Altered Homeostasis of CD4?Memory T Cells in
Allogeneic Hematopoietic Stem Cell Transplant
Recipients: Chronic Graft-versus-Host Disease
Enhances T Cell Differentiation and Exhausts
Central Memory T Cell Pool
Akiko Fukunaga, Takayuki Ishikawa, Masako Kishihata, Takero Shindo, Toshiyuki Hori,
Department of Hematology and Oncology, Graduate School of Medicine, Kyoto University, Kyoto, Japan
Correspondence and reprint requests: Takayuki Ishikawa, MD, PhD, Department of Hematology and Oncology,
Graduate School of Medicine, Kyoto University, 54 Shogoin-kawaharacho, Sakyo-ku, Kyoto, Japan 606-8507
Received March 27, 2007; accepted June 21, 2007
An increased risk of late infection is a serious complication after allogeneic hematopoietic stem cell transplan-
tation (AHSCT), especially for recipients with defective CD4?T cell recovery. Although chronic graft-versus-
host disease (cGVHD) negatively influences CD4?T cell reconstitution, the mechanisms leading to this defect
are not well understood. We found that the proportion of CD27?CD4?T cells was remarkably increased in
ASHCT recipients with cGVHD or with repetitive infectious episodes. Isolated CD27?CD4?T cells from
ASHCT recipients had significantly shortened telomere length, displayed enhanced vulnerability to activation-
induced cell death, and showed extremely reduced clonal diversity, when compared with CD27?CD4?T cells
from healthy donors. Also, CD27?CD4?T cells from AHSCT recipients easily lost their expression of CD27
in response to antigen stimulation regardless of cGVHD status. Taken together, these data indicate that
homeostasis of memory CD4?T cells from AHSCT recipients is altered, and that they easily transit into
CD27?effector memory T cells. Increased in vivo T cell stimulation observed in recipients with cGVHD
further promotes the transition to effector memory cells, a change that decreases the central memory CD4?
T cell pool and consequently weakens the recipient’s defense against persistently infecting pathogens.
© 2007 American Society for Blood and Marrow Transplantation
T cell memory
T cell reconstitution
An increased risk of fatal infection remains the
major problem for long-term survivors of allogeneic
hematopoietic stem cell transplantation (AHSCT).
After the immediate posttransplant period, AHSCT
recipients experience long-lasting immune deficiency
. Several studies have demonstrated that quantita-
tive and qualitative recovery of CD4?T cells both
play pivotal roles in protecting against late infections
and in improving the long-term outcome of AHSCT
recipients [2-4]. T cell reconstitution can involve both
a thymus-dependent pathway and a thymus-indepen-
dent pathway. In the early posttransplant period, a
thymus-independent peripheral expansion of memory
T cells occurs . Because the reconstitution of the
CD4?T cell pool is more dependent on thymic func-
tion , CD8?T cells predominate over CD4?T
cells for several months after AHSCT . In the
setting of myeloablative autologous transplantation in
adults, it is reported that CD4?T cell regeneration
and thymic enlargement occur by 6 months, and the
number and diversity of naïve CD4?T cells normal-
izes 2 years after transplantation, at least in nonelderly
patients . After AHSCT, reconstitution of CD4?T
cells is influenced not only by recipient age, but also
Biology of Blood and Marrow Transplantation 13:1176-1184 (2007)
? 2007 American Society for Blood and Marrow Transplantation
by sources of stem cells and, most important, by graft-
versus-host disease (GVHD) [6,8-11].
Chronic GVHD (cGVHD) is a vexing late com-
plication that deeply influences the long-term out-
comes of AHSCT recipients. The disease develops in
about 50% of AHSCT survivors, and for many is
resolved within 2 years following AHSCT [12-15].
Unfortunately, some patients experience refractory
disease for an extended period, and their probability of
long-term survival is lowered by opportunistic infec-
tion [12,16,17]. The presence of active cGVHD neg-
atively influence the recovery of CD4?T cells ;
however, how cGVHD impairs the functional recov-
ery of CD4?T cells is not yet well understood.
To clarify this mechanism, we analyzed the char-
acteristics of peripheral blood CD4?T cells in pa-
tients who received AHSCT at least 1 year earlier. We
also aimed to establish reliable clinical indicators for
use in identifying AHSCT recipients for whom rein-
forced protection against late infection would be
PATIENTS, MATERIALS, AND METHODS
Patients and Samples
Blood samples were obtained from 40 AHSCT
recipients and 15 healthy donors after receiving their
written informed consents. AHSCT recipients had
undergone transplantation 1 to 10 years prior. All
patients were confirmed to be in complete donor chi-
merism and in complete remission. To exclude the
effects on T cell characteristics derived from any con-
current infectious episodes, blood samples from fe-
brile patients were not drawn. The clinical character-
istics of AHSCT recipients are summarized in
Table 1. The presence of cGVHD was defined as
follows: dependency on immunosuppressive agents in
patients who had received transplantation ?3 years
before, or the presence of active symptoms for which
immunosuppressive therapy was required in patients
receiving transplantations 1 to 3 years earlier. All
studies using these blood samples were approved by
the Medical Ethics Committee of Kyoto University.
Monoclonal Antibodies and Flow Cytometry
Mononuclear cells from the peripheral blood of
AHSCT recipients or healthy donors (HDs) were
separated by Ficoll-Paque Plus (Amersham Pharmacia
Biotech, Piscataway, NJ) density gradient sedimenta-
tion. For analysis of cell surface antigens, cells were
incubated with an appropriate concentration of fluores-
cein isothiocyanate (FITC)-, phycoerythrin (PE)-, PE-
cyanin-5 (PC5)-, or allophycocyanin (APC)-conjugated
monoclonal antibody (mAb) or the corresponding iso-
type-matched control mAb (eBiosience, San Diego, CA)
in the dark at 4°C for 20 minutes. They were then
washed twice and analyzed by FACScalibur (BD Bio-
sciences, San Jose, CA) with CellQuest software (BD
Biosciences). MAbs used in this study included FITC-
FITC-CD57, PE-CD27, PE-CCR7, FITC-CD62L,
APC- CD45RA (BD PharMingen, San Jose, CA), and
PC5-CD4 (Beckman Coulter, Fullerton, CA). FITC-
and PE-conjugated mAbs were purchased from eBio-
science. For intracellular cytokine staining, cells were
suspended in culture medium consisting of RPMI 1640
(Invitrogen, Carlsbad, CA), 10% volume fetal calf serum
then stimulated with immobilized anti-CD3 mAb
(OKT3, 10 ?g/mL) and 2 ?g/mL soluble anti-CD28
mAb (clone48, agonistic antibody established in our
laboratory; T. Hori, unpublished data) for 5 hours in
the presence of 10 ?g/mL brefeldin A (BFA; Sigma-
Aldrich, St. Louis, MO). After cell surface staining,
cells were fixed with 2% formaldehyde (Wako Pure
Chemical Industries, Osaka, Japan) diluted in phos-
phate-buffered saline (PBS), and permeabilized with
0.2% saponin (Sigma-Aldrich) containing buffer.
After washing with saponin-containing buffer, cells were
Table 1. Characteristics of Patients, Donors, and Transplantations
Sex of patients
Median age of patients (range)
Stem cell source
Chronic myeloproliferative disease
Lymphoma or myeloma
Interval after AHSCT
1 to 3 years
3 to 5 years
more than 5 years
MSD indicates matched sibling donor; GVHD, graft-versus-
host disease; AHSCT, allogeneic hematopoietic stem cell
*Values are expressed as number of patients otherwise indicated.
**Cord blood donor is excluded.
Chronic GVHD Exhausts Central Memory CD4 T Cells
(IFN-?), FITC-conjugated anti-interleukin-2 (IL-2),
or FITC-conjugated isotype-matched control mAb.
All mAbs used for cytokine staining were obtained
from BD PharMingen.
CD4?T Cell Separation and Cell Sorting
CD4?T cells were isolated by a MACS CD4 mul-
tisort kit (Miltenyi Biotec, Bergisch Gladbach, Ger-
many), using MACS separation MS columns (Miltenyi
Biotec) according to the manufacturer’s recommenda-
tions. They were then stained with PE-CD27 and APC-
CD45RA and sorted into naïve (CD27?/CD45RA?),
CD27?memory (CD27?/CD45RA?), and CD27?
memory cells using FACSAria (BD Biosciences) with
FACSDiVa 4.1 software (BD Biosciences). In some ex-
periments, CD4?T cells from AHSCT recipients were
stained with PE-CD27 only and sorted into CD27?and
CD27?cells. Sorted T cell subsets displayed a purity of
at least 98%.
T Cell Stimulation and Short-Term Culture
Isolated naïve, CD27?memory, and CD27?
memory T cells were suspended in a culture medium
at a concentration of 5 ? 105/mL, and 1 mL of cell
suspension was transferred into a 48-well plate coated
with anti-CD3 mAb. The concentration of anti-CD3
mAb was either high (10 ?g/mL) or low (0.1 ?g/mL).
Anti-CD28 mAb (2 ?g/mL) was also added to the
culture. T cell stimulation with anti-CD3 and anti-
CD28 mAbs (CD3/28 stimulation) was performed for
3 days; the cells were then harvested, washed twice,
and incubated with culture medium supplemented
with 150 IU/mL IL-2 (Peprotec, Rocky Hill, NJ) at a
concentration of 2 ? 105/mL for an additional 4 days.
In some experiments, after 7-day cultivation, the cells
were harvested, washed, prepared at a concentration
of 5 ? 105/mL in culture medium, and restimulated
overnight on an anti-CD3 mAb-coated plate (10?g/
mL). T cell death was determined by propidium io-
dide (PI, Sigma-Aldrich) staining in which a final
concentration of 1 ?g/mL of PI was added to each cell
sample before flow cytometry analysis.
Relative telomere length (RTL) measurements of
CD27?memory and CD27?memory CD4?T cells
were performed by a flow-FISH method using a Dako
Telomere peptide nucleic acid (PNA) kit/FITC
(Dako, Glostrup, Denmark), according to the manu-
facturer’s instructions. In brief, isolated cells and con-
trol cells (1301 cell line; Interlab Cell Line Collection,
Genoa, Italy) were mixed in equal measure in hybridiza-
tion solution with or without FITC-labeled telomere
PNA probe for 10 minutes at 82°C; hybridized over-
night in the dark at room temperature; washed twice
with a wash solution at 40°C; resuspended in PBS
containing 2% FCS and PI (1 ?g/mL); and subjected
to flow cytometric analysis. Telomere fluorescence of
at least 100,000 events was measured at a low flow rate
to ensure low coefficients of variation. The RTL value
was calculated using the following formula: RTL ?
([mean FL1 sample cells with probe ? mean FL1
sample cells without probe] ? DNA index of control
cells) ? ([mean FL1 control cells with probe ? mean
FL1 control cells without probe] ? DNA index of
the DNA index of 1301 cells is 2 and the DNA index of
sample cells in this study is 1.
T Cell Receptor Repertoire Analysis and
Complementarity-Determining Region 3
RNA isolated from naïve, CD27?memory, and
CD27?memory cells was converted to double-
stranded complementary DNA (cDNA), and T cell
receptor (TCR) beta chain variable region (VB) rep-
ertoires were analyzed with an adaptor ligation poly-
merase chain reaction (PCR)-based microplate hy-
bridization assay . After PCR amplifications using
primers specific for each TCRVB, the 5=-biotinylated
PCR products were hybridized in microplate wells on
which 38 TCRVB-specific probes were immobilized,
and followed with quantitative ELISA. PCR products
were also subjected to analysis for complementary-
determining region 3 (CDR3) size spectratyping ;
they were diluted in dye solution (95% formamide, 10
mM EDTA, and 0.1% blue dextrane) and analyzed in
6% denatured acrylamide gel with an ALFred se-
quence analyzer (Pharmacia Biotech, Uppsala, Swe-
den). The data obtained were transferred to Fragment
manager software (Pharmacia Biotech).
We used a 2-sided Mann-Whitney U test (non-
paired samples) and a paired t-test (paired samples) for
an analysis of differences between the groups. P values
?.05 were considered statistically significant.
The Proportion of CD27?CD4?T Cells Increased
in AHSCT Recipients, Especially in Those with
During comparative analysis of peripheral CD4?
T cells from AHSCT recipients and HDs, we found
that the proportion of CD27?cells among total
CD4?T cells was substantially higher in AHSCT
recipients who had been transplanted more than 1
year previously. As shown in Figure 1A, identified 3
major subsets of CD4?T cells by dual staining, using
CD45RA and CD27. The CD45RA?/CD27?frac-
tion represents naïve T cells, and the CD27?fraction
A. Fukunaga et al.
is known to include terminally differentiated cells .
The CD45RA?/CD27?fraction is composed of cen-
tral memory T cells . When data from 19 AHSCT
recipients (interval after AHSCT ranged from 1.5 to
10 years, median 4 years) and 15 HDs were compared,
it was clear that the percentage of naïve T cells was
low, and that of CD27?memory cells was substan-
tially high within the CD4?T cell subset from
AHSCT recipients (Figure 1B-D). To search for a
correlation between the percentage of CD27?cells
among total CD4?T cells and the clinical parameters,
samples from an additional 21 patients who had re-
ceived AHSCT more than 1 year before were col-
lected. Statistical analysis revealed 2 variables associ-
ated with the increased proportion of the CD27?
fraction: the presence of cGVHD and a poor general
condition, defined by a Karnofsky score of less than 90
(Figure 1E). An analysis of clinical charts demon-
strated that repetitive infectious episodes reduced the
general condition of most patients, and many of them
(but not all) also suffered from cGVHD (data not
shown). The frequency of CD27?cells among CD4?
T cells did not correlate with recipient age, source of
stem cells, donor type, underlying disease, condition-
ing regimen, history of acute GVHD (aGVHD), or
interval after transplantation (data not shown). Al-
though van Leeuwen reported that individuals who
(CMV) showed increased frequency of CD27?
CD28?cells within CD4?T cells , in our exper-
iments, no pair of AHSCT recipient and donor was
found to be serologically negative for CMV at the
time of SCT; thus, we could not evaluate the corre-
lation between persistent CMV infection and the pro-
portion of CD27?cells.
We then analyzed surface phenotype and cytokine
profiles of CD27?CD4?T cells from AHSCT recipi-
ents. Multicolor flow cytometry revealed that CD27?
CD4?T cells preferentially expressed HLA-DR and
CD57 and expressed very little CCR7 or CD62L (data
not shown). Cytokine profiles of CD27?CD4?T cells
were determined by sorting CD4?T cells into CD27?
and CD27?fractions. CD3/28 stimulation followed by
intracytoplasmic cytokine-staining revealed that CD27?
cells produced IFN-? much more frequently than
CD27?cells, and the proportion of IL-2-producing
cells was comparable (data not shown). The expression
of surface molecules and cytokine-producing capacity
showed no significant difference between CD27?
CD4?T cells from AHSCT recipients and HDs (data
not shown). These results are consistent with previous
studies and demonstrating that CD27?CD4?T cells
are differentiated effector cells [21,23].
Shortened Telomere Length and Enhanced
Vulnerability to Activation-Induced Cell Death of
CD27?CD4?T Cells from AHSCT Recipients
To better understand the differentiated status of
CD27?cells in AHSCT recipients, we examined and
compared the telomere length of sorted CD4?T cell
fractions derived from AHSCT recipients and HDs.
CD4?T cells from HDs were sorted into naïve,
CD27?memory, and CD27?memory cells as shown
in Figure 2A. In CD4?T cells from AHSCT recipi-
ents, naïve cells were not separated from CD27?
memory cells because of the limited total cell number
available and the scarcity of naïve cells. Figure 2B
shows representative dot plots for the measurements
of relative telomere lengths (RTLs) of sample cells,
using the flow-FISH method. The RTL of CD27?
cells was shorter when compared with those of
CD27?cells in AHSCT recipients and HDs (Figure
2C). Most important, CD27?cells from AHSCT re-
cipients showed significantly shorter RTLs compared
with those from HDs (Figure 2C). Because previous
Figure 1. Proportion of CD27?cells among total CD4?T cells in
AHSCT recipients and healthy donors. Peripheral lymphocytes
from AHSCT recipients and HDs were stained with CD4, CD27,
and CD45RA and gates were set on CD4-positive lymphocytes. A,
Representative dot plots of CD45RA and CD27 are shown. The
percentage of CD27?CD45RA?naïve, CD27?CD45RA?mem-
ory, and CD27?memory cells are shown in the upper right of each
plot. The percentage of naïve (B), CD27?memory (C), and CD27?
memory cells (D) among total CD4?T cells was compared between
16 AHSCT recipients and 15 HDs. E, Comparison of the percent-
age of CD27?memory cells among total CD4?T cells between
AHSCT recipients in the presence or absence of cGVHD (left), and
with or without good physical condition (right). Samples from 40
AHSCT recipients were included. The median, 25th-75th percen-
tiles, 10th-90th percentiles, and outliers are represented by the
line in the box, shaded box, error bars, and open circles, respec-
tively. A Mann-Whitney U test was used for statistical analysis.
Chronic GVHD Exhausts Central Memory CD4 T Cells
work demonstrated that telomere length shortens in
the course of T cell differentiation , these results
indicated that CD27?CD4?T cells from AHSCT
recipients undergo more cell divisions and are more
differentiated than those from HDs.
We next evaluated the vulnerability to activation-
induced cell death (AICD), because differentiated T
cells are known to be susceptible to apoptosis .
Sorted CD27?and CD27?CD4?T cells from 8
AHSCT recipients (4 of them complaining of
cGVHD) and CD27?and CD27?memory CD4?T
cells from 9 HDs were subjected to CD3/28 stimula-
tion, cultivation with IL-2, and restimulation with
anti-CD3 mAb . As expected in every patient and
healthy donor, CD27?CD4?T cells were more sus-
ceptible to IL-2-induced AICD after a second stimula-
tion than were CD27?CD4?T cells (Figure 3A and B).
In accordance with a shorter telomere length, CD27?
CD4?T cells from AHSCT recipients showed more
cell deaths than those from HDs. CD27?CD4?T
cells from AHSCT recipients showed comparable vul-
nerability to AICD whether from patients with or
without cGVHD (Figure 3B).
Substantially Reduced T Cell Repertoire Diversity
of CD27?CD4?T Cells from AHSCT Recipients
For effective immunity against various pathogens,
broad T cell repertoire diversity is essential . Re-
ported experiments using total lymphocytes or sorted
CD4?T cells have shown that skewing of T cell
repertoire after stem cell transplantation continues for
an extended period [28,29]. To better understand im-
munologic reconstitution after AHSCT, sorted naïve,
CD27?memory, and CD27?memory CD4?T cells
were individually subjected to microplate hybridiza-
tion assay to determine TCRVB repertoires, and per-
form CDR3 spectratyping. As shown in Figure 4, we
found that naïve cells and CD27?memory cells from
AHSCT recipients showed diverse CDR3 size spec-
tratypes with a Gaussian distribution of TCR frag-
ments, identical to those from HDs . In con-
trast, the TCRVB repertoire of CD27?memory
CD4?T cells was prominently skewed, and CDR3
size spectratypes were characterized by oligoclonal
expansions, which was much more prominent in
Figure 3. IL-2-induced AICD assay of CD4?T cells from AHSCT
recipients and HDs. Isolated CD4?T cells from AHSCT recipients
were sorted into CD27?and CD27?T cells. Memory CD4?T
cells from HDs were sorted into CD27?memory, and CD27?
memory cells. Sorted cells were stimulated with 0.1 ?g/mL plate-
bound anti-CD3 mAb and 2 ?g/mL soluble anti-CD28 mAb for 3
days. Cells were then cultured with IL-2 for another 4 days and
restimulated with 10 ?g/mL plate bound anti-CD3 mAb overnight.
The collected cells were stained with PI and the ratio of PI-positive
cells was analyzed by flow cytometry. A, Representative histogram
of PI staining. Values shown denote the percentage of nonviable
cells. B, The vulnerability to IL-2-induced AICD was compared
between CD27?and CD27?cells from 8 AHSCT recipients and 9
HDs (left). A comparison of the ratio of PI-positive cells among
CD27?T cells from AHSCT recipients with or without cGVHD is
shown (right). Statistical analysis was performed using a paired
t-test (*) and a Mann-Whitney U test (**).
Figure 2. Shortened telomere length in CD27?CD4?T cells.
Magnetically isolated, peripheral blood CD4?T cells from
AHSCT recipients were stained with CD27 and sorted into CD27?
and CD27?cells. CD4?T cells from HDs were stained with CD27
and CD45RA and sorted into CD27?memory and CD27?memory
T cells. A, Representative dot plots of HD-derived CD4?T cells
before and after cell sorting are shown. B, Sorted T cells were
individually mixed in equal measure with control cell line 1301,
hybridized with or without an FITC-labeled PNA-telomere probe,
counterstained with PI, and analyzed by FACScalibur. The percent
RTL (relative telomere length) was calculated as described in the
Patients, Materials, and Methods section. C, The percent of RTL in
CD27?and CD27?cells from 6 AHSCT recipients and in CD27?
memory and CD27?memory cells from 6 HDs were analyzed by
flow-FISH. Statistical analysis was performed using a paired t-test
(*) and Mann-Whitney U test (**).
A. Fukunaga et al.