Transcription Profile of Cells Infected with Human T-cell Leukemia Virus Type I Compared with Activated Lymphocytes¹

HTLV-I³ is the etiologic agent of an aggressive and fatal disease termed ATL and of the neurodegenerative disease tropical spastic paraparesis/HTLV-I-associated myelopathy (1, 2, 3, 4). HTLV-I has also been less closely associated with uveitis, arthritis, infectious dermatitis, and immunosuppression (5, 6, 7). The principle target for HTLV-I infection in the lymphoid system are mature CD4+ CD45RO+ T lymphocytes, although other cell types of lymphoid origin have been infected by HTLV-I in vitro, including CD8+ T cells, B cells, and macrophages (8, 9, 10, 11).

The mechanism of oncogenic transformation of host T lymphocytes in ATL remains unclear, and to date there is no effective treatment for this disease. However, as with other cancers, altered gene expression of networks of genes are linked to ATL initiation and progression. Several studies (8, 9, 10, 11) have established that the viral transcriptional activator protein Tax plays a critical role in cellular transformation. Tax not only activates expression of viral genes via the viral LTR, but has also been reported to affect the expression or activity of several cellular genes. Several of these genes encode proteins involved in cell growth and cell death including proto-oncogenes, growth factors and their receptors, CDKs, and CDK inhibitors (8, 9, 10, 11).

In more recent studies, a role for the HTLV-I accessory proteins, p12, p30, and p13 in gene activation and cell signaling have been demonstrated. All three proteins have been shown to play a role in vivo for viral infectivity and replication (12, 13, 14, 15, 16). In addition, the p12 protein has been implicated in the MHC class I-mediated immune response and T-cell signaling (16, 17, 18). The viral protein p13 has also recently been reported to regulate cellular signaling pathways (19). Finally, the p30 protein is reported to play a role in transcriptional regulation with the use of Gal4-p30 fusion constructs (20).

DNA microarray technology has facilitated the development of a more complete and inclusive analysis of gene expression profiles in response to many stimuli for a variety of biological systems. Harhaj et al. (21) and de La Fuente et al. (22) used Atlas human cDNA arrays to analyze gene expression patterns in HTLV-I-infected PBMCs compared with uninfected PBMCs or HTLV-I-transformed C8166 cells compared with nonvirally transformed CEM cells, respectively. Ng et al. (23) have used NIH OncoChip cDNA arrays to analyze gene expression patterns of ∼2000 human genes in Tax-expressing Jurkat T lymphocytes. These studies have looked at small numbers of samples and identified a few candidate genes important for HTLV-I-induced pathogenesis. We have chosen GeneChip microarrays (Affymetrix, Inc.), containing oligonucleotide hybridization probes representative of >7000 genes, to perform a more comprehensive examination of the expression profiles for HTLV-I-immortalized and -transformed cell lines and compared these with the expression profile of normal activated PBLs. The results presented here extend earlier studies by identifying a significant number of new genes that have altered expression in HTLV-I-transformed cells compared with activated PBLs. We have identified several new response pathways involving G₂/M checkpoint control factors, DNA replication and licensing factors, transcriptional regulators, and kinase/phosphatase signaling molecules that are deregulated in HTLV-I-infected cells. Moreover, we found that by analyzing several HTLV-I cell lines, gene expression changes attributable to individual cell types were decreased.

We isolated PBMCs from healthy, HTLV-I-negative donors using Ficoll density gradient centrifugation. After removal of macrophages, cells were stimulated for 18 h with PHA (2 μg/ml) and then grown in RPMI 1640 medium supplemented with 10% fetal bovine serum, 100 units/ml penicillin-streptomycin, 2 mm glutamine, and 50 units/ml IL-2. Activated cells were expanded for 2 weeks and then total RNA was isolated. Resting cells were expanded for 2 weeks, washed twice in culture medium not containing PHA and IL-2, and total RNA was isolated 3 days later. HTLV-I-immortalized cells, Champ (ATL), Bes (ATL), and ACH.WT, were grown in complete medium supplemented with 50 units/ml IL-2. HTLV-I-transformed cells, C81 and Hut102, were grown in complete medium.

mRNA was isolated from total RNA by the RNeasy and Oligotex mRNA isolation procedures as outlined by the manufacturer (Qiagen). Experimental procedures for GeneChip were performed according to the Affymetrix GeneChip Expression Analysis Technical Manual (Affymetrix). Briefly, double-stranded cDNA was synthesized from mRNA with the SuperScript Choice system (Life Technologies, Inc.) and a T7-(dT) 24 (GENSET) primer. In vitro transcription was performed on the cDNA to produce biotin-labeled cRNA with an Enzo Transcription Kit (Enzo) as described by the manufacturer. The cRNA was linearly amplified with T7 polymerase, the biotinylated cRNA was cleaned with an RNeasy Mini Kit (Qiagen), fragmented to 50–200 nucleotides, and then hybridized to Affymetrix Hu6800 arrays. The arrays were then processed on the Affymetrix fluidics station and scanned on an HP GeneArray scanner. The intensity for each probe set of the array was captured with Affymetrix GeneChip Software, according to standard Affymetrix procedures. To determine the quantitative RNA level, the average differences representing the perfectly matched minus the mismatched for each gene-specific probe set was calculated. The GeneSpring Software (SiliconGenetics) was also used to examine the differential gene expression.

Total cellular RNA was isolated with Rnazol B (Tel-Test) as described by the manufacturer. RNA (5 μg) was converted to cDNA with RETROscript (Ambion) as described by the manufacturer. Samples were then PCR amplified with SuperTaq Plus polymerase (Ambion) to quantitate individual genes. The PCR primers for Tax were as follows: 5′-TGTTTGGAGACTGTGTACAAGGCG-3′ and 5′-CAGGCTGTCAGCGTGACGG-3′. The primers for IL-2Rα were as follows: 5′-GGTCCCAGGCAGAGAATCATA-3′ and 5′-AGAGGGAGAAGGGATGGAGGT-3′. PCR products were separated by agarose gel electrophoresis and quantitated with Fluorchem (Imgen Technologies).

Jurkat T cells were transiently transfected by electroporation with 4 μg of the Cdc25C promoter construct (C290-, C75-, CC51-, and C75M1-Luc) in the absence or presence of Tax (8 μg) as described previously (24). Luciferase activity was measured on a Berthold Luminometer and normalized to protein concentration.

To investigate the molecular basis of the HTLV-I-induced T-cell immortalization and transformation, we compared the gene expression profiles of normal uninfected PBLs with HTLV-I-infected cells. PBMCs from three healthy, HTLV-I-negative donors were cultured in the presence of IL-2 and PHA to allow the outgrowth of T lymphocytes. After 1 week of culture, the resulting cell population consisted of >95% CD4+ cells as determined by fluorescence-activated cell sorting analysis (data not shown). mRNA was isolated from either activated or resting PBLs, HTLV-I-immortalized cell lines, Champ, Bes, and ACH.WT (IL-2 dependent), and from the HTLV-I-transformed cells, C81 and Hut102 (IL-2-independent). mRNA samples were then processed to produce biotin-labeled cRNA probes, which were hybridized to the Affymetrix Hu6800 GeneChip as described in “Materials and Methods.” Arrays were processed with the Affymetrix fluidics station and scanner. Expression analysis files were generated for each sample, and a database was created.

Using GeneSpring Software, we performed a “cluster” or “tree” analysis of the cell gene-expression profiles. A hierarchical clustering (Fig. 1), which allowed visualization of a set of samples or genes by organizing them into a mock-phylogenetic tree, was performed for individual genes (across the top) and for each sample (down the side). In this tree, genes (or samples) having similar gene-expression patterns are clustered together. How far across the tree one goes until a subtree is found containing both genes (samples) is a measure of how closely correlated those genes (samples) are to each other. As expected, the gene-expression profile of the activated PBLs was most closely related to their unactivated counterparts (Fig. 1, lines A and B) and was distinct from the HTLV-I-infected cells (Fig. 1, lines C–G). It is interesting to note that the overall gene-expression profile of two HTLV-I-transformed cells lines, HUT102 and C81, fell into a closely related gene-expression cluster. This suggests that they are more closely related to each other than they are to the HTLV-I-immortalized cells. The HTLV-I-immortalized cell lines, Bes, obtained from the PBL culture of this ATL patient, and ACH.WT, lymphocytes immortalized with an infectious clone of HTLV-I, fell into a similar cluster. Interestingly, the HTLV-I-immortalized cell line Champ, also obtained through PBL culture of an ATL patient, fell into a separate cluster distinct from the HTLV-I-transformed cells (C81 and Hut102) and the HTLV-I-immortalized cells (Bes and ACH.WT). The shading across each sample/bar is an indication of the gene-expression levels for each of the 7300 genes. For each cell sample, a unique expression pattern emerges (Fig. 1). A section of the tree has been expanded to show an example of how individual genes performed. As shown, the genes GATA-2, PON2, and PLAGL2 are all overexpressed in the HTLV-I-infected cells as compared with the PBL samples. This form of clustering allows one to identify groups of genes that are expressed either similarly or opposite to the control sample.

From the expression data, we compiled a list of genes deregulated an average of 2-fold or greater in at least three of five HTLV-I-infected cell lines, as compared with activated PBLs (Table 1 and on-line data supplement¹). A 2-fold cutoff was chosen based on statistical information provided by Affymetrix. From this initial list, the genes were broken down into functional groups. Because the Affymetrix array contains a much larger number of genes than previous analyses, our results provide a more complete and inclusive “blueprint” of HTLV-I-induced gene expression. Because of size limitations, the list of 763 genes is provided in the on-line data supplement. Only 84 of the 763 differentially expressed genes have been reported previously (21, 23, 25). These genes are shaded in the on-line data supplement table and the results of individual studies from other laboratories are noted. The array analysis by Ng et al. (23) could not be included because the GenBank accession numbers were not available for this study. The results of the Affymetrix gene-expression analysis of several independent genes have been confirmed by quantitative RT-PCR (see below and data not shown).

One of the hallmark genes known to be overexpressed in HTLV-I-infected cells is the cytokine receptor IL-2Rα. IL-2Rα plays a pivotal role in the ability of HTLV-I-transformed cells to proliferate (26, 27, 28, 29). Tax has been demonstrated to activate expression of IL-2Rα through activation of the nuclear factor κB pathway (29). Not unexpectedly therefore, we found that IL-2Rα expression was elevated in each of the HTLV-I-infected cells (Table 1). Quantitative analysis of microarray hybridization data demonstrated that the level of IL-2Rα expression was increased 7–38-fold in the HTLV-I samples as compared with the activated PBLs (Table 1). The increased expression of IL-2Rα mRNA in the HTLV-I-transformed cells was confirmed by quantitative PCR (Fig. 2, B and C). Interestingly, the level of Tax protein as detected by Western blot analysis (Fig. 2 A) did not correlate completely with the level of IL-2Rα mRNA expression.

Consistent with recent observations of Mariner et al. (30), we observed that the level of IL-15Rα was elevated 5–10-fold in the HTLV-I-infected cell samples. The level of IL-15Rα expression was elevated in four of the various HTLV-I-infected cell samples (Table 1). Of note, the level of IL-15Rα expression in the C81 HTLV-I-transformed T-cell line was as low as was seen in the activated lymphocyte cells. Thus, the level of IL-15Rα expression in different HTLV-I-infected cells may vary significantly. These results are consistent with RNA and protein levels reported previously (30). Of the microarray reports, ours is the first to demonstrate differential IL-15Rα expression. To note, the IL-2Rβ chain common to both IL-15Rα and IL-2Rα signaling is also increased in the HTLV-I-infected cells except in the C81 sample (Table 1).

In agreement with the proposed IL-15 paracrine/autocrine loop (30, 31), we observed an increase in the level of IL-15 mRNA in four of five HTLV-I-transformed cells. These results are consistent with the work of Azimi et al. (31, 32) who demonstrated that the IL-15 promoter is transactivated by Tax and the level of IL-15 mRNA is increased 3–4-fold in HTLV-I-transformed cells lines. C81 cells appear to be an exception because the level of IL-15 mRNA expression is very low in this cell line.

In addition to the overexpression of cytokines including SCYA1, SCYA17, SCYA22, and P40 T-cell growth factor, there were also several signal pathway gene products that had increased expression (Table 1). We have previously shown that Tax activates expression of parathyroid hormone-releasing protein (PTHrP; see Ref. 33). In our current analysis, we see a 2–5-fold increase in PTHrP gene expression (Table 1). Our microarray analysis shows that TNFα (GenBank accession no. M16441) expression is elevated in C81, HUT102, and ACH-WT lymphocytes, which are immortalized with an infectious clone of HTLV-I. In contrast, very low levels of TNFα expression were observed in the Bes and Champ RNA samples.

We also observed that the expression level of gp34 was elevated in the HTLV-I-infected cells. This type II membrane protein belongs to the TNF superfamily (TNFSF4) and has been shown to stimulate T-cell proliferation and cytokine production. At slightly lower levels, we observed the elevated expression of TNFSF7 (CD27L), which also plays a role in T-cell activation. We saw an increase in the cell surface molecules CD58 and CD59 (Table 1). Many cell-surface signaling molecules were also decreased in expression in the HTLV-I-infected cells as compared with activated PBLs. These include integrin molecules (ITGAM, ITGAE, and CD11A), the adhesion molecule PECAM1, CD47, CDW52, CD37, CD27, CD7, CD20, CD16, and CD72 (Table 1); all of which play a role in lymphocyte signaling.

Phosphorylation has been shown to play a key role in regulating protein activity and cellular responses. Several kinases were overexpressed in the HTLV-I-transformed cells including mitogen-activated protein kinase family members PRKM7 and mitogen-activated protein kinase 3 (MARK3); tyrosine kinase Lyn; cell cycle regulatory kinases CDK4, CDK7, and CDK2; and CK1ε. The overexpression of CK1ε is of interest because we have recently demonstrated that the amino terminus of p53 is hyperphosphorylated at serine 15 and one other amino acid within amino acids 1–19 (34, 35). Sakaguchi et al. (36) and Dumaz et al. (37) have shown that threonine 18 can be phosphorylated by CK1ε. Interestingly, phosphorylation of threonine 18 is dependent on prior phosphorylation of serine 15 (37). The overexpression of Lyn is consistent with previous reports that Tax transactivates the promoter of this src family gene (38).

Recently, Ng et al. (23) reported increased levels of the MLK3 using cDNA arrays to study Tax-regulated genes in transformed Jurkat T lymphocytes. The studies by de La Fuente et al. (22) comparing a Tax/HTLV-I-expressing cell line (C81) with a transformed T-cell line (CEM), also noted an increase in MLK3. In contrast, our results, like those of Harhaj et al. (21) using Atlas (Clontech) filter arrays and Ruckes et al. (25) using subtractive hybridization, do not show differential expression of MLK3 in HTLV-I-infected cells compared with activated donor PBLs. The levels of MLK3 expression we observed are highlighted in Table 1. One explanation for the difference is the choice of control cells. The first studies use transformed cells, whereas our studies and those of Harhaj et al. (21) and Ruckes et al. (25) compared the HTLV-I-infected cells with activated PBLs. Perhaps nonviral transformed T-cells have a down-regulation of MLK3 and thus the Tax-expressing cells artificially appear to have an increase in MLK3.

Our studies are the first to show that the dual specific phosphatases DUSP2, DUSP4, and DUSP5 are increased ∼10-fold in the HTLV-I-infected cells as compared with the activated PBLs. These phosphatases have been implicated in the mitogenic signaling pathways involving erk1 and erk2 (39, 40, 41) and may play an important role in regulating signaling in the HTLV-I-infected cells.

Likewise, INPP1 is increased ∼10-fold in the HTLV-I-infected cells over activated PBLs. Inositol signaling is an important component of cellular processes including proliferation, differentiation, and apoptosis (42, 43, 44). INPP1 acts on both Ins(1,4)P₂ and Ins(1,3,4)P₄, which are key intermediary metabolites in several pathways such as DNA replication and cell cycle progression (42, 43, 44). INPP1 overexpression has recently been correlated with human colorectal cancer (45).

The HTLV-I Tax protein functions as a transcriptional transactivator of the viral LTR and several cellular promoters. Transcriptional activation by Tax is attributable, in part, to protein-to-protein interactions leading to enhanced DNA binding or transcriptional activity or increased nuclear accumulation of active transcription factors, causing activation of CREB, nuclear factor κB, and SRF. Tax has also been shown to stabilize the indirect binding of coactivator proteins such as CBP to the DNA, leading to an increase in transcription initiation and reinitiation. Our microarray data revealed the increased expression of ∼32 transcription regulators in the HTLV-I cells. Not all of these genes are likely to be directly activated by Tax, but may represent “secondary” effects of HTLV-I infection because of changes in cell metabolism and cell proliferation.

In recent reports, Harhaj et al. (21) and Ruckes et al. (25) analyzed the regulation of transcription factors in HTLV-I positive cells. Transcription factors of the Jun family including junB, Jun, and JUND, B94, Ets2, CDK7 (TFIIH), GTF2A1, Rel, and GATA-2 were overexpressed in the range of 3–14-fold. The results of our study are consistent with these reports. All of these transcription factors, with the exception of TAFII31 and YB-1, were found to be up-regulated in the HTLV-I cells (Table 1). Also included in our transcription factor group were several members of the CREB transcription factor group including ATF3, ATF6, and cAMP-responsive element modulator (CREM). As stated above, CREB/ATF plays an important role in HTLV-I-LTR transcription and thus viral expression and replication.

Consistent with the work of Sharma et al. (46), the transcriptional activator IRF4 which binds to the IFN-stimulated response element in the immunoglobulin λ light chain enhancer and plays a role in ISRE-targeted signal transduction mechanisms specific to lymphoid cells, is up-regulated up to 17-fold. Interestingly, IRF5 is also increased 3-fold in HTLV-I-transformed cells, whereas the family members IRF1 and IRF3 are decreased compared with PBLs (Table 1). This suggests that HTLV-I-infection may impact only a subset of IRF-regulated genes.

Several members of the cell cycle machinery have altered expression in HTLV-I-infected cells compared with control PBLs. The expression of the Cdc25C tyrosine phosphatase was increased 3–6-fold (Table 1) in HTLV-I-transformed cells. This protein functions as a dose-dependent inducer of mitotic control. It is required for progression of the cell cycle, through dephosphorylating Cdc2, activating kinase activity (Fig. 3 A). The fact that cyclin B and Cdc2 (CDK1) expression are also increased may further indicate a deregulation of G₂/M cell cycle control.

The Cdc25C promoter contains nuclear factor Y-binding sites important for the regulation of Cdc25C expression (47, 48). We have previously shown that Tax interacts directly with nuclear factor Y to allow activation of the MHC class II DQ promoter (24). To test whether Tax could activate the Cdc25C promoter, we performed cotransfection experiments in Jurkat T cells. As demonstrated in Fig. 3 (B and C), Tax expression activates the Cdc25C promoter by ∼13-fold. Deletion of the upstream promoter region to −75 reduced promoter activity, but did not diminish Tax transactivation (Fig. 3,C). The C75 promoter construct contains four copies of the Y box element. That the Y boxes are important for Tax transactivation is demonstrated by mutant pCDC25C75mY1, which contains base substitutions in the Y1 box. The expression from this plasmid was identical in the absence and presence of Tax (Fig. 3, B and C). These results correlate with the increased expression of Cdc25C mRNA seen in the microarray analysis. To note, although we consistently observed an increase in Cdc25C mRNA, posttranscriptional regulation is also a factor in the control of Cdc25C at the protein level (data not shown).

As with many oncogenic factors, deregulated checkpoint control occurs at several stages within the cell cycle. In addition to the increase in G₂/M phase regulatory factors, we see an increase in DNA replication licensing and elongation factors including MCM2, HsMCM6, MCM7, Cdc28 kinase 1 and 2, Cdc18L, RFC3, and RFC4 (Table 1). As described previously (49, 50, 51, 52), we see increased expression of the cell cycle regulators PCNA (2–7-fold), p21 (2–5-fold), CDK2 (2–6-fold), CDK4 (1–7-fold), and thioredoxin (2–7-fold). Concomitant with an increase in cdk levels, there is a decrease in the cdk inhibitor p19 (Table 1). Together these results suggest that HTLV-I infection impacts several steps in cell cycle regulation.

Experiments from numerous laboratories have defined apoptotic pathways and gene products that function to inhibit or accelerate cellular apoptosis. For example, HIAP-1 acts to repress apoptosis in mammalian cells, presumably by inhibiting the activity of caspases involved in cell death (53, 54). Interestingly, HIAP-1 and API1 (55, 56) are overexpressed 5–8-fold in HTLV-I cells. API1 functions in the cell to inhibit the action of specific caspases in the induction of cell death (55, 56). We also found that the apoptosis inhibitor BCL2L1 (Bcl-xL) was overexpressed in HTLV-I cells 4–7-fold. Similar to the findings of Ruckes et al. (25), we also find increased expression of the anti-apoptotic factor I-309. In contrast to the results of Harhaj et al. (21), we did not detect the overexpression of other apoptosis inhibitors such as dad1, ddlc1, HSP27, or NKEF in HTLV-I cells compared with activated T-cells.

In addition to the overexpression of genes that inhibit apoptosis, we also found genes that induce apoptosis were down-regulated. For example, the expression of caspase-8, a cysteine protease that functions in the initiation of the apoptotic proteolytic pathway (57, 58) is repressed in the HTLV-I cells. Depletion or inactivation of caspase-8 in cells is reported to prevent p53 transcription-independent apoptosis and significantly attenuate overall cell death induced by wild-type p53 (59). Similarly, caspase-4 and -6 expression was down-regulated in the HTLV-I cells.

It has been shown that decreases, as well as increases, in cellular gene expression are important regulatory events. Previous studies (21, 22, 23) have identified a limited number of repressed genes. In our study, we have identified 420 genes which are decreased 2-fold or greater in the HTLV-I cells are compared with control PBLs (on-line data supplement). Interestingly, several protein kinases were down-regulated in expression as compared with activated PBLs (Table 1). Very striking was the down-regulation of several protein kinase C isoforms including, β1, β2, θ, eta, ζ, and protein kinase C-like 2 kinase. Protein kinase C is a family of at least 10 structurally related enzymes that have been implicated in a variety of cellular responses. The protein kinase C (PKC) substrate inositol 1,4,5-triphosphate 3-kinase was also down-regulated, further suggesting that control of the pathway used in normal T-cell activation has been repressed in HTLV-I-infected cells.

We have used Affymetrix microarray technology to analyze and compare the gene expression profiles of ∼7300 genes in activated and HTLV-I-infected lymphocytes. Several new genes that have deregulated expression in HTLV-I-transformed cells have been identified including cellular receptors, cytokines, apoptosis inhibitors, kinases, checkpoint regulators, and transcription factors. Because all of the HTLV-I cells used in these studies are stably infected, it is not possible at this time to distinguish which genes are deregulated during the initial immortalization/transformation events and which genes are induced as the result of secondary effects of HTLV-I infection. In addition, a clear distinction between HTLV-I-immortalized versus -transformed cells cannot be made at this point and requires analysis of more HTLV-I-infected samples.

In examining the expression profiles, however, we noted several compelling changes. In particular, several alterations have occurred in the cell cycle/DNA repair pathways. We noted changes in the factors controlling G₂/M progression. Specifically, we have identified for the first time that mRNA expression of the Cdc25C tyrosine phosphatase was increased in HTLV-I-transformed cells. This protein phosphatase functions as a dose-dependent inducer of mitotic control (Fig. 3). Cdc25C is activated by hyperphosphorylation of the N-terminal domain. Several kinases, including Chk1, phosphorylate Cdc25C at Ser216 (60). Ser216 is phosphorylated throughout interphase, but not in mitosis. Ser216-phosphorylated Cdc25C is recognized and bound by 14-3-3, which may sequester Cdc25C in the cytoplasm, preventing it from interacting with Cdc2. At the appropriate time in the cell cycle, hypo- or unphosphorylated Cdc25C dephosphorylates Cdc2, activating kinase activity. The fact that cyclin B and Cdc2 (CDK1) expression are also increased may further indicate a deregulation of G₂/M cell cycle control. It will be of interest to examine the phosphorylation state of Cdc25C and Cdc2 in the HTLV-I-transformed cells.

As noted previously, several changes in the cytokine/cytokine receptor signal cascades are altered in HTLV-I-infected cells (for reviews see Refs. 9, 10, 11). We have extended these studies and have shown the expression profiles of several cytokines and their receptors including IL-15, IL-15Rα, IL-6, IL-6R, IL-7, and IL-7R, as well as T-cell signaling molecules. Importantly, kinases and phosphatases, known to play a role in signaling cascades, are also altered in HTLV-I-infected cells. Further investigation is needed to determine which pathways result in cellular immortalization/transformation and which are a result of immortalization/transformation.

The data presented in this study represents the most comprehensive analysis of gene expression patterns in HTLV-I-transformed cells to date. Of the 7300 genes analyzed in this study, the expression of ∼763 genes was deregulated >2-fold in the HTLV-I-transformed cells. Analysis of the deregulated genes, in terms of known function, allows several important conclusions. First, there is no single regulatory pathway that is solely targeted for activation/inactivation in the stably transformed cell. Rather, a network of interrelated pathways including cell proliferation, T-cell signaling, and immune regulation are deregulated. Second, there appear to be multiple alterations in the cell death/survival pathway that have occurred. Several proteins that increase the rate of apoptosis are down-regulated in the HTLV-I-transformed cell. Moreover, several proteins that inhibit apoptosis were up-regulated. Clearly, the overall pattern is to disrupt the cell cycle regulatory points and favor survival of the cell. Whether the pattern of gene expression is the same in all cells within the transformed cell population or whether what we are seeing is a global pattern of individual gene expression changes within the population await further analysis. It is also clear that the analysis of gene expression in the initial stages of infection and transformation would be of tremendous value in defining key transformation pathways.