Gene expression profiling of leukemic cells and primary thymocytes predicts a signature for apoptotic sensitivity to glucocorticoids

1475-2867-7-18 1475-2867 Primary research Gene expression profiling of leukemic cells and primary thymocytes predicts a signature for apoptotic sensitivity to glucocorticoids Miller L Aaron aamiller@utmb.edu Komak Spogmai spkomak@utmb.edu Webb M Scott mswebb4597@yahoo.com Leiter H Edward ehl@jax.org Thompson E Brad bthompso@utmb.edu

Department of Biochemistry and Molecular Biology, The University of Texas Medical Branch, Galveston, TX, USA

Department of Surgery, The University of Texas Medical Branch, Galveston, TX, USA

Diabetes and Obesity, Inflammatory Bowel Diseases, The Jackson, Laboratory, Bar Harbor, ME, USA

Cancer Cell International 1475-2867 2007 7 1 18 http://www.cancerci.com/content/7/1/18 18045478 10.1186/1475-2867-7-18 18 9 2007 28 11 2007 28 11 2007 2007 Miller et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Abstract

Background

Glucocorticoids (GC's) play an integral role in treatment strategies designed to combat various forms of hematological malignancies. GCs also are powerful inhibitors of the immune system, through regulation of appropriate cytokines and by causing apoptosis of immature thymocytes. By activating the glucocorticoid receptor (GR), GCs evoke apoptosis through transcriptional regulation of a complex, interactive gene network over a period of time preceding activation of the apoptotic enzymes. In this study we used microarray technology to determine whether several disparate types of hematologic cells, all sensitive to GC-evoked apoptosis, would identify a common set of regulated genes. We compared gene expression signatures after treatment with two potent synthetic GCs, dexamethasone (Dex) and cortivazol (CVZ) using a panel of hematologic cells. Pediatric CD4+/CD8+ T-cell leukemia was represented by 3 CEM clones: two sensitive, CEM-C7–14 and CEM-C1–6, and one resistant, CEM-C1–15, to Dex. CEM-C1–15 was also tested when rendered GC-sensitive by several treatments. GC-sensitive pediatric B-cell leukemia was represented by the SUP-B15 line and adult B-cell leukemia by RS4;11 cells. Kasumi-1 cells gave an example of the rare Dex-sensitive acute myeloblastic leukemia (AML). To test the generality of the correlations in malignant cell gene sets, we compared with GC effects on mouse non-transformed thymocytes.

Results

We identified a set of genes regulated by GCs in all GC-sensitive malignant cells. A portion of these were also regulated in the thymocytes. Because we knew that the highly Dex-resistant CEM-C1–15 cells could be killed by CVZ, we tested these cells with the latter steroid and again found that many of the same genes were now regulated as in the inherently GC-sensitive cells. The same result was obtained when we converted the Dex-resistant clone to Dex-sensitive by treatment with forskolin (FSK), to activate the adenyl cyclase/protein kinase A pathway (PKA).

Conclusion

Our results have identified small sets of genes that correlate with GC-sensitivity in cells from several hematologic malignancies. Some of these are also regulated in normal mouse thymocytes.

Background

This study was designed to find whether a common set of GC-regulated genes would be found in cell lines from several types of GC-sensitive malignancies. Several studies of malignant cells sensitive to GC-driven apoptosis identified various candidate genes regulated by GCs 12345. Such regulation precedes the final, irreversible activation of the apoptotic machinery. Though some genes had been detected in more than one study, detailed comparisons are difficult due to differing gene array platforms, bioinformatics methods, and cell conditions used. To avoid these problems we previously compared the genes regulated in a set of closely related clones from the CEM line of CD4, CD8 double-positive pediatric acute lymphoblastic leukemia (ALL) 6, discerning a set of genes regulated specifically in two clones sensitive to Dex-evoked apoptosis (CEM-C7–14 and CEM-C1–6) and not in a closely related Dex-resistant clone (CEM-C1–15). We now carry this work further. Treatment of the resistant clone with the phenylpyrazolo-steroid CVZ, a selective modulator of GR action 7, can provoke cell death 8910. Earlier work had also shown that CEM-C1–15, and its parental clone CEM-C1, can be converted to GC-sensitivity by activating the PKA pathway 1112. We therefore were able to compare the sets of genes regulated in C1–15 cells by Dex, which does not activate apoptosis, with those in the same cells treated with CVZ or converted to Dex-sensitive by activation of PKA. To ask whether overlapping gene sets would be found regulated in other types of malignant GC-sensitive cells, we selected three cell lines already known to be killed by GCs 1314 and derived from other lineages. SUP-B15 is a B-cell, Philadelphia chromosome positive pediatric ALL line grown from bone marrow blasts. SUP-B15 cells express multiple B-cell, but not T-cell lineage markers 15. The RS4;11 line, representing an adult form of ALL, was established from the bone marrow of a 32-year old Caucasian female ALL patient in relapse 16. These cells are characterized by the t(4;11) chromosomal abnormality. Kasumi-1 is an acute myeloblastic leukemia (AML) originally obtained from the peripheral blood of a 7-year old Japanese male 17. The cells contain a t(8;21) chromosomal abnormality which results in the expression of the AML1-ETO fusion protein, shown to bestow sensitivity to both extrinsic and intrinsic apoptotic mechanisms in leukemic cells 18. These myeloid cells possess the highly unusual property for AML of apoptotic sensitivity to GCs 14. As we found that indeed, a certain limited group of genes were GC-regulated in all the cells under conditions leading to GC-dependent apoptosis, we extended our process by comparing the above gene sets with genes regulated by GCs in mouse thymocytes, exquisitely GC-sensitive non-transformed cells.

A smaller set of genes were concordant with those found in the malignant human samples. Given that the experiments were done over a considerable period of time, and with evolving gene chip platforms, it is remarkable that a distinct set of genes were found to be GC-regulated during the pre-apoptotic phase, in cell lines from a variety of human GC-sensitive hematologic malignancies. While surely incomplete, the gene set found to be regulated in every instance could be useful in identifying GC-sensitive hematologic malignancies, if tested in clinical samples. We present it now to be of use to others engaged in examining such samples, noting that GC response is often also a predictor for response to many chemotherapies 1920.

Results

Comparison of time course of GC-dependent cell death in several sub-types of human leukemias

We have shown with CEM cells that although steroids must be continually present, their effects are reversible for many hours preceding the irreversible commitment to apoptosis. To validate their sensitivity of each cell type to GCs and define the limit of the pre-apoptotic, reversible "lag" phase, we treated with Dex or CVZ and followed viable cells over time. In preliminary experiments we determined the GC concentrations that maximized the apoptotic effect for each cell type (Methods). Counts of viable cells from GC-treated samples were compared to their matched vehicle controls (Fig. 1). Though its length varied, each human cell line showed a pre-apoptotic "lag" phase, during GC treatment of many hours, after which increased cell death was observed, except for resistant clone CEM-C1–15. When these were treated with CVZ or converted to Dex-sensitive through PKA stimulation 12, they also showed a lag. SUP-B15 cells were the most rapidly sensitive to GC; at a concentration of 10^-7M, Dex reduced viable cell numbers to 50% of controls by 24 hours. Sensitive clones CEM-C7–14 and CEM-C1–6 (the latter a spontaneous revertant from a resistant clone) required the longest period (40 hours) of exposure to GC before initiation of the apoptotic response. As we had observed previously for CEM-C7–14 cells, treatment with CVZ, a more potent GC with the ability to selectively modulate GR action 7, increased cell death over that by Dex. Both RS4;11 and Kasumi-1 cells reacted similarly in regard to timing, but with differences in sensitivity, as the Dex concentration required for maximum apoptotic potency of Kasumi-1 cells was 10-fold less than for RS4;11. Only a brief meeting report documented the GC-sensitivity of SUP-15 and RS4;11 cells 13. Our results confirm and extend that report.

Figure 1

Glucocorticoids kill multiple sub-types of leukemia

Glucocorticoids kill multiple sub-types of leukemia. Cells were seeded in triplicate at a starting density of 1 × 10⁵cells/ml and subsequently treated with vehicle or various concentrations of Dex (C7–14 and C1–15, symbol, open circle) or CVZ (C7–14 and C1–15, symbol, closed triangle). Dex-resistant CEM-C1–15 cells were restored to sensitivity by treatment with FSK and Dex (symbol, closed square, note: the "lag" time for this drug combination preceded the initial 24 hour count – dashed line 0–24 hours). Samples were analyzed by Trypan blue exclusion assay at various time-points thereafter. Presented is the percent of viable cells compared to matched control for the average of three independent replicates from vehicle vs. GC-treated at the time of evaluation. Error bars represent one standard deviation from the mean.

As is well known, primary mouse thymocytes in vitro rapidly respond to GCs with overt apoptosis 21. We found that under our experimental conditions, apoptosis of thymocytes began after about 1.5 hours of continued exposure in vitro to 10^-7M Dex (data not shown). Together, these data confirm and/or extend prior reports and establish the dose and timing parameters used during the remainder of the studies.

Sequential comparison of gene expression

For gene expression comparisons of all the malignant lines, cells were taken for RNA extraction during the late lag phase of pharmacological treatment, that is before overt apoptosis and at a time that removal of steroid would allow recovery of virtually all cells. This corresponded to the time at which empirical data showed a small fraction of the cells exhibiting phosphatidylserine eversion by annexin-v assay. Since thymocytes in primary culture are not growing, we extracted them after 1.5 hours of exposure to Dex, a time when cell viability did not differ appreciably from controls.

The data from all experiments were analyzed in a concatenated series (Fig. 2). In overview, the sequence was: 1) identification of those genes regulated ≥ 1.2-fold in every CEM cell type under conditions that produced GC-dependent apoptosis; 2) notation (asterisk, Table 1 and Additional file 1) of those genes found similarly regulated in Dex-treated but apoptosis-resistant CEM-C1–15 cells; 3) retention of genes regulated in CEM-C1–15 cells but in the opposite sense from the GC-sensitive cells; 4) application of Spotfire^®statistical software to the regulated genes common to CEM cells under all GC-sensitive conditions. In Additional file 1, all the genes from step 1 are retained, with notation if not unique to GC-sensitive cells, (e.g. seen in CEM-C1–15). This resulted in a "signature" list of genes always regulated in CEM cells under conditions in which they were GC-sensitive (Fig. 2A). Usually, these differences in mRNA levels were found to be statistically significant.

Figure 2

Flow chart of various comparative schemes elucidates multiple GC-mediated gene signatures

Flow chart of various comparative schemes elucidates multiple GC-mediated gene signatures. (A) CEM Signature: All genes regulated in common in the same sense in GC-treated CEM-C7–14, CEM-C1–6, CEM-C1–15 CVZ, and CEM-C1–15 Dex plus FSK cells were obtained (solid line box). This list was subsequently compared to Dex-treated resistant CEM-C1–15 cells and to mouse thymocytes. (B) Multiple Leukemia Signature: All genes regulated in common in the same sense in GC-treated CEM-C7–14, CEM-C1–6, RS4;11, SUP-B15, and Kasumi-1 cells were obtained (dashed line box). As before, this list was compared to Dex-treated resistant CEM-C1–15 cells and to mouse thymocytes. (C) Composite Signature: All genes regulated in common in the same sense in GC-treated CEM-C7–14, CEM-C1–6, RS4;11, SUP-B15, Kasumi-1, CEM-C1–15 CVZ, and CEM-C1–15 Dex plus FSK were compiled into a final list (spotted line box). This list was compared to Dex-treated resistant CEM-C1–15 and to mouse thymocytes.

Table 1

Gene expression signature for GC-sensitivity in CEM cells. (Signaling Network Genes)

Sensitive

Resensitized

Resistant

Name

Description

C7–14 Dx

C7–14 Z

C1–6 Dx

C1–15 Z

C1–15 Dx+F

C1–15 Dx

AK2

adenylate kinase 2

-1.4

-1.7

-1.6

-1.4

ARHGEF7

Rho guanine nucleotide exchange factor (GEF) 7

1.3

1.4

1.3

1.3

1.8

1.2

BCL2L11

BCL2-like 11 (apoptosis facilitator)

3.1

14.5

5.9

1.3

1.8

CALR

calreticulin

-1.9

-1.7

-1.6

-1.4

-1.6

-1.3

CDC6

CDC6 cell division cycle 6 homolog (S. cerevisiae)

-1.3

-1.2

-1.5

-1.3

-1.3

DDIT4

DNA-damage-inducible transcript 4

4.4

3.4

2.5

1.2

4.4

1.5

FGFR1

fibroblast growth factor receptor 1

1.6

3.1

1.4

1.7

1.7

1.3

FKBP4

FK506 binding protein 4, 59 kDa

-1.9

-2.1

-1.3

-1.5

-1.6

FKBP5

FK506 binding protein 5

6.6

11.2

4.4

3.8

2.9

2.8

HDAC1

histone deacetylase 1

-1.4

-1.2

-1.3

-1.2

-1.3

HRAS

v-Ha-ras Harvey rat sarcoma viral oncogene homolog

-1.3

-1.4

-1.3

-1.4

-1.4

ID1

inhibitor of DNA binding 1, dominant negative helix-loop-helix protein

-2.6

-1.7

-1.2

-1.9

-1.6

-1.4

IFRD1

interferon-related developmental regulator 1

-1.8

-1.2

-1.6

-1.6

-1.3

IL7R

interleukin 7 receptor

8.3

18.0

6.3

1.4

2.5

ITGA6

integrin, alpha 6

5.0

10.2

2.4

6.6

2.8

1.6

MAP4

microtubule-associated protein 4

-1.3

-1.8

-1.7

-1.3

MT1A

metallothionein 1A (functional)

1.3

1.3

2.2

1.4

1.4

MTHFD1

methylenetetrahydrofolate dehydrogenase (NADP+ dependent) 1

-1.4

-1.9

-1.5

-1.5

-1.2

MYC

v-myc myelocytomatosis viral oncogene homolog (avian)

-3.8

-4.6

-3.6

-1.4

-2.2

NFKBIA

nuclear factor of kappa light polypeptide gene enhancer in B-cells inhibitor, alpha

3.0

3.4

2.8

1.9

1.5

1.3

NME1

non-metastatic cells 1, protein (NM23A) expressed in

-1.5

-2.2

-1.6

-1.4

-1.3

NR3C1

nuclear receptor subfamily 3, group C, member 1 (glucocorticoid receptor)

4.2

6.6

2.1

1.7

1.6

OGT

O-linked N-acetylglucosamine (GlcNAc) transferase

1.3

1.7

2.0

1.6

1.3

PIAS2

protein inhibitor of activated STAT, 2

1.2

1.4

1.5

2.0

1.3

PIK3R1

phosphoinositide-3-kinase, regulatory subunit 1 (p85 alpha)

2.5

2.4

1.7

2.0

1.9

RBMS1

RNA binding motif, single stranded interacting protein 1

1.4

2.0

1.7

1.6

1.5

1.6

SCARB1

scavenger receptor class B, member 1

-1.7

-2.2

-1.2

-1.5

-1.2

-1.3

SRM

spermidine synthase

-1.8

-3.3

-1.7

-1.4

-1.4

TAGLN2

transgelin 2

-1.4

-2.0

-1.3

-1.2

TRAF3IP2

TRAF3 interacting protein 2

1.6

2.1

1.8

1.5

1.4

1.3

TSC22D3

TSC22 domain family, member 3

33.1

74.0

20.4

7.4

14.3

5.0

TXNIP

thioredoxin interacting protein

2.8

3.4

3.7

1.4

2.0

1.2

UBE2S

ubiquitin-conjugating enzyme E2S

-1.6

-1.6

-1.3

-1.6

-1.3

VCL

vinculin

1.8

1.6

1.9

1.3

1.5

YAF2

YY1 associated factor 2

1.8

2.4

1.6

1.7

1.3

1.6

Regulated transcripts from vehicle or GC-treated samples were evaluated and compared using Spotfire^®and Ingenuity^®bioinformatics software. Selection criteria for each gene were as follows: probe called present (Methods) on the microarray and regulated ≥ 1.2-fold (net 20% change). GC-mediated transcripts regulated in common in the same direction among sensitive CEM-C7–14 Dex (C7–14 Dx) and CVZ (C7–14 Z), CEM-C1–6 (C1–6 Dx), CEM-C1–15 CVZ (C1–15 Z), and CEM-C1–15 Dex plus FSK (C1–15 Dx+F) cells were obtained. This comparison resulted in 96 regulated genes (see Additional file 1). This list was subsequently compared to Dex-treated resistant CEM-C1–15 cells (C1–15 Dx). Shown is a subset of 35 genes from the original 96 that could be linked via a signaling network (symbol, closed diamond in Additional file 1). Bold type indicates statistically significant regulation p ≤ 0.05 between means of vehicle vs. GC-treated. Symbol, asterisk = gene "absent" by selection criteria.

Additional file 1

Gene expression signature for GC-sensitivity in CEM cells. Regulated transcripts from vehicle or GC-treated samples were evaluated and compared using Spotfire^®and Ingenuity^®bioinformatics software. Selection criteria for each gene were as follows: probe called present (Methods) on the microarray and regulated ≥ 1.2-fold (net 20% change). GC-mediated transcripts regulated in common in the same direction among sensitive CEM-C7–14 Dex (C7–14 Dx) and CVZ (C7–14 Z), CEM-C1–6 (C1–6 Dx), CEM-C1–15 CVZ (C1–15 Z), and CEM-C1–15 Dex plus FSK (C1–15 Dx+F) cells were obtained. The resulting 96 regulated genes are depicted. This list was subsequently compared to Dex-treated resistant CEM-C1–15 cells (C1–15 Dx) and to mouse thymocytes (C57/BL6 Dx). Genes linked via an Ingenuity^®generated signaling pathway (Network) are designated with (symbol, closed diamond). Opposite sense regulation (Opp. R) between human and mouse is indicated by (symbol, closed circle). Bold type indicates statistically significant regulation p ≤ 0.05 between means of vehicle vs. GC-treated. Symbol, asterisk = gene "absent" by selection criteria.

Click here for file

Similarly, a "signature" gene set was derived for GC sensitivity for all the leukemic cell lines: 1) compile genes regulated ≥ 1.2-fold in every cell line, without regard to statistical significance; 2) annotate genes Dex-regulated in resistant C1–15 cells; 3) apply statistics. Genes regulated similarly in every GC-sensitive cell type (Fig. 2B, C) comprise a possible "signature" for multiple leukemias. Again, in most cases these were found statistically significant.

Separately, each gene set was compared to the genes regulated by Dex in mouse thymocytes, to see whether some were also present in the pre-apoptotic phase in non-transformed mammalian cells.

Comparison of CEM cell clones reveals a signature for GC-evoked apoptosis

GC-sensitive clones CEM-C7–14 and CEM-C1–6 were treated in triplicate with vehicle alone or 10^-6M Dex. CEM-C7–14 cells were also treated with the more potent GR modulator CVZ. The CEM-C1–15 sister clone is resistant to as much as 4.0^-6Dex but can be killed by CVZ. Therefore these cells were treated with 10^-6M Dex or 10^-6M CVZ. Alternatively, we made use of the fact that CEM-C1–15 cells can be converted to Dex-sensitive by activating the adenyl cyclase/PKA signaling pathway 12 and (Fig. 1). We refined the gene set obtained from the steroids-only treatments by adding the requirement that they were also regulated in C1–15 cells after FSK plus Dex exposure. (See Fig. 2A, solid box.)

Initially, all genes were listed whose mRNAs if "present" as judged by Affymetrix software, were changed 20% or more in the two Dex-treated, Dex-sensitive clones, in one of these treated with 10-fold less CVZ, in the Dex-resistant clone treated with high-dose CVZ, and in the Dex-resistant clone treated with FSK plus Dex. After this list was prepared, the data was evaluated statistically with the exception of the FSK-sensitized C1–15 cell samples, for which only duplicate experiments were performed. This analysis resulted in 96 regulated genes, of which 48 were induced and 48 were repressed (see Additional file 1). Statistical analysis of the means from both control and GC-treated samples were used to generate p-values (bold type in the table). Of the 96 genes, 43 showed a suggestion of weak regulation by Dex in the resistant clone CEM-C1–15, though usually without reaching statistical significance. Of the 43, GSK3B and IL10RB were found to be regulated in opposite directions between Dex-sensitive and -resistant CEM clones. Through the use of Ingenuity^®pathway analysis, a signaling network was revealed, linking 35 of the 96 genes. (Fig. 3, Table 1, the 35 also are designated by, closed diamond in Additional file 1). It is obvious from the network that two major signal transduction nodes exist: 1) affiliation with induction of NR3C1 – the GR; and 2) affiliation with repression of MYC. Minor signaling nodes appear routed through HDAC1, NFKBIA, and PIK3R1. These data suggest that the gene set identified includes genes involved in a relevant signaling pathway, and are not just a random assortment that happen to be GC-regulated. In an independent study of the time course of response in CEM-C7–14 exposed to Dex, we found that the change in these genes was progressive over the interval of steroid exposure (M.S. Webb and E.B. Thompson, manuscript submitted).

Figure 3

A signaling network links genes regulated by GCs in CEM cells

A signaling network links genes regulated by GCs in CEM cells. Ingenuity^®bioinformatics pathway analysis tool was used to connect a subset of 35 genes from the CEM signatory list (Table 1, and symbol, closed diamond in Additional file 1) based upon a database of published observations. Symbols for genes representing specific categories of cellular molecules as well as interactive relationships are depicted in the legend. Color gradations are based upon gene regulation at the fold-change level. Red color: induced gene; green color: repressed gene. Fold-change data from CEM-C7–14 cells treated with Dex are presented as representative of the CEM signature.

To determine which of the conserved GC-regulated genes, if any, were also regulated in a naturally non-transformed sensitive lymphoid cell system, we compared regulation between human CEM cells and mouse thymocytes. The latter are exquisitely sensitive to GCs, undergoing apoptosis within hours of exposure instead of days as for the malignant cells 21. Examination of the conserved gene set yielded 34 transcripts that were regulated in thymocytes also. However, of these, only 20 were regulated in the same sense in the thymocytes; the other 14 were regulated in the opposite sense. (see Additional file 1, symbol, closed circle). These data suggest that there is a shared pathway of GC-regulated genes, relevant to apoptosis, in both malignant and normal mammalian lymphoid cells.

Genes regulated by GCs in multiple sub-types of GC-sensitive leukemias show a common genetic signature

To test for a common signature for GC-sensitivity among various leukemic cells, we compared cell lines representing three types of GC-sensitive leukemias with the CEM cells (Fig. 2B, dashed-line box). From literature reports of GC-sensitivity 1314, we chose SUP-B15 (pediatric B-cell ALL), RS4;11 (an adult B-cell ALL), and Kasumi-1 (a special type of GC-sensitive pediatric AML). As in CEM cells, preliminary experiments established both optimal concentrations of GCs and timing of phosphatidylserine membrane eversion for each cell line to capture gene expression data before irreversible onset of apoptosis. Comparison of GC-sensitive CEMs with RS4;11, SUP-B15, and Kasumi-1 identified 122 genes, 52 induced and 70 repressed, all regulated ≥ 1.2-fold in the same sense in all cell lines. These were then tested for statistical significance (see Additional file 2). Of these genes, 26 were also found to be regulated by Dex in the Dex-resistant CEM-C1–15 cells in the same sense and thus are not of themselves both necessary and sufficient for evoking apoptosis. Four genes, FOXO3A, SFRS2, SLC7A1, and PAI-RBP1 were regulated in a different sense (e.g. induced vs. repressed) between sensitive cells and the resistant CEM-C1–15 cells. As had been done with CEM cells alone, the 122 gene set was compared to the mouse thymocyte data. Thirty-three genes were also found regulated in the thymus; 19 in the same sense as in the GC-sensitive leukemias and 14 in the opposite sense (see Additional file 2 symbol, closed circle).

Additional file 2

Gene expression signature for GC-sensitivity in multiple sub-types of sensitive human leukemias. Messenger RNA was extracted and regulated genes were analyzed as previously stated. A list of GC-mediated transcripts regulated in common in the same sense among sensitive CEM-C7–14 Dex (C7–14 Dx) and CVZ (C7–14 Z), CEM-C1–6 (C1–6 Dx), RS4;11 (RS4 Dx), SUP-B15 (SUP Dx), and Kasumi-1 (Kas Dx) leukemic cells was generated. These comparisons resulted in 122 regulated genes. This list was subsequently compared to Dex-treated resistant CEM-C1–15 cells (C1–15 Dx) and to mouse thymocytes (C57/BL6 Dx). Opposite sense regulation (Opp. R) between human and mouse is indicated by (symbol, closed circle). Bold type indicates statistically significant regulation p ≤ 0.05 between means of vehicle vs. GC-treated. Symbol, asterisk = gene "absent" by selection criteria.

Click here for file

Quantitative real-time PCR confirms GC-regulation of an induced and repressed gene in multiple hematologic malignancies

As proof-of-principle that our approach was valid, we confirmed the regulation of two genes established as important for GC-sensitivity: 1) the pro-apoptotic Bcl-2 family member (BCL2-like 11, BCL2L11/Bim), a GC-induced gene; and 2) ornithine decarboxylase (ODC1), a MYC-dependent, GC-repressed gene 222324 in two cell lines evaluated by microarray (CEM-C7–14 and SUP-B15, see Additional file 2) and the GC-sensitive multiple myeloma cell line OPM-2, which was not evaluated for global gene changes in this array study. Fig. 4 shows the results from real-time PCR assays for BCL2L11 and ODC1 regulation in CEM-C7–14, SUP-B15, and OPM-2 cells treated with Dex for 24 hours. Both genes were confirmed as regulated, one induced and the other repressed in all three cell lines.

Figure 4

Quantitative real-time PCR confirms regulation of BCL2L11/Bim and ODC1 by Dex in multiple hematologic malignancies

Quantitative real-time PCR confirms regulation of BCL2L11/Bim and ODC1 by Dex in multiple hematologic malignancies. CEM-C7–14, SUP-B15, and OPM-2 cells were diluted to 4 × 10⁵cells/ml and treated with ethanol vehicle, 10^-6M Dex (C7–14 and OPM-2), or 10^-7M Dex (SUP-B15) for 24 hours. RNA was extracted and qRT-PCR was performed for BCL2L11/Bim and ODC1 transcripts. The log2 ratio for BCL2L11/Bim and ODC1 are presented. Regulation of both transcripts was determined to be statistically significant p ≤ 0.05 when compared to the internal standard after three independent PCR reactions.

Refinement of the GC-sensitivity signature by comparison of genes regulated in naturally sensitive and resistant-restored-to-sensitive cells

The gene set derived by comparing responses in all the cell clones and lines inherently sensitive to GC-evoked apoptosis (Fig. 2B) was compared to the set regulated by Dex-resistant CEM-C1–15 cells treated with high concentration CVZ (10^-6M) or to Dex in combination with FSK, the activator of the adenyl cyclase/PKA pathway (Fig. 2C, large dotted-line box). Genes regulated in common under all of these conditions were identified as possible significant players in the role of the GC-mediated apoptotic response (Table 2). A total of 27 genes met these criteria, and of these 17 were shown to be induced while 10 were observed as repressed. Nearly all genes were found to be statistically significantly regulated by GC (bold type) where the calculation could be applied. In Dex-resistant CEM-C1–15 cells, 14 of these 27 genes were also seen to be regulated by Dex.

Table 2

Gene expression signature for GC-sensitivity among sensitive and Dex-resistant-restored-to-sensitive leukemias

GC-response

Sen.

Resen.

Res.

Patient-derived line

Ped.

Adult

Ped.

Cell lineage

T-cell

B-cell

Myeloid

T-cell

Sub-type of leukemia

ALL

AML

ALL

Name

Description

C7–14 Dx

C7–14 Z

C1–6 Dx

RS4 Dx

SUP Dx

Kas Dx

C1–15 Z

C1–15 Dx+F

C1–15 Dx

AK2

adenylate kinase 2

-1.4

-1.7

-1.6

-1.4

-2.2

-3.3

-1.6

-1.4

BCL2L11

BCL2-like 11 (apoptosis facilitator)

3.1

14.5

5.9

1.2

3.1

5.1

1.3

1.8

BTG1

B-cell translocation gene 1, anti-proliferative

9.3

9.1

4.4

1.2

3.9

8.8

3.3

1.7

1.9

CD53

CD53 antigen

3.1

3.4

3.2

1.4

3.1

7.2

1.5

1.8

1.4

CDC6

CDC6 cell division cycle 6 homolog (S. cerevisiae)

-1.3

-1.2

-1.5

-1.7

-2.9

-4.7

-1.3

-1.3

DDIT4

DNA-damage-inducible transcript 4

4.4

3.4

2.5

5.5

4.6

22.4

1.2

4.4

1.5

DSCR1

Down syndrome critical region gene 1

5.0

8.2

4.8

1.5

7.7

7.9

1.4

1.6

EIF3S9

eukaryotic translation initiation factor 3, subunit 9 eta

-1.5

-2.2

-1.3

-1.2

-2.0

-2.2

-1.2

FADS1

fatty acid desaturase 1

-1.5

-4.1

-1.9

-1.5

-1.8

-1.9

-1.3

FKBP5

FK506 binding protein 5

6.6

11.2

4.4

7.0

21.4

13.7

3.8

2.9

2.8

IDH3A

isocitrate dehydrogenase 3 (NAD+) alpha

-1.5

-1.3

-1.5

-1.6

-1.7

-2.4

-1.2

-1.3

MARS

methionine-tRNA synthetase

-1.5

-1.6

-1.5

-1.5

-2.2

-1.2

-1.3

MEP50

WD repeat domain 77

-1.7

-1.4

-1.8

-1.8

-1.5

-1.4

-1.5

-1.3

MT1X

metallothionein 1×

1.4

2.1

2.2

1.6

1.8

1.7

1.3

1.4

NFKBIA

NFκB inhibitor, alpha

3.0

3.4

2.8

1.9

2.5

5.5

1.9

1.5

1.3

OGT

O-linked N-acetylglucosamine (GlcNAc) transferase

1.3

1.7

2.0

1.3

2.1

1.6

1.3

PA2G4

proliferation-associated 2G4, 38 kDa

-1.7

-1.9

-1.7

-1.3

-2.5

-3.5

-1.3

-1.2

PARD3

par-3 partitioning defective 3 homolog (C. elegans)

1.6

1.7

1.4

2.5

1.5

3.0

1.3

1.4

PRG1

proteoglycan 1, secretory granule

2.6

4.5

3.2

2.7

1.5

2.7

1.2

1.9

1.6

RAPGEF2

Rap guanine nucleotide exchange factor (GEF) 2

1.7

1.7

1.4

1.8

2.0

4.2

1.3

1.3

RBMS1

RNA binding motif, single stranded interacting protein 1

1.4

2.0

1.7

3.0

4.3

2.3

1.6

1.5

1.6

SCARB1

scavenger receptor class B, member 1

-1.7

-2.2

-1.2

-1.4

-4.9

-33.3

-1.5

-1.2

-1.3

TFPI

tissue factor pathway inhibitor

3.1

5.4

1.9

4.9

9.6

24.6

2.0

1.8

1.7

TSC22D3

TSC22 domain family, member 3

33.1

74.0

20.4

5.2

17.5

20.0

7.4

14.0

5.0

TXNIP

thioredoxin interacting protein

2.8

3.4

3.7

10.0

7.7

2.1

1.4

2.0

1.2

YAF2

YY1 associated factor 2

1.8

2.4

1.6

1.7

1.9

5.8

1.7

1.3

1.6

ZNF259

zinc finger protein 259

-1.4

-1.5

-1.8

-1.2

-1.3

-1.6

-1.3

-1.2

The 122 genes obtained from comparison of both pediatric (Ped.) and adult GC-sensitive (Sen.) leukemias (see Additional file 2) were compared to Dex-resistant, CVZ-sensitive CEM-C1–15 CVZ (C1–15 Z), and resistant-restored-to-sensitive (Resen.) CEM-C1–15 Dex plus FSK (C1–15 Dx+F) regulated transcripts. These comparisons resulted in 27 genes regulated in the same sense among all naturally GC-sensitive and resistant-restored-to-sensitive human leukemias. This list was subsequently compared to Dex-treated resistant (Res.) CEM-C1–15 cells (C1–15 Dx). Bold type indicates statistically significant regulation p ≤ 0.05 between means of vehicle vs. GC-treated. Symbol, asterisk = gene "absent" by selection criteria.

When we compared these 27 genes to Dex-treated thymocytes to obtain those conserved between human and mouse (Table 3) we found 10 genes (8 induced, 2 repressed). Two of the 10 were regulated in the opposite sense in the mouse cells (Table 3, OGT and PA2G4, symbol, closed circle). Six of the 10 genes were also regulated by Dex even in resistant CEM-C1–15 cells. All 10 however, were regulated in CEM-C1–15 cells treated with CVZ or with FSK plus Dex. We conclude that certain genes are regulated by GC in normal mouse thymocytes as they are in several types of GC-sensitive human leukemias. Among the leukemia types tested, we have identified a set of genes that appear to give a partial profile for sensitivity for GC-driven apoptosis. These genes may give clues to the pre-apoptotic pathway. We hope that this GC-sensitive profile encourages application to clinical samples.

Table 3

Conserved gene expression signature for GC-sensitivity in thymocytes and multiple leukemias

GC-response

Sen.

Resen.

Res.

Mouse vs. Human

Mouse

Ped.

Adult

Ped.