Main

The high-affinity receptor for IL-2 (1) is composed of three subunits: α (CD25), β (CD122), and γ (γcommon). Whereas the β and γ chains are constitutively expressed upon T lymphocytes, α expression is restricted to the early stages of thymocyte development and to activated mature T lymphocytes. Although the β and γ chains together can form an IL-2 receptor (IL-2R) of low affinity, the α chain cannot form a functional receptor in the absence of the others (2). The presence of the high-affinity receptor upon activated peripheral T cells is believed to be necessary for optimal proliferative responses to IL-2 after stimulation of the T-cell antigen receptor; however the role of the αβγ complex upon triple-negative thymocytes (CD3−, CD4−, CD8−) is less certain.

Deletion of the IL-2R γ chain in humans results in the development of severe combined immunodeficiency (3) (X-linked). X-linked severe combined immunodeficiency is characterized by a great reduction in or absence of peripheral T lymphocytes (3) due to the participation of the γ chain in multiple cytokine receptor complexes (i.e. with IL 2, 4, 7, 9, and 15). In mice, B-cell development is also ablated (4). In sharp contrast, the absence of CD122 manifests in the constitutive activation of B and T lymphocytes accompanied by the development of autoimmune responses resulting in premature death (5). The absence of CD25 in mice appears to cause the least severe impairment of the three. Young CD25-null mice have phenotypically normal T- and B-cell populations, whereas older animals exhibit enlarged lymphoid glands due to increased B- and T-cell populations (as the result of inefficient activation-induced cell death) and a propensity to develop autoimmune disorders (6).

We have shown that CD25 deficiency in humans is more severe than in mice. Presentation begins early in life with recurrent infections and dramatic lymphocyte infiltrates in multiple tissues (7). The deficiency of the α chain of the IL-2R was caused by a 4-bp deletion resulting in a complete lack of CD25 expression (8).

We show here that the consequence of CD25 deficiency is decreased apoptosis in the thymus, thus affecting negative selection. The result is the expansion of presumably autoreactive T-cell clones in multiple tissues causing inflammation.

METHODS

Tissue section preparation and staining.

Thymic biopsies from the patient and a normal control undergoing cardiac surgery were snap-frozen. Serial cryostat sections (4 mm) were mounted on glass slides and air-dried. Tissue sections were either stained with hematoxylin and eosin or immunostained with antibodies raised against CD1, CD25, bcl-2, or CD3, using a three-stage biotin-avidin-peroxidase staining procedure.

Flow cytometry.

EBV-transformed B cells were stained with monoclonal anti-CD25 conjugated with phycoerythrin or with isotype-matched control antibodies (Coulter) and were analyzed on an Epics Coulter V flow cytometer.

Western blots.

Peripheral blood lymphocytes (2 × 106) were lysed in SDS-sample buffer, electrophoresed on SDS 8% polyacrylamide gels, and transferred to nitrocellulose membranes (Hybond-C, Amersham). After nonspecific blocking with 5% milk solids in PBS, immunoblots with anti-IL-2R α and anti-IL-2R β (Santa Cruz Biotechnology) were performed in PBS containing 0.05% Tween 20 and 1% milk solids. Bound antibody was detected with the secondary reagent horseradish peroxidase conjugated with donkey antibody to rabbit immunoglobulin (Amersham) and developed by enhanced chemiluminescence (ECL, Amersham).

Reverse transcriptase-PCR.

RNA was extracted from patient peripheral blood lymphocytes by using Trizol (GIBCOBRL). Random priming was used in first-strand cDNA synthesis, and PCR was performed with primers corresponding to GGTCCCAAGGCTCAGGAAGATG and TGGTAAGAAGCCGGGAACAGACACAG, comprising CD25 5′ and 3′ untranslated sequences, respectively. The amplification conditions of 94°C denaturation for 30 s, 60°C annealing for 30 s, and 72°C extension for 60 s were used for 35 cycles in a Perkin-Elmer 9600 thermal cycler. PCR products were cloned in pUC19 and sequenced with the Sequenase kit (United States Biochemical) according to the instructions of the supplier.

Quantification of mRNA for Vβ families by PCR.

Specific mRNA species were quantified by PCR using synthetic DNA as an internal standard, as previously described (9). The technique involves coamplification of a target cDNA (produced from the corresponding mRNA by reverse transcription) and of the internal standard. The target cDNA and the internal standard use the same primer sequences but yield PCR products of different sizes that can be resolved on gel electrophoresis. In the exponential phase of the amplification, the amount of target cDNA was quantified by comparison with the amplification of various amounts of the internal standard. The internal standard containing the primer sequences for the 20 Vβ families of the T-cell receptor (TCR) was constructed by the technique of oligonucleotide overlap extension, as previously described (10). The design of the 5′ and 3′ primers used was based on published cytokine and cytokine-receptor cDNA sequences (9). Specific primers corresponding to mRNA sequences downstream from each 3′ PCR primer were used for reverse transcription, as previously described (9).

Total cellular RNA was prepared according to the method of Chomcyznski and Sacchi (11). RNA concentrations were measured spectrophotometrically (Beckman Instruments DU-40, Fullerton, CA, U.S.A.). The integrity of the RNA was confirmed by electrophoresis under denaturing conditions on a 1% agarose gel.

cDNA was prepared by reverse transcription using standard techniques. For quantitative analysis, the relevant bands were excised from the gel and the amount of incorporated isotope was determined by scintillation counting. The amount of radioactivity recovered from the internal standard gel bands was plotted against the corresponding internal standard DNA concentration, and the amount of cDNA present in each sample was calculated from the standard curve.

In situ detection of apoptotic cells.

In situ detection of apoptotic cells was carried out in paraffin-embedded sections of control and patient's thymuses using the Oncor ApopTag TM Plus kit (Concor, Gaithersburg, MD, U.S.A.). The method was as described by the manufacturer, in which residues of digoxigenin-nucleotide are catalytically added to the DNA by terminal deoxynucleotidyl transferase [TdT; (12)]; the incorporated nucleotides were then detected using an antidigoxigenin antibody conjugated to a fluorophore (fluorescein). The sections were counterstained with Oncor propidium iodide/antifade. Apoptotic cells were viewed by epifluorescence using standard fluorescein excitation (494 nm) and emission filters.

RESULTS AND DISCUSSION

Clinical and immunologic phenotype.

A male child of first-cousin parentage presented with cytomegalovirus (CMV) pneumonitis, persistent oral thrush, and candida esophagitis at the age of 6 mo. He subsequently developed chronic diarrhea and failed to thrive, probably due to chronic adenovirus gastroenteritis. From the age of 8 mo, lymphadenopathy and hepatosplenomegaly became increasingly apparent, with no significant abnormality of liver function. He subsequently required hospitalization for recurrent exacerbation of lung disease. By the age of 3 y, he had developed gingivitis, chronic inflammation of his mandible, and chronic lung disease involving the right upper lobe.

The patient was found to have anemia (Hb, 9.6 g/L) that was proven to be due to iron deficiency. There was no evidence of blood loss and, therefore, the cause of the iron deficiency was unknown. Although serum IgM levels were normal (0.2–1.5 g/L) and IgG somewhat increased (7–15 g/L; normal age-matched values are 4.5–14.3 g/L), IgA levels were low (<0.07 g/L; normal, 0.2–1 g/L). The absolute number of peripheral CD3+ T lymphocytes was relatively low (922 CD3+ cells/μL; normal range, 885–2270), with a significant reduction in CD3+ CD4+ cells (30% of peripheral mononuclear cells; normal range, 60–85%), resulting in an abnormal CD4/CD8 ratio of 1:1 (normal, 2:1). In vitro assays of peripheral lymphocyte function demonstrated a reduced responsiveness to stimulation by anti-CD3 (11% of control), phytohemagglutinin (20% of control), and other mitogens. Addition of exogenous IL-2 (500 U/mL) could not significantly rescue these proliferative responses. Reasoning that some of the patient's symptoms might stem from an autoimmune phenomenon, a 3-wk course of cyclosporin A was administered. Indeed, lymphadenopathy, hepatosplenomegaly, and coughing markedly improved, but diarrhea and abdominal pain worsened. This was accompanied by isolation of adenovirus, CMV, and torovirus from his stool. This further suggested an underlying severe immunodeficiency, which was proven 5 mo later by the patient's inability to reject an allogeneic skin graft that was obtained from a healthy donor (EBV, CMV, HIV, and hepatitis B negative). Consequently, the patient underwent an HLA-matched related (from his sister) allogeneic bone marrow transplant after cytoreduction. Engraftment was rapid, and a complete resolution of symptoms ensued. Recent follow-up visits revealed a sustained engraftment, as more than 95% of peripheral blood mononuclear cells were of donor origin. He is currently symptom-free, and his growth and development appear appropriate for his age.

Absence of IL-2R α chain (CD25) expression.

The lack of the patient's T-cell response to exogenous IL-2 suggested an aberration of the IL-2R. Indeed, immunohistochemical analysis of patient thymus sections revealed an absence of CD25 expression (not shown). Similarly, an EBV-transformed cell line derived from patient peripheral B lymphocytes did not express detectable CD25 by flow cytometry, whereas EBV-transformed lines from normal individuals typically contained 20–40% CD25-positive cells (Fig. 1A). Western blot analysis of lysates prepared from patient peripheral blood lymphocytes confirmed the absence of CD25 protein (Fig. 1B). In contrast, expression of the common γ chain was normal, and the IL-2R β chain was elevated in patient lymphocytes, presumably reflecting compensation for the lack of the α chain CD25.

Figure 1
figure 1

(A) EBV-transformed patient B lymphocytes (trace i) and normal control B lymphocytes (trace ii) were stained with anti-CD25 and analyzed by flow cytometry, demonstrating the absence of CD25 expression on patient cells (1.1% positive) compared with a normal control (44% positive). Isotype-matched antibody control staining of patient cells (1.8% positive) is shown in trace iii. LFL2, count. (B) Western blot analysis demonstrates the absence of IL-2Rα protein and elevated IL-2Rβ expression in patient peripheral blood lymphocytes compared with normal. (C) A 4-bp deletion in the patient's CD25 gene results in a frameshift in protein translation. The sequence around the deletion (bp 60–64) is shown (MUT) and compared with the sequence and reading frame of normal CD25 (WT).

Full size image

Two forms of CD25 (13–15) message were isolated from patient lymphocyte RNA by reverse transcriptase-PCR, representing the full-length sequence (coding the active receptor) and a secondary splicing product lacking exon 4 (coding an inactive receptor). All forms of the patient's CD25 message were found to contain a 4-bp deletion at bp 60–64, resulting in a translational frameshift (Fig. 1C). Translation proceeds for 20 amino acids before the deletion and resultant frameshift occurs, effectively ablating CD25 expression. A further 25 irrelevant amino acids are added before termination. The homozygous nature of the mutation was confirmed by analysis of sequences from the parents, who both demonstrated a normal and 4-bp-deleted allele.

IL-2Rα (CD25) is required for normal expression of CD1 in thethymus.

CD25 deficiency in this patient provided a unique opportunity to study its role in selection events in humans. Atypically for profound cellular immunodeficiency, the thymus was of normal size, although displaying minimal Hassall's corpuscles and a lack of distinct cortical-medullary demarcation (Fig. 2). Whereas immunohistochemical staining of thymus sections for CD3, CD2, CD4, CD8, and class I and class II MHC (major histocompatibility complex) proteins (not shown) revealed normal patterns of expression, staining for CD1 was completely negative (Fig. 2).

Figure 2
figure 2

An immunohistologic comparison of patient (left) and normal age-matched (right) thymus sections; hematoxylin and eosin (H&E) staining revealed fibrous septa separating lobules of lymphoid tissue as seen in normal thymus. However, although cortex and medulla are present in the patient thymus, there is a loss of the normally distinct demarcation between the two areas. Immunohistochemical analysis clearly showed lack of expression of CD1 (second row), whereas control samples stained strongly for all these markers. In the normal thymus, bcl-2 expression is restricted to the medulla with few positive cells present within the cortex, whereas the patient's sample displays a uniformly high level of bcl-2 expression (third row). Consequently, apoptotic figures in the patient's thymus were dramatically reduced (fourth row). (Figure is printed at 74% of original magnification.)

Full size image

The CD1 family consists of five members (a–e), all structurally related to class I MHC proteins (16). In normal thymus, CD1a is highly expressed on cortical thymocytes and dramatically down-regulated upon progression into the medulla. These proteins may function as nonclassical antigen-presenting molecules, as CD1b has been shown to present mycobacterial lipoglycan antigens (17, 18). A mutation in CD1 was unlikely to be the primary cause of immunodeficiency, as patient monocytes up-regulated CD1 expression normally upon in vitro stimulation with granulocyte/macrophage colony-stimulating factor and IL-4 (19) (not shown). However, to conclusively exclude this possibility, sequencing of patient CD1a cDNA was performed; it was found to be normal. This suggested a failure to provide appropriate signals for CD1 up-regulation in thymocytes. Currently, the significance of CD1 deficiency during T-cell differentiation is difficult to identify. It is possible that rather than serving simply as a nonclassical antigen-presenting molecule, CD1a may participate in mediating thymocyte-epithelium interactions and, in its absence, be partially responsible for the loss of the distinct corticomedullary boundary.

Role of the IL-2Rα (CD25) in apoptosis of thymocytes.

Normally, the induction of CD1 expression in cortical thymocytes precedes a dramatic decrease in the cellular level of the bcl-2 protein (15–18). The presence of the bcl-2 gene product acts to protect cells against programmed cell death, without promoting proliferation (20). Early immature triple-negative thymocytes and more mature double-positive (CD1 CD3hi) medullary thymocytes express high levels of bcl-2 and are relatively resistant to the induction of apoptosis (20–22). To permit immature thymocytes to undergo positive and negative selection, both of which frequently (95%) result in the induction of programmed cell death, a reduction in bcl-2 levels is believed to be required. Whereas normal thymocytes demonstrated distinct areas of low (cortical) and high (medullary) bcl-2 expression, staining of patient thymus sections revealed that all CD25-deficient thymocytes express high levels of bcl-2 protein (Fig. 2). The lack of regulation of bcl-2 expression suggests a possible dependency upon signals from either CD1 or CD25.

It has been reported that CD25+ CD4+ T cells maintain self-tolerance in mice by suppressing autoreactive cells (23). Indeed, even in the absence of thymic irregularities, autoreactive responses emerged in the absence of CD25+ T cells (23). This suggests that the lymphocytic infiltrates observed in human CD25 deficiency could be autoimmune in nature. The persistence of high bcl-2 expression in cortical thymocytes appears to support this view, possibly permitting some thymocytes bearing autoreactive T-cell antigen receptors to escape deletion, as seen in some transgenic mice overexpressing bcl-2 (24).

Normally, thymocytes, which express TCR that binds avidly to an MHC molecule (which is presenting self-derived antigens), are eliminated by apoptosis. To measure the extent of apoptosis in a patient's thymus, we have used an in situ detection method on paraffin-embedded sections obtained from a biopsy specimen. Residuals of digoxigenin-nucleotide were catalytically added to the DNA by terminal deoxynucleotidyl transferase. The incorporated nucleotides were then detected using antidigoxigenin antibody conjugated to a fluorescent dye. Stained paraffin sections of normal thymus showed positive fluorescent T-cell clamps particularly limited to the cortical areas of the thymus lobes. In contrast, the patient's sections, although equally populated with thymocytes, contained barely any positive cells (Fig. 2).

The results promote the hypothesis that CD25 is critical for negative selection of thymocytes, and, in its absence, autoreactive clones that are normally destined to die by apoptosis fail to do so and expand to peripheral tissues and cause damage.

Analysis of TCR Vβ repertoire in peripheral blood and tissueinfiltrates.

The phenotype of CD25 deficiency differs significantly from most other cellular immunodeficiencies, most noticeably with lymphocytic infiltration of multiple tissues (Fig. 3). Unlike CD25-/- mice, which have infiltrates limited to the gastrointestinal tract (6), human CD25 deficiency manifests dense lymphocytic infiltration of lung (Fig. 3, upper), liver (Fig. 3, lower), gut, soft tissue, and bone, accompanied by tissue atrophy and chronic inflammation.

Figure 3
figure 3

Light microscopy of hematoxylin and eosin-stained sections of patient lung biopsy (upper; ×18,250) shows dilatation of airways, partially collapsed lung tissue, and dense infiltration of lymphocytes in the bronchial wall as well as in the lung parenchyma. Specific staining identified no fungal elements, and no viral particles were identified by electron microscopy (not shown). In addition, hepatitis B and C virus, EBV, and CMV sequences were not detected by PCR. Light microscopy of liver sections (lower; ×18,250) shows preserved lobular architecture with marked infiltration of lymphocytes in the portal tracts. Consistent with the analysis of the lung biopsy, no fungal or viral elements were identified. Similar dense lymphocytic infiltrates were found in stomach and duodenal biopsies (not shown).

Full size image

To better understand the nature of the lymphocytic infiltration, we have studied the repertoire of peripheral blood as well as tissue T cells. The antigen-specific TCR is a heterodimeric transmembrane glycoprotein comprised of an α and β chain that is clonally distributed on T cells (25–28). These polypeptides are encoded in the germline by various dispersed gene regions including variable (V), diversity (D, for the β-chain gene), joining (J), and constant (C) gene segments. Functional α and β chains are formed during T-cell development by DNA rearrangements that generate VDJ genes, which are subsequently joined to a C region by RNA splicing (29, 30).

Diversity in the TCR repertoire is generated by random utilization of germline V, D, and J gene regions, junctional variation, and combinatorial association of α and β chains. Based on the sequence data of TCR genes and selection in the thymus, it is estimated that 1.9 × 1020 potential different TCR may be available in peripheral blood and tissues.

The available T-cell repertoire is largely determined by preferential rearrangement of certain TCR V gene families. Sequence analysis of the V region of the β chain was, therefore, widely used to determine the diversity of the T-cell repertoire. Normal individuals express all Vβ families, and sequence analysis typically detects unique sequences of every Vβ clone. In contrast, typical severe combined immunodeficiencies display a severely restricted repertoire with or without clonal expansion (9), whereas autoimmune disorders typically show preferential utilization and expansion of few Vβ families (31).

To understand the role of the IL-2Rα chain in determining the T-cell repertoire, we studied Vβ utilization of the patient's T cells The estimated percentages of T cells bearing different Vβ families in peripheral blood lymphocytes from the patient and the normal control are presented in Figure 4, showing that all Vβ families were represented in the patient sample. However, some variations in Vβ usage could be detected as Vβ2, Vβ10, and Vβ17 were overexpressed whereas Vβ1, Vβ12, Vβ15, and Vβ19 were slightly underrepresented when compared with the control samples.

Figure 4
figure 4

Profile of expression of Vβ families. The amplification of patient and control T-cell RNA through 25 cycles plotted vs the frequency of 20 different Vβ families.

Full size image

We next attempted to determine the clonality of the overexpressed Vβ groups. Sequencing individual Vβ2 clones revealed an identical complementarity-determining region in nine of 20 clones, whereas 11 were unique sequences. Similarly, Vβ10 showed four of nine redundant sequences, whereas no redundancy was detected in control normal T cell (24/24 unique sequences for Vβ2 and 17/17 unique sequences for Vβ10). In contrast, sequencing of Vβ17 did not reveal repeated clones (30/30 unique clones), suggesting polyclonal expansion of this Vβ group. Similar analysis performed on additional Vβ families revealed no redundant clones in Vβ1, 3–9, and Vβ11–20. At least 20 clones were sequenced from each of these Vβ groups with the exception of Vβ4 (nine unique clones), Vβ13 (eight unique clones), and Vβ19 (seven unique clones). Sequencing of TCR Vβ isolates from control T cells did not show oligoclonality.

We next studied patients' tissues for the presence of the expanded T-cell clones, summarized in Table 1. Because only limited amount of tissue was obtained in biopsies, we analyzed gut and lung specimens for the presence of Vβ2 and Vβ10. Indeed, in the gut, 3/5 clones of Vβ2 and 4/4 clones of Vβ10 were identical to the redundant clones identified in peripheral blood lymphocytes. Similar results were obtained by sequencing clones obtained from the lung including 7/13 clones of Vβ2, and only 2/15 of Vβ10 were identical to the redundant clone detected in peripheral T cells. Our experiments cannot exclude the possibility that other Vβ family-bearing T cells infiltrate the tissues because the specimen size did not allow for quantitation of other Vβ populations. Nevertheless, the results indicate that oligoclonal expanded clones are present in the tissues where significant damage occurred in the patient.

Table 1 Analysis of Vβ repertoire in blood and tissues
Full size table

Oligoclonality of T cells may reflect a restricted TCR repertoire frequently identified in severe combined immunodeficiency (SCID) patients (9), or, alternatively, it could represent a peripheral expansion of several clones reminiscent of autoimmune disorders. To determine which of these possibilities is applicable to a CD25-/- patient, we sequenced Vβ clones obtained from the thymus. Eleven of 11 clones of Vβ2, 15/15 of Vβ10, and 8/8 of Vβ17 were unique, and no redundancy was detected. The results indicate that the limited oligoclonality observed in the patient represents an expansion of autoreactive clones in peripheral blood and multiple tissues.

In conclusion, it is conceivable that a combination of at least two events, one intrathymic and the other peripheral, may combine to create the phenotype observed in CD25 deficiency: first, a potentially inefficient negative selection process rooted in abnormal thymic differentiation, as evidenced in the failure to express CD1 and to regulate bcl-2; second, a possible inability to control potential autoreactive cells within the periphery due to the absence of the CD25+ CD4+ T-cell subset (23). Inefficient activation-induced cell death, as observed in the CD25 knockout mouse (6), may also contribute to the phenotype observed. The lack of CD25 expression may also compromise the ability of peripheral T cells to proliferate after stimulation and, thus, result in the profound immunodeficiency observed.