Naturally occurring CD4 þ CD25 þ FOXP3 þ T-regulatory cells are increased in chronic myeloid leukemia patients not in complete cytogenetic remission and can be immunosuppressive Jose M. Rojas * , Lihui Wang * , Sally Owen, Katy Knight, Sarah J. Watmough, and Richard E. Clark Department of Haematology, The University of Liverpool, Liverpool, UK (Received 28 May 2010; revised 31 August 2010; accepted 2 September 2010) Objective. Clinical presentation of chronic myeloid leukemia (CML) requires not only the deregulated tyrosine kinase BCR-ABL, but also the failure of an immune response against BCR-ABLLexpressing cells. T-cell responses against BCR-ABL and other antigens are well- described, but their relevance to the in vivo control of CML is unclear. The suppressive role of naturally occurring T regulatory (T-reg) cells in antitumor immunity is well-established, although little is known about their role in modulating the T-cell response to BCR-ABL. Materials and Methods. Naturally occurring T-reg cells were characterized and quantified by flow cytometry in 39 CML patients and 10 healthy donors. Their function was studied by observing their effect on responses to purified protein derivative, a recall antigen, and on the response of an autologous T-cell line recognizing BCR-ABL. Results. T-reg cells were CD4 + , CD25 + , FOXP3 + , CD127 low , and CD62L high . T-reg numbers in patients in complete cytogenetic remission were significantly lower than in patients not in complete cytogenetic remission (p ! 0.01). T-reg cell depletion using anti-CD25 selection enhanced proliferative responses to purified protein derivative. Furthermore, the interferon- g and/or granzyme-B production of effector cells specific for viral peptides or a BCR-ABL HLA-A3Lrestricted peptide was inhibited when autologous T-reg cells were present. Conclusions. Taken together, these data suggest a role for T-reg cells in limiting immune responses in CML patients and this may include immune responses to BCR-ABL. The increased frequency of T-reg cells in patients with high levels of BCR-ABL transcripts indicates that an immune mechanism may be important in the control of CML. Ó 2010 ISEH - Society for Hematology and Stem Cells. Published by Elsevier Inc. Regulatory T (T-reg) cells were identified 3 decades ago in tumor models and provided an explanation for why highly immunogenic tumors continue to grow in spite of an anti- tumor immune response [1,2]. In 1995, Sakaguchi et al. demonstrated CD4 þ CD25 þ T cells in mice that had potent immune-suppressive activity both in vitro and in vivo [3]. Subsequently, a similar CD4 þ CD25 þ T-cell population was identified in rats and humans [4–9]. These cells were named ‘‘natural’’ T-reg cells because their detection in peripheral blood mononuclear cells did not require in vitro manipulation. More recently, the transcription factor Fork- head Box protein 3 (FOXP3) has been shown to be a key intracellular marker and essential developmental and func- tional factor of CD4 þ CD25 þ natural T-reg cells [10–12], although it has also been identified in nonT-lineage cells [13,14]. Natural T-reg cells constitute 5% to 10% of peripheral CD4 þ T cells in normal mice and humans. In mice, their suppressive role in anti-tumor immunity is well-established; in vivo depletion of natural T-reg cells by anti-CD25 mono- clonal antibodies induces rejection of inoculated immuno- genic tumors [15]. However, this treatment does not induce regression of established tumors, supporting the role of CD4 þ CD25 þ T-reg cells in indirectly facilitating the growth of immunogenic tumors [15,16]. Natural T-reg cells were first reported in human ovarian and lung tumor samples in 2001 [17], and have since been reported in a variety of cancers, including Hodgkin disease [18] and chronic lymphocytic leukemia [19]. In lymph nodes from *Drs. Rojas and Wang contributed equally to the study. Offprint requests to: Richard E. Clark, M.A., M.D., Department of Hae- matology, Royal Liverpool University Hospital, Prescot Street, Liverpool L7 8XP, UK; E-mail: [email protected]0301-472X/$ - see front matter. Copyright Ó 2010 ISEH - Society for Hematology and Stem Cells. Published by Elsevier Inc. doi: 10.1016/j.exphem.2010.09.004 Experimental Hematology 2010;38:1209–1218
Transcript
Experimental Hematology 2010;38:1209–1218
Naturally occurring CD4þ CD25þ FOXP3þ T-regulatorycells are increased in chronic myeloid leukemia patients not
in complete cytogenetic remission and can be immunosuppressive
Jose M. Rojas*, Lihui Wang*, Sally Owen, Katy Knight, Sarah J. Watmough, and Richard E. Clark
Department of Haematology, The University of Liverpool, Liverpool, UK
(Received 28 May 2010; revised 31 August 2010; accepted 2 September 2010)
*Drs. Rojas and Wa
Offprint requests to
matology, Royal Live
L7 8XP, UK; E-mail:
0301-472X/$ - see fro
doi: 10.1016/j.exph
Objective. Clinical presentation of chronic myeloid leukemia (CML) requires not only thederegulated tyrosine kinase BCR-ABL, but also the failure of an immune response againstBCR-ABLLexpressing cells. T-cell responses against BCR-ABL and other antigens are well-described, but their relevance to the in vivo control of CML is unclear. The suppressive roleof naturally occurring T regulatory (T-reg) cells in antitumor immunity is well-established,although little is known about their role in modulating the T-cell response to BCR-ABL.
Materials and Methods. Naturally occurring T-reg cells were characterized and quantified byflow cytometry in 39 CML patients and 10 healthy donors. Their function was studied byobserving their effect on responses to purified protein derivative, a recall antigen, and onthe response of an autologous T-cell line recognizing BCR-ABL.
Results. T-reg cells were CD4+, CD25+, FOXP3+, CD127low, and CD62Lhigh. T-reg numbers inpatients in complete cytogenetic remission were significantly lower than in patients not incomplete cytogenetic remission (p ! 0.01). T-reg cell depletion using anti-CD25 selectionenhanced proliferative responses to purified protein derivative. Furthermore, the interferon-g and/or granzyme-B production of effector cells specific for viral peptides or a BCR-ABLHLA-A3Lrestricted peptide was inhibited when autologous T-reg cells were present.
Conclusions. Taken together, these data suggest a role for T-reg cells in limiting immuneresponses in CML patients and this may include immune responses to BCR-ABL. Theincreased frequency of T-reg cells in patients with high levels of BCR-ABL transcriptsindicates that an immune mechanism may be important in the control of CML. � 2010ISEH - Society for Hematology and Stem Cells. Published by Elsevier Inc.
Regulatory T (T-reg) cells were identified 3 decades ago intumor models and provided an explanation for why highlyimmunogenic tumors continue to grow in spite of an anti-tumor immune response [1,2]. In 1995, Sakaguchi et al.demonstrated CD4þCD25þ T cells in mice that had potentimmune-suppressive activity both in vitro and in vivo [3].Subsequently, a similar CD4þCD25þ T-cell populationwas identified in rats and humans [4–9]. These cells werenamed ‘‘natural’’ T-reg cells because their detection inperipheral blood mononuclear cells did not require in vitromanipulation. More recently, the transcription factor Fork-head Box protein 3 (FOXP3) has been shown to be a key
ng contributed equally to the study.
: Richard E. Clark, M.A., M.D., Department of Hae-
rpool University Hospital, Prescot Street, Liverpool
nt matter. Copyright � 2010 ISEH - Society for Hematolo
em.2010.09.004
intracellular marker and essential developmental and func-tional factor of CD4þCD25þ natural T-reg cells [10–12],although it has also been identified in non�T-lineage cells[13,14].
Natural T-reg cells constitute 5% to 10% of peripheralCD4þ T cells in normal mice and humans. In mice, theirsuppressive role in anti-tumor immunity is well-established;in vivo depletion of natural T-reg cells by anti-CD25 mono-clonal antibodies induces rejection of inoculated immuno-genic tumors [15]. However, this treatment does notinduce regression of established tumors, supporting therole of CD4þCD25þ T-reg cells in indirectly facilitatingthe growth of immunogenic tumors [15,16]. Natural T-regcells were first reported in human ovarian and lung tumorsamples in 2001 [17], and have since been reported ina variety of cancers, including Hodgkin disease [18] andchronic lymphocytic leukemia [19]. In lymph nodes from
1210 J.M. Rojas et al./ Experimental Hematology 2010;38:1209–1218
metastatic melanoma patients, there is an increasedfrequency of CD4þCD25þ FOXP3þ T-reg cells capableof inhibiting CD4þCD25� T-cell and CD8þ T-cell func-tions [20]. Other studies have shown an increase inCD4þCD25þ T cells in the peripheral blood of cancerpatients [21].
Very little is known of the role of T-reg cells in chronicmyeloid leukemia (CML). Increased T-reg cell numbers areobserved in CML patients who relapse after allogeneic stemcell transplantation, suggesting that these cells may bedetrimental to the graft-vs.-leukemia effect of allogeneicstem cell transplantation [22]. These data indicate a possiblerole for T-reg cells in the immune control of CML.
In the present study, we investigated the role of T-regcells in CML patients. The naturally occurring T-reg cellphenotype was first characterized. T-reg cell numberswere then examined in 115 samples, and numbers werecorrelated to treatment parameters. Finally, the effect ofT-reg cell function on immune responses was explored inCML patient samples.
Materials and methods
Patient and normal donor samplesThe study was approved by the Liverpool Research EthicsCommittee and all subjects gave informed consent. Heparinizedblood samples from 39 CML patients in first chronic phase and10 healthy donors were collected and peripheral blood mononu-clear cells (PBMC) were obtained using standard gradient centri-fugation methods. Concomitant samples from patients wereobtained to assess BCR-ABL transcript levels. Table 1 providespatients’ details.
Flow cytometry surface stainingPBMC were washed in phosphate-buffered saline þ0.2% bovineserum albumin (Gen-Probe Diaclone SAS, Besancon, France) þ0.02% sodium azide (Sigma, St Louis, MO, USA) and incubatedfor 20 minutes on ice with the appropriate antibodies. In all exper-iments, the adequate isotype control antibodies were used. Cellswere stained with the following antibodies directly conjugatedwith fluorochrome: anti�CD4-fluorescein isothiocyanate (AbD,Serotec, Kidlington, UK), anti-CD25-Cy5 (AbD, Serotec, Kidling-ton, UK), anti�CD62L-Cy5 (e-Bioscience, Hatfield, UK), anti-CD127-Cy5 (e-Bioscience). Cells were either analyzed on a BDflow cytometer, or further stained for FOXP3 using intracellularstaining.
Flow cytometry intracellular staining for FOXP3PBMC labeled for surface markers were subsequently stained forFOXP3 using the FOXP3 staining kit from e-Bioscience followingmanufacturer’s instructions. Anti�FOXP3-phycoerythrin (PE)-conjugated antibody and its relevant immunoglobulin isotypecontrol (Rat IgG2a-PE) were used to stain the PBMC. Afterwashing, cells were analyzed on a BD flow cytometer.
BCR-ABL messenger RNA quantitationThe BCR-ABL messenger RNA transcript level was measured byquantitative real-time polymerase chain reaction as described
previously [23] using ABL as the control gene, in line with recentrecommendations [24]. Equivalence to complete cytogeneticresponses (CCRe) was defined as BCR-ABL/ABL transcript ratios!1% as demonstrated in our group previously [25].
CD25 isolation and depletionCD25 isolations and depletions were performed using the Dyna-beads� Regulatory CD4þCD25þ T cell kit according to manufac-turer’s instructions (Invitrogene, Paisley, UK). The efficiency ofthe isolations/depletions was assessed by flow cytometry. Typi-cally, the purity of the CD4þCD25� and the CD4þCD25þ popu-lations obtained with the kit was O85%. Data were discardedwhere sample purity was !85%.
Proliferation assays using 5-(and-6)-carboxyfluoresceindiacetate, succinimidyl ester (CFSE)The CD4þ population and CD4þCD25� population used in theseexperiments were obtained using the Dynal CD4þCD25þ isolationkit according to manufacturer’s instructions. Then 1 to 2 � 106
cells were labeled with 0.2 mM 5-(and-6)-carboxyfluorescein diac-etate, succinimidyl ester (CFSE) (Invitrogen, Carlsbad, CA, USA)and incubated for 7 minutes at 37�C. Cells were washed in mediaand resuspended in media for 30 minutes at 37�C to stabilize thestaining. After a final wash, cells were ready for antigen stimula-tion. For stimulation, labeled cells were incubated with purifiedprotein derivative (PPD) as antigen or no antigen as control. After3 to 5 days cells were harvested, washed, and analyzed by flowcytometry. Data are presented as proliferation normalized to thecontrol sample.
In vitro generation of peptide-specific T-cell linesOne to two � 106 PBMC (from HLA-A3þ CML patients) per wellwere seeded in 48-well plates and stimulated with 10 mg/mLpeptide (either HLA-DR�associated influenza hemagglutinin(307�319) peptide: PKYVKQNTLKLAT (HA307); HLA-A3 asso-ciated influenza A NP (265�273): ILRGSVAHK (NP265); PAnDR Epitope (PADRE, a non-natural CD4 epitope shown to stimu-late CD4þ T cells in vivo in CML [26]), aKXVAAWTLKAAaZCwhere X 5 L-cyclohexylalanine; Z 5 aminocaproic acid; a 5 D-alanine; BCR-ABL-A3: KQSSKALQR), 20 IU/mL interleukin(IL)-2 and 0.2 ng/mL IL-12. After 7 days, cells were restimulatedwith irradiated autologous PBMC pulsed with 10 mg/mL ofpeptide, 20 IU/mL IL-2. On day 14 cells were restimulated with50 IU/mL IL-2 and 5 mg/mL peptide. Cells were tested on day21 by ELISPOT for interferon-g (IFN-g) and granzyme-Bproduction.
Expansion of CD4þCD25þ and CD4þCD25� cellsFor the in vitro functional studies, the number of T-reg cells ob-tained after isolation was insufficient. CD4þCD25þ isolated cellswere therefore expanded using anti-CD3 antibody and IL-2 asdescribed by Wolf et al. [21]. Following purification, cell pheno-type was confirmed by antibody staining and flow cytometry.
Cocultures of effector cells with CD4þCD25þ cells andassessment of inhibition by ELISPOTThree to five � 104 peptide-specific T cells per well were culturedin the presence of 1.5 to 2.5 � 104 CD4þCD25þ T cells per well(ratio 2:1) and peptide-pulsed autologous irradiated PBMC (or un-pulsed as control) to assess the inhibitory effects of these cells. Ascontrol, these peptide-specific cells were also cultured with
Table 1. Patients’ details
UPN
Sokal
score
Age
(y) Sex
Transcript
type
Treatment (duration in month
at start of T-reg assessment)
Lymphocyte
counts (�109/L) Response
22 I 54 M e13a2 IM (10) 3.7 NCR
81 62 F e14a2 IM (25) þ EPIC (1) 0.7–1.1 CCR
43 L 36 F e13a2 IM (48) 1.5 NCR
90 I 67 M e14a2 IM (59) þ EPIC (34) 0.7–1.0 CCR to NCR
91 H 46 M e14a2 IM (58) þ EPIC (26) 1.0–-1.2 CCR
53 L 55 F e14a2 IM (27) 1.7 NCR
97 48 M e14a2 IM (49) þ EPIC (12) 1.7–2.0 CCR
98 I 67 F e14a2 IM (45) þ EPIC (24) 1.3–1.9 CCR
105 47 M e14a2 IM (44) þ EPIC (22) 1.1–1.5 CCR
80 H 52 F e14a2 IM (60) þ EPIC (30) 1.1–1.5 CCR
95 40 M e14a2 IM (35) þ EPIC (22) 1.3–1.6 CCR
101 H 26 M e14a2 IM (35) þ EPIC (18) 1.2–2.0 CCR
83 70 F e14a2 IM (64) þ EPIC (30) 1.3–1.8 CCR
33 56 M e14a2 IM (6) 1.3 CCR
56 I 42 M e14a2 IM (45) þ EPIC (32) 1.5–1.7 CCR
73 68 F e13a2 IM (87) 1.0–1.1 NCR
59 42 M e14a2 IM (51) þ EPIC (32) 1.5–1.9 CCR
54 I 41 F e14a2 IM (58) þ EPIC (9) 1.4 CCR
36 53 F e14a2 IM (60) þ EPIC (6) NA NCR
38 I 68 F e14a2 IM (36) þ EPIC (6) 1.2 NCR
82 61 F e14a2 IM (23) þ EPIC (6) NA CCR
79 I 56 F e14a2 IM (17) þ EPIC (5) then BMT 2.1–4.6 NCR then CCR
139 54 F e14a2 IM (60) þ EPIC (5) NA NCR
3 H 62 F e14a2 IM 800 OD (19) 1.1 CCR
47 40 M e14a2 IM (45) compliance problem NA NCR
75 H 59 F e14a2 IM (33) þ EPIC (4) 1.6–2.4 CCR
39 63 F e13a2 IM (60) then Nilotinib (18) 1.6–1.7 NCR
14 69 F e14a2 IM (60) 0.7 NCR
32 L 48 F e14a2 IM (10) 1.5 CCR
48 33 M e14a2 IM (72) then Nilotinib (17) 1.7 NCR
20 L 35 M e14a2 IM (13) then Nilotinib (2) 1.1 NCR
52 H 64 F e14a2 þ e13a2 IM (44) NA NCR
8 I 38 F e14a2 IM (33) þ EPIC (2) 0.7–1.2 NCR
112 37 F e13a2 IM (76) then Nilotinib (3) NA NCR
58 40 M e14a2 IM 800 OD (7) 1.5 CCR
115 L 25 M e14a2 IM (12) 3.1 NCR
44 H 49 F e13a2 IM (12) NA NCR
28 I 33 F e14a2 IM (21) 1.3 CCR
CCRe 5 complete cytogenetic response equivalence; EPIC 5 evaluation of peptide vaccination in CML; F 5 female; IM 5 imatinib; Lymph 5 lymphocyte
counts; M 5 male; NA 5 not available; Neut 5 neutrophil counts; NCR 5 no CCRe; UPN 5 unique patient number; WBC 5 white blood cell counts.
1211J.M. Rojas et al./ Experimental Hematology 2010;38:1209–1218
autologous CD4þCD25� T cells. Proliferative responses weremeasured by CFSE-labeled peptide-specific cells and flow cytom-etry analysis. IFN-g and granzyme-B productions were measuredby ELISPOT assays according to manufacturer’s instructions(Gen-Probe, Diaclone SAS, Besancon, France).
Results
Naturally occurring T-reg cellsin CML express CD4, CD25, and FOXP3PBMC from CML patients were stained for CD4, CD25,and FOXP3 with the appropriate immunoglobulin isotypecontrols. Analysis was performed on tightly gated livelymphocytes (Fig. 1). The dot-plots in Figure 1A illustrate
the presence of a CD4þCD25þFOXP3þ cell population.CD4þFOXP3þ cells were also stained for the T-reg markerCD127 (for which expression is low on T-reg [27,28]) andthe activation marker CD62L (which is shed upon T-cellactivation [29]) in order to confirm that these cells representthe T-reg population and not activated T cells (Fig. 1B). Asshown in Figure 1C, CD4þFOXP3þ cells display a typicalT-reg cell phenotype: CD25high CD62Lhigh and CD127low.
T-reg cell numbers are increased in CMLpatients with high BCR-ABL transcriptsPBMC were isolated from 39 CML patients and 10 healthydonors at different time points, resulting in 115 samplesanalyzable from patients and 25 from healthy donors.
Figure 1. Characterization of naturally occurring T-reg cells in CML. (A) PBMC from CML patients were stained with anti-CD4, anti-CD25, and anti-FOXP3
antibodies and their appropriate isotype controls and analyzed by flow cytometry. Analysis was performed on tightly gated lymphocytes and markers were set
using the isotype control staining (positive eventsO0.2% of negative control). These dot-plots illustrate thatO80% of CD4þFOXP3þ are also CD25þ. (B) Tofurther characterize these cells PBMC from CML patients were also stained for either CD62L or CD127. Examples of staining are shown where isotype control
staining is labeled as IC, and CD25, CD62L or CD127 staining is shown in black for CD4þFOXP3� cells and gray in CD4þFOXP3þ. CD4þFOXP3þ cells
were also CD25high, CD62Lhigh, and CD127low. (C) These stainings were performed in 15 cases and data are displayed as the average of the geometrical means
in the CD4þFOXP3þ and CD4þFOXP3� cell populations. Statistical significance was demonstrated using Student’s paired t test (***p ! 0.001).
1212 J.M. Rojas et al./ Experimental Hematology 2010;38:1209–1218
PBMC were stained for CD4, CD25, and FOXP3, and T-regcell percentage was measured on tightly gated lymphocytesand confirmed on gated CD4þ events, to exclude circu-
lating leukemic mononuclear cells. Patient details areshown in Table 1. No correlation was found betweenT-reg cell numbers and the duration of imatinib treatment,
1213J.M. Rojas et al./ Experimental Hematology 2010;38:1209–1218
BCR-ABL peptide vaccination, or white blood/lymphocyte/neutrophil cell counts. Figure 2 illustrates the levels ofCD4þCD25þ and CD4þFOXP3þ cells stratified accordingto their BCR-ABL transcript levels. All patients includedin the analysis had normal white blood cell counts. Inpatients not in complete cytogenetic remission equivalence(CCRe; BCR-ABL/ABL transcript ratio O1%), the mediannumber of CD4þCD25þ events in the lymphocyte popula-tion was 2.54%, whereas in cases in CCRe the mediannumber was 1.45% (p ! 0.01, two-tailed Mann-Whitneytest). Median number of CD4þCD25þ events was alsosignificantly higher in cases not in CCRe than in healthydonor cases (2.54% vs. 1.78%; p ! 0.01 two-tailed
Figure 2. Correlation between T-reg cell numbers and BCR-ABL transcript level
their concomitant BCR-ABL/ABL ratio. Cases where BCR-ABL/ABL ratios are a
ratios are !1% as CCRe. (Right panels) Cases with high levels of BCR-A
CD4þFOXP3þ events in the lymphocyte population (gate R1) than cases with
tailed Mann-Whitney test). (Left panels) These data were confirmed by analyz
the possibility of underestimating the T-reg population in the lymphocyte gate d
Mann-Whitney). Similarly, the median number ofCD4þFOXP3þ events in the lymphocyte population was1.62% in CCRe cases vs. 3.14% in cases not in CCReand 1.74% in healthy donor cases (p ! 0.01 for noCCRe vs. CCRe or healthy donor, two-tailed Mann-Whitney test). No differences in T-reg cell events weredetected between the healthy donor cases and the CCReCML cases. This correlation was confirmed with gatingperformed on CD4þ events instead of the whole lympho-cyte population. However, the correlation between T-regnumbers and transcript levels was not linear. Interestingly,in one case, the BCR-ABL/ABL ratio rose above 1% duringthe study (Fig. 3), and this was accompanied by an increase
s. The CD4þCD25þ and CD4þFOXP3þ events were grouped depending on
bove the equivalence to CCR (1%) are labeled as no CCRe and cases where
BL transcripts had a significantly higher number of CD4þCD25þ and
low levels of BCR-ABL transcripts or healthy donors (p ! 0.001, two-
ing the CD25þ and FOXP3þ events in the CD4þ population to rule out
ue to the presence of circulating leukemic mononuclear cells.
0
1
2
3
4
5
0 5 10Months
CD
4+
CD
25+
FO
XP
3+
0
2,5
5
7,5
10
BC
R-A
BL
:A
BL
CD4+CD25+FOXP3+ events BCR-ABL:ABL
Case 90
Figure 3. Relationship between CD4þCD25þFOXP3þ events and BCR-
ABL/ABL ratio in patients with changing levels of transcripts. During
the course of T-reg assessment the transcripts levels rose in case 90 leading
the later samples to be classified as no CCRe. Interestingly, the level of
CD4þCD25þFOXP3þ events detected also increased at the same time
points.
Figure 4. Proliferation to PPD after CD25 depletion. PBMC from CML
patients were enriched in CD4þ T cells and labeled with CFSE. Cells
were tested for proliferation to the recall antigen PPD either directly or
after depletion of CD25þ cells. As control, cells were culture in the
absence of PPD. (A) Shows a representative example of these experiments.
The proliferation to PPD was increased in the CD25-depleted cells when
compared to the whole CD4þ fraction. (B) Summarized data of these
proliferation experiments. Proliferation was normalized to the control
(unstimulated cells). In all six experiments, proliferation to PPD was
increased in the CD25-depleted population.
1214 J.M. Rojas et al./ Experimental Hematology 2010;38:1209–1218
in CD4þCD25þFOXP3þ events, consistent with a linkbetween the T-reg cell number and the BCR-ABL transcriptlevels.
T-reg cells limit T-cell responsesto recall antigens in CML casesThe effect of T-reg cells on the immune responses to recallantigens in CML was investigated using CD4-enrichedPBMC from patients either directly (CD4þ fraction) orafter CD25 depletion (CD4þCD25� cell fraction). TheCD4þ and CD4þCD25� fractions were labeled withCFSE and cultured with PPD as antigen or without antigenas control. Figure 4 shows a proliferative response to PPDin the CD4þ fraction. Depletion of CD25þ cells furtherenhances proliferation. Similar results were also seen inall five other cases studied; these data are summarized inFigure 4B; and suggest that CD4þCD25þ cells may limitthe proliferative response to PPD.
For functional studies of their inhibitory potential,CD4þCD25þ cells were expanded, using anti-CD3 anti-body and IL-2. As control, the CD4þCD25� cell populationwas similarly expanded. These expanded cell populationswere cocultured with autologous T cell lines specific forinfluenza viral peptides or the PADRE peptide. PADREwas linked to BCR-ABL peptides in our clinical trial ofBCR-ABL peptide vaccination known as EPIC (reportedin full in [26]). All patients in these functional experimentshad previously taken part in EPIC, and therefore PADRE isa recall antigen for them. Figure 5 shows representativeexamples of these coculture experiments. In the presenceof peptide, peptide-specific T-cell lines produced IFN-gand/or granzyme B; this cytokine production was abolishedon coculture with autologous CD4þCD25þ cells but not byCD4þCD25� cells. These data demonstrate thatCD4þCD25þ T cells can inhibit peptide-specific responsesin CML.
T-reg cells inhibit T-cell responsesto a HLA-A3-restricted BCR-ABL peptideT-cell lines were also generated to a HLA-A3�restrictedBCR-ABL peptide in three CML cases (Fig. 6). In all threecases, specific granzyme-B production was observed to theBCR-ABL peptide. Cytokine production was decreasedwhen autologous CD4þCD25þ cells were present in theculture but not in the presence of autologous CD4þCD25�
cells, suggesting that CD4þCD25þ cells can limitanti�BCR-ABL immune responses in CML.
Figure 5. Inhibition of responses to PADRE, HA307, and NP265 peptides by CD4þCD25þ cells. Effector cells specific for either peptide were cocultured with
autologous CD4þCD25þ cells (or CD4þCD25� cells as control). The specific production of IFN-g was inhibited in the presence of CD4þCD25þ cells. Data
presented are representative experiments.
1215J.M. Rojas et al./ Experimental Hematology 2010;38:1209–1218
DiscussionT-reg cells have been shown to limit anti-tumor immuneresponses in a variety of cancers. However, there havebeen few studies of their role in CML. We and othershave previously shown that CML cells can be targeted bycytotoxic T lymphocytes and that CML cells displaytumor-specific peptides on the cell surface [30]. In spite
of this, CML cells persist in patients. T-reg cells mayplay a role in limiting this existing immunity in CML insome patients. We therefore decided to investigate the pres-ence and role of naturally occurring T-reg cells in CML.
Naturally occurring T-reg cells have been defined asCD4þCD25þFOXP3þ. We found that this phenotype couldalso define naturally occurring T-reg cells in CML. Because
Effector Peptide CD4+
CD25+
CD4+
CD25-
Effector Peptide CD4+
CD25+
CD4+
CD25-
Effector Peptide CD4+
CD25+
CD4+
CD25-
Case 9
Case 15
Case 17
Granzyme-B spots for 50,000 cells
Granzyme-B spots for 50,000 cells
Granzyme-B spots for 50,000 cells
0 5 10 15 20 25 30
0 5 10 15 20 25 30
0 5 10 15 20 25 30 35 40
Figure 6. Inhibition of response to HLA-A3�restricted BCR-ABL peptide by CD4þCD25þ cells. Effector cells specific for the HLA-A3�restricted BCR-
ABL peptide were cocultured with autologous CD4þCD25þ cells (or CD4þCD25� cells as control). In all three cases where BCR-ABL peptide-specific cells
were detected, the specific production of granzyme-B was inhibited when CD4þCD25þ cells were present in the culture.
1216 J.M. Rojas et al./ Experimental Hematology 2010;38:1209–1218
CD25 and FOXP3 can also be expressed on some activatedT cells [31], we co-stained patient’s PBMC with CD127and CD62L to confirm that the populations used for anal-ysis were truly T-reg. The expression of CD127 has previ-ously been demonstrated to be low on T-reg [27,28].CD62L was used as an activation marker, as this moleculeis shed upon T-cell activation [29]. We confirmed that the
naturally occurring T-reg cells in CML can be defined asCD4þ CD25high FOXP3þ CD127low CD62Lhigh.
Serial patient samples were stained for CD4, CD25, andFOXP3 to estimate the percentages of T-reg cells in CMLpatients, and correlated with disease burden as assessedby real-time polymerase chain reaction. A correlationbetween the amount of T-reg cells present in the samples
1217J.M. Rojas et al./ Experimental Hematology 2010;38:1209–1218
and the BCR-ABL transcript level was found. T-reg cellnumbers are increased in cases not in CCRe whencompared to healthy subjects or CML cases in CCRe. Nocorrelation was found with other parameters like imatinibduration or BCR-ABL peptide vaccination. It remainsunclear whether this increase in T-reg cells suppressessome of the immune control of CML resulting in anincreased of the CML burden, or is only a consequenceto the expansion of CML cells. Nadal et al. have shownthat T-reg cells are increased in patients who relapse afterallogeneic stem cell transplant [22]. Taken together, it istherefore plausible that T-reg cells limit immune responsesin all CML cases.
We therefore decided to further investigate the func-tional role of T-reg cells in vitro. When CD4þ cells fromCML patients were depleted of CD25þ cells, we wereable to demonstrate an increase in proliferation to the recallantigen PPD, suggesting that T-reg cells are likely to limitimmune responses in CML. This observation wasconfirmed when we investigated the potential of isolatedT-reg cells for inhibiting peptide-specific responses. Theaddition of T-reg cells to T cell lines specific for influenzapeptides or PADRE peptide abolished the peptide-specificproduction of IFN-g and granzyme-B, indicating the impor-tance of T-reg cells in limiting immune responses. Impor-tantly, T-reg cells also inhibited the T-cell production ofgranzyme-B to a HLA-A3�restricted BCR-ABL peptide.These data suggest that T-reg cells can inhibit anti-CMLimmune responses. To the best of our knowledge, thisconstitutes the first evidence that T-reg cells can potentiallyinhibit anti-CML immune responses.
Imatinib and other tyrosine kinase inhibitors have beenreported to cause impairment of T-cell and dendritic cellfunction in vitro [32,33], and there is concern that thismay have clinical implications. However, in CML patients,imatinib may improve helper T-cell and dendritic cell func-tion toward normal [34,35]. Furthermore, imatinib may alsoimpair T-reg proliferation and also their expression of func-tional markers [36,37], suggesting that the overall net effectof imatinib on the immune system may be complex and ofuncertain clinical relevance. In a recent report, T-regnumbers were found to rise during imatinib but not dasati-nib therapy [38]. In the present study, we find no correlationbetween T-reg numbers and the duration of imatinib or totalTKI treatment, suggesting that T-reg activity is related moreto the level of disease burden rather than the cumulativeeffects of TKIs.
In conclusion, T-reg cells are readily detectable in CMLpatients and appear increased in patients with poorlycontrolled CML (as estimated by BCR-ABL transcriptlevels). We have also demonstrated that these T-reg cellsare involved in limiting immune responses in vitro. It istherefore plausible that naturally occurring T-reg cells areinvolved in limiting anti-tumor responses in CML. Aclinical trial of immunotherapy targeting CML-associated
antigens, whether peptide- or DNA-based, should thereforecarefully consider the potential inhibitory effects of T-regcells. These data also support the intriguing possibility ofclinically manipulating anti-CML immune response,although this is not without risk of provoking autoimmunereactions.
AcknowledgmentsThe study was supported by a clinical research grant from theLeukaemia Research Fund of Great Britain (now renamedLeukaemia and Lymphoma Research).
Conflict of Interest DisclosureNo financial interest/relationships with financial interest relatingto the topic of this article have been declared.
References1. Umiel T, Trainin N. Immunological enhancement of tumor growth by
