Introduction

Many studies of T-cell activation have been performed using either the human leukemic Jurkat cell line (Black et al, 1997 or primary T cells (Cooper and Pellis, 1998; Ellisen et al, 2001). T-cell activation is associated with regulation of multiple intracellular signaling events, mediated by either protein tyrosine kinases and serine/threonine kinases (Rudd, 1999; Tsuchida et al, 1999). Several key processes have been well documented during T-cell activation, including the activation of phospholipase C-γ, which is required for TCR–dependent activation of IL-2 (Lindholm et al, 1999). Up-regulation of protein kinase C, a downstream target of phospholipase C-γ (Cooper and Pellis, 1998; Ghaffari-Tabrizi et al, 1999), results in the stimulation of AP-1 transcriptional activity, which activates the transcription of c-Jun-N-terminal kinase/stress-activated protein kinase (Ghaffari-Tabrizi et al, 1999). Other signaling molecules such as mitogen-activated/extracellular signal-regulated kinase and calcium ions are also involved in T-cell activation (Tsuchida et al, 1999).

Both the leukemic Jurkat T-cell line (Black et al, 1997; Ghaffari-Tabrizi et al, 1999; Tanaka et al, 1997; Whisler et al, 1994) and primary PB T cells (Cooper and Pellis, 1998; Ellisen et al, 2001; Whisler et al, 1993) are commonly used as a model for human T-cell activation. Genomic expression programs in primary PB T-cell activation have been reported (Diehn et al, 2002; Ellisen et al, 2001; Feske et al, 2001); however, no genomic-scale comparison has been performed to analyze the activation-induced gene expression profiles of these two T cell types.

The use of cDNA microarray technology is an effective approach for analysis of gene expression profiles in T-cell activation (Ellisen et al, 2001; Teague et al, 1999). The technology is based on two-color fluorescence hybridization in which two cDNA populations (from test and reference samples) are labeled separately with red or green fluorochrome and are subsequently hybridized to a microarray containing thousands of predeposited cDNAs (Schena et al, 1995). The ratio of fluorescence intensity (red/green) reflects the relative expression level between the tested cDNA populations (such as the test sample and the control sample). The technology has been used in the analysis of gene expression patterns, signaling pathways, and identification of new functional genes on a genomic scale (Chtanova et al, 2001; Ellisen et al, 2001; Teague et al, 1999).

In this study we sought to compare the gene expression profiles of phenotypically purified human CD2+ T cells with those of Jurkat T cells during PHA-induced activation. The results show that general patterns of gene expression profiles were similar during activation of Jurkat T cells and peripheral blood (PB) T cells. However, a subset of genes showed differential expression with varying degrees of gene regulation between the Jurkat and PB T cells. This is the first comparative analysis of transcription profiles on a genomic scale between primary T cells and leukemic Jurkat T cells during T-cell activation (Gonzales and Bowden, 2002; Rowan et al, 1995).

Results and Discussion

Overall Gene Expression Patterns Between Jurkat and PB T Cells During Activation

The baseline similarity of gene expression between Jurkat and PB T cells was evaluated by hybridizing cDNA from resting CD2+ PB T cells with cDNA from nonactivated Jurkat T cells. The correlation coefficient between gene expression profiles of the two T cell types was 0.674 (Fig. 1), indicating that the two T cell types were significantly different. By contrast, cohybridization of PB T-cell mRNA with themselves revealed a correlation coefficient of 0.99 (data not shown).

Figure 1
figure 1

Overall expression profile of 6912 genes between resting peripheral blood (PB) T cells and Jurkat T cells. Red and green data points indicate overexpressed and underexpressed genes, respectively. Genes above the upper blue lines represent those that were overexpressed by greater than 2-fold, whereas genes below the lower blue line represent those that were underexpressed by more than 2-fold.

Full size image

To compare the transcriptional profiles of PB T cells and Jurkat T cells during activation, Cy-5-cDNAs from either activated Jurkat or PB T cells were hybridized with Cy-3-cDNAs from their nonactivated counterparts. Genes that were expressed greater than 2-fold (2-fold higher than control) after activation were considered overexpressed, whereas genes that were expressed less than 0.5 fold (2-fold lower than control) were considered underexpressed. After 24 hours of activation, approximately 300 genes (4.3% of 6912 genes tested) were overexpressed in each T cell type (Table 1). The number of underexpressed genes was much lower than that of the overexpressed genes in both T cell types, with only 79 genes (1.1% of 6912 genes tested) in PB T cells and 37 genes (0.5% of 6912 genes tested) in Jurkat T cells. The ratio of up-regulated genes, expressed sequence tags (ESTs), and unknown genes are also shown in Table 1; the accession numbers of these ESTs and unknown genes are listed in Tables 2 and 3. A significant portion of the up-regulated genes represents ESTs and unknown genes, highlighting the limited knowledge we have regarding the genes involved in T-cell activation. In PB T cells, the majority of the 6912 tested genes were expressed at similar levels (between the upper and lower blue lines in Fig. 1) during activation. This result is consistent with other reports that resting T cells are not in a quiescent stage; instead, there are many cellular transcriptional activities that are needed to maintain viability and mobility of unstimulated T cells (Boise et al, 1995; Teague et al, 1999; Vella et al, 1998). Jurkat T cells are derived from a T-precursor lymphoblastic lymphoma (Black et al, 1997). As a result, many of the genes were expected to be constitutively expressed before activation treatment; however, Jurkat T cells responded to activation treatment similar to resting T cells, as reflected by the similar numbers of up-regulated genes in both T cell types (Table 1).

Table 1 Numbers of Differentially Expressed Genes During Activation of Jurkat and PB T Cellsa
Full size table
Table 2 GenBank Accession Numbers of Overexpressed ESTs and Unknown Genes (≥2-fold) During PB T-Cell Activationa
Full size table
Table 3 GenBank Accession Numbers of Overexpressed ESTs and Unknown Genes (≥2-fold) During Jurkat T-Cell Activationa
Full size table

Functional Groups of Similarly Expressed Genes in Both Jurkat and PB T Cells During Activation

Analysis of the 300 genes that were overexpressed in each cell type revealed that many were unknown genes and ESTs. This suggests the existence of many novel, as yet uncharacterized, genes that are involved in T-cell activation. Activation of T cells leads to expression of multiple genes that are involved in cellular events such as proliferation, cytoskeletal construction, apoptosis, and secretion of cytokines (Ellisen et al, 2001; Teague et al, 1999). To determine the functional groups of genes that were commonly up- or down-regulated in T-cell activation, we examined the identities of 93 overexpressed and 15 underexpressed genes in both Jurkat and PB T cells. Of the genes that were up- and down-regulated by 1.5 fold or more, the genes with known functions were grouped together (Tables 4 and 5). Among the known up-regulated genes (Table 4), several genes involved in cellular proliferation, cell adhesion, and cell cycle progression were identified, including CDC kinases, integrin, and Rho-activating protein-encoding genes (Klekotka et al, 2001; Lew and Reed, 1993; Welsh and Assoian, 2000). Genes encoding for protein synthesis machinery (eukaryotic translational termination factor 1, ribosomal proteins L7a, L31, L21), cytoskeletal elements (myosin, actin, tubulin, calponin), and cell surface antigens (CD68, CD36, CD47) were also commonly up-regulated in both cell types during activation. Interestingly, a group of KIAA genes encoding proteins with potential signaling function, involved in cell migration (Chen et al, 2002) and metastasis (Lee et al, 2002), were also up-regulated upon activation.

Table 4 Functional Groups of Genes That Were Up-Regulated in Both Jurkat and PB T Cells During Activationa
Full size table
Table 5 Functional Groups of Genes That Were Down-Regulated During Activation of Both Jurkat and PB T Cellsa
Full size table

The down-regulated genes in both activated Jurkat and PB T cells consisted mostly of genes encoding CDK inhibitors, proapoptotic caspases, and transcription factors C-Fos and Jun-B (Table 5). Down-regulation of CDK inhibitors is critical for inhibition of cell cycle progression. The stimulation of cell proliferation via PHA activation has been shown to induce the down-regulation of p27Kip and p21Waf CDK inhibitors, resulting in the release of cells into G2/S phase of the cell cycle (Polyak et al, 1994; Sawada et al, 1996). The fact that some cytokines such as IL-4 (Boise et al, 1995; Vella et al, 1998) can only induce activated T cells (but not resting T cells) to proliferate also supports these data.

Caspases are a group of proteases that are generally associated with apoptosis. Activation of several caspases including caspase-1, -3, -8, and -9 has been reported to induce T-cell apoptosis (MacFarlane et al, 2000; Morley et al, 2000; Ruiz et al, 2001; Takahashi et al, 2001; Varghese et al, 2001). Our data showed that several caspases (caspase-1, -3, -4, -7, -8, -9) were down-regulated in both Jurkat and PB T cells after 24 hours of activation (Table 5). This occurred during the period of T-cell proliferation in which cell numbers increased by 3-fold (data not shown). This is consistent with a previous report demonstrating that the activation of caspase 1 and 3 in CD4+ and CD8+ T cells was time dependent. The maximum activities of caspase-1 and -3 were observed after the proliferative period, at Days 7 and 10 (Ruiz et al, 2001).

The proto-oncogenes jun-B and c-fos are generally involved in proliferation-associated transcriptional activation (Sylvester et al, 1998; Wang et al, 1997); thus, our results of their down-regulation were unexpected. Our results are, however, in keeping with those of Ellisen et al (2001) who observed down-regulation of jun-B and c-fos in ConA-treated PB T cells. In our experiments, jun-B and c-fos were down-regulated in both Jurkat and PB T cells during activation; this is consistent with other reports (Kreuzer et al, 1997; Passegue and Wagner, 2000) that C-Fos and Jun-B may be involved in the transcriptional activation of antiproliferative genes such as p16INK4a.

Diehn et al (2002) performed a study of gene expression profiles during PB T-cell activation. To compare our results with theirs, we used their web-based GeneXplorer software and Figure 1 (available at http://genome-www.stanford.edu/costimulation) to search for the expression levels of 22 randomly selected genes from our list of genes that were up- or down-regulated during T-cell activation. The expression levels of these genes showed a high level of concordance between the two studies (18 of 22). Among the consistently expressed genes, there were proliferation-related genes (four), CDK inhibitors (four), caspases (three), and transcription factors (two).

Differences in Gene Expression Between Jurkat and PB T Cells During Activation

Interestingly, many activation-induced genes in PB T cells were constitutively overexpressed to similar levels in the non-activated Jurkat T cells (Fig. 2, A and C). Of the 300 overexpressed (by 2-fold) genes and ESTs, more than one-third (118 genes and ESTs) were expressed at similar levels in nonactivated Jurkat cells (Fig. 2A, Fig. 3). Although the vast majority of genes were similarly expressed between Jurkat and PB T cells during activation, one of the most significant differences between the expression profiles of the two cell types was the total number of genes that were expressed greater than 4-fold after activation (Table 1). Although 71 genes and ESTs were expressed greater than 4-fold in PB T cells, only three genes were overexpressed to this degree in Jurkat T cells. Among the ~300 2-fold up-regulated genes during T-cell activation, there were only 93 genes (31%) that were common between the two T cell types (Fig. 3).

Figure 2
figure 2

Distribution of genes overexpressed greater than 2-fold during activation. A, Of the 6912 genes tested, 300 were overexpressed greater than 2-fold in PB T cells (a). Among these 300 overexpressed genes, 93 genes were also overexpressed in Jurkat T cells during activation (b), and 118 genes were expressed more than 2-fold higher in nonactivated Jurkat than in resting PB T cells (c). B, Profile of 301 genes overexpressed more than 2-fold in Jurkat T cells (a) during activation, in PB T cells (b) during activation, and in resting PB T cells compared with nonactivated Jurkat T cells (c). C, Profile of 71 genes up-regulated by 4-fold in PB T cells (a) during activation, in Jurkat T cells (b) during activation, and in resting PB T cells compared with nonactivated Jurkat T cells (c). Red points indicate overexpressed genes and green points represent underexpressed genes. Genes above the upper blue line were overexpressed by more than 2-fold, whereas genes below the lower blue line were underexpressed by more than 2-fold.

Full size image
Figure 3
figure 3

A Venn diagram showing the overlaps of up-regulated genes during T-cell activation. Among the 300 genes that were up-regulated more than 2-fold (left circle) during PB T-cell activation, 93 of them were also overexpressed more than 2-fold during activation of Jurkat T cells (right circle). Approximately 118 genes (bottom circle) up-regulated during PB T cell activation were constitutively up-regulated in nonstimulated Jurkat T cells compared with resting PB T cells.

Full size image

Among the 71 genes and ESTs overexpressed by 4-fold or higher in activated PB T cells, more than half were constitutively overexpressed in nontreated Jurkat cells (Fig. 2C). These data suggest that when using Jurkat cells as model of T-cell activation, the number of activated genes may be underrepresented because many are constitutively overexpressed. The constitutively overexpressed genes (greater than 2-fold) in Jurkat cells (Table 6) most likely reflect the activated status of these leukemic T cells and probably are important in the pathogenesis of the neoplasm. Similar observations have been made when expression of phosphoinositide (PI)3-kinase was compared between PB T cells and Jurkat cells (Astoul et al, 2001). The finding that many of the constitutively overexpressed genes in Jurkat cells were also overexpressed in activated PB T cells suggests that genes induced during activation of T cells may be involved in leukemogenesis. Indeed, growth factors such as transforming growth factor-β and TNF activate several signaling pathways including the PI3-kinase pathway (Kim et al, 2002) and the nuclear factor-κB pathway (Natoli et al, 1998) to ultimately induce cell survival and inhibit apoptosis. The expression of transcription factors such as jun and fos are also activated by PI3-kinase and AKT (Duan et al, 2002; Walsh et al, 2002) and contribute to T-cell survival. Cross-linking of cell surface proteins such as CDW52 (CAMPATH-1 antigen), a glycopeptide antigen expressed on T and B lymphocytes, by anti-CD52 antibody mediates effective growth inhibition, supporting its important role in T-cell transformation (Rowan et al, 1998). Another cell surface antigen, endoglin (a recently recognized component of the transforming growth factor-β receptor complex), is a strong marker of neovascularization and may thus play a role in Jurkat T-cell leukemogenesis (Calabro et al, 2003). Furthermore, several other genes in this list (Table 6) may be influenced by the persistent activation of the PI3-kinase pathway. For example, matrix metalloproteinase secretion in some cell types occurs via the PI3 kinase-AKT pathway (Ellerbroek et al, 2001). Matrix metalloproteinases are critical for cell adhesive and invasive properties of malignant cells including leukemic T cells (Stetler-Stevenson et al, 1997) and may contribute to the neoplastic phenotype of Jurkat T cells. Further functional studies of these differentially expressed genes may lead to identification of novel leukemia-specific genes.

Table 6 Constitutively Overexpressed Genes and ESTs in Nontreated Jurkat T cellsa in Comparison to Nonactivated PB T Cells
Full size table

Validation of Gene Expression by Real-Time RT-PCR

To validate the gene expression patterns observed in our microarray experiments, we used the same source of mRNAs to perform quantitative fluorescent real-time RT-PCR (Elenitoba-Johnson et al, 2002). The RT-PCR quantitation of nine selected gene transcripts (Table 7) showed that the trend of gene expression during activation of both Jurkat and PB T cells was consistent between the microarray and RT-PCR data. The overall correlation coefficient between relative expression levels from the microarray analysis and from RT-PCR was 0.81. Using this methodology, our previous validation of microarray data has shown approximately 70% concordance (Fillmore et al, 2002; Robetorye et al, 2002).

Table 7 Validation of Microarray Gene Expression Data by Real-Time RT-PCR
Full size table

In summary, comparative microarray studies of PHA-induced activation of PB T cells and Jurkat cells reveal that genes associated with proliferation, protein synthesis, cytoskeletal construction, and signaling were similarly up-regulated, whereas genes associated with cell cycle inhibition and apoptosis were down-regulated at similar levels. A subset of genes induced during PB T-cell activation, however, were constitutively expressed at high levels in nonactivated Jurkat cells, suggesting a role for these genes in the neoplastic transformation and/or biology of T-ALL. Finally, with the exception of a small subset of genes, the high degree of overlap between the gene expression profiles of the two cell types suggests that Jurkat T cells remain a suitable model for T-cell activation.

Materials and Methods

Isolation of Human PB T Cells

Fresh PB was obtained from normal donors. To isolate T cells, buffy coat samples were collected from blood using Ficoll-Paque (Amersham Pharmacia, Piscataway, New Jersey), following the manufacturer’s instructions. CD2+ T cells were then isolated by Dynal beads coated with CD2-specific mAbs (Dynal CD2 CELLection Kit; DYNAL Biotech, Oslo, Norway), following the manufacturer’s protocol.

T-Cell Activation

Both CD2+ T cells and leukemic Jurkat T cells were cultured in RPMI-1640 (Life Technologies, Rockville, Maryland) containing 10% fetal bovine serum and activated in the same medium supplemented with 1% (v/v) PHA (Life Technologies, Carlsbad, California) for 24 hours at 37° C in 5% CO2.

Separation of Total RNA and mRNA

Total RNA was extracted by using the TRIZol reagent (Life Technologies, Carlsbad, California). Messenger RNA was purified from total RNA using poly(T)-coated beads (Oligotex mRNA Isolation Kit; Qiagen, Valencia, California), according to the manufacturer’s protocol.

Preparation and Hybridization of Fluorescent-Labeled cDNA

For each comparative array hybridization, fluorescent cDNA was synthesized by reverse transcription from a test sample (activated T-cell mRNA or activated Jurkat mRNA) in the presence of Cy5-dUTP and from a reference sample (nonactivated T-cell mRNA or Jurkat mRNA) with Cy3-dUTP, using the Superscript II RT kit (Life Technologies, Carlsbad, California) with several modifications. The RT reaction mixture contained 0.5 mm dATP, dCTP, and dGTP, 0.2 mm dTTP, and 0.1 mm Cy5- or Cy3-dUTP (Amersham Pharmacia). The labeling reaction was stopped by the addition of 5 μl of 500 mm EDTA. The mRNA was hydrolyzed by adding 10 μl of 1 N NaOH and heating to 65° C for 1 hour. The resulting alkaline solution was neutralized with 1 m Tris-HCl (pH 7.5). Unincorporated nucleotides were removed using a Biospin 6 column (Bio-Rad Laboratories, Hercules, California). The two fluorescently labeled cDNA samples were mixed, concentrated using a Microcon 30 microconcentrator (Amicon Inc., Beverly, Massachusetts), and hybridized to an arrayed slide (Research Genetics, Huntsville, Alabama) containing 6912 clones as described previously (Manos and Jones, 2001). Hybridization patterns were captured electronically using a two-color confocal laser microscope (Molecular Dynamics, Sunnyvale, California).

Microarray Experimental Design and Data Analysis

To directly analyze the changes in transcriptional profile induced by activation, mRNA from PHA-stimulated cells were hybridized directly with their nonactivated counterpart. As a control, Cy5-labeled cDNA from CD2+ T cells was also hybridized with Cy3-cDNA from Jurkat cells. Each hybridization was performed in duplicate and repeated twice. Microarray analysis was performed using GeneSpring 4.0.4 (Silicon Genetics, Redwood City, California). Relative levels of transcripts between test and reference samples were shown as ratio of Cy5 signal (red fluorescence) versus Cy3 (green fluorescence) signal. Two-fold overexpression of a gene was indicated when the Cy5 signal intensity from the test sample was twice that of Cy3 (from the reference sample), and 2-fold underexpression was indicated by a Cy3 signal intensity twice as high as that of Cy5. To ensure that the reported fold changes in gene expression were significant, p < 0.05 was used as a cut-off value so that only those fold changes with p values < 0.05 were presented.

Quantitative Fluorescent Real-Time RT-PCR

RNA was reverse transcribed using the First-Strand cDNA Kit (Life Technologies, Carlsbad, California). After cDNA synthesis, 1 μl of cDNA template was used for each PCR reaction. The forward and reverse primers used for PCR reactions were as follows: 5′-CGACCACTTTGTCAAGCTCA-3′ (forward, GAPDH), 5′-AGGGGAGATTCAGTGTGGTG-3′ (reverse, GAPDH); 5′-CAAGGGTCTTGGGTCCTA-3′ (forward, RPS 14), 5′-TGCCCTTGTGAACTGGTG-3′ (reverse, RPS 14); 5′-GCACGACAACATCTCATTGG-3′ (forward, NFKB1), 5′-TCCCAAGAGTCATCCAGGTC-3′ (reverse, NFKB1); and 5′-TCTCTCAAGCTCGCCTCTTC-3′ (forward, Jun B), 5′-ACGTGGTTCATCTTGTGCAG-3′ (reverse, JunB). Real-time PCR was conducted using a LightCycler DNA Master SYBR Green I Kit (Roche Molecular Biochemicals, Indianapolis, Indiana), following the manufacturer’s instructions with the following thermal cycling conditions: 95° C for 30 seconds, followed by 45 cycles of 95° C for 0 seconds, 55° C for 5 seconds, and 72° C for 10 seconds. Each gene was normalized to a housekeeping gene (GAPDH) before fold change was calculated based on the signal intensities at the midexponential stages of the real-time amplification curves.