It’s All in the Interaction: Quantitating Gene Networks

Toxicologists who use microarrays hope to uncover relationships that link gene expression data to signal transduction pathways, gene networks that are often used to describe the sequence of biochemical events controlling cellular function. The large quantities of data generated by microarray studies generally are examined qualitatively—for example, by comparing whether one gene is turned on relative to another. These qualitative relationships, however, fail to describe how genes in a network influence each other. Still in their infancy are tools that quantitate the complex relationships within gene networks more comprehensively than simple correlations between pairs of genes. Now, for the first time, researchers describe a new quantitative statistical technique that assesses the interactions of genes in a network [EHP 112:1217–1224]. 
 
The team, led by Hiroyoshi Toyoshiba of the NIEHS Laboratory of Computational Biology and Risk Assessment, created a statistical software program that verifies concurrently that the expression of one gene is linked to the expression of several others. The first proof-of-concept demonstration evaluated genes that are directly responsive to tetrachlorodibenzo-p-dioxin (TCDD; a ubiquitous environmental pollutant and known human carcinogen) and their effect on the retinoic acid signal transduction pathway. 
 
Signal transduction pathways respond to different environmental conditions; they are like molecular circuits that detect and integrate diverse external signals to alter gene transcription. This results in changes in enzyme activities as well as the production of abnormal levels of proteins, which further results in changes in biochemical processes. Alterations in signal transduction pathways can lead to cancer and other disorders. 
 
Toyoshiba and colleagues had earlier identified genes that are altered in lung airway epithelial cells after exposure to TCDD. Starting with microarrays composed of 2,000 genes that are known to be expressed in response to environmental toxicants, the researchers had identified 11 genes that responded significantly to TCDD in two different lung cell lines. These genes appeared to be involved in the effects of TCDD on the retinoic acid signal transduction pathway. 
 
The researchers constructed a hypothetical model of the retinoic acid signal transduction pathway that describes how the 11 genes interrelate. Based on published reports on retinoic acid metabolism, the model postulated that dietary vitamin A (retinol) is converted first to retinal and then to retinoic acid by alcohol dehydrogenases and, possibly, by cytochrome P450 enzymes. Once synthesized, retinoic acid enters the cell nucleus. There, it binds retinoic acid receptor beta, which, in turn, alters the expression of genes that may play a role in tumor formation. The hypothetical model included genes that produce three alcohol dehydrogenases, a cytochrome P450 enzyme, retinoic acid binding proteins and receptors, and four nuclear proteins. 
 
Following exposure to three concentrations of TCDD, the expression levels of the 11 genes were calculated relative to unexposed controls. Statistical methods were applied to these data to test the hypothetical linkages between TCDD-responsive genes and the retinoic acid signal transduction pathway. These tests confirmed strong linkages between the genes included in the hypothetical model. 
 
Epidemiological studies show a strong association between TCDD and lung cancer; the model offers a potential explanation for how TCDD damages the lungs. TCDD appears to activate genes associated with the synthesis of retinoic acid, which—through the retinoic acid signal transduction pathway—turns on nuclear genes that promote cell proliferation and carcinogenesis. Scientists can focus future experiments on particular genes directly related to TCDD-induced tumor progression. 
 
The new statistical tool makes it possible to understand biological pathways in cells, tissues, organs, and whole organisms. It can be expanded to include other relevant data, such as protein levels in cells. These data can be combined with pharmacological models to present a true systems biology approach to quantifying risks from exposure to xenobiotics such as TCDD, suggest the authors. Other researchers can obtain the statistical software by contacting laboratory director Christopher Portier at portier@niehs.nih.gov.


Background
Human immunodeficiency virus type 1 (HIV-1) viral protein R (Vpr), a virion-associated protein with a calculated molecular weight of 12.7 kilodalton (kD), is highly conserved among HIV, simian immunodeficiency virus (SIV) and other lentiviruses [1][2][3]. During the acute phase of the viral infection, Vpr is preferentially targeted by the HIVspecific CD8 T-lymphocytes [4,5]. Increasing evidence suggests that Vpr plays an important role in the viral life cycle and pathogenesis. For example, Vpr is required both in vitro and in vivo for viral pathogenesis and efficient viral infection of non-dividing host cells such as monocytes and macrophages [6,7]. Rhesus monkeys, chimpanzees and human subjects infected with Vpr-defective viruses have a slower disease progression often accompanied by reversion of the mutated vpr genes back to the wild type phenotype [8][9][10][11][12].
Vpr displays several distinct activities in host cells. These include induction of cell cycle G2 arrest [13][14][15][16][17] and cell killing [18]. The cell cycle G2 arrest induced by Vpr is thought to suppress human immune functions by preventing T cell clonal expansion [19] and to provide an optimized cellular environment for maximal levels of viral replication [8]. In addition, Vpr induces cell death, which may contribute to the depletion of CD4+ T-cells in HIV-infected patients [12,18]. Whether Vpr-induced G2 arrest and cell death are functionally independent of each other is currently of controversial. There are reports suggested that these two activities are separable both in fission yeast and mammalian cells [20][21][22][23][24]; others suggested that Vpr-induced apoptosis is cell cycle dependent [25,26]. Reasons for these discrepancies are not clear at the moment. In an earlier report, we demonstrated that overexpression of fission yeast (Schizosaccharomyces pombe) Hsp16 specifically suppresses Vpr activities, resembling cellular stress responses to heat shock, [27]. Here, we further show that this suppression is mediated by a heat shock factor (Hsf)-mediated mechanism. Furthermore, we have also tested the suppressive effect of Hsp16 on wild type and a F34I mutant Vpr. The wild type Vpr induces cell cycle G2 arrest and cell death, the F34IVpr mutant is incapable of inducing cell death but retains its ability to induce cell cycle G2 arrest both in fission yeast [21,27,28] and mammalian cells ( [29]; our unpublished data) Thus, examination of the wild type and the F34I mutant Vpr enable us to investigate these two Vpr activities separately. In addition, the highly conserved Vpr effect on cell cycle G2/M regulation and cell survival makes fission yeast a particularly useful model to study mechanisms of these Vpr activities (For review of this subject, see [30][31][32][33][34]). Interestingly, vpr gene expression appears to trigger a moderate increase in Hsp16 levels but counteracts heat shock-mediated elevation of Hsp16. Together, our findings suggest a highly conserved and dynamic inter-play between vpr gene expression and cellular heat shock response involving heat shock proteins.

Endogenous Hsp16 is responsive to vpr gene expression
We previously identified fission yeast Hsp16 as a potent Vpr suppressor [27]. Analysis of hsp16 expression in S. pombe Q1649 strain, in which the hsp16 gene is tagged with GFP and is under the control of its native promoter [35], demonstrated that both the wild type Vpr and the mutant protein (Vpr') elicited Hsp16 production (Fig.  1A). The mutant Vpr', in which phenylalanine in position 34 was replaced with isoleucine (F34IVpr), was used in this study to measure Vpr-induced cell cycle G2 induction because the wild type Vpr kills cells. Vpr' has lost its ability to induce cell killing but retains its capacity to induce G2 arrest as previously shown both in human (our unpublished data) and yeast cells [21,27,28].
We next tested whether the expression of endogenous hsp16 is responsive to vpr gene expression. Both the wild type vpr and F34I mutant vpr genes were induced by depleting thiamine from the EMM medium as previously described [36,37]. As shown in Fig. 1B, expression of wild type vpr or mutant vpr' under normal growth conditions elicited a moderate increase of the Hsp16 protein level (Fig. 1B, lanes 3 and 5). The faint protein band in lane 2 could possibly be due to low level of vpr expression even when the inducible promoter is repressed [36]. Together, these observations suggest that Hsp16 production is responsive to vpr gene expression. These results are consistent with our studies in mammalian cells where vpr gene expression stimulates expression of HSP27, a human paralogue of Hsp16 (Our unpublished data).

Overproduction of Hsp16 suppresses viral infection in CD4-positive T-cells and macrophages
Vpr activities have been implicated as positive factors for HIV-1 replication [6,8,38]. Consistent with these activities, Vpr has been shown to increase viral replication 2 to 4 fold in proliferating T lymphocytes [8,39,40] but its activities are required for viral infection in non-dividing cells such as macrophages [6,7]. Responsive expression of human HSP27 and yeast hsp16 to Vpr suggest a possible and highly conserved cellular activity against Vpr. Indeed, we have showed previously that overproduction of Hsp16 reduces viral replication in CD4-positive T-cells in a Vprdependent manner [27].
To further delineate the suppressive effect of Hsp16 on Vpr, here we tested the effect of Hsp16 on viral replication in CD4-postive cells infected by a viral strain IIIB, in which the vpr gene has a frame shift mutation at codon 73 resulting in a truncated Vpr protein that misses 24 a.a. at its C-terminus [8,11,41]. The C-terminal Vpr is responsi-ble for a number of Vpr activities including protein dimerization [42], cell cycle G2 arrest and cell death [20,43]. We established a CD4+ H9 cell line stably producing high level of yeast Hsp16 (Fig. 2A). These H9 cells were then infected with a HIV-1 Vpr-positive laboratory strain LAI. To test the potential effect of Hsp16 on viral replication, p24 antigen was measured in culture supernatants over a period of 21 days after infection. As shown in Fig. 2B and consistent with our previous findings [27], a consistent but moderate reduction of HIV-1 viral replication was observed in cells expressing hsp16. For example, levels of p24 antigen steadily increased in HIV-infected cells expressing the vector control from day 3 to day 21 of HIV-1 infection indicating successful viral infection. (Fig. 2Ba). However, a 1.5 to 4.5-fold reduction in p24 antigen levels was detected in HIV-infected cells expressing Hsp16 from day 10 to 21 after viral infection. No detectable p24 antigen was observed in mock-infected cell over the entire experimental period. To ensure the observed viral inhibition by Hsp16 is not cell line-specific, we examined another CD4-positive cell line, CEM-SS, which was also derived from T lymphocytes [44]. A similar suppressive Endogenous Hsp16 is responsive to vpr gene expression Figure 1 Endogenous Hsp16 is responsive to vpr gene expression. (A) Expression of hsp16 was measured through GFP green fluorescence as shown by gfp-hsp16 fusion protein expression. Cells were grown under normal growth conditions and expression of the wild type vpr (Vpr) or mutant F34I vpr (Vpr') was induced in thiamine depleted EMM medium as previously described [35]. Photographs were taken 24 hrs after gene induction. Small panel in A-a shows cells without green fluorescence. (B) Comparison of the Hsp16 protein levels in the presence and absence of Vpr as shown by Western blot analysis. The vpr or vpr'-expressing cells were collected at the same time as in panel (A). Lane 1 shows wild type SP223 cells without plasmid; lanes 2 and 4 show cells with vpr gene expression repressed; lanes 3 and 5 -cells with vpr gene expression induced. Note that elevation of Hsp16 shown in lane 2 is most likely due to leakage of nmt1 promoter and low level gene expression under these conditions [36]. LC, protein loading control. A protein band that nonspecifically reacted to the antibody was used as a protein loading control. GI, gene induction. effect on viral replication (1.5 to 3.1-fold reduction) was also observed in the CEM-SS cells that stably express hsp16 genes ( Fig. 2B-c).
To examine whether Hsp16 retains its suppressive effect on viral replication when 24 aa of the C-terminal Vpr is removed, we repeated the same infection experiments in the H9 and CEM-SS cells using the C-terminal truncated Vpr-carrying viral strain IIIB. As shown in Fig. 2B-b and Fig. 2B-d, the kinetics of viral replication were essentially indistinguishable between cells with or without Hsp16, suggesting that Hsp16 has lost its inhibitory effect on viral replication in the absence of C-terminal end of Vpr.
The above data suggest the suppressive effect of Hsp16 is specific to Vpr. Since Vpr is required for viral infection in non-dividing cells such as macrophages, we next tested the potential effect of Hsp16 on HIV infection in macrophages. Purified fission yeast Hsp16 protein was added to primary human macrophages infected with HIV-1 ADA with increasing concentration from 1, 5 to 10 μg/ml of cells. Viral replication was followed 7 and 10 days after infection by measuring the reverse transcriptase (RT) activities in culture supernatants. To avoid potential interference of endotoxin that often presents in purified recombinant proteins [45], purified Hsp16 was treated with 10 μg/ml Polymyxin B (PMB)-agarose that was shown to efficiently remove endotoxin [45]. As shown in Fig. 2C-a, infected macrophages without removing endotoxin (-PMB) almost completely eliminated viral replication at day 7 after infection; about 3.5 to 7-fold decrease of viral infection was observed at day 10 with 1 or 5 μg/ml of Hsp16. No viral activity was detectable at 10 μg/ml level. After removing the possible endotoxin from Hsp16, reduced but still significant reduction of viral replication was observed both at day 7 and day 10 after infection. 2.6 to 8.0-fold decrease of viral replication were seen at day 7 with 1.7 to 5.5-fold reduction of viral replication was observed in day 10. These data suggest a dose-dependent suppression of viral replication by Hsp16 in macrophage.
To ensure the observed effect was indeed due to Hsp16, as a control, we also tested the potential effect of purified HSP27. 10 μg/ml of HSP27 with the same level of PMB (10 μg/ml) was added to HIV-1 ADA -infected macrophages the same way as we did for Hsp16. RT activities were measured over time. As shown in Fig. 2C-b, no significant differences were seen during the entire 24 days after infection. These data suggest a dose-dependent suppression of viral replication by Hsp16 in macrophage. Together, these data show that overproduction of Hsp16 specifically inhibits HIV-1 infection possibly by targeting the Vpr activates.

Heat shock factor is the key regulator for the elevation of Hsp16 and heat shock-mediated suppression of Vpr
Overexpression of hsp16 by itself has no any obvious effect on cell length or morphology [27,35]. However, our earlier data showed that overexpression of hsp16 or high temperature (36°C) suppressed Vpr-induced G2 arrest as measured by cell elongation in fission yeast [27], indicating a potential and specific suppressive effect of Hsp16 on Vpr. Since Hsp16 can be activated by host cellular stress responses through heat shock factor (Hsf)-mediated pathway, we next investigated the potential involvement of heat shock factor (Hsf) in the heat shock-mediated suppression of Vpr. There is only one Hsf in S. pombe. However, deletion of hsf1 is lethal in yeast [46], thus we were unable to test the deletion effect of Hsf on the Vpr activities. Instead, we overexpressed the hsf1 gene from a pART1-hsf1 plasmid where it is controlled by an exogenous and constitutively expressing adh promoter. Since no specific antibody against Hsf1 is available, we used Hsp16 as a marker for hsf1 expression [35]. As shown in Fig. 3Aa, b, empty pART1 plasmid had no effect on Vpr'-induced cell elongation. Cells were 17.7 ± 0.7 μm in length 30 hrs after vpr gene induction [47]. Expression of hsf1 by itself in S. pombe cells gave rise to slightly shorter (6.1 ± 0.1 μm) than normal cells ( These observations suggested that Hsf1 is probably the major cellular factor that contributes to the anti-Vpr activities. To verify this finding, we further examined whether heat shock treatment can induce additional shortening of cells besides the suppressive Hsf1 effect. If additional cellular factors are involved in suppressing Vpr' during the cellular heat shock response, we would expect to see shorter cell length than when Hsf is overexpressed alone. The same experiment as described above was repeated at elevated temperature (36°C). No additional shortening of cells beyond the length observed with Hsf1 overexpression at normal temperature was seen when vpr-expressing cells were grown at 36°C with overproduced Hsf1 (Fig.  3B). Together, results of these experiments suggest that Hsf is the key cellular regulator of heat shock-mediated suppression of the Vpr activities. To neutralize the effect of potential contamination with endotoxin [45], 10 μg/ml of Polymyxin B (PMB) was added [45]. Results are mean ± SE of triplicates. Heat shock factor is responsible for Hsp16 elevation and heat shock-mediated suppression of the Vpr activities . Because of the lack of an antibody against fission yeast anti-Hsf1 and our interest in monitoring Hsf1mediated Hsp16 elevation, Hsp16 protein production was used here as marker for Hsf1 activity [35]. Vpr + hsf1

Vpr counteracts Hsp16 elevation induced by heat treatment at the transcriptional level
Even though induction of cellular heat shock response by heat treatment suppresses vpr'-induced cell cycle G2 delay, surprisingly, the same heat treatment was not able to block Vpr-induced cell death in RE007 cells which express the wild type vpr (Fig. 4A-3, bottom plate). Inability of colony formation at high temperature is not due to lack of vpr expression because Western blot analysis showed that heat treatment does not affect the Vpr protein level ( Fig.  4B; [27]). Since heat treatment induces high levels of Hsp16 [35] and artificial overproduction of Hsp16 suppresses Vpr-induced cell death at both temperatures ( Fig.  4A-a and 4A-4, bottom), it was puzzling why heat treatment only suppresses Vpr-induced G2 arrest but it does not suppress Vpr-induced cell killing. One potential explanation is that wild type Vpr may actually prevent heat-induced elevation of Hsp16. To test this possibility, we measured protein levels of Hsp16 in the presence and absence of Vpr using different methods. One was to observe the fluorescent signal emitted by the GFP-Hsp16 fusion protein in a S. pombe Q1649 strain, in which the hsp16 gene is tagged with GFP and is under the control of its native promoter ( Fig. 5A; [35]). Changes in the Hsp16 protein level were further quantified by measuring fluorescent intensity (FI) using a luminescence spectrophotometer [35,48,49]. In addition, Western blot analysis was also carried out to measure endogenous Hsp16. Two heat treatment methods were used to delineate the potential effect of Vpr on the Hsp16 protein levels. Acute heat shock (45°C for 15 min) was used to transiently activate Hsp16, and the Vpr effect was measured 2 hrs after the heat shock.
As an alternative method, constant high temperature was used for lasting elevation of Hsp16, and the effect of Vpr on Hsp16 was measured 48 hrs after cell culturing at 36°C.
Under the normal growth conditions, Hsp16 protein expression is typically very low or undetectable (FI = 0.1 ± 0.3; Fig. 5A-a; Fig. 5B-a,b, lane 1 [35]). When these cells were subjected to an acute heat shock (45°C for 15 min), a significant increase (FI = 5.9 ± 0.2) in the Hsp16 protein level was observed 2 hr after heat shock in cells that either had no vpr-containing plasmid (Fig. 5A-d; Fig. 5B-a, b, lane 2) or vpr gene expression was suppressed ( Fig. 5B-a,  lanes 3,5). In contrast, the level of Hsp16 (FI = 3.1 ± 0.6) was markedly decreased when wild type vpr was expressed under the same heat shock conditions (Fig. 5A-e; Fig. 5Ba, lane 4). Similar Hsp16 elevation (Fig. 5A-g; Fig. 5B-b, lane 2-4) was also observed in cells grown under constant high temperature at 36°C. Consistent with the observation shown in acute heat shock experiment, Hsp16 protein level was diminished in the vpr-expressing cells cultured at 36°C for 48 hrs (Fig. 5B-b,  However, after prolonged (48 hrs) incubation of vprexpressing cells at constant high temperature, both the wild type and mutant Vpr were able to eliminate Hsp16 elevation ( Fig. 5B-b, lane 5-6). Taken together, these observations provide an explanation to our finding that heat treatment suppresses the Vpr'-induced cell cycle defect but does not protect against Vpr-induced cell killing because the F34I mutation in Vpr' may have attenuated the ability of Vpr to down-regulate Hsp16 thus allowing elevated Hsp16 to suppress activity of Vpr'. Therefore, wild type Vpr specifically counteracts activation of Hsp16 in response to vpr gene expression or heat treatment.
It is of interest to note that overexpression of hsp16 under the control of an exogenous nmt1 promoter suppressed Vpr-induced cell killing in the wild type cells (Fig. 4A-2 bottom panel; [27]) suggesting that the counteracting effect of Vpr on Hsp16 is specifically targeted to the hsp16 promoter, i.e., occurs at the transcriptional level. Attempting to confirm this possibility, we further tested whether overexpression of hsp16 under the same nmt1 promoter was also capable of suppressing Vpr-induced cell death at 36°C when Vpr has the strongest counteracting effect on Hsp16. As shown in the bottom panel of Fig. 4A-4, overproduction of Hsp16 was indeed capable of blocking Vprinduced cell death at both high (36°C) and normal growth temperature (30°C). Therefore, Vpr counteracts Hsp16 elevation induced by heat treatment most likely at the transcriptional level.

Discussion
In this report, we provide evidence that Hsf1 is the main regulator responsible for Hsp16 elevation and anti-Vpr responses in fission yeast cells. The fact that Hsf1 is responsible for Hsp16-mediated response to Vpr indicates that the effect of Hsp16 on Vpr resembles the cellular heat shock responses. Indeed, overexpression of hsf1 completely reduced Vpr'-induced cell cycle G2 arrest as shown by reversion of the cell elongation ( Fig. 3A-d vs. b), shift of the cellular DNA content from G2 to G1 (Fig. 3A-b) and additional heat treatment of hsf-expressing cells did not significantly enhance the suppressive effect of Hsf1 on Vpr ( Fig. 3B-h vs. d).
Even though vpr gene expression triggers Hsp16 elevation, Vpr appears to prevent further elevation induced by heat treatment (Fig. 5A-e; Fig. 5B-a, lane 4; Fig. 5B Fig. 5A, in which a GFP reporter was fused with the endogenous hsp16 promoter [35] and expression of vpr eliminated the Hsp16 elevation ( Fig. 5Ae, h). Although the molecular mechanism underlying this transcriptional suppression of hsp16 is unclear at the moment, the fact that Hsf activates Hsps through binding of the Hsp promoters [50] and overexpression of hsf1 or hsp16 through an exogenous adh or nmt1 promoter alleviates the Vpr activity ( Fig. 3A; [27]) support the idea that Vpr may affect expression of hsp16 through competition with Hsf for control of hsp16 expression. One possible scenario is that Vpr may inhibit Hsf1 that results in reduced transcription of hsp16. Alternatively, since Vpr is a weak transcriptional activator through binding to the transcriptional factor Sp1 [51], it is also possible that Vpr may compete with Hsf1 by binding to the Sp1 region of the hsp16 promoter. Obviously additional tests are needed to elucidate these possibilities. Interestingly, only the wild type Vpr was able to inhibit Hsp16 at early hours (23 hrs) after induction, as a single amino acid substitution from phenylalanine to isoleucine at position 34 of Vpr attenuated its ability to suppress the increase of Hsp16 after acute heat shock (Fig. 5A-i vs. h; Fig. 5B, lane 6 vs. lane 4). In fact, an even higher level of Hsp16 was observed. This is presumably due to the inability of Vpr' to compete with Hsf-mediated Hsp16 elevation. Thus, assuming that expression of hsp16 is responsive to both the presence of Vpr and heat shock treatment, this larger increase of Hsp16 could be an additive effect. Since amino acid substitution at residue 34 of Vpr diminishes the ability of Vpr to induce cell death but retains induction of G2 arrest [20,28,29], a plausible possibility is that suppression of Hsp16 and induction of cell death by Vpr share common pathways.
It should be mentioned that whether Vpr-induced G2 arrest and cell death are two functionally independent activities is still of debate. Earlier reports suggested that these two activities are separable both in fission yeast and Expression of hsp16 under exogenous nmt1 promoter rather than endogenous promoter suppresses Vpr-induced cell killing mammalian cells [20][21][22][23][24] [52]. However, recent reports indicated Vpr-induced apoptosis is cell cycle dependent [25,26]. Although reasons for these discrepancies are not completely clear at the moment, it is noticed that apoptosis shown in the Andersen's study describes a late event as cells were collected 48-72 hrs after viral infection [26]. Prolonged cell cycle G2 arrest results in apoptosis. Thus, it is not surprising to find that apoptosis described in the Andersen's study is ANT-independent and ANT-dependent apoptosis was documented previously [53]. Additional difference between the apoptosis described by Andersen et al from others is also noticed in the examination of two Vpr mutations. The R77Q and I74A mutants, which separate the apoptosis and G2 arrest induced by Vpr [24,54], showed no separation between the G2 induction and apoptosis. In our study, the F34IVpr mutant is unable to induce cell death but retains its ability to induce cell cycle G2 arrest both in fission yeast [21,27,28] and mammalian cells ( [29]; our unpublished data) It thus allowed us to differentiate the effect of a wild type Vpr vs. a mutant Vpr that only confers the inhibitory effect on cell cycle regulation.
Responsive elevation of fission yeast Hsp16 and its human paralogue HSP27 (our unpublished data) suggests that the cellular heat stress-like responses might be antagonistic to Vpr. Indeed, we previously showed that overexpression of hsp16 and human HSP27 suppress the Vpr activities, including cell cycle G2 arrest and cell killing, both in fission yeast and human cells ( [27]; our unpublished data). However, the suppressive effect of yeast Hsp16 and human HSP27 on Vpr are not identical. Overproduction of Hsp16 completely eliminated all of the Vpr activities including the positive role of Vpr in supporting viral replication in macrophages ( Fig. 2C-a). Under the same condition, however, HSP27 has no clear suppressing effect against Vpr in macrophages ( Fig. 2C-b). One possible difference between these two HSPs is that Hsp16 associates directly with Vpr [27] but no clear HSP27-Vpr interaction was detected both in vitro and in vivo (our unpublished data). Unlike Hsp16, overexpression of HSP27 is unable to block nuclear transport capacity of Vpr (our unpublished data). Since nuclear transport of Vpr is required for HIV-1 infection in non-dividing cells such as macrophages [7,55,56], the inability of HSP27 to block nuclear import of Vpr could potentially explain why it has no effect on HIV-1 infection in macrophages.
There appears to be a dynamic interaction between vpr gene expression and activation of Hsp16 in fission yeast.
Results of our parallel studies in mammalian cells indicated a similar dynamic and antagonistic interaction between Vpr and HSP27 (our unpublished data). This finding is not surprising because activation of heat shock proteins by Hsf1 is a highly conserved cellular process among all eukaryotic cells [50]. All eukaryotes encode at least one heat shock factor that is believed to regulate transcription of heat shock genes. This protein binds to a regulatory sequence, i.e., the heat shock element, that is absolutely conserved among eukaryotes [50]. Based on the data presented, we hypothesize that expression of vpr or HIV infection elicits a transient activation of the small heat shock proteins (sHsps) of eukaryotes through an Hsfmediated pathway. Activation of these sHsps is most likely a part of the cellular antiviral reaction to HIV infection and specifically to Vpr. However, these stress responses are normally not sufficient to suppress the Vpr activities because of active counteraction from Vpr. Importantly, however, the Vpr activities could be completely blocked when sHsp's are produced under control of an exogenous promoter thus avoiding transcriptional inhibition by Vpr. Since the Vpr-specific activities have been linked to such clinical manifestation of AIDS as activation of viral replication [57], suppression of host immune responses [19] and depletion of CD4+ T-lymphocytes [12,58], this finding could potentially provide a new approach to reducing Vpr-mediated detrimental effects in HIV-infected patients by stimulating expression of sHsps.

Maintenance and growth of mammalian and yeast cells
Genotypes and sources of S. pombe strains, mammalian cell lines and plasmids used in this study are summarized in Table 1. CD4-positive H9 and CEM-SS cells were grown in RPMI 1640 medium supplemented with 10% heatinactivated fetal calf serum (FCS) and 100 unit/ml of penicillin/streptomycin. Gene inductions of HIV-1 vpr and other cellular genes under the control of the nmt1 promoter in fission yeast have been described previously [27,37]. Cells containing the plasmid with the nmt1 promoter were first grown to stationary phase in the presence of 20 μM thiamine. Cells were then washed three times with distilled water, diluted to a final concentration of approximately 2 × 10 5 cells/ml in 10 ml of the appropriately supplemented EMM medium with or without thiamine. Cells were examined approximately 24 hours after gene induction. Fission yeast cells were normally grown at 30°C with constant shaking at 250 rpm unless otherwise specified.
Induction of cellular heat shock responses were conducted as previously described [27,35,59]. Briefly, cultures were first grown as mentioned above to fully express vpr and then exposed to either an acute heat shock at 45°C for 15 min or grown at consistent high temperature at 36°C for an indicated period of time.

Measurement of the Vpr-specific activities in fission yeast
All of the functional assays used to measure Vpr-specific activities, i.e. cell cycle G2 arrest and cell death induced by Vpr, have been described previously [20,37,60]. Vprinduced G2 arrest can be specifically measured in fission yeast based on a number of cellular endpoints [21,37].
Here we used F34IVpr (Vpr')-induced cell elongation as a marker of G2 arrest [47,61]. The mutant Vpr', in which phenylalanine was replaced with isoleucine in position 34 (F34IVpr) was used instead of the wild type Vpr because Vpr' has lost its ability to induce cell killing but retains its capacity to induce G2 arrest, as previously shown both in human (our unpublished data) and fission yeast cells [21,27,28]. The use of cell elongation, also known as "cdc phenotype", as a marker for cell cycle G2/M delay is a standard approach in fission yeast [47,[61][62][63]. Cell images were first captured on a Leica microscope and the cell length was determined using OpenLab software. Average and standard deviation of cell length were calculated based on three independent experiments, each counting at least 100 cells. To verify Vpr-induced cell cycle G2 arrest, flow cytometric analysis is carried out as previously described [37] Induction of cell death by Vpr was detected by inability of vpr-expressing cells to form colonies on agar plates as previously described [60]. Briefly, S. pombe cells containing the pYZ1N::vpr constructs were first grown on a selective leucine-free minimal EMM plate under vpr-repressing condition. A loopful of viable cells was streaked onto vprinducing or vpr-repressing EMM plates and incubated at 30°C for 3-4 days. Inability to form colonies on the vprinducing EMM plates but normal growth on the vprrepressing EMM plates is indicative of Vpr-induced cell killing [20,64].

Viral infections
To evaluate the suppressive effect of Hsp16 on viral replication in proliferating CD4-positive T-lymphocytes, H9 and CEM-SS cells [41] that stably express a plasmid control or hsp16 were established [27]. 3 × 10 6 to 5 × 10 6 of these H9 or CEM-SS cells were either mock infected or infected with 2.0 × 10 3 TCID 50 of HIV-1 LAI or IIIB. The HIV-1 LAI strain carries a wild type vpr gene; the vpr gene in the IIIB strain has a frame shift mutation at codon 73, which results in a truncated Vpr protein missing its C-terminus [8,11,41]. Equal infection of the cells was further verified by measuring viral RNA levels 24 hr after viral inoculation using the Roche Monitor assay following the manufacturer's instructions. Viral replication was determined by p24 antigen levels using a commercially available HIVAG-1 polyclonal antigen kit (Abbott Laboratories, Abbott Park, IL).
Monocyte-derived macrophages (MDMs) were prepared from peripheral blood mononuclear cells (PBMCs) by adherence to plastic as described previously [65].

Fluorescence microscopy
A Leica fluorescence microscope DMR equipped with a high performance CCD camera (Hamamatsu) and Open-Lab software (Improvision, Inc., Lesington, MA) was used for all imaging analyses. Fission yeast cells were collected onto a regular glass slide and covered with cover slip. For the observation of green fluorescent protein, we used a Leica L5 filter, which has an excitation of 480/40 (460-500 nm) and emission of 527/30 (512-542 nm). Induction of cellular heat shock responses were conducted as previously described [35,59]. The level of hsp16 gene expression was quantified by measuring fluorescent signal emitted from the GFP-Hsp16 fusion protein as described previously [35,48,49]. For DNA staining, cells were counterstained with 1 μg/ml DAPI, which was observed with a Leica A8 filter with an excitation of 360/40 (340-380 nm) and emission of 470/40 (450-490 nm). Cell length was measured individually on the captured images using the OpenLab software. Statistical significance of differences in cell length was determined using the t-test for paired samples.

Western blot analysis
For Western blot analysis, mammalian cells were harvested and rinsed with ice-cold HEPES-buffered saline (pH 7.0), then lysed in an ice-cold cell lysis buffer [20 mM Tris-HCl, pH7.6, 150 mM NaCl, 1 mM EDTA, 0.5% Nonidet P-40, 1 mM DTT, 5 μM Trichostatin A, 1 mM sodium orthovanadate, 1 mM PMSF, 1 mM NaF and complete protease inhibitors (Roche Applied Science)]. Cellular lysates were prepared and the protein concentration was determined using the Pierce protein assay kit. For immunoblotting, an aliquot of total lysate (50 μg of proteins) in 2× SDS-PAGE sample buffer (1:1 v/v) was electrophoresed and transferred to a nitrocellulose filter. Filters were incubated with appropriate primary antibody in Tris-buffered saline (TBS, pH 7.5) and 5% skim milk or 5% BSA overnight. After washing, the filter was incubated with secondary antibody in TBS-Tween-20 (TBS-T) buffer for 1 h. Protein bands were visualized by an ECL detection system.
For Western blot analysis of fission yeast proteins, cells were washed once with water prior to adding cold stop buffer (150 mM NaCl, 50 mM NaF, 10 mM EDTA, 1 mM NaN 3 , pH 8.0). Cells were then collected by centrifugation and resuspended in 3 volumes of HB buffer (25 mM MOPS, pH7.2, 60 mM β-glycerophosphate, 15 mM pnitrophenyl phosphate, 15 mM MgCI 2 , 15 mM EGTA, 1% Triton X-100, 1 mM DTT). A mixture of protease inhibitors (1 mM PMSF, 20 μg/ml leupeptin, 40 μg/ml aprotinin and 0.1 mM sodium vanadate) and a commercial complete mini protease inhibitor cocktail (Roche, one tablet per 7 ml) was added immediately before lysis by glass-bead agitation. The cells were disrupted for 60 sec using a bead beater (Biospec Products, Bartlesville, OK). Cell breakage was checked under the microscope and disruption was repeated 3-5 times if necessary. Protein concentration was measured using the BCA protein assay kit (Pierce). Equal amounts of protein (30 μg) were resolved on a SDS-polyacrylamide gel and transferred to a nitrocellulose membrane. HIV-1 Vpr or Hsp16 protein levels were assayed by immunoblotting procedures using anti-Vpr serum (generated in our laboratory) and anti-Hsp16 serum (Paul Young's laboratory), respectively. Immunoblots were developed using the enhanced chemiluminescent (ECL) system (Pierce).

HIV-human immunodeficiency virus
Vpr-viral protein R

Hsp-heat shock protein
Hsf-heat shock factor