Gene-expression profiles of peripheral blood mononuclear cell subpopulations in acute graft-vs-host disease following cord blood transplantation

Experimental Hematology 2008;36:1760–1770

Gene-expression profiles of peripheral blood mononuclear cell subpopulations in acute graft-vs-host disease following cord blood transplantation Naoyuki Takahashia, Noriharu Satob,c, Satoshi Takahashid, and Arinobu Tojod a Department of Advanced Medical Science; bDepartment of Applied Genomics, The Institute of Medical Science, The University of Tokyo, Tokyo, Japan; cSato Clinic of Internal Medicine, Tokyo, Japan; dDivision of Molecular Therapy, The Advanced Clinical Research Center, The Institute of Medical Science, The University of Tokyo, Tokyo, Japan

(Received 9 June 2008; revised 9 July 2008; accepted 25 July 2008)

Objective. Compared with allogeneic hematopoietic stem cell transplantation using other sources, cord blood (CB) transplantation (CBT) has clinical advantages in terms of incidence and severity of acute graft-vs-host disease (GVHD), despite using allogeneic stem cells with more human leukocyte antigen mismatches. However, detailed pathophysiology of acute GVHD developed after CBT has not yet been elucidated. In this study, we aimed to clarify the molecular mechanism of acute GVHD after CBT. Materials and Methods. We performed microarray expression profiling of immunoregulatory genes on each of four subpopulations (CD4+, CD8+, CD14+, and CD56+) of peripheral blood mononuclear cells (PBMCs), which were taken from eight patients with hematologic malignancies who suffered from acute GVHD after unrelated CBT. Results. We identified 55 genes, which were differentially expressed during acute GVHD compared to recovery phase. Among them, 22 showed differential expression concurrently in multiple PBMC subpopulations. In particular, five genes (TNFSF10/TRAIL, IL1RN, IFI27, GZMB, and CCR5) were upregulated and three genes (CLK1, TNFAIP3 and BTG1) were downregulated in at least three out of four subpopulations during acute GVHD. In addition, downregulation of antiinflammatory factors, such as TNFAIP3, KLF2, ZFP36, and BTG1, seems to be involved in acceleration of immune response, thus exacerbation of acute GVHD. Meanwhile, differential expression of several genes, such as CCL5, TNFAIP3, KLRB1/CD161, BY55/CD160, and PTGS2/COX2, was assumedly affected by the developmental immaturity of CB-derived cells. Conclusions. These results will contribute to the understanding of molecular mechanism underlying the behavior of inflammatory cells during acute GVHD following CBT. Ó 2008 ISEH - Society for Hematology and Stem Cells. Published by Elsevier Inc.

Acute graft-vs-host disease (GVHD) is a frequent serious complication after allogeneic hematopoietic stem cell transplantation and a major cause of posttransplantation mortality. Diverse cell populations, such as T cells, antigenpresenting cells, natural killer (NK) cells, regulatory T cells, NK T cells, and gd T cells, have been reported to play immunoregulatory roles in acute GVHD [1]. However, the molecular mechanism underlying the behavior of these Offprint requests to: Naoyuki Takahashi, M.D., Ph.D, Department of Advanced Medical Science, The Institute of Medical Science, The University of Tokyo, 4-6-1, Shirokanedai, Minato-ku, Tokyo, 108-8639, Japan; E-mail: [email protected]

inflammatory cell populations in acute GVHD is not fully determined. Recently, several analyses of gene-expression profiling of target organ tissue, including liver [2] and skin [3,4], in murine acute GVHD have been reported. These studies provided comprehensive information on molecular mechanism of acute GVHD. However, in these studies, geneexpression profiling was performed on the target organ tissue as a whole, which included diverse inflammatory cell populations as well as tissue cells. The recruited inflammatory cells will significantly differ in type and number, according to the time since transplantation as well as the intensity of allostimulation. Thus, changes in gene expression for the whole tissue samples are thought to be

0301-472X/08 $–see front matter. Copyright Ó 2008 ISEH - Society for Hematology and Stem Cells. Published by Elsevier Inc. doi: 10.1016/j.exphem.2008.07.007

N. Takahashi et al./ Experimental Hematology 2008;36:1760–1770

greatly affected not only by the change in gene expression for various tissue constituting cells, but also by the change in type and number of the recruited inflammatory cells. Therefore, it is difficult to infer from the results of these studies the roles of individual inflammatory cell populations in the pathophysiology of acute GVHD. In addition, it is yet unclear to what extent the knowledge about molecular and cellular mechanism of murine models can be applied to human acute GVHD, considering the differences in the immune system between human and mice. Compared with allogeneic hematopoietic stem cell transplantation using other sources, cord blood (CB) transplantation (CBT) has clinical advantages in terms of incidence and severity of acute and chronic GVHD despite using allogeneic stem cells with more human leukocyte antigen mismatches [5,6]. Although the blunted alloresponsiveness of CB-derived cells has been believed to be attributed to their naı¨ve character, detailed molecular mechanism, particularly in clinical setting, has not yet been fully elucidated. Here, we performed expression profiling of immunoregulatory genes on major subpopulations of peripheral blood mononuclear cells (PBMCs), namely, CD4þ, CD8þ, CD14þ, and CD56þ cells, to examine the molecular mechanism underlying the behavior of the inflammatory cells in human acute GVHD following CBT. Furthermore, we weighed differential gene expression among PBMC subpopulations to identify factors common to multiple subpopulations, which may play key roles in pathophysiology of acute GVHD.

Materials and methods Patients The study includes data from eight Japanese patients with hematologic malignancies who suffered from acute GVHD (grade II: n 5 7, grade I: n 5 1) after unrelated CBT between 2004 and 2006 at the Institute of Medical Science, the University of Tokyo. Acute GVHD was graded according to the criteria of Glucksberg et al. [7]. Five patients (IMS8, 11, 13, 18, 25) received total body irradiation (TBI) (12 Gy), cytosine arabinoside (12 g/m2) with recombinant human granulocyte colony-stimulating factor (G-CSF) (5 mg/ kg/day), and cyclophosphamide (120 mg/kg). Two patients (IMS1 and IMS5) received TBI (12 Gy), cytosine arabinoside (24 g/m2) with G-CSF (5 mg/kg/day), and fludarabine (120 mg/m2). One patient (IMS13) received TBI (12 Gy), fludarabine (120 mg/m2), and melphalan (140 mg/m2). As prophylaxis against GVHD, six patients (IMS1, 5, 8, 11, 18, and 25) received standard protocol with cyclosporine administered daily at 3 mg/kg from day 1 and methotrexate administered at 15 mg/m2 on day 1, followed by 10 mg/m2 on days 3 and 6. Two patients (IMS13 and 17) received standard protocol with cyclosporine administered daily at 3 mg/kg from day 1 and methotrexate administered at 15 mg/m2 on day 1, followed by 10 mg/m2 on day 3. Detailed clinical and laboratory data are summarized in Table 1. The study was approved by the institutional review board and written informed

1761

consents were obtained from all patients in accordance with the Declaration of Helsinki. Microarray In this study, we used two versions of custom-made oligonucleotide DNA microarrays Genopal (Mitsubishi Rayon, Tokyo, Japan) with probes for immunoregulatory genes. Detailed description of these microarrays has been deposited in Gene Expression Omnibus (GEO, http://www.ncbi.nlm.nih.gov/geo/) with accession numbers, GPL6487 for IMShuman0.6Kv.1 (hereinafter designated as ‘‘Array v.1’’) with probes for 594 genes, and GPL6488 for IMShuman0.6Kv.2 (hereinafter designated as ‘‘Array v.2’’) with probes for 21 genes in addition to the 594 genes. Blood from three patients (IMS1, 5, and 8) and five patients (IMS11, 13, 17, 18, and 25) were analyzed using Array v.1 and Array v.2, respectively. Detailed description of experimental protocols and data processing has been deposited in GEO database with accession number GSE10572. Briefly, 10 mL peripheral blood was taken twicedat the acme of macroscopic cutaneous abnormality [designated as ‘‘acute GVHD phase’’ and abbreviated as aGVHD(þ)] and at the time point when macroscopic cutaneous abnormality resolved [designated as ‘‘recovery phase‘‘ and abbreviated as aGVHD()]. Promptly after blood collection, CD8þ, CD56þ, CD4þ, and CD14þ cells were isolated from PBMCs in this order by employing CD8 MicroBeads, CD56 MicroBeads, CD4 MicroBeads, and CD14 MicroBeads (Miltenyi Biotec GmbH, Bergisch Gladbach, Germany), respectively, and then lyzed immediately in TRIZOL reagent (Invitrogen, Carlsbad, CA, USA). However, only CD8þ and CD56þ populations were isolated from patient IMS1. Purified total RNA was reverse-transcribed using oligo (dT) primer with T7 promoter, and transcribed in vitro to synthesize biotin-modified antisense RNA. For some specimens that contained too few cells to obtain sufficient amount of antisense RNA by single round of amplification, two rounds of amplification were performed. Microscopically, monocytes contained in CD4þ subpopulation were !2%. Data processing To identify differentially expressed genes in acute GVHD phase vs recovery phase, the raw signal intensity data were processed as follows: To obtain gene-specific signal intensity data, the raw value for l A (negative control) was subtracted. Secondly, paired gene-specific data [aGVHD(þ) vs aGVHD()] were discarded, if both values were lower than a threshold of 100 counts/second. If only one of the paired gene-specific values was !100 counts/second, it was replaced by 100 counts/second. Thirdly, these filtered gene-specific data were normalized, using lowess function (‘‘stats’’ package) in R language [8], with the lowess smoothing parameter f set to 0.4. Finally, significance analysis of microarray (SAM) [9] was carried out to evaluate differential gene expression during acute GVHD compared to recovery phase, for the five patients (IMS11, 13, 17, 18 and 25), using ‘‘samr’’ package in R language, with false discovery rate !10%. For conclusive confirmation of the statistical results for the five patients, SAM was carried out supplementarily, for the common 594 genes shared by Array v.1 and Array v.2, using the data from all the eight patients. To exclude genes with relatively low expression or small change, we selected genes according to the following criteria. 1)

N. Takahashi et al./ Experimental Hematology 2008;36:1760–1770

1762

Table 1. Clinical characteristics and laboratory data Acute GVHD Stage Case (age/gender)

Diagnosis

Proportion (%)

Grade Skin Gut Liver

IMS1 (35/M)

AML (M2)

II

3

1

0

IMS5 (43/F)

AML (M3)

II

3

1

0

IMS8 (53/F)

AML (M2)

II

2

1

0

IMS11 (50/F)

AML (M2)

II

3

0

0

IMS13 (49/M)

AML (M3)

II

3

0

0

IMS17 (24/F)

NHL (LBL) I

2

0

0

IMS18 (44/M)

MDS/AML

II

3

0

0

IMS25 (38/M)

ALL (L2)

II

3

0

0

Phase

Days after CBT CD4 CD8 CD14 CD56

aGVHD(þ) aGVHD() aGVHD(þ) aGVHD() aGVHD(þ) aGVHD() aGVHD(þ) aGVHD() aGVHD(þ)

48 194 26 132 71 111 34 119 41

NA NA NA NA NA NA 7.5 5.5 74.6 10.9 41.8 29.1 11.4 11.9 54.5 13.0 22.6 49.6 10.8 15.8 48.9 5.7 71.4 10.6 7.4 5.4 73.6

NA NA 12.4 18.2 22.3 14.8 24.5 12.2 13.7

aGVHD()

147

20.1

40.3

13.4

26.2

aGVHD(þ) aGVHD() aGVHD(þ) aGVHD()

30 86 28 105

17.8 16.2 6.7 15.5

37.4 38.1 12.6 4.9

32.1 26.7 71.4 70.5

12.7 19.0 9.2 9.0

aGVHD(þ) aGVHD()

28 124

3.7 16.5

12.2 38.5

76.5 30.8

7.6 14.3

Meda

Array

CyA 120 mg (IV) None CyA 70 mg (IV) CyA 30 mg (PO) CyA 30 mg (IV) CyA 30 mg (PO) CyA 70 mg (IV) CyA 50 mg (PO) CyA 50 mg (IV), PSL 15 mg (PO) CyA 10 mg (PO), PSL 10 mg (PO) CyA 120 mg (IV) None CyA 90 mg (IV) CyA 40 mg (PO), PSL 17.5 mg (PO) CyA 50 mg (IV) None

v.1 v.1 v.1 v.2 v.2

v.2 v.2

v.2

Percentage of each subpopulation is calculated regarding the sum of the cell numbers of four subpopulations as total. And the significant difference in percentage of subpopulations among samples justifies examining each subpopulation individually instead of entire peripheral blood mononuclear cells altogether. ALL 5 acute lymphoblastic leukemia; AML 5 acute myelogenous leukemia; CyA 5 cyclosporine; IV 5 intravenously; LBL 5 lymphoblastic lymphoma; MDS/AML 5 secondary AML developed from myelodysplastic syndrome; NA 5 not applicable; NHL 5 non-Hodgkin’s lymphoma; PO 5 orally; PSL 5 prednisolone. a Daily dosage of immunosuppressants receiving at the time of sampling.

Average of maximum values of signal data pairs [aGVHD(þ) vs aGVHD()] is O1500 counts/second; 2) average lowess-normalized logarithmic ratio is O0.5 or O 0.5. Quantitative reverse transcriptase polymerase chain reaction To validate microarray data, quantitative real time reverse transcriptase (RT)-polymerase chain reaction (PCR) analysis was performed for 10 genes for which validated TaqMan Gene Expression Assays are available (Table 2). Because the remaining complementary DNA, which had been obtained by reverse-transcription of total RNA with oligo (dT) primer bearing a T7 promoter, was

used as a template for PCR, TaqMan probes and primers were selected so that their target regions lied within 1.7 kb from 30 -ends.

Results Microarray data are shown in Figure 1. A total of 55 genes, which included many interferon (IFN)-induced genes, were differentially expressed during acute GVHD compared to recovery phase. The number of differentially expressed genes in CD4þ subpopulation was the largest among the

Table 2. TaqMan primers and probes used in the study Gene name

Accession number

BTG1 BY55/CD160 CCL5/RANTES GZMB KLF2 KLF4 KLRB1/CD161 LTB NPM1 TNFSF10/TRAIL ACTB GAPDH

NM_001731.1 NM_007053.2 NM_002985.2 NM_004131.3 NM_016270.2 NM_004235.3 NM_002258.2 NM_009588.1 NM_199185.2 NM_003810.2 NM_001101.2 NM_002046.3

a

Catalogue number

Distance from 30 a

TaqMan Gene Expression Assay: Hs00982890_m1 TaqMan Gene Expression Assay: Hs00199894_m1 TaqMan Gene Expression Assay: Hs00174575_m1 TaqMan Gene Expression Assay: Hs00188051_m1 TaqMan Gene Expression Assay: Hs00360439_g1 TaqMan Gene Expression Assay: Hs00358837_g1 TaqMan Gene Expression Assay: Hs00174469_m1 TaqMan Gene Expression Assay: Hs00242739_m1 TaqMan Gene Expression Assay: Hs02339479_g1 TaqMan Gene Expression Assay: Hs00234356_m1 Predeveloped TaqMan Assay Reagent: #4326315E Predeveloped TaqMan Assay Reagent: #4326317E

1331 801 979 288 1489 1234 591 605 777 1259 1753 1152

Distance from 30 is calculated by subtracting ‘‘Assay Location’’ (distance from 50 end) from full length of mRNA, according to manufacturer’s information.

N. Takahashi et al./ Experimental Hematology 2008;36:1760–1770

1763

Figure 1. Expression profile of genes that showed significant change during aGVHD compared to recovery phase. Heat map shows logarithmic ratio of gene expression [aGVHD(þ) vs aGVHD()] for each patient, using Array v.2 (for IMS11, 13, 17, 18, and 25) or Array v.1 (for IMS1, 5, and 8). Data using Array v.1 (IMS1, 5, and 8) are presented only if the change in gene expression was judged to be significant by SAM for all the eight patients. In ‘‘Gene name’’ column, genes upregulated in multiple subpopulations are in red and genes downregulated in multiple subpopulations are in green, while genes whose differential expression was confined to one subpopulation are in black. In addition, genes whose expression was validated by real-time PCR are underlined. Genes included in Array v.2 but not in Array v.1 are accompanied with ‘‘*.’’ SAM results for the five patients (IMS11, 13, 17, 18, and 25) are presented, as well as those for all the eight patients, with means and standard deviations of lowess-normalized logarithmic ratios (Mfit) and q values.‘‘d’’ indicates that the gene did not show significant change by SAM for all the eight patients. yLowess-normalized logarithmic ratio of gene expression [aGVHD(þ) vs aGVHD()].zStandard deviation. xLowest false discovery rate (%).

1764

N. Takahashi et al./ Experimental Hematology 2008;36:1760–1770

Figure 1. (continued).

four subpopulations, which seems to reflect the indispensable role of helper T cells in pathophysiology of acute GVHD [1], although expression of selected genes were

examined with this array. It was confirmed by quantitative real time RT-PCR that microarray data reflected gene expression properly (Fig. 2).

N. Takahashi et al./ Experimental Hematology 2008;36:1760–1770

1765

Figure 2. Validation of microarray data by real-time PCR. Real-time PCR was performed in duplicate for all samples, on an ABI PRISM 7700 Sequence Detection system (Applied Biosystems, Foster City, CA, USA), using TaqMan Universal PCR Master Mix (Applied Biosystems). The thermal cycler conditions were as follows: heating at 95 C for 10 minutes, followed by 50 cycles of denaturing at 95 C for 15 seconds and annealing and extension at 60 C for 1 minute. Signal intensities were normalized using two kinds of endogenous controls [ACTB (left) or GAPDH (right)]. The logarithmic ratios of signal intensities (aGVHD(þ) vs aGVHD ()] obtained by real-time PCR (DDCt) and by microarray are represented as a pair of heat maps. The mean and standard deviation (SD) of logarithmic ratio for each gene are indicated on the right side of each heat map. Genes upregulated in multiple subpopulations are in red and genes downregulated in multiple subpopulations are in green, while genes whose differential expression was confined to one subpopulation are in black, as in Figure 1. *DDCt was obtained according to ‘‘Comparative CT method’’ recommended by the manufacturer. yLogarithmic ratio of gene expression [aGVHD(þ) vs aGVHD()]. zStandard deviation.

Of 55 genes, 22 showed significant change in expression, which, surprisingly, was common to multiple PBMC subpopulations in terms of direction (Fig. 3). Among the 22 genes, TNFSF10/TRAIL, which regulates cell death, immune response, and inflammation, showed significantly higher expression during acute GVHD compared to recovery phase in all the four subpopulations, which was confirmed by real-time PCR (Fig. 2). Contrastingly, CLK1, which regulates splicing, showed significantly lower expression in four subpopulations. In three of four subpopulations, two anti-proliferative and/or anti-inflammatory genes (TNFAIP3 and BTG1) showed significantly lower expression during acute GVHD compared to recovery phase. TNFAIP3 is a zinc finger protein that is critical for limiting inflammation by terminating tumor necrosis factor (TNF)-ainduced nuclear factor kB responses [10], while BTG1, which inhibits proliferation with G1 arrest, promotes apoptosis [11] and differentiation in multiple cell types [12]. Meanwhile, four genes (IFI27, IL1RN, CCR5, and GZMB) showed higher expression during acute GVHD in three subpopulations. IL1RN is another anti-inflammatory protein, which inhibits activity of interleukin (IL) 1, one of key mediators of acute GVHD [13]. GZMB, a serine protease that is packaged with perforin in cytotoxic granules of NK and CD8þ T cells, showed upregulated expression during acute GVHD not only in cytolytic lymphocytes (CD8þ and CD56þ cells) but also in CD4þ cells with comparable signal intensities, which was confirmed by real-time PCR (Fig. 2).

In two of four subpopulations, three genes (KLF2, MICA, and ITGB1) showed lower expression during acute GVHD. Among them, KLF2, a member of Kruppel-like family transcription factors, which has been reported to be a quiescence factor for T lymphocytes [14,15] and monocytes [16], was downregulated in CD4þ and CD14þ subpopulations, which was confirmed by real-time PCR (Fig. 2). Meanwhile, 11 genes (IFIT1, IFIT2, IFI16, IFI30, IFI35, CCR1, IL10RB, IL18RAP, CASP1, HSP105B, and ADPRTL1) showed higher expression during acute GVHD in two subpopulations. IFI30, an IFN-ginducible lysosomal thiol reductase that plays an inhibitory role in T-cell activation [17], was upregulated in CD4þ and CD8þ cells. Among chemokine-related genes, CCR1 was upregulated in CD4þ and CD8þ subpopulations and CCR5 was upregulated in CD4þ , CD8þ, and CD56þ subpopulations, as is the case with bone marrow or peripheral blood stem cell transplantation [18,19]. Contrastingly, CCL5/RANTES, a ligand of CCR1 as well as CCR5, was downregulated in CD4þ subpopulation with the second lowest logarithmic ratio, which was confirmed by real-time PCR (Fig. 2). Among the remaining 33 genes whose differential expression was confined to one subpopulations, zinc finger proteins, such as KLF4 and ZFP36/tristetraprolin, showed downregulated expression in CD14þ subpopulation during acute GVHD. KLRB1/CD161, a gene encoding NK cell receptor, was downregulated in CD4þ subpopulation with the lowest

1766

N. Takahashi et al./ Experimental Hematology 2008;36:1760–1770

Figure 3. Concurrent differential expression in multiple PBMCs subpopulations [aGVHD(þ) vs aGVHD()]. Heat map shows the maximum values of ACTB-normalized logarithmic intensity pairs [aGVHD(þ) vs aGVHD()] averaged over the five patients (IMS11, 13, 17, 18, and 25) in each PBMC subpopulation. A blank cell signifies that average could not be obtained due to threshold processing. To facilitate visualization of change in expression common to multiple PBMC subpopulations, the upregulation and downregulation of genes are indicated by arrows, according to the microarray result in Figure 1. As in Figure 1, genes upregulated in multiple subpopulations are in red and genes downregulated in multiple subpopulations are in green, while genes whose differential expression was confined to one subpopulation are in black. In addition, genes whose expression was validated by real-time PCR are underlined. Genes included in Array v.2 but not in Array v.1 are accompanied with ‘‘*’’. yMaximum value of ACTB-normalized logarithmic intensity pairs [aGVHD(þ) vs aGVHD()] averaged over the five patients (IMS11, 13, 17, 18, and 25).

logarithmic ratio, while its expression did not show significant change in CD8þ and CD56þ cells. BY55/CD160, another NK cell receptor, was downregulated in CD8þ subpopulation with the lowest logarithmic ratio, while its expression did not show significant change in CD56þ cells. PTGS2/COX2, a key enzyme that is responsible for the prostanoid biosynthesis in inflammation and mitogenesis, was downregulated in CD14þ subpopulation with the lowest logarithmic ratio.

Discussion So far, several gene-expression analyses of murine target organ tissue during acute GVHD have been reported

[2–4]. Ichiba et al. [2] analyzed gene expression during development of acute hepatic GVHD, using murine model of allogeneic bone marrow transplantation (BMT). They showed that genes including Stat1, Ifit1, Ifi16/Ifi204, and Ccl5 were upregulated on day 7 after BMT, before development of histological change in the liver, while Il1rn was upregulated on day 35, when hepatic GVHD was histologically evident [2]. Sugerman et al. [3] analyzed gene expression during development of acute cutaneous GVHD, using murine model of allogeneic BMT, and showed upregulated expression of genes including Ccl5, Ccr1, Ccr5, Gzmb, Ltb, Casp1, Ifngr, Ifi16/Ifi204, Ifi30, Ifit1, Stat1, and Il1rn during GVHD [3]. Zhou et al. [4] also analyzed gene-expression profiles during development

N. Takahashi et al./ Experimental Hematology 2008;36:1760–1770

of cutaneous GVHD, using murine model of allogeneic BMT, and showed upregulated expression of genes including Ccl5, Ccr1, Ccr5, and Sell in acute phase (days 7, 30, and 60) [4]. In these three studies, gene-expression profiling was performed on whole tissue samples, without separating inflammatory cells according to types. Therefore, it is difficult to infer from the results of these three studies the roles of individual inflammatory cell populations in the pathophysiology of acute GVHD. Although cautiousness is required in this sense when comparing the results, the upregulation, during acute GVHD, of IFN-induced genes (IFNGR, IFI16, IFI30, IFIT1, and STAT1), chemokine receptor genes (CCR1 and CCR5), an acute phase protein gene (IL1RN), apoptosis regulator genes (GZMB, LTB, and CASP1), and an adhesion molecule gene (SELL), which was reported in the three studies mentioned previously [2–4], is apparently consistent with our data. All of these three studies showed upregulated expression of CCL5 in murine liver or skin during acute GVHD. In our study, however, CCL5 showed downregulated expression with the second lowest logarithmic ratio in CD4þ PBMC subpopulation, which was confirmed by real-time PCR. Because CCL5 is reported to be induced in vitro synergistically by IFN-g and TNF-a [20], which are key mediators of acute GVHD [13], downregulation of CCL5 during acute GVHD is a novel result. Compared to adult blood cells, expression of CCL5 by CB-derived CD4þ T cells is reported to be greatly reduced both at basal state and upon CD3mediated stimulation [21], which presumably explains the repressed CCL5 expression during acute GVHD following CBT. CCL5-deficient mice showed defects in T-cell function, such as polyclonal and antigen-specific proliferation, as well as IFN-g and IL-2 production [22]. Therefore, repressed expression of CCL5 during acute GVHD might be linked to the low incidence and severity of acute GVHD following CBT. TNFAIP3, an anti-inflammatory protein which terminates TNF-ainduced nuclear factorkB responses, was downregulated in CD4þ, CD8þ, and CD14þ subpopulations, although TNFAIP3 has been reported to be induced in vitro by proinflammatory stimuli, such as TNF-a [10], one of key mediators of acute GVHD [13]. Compared to adult monocytes, CB-derived monocytes have been shown to express significantly less TNFAIP3 upon lipopolysaccharide stimulation [23], which seems to be in accord with the downregulation of TNFAIP3 during acute GVHD. KLRB1/CD161, a gene-encoding NK cell receptor, showed the lowest logarithmic ratio in CD4þ subpopulation with signal intensity comparable to those of CD8þ and CD56þ cells. It has been shown that CD161 is expressed on around 20% of human CD4þ T cells, and human CD4þCD161þ T cells are mostly of central or effector memory phenotype [24]. Therefore, repressed expression of CD161 in CD4þ subpopulation during acute GVHD

1767

seems to be attributed to the naı¨ve phenotype of CB-derived cells early after transplantation, namely during acute GVHD phase, compared to recovery phase. Unexpectedly, BY55/CD160, another NK cell receptor, was downregulated in CD8þ subpopulation with the lowest logarithmic ratio, although its expression on CD8þ T cells has been reported to be increased upon CD3-mediated stimulation in vitro [25]. Because CD160 expression on CD8þ T cells is exclusively associated with cytolytic effector activity [26], our result presumably reflects low cytolytic effector activity of CB-derived CD8þ T cells [27] early after transplantation (during acute GVHD phase) compared to recovery phase. Downregulation of PTGS2/COX2 in CD14þ subpopulation is also an unexpected result, as PTGS2 has been shown to be induced in many kinds of cells, including monocytes, by inflammatory stimuli in vitro [28]. Expression of PTGS2 is significantly less in CB-derived monocytes compared to adult monocytes [23], which presumably is in accord with downregulation of PTGS2 early after transplantation (during acute GVHD phase). Among anti-inflammatory factors, IL1RN, and IFN-induced gene IFI30 were upregulated, while TNFAIP3, KLF2, ZFP36, and BTG1 were downregulated during acute GVHD. Upregulation of IL1RN and IFI30 has been shown in previous reports [2,3], and presumably constitutes a component of negative feedback loop against proinflammatory stimulation. On the other hand, downregulation of TNFAIP3, KLF2, ZFP36, and BTG1, has not yet been reported, and their downregulations seems to be involved in acceleration of immune response, thus exacerbation of acute GVHD. If gene-expression profiling is performed on whole tissue sample, expressional change of a gene, which was upregulated in recruited inflammatory cells, tends to be overestimated, as the control tissue sample contains significantly less number of inflammatory cells of corresponding types. In contrast, expressional change of a gene, which was downregulated in recruited inflammatory cells, tends to be underestimated, as the decrease in gene expression will be counterbalanced by the increase in cell number. This seems to be one of the possible reasons why downregulation of these anti-inflammatory factors has not been identified by the microarray analyses performed so far on whole target organ tissue of murine acute GVHD models [2–4]. KLF2, a quiescence factor for T lymphocytes [14,15] and monocytes [16], showed repressed expression in CD4þ and CD14þ subpopulations. In endothelial cells, KLF2 is also a regulator of activation in response to proinflammatory stimuli [29] and statin exerts anti-inflammatory effects by upregulating KLF2 [30]. Recently, it has been shown that statin prevented acute GVHD by T-helper type 2 polarization [31]. Statin may prevent acute GVHD also by upregulating KLF2 in immune cells as is the case with endothelial cells, although this awaits further study.

1768

N. Takahashi et al./ Experimental Hematology 2008;36:1760–1770

ZFP36, a zinc finger protein that interferes with TNFa production by destabilizing its messenger RNA, was downregulated in CD14þ subpopulation, although ZFP36 has been shown to be induced in macrophages by proinflammatory stimuli, such as TNF-a [32]. Targeted disruption of ZFP36 in mice resulted in diffuse systemic inflammation, including inflammatory arthritis, dermatitis, and immunologic abnormalities [33]. In PBMCs from patients with rheumatoid arthritis (RA), one of common chronic systemic inflammatory diseases, TNF-a was overexpressed and ZFP36 was underexpressed with significantly lower ratio of ZFP36 to TNF-a, compared with healthy control cells [34], suggesting causal relationship between ZFP36 and dysregulated expression of TNF-a. Furthermore, PBMC from RA patients showed shorter duration of upregulation of ZFP36 in response to lipopolysaccharide compared with healthy control cells [35]. Hence, downregulation of ZFP36 in CD14þ cells during acute GVHD may reflect similar defects in ZFP36 production upon inflammatory stimuli, as in RA. Unexpectedly, KLF4, a zinc finger transcription factor that plays an essential role in myelopoiesis and monocyte differentiation [36], was downregulated in CD14þ subpopulation, which was confirmed by real-time PCR, although KLF4 has been shown to be upregulated in macrophages in vitro by proinflammatory cytokines, such as IFN-g and TNF-a [37]. TRAIL showed upregulated expression in all the four PBMC subpopulations during acute GVHD, which seems to be consistent with the in vitro result that TRAIL is induced in these subpopulations upon stimulation [38,39], as well as previous clinical results [40]. It has been shown that TRAIL directly inhibits activation of human T cells by blocking calcium influx, without inducing T-cell anergy or apoptosis [41]. Hence, upregulation of TRAIL in multiple subpopulations may also constitute a component of negative feedback loop against proinflammatory stimulation. GZMB showed upregulated expression during acute GVHD not only in cytolytic lymphocytes (CD8þ and CD56þ subpopulations), but also in CD4þ cells. Although its expression has long been believed to be limited to CD8þ and NK cells, it was recently shown that CD3/CD46 activation, converted naı¨ve CD4þ T cells into IL-10 producing regulatory CD4þ T cells, which expressed GZMB and perforin and had cytotoxic activity for allogeneic as well as autologous target cells in a perforin-dependent manner [42–44]. In addition, virus-stimulated plasmacytoid dendritic cells was shown to induce allogeneic naı¨ve CD4þ T cells to differentiate into IFN-g/IL-10producing cytotoxic regulatory T cells, which also expressed GZMB and perforin [45]. In vivo, IFN-g/IL-10producing CD4þ T cells have been isolated from patients with chronic infectious disease, such as tuberculosis [46] and hepatitis C [47]. One may think, therefore, upregulation of GZMB in CD4þ subpopulation would somehow reflect the increase

of those cytotoxic regulatory CD4þ T cells during acute GVHD, although further study is required. Purity of the PBMC subpopulations sorted by MicroBeads is inferred to be ranged from 85% to 90%, based on the results of preliminary single-color fluorescence-activated cell-sorting analysis of normal volunteers’ PBMCs, using fluorescein isothiocyanateconjugated antibodies against the corresponding cell surface markers (data not shown). To assess the influence of the cells that were mixed in through nonspecific binding to MicroBeads, on gene expression of the PBMC subpopulations, we reviewed the signal intensity data of the 55 differentially expressed genes, after conducting normalization using ACTB and GAPDH as controls. If the mixed-in cells should dominate gene expression of the PBMC subpopulations isolated by MicroBeads, the normalized signal intensities would be quite similar among the four PBMC subpopulations. However, as shown in Supplementary Figure 1, the normalized signal intensities tended, in general, to show gene-specific patterns over the four subpopulations, and these patterns tended, in general, to be similar among the five patients, in both acute GVHD phase and recovery phase, which seems to exclude the possibility of dominant influence of the mixed-in cells on gene expression of the PBMC subpopulations. Furthermore, there was a significant difference among the number of differentially expressed genes of PBMC subpopulations (41 genes in CD4þ cells, 24 genes in CD8þ cells, 13 genes in CD14þ cells, and 9 genes in CD56þ cells). Therefore, the probability that the concurrent expressional change of the 22 genes in multiple PBMC subpopulations would derive from those mixed-in cells, seems to be low. In this study, we examined gene-expression profiles of individual PBMC subpopulations, precursors of inflammatory cells recruited to and retained in the target organs. Further study is necessary to determine to what extent the expression profiles of circulating cells represent those of inflammatory cells infiltrating into target organs in situ. In addition, the CD4þ, CD8þ, CD14þ, and CD56þ PBMCs are not necessarily homogeneous populations. Because of the limitation of the number of cells that can be obtained clinically, further study with other methodology for analyzing a minute amount of RNA, such as real-time PCR, would be necessary to understand in detail the roles of moresubdivided cell populations in acute GVHD. To summarize, we identified 22 genes that showed differential expression concurrently in multiple PBMC subpopulations, by profiling the immunoregulatory genes in PBMC subpopulations. Among them, five genes (TRAIL, IL1RN, IFI27, GZMB, and CCR5) were upregulated and three genes (CLK1, TNFAIP3 and BTG1) were downregulated in at least three of four subpopulations during acute GVHD. These eight genes seem to be candidates which may play key roles common to multiple subpopulations in the pathophysiology of acute GVHD. Moreover,

N. Takahashi et al./ Experimental Hematology 2008;36:1760–1770

downregulation of anti-inflammatory factors, such as TNFAIP3, KLF2, ZFP36, and BTG1, seems to be involved in acceleration of immune response, thus exacerbation of acute GVHD. Further study is required to clarify whether therapeutic augmentation of these factors may ameliorate acute GVHD. In addition, differential expression of CCL5 as well as TNFAIP3, CD161, CD160, and PTGS2 was assumedly affected by the developmental immaturity of CB-derived cells and this might be somehow involved in the relatively low incidence and low severity of acute GVHD following CBT.

Acknowledgment This work was partially supported by grants from Chugai Pharmaceutical Company for study of gene expression in myelodysplastic syndrome and a search for novel therapeutic targets in hematologic malignancies using Gene Chip. Financial disclosure The authors declare no competing financial interests.

References 1. Morris ES, Hill GR. Advances in the understanding of acute graftversus-host disease. Br J Haematol. 2007;137:3–19. 2. Ichiba T, Teshima T, Kuick R, et al. Early changes in gene expression profiles of hepatic GVHD uncovered by oligonucleotide microarrays. Blood. 2003;102:763–771. 3. Sugerman PB, Faber SB, Willis LM, et al. Kinetics of gene expression in murine cutaneous graft-versus-host disease. Am J Pathol. 2004;164: 2189–2202. 4. Zhou L, Askew D, Wu C, Gilliam AC. Cutaneous gene expression by DNA microarray in murine sclerodermatous graft-versus-host disease, a model for human scleroderma. J Invest Dermatol. 2007;127:281– 292. 5. Takahashi S. Leukemia: cord blood for allogeneic stem cell transplantation. Curr Opin Oncol. 2007;19:667–672. 6. Takahashi S, Ooi J, Tomonari A, et al. Comparative single-institute analysis of cord blood transplantation from unrelated donors with bone marrow or peripheral blood stem-cell transplants from related donors in adult patients with hematologic malignancies after myeloablative conditioning regimen. Blood. 2007;109:1322–1330. 7. Glucksberg H, Storb R, Fefer A, et al. Clinical manifestations of graftversus-host disease in human recipients of marrow from HLA-matched sibling donors. Transplantation. 1974;18:295–304. 8. Gentleman RC, Carey VJ, Bates DM, et al. Bioconductor: open software development for computational biology and bioinformatics. Genome Biol.. 2004;5:R80. 9. Tusher VG, Tibshirani R, Chu G. Significance analysis of microarrays applied to the ionizing radiation response. Proc Natl Acad Sci U S A. 2001;98:5116–5121. 10. Lee EG, Boone DL, Chai S, et al. Failure to regulate TNF-induced NF-kappaB and cell death responses in A20-deficient mice. Science. 2000;289:2350–2354. 11. Nahta R, Yuan LX, Fiterman DJ, et al. B cell translocation gene 1 contributes to antisense Bcl-2-mediated apoptosis in breast cancer cells. Mol Cancer Ther. 2006;5:1593–1601. 12. Busson M, Carazo A, Seyer P, et al. Coactivation of nuclear receptors and myogenic factors induces the major BTG1 influence on muscle differentiation. Oncogene. 2005;24:1698–1710.

1769

13. Jaksch M, Mattsson J. The pathophysiology of acute graft-versus-host disease. Scand J Immunol. 2005;61:398–409. 14. Kuo CT, Veselits ML, Leiden JM. LKLF: a transcriptional regulator of single-positive T cell quiescence and survival. Science. 1997;277: 1986–1990. 15. Buckley AF, Kuo CT, Leiden JM. Transcription factor LKLF is sufficient to program T cell quiescence via a c-Mycdependent pathway. Nat Immunol. 2001;2:698–704. 16. Das H, Kumar A, Lin Z, et al. Kruppel-like factor 2 (KLF2) regulates proinflammatory activation of monocytes. Proc Natl Acad Sci U S A. 2006;103:6653–6658. 17. Barjaktarevic´ I, Rahman A, Radoja S, et al. Inhibitory role of IFNgamma-inducible lysosomal thiol reductase in T cell activation. J Immunol. 2006;177:4369–4375. 18. Jaksch M, Remberger M, Mattsson J. Increased gene expression of chemokine receptors is correlated with acute graft-versus-host disease after allogeneic stem cell transplantation. Biol Blood Marrow Transplant. 2005;11:280–287. 19. Wysocki CA, Panoskaltsis-Mortari A, Blazar BR, Serody JS. Leukocyte migration and graft-versus-host disease. Blood. 2005;105:4191– 4199. 20. Lee AH, Hong JH, Seo YS. Tumour necrosis factor-alpha and interferon-gamma synergistically activate the RANTES promoter through nuclear factor kappaB and interferon regulatory factor 1 (IRF-1) transcription factors. Biochem J. 2000;350:131–138. 21. Kaminski BA, Kadereit S, Miller RE, et al. Reduced expression of NFAT-associated genes in UCB versus adult CD4þ T lymphocytes during primary stimulation. Blood. 2003;102:4608–4617. 22. Makino Y, Cook DN, Smithies O, et al. Impaired T cell function in RANTES-deficient mice. Clin Immunol. 2002;102:302–309. 23. Jiang H, Van De Ven C, Satwani P, Baxi LV, Cairo MS. Differential gene expression patterns by oligonucleotide microarray of basal versus lipopolysaccharide-activated monocytes from cord blood versus adult peripheral blood. J Immunol. 2004;172:5870–5879. 24. Takahashi T, Dejbakhsh-Jones S, Strober S. Expression of CD161 (NKR-P1A) defines subsets of human CD4 and CD8 T cells with different functional activities. J Immunol. 2006;176:211–216. 25. Tsujimura K, Obata Y, Matsudaira Y, et al. Characterization of murine CD160þ CD8þ T lymphocytes. Immunol Lett.. 2006;106:48–56. 26. Rey J, Giustiniani J, Mallet F, et al. The co-expression of 2B4 (CD244) and CD160 delineates a subpopulation of human CD8þ T cells with a potent CD160-mediated cytolytic effector function. Eur J Immunol. 2006;36:2359–2366. 27. Szabolcs P, Park KD, Reese M, Marti L, Broadwater G, Kurtzberg J. Coexistent naı¨ve phenotype and higher cycling rate of cord blood T cells as compared to adult peripheral blood. Exp Hematol. 2003;31:708–714. 28. Smith WL, Garavito RM, DeWitt DL. Prostaglandin endoperoxide H synthases (cyclooxygenases)-1 and -2. J Biol Chem. 1996;271:33157– 33160. 29. SenBanerjee S, Lin Z, Atkins GB, et al. KLF2 is a novel transcriptional regulator of endothelial proinflammatory activation. J Exp Med. 2004;199:1305–1315. 30. Sen-Banerjee S, Mir S, Lin Z, et al. Kruppel-like factor 2 as a novel mediator of statin effects in endothelial cells. Circulation. 2005;112: 720–726. 31. Zeiser R, Youssef S, Baker J, Kambham N, Steinman L, Negrin RS. Preemptive HMG-CoA reductase inhibition provides graft-versus-host disease protection by Th-2 polarization while sparing graft-versusleukemia activity. Blood. 2007;110:4588–4598. 32. Carballo E, Lai WS, Blackshear PJ. Feedback inhibition of macrophage tumor necrosis factor-alpha production by tristetraprolin. Science. 1998;281:1001–1005. 33. Carrick DM, Lai WS, Blackshear PJ. The tandem CCCH zinc finger protein tristetraprolin and its relevance to cytokine mRNA turnover and arthritis. Arthritis Res Ther. 2004;6:248–264.

1770

N. Takahashi et al./ Experimental Hematology 2008;36:1760–1770

34. Sugihara M, Tsutsumi A, Suzuki E, et al. Effects of infliximab therapy on gene expression levels of tumor necrosis factor alpha, tristetraprolin, T cell intracellular antigen 1, and Hu antigen R in patients with rheumatoid arthritis. Arthritis Rheum. 2007;56:2160–2169. 35. Fabris M, Tolusso B, Di Poi E, et al. Mononuclear cell response to lipopolysaccharide in patients with rheumatoid arthritis: relationship with tristetraprolin expression. J Rheumatol. 2005;32:998– 1005. 36. Feinberg MW, Wara AK, Cao Z, et al. The Kruppel-like factor KLF4 is a critical regulator of monocyte differentiation. EMBO J. 2007;26: 4138–4148. 37. Feinberg MW, Cao Z, Wara AK, Lebedeva MA, Senbanerjee S, Jain MK. Kruppel-like factor 4 is a mediator of proinflammatory signaling in macrophages. J Biol Chem. 2005;280:38247–38258. 38. Smyth MJ, Cretney E, Takeda K, et al. Tumor necrosis factor-related apoptosis-inducing ligand (TRAIL) contributes to interferon gammadependent natural killer cell protection from tumor metastasis. J Exp Med. 2001;193:661–670. 39. Schmaltz C, Alpdogan O, Kappel BJ, et al. T cells require TRAIL for optimal graft-versus-tumor activity. Nat Med. 2002;8:1433–1437. 40. Nomura S, Ishii K, Inami N, et al. Role of soluble tumor necrosis factor-related apoptosis-inducing ligand concentrations after stem cell transplantation. Transpl Immunol. 2007;18:115–121.

41. Lu¨nemann JD, Waiczies S, Ehrlich S, et al. Death ligand TRAIL induces no apoptosis but inhibits activation of human (auto)antigenspecific T cells. J Immunol. 2002;168:4881–4888. 42. Grossman WJ, Verbsky JW, Tollefsen BL, Kemper C, Atkinson JP, Ley TJ. Differential expression of granzymes A and B in human cytotoxic lymphocyte subsets and T regulatory cells. Blood. 2004;104:2840– 2848. 43. Grossman WJ, Verbsky JW, Barchet W, Colonna M, Atkinson JP, Ley TJ. Human T regulatory cells can use the perforin pathway to cause autologous target cell death. Immunity. 2004;21:589–601. 44. Price JD, Schaumburg J, Sandin C, Atkinson JP, Lindahl G, Kemper C. Induction of a regulatory phenotype in human CD4þ T cells by streptococcal M protein. J Immunol. 2005;175:677–684. 45. Kawamura K, Kadowaki N, Kitawaki T, Uchiyama T. Virus-stimulated plasmacytoid dendritic cells induce CD4þ cytotoxic regulatory T cells. Blood. 2006;107:1031–1038. 46. Gerosa F, Nisii C, Righetti S, et al. CD4(þ) T cell clones producing both interferon-gamma and interleukin-10 predominate in bronchoalveolar lavages of active pulmonary tuberculosis patients. Clin Immunol. 1999;92:224–234. 47. MacDonald AJ, Duffy M, Brady MT, et al. CD4 T helper type 1 and regulatory T cells induced against the same epitopes on the core protein in hepatitis C virus-infected persons. J Infect Dis. 2002;185:720–727.

N. Takahashi et al./ Experimental Hematology 2008;36:1760–1770

Appendix

1770.e1

N. Takahashi et al./ Experimental Hematology 2008;36:1760–1770

1770.e2

N. Takahashi et al./ Experimental Hematology 2008;36:1760–1770

1770.e3

N. Takahashi et al./ Experimental Hematology 2008;36:1760–1770

1770.e4