Cancer Immunology, Immunotherapy

, Volume 64, Issue 7, pp 791–804

Two distinct effector memory cell populations of WT1 (Wilms’ tumor gene 1)-specific cytotoxic T lymphocytes in acute myeloid leukemia patients

  • Yoshiki Nakae
  • Yoshihiro Oka
  • Fumihiro Fujiki
  • Soyoko Morimoto
  • Toshio Kamiya
  • Satoshi Takashima
  • Jun Nakata
  • Sumiyuki Nishida
  • Hiroko Nakajima
  • Naoki Hosen
  • Akihiro Tsuboi
  • Taiichi Kyo
  • Yusuke Oji
  • Kenji Mizuguchi
  • Atsushi Kumanogoh
  • Haruo Sugiyama
Original Article

DOI: 10.1007/s00262-015-1683-7

Cite this article as:
Nakae, Y., Oka, Y., Fujiki, F. et al. Cancer Immunol Immunother (2015) 64: 791. doi:10.1007/s00262-015-1683-7

Abstract

Wilms’ tumor gene 1 (WT1) protein is a promising tumor-associated antigen for cancer immunotherapy. We have been performing WT1 peptide vaccination with good clinical responses in over 750 patients with leukemia or solid cancers. In this study, we generated single-cell gene-expression profiles of the effector memory (EM) subset of WT1-specific cytotoxic T lymphocytes (CTLs) in peripheral blood of nine acute myeloid leukemia patients treated with WT1 peptide vaccine, in order to discriminate responders (WT1 mRNA levels in peripheral blood decreased to undetectable levels, decreased but stayed at abnormal levels, were stable at undetectable levels, or remained unchanged from the initial abnormal levels more than 6 months after WT1 vaccination) from non-responders (leukemic blast cells and/or WT1 mRNA levels increased relative to the initial state within 6 months of WT1 vaccination) prior to WT1 vaccination. Cluster and principal component analyses performed using 83 genes did not discriminate between responders and non-responders prior to WT1 vaccination. However, these analyses revealed that EM subset of WT1-specific CTLs could be divided into two groups: the “activated” and “quiescent” states; in responders, EM subset of the CTLs shifted to the “quiescent” state, whereas in non-responders, those shifted to the “activated” state following WT1 vaccination. These results demonstrate for the first time the existence of two distinct EM states, each of which was characteristic of responders or non-responders, of WT1-specific CTLs in AML patients, and raises the possibility of using advanced gene-expression profile analysis to clearly discriminate between responders and non-responders prior to WT1 vaccination.

Keywords

WT1 CTL Single-cell Gene-expression profiles 

Abbreviations

AML

Acute myeloid leukemia

BM

Bone marrow

CTLs

Cytotoxic T lymphocytes

EM

Effector memory

ER

Early relapse

GO

Gene ontology

HCR

Hematological complete remission

MRD

Minimal residual disease

PB

Peripheral blood

PBMCs

Peripheral blood mononuclear cells

PCA

Principal component analysis

PC1

First principal component

PC2

Second principal component

TAA

Pan-tumor-associated antigen

WT1

Wilms’ tumor gene 1

Introduction

Wilms’ tumor gene 1 (WT1) was isolated as the responsible gene for Wilms’ tumor, a childhood renal tumor, and defined as a tumor-suppressor gene [1, 2]. WT1 encodes a transcription factor that regulates many important genes and plays an essential role in normal cell proliferation, differentiation, and embryogenesis [3, 4, 5]. Despite its initially reported tumor-suppressor functions, WT1 is overexpressed in leukemia and various types of solid cancers [6, 7] and plays an important role in leukemogenesis and tumorigenesis through an oncogenic function, such as promotion of cell proliferation, motility, suppression of cell differentiation, and apoptosis [8, 9].

WT1 protein is a pan-tumor-associated antigen (TAA) that is overexpressed in not only leukemia but also in various types of solid cancers and is regarded as one of the most promising TAAs for cancer immunotherapy [10]. Currently, WT1-targeted immunotherapy is being actively performed and the accumulating evidence shows that HLA class I-restricted WT1 peptide vaccines are safe and effective for leukemia and various types of advanced solid cancers [11, 12, 13]. We had previously reported successful WT1 vaccination in patients with acute myeloid leukemia [14, 15, 16, 17], chronic myeloid leukemia [18, 19], myelodysplastic syndromes [20, 21], multiple myeloma [22], glioblastoma multiforme [23, 24, 25, 26], malignant melanoma [27], and lung [28], breast [17], pancreatic [29, 30], ovarian [31, 32], uterine [33, 34], and salivary gland cancers [35, 36], as well as infantile sarcomas including rhabdomyosarcoma [37, 38]. These clinical studies let us notice the importance of discriminating responders from non-responders prior to WT1 vaccination in order to improve the clinical efficacy (see the section entitled “Patients and samples” for detailed definition of responders and non-responders).

One method for discriminating responders from non-responders prior to WT1 vaccination is selection of patients with higher frequencies of WT1-specific cytotoxic T lymphocytes (CTLs) [16, 17, 29]. We previously reported that among pancreatic cancer patients, responders had higher frequencies of effector memory (EM) subset of WT1-specific CTLs, which play an essential role in maintenance of WT1-specific CTLs, than non-responders prior to WT1 vaccination. Furthermore, the EM subset’s frequencies increased in responders, but not in non-responders, upon repeated WT1 vaccination [29]. However, the prediction of responders based on the EM subset’s frequencies was insufficient because some patients well responded to WT1 peptide vaccine despite the absence of increase in the EM subset’s frequencies prior to the vaccination. Therefore, development of the novel methods to discriminate clearly between responders and non-responders prior to WT1 vaccination is necessary. However, at present time, there is no method for selection of responders other than measuring of the frequencies of WT1-specific CTLs.

The results obtained thus far indicated that the clear discrimination between responders and non-responders by quantity (frequency) of EM subset of WT1-specific CTLs was difficult. Therefore, the discrimination by quality of EM subset of the CTLs should be needed. One way to determine their quality would be to analyze their gene-expression profiles and then identify differences in the profiles between responders and non-responders. However, when performed on a cell population, such as EM subset of WT1-specific CTLs, traditional high-throughput assays (including gene-expression arrays) only provide us with the average gene-expression levels for the cell population [39, 40, 41]. Therefore, we focused on analyzing gene-expression profiles at single-cell level.

In the present study, we analyzed expression profiles of 83 genes at single-cell levels in EM subset of WT1-specific CTLs, obtained before and after WT1 vaccination, from nine acute myeloid leukemia patients treated with WT1 peptide vaccine, and demonstrated that EM subset of WT1-specific CTLs can be divided into two distinct populations in the “activated” and the “quiescent” states by differential gene-expression profiles, and those in responders shifted from the “activated” toward the “quiescent” state, while those in non-responders stayed in the “activated” state following WT1 peptide vaccination. These results clearly demonstrated for the first time the existence of two distinct populations of EM subset of WT1-specific CTLs: the “activated” and the “quiescent” states, with different gene-expression profiles. The results should provide us with a novel insight into understanding of responders and non-responders to WT1 immunotherapy.

Materials and methods

Patients and samples

Nine AML patients were treated with WT1 peptide vaccine with approval of the Ethical Review Boards of the Faculty of Medicine, Osaka University, and the Hiroshima Red Cross and Atomic Bomb Survivor Hospital. AML patients aged 16–80 were eligible if their diseases were resistant to conventional chemotherapy. Patients who had received hematopoietic stem cell transplantation were excluded. Inclusion criteria were: (1) overexpression of the WT1 gene in leukemic cells, as determined by RT-PCR; (2) HLA-A*24:02 positivity; (3) estimated survival of >2 months; (4) performance status from 0 to 2 (Eastern Cooperative Oncology Group); (5) no severe impairment of organ function; and (6) no administration of chemotherapy, immunotherapy, immunosuppressive therapy, or radiotherapy within 4 weeks before WT1 vaccination. All the patients except patient 6 were in hematological complete remission (HCR), in which blast cells accounted for <5 % of bone marrow (BM) cells. Patient 6 was in early relapse (ER), in which blast cells accounted for 5–20 % of BM cells. Since all patients enrolled were at high risk of relapse, maintenance of unchanged WT1 mRNA levels for more than 6 months was considered to reflect the effects of WT1 vaccination. Therefore, patients in whom WT1 mRNA levels in peripheral blood (PB) decreased to undetectable levels, decreased but stayed at abnormal levels, were stable at undetectable levels, or remained unchanged from the initial abnormal levels more than 6 months after WT1 vaccination were defined as responders. On the other hand, the patients in whom leukemic blast cells and/or WT1 mRNA levels increased relative to the initial state within 6 months of WT1 vaccination were defined as non-responders. Patients’ characteristics are shown in Table 1. After written informed consent was given, 5 ml of PB was obtained from each patient.
Table 1

Characteristics of nine AML patients who were treated with WT1 peptide vaccine

Patients

Age

Sex

WT1 peptide

FAB classification

Status at entry

DFS (months)

Clinical effects responder/non-responder

WT1mRNA levels

1

49

F

M

M1

HCR

>120

Responder

Decreased but stayed at abnormal levels

2

60

F

M

t-MN

HCR

>120

Responder

Decreased but stayed at abnormal levels

3

32

M

N

M4

HCR

>120

Responder

Decreased but stayed at abnormal levels

4

69

F

M

M1

HCR

>33

Responder

Decreased to undetectable levels

5

75

M

M

M2

HCR

10a

Responder

Stable at undetectable levels

6

61

M

M

M2

ER

1b

Non-responder

Increased

7

68

M

M

M2

HCR

6b

Non-responder

Increased

8

75

F

M

M4

HCR

2b

Non-responder

Increased

9

63

M

M

M5

HCR

2b

Non-responder

Increased

M and N indicate modified and natural WT1 peptide, respectively. AML was classified according to WHO classification 2008 on myeloid neoplasms. M1: AML without maturation, M2: AML with maturation, M4: acute myelomonocytic leukemia, M5: acute monoblastic and monocytic leukemia, t-MN: therapy-related myeloid neoplasms, HCR: hematological complete remission (blasts <5 % in BM), ER: early relapse (5 % ≤ blasts < 20 % in BM), DFS: disease-free survival

aOverall survival was 72 months

bDeath due to relapse within 1 month

Preparation of WT1 peptide vaccine and the vaccination schedule

A natural 9-mer WT1 peptide (amino acids 235–243, CMTWNQMNL) and the modified 9-mer WT1 peptide (amino acids 235–243, CYTWNQMNL), in which Y was substituted for M at amino acid position 2 (anchor position) of the natural WT1 peptide, were used for immunization. WT1 peptides (GMP grade) were purchased from Multiple Peptide Systems (San Diego, CA, USA) and Peptide Institute (Osaka, Japan) as lyophilized peptides. After written informed consent was given, skin test-negative patients were intradermally injected with 3.0 mg of HLA-A*24:02-restricted, natural or modified 9-mer WT1 peptide emulsified with Montanide ISA51 adjuvant. WT1 vaccination was scheduled to be performed biweekly for patients 1, 2, and 3, and weekly for the other patients, according to the protocols.

Flow cytometric analysis and single-cell sorting

After 16–24 h of culture of thawed peripheral blood mononuclear cells (PBMCs) in complete media composed of X-vivo 15 (Lonza, Wakersville, MD, USA) supplemented with 10 % AB serum(Gemini Bio-Products, West Sacramento, CA, USA), 1.0 ng/ml of IL7 (PeproTech, London, UK), and 1.0 ng/ml of IL15 (PeproTech, London, UK), non-adherent cells were resuspended in FACS buffer composed of PBS, 0.02 % sodium azide and 2 % FBS and then incubated with phycoerythrin (PE)-conjugated HLA-A*24:02/WT1 235–243 tetramer (MBL, Tokyo, Japan) at 37 °C for 30 min. Subsequently, these cells were stained at 4 °C for 25 min in the dark with a mixture of mAbs [anti-CD3-PacificBlue (BD Pharmigen, San Diego, CA, USA), anti-CD8-APC-Cy7 (BD Biosciences, San Jose, CA, USA), anti-CD45RA-FITC (eBioscience, San Diego, CA, USA), and anti-CCR7-PE-Cy7 (BD Biosciences)], washed three times, resuspended in FACS buffer, and stained with 7AAD to exclude non-viable cells. In this study, CD3, CD8, and WT1235 tetramer-positive T lymphocytes were defined as WT1 tetramer+ CD8+ T cells. Based on CCR7 and CD45RA expression, WT1 tetramer+ CD8+ T cells were phenotypically classified into four subsets [42, 43]: naïve (CCR7+ CD45RA+), central memory (CCR7+ CD45RA−), EM (CCR7− CD45RA−), and terminally differentiated effector (CCR7− CD45RA+). Flow cytometric analysis was performed using a FACSAria instrument (BD Biosciences), and data were analyzed using the FACS Diva software (BD Biosciences). EM subset of WT1 tetramer+ CD8+ T cells were sorted as single cells on the FACSAria.

Comprehensive measurement of single-cell gene-expression levels

Single cells were sorted on the FACSAria into 96-well plates containing Platinum Taq polymerase, SuperScript III reverse transcriptase (Invitrogen, Carlsbad, CA, USA), a mixture of 50 nM each of DELTAgene Assay primer pairs [44, 45] (Fluidigm, San Francisco, CA, USA) (Supplementary Table S1), CellsDirect One-Shot qRT-PCR buffer (Invitrogen), and SUPERase In RNase inhibitor (Invitrogen). Immediately after cell sorting, samples were centrifuged, reverse-transcribed into cDNA (50 °C for 15 min, 95 °C for 2 min), and preamplified for 20 PCR cycles (each cycle: 95 °C for 15 s and 60 °C for 4 min). The resultant preamplified single-cell cDNA was stored at −80 °C until analysis. Each cDNA sample was then separated into 48 reaction tubes for further qPCR using the BioMark 48.48 dynamic array nanofluidic chip (Fluidigm) and then reacted according to the manufacturer’s protocol [46].

Data analysis for gene-expression levels measured by single-cell PCR

Single-cell PCR data were analyzed using the Fluidigm real-time PCR analysis software (Fluidigm). In each experiment, 48 samples containing two positive and two negative controls were analyzed. As positive controls, Universal cDNA-Random primer from Human normal tissues (BioChain, San Francisco, CA, USA) was used. Single-cell samples in which the expression of the house-keeping gene GAPDH was undetectable or detected at extremely low (Ct > 35) were removed from the analysis. Gene-expression analysis was performed by the ∆∆Ct method using GAPDH as a normalization control. Hierarchical clustering and principal component analysis (PCA) were performed with Ward’s method using the Jmp® 10 software (SAS, Cary, NC, USA). Ninety-six genes that were considered to be important for cell proliferation, apoptosis, exhaustion, transcription, and cytokine production were examined [47]. Four genes (GAPDH, GUSB, HSP90, and HPRT1) were considered as candidates for internal control, but only GAPDH was finally chosen because its expression was most stable. Seven genes (TNF, PDCD1, KLF2, HLA-DRB1, TNFRSF7, CD58, and TNFRSF9) were excluded from the analysis because these genes were not expressed in any of the single-cell samples. Hierarchical clustering and PCA were performed for the remaining 83 genes that were differentially expressed in single cells. Gene ontology (GO) enrichment analysis was performed for the 31 genes that exhibited a loading value of 0.2 or more for the first principal component (named “List 1,” as shown in Supplementary Table S2a) and the 13 genes that exhibited a loading value of −0.2 or less for that component (named “List 2,” as shown in Supplementary Table S2b). TargetMine was used to identify the GO terms enriched for each of the selected gene lists by setting the entire set of 83 genes as the reference population [48]. The statistical significance of the enrichment was evaluated using Fisher’s exact test with the Benjamini–Hochberg method for multiple testing correction. GO terms related to biological process, cellular component, and molecular function were examined with an adjusted P < 0.1.

Results

No significant difference in the frequencies of WT1 tetramer+ CD8+ T cells in PB between responders and non-responders

HLA-A*24:02-positive AML patients in HCR (HCR: blasts <5 % in BM) or ER (ER: 5 % ≤ blasts < 20 % in BM) were vaccinated with HLA-A*24:02-restricted WT1235 peptide (Table 1).

PBMCs were examined for WT1 tetramer+ CD8+ T cells using WT1 tetramer assay. The frequencies of WT1 tetramer+ CD8+ T cells were 0.11–0.25 % (mean 0.19 %) and 0.06–0.47 % (mean 0.18 %) in responders and non-responders, respectively, prior to WT1 vaccination. Thus, there was no significant difference in the frequencies of WT1 tetramer+ CD8+ T cells between responders and non-responders (Fig. 1). Based on CCR7 and CD45RA expression, WT1 tetramer+ CD8+ T cells could be divided into four subsets: naïve (CCR7+ CD45RA+), central memory (CCR7+ CD45RA−), EM (CCR7− CD45RA−), and terminally differentiated effector (CCR7− CD45RA+) (Fig. 1b). Higher frequencies of EM subset of WT1 tetramer+ CD8+ T cells prior to WT1 vaccination correlated with better responses to WT1 vaccination [29]. However, there was no significant difference in the frequencies of the EM subset between responders and non-responders (data not shown). Therefore, the difference in the expression levels of various functional genes in EM subset of WT1 tetramer+ CD8+ T cells between responders and non-responders was examined in detail at the single-cell levels.
Fig. 1

Frequencies of WT1 tetramer+ CD8+ T cells in PB of WT1 peptide-vaccinated AML patients. a Representative data of flow cytometric analysis using WT1235 tetramer. CD3+, CD8+, 7AAD−, and WT1235 tetramer+ T cells were defined as WT1 tetramer+ CD8+ T cells. b WT1 tetramer+ CD8+ T cells were classified into four distinct differentiation subsets according to their cell surface expression of CCR7 and CD45RA as follows: (1) CCR7+ CD45RA+ (naïve) cells, (2) CCR7+ CD45RA− (central memory) cells, (3) CCR7− CD45RA− (effector memory) cells, and (4) CCR7− CD45RA+ (terminally differentiated effector) cells. Representative data from patient 5 are shown. N naïve, CM central memory, EM effector memory, E terminally differentiated effector. c Frequencies of WT1 tetramer+ CD8+ T cells in responders and non-responders prior to WT1 vaccination. Bars indicate mean values of the frequencies. res responders, non-res non-responders

Single-cell-based gene-expression analysis of EM subset of WT1 tetramer+ CD8+ T cells

EM subset of WT1 tetramer+ CD8+ T cells were sorted as single cells, and the single cells were quantified for the expression of 84 genes (including one house-keeping gene) responsible for transcription, proliferation, apoptosis, exhaustion, and cytokine production [47] using TaqMan real-time PCR with the BioMark system.

A total of 162 and 145 single cells were obtained from five responders and four non-responders, respectively, prior to WT1 vaccination, and hierarchical clustering analysis was performed (Fig. 2). We expected that this analysis could classify the total of 307 single cells into several groups according to the expression levels of the 83 genes. This analysis unequivocally divided the single cells into two populations: group A (137 single cells) and group B (170 single cells), which were characterized by distinct gene-expression profiles. Group A included 59 of 162 single cells from responders and 78 of 145 single cells from non-responders, whereas group B included the remaining 103 and 67 single cells from responders and non-responders, respectively. Therefore, hierarchical clustering analysis could not clearly discriminate between responders and non-responders prior to WT1 vaccination.
Fig. 2

Single-cell gene-expression profiles in EM subset of WT1 tetramer+ CD8+ T cells prior to vaccination. EM subset of WT1 tetramer+ CD8+ T cells were single-cell-sorted from PB of the nine AML patients. Gene-expression heatmap shows gene-expression profiles for individual 307 cells from all nine patients prior to vaccination, obtained after two-way hierarchical clustering using the Euclidean distance and Ward agglomeration methods. Expression of the indicated mRNA transcripts was measured by quantitative RT-PCR. Heatmap shows expression levels of 83 genes for 307 single cells isolated from five responders (horizontal red line) and four non-responders (horizontal blue line). Red, white, and blue colors indicate high, intermediate, and low expression

Change in gene-expression profiles in EM subset of WT1 tetramer+ CD8+ T cells characteristic to responders and non-responders following WT1 vaccination

Next, PCA, a form of multivariate analysis that is easy to understand visually, was performed to weight statistically the involvement of each gene in responders and non-responders, to make principal components, and to evaluate comprehensively one each of single cells for the involvement in responders and non-responders. Although PCA is statistically equivalent to hierarchical clustering analysis, this approach makes it possible representing gene-expression profiles of individual single cells in two dimensions, allowing groups of cells to be easily recognized.

First, PCA was applied to the analysis of gene-expression profiles of EM subset of WT1 tetramer+ CD8+ T cells prior to WT1 vaccination (Fig. 3a). The first principal component (PC1) explained 10.2 % of the observed variance, and the second principal component (PC2) explained 3.7 %. A projection of the gene-expression profiles onto PC1 clearly divided the single cells into two distinct populations with PC1 ≥0 (group C) or PC1 <0 (group D) (Fig. 3a). As expected statistically, group C and group D were identical to group A and group B, respectively. Therefore, groups A and B were used here instead of groups C and D for convenience. The gene-expression profiles obtained from 218 to 215 single cells from responders and non-responders, respectively, at 3–6 and 10–14 weeks post-WT1 vaccination, were projected onto the two principal components. Although PCA patterns changed following WT1 vaccination, they could not discriminate between responders and non-responders (Fig. 3a).
Fig. 3

Substantial change in gene-expression profiles in EM subset of WT1 tetramer+ CD8+ T cells following WT1 vaccination. a PCA of single-cell gene-expression profiles prior to WT1 vaccination, 3–6, and 10–14 weeks post-WT1 vaccination. Each symbol represents an individual cell derived from responders (closed circle) and non-responders (cross). Groups C and D were identical to groups A and B, respectively. b PCA for individual patients at each time point. Ninety percent tolerance ellipses are shown, reflecting the distribution of individual single cells from each patient prior to WT1 vaccination, 3–6, and 10–14 weeks post-WT1 vaccination. Samples from patients 1 and 3 at 3–6 weeks post-WT1 vaccination and from patients 6, 8, and 9 at 10–14 weeks post-WT1 vaccination were not available for analysis. Three (patients 6, 8, and 9) of four non-responders had already died before the time point of 10–14 weeks; therefore, only one non-responder (patient 7) was analyzed

Next, the gene-expression profiles of each patient were projected onto the two principal components (Fig. 3b). Overall, responders and non-responders exhibited different tendencies with respect to the aforementioned groups. Among responders, patient 1 completely shifted from group A to group B at 10–14 weeks post-WT1 vaccination; patient 2 gradually shifted from group A to group B following WT1 vaccination and stayed in the boundary at 10–14 weeks post-WT1 vaccination; patient 3 stayed in group B following WT1 vaccination, although the gene-expression profiles were not determined at 3–6 weeks post-WT1 vaccination; and patient 4 moved from group B toward group A following WT1 vaccination and stayed in the boundary at 10–14 weeks post-WT1 vaccination. Although patient 5 was in group B prior to WT1 vaccination and then distributed widely across the two groups at 3–6 weeks post-WT1 vaccination, this patient completely shifted to group B at 10–14 weeks post-WT1 vaccination. Among the non-responders, two patients (patients 6 and 7) belonged to group A and the remaining two patients (patients 8 and 9) belonged to group B, prior to WT1 vaccination. Surprisingly, patients 8 and 9 shifted from group B to group A at 3–6 weeks post-WT1 vaccination, whereas patients 6 and 7 stayed in group A. Therefore, all four non-responders accumulated in group A at 3–6 weeks post-WT1 vaccination. Patient 7 continued to stay in group A until 10–14 weeks post-WT1 vaccination because the remaining three non-responders died of aggravation as a result of leukemia before that time.

Difference between groups A and B in biological characteristics of EM subset of WT1 tetramer+ CD8+ T cells

As described above, EM subset of WT1 tetramer+ CD8+ T cells were divided into two groups, A and B, based on their gene-expression profiles. Because the principal components consisted of variable contributions from all 83 genes analyzed, the individual contributions of the 83 genes to the two groups were weighted (Fig. 4). Higher expression of genes with positive values for PC1 loading contributed more to inclusion of the single cells in group A, whereas higher expression of genes with negative values for PC1 loading contributed more to inclusion of the single cells in group B. To examine the biological characteristics of single cells in groups A and B, GO enrichment analysis was performed using TargetMine [48] based on the PC1 loading value.
Fig. 4

Evaluation of the contribution of each of 83 genes to the PC1. Five genes (RGS1, IFNG, HLA-DPA1, CCR5, and TCF7) whose PC1 loading scores were highest and five genes (PTGER2, LGALS1, FOXO3, CXCR1, and BY55) whose PC1 loading scores were lowest are shown. CCL5 contributed to all GO terms related to activation of metabolism and cell motility, and BAD contributed to all GO terms related to autophagy and activation of catabolism and proteolysis

Genes with positive and negative PC1 loading values contributed to groups A and B, respectively. To select genes that would be more useful for GO enrichment analysis, cutoff values for PC1 loading were determined. When cutoff values of ≥0.1 and ≤−0.1, ≥0.2 and ≤−0.2, ≥0.3 and ≤−0.3, ≥0.4 and ≤−0.4, and ≥0.5 and ≤−0.5 were temporarily set and GO enrichment analysis was performed using the genes selected by each cutoff value, cutoff values of ≥0.2 and less ≤−0.2 yielded GO terms that were strikingly different between group A and group B. Therefore, GO enrichment analysis was performed using genes with PC1 loading values of ≥0.2 and ≤−0.2.

Thirty-one genes with PC1 loading value ≥0.2 were selected as genes that positively contributed to inclusion of single cells in group A (List 1 in Supplementary Table S2a), whereas thirteen genes with PC1 loading value ≤−0.2 were selected as genes that positively contributed to inclusion of single cells in group B (List 2 in Supplementary Table S2b). Next, GO enrichment analysis was performed for both sets of genes, using all 83 genes as the reference set. Biological process-related GO terms, which were statistical terms that did not directly relate to biological function or cellular components, were enriched by setting an adjusted P value <0.1 for genes in lists 1 and 2 in comparison with the reference set of 83 genes (Table 2). Molecular function- and cellular component-related GO terms were not enriched in a statistically significant manner.
Table 2

Biological process-related GO terms enriched for lists 1 and 2

Ancestor GO term

GO term

List 1

CCL5

IFNG

PIM1

CDKN1B

BMI1

TP53

RGS1

MAP3K5

BCL2L11

DUSP4

FAS

CXCR4

CCR5

Regulation of molecular function

Regulation of molecular function

 

Positive regulation of molecular function

 

 

 

Regulation of catalytic activity

Regulation of catalytic activity

 

Positive regulation of catalytic activity

 

 

 

Protein metabolic process

Regulation of cellular protein metabolic process

 

 

 

 

Regulation of protein metabolic process

 

 

 

 

Regulation of protein modification process

 

 

 

 

Cellular protein modification process

 

 

 

 

Protein modification process

 

 

 

 

Positive regulation of protein modification process

 

 

   

 

Macromolecule modification

Macromolecule modification

 

 

 

 

Phosphorus metabolic process

Regulation of phosphate metabolic process

 

 

 

 

Regulation of phosphorus metabolic process

 

 

 

 

Phosphorus metabolic process

 

 

 

 

Phosphate-containing compound metabolic process

 

 

 

 

Regulation of protein phosphorylation

 

 

 

 

 

Positive regulation of protein phosphorylation

  

 

   

 

Positive regulation of phosphorus metabolic process

  

 

   

 

Positive regulation of phosphorylation

  

 

   

 

Positive regulation of phosphate metabolic process

  

 

   

 

Regulation of nitrogen compound metabolic process

Regulation of nucleobase-containing compound metabolic process

      

Regulation of nitrogen compound metabolic process

      

Biosynthetic process

Regulation of biosynthetic process

       

Regulation of macromolecule biosynthetic process

       

Regulation of cellular biosynthetic process

       

Regulation of cellular macromolecule biosynthetic process

       

Heterocycle biosynthetic process

       

Aromatic compound biosynthetic process

       

Nucleobase-containing compound biosynthetic process

       

Cellular nitrogen compound biosynthetic process

       

Organic cyclic compound biosynthetic process

       

RNA metabolic process

RNA biosynthetic process

       

RNA metabolic process

       

Regulation of RNA metabolic process

       

Regulation of RNA biosynthetic process

       

Localization of cell

Cell migration

          

Cell motility

          

Leukocyte migration

          

localization of cell

          

Catabolic process

Autophagy

             

Regulation of autophagy

             

Positive regulation of catabolic process

             

Positive regulation of cellular catabolic process

             

Positive regulation of proteolysis

Positive regulation of proteolysis

             

Ancestor GO term

GO term

List 1

List 2

SELL

IL10

NFKB1

DAZL

TCF7

BCL6

Notch2

STAT5

BAD

BCL2L1

TNFRSF1B

Regulation of molecular function

Regulation of molecular function

 

        

Positive regulation of molecular function

 

        

Regulation of catalytic activity

Regulation of catalytic activity

  

        

Positive regulation of catalytic activity

           

Protein metabolic process

Regulation of cellular protein metabolic process

 

       

Regulation of protein metabolic process

 

       

Regulation of protein modification process

           

Cellular protein modification process

           

Protein modification process

           

Positive regulation of protein modification process

           

Macromolecule modification

Macromolecule modification

           

Phosphorus metabolic process

Regulation of phosphate metabolic process

           

Regulation of phosphorus metabolic process

           

Phosphorus metabolic process

           

Phosphate-containing compound metabolic process

           

Regulation of protein phosphorylation

           

Positive regulation of protein phosphorylation

           

Positive regulation of phosphorus metabolic process

           

Positive regulation of phosphorylation

           
 

Positive regulation of phosphate metabolic process

           

Regulation of nitrogen compound metabolic process

Regulation of nucleobase-containing compound metabolic process

 

 

   

Regulation of nitrogen compound metabolic process

 

 

   

Biosynthetic process

Regulation of biosynthetic process

 

   

Regulation of macromolecule biosynthetic process

 

   

Regulation of cellular biosynthetic process

 

   

Regulation of cellular macromolecule biosynthetic process

 

   

Heterocycle biosynthetic process

 

 

   

Aromatic compound biosynthetic process

 

 

   

Nucleobase-containing compound biosynthetic process

 

 

   

Cellular nitrogen compound biosynthetic process

 

 

   

Organic cyclic compound biosynthetic process

 

 

   

RNA metabolic process

RNA biosynthetic process

 

 

   

RNA metabolic process

 

 

   

Regulation of RNA metabolic process

 

 

   

Regulation of RNA biosynthetic process

 

 

   

Localization of cell

Cell migration

         

Cell motility

         

Leukocyte migration

         

localization of cell

         

Catabolic process

Autophagy

        

 

Regulation of autophagy

        

 

Positive regulation of catabolic process

        

 

Positive regulation of cellular catabolic process

        

 

Positive regulation of proteolysis

Positive regulation of proteolysis

        

 

GO terms with P < 0.1 are shown. GO terms were arbitrarily classified into ancestor GO terms according to the definition of each GO term. GO terms in the leftmost column are the arbitrary ancestor GO terms, and the terms in the left second column are GO terms. Genes that compose the GO terms are shown in the upper row. Each GO term is not placed according to its P value, but is clustered based on its ancestor GO terms

GO terms obtained from gene function in lists 1 and 2 differed from each other. GO terms obtained from genes in List 1 were related to activation of metabolism and motility. On the other hand, GO terms obtained from genes in List 2 were related to autophagy and activation of catabolism and proteolysis. Therefore, these results indicated that the cells in group A were in a metabolically and biosynthetically active state, whereas the cells in group B were in an anti-apoptotic and quiescent state.

CCL5 (Chemokine C–C motif ligand 5) contributed to all GO terms related to activation of metabolism and cell motility in List 1, whereas BAD (Bcl-2-associated agonist of cell death) contributed to all GO terms related to autophagy and activation of catabolism and proteolysis in List 2. These results indicated that these two genes play an important role in the functional regulation of EM subset of WT1 tetramer+ CD8+ T cells.

Discussion

T cell metabolism of activation and differentiation is mainly regulated by TCR stimulation by antigens [49, 50]. EM subset of T cells cease or slow cell division in the absence of TCR stimulation by antigens, whereas they start to proliferate rapidly in the presence of TCR stimulation by antigens. GO enrichment analysis revealed that EM subset of WT1-specific CTLs in group A were “activated,” as well as functionally and metabolically up-regulated, and thus were in a metabolic, biosynthetic, and locomotive state. Therefore, the EM subset in group A were designated as “activated” state for further discussion.

By contrast, catabolism including autophagy is induced following TCR stimulation by antigens. Autophagy plays an important role in prevention of apoptosis of T cell [51, 52]. For example, autophagy-defective T cells are susceptible to apoptosis, whereas CD4+ T cells rapidly proliferate in a neoplastic manner when autophagy in CD4+ T cells increases and prevents apoptosis following infection with HTLV-2 [52]. The GO term “positive regulation of proteolysis,” obtained from genes in List 2, could refer to “cellular catabolic process,” including autophagy. Therefore, the present results indicated that EM subset of WT1-specific CTLs in group B were in an anti-apoptotic and quiescent state with characteristics of memory T cells. Thus, the EM subset in group B were designated as “quiescent” state for further discussion.

Prior to WT1 vaccination, EM subset of WT1-specific CTLs from two of five responders and two of four non-responders were “activated,” whereas those from three of five responders and two of four non-responders were “quiescent.” Therefore, prior to WT1 vaccination, multivariate analysis using our gene set could not discriminate between responders and non-responders. Surprisingly, 3–6 weeks post-WT1 vaccination, EM subset of the CTLs from all of four non-responders shifted to the “activated” state and remained in this state until 10–14 weeks post-WT1 vaccination, although three of four non-responders died of aggravation of leukemia prior to this time points. On the other hand, EM subset of the CTLs from responders showed a strikingly different picture. EM subset of the CTLs initially spread to both the “activated” and “quiescent” states at 3–6 weeks post-WT1 vaccination. However, at 10–14 weeks post-WT1 vaccination, EM subset of the CTLs from three responders (patients 1, 3, and 5) shifted to the “quiescent” state and those from the remaining two responders moved toward the “quiescent” state from the “activated” state. It appears difficult to explain why responders (patients 2 and 4) moved partially toward the “quiescent” gene signature at 10–14 weeks. WT1 mRNA levels in PB of both cases decreased after vaccination; however, there may be a possibility that minimal residual disease (MRD) was retained at 10–14 weeks that the MRD stimulated a WT1-directed immune response and that the immune response interrupted complete shift to the “quiescent” state.

It is not easy to explain why EM subset of WT1-specific CTLs in responders shifted to the “quiescent” state, nor why those in non-responders stayed in the “activated” state. However, one plausible explanation may be as follows: (1) Prior to WT1 vaccination, EM subset of WT1-specific CTLs in patients whose immune cells were spontaneously and effectively primed by endogenous WT1 antigen from WT1-expressing leukemia cells were in the “activated” state, whereas those in patients in whom priming did not occur effectively by undetermined reasons were in the “quiescent” state. The presence or absence of effective priming by endogenous WT1 antigen prior to WT1 vaccination could not become a prediction factor for discrimination between responders and non-responders. (2) At 3–6 weeks post-WT1 vaccination, EM subset of WT1-specific CTLs from two non-responders (patients 8 and 9) changed from the “quiescent” to the “activated” state, probably as a result of effective priming and resultant stimulation of WT1-specific CTLs by the injection of WT1 peptide antigen along with immune stimulatory adjuvant, probably followed by immune stimulation by endogenous WT1 antigen from leukemia cells. On the other hand, EM subset of the CTLs from two (patients 4 and 5) of three responders partially changed from the “quiescent” to the “activated” states and spread widely toward the “activated” state due to the similar reason described above. (3) At 10–14 weeks post-WT1 vaccination, in responders, leukemia cell burden significantly decreased, and as a result, stimulation of WT1-specific CTLs by endogenous WT1 antigen weakened, resulting in the shift of EM subset of WT1-specific CTLs from the “activated” to the “quiescent” state. For example, EM subset of the CTLs, which distributed widely between the “quiescent” to the “activated” state in responder patient 5, completely shifted to the “quiescent” state, probably due to a reduction in stimulation of WT1-specific CTLs by endogenous WT1 antigen as a result of a large reduction in leukemia cell burden. On the other hand, in non-responders, EM subset of the CTLs continued to be “activated” by endogenous WT1 antigen from a large amount of residual leukemia cells, resulting in persistence of the “activated” state.

If this plausible explanation is correct, it raises an interesting question: Why do EM subset of WT1-specific CTLs in responders shift to the “quiescent” state regardless of the successive stimulation of WT1-specific CTLs by repeated WT1 vaccination? We previously reported that in three AML patients who were successively WT1 peptide-vaccinated for over 7 years and have remained in continuous complete remission until now, only a small fraction of WT1-specific CTLs were in the “activated” state. In view of this result, it might be reasonable to speculate that because a small fraction of WT1-specific CTLs in responders were in the “activated” state, the present analysis could not detect such “activated” ones.

The present study could not discriminate between responders and non-responders based on difference in gene-expression profiles in EM subset of WT1-specific CTLs prior to WT1 vaccination. However, the present results raise the possibility of discrimination between responders and non-responders prior to WT1 vaccination by the gene-expression profile analysis of more suitable genes. The 83 genes used for gene-expression profile analysis were arbitrarily selected from among many genes related to metabolism, catabolism, and cell mobility. In a more advanced gene-expression profile analysis, firstly, genes that are strikingly differently expressed between EM subset of WT1-specific CTLs of responders and those of non-responders prior to WT1 vaccination would be selected by gene chip analysis, and then, GO enrichment analysis would be performed using these gene sets. This more advanced GO enrichment analysis would discriminate between responders and non-responders prior to WT1 vaccination. Thus, prediction of responders and non-responders based on genes that are strikingly differently expressed prior to WT1 vaccination may be possible. However, in that case, the prediction would be done only by the expression levels of genes of interest, but would not consider the biological functions of those genes. On the other hand, GO enrichment analysis using a gene set selected by gene chip analysis, as described above, would well consider, evaluate, and weight the biological meaning of genes of interest. Therefore, for discrimination of responders from non-responders prior to WT1 vaccination, GO enrichment analysis should have priority over simple comparison of gene-expression levels.

Acknowledgments

The authors thank the nursing teams for their care of the patients in this study and Ms. Tomoe Umeda for coordination of clinical research. This study was partially supported by the Japan Society for the Promotion of Science (JSPS) through grants for Scientific Research, and Grants-in-Aid for Young Scientists from the Ministry of Education, Science, Sports, Culture and Technology and the Ministry of Health, Labour and Welfare of Japan (Grant Nos. 24591164, 25830116, 25430186 and 25293079). The Department of Cancer Immunology is a department in collaboration with Otsuka Pharmaceutical Co., Ltd., and is supported with a grant from the company. The company had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Supplementary material

262_2015_1683_MOESM1_ESM.pdf (66 kb)
Supplementary material 1 (PDF 66 kb)

Copyright information

© Springer-Verlag Berlin Heidelberg 2015

Authors and Affiliations

  • Yoshiki Nakae
    • 1
  • Yoshihiro Oka
    • 1
    • 2
    • 3
  • Fumihiro Fujiki
    • 2
  • Soyoko Morimoto
    • 2
  • Toshio Kamiya
    • 8
  • Satoshi Takashima
    • 1
  • Jun Nakata
    • 4
  • Sumiyuki Nishida
    • 1
  • Hiroko Nakajima
    • 2
  • Naoki Hosen
    • 5
  • Akihiro Tsuboi
    • 4
  • Taiichi Kyo
    • 6
  • Yusuke Oji
    • 5
  • Kenji Mizuguchi
    • 7
  • Atsushi Kumanogoh
    • 1
    • 3
  • Haruo Sugiyama
    • 8
  1. 1.Departments of Respiratory Medicine, Allergy and Rheumatic Diseases, Graduate School of MedicineOsaka UniversitySuitaJapan
  2. 2.Department of Cancer Immunology, Graduate School of MedicineOsaka UniversitySuitaJapan
  3. 3.Department of Immunopathology, Immunology Frontier Research Center (World Premier International Research Center)Osaka UniversitySuitaJapan
  4. 4.Department of Cancer Immunotherapy, Graduate School of MedicineOsaka UniversitySuitaJapan
  5. 5.Department of Cancer Stem Cell Biology, Graduate School of MedicineOsaka UniversitySuitaJapan
  6. 6.Department of HematologyHiroshima Red Cross and Atomic Bomb Survivor HospitalHiroshima-CityJapan
  7. 7.National Institute of Biomedical InnovationIbarakiJapan
  8. 8.Department of Functional Diagnostic Science, Graduate School of MedicineOsaka UniversitySuitaJapan

Personalised recommendations