TEM of the late memory effector phenotype could reflect an altered EBV viral load set-point in this patient, considering that he remained asymptomatic in spite of his high EBV burden [10,11]. These data suggest that SH2D1A deficiency appears to be less important than the environmental challenge for the outcome of the CD8+ lymphocyte differentiation pattern.

Patients and methods

XLP patients 4 and 9

Two surviving hypogammaglobulinaemic siblings of an established XLP family were studied. Four siblings had died previously of different causes related to the XLP condition. Among the three surviving brothers, patient 6 had no SH2D1A mutation, and the surviving affected siblings (patients 4 and 9) are subjects of the current investigation. The XLP diagnosis was performed by analysis of restriction fragment length polymorphisms and the results were confirmed later by direct sequence analysis [11]. The inactivating mutation identified in this family resulted from the substitution of a G for a C nucleotide at position 383 within SH2D1A exon 1. Patient 4, now 42 years old, developed a tonsilar lymphoma at 4 years of age that was treated with a combination of chemo- and radiotherapy followed by monthly infusions of IgG. Patient 9 had also received monthly infusions of IgG since 14 years of age. At 26 years of age he developed AIM and was treated successfully with humanized anti-CD20 monoclonal antibody (Rituximab) [11] in combination with acyclovir and continued intravenous IgG infusions. Both patients are now hypogammaglobulinaemic. EBV infection had been confirmed previously in both patients by determination of EBV viral load above 500 viral copies/106 leucocytes [11]. The blood samples used in these investigations were obtained on a monthly basis before IgG infusion, after more than 2 years of the initial XLP marker disease. Informed consent was obtained. A summary of XLP patients' characteristics is given in Table 2. For these studies, normal control individuals (N) consisted of asymptomatic adults (25–40 years of age) with positive EBV serology [IgG anti-virus capsid antigen(anti-VCA) + 1 : 16–1 : 32], indicative of past EBV infection. In addition, blood samples were drawn from three young adults who had suffered AIM more than 2 years before the study as controls of past, symptomatic infection.

Table 2.  Summary of patients 4 and 9 data.
 Patient 4Patient 9
  1. AIM: acute infectious mononucleosis; EBV: Epstein–Barr virus; XLP: X-linked lymphoproliferative disease.

Age XLP presentation3 years26 years
Marker XLP diseaseLymphomaAIM
TreatmentChemotherapyAcyclovir, corticoids, Rituximab
EBV viral burdenLowHigh

EBV viral load

EBV in peripheral blood samples was assayed at Centro Diagnostico IATRIA, Buenos Aires, Argentina. EBV viral load was investigated using a polymerase chain reaction (PCR) assay by amplifying a conserved 210 base pairs (bp) sequence within the EBER 1 gene of EBV-1 and EBV-2 in the presence of an internal quantification standard. Results were expressed as the number of EBV viral copies per 1 × 106 peripheral blood leucocytes or per 1 × 105 lymphocytes, with a range of detection of 25–800 million viral copies.

Leucocyte phenotype studies

Lymphocyte subsets were determined by flow cytometric analysis using whole blood that had been treated briefly with a fluorescence activated cell sorter (FACS) lysing reagent (Becton Dickinson Biosystems, Buenos Aires, Argentina). Fluorescein isothiocyanate (FITC)-, phycoerythrin (PE)- or peridinin chlorophyll (PerCP)-labelled, anti-CD45RA, anti-CD45RO, anti-CD4, anti-CD8, anti-CD27, anti-CD28, anti-CCR7, anti-CD62L and anti-CD57 monoclonal antibodies (Becton Dickinson, San José, CA, USA) were used at the concentrations recommended by the suppliers. Intracellular staining with PE-anti-perforin (Becton Dickinson) was carried out in peripheral blood mononuclear cells (PBMC) that were permeabilized with Fix and Perm solution (Caltag Laboratories, Burlingame, CA). The cells were evaluated using a FACScan cytometer and CellQuest software (Becton Dickinson).

Intracellular IFN-γ determination

As a marker of specific cell-mediated immune response against EBV, the induction of intracellular IFN-γ by stimulation with an EBV lysate was determined [13]. Briefly, 4 × 106 cells of the 95·8 EBV-positive continuous cell line were suspended in 4 ml of complete RPMI-1640 medium containing penicillin, streptomycin and 10% fetal calf serum (FCS) (Bioser, Buenos Aires, Argentina). Cells were washed, resuspended in 1·8 ml of sterile distilled water and incubated at 37°C for 20 min. Then, tonicity was reconstituted with 200 µl 10× concentrated phosphate buffered saline (PBS), centrifuged again and the cell-free supernatant was saved as a source of EBV antigens. For the induction of intracellular IFN-γ, the pellets corresponding to 1 × 106 cells from patients 4, 9 or N were incubated with 1 ml of the EBV lysate or with 1 ml of complete RPMI-1640 medium in 24-well plastic plates for 12, 24 or 48 h; 1 µl brefeldin (Golgi Plug; Becton Dickinson) was added 12 h before labelling with PerCP-labelled anti-CD4 or anti-CD8 monoclonal antibodies (Becton Dickinson). Cells were permeabilized with Fix and Perm (Becton Dickinson) following the suppliers' instructions, washed with PBS containing 0·5% bovine serum albumin (BSA–PBS) and incubated for 1 h at 4°C with PE-labelled anti-IFN-γ (IFN–PE; Pharmingen, Becton Dickinson) or with IgG1–PE (Pharmingen, Becton Dickinson). The percentage of CD4 or CD8 cells containing intracellular IFN-γ was calculated from the flow cytometry dot plots.


Memory phenotype of XLP CD4 and CD8 T lymphocytes

Expression of CD45RA has been associated mainly with naive T cells, while CD45RO is associated with memory T lymphocytes. Analysis of the expression of these markers revealed that, in contrast to controls (N and AIM), expression of CD45RA was lower and CD45RO higher in CD4+ T lymphocytes of the patients, suggesting a predominant memory phenotype (Fig. 1a, b). Similar results were observed with CD8+ T lymphocytes of patient 4 (relatively low CD45RA and high CD45RO expression). In patient 9, however, expression of CD45RA and CD45RO in CD8+ T cells was similar to that of controls. Because CD45RA appears to be re-expressed as memory CD8+ T lymphocytes differentiating into late effectors we studied co-expression of CD45RA and CCR7, taking into account that late effector T lymphocytes do not express this chemokine receptor. Figure 1c shows that CD4+ and CD8+ cells bearing both CCR7 and CD45RA (naive-like phenotype) were lower than in the N and AIM controls in both XLP patients. Analysis of expression of CD62L (L-selectin), a molecule that has been associated with naive or central memory T cells in combination with CCR7, confirmed that the frequency of CD8 T lymphocytes with potential for lymph node homing was lower in XLP patients 4 and 9 than in controls (Fig. 1d). This suggests that while in both patients' effector memory T cells predominate, XLP patients 4 and 9 differed in the quality of memory CD8+ T cells (most CCR7 CD62L CD8+ T cells were CD45RA+ in patient 9 compared to CCR7 CD62L CD45RO+ in patient 4).

Figure 1.

Memory phenotype of X-linked lymphoproliferative disease (XLP) CD4 and CD8 T lymphocytes. Expression of CD45RA (a) and CD45RO (b) was assessed by flow cytometry in CD4+ or CD8+ T lymphocytes. Co-expression of the chemokine receptor CCR7 and CD45RA (c) or of the homing molecule CD62L and CCR7 (d) was determined in the CD4 or CD8 region of viable lymphocytes. A gate for CD4 or CD8 was established by using peridinin chlorophyll (PerCP)-labelled anti-CD4 or anti-CD8 monoclonal antibodies. Fluorescein isothiocyanate (FITC)-labelled anti-CCR7 and phycoerythrin (PE)-labelled anti-CD45RA or anti-CD62L were used. Mean ± standard error of the mean is shown for eight determinations in samples drawn for 10 months from patients 4 and 9; from eight different normal individuals (N) and from three normal young adults who suffered acute infectious mononucleosis (AIM) for 2 years before the study (AIM). Statistical differences (t-test).

Expression of CD27 and CD28 in CD4+ and CD8+ T lymphocytes of XLP patients

In order to determine the differences in the type of memory effector T cells present in peripheral blood of both XLP patients, expression of surface markers that have been associated with T lymphocytes with different memory phenotype (early/intermediate CD27+ CD28+ T cells or late CD27 CD28 effector T cells) was studied by flow cytometry. Expression of CD27+ CD28+ was markedly reduced in CD8+ T lymphocytes of patient 9 when compared to N and AIM controls and to patient 4, suggesting a predominance of CD8+ T lymphocytes with a late effector memory phenotype (Fig. 2a, b). Although less pronounced than in CD8+ T cells, CD27+ CD28+ expression was also lower than normal in CD4+ T lymphocytes of patients 9 and 4 (Fig. 2a).

Figure 2.

Expression of CD27 and CD28 in CD4 and CD8 T lymphocytes of X-linked lymphoproliferative disease (XLP) patients. Co-expression of CD27 and CD28 was determined in CD4 and CD8 T lymphocytes by triple staining with fluorescein isothiocyanate (FITC)-labelled anti-CD27 and phycoerythrin (PE)-labelled anti-CD28 on a gate defined with peridinin chlorophyll (PerCP)-labelled anti-CD4 or anti-CD8 (a). An example of a dot plot of the CD8 region for a normal control (N), patient 4, patient 9 and a normal individual who recovered from acute infectious mononucleosis (AIM) for 2 years before the study (AIM) is shown (b). The absolute number of CD27+, CD28+ CD4 or CD8 T cells of XLP patients 4 and 9 as well as that of control AIM and N individuals is shown. Total lymphocyte counts and CD8 T lymphocyte counts were similar in both XLP patients 4 and 9 and higher in XLP than in N or AIM (total lymphocytes/mm3, patient 4: 2513 ± 149; patient 9: 2592 ± 184; N: 1380 ± 9; AIM: 1579 ± 48; CD8 T lymphocytes/mm3, 1973 ± 499; patient 9: 2019 ± 489; N: 628 ± 172; AIM: 694 ± 149). CD4 T lymphocyte counts were lower in XLP patients than in N or AIM (CD4 T lymphocytes/mm3, 585 ± 137; patient 9: 615 ± 23; N: 791 ± 165; AIM: 961 ± 81) (c). Statistical differences (t-test).

Total lymphocyte counts were higher in both XLP patients than in controls (lymphocytes/mm3, patient 4: 2513 ± 149; patient 9: 2592 ± 184; N: 1380 ± 9; AIM: 1579 ± 48).

Considering the absolute lymphocyte counts, CD27+ CD28+ CD8+ T cells were two to three times higher in patient 4 than in patient 9 and controls (N and AIM), while in both patients, the number of CD27+ CD28+ CD4+ T cells was lower than in N and AIM (Fig. 2c).

Expression of CD70 in CD19+ B lymphocytes of XLP patients

It has been reported that polyclonal B cell activation leads to protective CD8+ T cell effector memory via retained CD27 expression on memory cytotoxic CD8+ lymphocytes. During secondary antigen encounter, interaction of CD70 present in B lymphocytes with CD27 present on memory CD8+ T cells would lead to persistent CD27 expression and improved protection. Because memory CD27+ B lymphocytes were reduced in XLP patients [11] and B lymphocytes are thought to provide signals for retained CD27 expression on memory CD8+ T lymphocytes [9], we evaluated if CD70 expression was also impaired in B lymphocytes from XLP patients and if this could be related to CD27 persistence in CD8+ memory T lymphocytes. The results of Table 3show that the proportion of CD19+ lymphocytes expressing CD70, as well as that of those with simultaneous CD70 and CD27 expression, was similar in both XLP patients and markedly reduced when compared to that of N and AIM.

Table 3.  CD70 and CD27 expression in B lymphocytes of X-linked lymphoproliferative disease (XLP) patients.
 CD70+ B lymphocytes (%)CD70+, CD27+ B lymphocytes (%)
  1. Expression of CD70 and CD27 was determined on B lymphocytes gated with anti-CD19 in XLP patients, normal individuals who recovered from acute infectious mononucleosis over 2 years before the study (AIM) and from normal controls with positive anti-EBV serology (N). Statistical differences (t-test): XLP patient 4 versus XLP patient 9, non-significant; XLP patient 4 and XLP patient 9 versus N and AIM, P < 0·01.

XLP, patient 4 (n = 5)1·98 ± 1·280·95 ± 0·43
XLP, patient 9 (n = 4)1·74 ± 0·960·46 ± 0·29
AIM (n = 3)9·16 ± 3·267·22 ± 3·11
N (n = 6)14·45 ± 6·210·02 ± 5·46

Senescence and perforin expression in XLP

Highly differentiated CD8+ T cells exhibit characteristics of replicative senescence, including increased expression of CD57 [14]. The results of CD57 expression studies performed on PBMC of XLP patients 4 and 9 indicates that in accordance to the CD27, CD28 phenotype, CD8+ T memory cells from patient 9 exhibited higher CD57 levels than those of patient 4, N or AIM controls (Fig. 3a). This was also coincident with higher expression of perforin in the CD8 T cells of patient 9 than in those of patient 4 or controls (Fig. 3b), confirming the predominance of a late effector memory phenotype.

Figure 3.

Senescence and perforin expression in X-linked lymphoproliferative disease (XLP). Expression of the senescence marker CD57 was determined in CD4 and CD8 T lymphocytes by flow cytometry for patients 4 and 9 (n = 8), for normal controls (N, n = 8) and for normal individuals who recovered from acute infectious mononucleosis (AIM) (AIM, n = 3) (a). Expression of perforin was determined on permeabilized CD4 and CD8 T cells in the same groups (b). Statistical differences (t-test).

Persistently high EBV viral load and CD8+ memory effector phenotype in XLP patients

Because we have observed previously persistently high EBV viral loads in patient 9 up to 18 months post-infection, we assayed the EBV viral loads in this patient and in his brother, patient 4, 1–3 years later, trying to determine if this fact was associated with the differences observed in the quality of CD8 memory T cells in both individuals. After 2–4 years of initial EBV infection, leucocytes from patient 9 repeatedly showed high EBV viral load in association with the prevailing late CD8 T cell effector memory phenotype (CD27, CD28). In contrast, leucocytes from patient 4 were weakly positive or gave negative results (Table 4). Interestingly, no clinical symptoms of persistent EBV infection were observed in either patient.

Table 4.  Epstein–Barr virus (EBV) viral load in X-linked lymphoproliferative disease (XLP) patients.
DateXLP patients
Patient 4Patient 9
EBV DNA copies/106 leucocytesEBV DNA copies/105 lymphocytesEBV DNA copies/106 leucocytesEBV DNA copies/ 105 lymphocytes
  1. Results are expressed as the number of EBV viral copies per 1 × 106 peripheral blood leucocytes or per 1 × 105 lymphocytes. The range of detection of this assay is 25–400 million viral copies. No viral copies were detected in the N group. In AIM controls, only one of the three individuals had 215 EBV DNA copies per 1 × 106 peripheral blood leucocytes; the other two were negative.

7 October 2003Non-detectableNon-detectable2164503
28 June 2005Non-detectableNon-detectable65491601
24 September 200526453113903340

Cell-mediated immunity to EBV in XLP patients 4 and 9

Humoral immunity to EBV is difficult to evaluate in XLP patients because they receive exogenous IgG as replacement therapy for hypogammaglobulinaemia. Therefore, as a surrogate marker of cellular immunity, EBV-induced intracellular expression of IFN-γ was assessed in XLP patients and in N controls. Figure 4 shows that expression of IFN-γ was higher in patient 4 than in patient 9 and in N controls, peaking at 24 h both in CD4 and in CD8 T lymphocytes. In patient 9 IFN-γ expression in CD4 lymphocytes, although lower than that of patient 4, was above the values observed for N controls.

Figure 4.

Induction of intracellular interferon (IFN)-γ by stimulation with Epstein–Barr virus (EBV) antigens. Intracellular IFN-γ was studied by flow cytometry after stimulation for 12, 24 and 48 h with lysates from the EBV-positive 95·8 cell line. The values obtained in the brefeldin controls (0–0·38%) in the absence of EBV were subtracted. Mean ± standard error of the mean is given for each culture time.


CD8+ MHC class I-restricted cytotoxic T lymphocyte responses are essential in controlling virus infections. After exposure to virus in normal individuals, CD8+ T cells are activated to kill virus-infected cells or inhibit viral replication. This can be achieved through diverse effector mechanisms, such as cytolysis and cytokine release [15].

Relatively brief periods of exposure to antigen (24 h) are sufficient to programme clonal expansion, expression of effector functions and differentiation into the memory cells important for the control of chronic persistent viral infections such as EBV infection [16].

The initial expansion phase of CD8+ T lymphocytes is followed by a contraction phase and by generation of memory effector cells. Memory CD8+ T cells are phenotypically and functionally heterogeneous, ranging from naive and early memory T cells (CD27+ CD28+) to intermediate (CD27+ CD28) and fully differentiated (CD27 CD28) subsets [6,8]. Memory CD8+ T cells that lack expression of lymph node (LN) homing receptors (CD62L and the chemokine receptor CCR7) are termed effector memory cells (TEM), whereas memory T cells expressing CCR7 and CD62L are termed central memory cells (TCM) [12]. On the basis of experiments performed using mouse T cell receptor (TCR)-transgenic CD8+ memory T cells specific for an epitope of lymphocytic choriomeningitis virus (LCMV), Wherry et al. have proposed that TCM (capable of self-renewal and rapid responsiveness) are more efficient in mediating protective immunity than CD8+ TEM [17].

In this study, we have analysed (many months after the original EBV challenge) the phenotype of CD8 and CD4 lymphocytes in two XLP brothers with the same SH2D1A-inactivating mutation who suffered different initial exposure to EBV. In one of them (patient 9), B lymphocyte-ablative therapy (anti-CD20) was given during the acute phase of the disease in order to reduce the number of EBV target cells [10].

In normal individuals, up to 50% of the CD8+ T lymphocytes may be EBV-specific during AIM [18]. CD8 cell numbers decrease and a pool of memory cells is established following recovery from AIM. This pool of CD8 memory T cells persists over many years, and their phenotype corresponds mainly to that of cells of the ‘early/intermediate’ phenotype [6]. Because patients with XLP who suffer AIM experience expansion and activation of CD8+ T cells with subsequent inflammatory consequences, the final profile of memory effector cells could differ from that of normal individuals or of patients who did not suffer AIM. The two patients involved in this study provided an opportunity to analyse this, as their initial exposure to EBV was different and they were subject to different treatments.

Encounter with antigen leads to loss of the expression of the high molecular weight isoform CD45RA and appearance of the low molecular weight isoform CD45RO [19,20]. Because of this, CD45RO expression is associated generally with memory cells, while CD45RA expression is taken as a marker of naive cells. In a recent study, the frequency of the subsets of CD4+ T lymphocytes having naive (CD45RA+ CD27+), conventional memory (CD45RA CD27+) or effector memory phenotype (CD45RA CD27) in a series of 12 XLP patients were similar to that of N controls [21], suggesting that SH2D1A deficiency did not alter differentiation of CD4+ T cells, although their number was reduced. In contrast, in this study, both the percentage of CD45RA+ CD4+ T lymphocytes and that of CD45RA+ CCR7+ CD4+ T lymphocytes were reduced in patients 4 and 9 (Fig. 1), indicating the predominance of an effector memory phenotype. CD45RO was also higher than that of controls in both XLP patients CD4+ T cells and in patient 4 CD8+ T cells, as expected for memory cells. However, CD45RA expression was significantly higher in the CD8+ T lymphocytes of patient 9 than in patient 4 or N and AIM controls. This can be explained because CD45RA can be re-expressed on highly differentiated CD8+ T cells that lose CD27 expression [22]. Increased numbers of this subpopulation (CD8+ CD45RA+ CD27) had been detected previously in XLP patients and in older individuals. Furthermore, in one of the XLP patients reported, this phenotype prevailed in the CD8+ T cell population that was EBV-specific in association with shorter telomere length, indicative of replication senescence [23]. In the present study, we observed that in both XLP patients 4 and 9, expression of CCR7 and of the homing molecule CD62L were reduced in CD4 T lymphocytes, confirming differentiation to memory cells [16]. Moreover, the proportion of CCR7+ CD62L+ cells was even lower in CD8 T lymphocytes, indicating that TEM cells predominate in the CD8+ memory cell compartment. However, memory CD8+ T lymphocytes were qualitatively different in both siblings. In patient 4, who did not have AIM as a triggering XLP event, the phenotype of memory CD8+ lymphocytes was similar to that of N. The fact that this patient suffered symptomatic EBV infection in adult age was not sufficient to explain impaired expression of CD27 and CD28 in CD8+ T lymphocytes, as normal young adults who recovered from AIM had levels of CD27 and CD28 that were similar to those of N and of patient 4. Low CD27 expression in the CD8+ T lymphocytes of patient 9 could be due, in part, to insufficient interaction of B cells and CD8+ T lymphocytes following the initial EBV infection. It has been shown that interaction of CD70 expressed on activated B cells during the contraction phase of the cytotoxic response is necessary for the retained CD27 expression on memory CD8+ T lymphocytes [9]. Because B lymphocytes were absent during at least 6 months in patient 9 because he received anti-CD20 treatment [10] during the contraction phase of the cytotoxic response, this factor, in addition to the magnitude of the antigen impact, cytokine release and inflammation, must be taken into account. Assay of CD70 expression on the remaining B lymphocytes revealed a marked decrease in the proportion of CD70-expressing B lymphocytes and CD70+ memory B cells (Table 3) in both patients, irrespective of the differences in the CD8+ memory phenotype (late in patient 9 compared to early in patient 4). Therefore, if deficient CD70-dependent stimulation was a factor affecting long-term CD27 expression in CD8+ T cells in patient 9, absence of CD70+ B lymphocytes during the contraction phase caused by anti-CD20 treatment was probably more important than the persistently low levels of CD70 found later in both patients.

In accordance with the late TEM phenotype [6] of the CD8+ T lymphocytes of patient 9, the intracellular perforin content was higher than that of patient 4, N and AIM controls. As expected for a highly differentiated phenotype (CD27 CD28), expression of CD57, a molecule that has been associated with activated senescent cells [12], was high both in CD4+ and in CD8+ T lymphocytes from patient 9. On the other hand, memory CD8+ T lymphocytes from his brother, patient 4, with low or undetectable EBV load as well as those from N with positive EBV serology, or of individuals who had suffered AIM and had recovered from infection, corresponded to the ‘early’ effector phenotype [6] with low CD57 expression [12].

Taken together these results suggest, in agreement with the results of Ma et al. [21], that SH2D1A deficiency does not impede differentiation of T lymphocytes. The ‘early’ memory phenotype observed in patient 4 appears to be associated with lower (initial and/or persistent) antigen exposure concurrent with a milder inflammatory milieu during the initial EBV encounter. On the other hand, the ‘late’ phenotype of the CD8+ T lymphocytes of patient 9 could be the result of absence of CD70 stimulus during therapeutic B lymphocyte ablation or of continued exposure to a high EBV load persisting after recovery of AIM. When EBV-induced IFN-γ expression was evaluated as a marker of cell-mediated immunity against EBV, the CD4 and CD8 lymphocytes of patient 4 showed a higher response than patient 9 and N controls (Fig. 4), confirming the immunophenotype results. N controls had a very weak CD8 response and virtually no CD4 response, as expected, because no additional co-stimulatory agents were added [13,24]. It is important to note that neither patient had symptoms of ongoing EBV-related disease [25], suggesting that in spite of the differences in the phenotype of CD8+ and CD4+ T effector cells observed in both patients, the immune system was capable of controlling EBV expansion and its pathological consequences.

In contrast to results reported for post-transplant patients [26], it has been shown that a high EBV viral load may not always be associated with the increased occurrence of EBV-associated malignancy [27]. For instance, the high EBV viral load that is observed frequently after HIV infection has no predictive value for the occurrence of AIDS-related non-Hodgkin's lymphoma [28].

While increased frequency of EBV-related non-Hodgkin's lymphoma (NHL) is observed in XLP patients who survive the acute EBV infectious episode, it is not possible to predict that patient 9 will be more at risk of developing NHL than patient 4 in relation to their different EBV viral burden. In summary, although more patients with SH2D1A deficiency should be studied, these results suggest that the nature of the initial encounter of XLP patients with EBV and the magnitude of the initial EBV-related disease and/or B cell-ablative treatment were more important than SH2D1A deficiency in order to determine the long-lasting differentiation pattern of the effector CD8 T lymphocytes, which persisted unchanged for more than 3 years.


We gratefully acknowledge the continued collaboration of patients 4 and 9. We are indebted to Dr N. Riera, Dr N. Galassi and Ms M. Felippo for help and discussion on flow cytometry studies. This work was supported by grants from Fundacion Rene Baron, SECyT (PICT 10736–2002) and CONICET (PIP 5679), Buenos Aires, Argentina.