Molecular and flow cytometric characterization of the CD4 and CD8 T-cell repertoire in patients with myelodysplastic syndrome


J. Joseph Melenhorst, Stem Cell Allotransplantation Section, Hematology Branch, National Heart, Lung, and Blood Institute, National Institutes of Health, Building 10, Room 7c103, 9000 Rockville Pike, Bethesda, MD 20892, USA. E-mail:


Summary. We studied 18 patients with myelodysplastic syndrome (MDS), measuring clonality and T-cell receptor Vbeta (TCRBV) expression of CD4 and CD8 T cells by polymerase chain reaction and by flow cytometric analysis of TCRBV families. The CD4 and CD8 T-cell repertoire in most MDS patients is characterized by an abnormal TCRBV-restricted expansion of T cells in CD4 and CD8 cells, and increased expression of the CD8 effector marker CD57 of multiple TCRBV in CD8 cells. Clonality analysis of CD4 and CD8 cells showed that seven of 10 patients analysed had a major clone in the CD8 cells but not in CD4 cells. Furthermore, in one patient we found that both the CD57 and CD57+ fraction contained the clone (which was absent from the TCRBV-negative fraction). These data suggest that, in MDS, multiple T-cell expansions can be found in both helper and cytotoxic T cells, and that, in the CD8 cells, T cells functionally differentiate in vivo from memory to effector T cells. Together, these data support the hypothesis of the involvement of T cells in the pathogenesis of MDS.

Myelodysplastic syndrome (MDS) is a haematopoietic disorder of clonal pluripotent stem cells, bone marrow failure and varying degrees of pancyctopenia. The lineage involved in MDS is mostly the myeloid series alone, and sometimes B cells, but without the involvement of T lymphopoiesis (Janssen et al, 1989; Van Kamp et al, 1992; Weimar et al, 1994; Kroef et al, 1997). There is increasing evidence for an autoimmune process in the pathogenesis of the bone marrow failure that accompanies MDS (Hellstrom-Lindberg et al, 2000). Immune suppression with antithymocyte globulin (ATG) and cyclosporine has been shown to partially improve bone marrow function in about half of the MDS patients (Biesma et al, 1997; Molldrem et al, 1997). Furthermore, haematological response to ATG correlated with disappearance of T cell-mediated haematopoietic progenitor cell inhibition and alterations in the T-cell receptor Vbeta (TCRBV) repertoire (Molldrem et al, 1998).

Antigen-driven proliferation of T cells can produce clonal or oligoclonal expansions in peripheral T cells. Such T-cell expansions have been investigated using molecular techniques in a variety of diseases, including haematopoietic malignancies, such as hairy cell leukaemia (Kluin-Nelemans et al, 1996), myeloma (Moss et al, 1996; Halapi et al, 1997; Raitakari et al, 2000) and chronic lymphocytic leukaemia (Sherman et al, 1993; Farace et al, 1994; Serrano et al, 1997). The TCRBV repertoire in MDS was shown to contain oligoclonal expansions using a spectratype assay (Epperson et al, 2001). Here we looked for evidence of ongoing T-cell activation in patients with MDS by quantifying the T-cell BV repertoire in CD4 and CD8 T cells in MDS patients compared with that of age-matched controls, and by measuring the expression of the effector cell marker CD57 in individual BV families. We found quantitative differences in TCRBV expansion in MDS patients together with increased CD57 expression, suggesting ongoing T-cell activation in MDS.

Materials and Methods

Patients.  Eighteen patients, median age 65 years (range 23–81 years), were studied (Table I). Patients met the diagnostic criteria of MDS, with dysplasia in two or three lineages with or without karyotypic abnormalities. They had red blood cell transfusion dependence with or without neutropenia or thrombocytopenia and varying degrees of marrow cellularity. Patients gave written informed consent for investigation and treatment with antithymocyte globulin (ATG) with or without cyclosporine under the National Institutes of Health (NIH) protocols 95-H-189 and 98-H-0122. Peripheral blood for these studies was obtained before immunosuppressive treatment was given. The mononuclear cell (PBMC) fraction was isolated by Ficoll-Isopaque density centrifugation and cryopreserved in liquid nitrogen until further use.

Table I.  Patient characteristics.
× 109/l
× 109/l
RBC transfusion
  1. F, female; M, male; RA, refractory anaemia; RARS, refractory anaemia with ring sideroblasts; RAEB, refractory anaemia with excess of blasts; RBC, red blood cell.

MDS0348FRARS46xx (28/31); −7; +83·69310+
MDS0661MRA46xy (8); 45x, –y (12)0·9020+
MDS0864MRARS46xy; del 11; dup 1, del 170·36115+
MDS1471FRARS46x, t(x;21)0·94158+

Antibodies.  Monoclonal antibody to CD57 (IgM) was purchased from Beckman/Coulter (Fullerton, CA, USA), and antibodies to CD4 and CD8 were obtained from Becton Dickinson (San Diego, CA, USA). The antibodies to the TCRBV proteins were obtained from Biodesign (Saco, ME, USA). All 21 BV-specific antibodies and the CD57, CD4 and the CD8 antibodies were titred, using 0·3 × 106 PBMC per sample, to obtain optimal and consistent staining of T cells.

Flow cytometric analysis.  PBMC from healthy controls and patients were thawed and cultured overnight in Roswell Park Memorial Institute (RPMI) medium/10% fetal calf serum (FCS) with a small amount of DNase. The next day, the cells were washed with phosphate-buffered saline (PBS)/0·2% bovine serum albumin (BSA) and stained with CD4 peridinin chlorophyll (PerCP) CD8a-allophycocyanin (APC) and CD57-fluorescein isothiocyanate (FITC), and either one of the phycoerythrin (PE)-conjugated anti-TCRBV antibodies. Quadrants were set using CD8α-APC and CD4-PerCP, and TCRBV expression was determined within each subset. Furthermore, CD57 expression per TCRBV was examined in CD8 cells. Data acquisition was performed on a Becton Dickinson facscalibur and analysed using the cellquest software. To ensure that sufficient numbers of TCRBV+ T cells were obtained for statistically significant analysis, at least 500 events per CD4+ or CD8+ cells were acquired per TRCBV.

Flow cytometric purification of CD8+ TCRBV+ T cells and analysis of clonality.  To determine the TCRG rearrangement pattern in the CD8+ T-cell subsets in one MDS patient, cells were stained with CD8α-PC5, anti-TCRBV8 and CD57-FITC, and sorted on a Coulter flow cytometer. Viable lymphocytes were gated using the forward and side scatter diagrams, and sort gates for CD8+CD57 and for CD8+CD57+ cells that did or did not express the BV protein were set over viable lymphocyte population. Clonality of these cells was examined using the T-cell receptor-γ (TCRG) polymerase chain reaction (PCR) as described below.

T-cell clonality.  Clonality of T cells was determined by amplifying TCRG gene rearrangements as reported (Melenhorst et al, 2001). Briefly, genomic DNA was isolated from flow-sorted cells or from cells isolated from PBMC using CD4 or CD8 immunomagnetic beads (Dynal A.S., Oslo, Norway), with GeneReleaser (BioVentures, Murfreesboro, TN, USA). TCRG rearrangements were amplified using the Vγ11 primer (complementary to the first eight Vγ gene segments) and primer γ101 (which anneals to Vγ10 and Vγ11 gene segments), in combination with either Jγ11 or Jp11 was used, which are specific for Jγ1 and Jγ2, and Jγp1 and Jγp2 respectively. The PCR products were separated on a 20% non-denaturing polyacrylamide gel and the DNA was stained with ethidium bromide.

Statistics.  Normal TCRBV expression values were generated from the analyses of PBMC of 23 healthy controls, ranging in age from 20 to 74 years. A normal cut-off was set by taking the average plus 2 × the standard deviation (2SD) of the TCRBV expression levels in CD4 cells, CD8 cells and the CD57 +/– ratio. The average and standard deviation for each TCRBV were calculated after excluding the two largest values. For each subject, the number of TCRBV exceeding the normal cut-off was calculated. In theory, this number could range from 0 to 21. The Wilcoxon rank sum test was used to determine whether the number of extreme TCRBV values differed between patients and controls.

To determine whether the magnitude of BV overexpression differed between patients and controls, we derived an ‘expression coefficient’ by subtracting the mean normal value from the measured value and dividing by the SD obtained for that particular BV family. For each individual, the largest of these 21 standardized values was chosen and the Wilcoxon rank sum test was used to determine whether the expansion coefficient differed between patients and controls.


TCRBV Expression

Healthy controls.  In 23 healthy controls, the TCRBV antibody panel identified 63 ± 12% and 53 ± 15% of the BV proteins in CD4 and CD8 cells respectively. Figure 1 shows the normalized mean distribution of BV families for CD4 cells (Fig 1A) and CD8 cells (Fig 1B) ranked by frequency. Also shown is the coefficient of variation (CV) for each BV family, which identified some BV families as being highly variable (e.g. BV18, BV7, BV3) and some with low variability (e.g. BV2, BV14, BV17). While the frequency pattern was similar in CD4 and CD8 cells, CD4 cells showed lower variability than CD8 cells.

Figure 1.

TCRBV expression in CD4 cells (A) and CD8 cells (B) of 23 healthy donors. Shown are the average expression level with standard deviation (solid lines) and the coefficient of variation (triangular symbols).

Using this normal range, 45 TCRBV frequencies in CD8 cells in 16/23 individuals and 43 TCRBV frequencies in CD4 cells in 10/23 individuals were outside the mean +2SD. There was a trend for greater abnormal frequencies in the oldest individuals: only one of 14 individuals under 54 years showed abnormal BV expansion in CD4 cells and one quarter of CD8+ TCRBV expansions also occurred in the younger group. A linear regression of the rank of the number of cells exceeding the threshold on age was performed. Older controls had more CD4 and CD8 cells exceeding the threshold (P < 0·01 in each case).

MDS patients.  Expansions greater than mean +2SD of the normal range in one or more TCRBV families were found in 17/18 individuals: 17 in CD8 cells and 17 in CD4 cells. (Fig 2A and B and Table II). The number of MDS patients showing abnormal TCRBV expansion was significantly greater than in controls, both in CD4 cells and CD8 cells (P = 0·002 and P = 0·002 in CD4 cells and CD8 cells respectively). Furthermore, the degree of expansion of particular BV families was often much greater than that seen in the normal controls (Fig 2A and B, numbers 1–4 are examples). A linear regression of the rank of the number of cells exceeding the threshold on age was performed. Older patient age was not associated with either the number of CD4 or CD8 cells exceeding the threshold (P > 0·5 in each case).

Figure 2.

Figure 2.

Examples of TCRBV expression in CD4 cells (A) and in CD8 cells (B), and CD57 expression in CD8 cells (C) of a 54-year-old and a 70-year-old donor in 1 and 2, respectively, and two MDS patients (3, 60-year-old patient; 4, 67-year-old patient). (A and B) The TCRBV expression is shown as bars, while the cut-off (mean plus 2SD in the control population) is shown as a broken line. (C) The CD57+/CD57 ratio from the same individuals as in A and B is shown as filled symbols, while the cutoff is shown as a broken line with open symbols.

Figure 2.

Figure 2.

Examples of TCRBV expression in CD4 cells (A) and in CD8 cells (B), and CD57 expression in CD8 cells (C) of a 54-year-old and a 70-year-old donor in 1 and 2, respectively, and two MDS patients (3, 60-year-old patient; 4, 67-year-old patient). (A and B) The TCRBV expression is shown as bars, while the cut-off (mean plus 2SD in the control population) is shown as a broken line. (C) The CD57+/CD57 ratio from the same individuals as in A and B is shown as filled symbols, while the cutoff is shown as a broken line with open symbols.

Figure 2.

Figure 2.

Examples of TCRBV expression in CD4 cells (A) and in CD8 cells (B), and CD57 expression in CD8 cells (C) of a 54-year-old and a 70-year-old donor in 1 and 2, respectively, and two MDS patients (3, 60-year-old patient; 4, 67-year-old patient). (A and B) The TCRBV expression is shown as bars, while the cut-off (mean plus 2SD in the control population) is shown as a broken line. (C) The CD57+/CD57 ratio from the same individuals as in A and B is shown as filled symbols, while the cutoff is shown as a broken line with open symbols.

Table II.  Expansion in CD4 cells and CD8 cells in MDS patients and age-matched controls, and the comparison of increase in CD57+/CD57 ratio in CD8 cells in both groups.
Expressions inNumber exceeding
mean + 2SD
(Wilcoxon Rank sum test)
  • *Number of TCRBV and CD57+-expressing CD8 cells expanded per group.

  • Amplitude of standardized TCRBV and CD57 expression levels.

CD8 TCRBV*0710·002
> 4310 
CD4 TCRBV01310·002
> 458 
CD57 ± ratio0610·002
> 4514 
CD8 expression
> 828 
CD4 expression
> 823 
CD57 ± ratio
> 868 

In both CD4 cells and CD8 cells, the amplitude of TCRBV expansion was significantly higher than in controls (P = 0·0027 and P = 0·0061 in CD4 cells and CD8 cells respectively). These findings remained significant when controlling for age, using linear regression with the rank of CD4 or CD8 exceeding the threshold as the dependent variable (P = 0·03 in each case). Thus, there was a quantitative difference in the degree (number and magnitude) of BV family abnormalities seen in MDS when compared with normals. Some patients had marked BV expansion in CD8 cells. In two of these cases, the CD8 cells were clonal, as determined by TCRG PCR, and five of 10 patients had a clonal rearrangement with a polyclonal background (Table III).

Table III.  Molecular analysis of CD4 cells and CD8 cells of MDS patients.
PatientTCRG data in
PBMCCD4 cellsCD8 cells
  1. PBMC, and the CD4 and CD8 fractions were analysed by TCRG PCR. C, clonal; p w/C, clone with polyclonal background; p, polyclonal; nt, not tested.

MDS02 pp with C
MDS04 pp with C
MDS07p w/Cntnt
MDS09Cpp with C
MDS14 pp
MDS15Cpp with C
MDS18Cpp with C

To determine whether the BV dominance represented a clonal expansion, the dominant fraction (BV8) was purified from peripheral blood of one MDS patient by flow sorting and tested for clonality using TCRG PCR. About half of these cells expressed the effector cell marker CD57. As shown in Fig 3, the BV8 fraction was clonally rearranged, but the clonal rearrangement was absent from the BV8-negative fraction. Furthermore, the clone was present in both the CD57/memory cell and the CD57+/effector cell fractions of CD8+BV8+ cells, suggesting that the T-cell clone was able to differentiate from memory to effector cells, the latter of which have been shown to be end-stage cells (Hamann et al, 1997, 1999; Posnett et al, 1999; Melenhorst et al, 2001).

Figure 3.

Molecular evidence that the dominant TCRBV8 contained the clone, and that this clone is CD57 negative and positive. (A) The CD8 cells of a MDS patient overexpress CD57 (filled portion of each bar) and TCRBV8-expressing cells dominate the TCRBV repertoire in this patient. (B) CD8 cells were sorted into BV8-negative fractions (lanes 1), BV8-positive cells (lanes 2), BV8+CD57 cells (lanes 3) and BV8+CD57+ cells (lanes 4).

CD57 expression in CD8 cells.  The proportion of each BV family in CD8 cells expressing the effector T-cell marker CD57 was compared in MDS and controls. Seventeen controls had an increased CD57 expression in one (seven controls), two (four controls) or more than three BV families. Figure 2C shows examples of two controls (1, 2) and two patients (3, 4).

In 17/18 of the MDS patients, an expanded CD57+/CD57 ratio was identified and the difference with the control group was significant (P = 0·0019). Twenty-six of 77 expanded TCRBV in CD8 cells contained a higher than normal proportion of CD57-expressing cells (Table II). Furthermore, not only the number of CD57-expressing CD8 cells was increased, but also the level of CD57 expression per BV was significantly increased (P = 0·01), indicating that these expanded BV families included abnormally high numbers of effector T cells.


Quantitative analysis of the T-cell repertoire by flow cytometric measurement of TCRBV families identifies a pattern of TCRBV frequencies that appears to be determined repertoire selection in the thymus and periphery (Goldrath & Bevan, 1999), by peripheral antigen-driven selection of individual T-cell clones and intrinsic factors (Brinchmann et al, 1996; and data not shown). Our results confirm the reproducibility of the TCRBV family repertoire in a population of normal individuals randomly selected for human leucocyte antigen (HLA) type, and the frequency distribution described here conforms closely with that described by Van den Beemd et al (2000) in 36 normal individuals and Muraro et al (2000) in 64 normal donors. We confirm that the repertoire of the CD4 subset and the CD8 T-cell subset follow a strict rank order of BV usage (Brinchmann et al, 1996). We also found that the TCRBV usage in CD4 cells corresponded with that of the CD8 cells. Our results also identify a spectrum of interindividual variability of BV families. Thus, we found BV18, BV7 and BV3 to be the most variable with CVs in both CD4 and CD8 subsets of > 0·53, and BV2, BV17 and BV21·3 to be the least variable with CV < 0·44. Superimposed upon this regular pattern of BV family frequency, we identified outlying values for some BV families, especially in older individuals. We here confirm previous findings that the older individuals have a significantly more skewed TCRBV repertoire compared with younger persons (Posnett et al, 1994).

Using molecular biological techniques, clonal CD8 T-cell expansions have been found by some investigators in the healthy population and mostly in elderly individuals (Schwab et al, 1997; Wack et al, 1998; Globerson & Effros, 2000; LeMaoult et al, 2000). These T-cell expansions could be identified using flow cytometry (Clarke et al, 1994; Posnett et al, 1994), but the clonal expansion in aged, normal donors cannot always be detected using this technique (Ruiz et al, 1996; Schwab et al, 1997; Esparza et al, 1998) and may thus represent minor expansions. Also, effector cells are more clonal than naive and memory cells (Gorochov et al, 1994; Hamann et al, 1999). Clonal T-cell expansions are considered to represent clonal or oligoclonal expansion of possibly autoreactive T cells and is most marked in CD8 populations, predominantly effector cells (Hamann et al, 1999).

Using tetramer technology, it has recently been shown that persistent T-cell responses can range from 0·1% to 2% to even 50% of CD8 T cells (McMichael & O'Callaghan, 1998). These figures were much higher than detected by limiting dilution assay (LDA). Epitope-specific CD8 T-cell responses in a few BV families may be up to 500-fold higher compared with LDA (McMichael & O'Callaghan, 1998; Maini et al, 1999).

Such antigen-driven T-cell expansions should thus be detectable with monoclonal antibodies recognizing the BV domain. Using spectratyping and molecular techniques, there is increasing evidence for abnormal clonal expansions in MDS. However, the question arises whether the expansions observed in this relatively elderly patient population differ significantly from the clonal perturbations seen in otherwise healthy individuals with increasing age. Here we applied BV repertoire analysis in MDS to rigorously test whether TCRBV variations in MDS differ qualitatively or quantitatively from the perturbations known to occur in older individuals without MDS. It is immediately apparent from Figs 1 and 2 and Table II that MDS patients have an increased number of abnormally expanded TCR compared with the age-matched controls, even after correction for age. Furthermore, the degree of expansion was several orders greater than that seen in older normal individuals. Both CD4 and CD8 subsets showed abnormal expansion in MDS patients. Thus, overall the BV repertoire of MDS patients was significantly more irregular in the representation of individual BV families. Several BV families were repeatedly abnormally expanded in CD4 cells (BV5·2 in both the patients and donors) but not in CD8 cells. These findings support the existence of expanded oligoclonal T-cell populations in MDS.

In a primary CD8 T-cell response, a multitude of naïve T cells start proliferating in response to recognition of their cognate antigen (expansion phase) and, once the antigen has been cleared, the majority undergo apoptosis (about 95%) (contraction phase), leaving behind a small population of memory cells (memory population formation) (reviewed in Goldrath & Bevan, 1999; Maini et al, 1999). In a secondary response, the repertoire responding to the same antigenic stimulus will be essentially the same in clonal composition (Lin & Welsh, 1998; Sourdive et al, 1998; Blattman et al, 2000; Callan et al, 2000). However, the situation is different if the antigenic stimulus persists. It has been demonstrated in viral immune responses that the clonal composition can be stable (Masuko et al, 1996) or decreased in chronic T-cell responses, and eventually may leave behind an oligoclonal or even clonal population (Masuko et al, 1996; Wilson et al, 1998; Smith et al, 2000), or even continue to evolve, with the appearance and disappearance of clones (Goulder et al, 2001). Thus, the apparently oligoclonal nature of TCRBV expansions seen in MDS are consistent with a persistent antigen-driven immune response. The increase of CD57 expression by CD8 cells supports this argument. Many of these expanded BV families showed increased expression of this CD8 T-cell effector marker (Hamann et al, 1997). As the features of expanded TCRBV families expressing CD57 are seen in active immune responses such as acute graft-versus-host disease (Leroy et al, 1986; Fukuda et al, 1994) and viral infection (Wursch et al, 1985; Joly et al, 1989; Labalette et al, 1994; Wang et al, 1995; Dolstra et al, 1996), our findings suggest that ongoing immune responses are occurring in MDS. While these changes could be indicative of an autoimmune process, it can be argued that they might have arisen from the immune stimulation from blood transfusion or repeated infections. Against this possibility was our finding that no TCRBV expansion was observed in younger individuals with haemoglobinopathies receiving regular blood transfusions. Furthermore, none of the patients studied were actively infected at the time of study (Epperson et al, 2001).

There is accumulating but indirect evidence that autoimmune processes are involved in the pathogenesis of bone-marrow-failure MDS. Some patients with MDS show haematological recovery, following treatment with ATG or cyclosporine (Biesma et al, 1997; Molldrem et al, 1997). T cells from MDS patients can suppress autologous bone marrow progenitor cell growth and there are reports of clonal T-cell expansions occurring in MDS (Molldrem et al, 1998). The results reported here extend our earlier findings in MDS, using TCRBV spectratyping (Epperson et al, 2001), and demonstrate that both CD4+ and CD8+CD57+ T-cell expansions commonly occur in these disorders. Although the series is too small to comment upon any specific features of particular MDS subtypes, it is clear that immune abnormalities occur in MDS refractory anaemia (RA), RA with excess blasts (RAEB) and RA with ring sideroblasts (RARS). The improvement in haematopoiesis following ATG in all of these subtypes is consistent with the possibility that an immune process causing myelosuppression is widespread in MDS. Further studies are needed to determine whether BV repertoire analysis could be useful in predicting responders to immunosuppressive treatment or could be correlated with the outcome after immunosuppressive treatment.