Translocation of the Dictyostelium TRAP1 homologue to mitochondria induces a novel prestarvation response

Tumor necrosis factor (TNF) receptor-associated protein 1 (TRAP-1) was initially identified using the yeast two-hybrid system as a novel protein that binds to the intracellular domain of the type 1 receptor for TNF (TNFR-1IC) (Song et al., 1995). The retinoblastoma protein (Rb) and tumor suppressors EXT1 and EXT2 have also been reported as its binding partners (Chen et al., 1996; Simmons et al., 1999). Additionally, Coller et al. have suggested that TRAP-1 is part of a pathway leading to MYC-mediated apoptosis (Coller et al., 2000). In summary, it appears that TRAP-1 plays roles in cell cycle progression, cellular differentiation and apoptosis, but its precise function remains to be elucidated. Although the TRAP-1 protein shows significant homology to the 90 kDa molecular chaperone Hsp90 and is predominantly located in mitochondria in several cell lines (as expected since it contains a mitochondrial localization sequence at its N-terminus), specific extramitochondrial localizations of TRAP1 have also been observed (Cechetto and Gupta, 2000). All partners identified as interacting with TRAP-1 are in fact extramitochondrial proteins, suggesting possible functions outside as well as inside mitochondria. We previously reported that the Dictyostelium homologue (Dd-TRAP1) of TRAP-1 predominantly localizes to the cell cortex together with F-actin during growth, and that it translocates to mitochondria early in differentiation (Morita et al., 2002).

Dictyostelium discoideum has a well-defined life cycle that provides a good system for elucidating the mechanisms governing the transition from growth to differentiation. Vegetative Dictyostelium discoideum cells grow and proliferate as long as external nutrients are available. Upon deprivation of nutrients, however, starving cells progress through the cell cycle to a particular point (putative shift point; PS-point) in the mid-late G₂ phase of cell cycle and enter the differentiation phase from this point (Maeda et al., 1989). Along with the initiation of the developmental program, cells acquire aggregation competence and EDTA-resistant cohesiveness to form aggregates by chemotaxis to cAMP (Gerisch, 1961; Bonner et al., 1969). Thus growth and differentiation are temporally separated from each other and easily controlled by nutritional conditions. Using this advantage we have identified several genes that are specifically expressed in response to the initial differentiation from the PS-point (reviewed by Maeda et al., 2002). By contrast, a part of the differentiation program has been shown to begin before starvation (prestarvation response, PSR); premature expression of early developmental genes [prestarvation genes; for example, dscA (discoidin I) and car1 (carA; cAMP receptor)] is induced as the density of growing cells increases (Clarke et al., 1987; Rathi and Clarke, 1992). Vegetatively growing cells sense their own cell density by secreting factors (prestarvation factors) that induce the expression of the prestarvation genes when they accumulate above a threshold level, and thus allow the relatively sharp onset of development. Although two kinds of prestarvation factors (PSF and PSF′, renamed PSF-1 and PSF-2, respectively) have been reported (Clarke et al., 1987; Maeda and Iijima, 1992), a detailed signal transduction pathway of PSR remains to be determined.

As presented here, Dd-TRAP1 translocates from the cell cortex to mitochondria as the density of growing cells increases, and allows the prompt transition of cells from growth to differentiation through a novel prestarvation factor (PSF-3) in growth medium. From the analyses of conditional knockdown or overexpressing mutants of Dd-TRAP1, it is evident that Dd-TRAP1 is coupled to a novel PSR, including induction of differentiation associate-1 (dia1) and dia3 expression but not discoidin I and car1 expression as the density of growing cells increases.

Vegetative cells of Dictyostelium discoideum Ax-2 (clone MS) were grown axenically in growth (PS)-medium, as previously described (Morita et al., 2002). Transformed cells overexpressing Dd-TRAP1 were grown axenically by shake culture in PS-medium containing 50 μg/ml of G418. To allow cells to differentiate, growth-phase cells at different cell densities were harvested, washed twice in BSS (10 mM NaCl, 10 mM KCl, 2.3 mM CaCl₂) (Bonner, 1947) as starvation medium, and plated either in a 24-well titer plate at a density of 5×10⁵ cells/cm² or on 1.5% non-nutrient agar at a density of 2.5×10⁵ cells/cm², followed by incubation at 22°C. Conditioned growth medium (CGM) was prepared as follows: Ax-2 cells were grown in PS-medium up to a high cell density (>8×10⁶ cells/ml), and the supernatant of culture were withdrawn by centrifugation as CGM.

pDNeo2 (Witke et al., 1987) was used as the original vector for the preparation of Dd-TRAP1-overexpressing transformants. This vector was digested by BamHI and PstI, and then ligated overnight with the dd-trap1 gene that had been prepared by digesting a cDNA clone SLB414 (a kind gift from the Dictyostelium cDNA project in Japan) with BamHI and PsII, to obtain pDNeo2-Dd-TRAP1. The pDNeo2-Dd-TRAP1 vector was introduced into Ax-2 (clone MS) cells by electroporation as described previously (Howard et al., 1988). Transformed cells were selected in PS-medium containing 10 μg/ml of G418. After 1 week of the selection, the transformants were cloned and cultured in PS-medium containing 50 μg/ml of G418.

Fragments of the dd-trap1 gene for RNAi were obtained by PCR with primers containing appropriate restriction sites on their ends (trap1737, 5′-cccaagcttggatccgaattcgtcaacggcgttat-3′; trap-539f, 5′-aaactgcagttgaggatgcatcaattcca-3′; trap-981f, 5′-aaactgcagattcccaatcgctttgaatg-3′). The first PCR product (amplified with trap1737 and trap-539f) was ligated tail to tail with the second PCR product (amplified with trap1737 and trap-981f). This construct was inserted into the vector MB38, which is the response plasmid of tetracycline-regulated gene expression system (tet-off system) for Dictyostelim cells (TRAP1-RNAi MB38) (Blaauw et al., 2000). The transactivator plasmid of tet-off system (MB35) was initially introduced into Ax-2 cells by electroporation, and the transformed cells (MB35 cells) were selected and cloned with 30 μg/ml of G418. The TRAP1-RNAi MB38 vector was introduced into MB35 cells by electroporation, and double-transformed cells (TRAP1-RNAi cells) were selected and cloned with 30 μg/ml of G418 and 10 μg/ml of blasticidin S. TRAP1-RNAi cells were subcultured in PS-medium with 30 μg/ml of G418, 10 μg/ml of blasticidin S and 20 μg/ml of tetracycline (Sigma). To monitor the development of TRAP1-RANi cells with food bacteria, the bacterial lawn was prepared by incubating Escherichia coli (B/r) on 3LP plate (0.3% Bacto peptone (Difco), 0.3% lactose, 2% agarose). Then, TRAP1-RANi cells cultured axenically in the absence of tetracycline for 2 days were placed as a small droplet on the bacterial lawn and incubated for 2 days at 22°C to allow development.

Total RNAs were prepared from vegetatively growing cells or cells starved in non-nutrient medium, using TRIzol Reagent (Life Technologies) as described by the manufacturer (GibcoBRL). Northern hybridization was carried out as previously described (Inazu et al., 1999).

Cells that had been allowed to adhere to coverslips were fixed twice with 99.6% methanol including 0.1% glutaraldehyde (10 minutes for each). The fixed samples were stained with the anti-Dd-TRAP1 antibody, as previously described (Morita et al., 2002). In the case of double-staining with MitoTracker, cells were incubated in BSS containing 1 μM MitoTracker Orange CMTMRos (Molecular Probes) for 30 minutes before the fixation.

Ax-2 cells growing at a low cell density (1.0-1.5×10⁶ cells/ml) or a high cell density (1.0-1.5×10⁷ cells/ml) were separately harvested and suspended in electroporation buffer [EB; 10 mM phosphate buffer (pH 6.2), 50 mM sucrose] at a density of 2×10⁷ cells/ml. 400 μl of the cell suspension were placed on ice for 10 minutes, and then electroporated in EB containing 5 μl of the Dd-TRAP1 antibody, non-immune serum or BSS. The samples were placed on ice for 10 minutes to heal the damage of electroporation, and washed with PS-medium or BSS. To allow cells to differentiate, starved cells were plated in a 24-well titer plate at a density of 5×10⁵ cells/cm² and incubated at 22°C.

In cells growing at a low cell density (<5×10⁵ cells/ml), Dd-TRAP1 was mainly localized in the cortical region, as previously reported (Fig. 1A). When the cell density reached over 2-3×10⁶ cells/ml, Dd-TRAP1 translocated to the mitochondria, although a small amount was retained in the cell cortex (Fig. 1A,B). This behavior of Dd-TRAP1 raises the possibility that growing cells secrete a factor(s) that induces Dd-TRAP1 translocation, and that this factor(s) accumulates in growth medium as the cell density increases. To examine the effect of secreted factors on Dd-TRAP1 translocation, Ax-2 cells growing at low cell density (5×10⁵ cells/ml), in which Dd-TRAP1 was localized in the cell cortex, were harvested and incubated in conditioned growth medium (CGM) in which Ax-2 cells had been grown up to the late exponential growth phase (8×10⁶ cells/ml). As a result, Dd-TRAP1 was found to quickly translocate to mitochondria within 1 minute of incubation even at low density (Fig. 1C), suggesting that the CGM contains a prestarvation factor(s) that induces Dd-TRAP1 translocation. Maeda and Iijima reported that when cells growing at a high cell density (5×10⁶ cells/ml) are transferred to fresh growth medium, their developmental competence, acquired by the prestarvation response (PSR), is lost within 30 minutes of incubation (Maeda and Iijima, 1992). Similarly, when Ax-2 cells growing at the late exponential growth phase (8×10⁶ cells/ml) were transferred to fresh growth medium, Dd-TRAP1 located in mitochondria quickly returned to the cell cortex within 30 minutes of incubation in fresh growth medium, as shown in Fig. 1D.

To analyze whether Dd-TRAP1 is involved in PSR, we have tried repeatedly to isolate Dd-TRAP1-null cells by homologous recombination. Although many clones (n=30) with the Bsr-resistance gene cassette were analyzed by western blotting using the Dd-TRAP1 antibody, all of them expressed Dd-TRAP1 at normal levels (data not shown). We also attempted to isolate transformants in which the expression of Dd-TRAP1 was suppressed by antisense RNA-mediated gene inactivation. In spite of many trials, transformants with strongly reduced Dd-TRAP1 levels were never obtained. Therefore, it is most likely that Dd-TRAP1 is essential for vegetative growth, as is the case for other members of the Hsp90 family. We then tried to isolate Dd-TRAP1 conditional knockdown transformants by a combination of RNA interference technique (RNAi) and tetracycline-regulated gene expression system (tet-off system) (Martens et al., 2002; Blaauw et al., 2000). We designed a construct for expression of double-strand RNA (dsRNA) directed against the middle part of dd-trap1 (Fig. 2A), and introduced this construct into MB35 cells that had been transformed by the transactivator plasmid of the tet-off system to produce TRAP1-RNAi cells. In TRAP1-RNAi cells, the expression of the dsRNA was regulated by the tet-off system. In the presence of 20 μg/ml of tetracycline, the Dd-TRAP1 expression was scarcely suppressed and TRAP1-RNAi cells grew normally with a doubling time of 8.9 hours (doubling time of parental MB35 cells; 8.4 hours) (Fig. 2B). However, the level of Dd-TRAP1 expression in TRAP1-RNAi cells was gradually decreased during incubation in the absence of tetracycline, reaching approximately 10% of their initial value by 3 days. As the level of Dd-TRAP1 expression was decreased, the growth rate of TRAP1-RNAi cells declined considerably (t; >24 hours) (Fig. 2B). However, the effects of the dsRNA on Dd-TRAP1 expression and cell growth were gradually nullified after 5 days of incubation. This might be due to the appearance of cells in which Dd-TRAP1 expression was less suppressed allowing them to grow normally. In addition, the expression levels of other members of the Dictyostelium Hsp90 family, Dd-GRP94 and Hsc90, did not change during the whole course of incubation, which suggests that the dsRNA of dd-trap1 does not have any non-specific effects.

When starved MB35 cells were developed under submerged conditions, they acquired an aggregation competence and aggregated into streams after 6-7 hours of starvation (Fig. 3A). By contrast, starving TRAP1-RNAi cells, which had been cultured for 2 days in the absence of tetracycline to suppress the expression of Dd-TRAP1, showed considerably delayed differentiation (Fig. 3A). To determine whether TRAP1-RNAi cells can induce a prestarvation response, we examined the expression pattern of several prestarvation genes in axenically growing TRAP1-RNAi cells. As previously described (Wetterauer et al., 1995), the expression of discoidin I and car1 genes was induced by a prestarvation response in axenically growing MB35 cells. Unexpectedly, however, no significant differences in their expression were detected between growing TRAP1-RNAi cells and MB35 cells. The dia1 and dia3 genes were initially isolated as differentiation-associated genes, specifically expressed in cells differentiating from the PS point (Inazu et al., 1999; Hirose et al., 2000). In the present work, we found that these genes are moderately expressed even in vegetative Ax-2 cells, and their expression gradually increased and reached a peak at the late exponential growth phase, followed by a decrease or disappearance at the stationary growth phase. This finding indicates that dia1 and dia3 are prestarvation genes as well as differentiation-associated genes. Interestingly, the prestarvation response did not induce expression of dia1 and dia3 in TRAP1-RNAi cells (Fig. 3B). In addition, when TRAP1-RNAi cells were spotted on the bacterial lawn to allow them to develop, they formed aggregates much further away from the feeding edge than those formed by parental MB35 cells (Fig. 3C). This strongly suggests that growing TRAP1-RNAi cells fail to calculate correctly the cell density of their population relative to the density of the remaining nutrients because of the defect in their prestarvation response.

We have isolated TRAP1-overexpressing (TRAP1^OE) cells by introduction of the vector (pDNeo2-Dd-TRAP1) into Ax-2 cells. After a densitometric analysis of western blots, the expression level of Dd-TRAP1 in TRAP1^OE cells was found to be approximately three times higher than in parental Ax-2 cells. Importantly, in TRAP1^OE cells the early localization of Dd-TRAP1 to mitochondria occurred even at a low cell density (0.5-1.0×10⁶ cells/ml) (Fig. 4A). We examined the expression patterns of prestarvation genes (discoidin I, car1, dia1 and dia3) in axenically growing TRAP1^OE cells. As was expected from the above results obtained using TRAP1-RNAi cells, the expression patterns of discoidin I and car1 in TRAP1^OE cells were the same as those in parental Ax-2 cells (Fig. 4B). By contrast, the expression of dia1 and dia3 was significantly increased in TRAP1^OE cells, possibly through enhancement of a PSR pathway via overexpression of Dd-TRAP1 (Fig. 4B). TRAP1^OE cells exhibited a growth rate slightly slower than that of parental Ax-2 cells and occasionally formed cell aggregates even in growth medium, which were never seen with parental Ax-2 cells (Fig. 4C). This phenotype is somewhat similar to that of transformants exhibiting early PSR (Souza et al., 1998; Colosimo and Katz, 2001). When these TRAP1^OE aggregates were treated with 10 mM EDTA, they did not dissociate (data not shown), indicating that the cell contacts are mediated by EDTA-resistant adhesion molecules, which are usually expressed only during the differentiation phase (Gerisch, 1961). Moreover, when single cells that had not participated in aggregate formation were gently shaken in 16 mM phosphate buffer (pH 6.4), they rapidly acquired the EDTA-resistant adhesiveness (Fig. 4D). To test whether TRAP1^OE cells forming aggregates can take in external nutrients for proliferation, they were cultured for 60 minutes in PS-medium containing 3 mg/ml of FITC-dextran (FD50) as a fluid-phase marker to monitor their pinocytotic activity. Since little or no pinocytotic activity was detected in TRAP1^OE cells, they must be starving even in the presence of external nutrients. More notably, some cells in the aggregate were found to contain prespore-specific vacuoles (PSVs), the organelles specifically present in prespore cells (data not shown). These results indicate that at least a proportion of TRAP1^OE cells undergo an early transit into the differentiation phase even in the presence of nutrients, possibly through PSR-enhancement.

The 65-70 kDa glycoprotein (PSF, renamed PSF-1) is the best-established prestarvation factor (Clarke et al., 1987; Clarke et al., 1988). Its activity is usually measured as the increased discoidin I expression in growth phase cells, but Dd-TRAP1 is not involved in the expression of discoidin I as described in the preceding section (Fig. 3B, Fig. 4B). Burdine and Clarke have reported that prestarvation genes such as discoidin I and PDE (cAMP phosphodiesterase; pdsA) are barely induced in PKAcat-null cells, but their expression is normal in Gβ-null cells, suggesting that the PSR as assayed by discoidin I expression is regulated by PKA, but not by the G-protein β subunit (Burdine and Clarke, 1995). Importantly, the translocation of Dd-TRAP1 to mitochondria is observed both in Gβ-null and in PKAcat-null cells (Fig. 5A), suggesting that neither is required for the translocation of Dd-TRAP1 to mitochondria. PSF-1 is sensitive both to protease and to heat treatment (Clarke et al., 1988; Rathi and Clarke, 1992), but another prestarvation factor (PSF′, renamed PSF-2) (Maeda and Iijima, 1992) is fairly stable to heat treatment. Conditioned growth medium (CGM) treated with protease (500 μg/ml of pronase E) for 1 hour at 37°C was found to retain activity, but activity was lost if it was boiled for 15 minutes before use (Fig. 5B). In addition, a crude fraction obtained by 80% saturated (NH₄)₂SO₄-fractionation of CGM was found to have a potent capacity for transferring the cortical Dd-TRAP1 into mitochondria. These results indicate that Dd-TRAP1 translocation from the cell cortex to mitochondria is induced by a novel prestarvation factor (referred to as PSF-3), differing from PSF-1 and PSF-2.

Starving Ax-2 cells aggregated into streams after 7-8 hours of incubation and then formed tight mounds under submerged conditions (Fig. 6Aa,b). By contrast, starving TRAP1^OE cells developed more slowly and remained as aggregation-streams even after 24 hours of starvation (Fig. 6Ac,d), even though they had experienced an early PSR during growth. The aggregation-streams formed by TRAP1^OE cells never developed to tight mounds during a prolonged time (3 days) of starvation. In addition, 30% or more plated TRAP1^OE cells remained as non-aggregated single cells. Such abnormal morphogenesis was also observed when starved TRAP1^OE cells were incubated on 1.5% non-nutrient agar (Fig. 6Ae-h). To examine how the enforced expression of Dd-TRAP1 leads to a delay in development, the expression patterns of several developmental genes in TRAP1^OE cells were compared with those in parental Ax-2 cells. As shown in Fig. 6B, car1 expression was only slightly reduced at 2-4 hours of starvation in TRAP1^OE cells, and dia3 expression was severely suppressed until 6 hours of starvation. The expression of the gene lagC, which is a postaggregative gene, was also delayed in TRAP1^OE cells, thus being consistent with their slow aggregation. By contrast, expression of dia1 and csA genes was found to be greatly increased in TRAP1^OE cells. Expression of dia1 has been reported to negatively regulate the transition from growth to differentiation (Hirose et al., 2000). To know if the delayed morphogenesis of TRAP1^OE cells is caused by overexpression of the dia1 gene, Dd-TRAP1 was overexpressed in a dia1-null mutant (TRAP1^OE/dia1^– cells). Under submerged conditions, aggregation of starved TRAP1^OE/dia1^– cells was delayed and many of them remained as single cells even after 22 hours of starvation, as is the case for TRAP1^OE cells (Fig. 6Ca-d). However, a few populations of TRAP1^OE/dia1^– cells were able to form tiny tight mounds, in contrast to TRAP1^OE cells, which never formed tight mounds under submerged conditions. This slight difference between TRAP1^OE cells and TRAP1^OE/dia1^– cells might be due to the difference in clones used as parental cells: dia1^– cells are derived from Ax-2 (clone 8A) cells, whereas TRAP1^OE cells are from Ax-2 (clone MS) cells. The 8A cells develop almost normally without delay when harvested at a relatively low cell density of around 1×10⁵ cells/ml and starved. Actually, Dd-TRAP1^OE cells isolated from Ax-2 (clone 8A) cells were also capable of forming tiny tight mounds under submerged conditions, although their developmental program was markedly delayed. When starved TRAP1^OE/dia1^– cells were developed on 1.5% non-nutrient agar, they needed a longer time to aggregate than parental dia1^– cells (Fig. 6Ce-h). Moreover, TRAP1^OE/dia1^– cells formed numerous aggregation centers and hence many tiny mounds, resulting thus in the formation of tiny fruiting bodies.

To know whether the delay of differentiation in TRAP1^OE cells is reversed by exogenously added cAMP pulses, cells growing at 1-3×10⁶ cells/ml were harvested and starved for 6 hours with or without exogenous cAMP pulses under submerged conditions. Pulsed TRAP1^OE cells acquired aggregation-competence earlier (after 6.5 hours of starvation) than those without cAMP pulses (Fig. 7A). Therefore, a defect in the cAMP signaling system may be one of the causes for delayed differentiation in TRAP1^OE cells. However, TRAP1^OE cells that were starved with cAMP pulses were unable to form tight mounds even after 24 hours of starvation under submerged conditions (Fig. 7A). Further, the expression of car1 was barely induced by cAMP pulses in TRAP1^OE cells, at least during 4 hours of starvation (Fig. 7B).

Fig. 7C shows the relation of cell densities in growth medium at the time-point of starvation to elapsed time required for acquisition of aggregation-competence. When Ax-2 (clone MS) cells growing at 5×10⁵ cells/ml were harvested and starved, they needed 10 hours or more to acquire aggregation competence and failed to form tight aggregates under submerged conditions. Ax-2 cells growing above 1×10⁶ cells/ml, however, acquired aggregation competence after 5-8 hours of starvation (Fig. 7C), and they were able to form tight aggregates. As described in the preceding section, TRAP1^OE cells grown to less than 4×10⁶ cells/ml failed to acquire aggregation-competence until 10 hours or more of starvation and never formed tight mounds, as observed in Ax-2 cells growing at low cell densities (<1×10⁶ cells/ml) (Fig. 7C). However, when TRAP1^OE cells grown to relatively high densities (>5×10⁶ cells/ml) were harvested and starved, their developmental ability was largely restored, resulting in formation of tight mounds under submerged conditions (Fig. 7C). This raised the possibility that the phenotype observed in TRAP1^OE cells might be closely related to the subcellular localization of Dd-TRAP1 at the onset of starvation. To test this possibility, the localization of Dd-TRAP1 in starving TRAP1^OE cells was compared with that in parental Ax-2 cells. In Ax-2 cells harvested at low densities (<2×10⁶ cells/ml), Dd-TRAP1 translocation from the cell cortex to mitochondria began after starvation and was completed by 6 hours of starvation (Fig. 7Da) (Morita et al., 2002). In TRAP1^OE cells harvested at a high density (8×10⁶ cells/ml), however, most Dd-TRAP1 was already translocated to mitochondria before starvation, and translocation was complete just after starvation (Fig. 7Dc). Here it is worthy to note that in TRAP1^OE cells harvested at low cell densities (1-2×10⁶ cells/ml) and starved, most of Dd-TRAP1 is translocated to mitochondria, but a significant amount of Dd-TRAP1 remained in the cell cortex even after 6 hours of starvation (Fig. 7Db).

We also attempted to inhibit the function of Dd-TRAP1 by introducing directly the Dd-TRAP1 antibody into cells. For this, Ax-2 cells growing at a low cell density (1.0-1.5×10⁶ cells/ml) were harvested and electroporated in EB containing 5 μl/ml of anti-Dd-TRAP1 serum. For the controls, cells were electroporated with the non-immune serum or without any sera. After electroporation, cells were incubated either in growth medium or in starvation medium to observe the effects of Dd-TRAP1 antibody on cell proliferation and differentiation. The cells electroporated with the non-immune serum or the Dd-TRAP1 antiserum grew at the same rate as those electroporated without any sera, suggesting that the incorporated serum has no toxic effect on cells. By contrast, cells into which the Dd-TRAP1 antibody had been introduced exhibited delayed aggregation and never formed tight aggregates, even after 24 hours of incubation (Fig. 8A). This effect was caused by the inhibition of the Dd-TRAP1 translocation. In Ax-2 cells with the Dd-TRAP1 antibody, a significant amount of Dd-TRAP1 remained in the cell cortex even after 6 hours of starvation, whereas Dd-TRAP1 was completely translocated to mitochondria in the control cells (Fig. 8B). To determine whether the inhibitory effect of the Dd-TRAP1 antibody on differentiation is due to failure of Dd-TRAP1 to be translocated to mitochondria, we introduced the Dd-TRAP1 antibody into Ax-2 cells growing at a high cell density (1.0-1.5×10⁷ cells/ml), in which Dd-TRAP1 had already been translocated to mitochondria. Our results clearly showed that the introduced Dd-TRAP1 antibody has no effect on early differentiation of the starved cells (Fig. 8C).

TRAP-1 was initially identified as a novel protein that binds to TNF receptor type 1 (Song et al., 1995). Although several studies suggest that TRAP-1 plays roles in cell cycle progression, cellular differentiation and apoptosis, this is the first report demonstrating a crucial function of TRAP-1 as a mediator of the prestarvation response (PSR) in Dictyostelium cells. The early aggregate formation, as noticed in TRAP1^OE cells, is observed in a transformant overexpressing YakA, a protein kinase required for the initiation of development (Souza et al., 1998). The expression of YakA is upregulated by PSR, and the overexpression of YakA during growth induces an arrest of cell cycle and the formation of small cell clumps (Souza et al., 1998). In HK19 cells having an increased sensitivity to prestarvation factors, aggregate formation is induced by exogenously applied cAMP pulses even in the presence of food bacteria (Colosimo and Katz, 2001).

Clarke et al. (Clarke et al., 1987; Clarke et al., 1988) have identified a 65-70 kDa glycoprotein as one of the prestarvation factors (PSF-1). Most studies concerning PSF-1 have monitored the expression level of the gene discoidin I (Wetterauer et al., 1993; Zeng et al., 2000). Another prestarvation factor (PSF-2) has been found by Maeda and Iijima (Maeda and Iijima, 1992). We report here a novel pathway of PSR that is mediated by Dd-TRAP1 and differs from the previously reported pathway: (1) the novel PSR is not accompanied by increased expression of discoidin I; (2) PSF-3, which is a novel prestarvation factor for the activity of Dd-TRAP1 translocation, has distinct chemical properties from PSF-1 and PSF-2; (3) although PSR requires the activity of PKA (as realized by discoidin I expression), the PKA activation is not involved in the novel pathway. Originally dia1 and dia3 genes were reported to be specifically expressed in response to the initial differentiation of Ax-2 cells from the PS-point (Inazu et al., 1999; Hirose et al., 2000). However, in this study these genes were found as prestarvation genes as well as differentiation-associated genes. Furthermore, four PS-point-specific genes have also been regarded as prestarvation genes [dia1 and dia3, in this work; car1/quit1 (Rathi and Clarke, 1992); dia2 (S. Hirose, C. Pears, T. Mayanagi, A.A. and Y.M., unpublished). Huang and Pears have reported that the discoidin I gene is preferentially expressed in cells starved at the G₂ phase near the PS-point in the cell cycle, which suggests that discoidin I is also a PS-point-specific gene (Huang and Pears, 1999). These facts suggest that the biological significance of PSR is to allow readily the cells to exist from the PS-point toward the differentiation phase in response to increased cell density. Actually, TRAP1-RNAi cells formed aggregates far apart from the feeding edge, where food bacteria are still present. In addition, HBW3 (a chemically induced mutant with an unknown molecular defect that expresses discoidin I and develops rapidly) or gdt1^– cells (expresses discoidin I early and at high levels and prematurely enters the differentiation pathway), which show early PSR, form aggregates close to the feeding edge on the bacterial lawn (Primpke et al., 2001). Interestingly, Araki et al. have reported that cells starved at the mid-late G₂ phase (just before the PS-point) differentiate into viable spores, whereas the cells starved at the late G₂ phase (just after the PS-point) differentiate into non-viable stalk cells (Araki et al., 1997). Therefore, the earlier entry to differentiation phase by PSR may be strategically advantageous for survival of cells to the next generation.

Although Dd-TRAP1 induced the prestarvation response with enhanced expression of differentiation-associated genes, TRAP1^OE cells exhibited the delayed differentiation after starvation (Fig. 6A). In TRAP1^OE cells, the bulk of Dd-TRAP1 was translocated to mitochondria to induce the premature prestarvation response, but a significant amount of Dd-TRAP1 was still retained in the cell cortex for a long period of starvation (Fig. 7Da,b). This raised the possibility that Dd-TRAP1 located in the cortex might cause delayed differentiation. This idea is supported by the fact that introduction of the Dd-TRAP1 antibody into cells partially keeps Dd-TRAP1 in the cell cortex, resulting in delayed differentiation. The delayed differentiation of TRAP1^OE cells was partially canceled by exogenously applied cAMP pulses, but the delayed expression of car1 was not recovered by the pulses (Fig. 7A,B). This suggests Dd-TRAP1 may negatively regulate the cAMP signaling pathway upstream of car1 expression.

We have attempted to disrupt the dd-trap1 gene by homologous recombination, but failed to obtain dd-trap1-null cells. In general, the members of the Hsp90 family are necessary for cell growth, and it is difficult to prepare their knockout strains. Consequently, we attempted to isolate the Dd-TRAP1 conditional knockdown transformants by a combination of RNAi and the tet-off system (TRAP1-RNAi cells). In TRAP1-RNAi cells growing in the absence of tetracycline, the expression level of Dd-TRAP1 protein was markedly decreased, resulting in a growth rate much slower than that of TRAP1-RNAi cells grown in the presence of tetracycline and that of parental MB35 cells. This indicates that Dd-TRAP1 is necessary for cell growth, as is the case for other members of the Hsp90 family.

In the Hsp90 family, TRAP-1 has a mitochondrial localization sequence at its N-terminus (Song et al., 1995; Felts et al., 2000) and is located predominantly in mitochondria in several cell lines (Felts et al., 2000; Cechetto and Gupta, 2000). However, TRAP1 has also been identified as an interacting partner for several extramitochondrial proteins (the type 1 TNF receptor, TNFR-1; the retinoblastoma protein, Rb; and EXT1, EXT2) (Song et al., 1995; Chen et al., 1996; Simmons et al., 2000). In mammalian cells, extramitochondrial localization of TRAP1 has been observed in secretary granules, nuclei, and at the cell surface (Cechetto and Gupta, 2000). Mitochondrial Hsp60 has been found at a discrete extramitochondrial sites, including the cell surface, cytoplasmic vesicles and granules, peroxisomes, and ER (Cechetto et al., 2000). In Dictyostelium cells, newly synthesized Dd-TRAP1 (80 kDa) with a mitochondrial localization sequence is translocated to the cell cortex after a brief and temporal stay in mitochondria (Morita et al., 2002). Coupled with an increase of cell density at the growth phase, Dd-TRAP1 is again and rapidly translocated from the cell cortex to mitochondria by PSF-3-mediated PSR. The molecular mechanism by which the processed Dd-TRAP1 (73 kDa) devoid of any mitochondrial localization sequences is able to come back to mitochondria is presently puzzling. Importantly, Ledgerwood et al. have reported that exogenously added TNF is found to be delivered to mitochondria (Ledgerwood et al., 1998). This raises the possibility that in cells activated by TNF, TRAP-1 might transfer TNF itself and other related proteins from the cell surface to mitochondria.

We thank Rob Kay and Jean-Paul Rieu for their critical reading and insightful comments. We are grateful to the Dictyostelium cDNA project in Japan for kind gift of the cDNA clone SLB414. This work was supported by a Grant-in-Aid (14654170) from JSPS. This work was also funded by the Mitsubishi Foundation.