Cell-cycle checkpoint for transition from cell division to differentiation


Author to whom all correspondence should be addressed.
Email: kjygy352@ybb.ne.jp


In general, growth and differentiation are mutually exclusive, but they are cooperatively regulated during the course of development. Thus, the process of a cell’s transition from growth to differentiation is of general importance for the development of organisms, and terminally differentiated cells such as nerve cells never divide. Meanwhile, the growth rate speeds up when cells turn malignant. The cellular slime mold Dictyostelium discoideum grows and multiplies as long as nutrients are supplied, and its differentiation is triggered by starvation. A critical checkpoint (growth/differentiation transition or GDT point), from which cells start differentiating in response to starvation, has been precisely specified in the cell cycle of D. discoideum Ax-2 cells. Accordingly, integration of GDT point-specific events with starvation-induced events is needed to understand the mechanism regulating GDTs. A variety of intercellular and intracellular signals are involved positively or negatively in the initiation of differentiation, making a series of cross-talks. As was expected from the presence of the GDT point, the cell’s positioning in cell masses and subsequent cell-type choices occur depending on the cell’s phase in the cell cycle at the onset of starvation. Since novel and multiple functions of mitochondria in various respects of development including the initiation of differentiation have been directly realized in Dictyostelium cells, they are also reviewed in this article.


Growth and differentiation are fundamental characteristics of the cell. In general, they are mutually exclusive but are cooperatively regulated throughout development. Thus, the process of a cell’s switching from growth to differentiation is of great importance not only for the development of organisms but also for malignant transformation, in which this process is reversed. When most cells are terminally differentiated, they must exit the cell cycle. In general, there is a checkpoint (restriction point: RP) in G1-phase; cells that should cease division exit the cell cycle and enter the G0-phase. However, the precise cell-cycle-position for G1/G0 transition still remains to be specified.

Dictyostelium discoideum is a social amoeba whose life cycle consists of two distinct phases – growth and differentiation – that are easily controlled by nutritional conditions. D. discoideum (strain Ax-2) cells grow and multiply by mitosis as long as nutrients are supplied. Upon exhaustion of nutrients, however, starving cells initiate differentiation to acquire aggregation competence and form multicellular structures by means of chemotaxis toward cyclic adenosine monophosphate (cAMP) and ethylenediaminetetraacetic acid (EDTA)-resistant cohesiveness. Subsequently, the cell aggregate (mound) undergoes a series of well-organized movements and zonal differentiation to form a migrating slug. The slug eventually culminates to form a fruiting body consisting of a mass of spores (sorus) and a supporting cellular stalk. At the slug stage, a clear pattern along the anterior–posterior axis is established; prestalk cells, which finally differentiate into stalk cells during culmination, are located in the anterior one-fourth, while prespore cells destined to differentiate eventually into spore cells occupy the posterior three-fourths of the slug. The life cycle of Dictyostelium cells is relatively simple, but it contains almost all of the cellular processes (movement, adhesiveness, differentiation, pattern formation, etc.) essential for the establishment of multicellular organization. In basically haploid Dictyostelium cells, gene disruptions by homologous recombination are available for analysis of precise gene functions. Insertional mutagenesis by the restriction enzyme-mediated integration (REMI) method has also been established to isolate and characterize intrigued functional genes (Kuspa & Loomis 1992).

Similarly to most higher eukaryotic cells, extracellular signals control the transition of Dictyostelium cells from growth to differentiation. The elucidation of these signals and their pathway toward the switch in the genetic program must provide insights into general mechanisms for the initiation of cell differentiation. During the growth phase, Dictyostelium cells continuously synthesize and secrete autocrine factors that accumulate in a cell-density-dependent manner. At appropriate concentrations these factors induce changes in gene expression and prepare cells for the initiation of differentiation. Cells are able to detect the levels of prestarvation factors (PSFs) secreted by growing cells and prepare the initial differentiation before the onset of starvation (Clarke et al. 1988, 1992; Maeda & Iijima 1992; Morita et al. 2004). The “prestarvation response” (PSR) occurs during increases in PSF levels and decreases in nutrients, and can be detected a few generations before actual starvation occurs. When external nutrients are depleted and cells stop growing, starving cells secrete another glycoprotein, conditioned medium factor (CMF), which is essential for establishment of cAMP signaling and the initiation of cell aggregation (Gomer et al. 1991; Yuen et al. 1995).

Using the temperature shift method for cell synchronization, we have precisely specified a checkpoint (a GDT point; formerly referred to as a PS point) of growth/differentiation transition (GDT) in the cell cycle of a D. discoideum cell (Fig. 1), and demonstrated cell-cycle phases at the onset of starvation are particularly critical for the cell’s positioning in cell masses and subsequent cell-type choice (Maeda 2005). In this review, I will survey cellular and molecular events occurring during the GDT in Dictyostelium cells, and illustrate the mechanism of cell cycle-dependent pattern formation and also the significance of mitochondria in a variety of cellular activities including the initiation of differentiation.

Figure 1.

 A growth/differentiation checkpoint (GDT point) in the cell cycle of a Dictyostelium discoideum Ax-2 cell. The doubling time of axenically growing Ax-2 cells is about 7.2 h and most of their cell cycle is composed of G2-phase with little or no G1-phase and a short period of M- and S-phases. A specific checkpoint (referred to as the GDT point) of GDT is located at the mid–late G2-phase (just after T7 and just before T0). Ax-2 cells progress through their cell cycle to the GDT point, irrespective of the presence or absence of nutrients, and enter the differentiation phase from this point under starvation conditions (Maeda et al. 1989). T0, T1, and T7 indicates 0, 1, and 7 h, respectively, after a temperature shift from 11.5 to 22.0°C for cell synchrony. (Basically from Maeda 2005).

Growth/differentiation transition point (GDT-point) in the cell cycle

Cell proliferation is finely regulated by extracellular signals such as growth factors, and there are some checkpoints monitoring the exact progression of the cell cycle, e.g. the G2-phase checkpoint for DNA damage (Hartwell & Weinert 1989) and the M-phase for spindle formation (Chen et al. 1998; Jin et al. 1998). It has been shown that a specific checkpoint regulating the transition from growth to differentiation exists in the G1-phase (Sherr 1996). However, the cell-cycle position of the checkpoint for the GDT has not been precisely determined yet. Based on much experimental data obtained by synchronized D. discoideum Ax-2 cells, we have succeeded in specifying a strict checkpoint of GDT at the mid–late G2-phase of the cell cycle, as shown in Figure 1 (Maeda et al. 1989; Maeda 1993, 2005).

Evidence showing the existence of a GDT-point in the cell cycle

During axenic growth of Ax-2 cells by shake culture, cells have a doubling time of 7–8 h at 22.0°C. Several studies including autoradiographic and fluorimetric analyses (Ohmori & Maeda 1987) have revealed that the cell cycle of vegetative Ax-2 cells is composed of (i) a very short period of M-phase (15 min or less); (ii) little or no G1-phase; (iii) 30 min or less of S-phase; and (iv) a long period (about 6.5 h) of G2-phase, though some studies have claimed the presence of G1-phase (Katz & Bourgignon 1974; Zada-Hames & Ashworth 1978b; MacDonald & Durston 1984; Azhar et al. 2001). If true, the absence of a G1-phase in the cell cycle is not specific to Ax-2 cells, and is generally noticed in rapidly dividing cells such as embryonic cells and Amoeba proteus.

For precise analysis of cell-cycle-related events, good methods are required for inducing synchronous growth of cells. In cultures of Ax-2 cells, three main methods have been used for cell synchrony: (i) a mitosis wash-off procedure (Durston et al. 1984); (ii) a stationary phase release method in which the stationary phase cells are released from growth inhibition by transfer into fresh medium (Weijer et al. 1984a; Wang et al. 1988); and (iii) a temperature shift (cold-shock) method in which exponentially growing cells at 22.0°C are shifted to a low temperature of about 11.5°C, shaken for 20.0 h, and then reincubated to 22.0°C (Maeda 1986). The wash-off procedure and the stationary phase release method result in only partial synchrony. Stationary-phase cells are arrested at a relatively critical point of the G-2 phase and grow unimodally in fresh growth medium over a 2–3 h period after a lag phase of 1–2 h (Weijer et al. 1984a; Maeda 1986). In contrast, the temperature-shift method gives highly synchronous growth of cells; after the reshift of low-temperature-treated cells to 22.0°C, cell doubling occurs over about a 2-h period after a lag phase of about 1 h. Cell synchrony induced by the cold shock is well maintained at least until the beginning of the second M-phase. This method is simple and practical, thus being useful for detailed studies at the cellular and molecular levels on growth regulation, GDT, and cell-cycle-dependent pattern formation of developing cells. The cell cycle of Ax-2 cells was also found to proceed in differentiating prespore cells during the mound-tipped aggregate stage as well as during the vegetative growth phase (Zimmerman & Weijer 1993; Araki & Maeda 1998). Recently, Muramoto & Chubb (2008) have demonstrated using Ax-2 cells labeled by a live-cell S-phase marker that there is a cell-cycle checkpoint that operates at the G2/M transition in response to DNA damage, and that a large proportion of cells undergo nuclear DNA synthesis during multicellular development. The cell cycle progression during the multicellular stage appears to occur coupled with prespore differentiation, and has a principal implication for cell sorting recognized during pattern formation, as discussed later.

When T7 cells (7 h after the shift up from 11.5 to 22.0°C) are starved, they initiate the differentiation more rapidly compared to cells at other cell-cycle phases, acquiring the chemotactic sensitivity to cAMP and EDTA-resistant cohesiveness (Ohmori & Maeda 1987). This is consistent with the fact that starved T7 cells exhibit the earliest expression of the cAMP receptor1, carA (Abe & Maeda 1994), as described in the next section. In contrast to T7 cells, T0 and T1 cells require a longer time for differentiation. Here it is of importance to note that T1+6 cells (T1 cells starved and shaken for a further 6 h in the absence of external nutrients) exhibit essentially the same developmental features as T7 cells (Maeda et al. 1989), thus indicating that T1 cells progress through their cell cycle up to the phase of T7, independently of nutritional conditions, during the 6 h of incubation. That is, a critical GDT point (formerly referred to as the PS point) is located just after T7 and just before T0, and Ax-2 cells at any cell-cycle position progress in their cell cycle to the GDT point, irrespective of the presence or absence of nutrients, and enter the differentiation phase from this point under starvation conditions (Fig. 1).

Molecular events occurring around the GDT

We analyzed gene expressions associated with the initial step of GDT using the temperature-shift method for cell synchrony. Six genes (quit1, 2, 3 and dia1, 2, 3) were isolated as being expressed specifically or differently in Ax-2 cells starved just before the GDT point by means of differential plaque hybridization and differential display (Abe & Maeda 1994, 1995; Okafuji et al. 1997; Chae et al. 1998; Inazu et al. 1999; Hirose et al. 2000). That is, genes expressed specifically or differently in T7+2 cells (T7 cells starved for 2 h; differentiating cells from the GDT-point) but not in T1+2 (T1 cells starved for 2 h cells; cells being around the M-phase of the cell cycle), T4+2 cells (T4 cells starved for 2 h; cells; cells being just before the GDT-point) and T9 cells (T7 cells further grown for 2 h in growth medium; growth-phase cells) were screened and isolated. The phosphorylated levels of some proteins have been also shown to be markedly altered, coupled with differentiation of cells from the GDT point. These molecular events, specifically or predominantly occurring during GDT, are summarized in Table 1.

Table 1.   Molecular events specifically induced around the GDT-point in the cell cycle of Dictyostelium discoideum Ax-2 cells
GenemRNA or proteinExpression patternReferences
carAcAMP receptor 1 (CAR1)Specifically expressed in response to starvation around the GDT-pointAbe & Maeda (1994)
dia3Mitochondrial proteins including ribosomal protein S4 (RPS4)Specifically expressed in response to starvation around the GDT-pointInazu et al. (1999); Hosoya et al. (2003); Chida et al. (2008)
dia2A novel protein (DIA2; 16.9 kDa)Specifically expressed in response to starvation around the GDT-pointChae et al. (1998); Hirata et al. (2008)
dia1A novel protein (DIA1; 48.6 kDa)Specifically expressed in response to starvation around the GDT-pointHirose et al. (2000, 2005)
caf1A novel Ca2+-binding protein (CAF-1; 19.5 kDa)Predominantly expressed in response to starvation around the GDT-pointAbe & Maeda (1995); Itoh et al. (1998)
AnVIIAnVII mRNA (encoding annexin VII)Predominantly expressed in response to starvation around the GDT-pointBonfils et al. (1994)
discADiscoidin IPredominantly expressed in response to starvation around the GDT-pointDoring et al. (1995)
Quit3Anti-AnVII mRNA (no coding region)Predominantly expressed during the vegetative growth phaseOkafuji et al. (1997)
PhosphoproteinChange in the phosphorylated stateReferences
Elongation factor-2 (EF-2A; 101 kDa)Specifically failed to be phosphorylated in response to starvation around the GDT-pointAkiyama & Maeda (1992); Watanabe et al. (2003); Yoshino et al. (2007)
Heat-shock protein 90 (GRP94; 90 kDa)Specifically failed to be phosphorylated in response to starvation around the GDT-pointAkiyama & Maeda (1992); Morita et al. (2000)
48 kDa proteinSpecifically shifted to 50 kDa in response to starvation around the GDT-pointFurukawa & Maeda (1994)

Cyclic AMP receptor 1 (CAR1).  The coding region of quit1 is identical to that of the cAMP receptor 1 (carA) gene (Abe & Maeda 1994), which is specifically expressed in T7+2 cells, but not in T1+2, T4+2 and T9 cells. CAR1, a G-protein-linked surface cAMP receptor, exerts a central role in Dictyostelium development including cell aggregation; its disruption by homologous recombination and antisense RNA results in the failure of transformed Ax-3 cells to differentiate (Sun et al. 1990; Sun & Devreotes 1991), thus providing evidence of the role of CAR1 in the exit of cells into differentiation and also the real existence of the GDT point in the Dictyostelium cell cycle.

DIA3 (mitochondrial ribosomal protein S4; mt-RPS4).  A mitochondrial gene cluster (nad11, nad5, rps4, rps2, and nad4L) including ribosomal protein S4 (rps4), is specifically expressed in response to starvation around the GDT point and plays important roles in the initiation of cell differentiation in Ax-2 cells. The rps4 gene is present as a single copy in mt-DNA, but the copy number must be multiple, because numerous mitochondria are contained in a cell. In spite of this situation, we tried homologous recombination to determine the function of mt-rps4, by inactivating the subpopulation of the mt-rps4 gene (Inazu et al. 1999). As was expected, partial disruption of mt-rps4 was found to cause impaired differentiation, thus resulting in the failure of many cells to aggregate even after a prolonged time of starvation (Inazu et al. 1999). Transformants (rps4AS cells) generated by antisense-mediated gene inactivation also exhibit markedly delayed differentiation; most of them showed no sign of aggregation and remained as round-shaped single cells even after 16 h of incubation (Hosoya et al. 2003). In contrast, overexpression of the mt-rps4 mRNA in the extramitochondrial cytoplasm enhances the initial step of cell differentiation including aggregation and induces a lot of small tight aggregates under the submerged conditions (Inazu et al. 1999). Here it is of interest to note that the antisense-rps4 RNA synthesized in the extramitochondrial cytoplasm is effective as the partial disruption of mt-rps4 gene. This seems to indicate that a trace of the rps4 mRNA and/or RPS4 protein, both of which are synthesized in mitochondria, may be released to the extramitochondrial cytoplasm. Alternatively, it is also possible that the antisense-mt-rps4 RNA may enter mitochondria to inactivate mt-rps4 mRNA and mt-RPS4 synthesis. Based on PSORTII prediction, when the mt-RPS4 protein is released into the cytosol, it is predicted to move preferentially to the nucleus. It was confirmed by the immunohistochemical method using the anti-mt-RPS4 antibody that in the rps4OE cells the mt-RPS4 protein produced in the cytoplasm is capable of moving to the nucleus (Hosoya et al. 2003). Again, it is worth noting that the partial inactivation (mitochondrial heteroplasmy) of the rps4 gene greatly impairs differentiation, including cell aggregation. Although the fact that only the RPS4 protein of Dictyostelium has several nuclear localization signals is puzzling, at least a part of the mt-RPS4-protein seems to work in the nucleus to regulate cell differentiation. In other organisms, their mt-RPS4 proteins lack nuclear localization signals. It is generally difficult for proteins located in the mitochondrial matrix to go out to the cytosol, because mitochondria are partitioned by two (outer and inner) membranes. Recently, however, several mitochondrial proteins like apoptosis-inducing factor (AIF; Daugas et al. 2000), endonuclease G (Ohsato et al. 2002), and heat shock protein 70 (Hsp70; Susin et al. 1999) have been shown to move to the nucleus in response to apoptosis and heat shock. All of these proteins are encoded by the nuclear genome DNA, followed by translocation to the mitochondrion and then again to the nucleus. Accordingly, the behavior of Dictyostelium RPS4 is unique in that it is encoded by the mtDNA.

Provided that it is possible to completely inactivate rps4 expression, the rps4-null cells should never differentiate from the GDT point in response to starvation. To test this, we have explored a new method for specifically disrupting a mitochondrial gene (rps4), by a combination of homologous recombination and delivery of an appropriate restriction endonuclease (SfoI) into mitochondria (Chida et al. 2008). First, mitochondrially targeted SfoI whose expression is under control of the tetracycline (Tet)-regulated gene expression system was introduced into cells heteroplasmic with respect to the rps4 gene. The heteroplasmic cells were produced by homologous recombination by use of the construct in which the unique SfoI site and 5′-half of the rps4 coding region were deleted not to be digested by SfoI, and therefore their mitochondria have both the wild-type mtDNA and the mutant mtDNA with the disrupted rps4 gene (Fig. 2A). In response to removal of Tet from growth medium, SfoI is selectively delivered into mitochondria and digests only the wild-type mtDNA but not the mutated rps4, thus giving rps4-null cells with only the mutated mtDNA, under the Tet-minus condition. A starting material, LpCSfo cells in which pCoxIV (MTS)-SfoI is expressed under the Tet-minus condition stop growing in growth medium within 48 h after their transfer to Tet-minus medium, because they are converted to a rho-zero state in which mtDNA is completely eliminated (Fig. 2B). However, LpCSfoHR(−Tet) cells (rps4-null cells) have almost the same number of apparently intact mitochondria and are able to grow normally with nearly the same doubling-time, as in the case of parental MB35 cells, indicating that the intact rps4 gene is not required for growth. As expected, starving rps4-null cells (LpCSfoHR(−Tet) cells) exhibit a great delay of differentiation: no sign of cell aggregation is noticed even after 16 h of incubation, though LpCSfoHR(+Tet) cells show normal development (Fig. 2C). This suggests that the mt-RPS4 protein and/or mt-rps4 RNA may be essential for gene expressions required for cell aggregation. The new method presented here will provide a powerful tool for precisely determining the functions of individual mitochondrial genes as well as for genetic therapy of mitochondrial diseases.

Figure 2.

 Strategy for creating rps4-null cells and their phenotypes. (A) As a starting material, LpCSfo cells in which pCoxIV (MTS)-SfoI is expressed under the tetracycline-minus (−Tet) condition, were prepared. Subsequently, the mutant rps4 gene (Mut-mtDNA for homologous recombination), in which the upstream SfoI-site and a 5′-half of rps4 coding region were deleted, was introduced into LpCSfo cells to obtain heteroplasmic transformants (LpCSfoHR cells) with mitochondria consisting of the Mut-mtDNA and wild-type mtDNA (Wt-mtDNA). Coupled with removal of Tet from growth medium, the fusion protein MTS-SfoI synthesized in the cytoplasm is exclusively transferred into mitochondria of LpCSfoHR cells and selectively digests Wt-mtDNA but not Mut-mtDNA. Since the digested Wt-mtDNA is not duplicated, the Mut-mtDNA becomes dominant during the course of growth under the −Tet condition, thus eventually giving rps4-null cells. (B) These cells were grown in growth medium with (+Tet) or without (−Tet) teteracyclin. Membrane potential of mitochondria was visualized by staining of cells with MitoTracker Orange. As was expected, the staining of mitochondria was almost completely vanished in LpCSfo cells grown without Tet, because their mtDNA with an intact SfoI site would be cleaved by SfoI eventually to become a ρ0 state devoid of mitochondrial DNA. Bars, 20 μm. (C) Development of starved MB35 cells and LpCSfoHR cells on agar. MB35 cells and LpCSfoHR cells grown with (+Tet) or without (−Tet) tetracycline were washed twice in BSS and plated on 1.5% non-nutrient agar at a density of 5 × 106 cells/cm2. This was followed by incubation for the indicated times at 22°C. Bars, 0.5 mm. (Basically from Chida et al. 2008).

DIA2.  The dia2 gene encodes a novel lysine- and leucine-rich protein (DIA2) with a predicted cell mass of 16.9 kDa, and the mRNA accumulates in differentiating cells starved just before the GDT point, while there is no detectable expression in vegetatively growing cells (Chae et al. 1998). The expression pattern of dia2 during the whole course of development is quite similar to that of the cAMP receptor 1 (carA). The DIA2 protein seems to be essential for cell viability, because we failed to obtain dia2-null cells by means of homologous recombination. Antisense-mediated gene inactivation of dia2 greatly inhibits the progress of differentiation, coupled with marked suppression of carA expression, and stops their development at the mound stage (Fig. 3), though dia2AS cells in which the dia2 expression is partially inactivated exhibit almost normal proliferation in growth medium (Chae et al. 1998). Thus dia2 expression appears to play an essential role in the initiation of differentiation, closely relating to the cAMP signaling system. In fact, we have recently demonstrated that both of the aca (adenylate cyclase A) and carA expressions in dia2AS cells produced by the antisense RNA method are delayed and lower compared with those in parental Ax-2 cells, thus resulting in impaired cAMP release from dia2AS cells, and that the aggregative defect in dia2AS cells is recovered by externally added cAMP pulses (Hirata et al. 2008). Interestingly, the DIA2 protein was found to change its location from the endoplasmic reticulum (ER) to prespore-specific vacuoles (PSVs), and exocytosis of PSVs from prespore cells and the subsequent spore differentiation are almost completely impaired in dia2AS cells (Hirata et al. 2008).

Figure 3.

 Developmental behaviors of starved dia2AS cells and parental Ax-2 cells on agar. Exponentially growing cells were harvested, washed twice in BSS (Bonner’s salt solution; Bonner 1947), and spread on 1.5% non-nutrient agar at a density of 5 × 105 cells/cm2, followed by incubation for the indicated times at 22.0°C. dia2AS cells exhibit considerably delayed aggregation compared to Ax-2 cells (A, B) and form large aggregates (D). Each aggregate of dia2AS cells subdivides into smaller mounds (F) and stopped its development at this stage. Thus, dia2AS cells fail to form migrating slugs or fruiting bodies (C, E) as formed by Ax-2 cells. Bar, 1 mm. (Basically from Hirata et al. 2008).

DIA1.  Overexpression of another novel gene, dia1, was found to rather impair the progression of differentiation, possibly coupled with the reduced expression of early genes such as carA, and the inhibitory effect of enforced dia1 expression is almost completely nullified by externally applied cAMP pulses (Hirose et al. 2000). In contrast, antisense RNA-mediated inactivation of dia1 enhances the initial step of differentiation, as exemplified by precocious expression of carA and other early genes (Hirose et al. 2000). The dia1 mRNA with 1368 bp of open reading frame is deduced to encode a 48.6 kDa protein (DIA1). The DIA1 protein is highly serine rich, particularly in the C-terminal region, and is predicted to be GPI (glycosyl phosphatidyl inositol) anchored at the cell membrane. Considering the abovementioned function of dia1 expression, the DIA1 protein seems to be negatively coupled with CAR1-associated events. Provided that dia1 expression transiently suppresses the progression of differentiation in T7 cells located just before the GDT point, it is possible that the time difference (about 7 h) between T0 (just after the GDT point) and T7 cells may be shortened, thus allowing both of the T0 and T7 cells to coordinately participate in forming a common aggregate.

Recently, the dia1 gene has been shown to be adjacent to a vegetative gene, impA, on chromosome 4, with an intergenic region of 654 bp (Hirose et al. 2005). The 645 region separates the start codons of dia1 and impA, which are transcribed in opposite directions, and there is a pair of nearly identical 20 bp sequences in the region proximal to dia1 that appear to stimulate expression of impA about twofold (Hirose et al. 2005). Although the implication of the closeness of dia1 and impA is presently unknown, it is of interest to note that a highly related 20 bp sequence is found just upstream of carA specifically expressed at the GDT.

The possible functions of other genes (caf1, anti-annexin VII RNA), which are predominantly expressed during the GDT, have been reviewed in my previous article (Maeda 2005). To elucidate GDT signaling, several studies have been done using REMI mutants with defects in the expression of the discoidin I gene, monitoring misexpression of discoidin by colony blots using a monoclonal anti-discoidin antibody. Zeng et al. (2000) isolated eight mutants that exhibit overexpression of discoidin I: they display normal morphogenesis after starvation but show premature entry into the differentiation phase. The disrupted gene was named gdt1. gdt1 is expressed in growing cells, and the levels of mRNA and protein appear to increase with cell density, followed by a rapid decrease with the onset of differentiation. gdt1 encodes a 175 kDa protein with four putative transmembrane domains. The amino-acid sequence of the C-terminus displays some similarity to the catalytic domain of protein kinases. The response to folate, a negative regulator of discoidin expression, is not impaired in gdt1-null cells. Cells lacking the Gα2 protein display a loss of discoidin expression and never aggregate. Interestingly, gdt/Gα2 double mutants exhibit no aggregation but strong discoidin expression, thus suggesting that gdt1 is a negative regulator of GDT downstream or in a parallel pathway to Gα2 (Zeng et al. 2000). Recently, gdt2 has been identified by Chibalina et al. (2004) as another gene that represses expression of discoidin. While GDT1 and GDT2 are similar in many ways, GDT2 contains a well conserved protein kinase domain, unlike GDT1 whose kinase domain is presumably nonfunctional. In addition, gdt2 and gdt1 mRNA are regulated differently, with gdt2 expressed throughout development. The phenotypes of gdt2- and gdt1-null cells are somewhat similar but not identical. For example, gdt2-null cells are able to grow at a normal rate, unlike gdt1-null cells. protein kinase A (PKA) levels and activity are essentially normal in growing gdt2-null cells, implying that GDT2 regulates a signaling pathway that acts separately from PKA (Chibalina et al. 2004). As other genes regulating the GDT process, amiA and amiB have been reported (Kon et al. 2000; Nagasaki et al. 1998). amiA is homologous to a yeast gene of unknown function, while amiB does not show homology to any known genes. In contrast to dia1, gdt1, and gdt2, both genes positively regulate GDT possibly via regulation of adenylate cyclase expression. Until now, however, it remains to be determined if the expression patterns of these four genes (gdt1, gdt2, amiA, and amiB) are GDT-point specific or starvation specific.

In general, phosphorylation and dephosphorylation of proteins play essential roles in cell-cycle regulation and oncogenesis. The phosphorylated level of proteins changes markedly in a cell-cycle-dependent manner: p105-RB, the product of the retinoblastoma (RB) tumor suppressor gene (Friend et al. 1986; Huang et al. 1988; Bookstein et al. 1990), is maximally phosphorylated at the S-phase of the cell cycle, while the protein is dephosphorylated at the G0- and G1-phases in vertebrate cells (Buchkovich et al. 1989; Goodrich et al. 1991). RB dephosphorylation also occurs in response to the induction of differentiation in several human leukemia cell lines by phorbol ester or retinoic acid treatment (Chen et al. 1998). In Dictyostelium, cAMP acts to facilitate early differentiation by activation of the catalytic subunit of the cAMP-dependent PKA (Simon et al. 1989). In light of these findings, studies on the phosphorylated state of proteins are of particular importance to elucidate the molecular mechanism of GDT. As shown in Table 1, evidence has been obtained indicating that the phosphorylation level of the 90 kDa protein (GRP94: glucose-regulated protein 94; endoplasmic reticulum Hsp90) and 101 kDa protein (EF2A: elongation factor-2A) is greatly reduced in response to the initiation of differentiation from the GDT point (Akiyama & Maeda 1992). Since the possible functions of these phosphoproteins have been reviewed in Maeda (2005), see the review article for details.

There are numerous unanswered questions about the mechanism of GDT. Complex interactions between positive and negative regulators must be working to control the GDT in Dictyostelium development. Elucidation of signaling networks of the GDT-associated molecules, particularly of their integration with starvation-induced events, would be helpful to understand the precise mechanism of initial differentiation.

Intercellular signals required for the initiation of cell differentiation

Dictyostelium cells survive adverse conditions, such as starvation, by gathering into communities that provide optimum conditions for differentiation into spores with strong resistance against various physical and chemical stresses. Depending on the circumstances, they have other responses possible too – differentiation to microcysts and macrocysts formed by a sexual process. In the laboratory, differentiation is induced by abruptly washing away nutrients. In the soil, however, depletion of nutrients is more gradual and the cells have mechanisms to sense when hard times such as starvation are approaching. There is a density-sensing mechanism (PSR; prestarvation response) that is working during growth and prepares growing cells to induce the initial step of differentiation including the acquisition of aggregation-competence. Prestarvation factors (PSFs) are synthesized during growth and accumulate in the microenvironment such as growth medium according to the density of the cells. Three kinds of PSFs (PSF-1, PSF-2, and PSF-3) have been reported. Conditioned medium factors (CMFs) secreted by starving cells are essential for establishing cAMP signaling and the initiation of aggregation (Gomer et al. 1991; Iijima et al. 1995; Yuen et al. 1995). Also, several counting factors (CFs) that are also secreted by starving cells help to assess their density at a slightly later stage (during aggregation) (see Tang & Gomer 2008).


Exponentially growing cells of Dictyostelium discoideum (strains NC-4 and Ax-3) produce a soluble substance that accumulates in the medium in proportion to cell density; this substance (referred to as PSF-1 for convenience) regulates the production of certain proteins previously thought to be induced by starvation (Clarke et al. 1987). During growth, Dictyostelium cells monitor the relative concentrations of PSF-1 and food bacteria such as Escherichia coli. When PSF-1 reaches a sufficiently high level relative to the concentration of bacteria, synthesis of PSF-1-regulated proteins such as discoidin I, cAMP receptor 1 and cell-adhesion molecule gp24 is induced (Clarke et al. 1992; Rathi & Clarke 1992; Shatzle et al. 1992). The food bacteria are shown to inhibit the response of Dictyostelium cells to PSF-1; the bacteria do not inactivate PSF-1 or inhibit its production; instead, they affect the ability of NC-4 cells to detect PSF-1, possibly by binding to the same cell surface receptor (Clarke et al. 1988). In the absence of bacteria, as during axenic growth of Ax-3 cells, the PSR occurs at much lower cell densities, probably accounting for the presence of certain developmentally regulated mRNAs and proteins in axenic cultures.

PSF-1 with a mass of 65–70 kDa is sensitive to proteases and to heat. Using partially purified PSF-1, a number of genes including secreted cAMP phophodiesterase (pdsA), which are induced by starvation, have been induced in growing cells (Lacombe et al. 1986). However, the precise role of PSF-1 remains to be elucidated; it has not been purified to homogeneity, nor has its gene been cloned. The synthesis of PSF-1 declines after starvation, coupling with cell differentiation, and PSF-1 does not enhance further differentiation in the absence of starvation. The steps between binding of PSF-1 to cells and induction of various prestarvation genes require protein synthesis and seem to be complex.


Another PSF (PSF-2) that is sensitive to protease treatment but is heat stable was found by Maeda & Iijima (1992). D. discoideum Ax-2 cells harvested from growth medium at relatively high densities (above 1 × 106 cells/mL) differentiate normally after starvation, while those harvested at low densities (below 5 × 105 cells/mL) never differentiate because of a failure of cells to be exposed to a sufficient concentration of PSF-2 during growth. When Ax-2 cells are allowed to grow to the stationary phase (above 2 × 107 cells/mL), they exhibit delayed development after starvation. Although the reason for this delay is presently unknown, it is possible that PSF-inhibitors as well as growth inhibitors may accumulate in the medium during the stationary growth phase. Differentiation competence, once acquired, is lost within 30 min, following introduction of cells to fresh growth medium. During the erasure event, differentiation characters are lost in distinct steps; firstly cells lose the ability to release cAMP, and then, parallel with erasure stabilization, cells also lose cAMP-directed chemotaxis (Soll & Mitchell 1982; Varnum & Soll 1984). On the other hand, a relatively long time (3–5 h) is needed for noncompetent cells to achieve differentiation competence when transferred to conditioned growth medium (CGM) that probably contains a sufficient concentration of PSF-2. Differentiation competence is also induced by addition of partially purified PSF-2. By means of column chromatography, the activity of PSF-2 was found to be in the fraction with an estimated molecular mass of 30–40 kDa (Maeda & Iijima 1992). The differences in heat stability and molecular weight indicate that PSF-1 and PSF-2 are different molecular species, but a functional relationship between the two cannot be excluded until a single preparation has been assayed for both activities.


Recently, the Dictyostelium homologue (Dd-TRAP1) of TRAP-1 (tumor necrosis receptor-associated protein 1), which is a molecular chaperone belonging to the heat-shock protein 90 (hsp90) family, has been shown to translocate from the cell cortex to mitochondria as the density of growing cells increases, which allows the prompt transition of cells from growth to differentiation through a novel prestarvation factor (PSF-3) in growth medium (Morita et al. 2002, 2004). When Ax-2 cells growing at low cell density (5 × 105 cells/mL), in which Dd-TRAP1 is localized in the cell cortex, were harvested and incubated in conditioned growth medium in which Ax-2 cells had been grown up to the late exponential growth phase (and therefore contained a sufficient amount of PSF- 3), Dd-TRAP 1 was found to quickly translocate to mitochondria within 1 min of incubation even at low density. As described above, when cells growing at a high cell density (5 × 106 cells/mL) are transferred to fresh growth medium, their differentiation competence, acquired by the PSR, is lost within 30 min of incubation (Maeda & 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 min of incubation in fresh growth medium (Morita et al. 2004). The activity for the translocation of Dd-TRAP1 to mitochondria is resistant to protease but is lost by boiling of the CG M for 15 min before use, thus indicating that the factor (PSF-3) is different from PSF-1 and PSF-2. The activity of PSF is usually measured as the increased discoidin I expression in growth-phase cells, but Dd-TRAP1 is not involved in discoidin I expression.

Burdine & Clarke (1995) have reported that prestarvation genes such as discoidin I and the 2.4 kb PDE (phosphodiesterase) are barely induced in PKA-cat (cAMP-dependent protein kinase, catalytic subunit)-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. Importantly, the translocation of Dd-TRAP1 to mitochondria is observed both in Gβ-null cells and in PKA-cat-null cells (Morita et al. 2004). This indicates that neither PKA nor Gβ is required for the translocation via the novel PSF-3-mediated PSR. The knockdown mutant of Dd-TRAP1 (TRAP1-RNAi cells) exhibits a significant defect in PSR (Morita et al. 2004, 2005). Although TRAP1-RNAi cells show normal expression of classic prestarvation genes such as dscA (discoidin I) and carA, the expression of differentiation-associated genes (dia1 and dia3) induced by the PSR is markedly repressed. In contrast, transformants overexpressing Dd-TRAP1 show an early PSR and also increased expression of dia1 and dia3 in a cell-density dependent manner in growth medium (Morita et al. 2004).

As mentioned above, D. discoideum Ax-2 cells secrete at least three different kinds of PSFs that accumulate in proportion to cell density during growth and induce subsequent differentiation after starvation. The features of PSF-1, PSF-2, and PSF-3 are summarized in Table 2. The precise roles and action mechanisms of the PSFs will be defined when pure PSFs, the specific antibodies against them, and the genes encoding them are available in further studies. Here it is of importance to note a good correlation between the prestarvation genes and the GDT point-specific genes, as shown in Table 3.

Table 2.   Prestarvation factors (PSFs) secreted from growing Dictyostelium cells and molecular events quickly caused by starvation
PSF-speciesTreatment withFunctionReferences
PSF-1SensitiveSensitiveInduction of prestarvation genes such as discoidin IClarke et al. (1988)
PSF-2ResistantSensitiveEnhancement of differentiation under semi-starvation conditionsMaeda & Iijima (1992)
PSF-3SensitiveResistantInduction of differentiation associated genes such as dia1 and dia3.
Translocation of Dd-TRAP1 from the cell cortex to mitochondria
Morita et al. (2004)
Change of gene or protein coupled with starvationDetails of the changeReferences
srsA encoding a novel protein (6.4 kDa)Quite immediately expressed in the first 5 min after starvationSasaki et al. (2008)
Ribosomal protein S6 (RPS6; 32 kDa)Completely dephosphorylated in response to starvationAkiyama & Maeda (1992); Ishii et al. (2009)
165 kDa proteinQuickly phosphorylated at the tyrosine residue(s) coupled with starvationItakura et al. (unpubl. data)
Table 3.   Correlation between the growth/differentiation transition point (GDT-point) specific genes and the prestarvation genes
GeneGDT-point specific? (reference)Prestarvation gene? (reference)
  1. From Maeda (2005). nd, not determined.

discoidin IYes (Huang & Pears 1999)Yes (Clarke et al. 1987)
carA (cAMP-receptor 1)Yes (Abe & Maeda 1994)Yes (Rathi & Clarke 1992)
dia1Yes (Hirose et al. 2000)Yes (Morita et al. 2004)
dia2Yes (Chae et al. 1998)Yes (Hirose, unpubl. data)
dia3 (mitochondrial rps4 gene)Yes (Inazu et al. 1999)Yes (Morita et al. 2004)
caf-1Yes (Abe & Maeda 1995)nd
pde (phosphodieterase)ndYes (Rathi & Clarke 1992)

Conditioned medium factors and counting factors

Starving cells are known to secrete several kinds of CMFs and counting factors (CFs), both of which are involved in establishing cAMP signaling and sensing own cell density in the population, respectively. The functions of these factors will be reviewed by R. H. Gomer in this special issue. At all events, several diffusible factors secreted during the growth or early differentiation phase work as intercellular communicators that enable starving Dictyostelium cells to differentiate normally.

Early events induced by starvation

Starvation is an environmental element essential to triggering cell differentiation and morphogenesis, but it is not enough: for the initiation of differentiation the signals guided by starvation must be integrated into specific events coupled with the growth/differentiation checkpoint (GDT point) in the Dictyostelium cell cycle.

To study individual proteins synthesized after the cells were starved, early experiments were done using two-dimensional polyacrylamide gel electrophoresis (2D-PAGE) analysis. When cells were suddenly removed from growth medium, most proteins are reduced in the hours after starvation, but the synthesis of several proteins is transiently induced (Margolskee & Lodish 1980). At the time this work was done, however, it was impossible to determine what these proteins were. More recent studies have revealed a number of early transcripts induced by starvation. Singleton et al. (1988) reported the cycloheximide-resistant induction of several genes, though their precise roles in early differentiation remain to be elucidated, because their genes have not been disrupted. When the V4 gene, isolated by McPherson & Singleton (1992), was inactivated by the antisense RNA, the transcription of vegetative growth phase genes failed to be deactivated, thus leading to a reduction in the transcription of genes that are involved in the events of chemotaxis to cAMP. Among the genes that are repressed during early differentiation, the transcription of several ribosomal protein genes has been shown to be rapidly reduced after starvation begins (Ken & Singleton 1994).

YakA-mediated signal transduction.  Protein kinase A plays a critical role during the early stage of differentiation and at all later stages. The PKA of Dictyostelium cells is a dimer consisting of one regulatory subunit (PKA-R) and one catalytic subunit (PKA-cat), rather than the tetramer of higher organisms (De Gunzburg et al. 1984). Expression of the PKA-R gene starts after a few hours of starvation and continues during the entire course of development (Mutzel et al. 1987). The PKA-cat gene is exposed at low levels during growth and at high levels following starvation. Removing the PKA-R by deleting its gene releases PKA-cat from inhibition and causes differentiation to proceed very rapidly (Simon et al. 1992). Overexpression of PKA-cat also causes rapid differentiation (Anjard et al. 1992; Mann et al. 1992). By contrast, deletion of PKA-cat causes an aggregateless phenotype without affecting growth. Several genes involved in chemotaxis – acaA (adenylate cyclase A), pdiA (phosphodiesterase inhibitor A), and carA (the major cAMP receptor in early differentiation) – are never transcribed in the absence of PKA-cat (Mann et al. 1997; Wu et al. 1995). The failure of PKA-null cells to differentiate is not due to the loss of adenylate cyclase, because this function can be replaced by expressing the PKA-cat gene under the control of a constitutive active promoter (actin 15) without restoring differentiation (Mann et al. 1997). Recent evidence suggests that the role of PKA is fulfilled by a series of sensor histidine kinases that integrate with the cAMP signaling events (Loomis 1998; Thomason et al. 1998).

Kuspa and his colleagues have found a gene encoding the protein kinase YakA, a homologue of yeast Yak1p growth-regulating protein kinase, from a REMI library (Souza et al. 1998). The expression of yakA is required for turning off growth-phase genes and for induction of differentiation-associated genes (Souza et al. 1998). yakA-null cells divide and multiply more rapidly compared to parental cells, reducing their size. PKAcat mRNA appears normal in yakA-null cells, but the enzyme activity of PKA does not exhibit the characteristic increase after 5 h of starvation. PKA-dependent genes are never expressed in yakA-null cells. Importantly, yakA-null cells, just like those of V4 mutants, cannot turn off vegetative genes that are expressed in growth-phase cells. In contrast, overexpression of YakA in a conditional manner causes arrest of growth and exhibits the expression of differentiation-associated genes even in the presence of nutrients. The yeast Yak1p is capable of mediating growth to differentiation transition, and YakA is also able to substitute for Yak1p, thus indicating that YakA shares many functions with the yeast Yak1p (Souza et al. 1998).

To search additional components relating to YakA signaling, Souza and her colleagues tried a REMI suppressor screening on yakA-null cells and identified a gene, pufA, that can change the aggregateless phenotype of yakA-null cells to a normal one (Souza et al. 1999). pufA encodes a member of the Puf (pumilio/FBP) family of proteins, which functions in the translational control of key regulators for anterior–posterior patterning in Caenorhabditis elegans and Drosophila melanogaster (Zhang et al. 1997; Forbes & Lehmann 1998; Wharton et al. 1998; Zamore et al. 1999). PKA is a likely candidate for regulation by PufA, because the Dictyostelium PKA mRNA has sequences related to the NREs (Nanos response elements) of the Drosophila hunchback protein (Souza et al. 1999). pufA-null cells show a phenotype similar to YakA-overexpressing cells in their rapid differentiation, precocious expression of adenylate cyclase A (ACA) and increased PKA activity, while in yakA-null cells the pufA mRNA is well retained even after 2 h of starvation, thus indicating that YakA is required for the loss of pufA mRNA at the onset of differentiation coupled with starvation (Souza et al. 1999).

Ca2+-related events.  Calcium ions (Ca2+) are believed to be a key factor in numerous cellular processes. Cytosolic free Ca2+ concentration ([Ca2+]i) in Dictyostelium cells changes significantly in a cell-cycle-dependent manner (Azhar et al. 2001). Ax-2 cells starved in the S- and early G2-phases of the cell cycle have relatively high [Ca2+]i levels and display a tendency to become stalk cells, while those starved in the mid- or late-G2 phase have lower [Ca2+]i levels and tend to become spores (Saran 1999; Azhar et al. 2001). [Ca2+]i also increases strikingly in response to starvation and reaches a peak within 30 min of starvation, and [Ca2+]i levels artificially elevated by thapsigargin and nigericin are capable of mimicking starvation responses to induce differentiation (Tanaka et al. 1998).

Changes in the phosphorylation levels of proteins.  The serine residue(s) of the 32 kDa protein is rapidly and completely dephosphorylated within 30 min of starvation (Akiyama & Maeda 1992). Recently, this phosphoprotein was identified as a homologue (Dd-RPS6) of ribosomal protein S6 (RPS6) that is an essential member of protein synthesis (Ishii et al. 2009). The dephosphorylation of Dd-RPS6 is completely inhibited by application of a highly specific inhibitor of protein phosphatase PP1 (ATP-Mg2+-dependent serine/threonine-specific phosphatase) and PP2 (polycation-stimulated serine/threonine-specific phosphatase), okadaic acid (OA), or calyculin A (CL-A) to starvation medium. Interestingly, the Dd-RPS6 is highly phosphorylated in starvation medium containing 0.5 mmol/L CL-A, and the hyperphosphorylation of the protein seems to induce the progression of the cell cycle through the M- and S-phases up to the early G2-phase even in the absence of nutrients (Akiyama & Maeda 1992). As expected, Dd-RPS6 is absolutely required for cell survival; we failed to obtain antisense-RNA mediated cells and Dd-rps6-null cells by homologous recombination in spite of many trials. Somewhat surprisingly, Dd-RPS6-overexpressing cells exhibit the delayed initiation of differentiation triggered by starvation (Ishii et al. 2009). Thus, it is likely that the proper expression of Dd-RPS6 may be of importance for the initiation of differentiation.

srsA: a gene expressed quite quickly in response to starvation.  Recently, we have isolated a novel gene, srsA, as one rapidly expressed in the first 5 min following the removal of nutrients (Sasaki et al. 2008). This gene encodes a small protein (a predicted molecular mass of 6.4 kDa) with no significant similarity to previously characterized proteins (Table 2). Disruption of srsA results in delayed expression of early genes acaA and carA that encode adenylate cyclase and the cAMP receptor necessary for chemotactic aggregation, respectively (Sasaki et al. 2008). Unexpectedly, however, srsA-overexpression also results in delayed aggregation, thus suggesting that the proper expression of srsA may be critical for starvation response and subsequent differentiation, as in the case of Dd-RPS6.

Cell-cycle dependent differentiation and pattern formation

The cell population of D. discoideum is heterogeneous before differentiating into prestalk and prespore cells, as is well exemplified by the difference in the size and buoyant density of cells. It has been demonstrated using various fractionation methods for preaggregative cells that two classes separated by centrifugation behave differently to be sorted out into the anterior (prestalk) zone or the posterior (prespore) zone of migrating slugs (Takeuchi 1969; Bonner et al. 1971; Maeda & Maeda 1974). Here it is of importance to know what is a naturally occurring variable for cell sorting during normal development. Without any manipulation of growing cells for cell synchronization, the cell-cycle phases of exponentially growing cells must be randomly distributed, thus resulting in a difference in the cell-cycle positions of cells at the onset of starvation. Accordingly, it is quite possible that the heterogeneities, such as cell size and buoyant density, may be due to different cell-cycle positions at which the cells are starved. In fact, many studies (MacDonald & Durston 1984; Weijer et al. 1984b; Maeda 1993, 1997; Takeuchi et al. 1994; Maeda et al. 2002) have shown that Dictyostelium cells sort out in a cell-cycle-dependent manner during the establishment of the prestalk–prespore pattern in a migrating slug.

When synchronized cell populations of D. discoideum Ax-2 at various phases of the cell cycle (referred to as Tt cells, obtained by the temperature shift method: shift-up from 11.5 to 22.0°C) are starved, washed, and allowed to develop on non-nutrient agar, they exhibit different developmental features during further culture. For example, cells starved just before mitosis (M-phase), like T1 cells (cells 1 h after shift-up of temperature, just after the GDT point), are preferentially sorted out into the anterior prestalk zone of slugs, whereas T7 cells (cells 7 h after shift-up of temperature) starved at the mid-to-late G2-phase (just before the GDT point) are sorted out into the posterior prespore zone (Ohmori & Maeda 1987; Araki et al. 1994, 1997). The time course of cell aggregation also varies in a cell-cycle-related manner. As was expected, T7 cells exhibit the most rapid differentiation, because they enter the differentiation phase immediately from the GDT point coupled with starvation and acquire chemotactic activities to cAMP and EDTA-resistant cohesiveness much earlier than T1 cells starved just after the GDT point (Ohmori & Maeda 1987). Accordingly, it is most likely that T7 cells may function as autonomously signaling aggregation centers and attract chemotactically neighboring cells starved at other phases of the cell cycle. This is supported by the fact that T7 cells transformed by a vector (pAct15-Gal) bearing bacterial β-galactosidase are preferentially located in the central region of an aggregating cell mass on mixed culture with nonsynchronized (nontransformed) cells (Araki et al. 1994; Huang et al. 1997; Huang & Pears 1999). In this connection, the number of cell aggregates formed from nonsynchronized Ax-2 cells increases in proportion to that of externally added T7 cells (unpubl. data). The sorting behaviors of T1 and T7 cells during the entire course of development are shown schematically in Figure 4. T1 cells starved at the very late G2-phase aggregate more slowly than T7 cells, but are then sorted out into the apical tip of tipped aggregates, thus locating predominantly in the anterior prestalk zone of slugs to establish the anterior T1/posterior T7 pattern. Such behaviors of the synchronized cells have been confirmed by several studies (e.g. Azhar et al. 2001; Weeks & Weijer 2006). At the mound stage, both T1 and T7 cells exhibit a temporarily uniform distribution throughout the cell mass, the former being in transit to the tip region and the latter to the basal region. Thus, T1 and T7 cells exchange their relative position in the cell masses between the aggregation-stream and tipped-aggregate stages (Fig. 5A). Although the precise mechanism of the interchange remains to be elucidated, it is noteworthy that most T7 cells progress in their cell cycle after formation of multicellular structures (mounds) (Araki & Maeda 1995; Araki et al. 1997), possibly coupled with prespore differentiation, while starving T1 cells double by passing through the M-phase in the cell cycle but never progress through the cell cycle after that (Fig. 5B). In this connection, many studies have revealed that in Ax-2 cells the progression of S-phase is specifically reinitiated in prespore cells (Durston & Vork 1978; Zimmerman & Weijer 1993), and that the cell division and nuclear DNA synthesis actually occur around the mound stage (Zada-Hames & Ashworth 1978; Araki & Maeda 1995, 1998; Weeks & Weijer 2006; Muramoto & Chubb 2008). This might relate to the finding that the duplication of mitochondrial DNA occurs in differentiating prespore cells but not in prestalk cells (Shaulsky & Loomis 1995). Thus it seems to be possible that preferential re-entry of T7 cells (prespore cells) into the cell cycle during the early mound stage might be coupled with the loss (dedifferentiation) of once acquired differentiation characteristics such as the chemotactic ability to cAMP signals as compared with T1 cells at the early mound stage, as shown schematically in Figure 5.

Figure 4.

 Diagram showing cell-cycle-dependent sorting behaviors and pattern formation in Dictyostelium development. T7 cells (red cells) starved just before the growth/differentiation checkpoint (GDT point) may work as autonomously signaling centers for chemotaxis and locate preferentially around aggregation centers, while T1 cells (blue cells) starved just after the GDT point exhibit much slower aggregation and are preferentially located at the peripheral region of aggregation streams and early mounds. During further morphogenesis, however, T1 cells are sorted out into the apical region of tipped aggregates, located at the anterior prestalk zone of migrating slugs, and eventually differentiate into stalk cells in fruiting bodies. In contrast, T7 cells are sorted out into the posterior prespore zone of migrating slugs, establishing the anterior T1/posterior T7 pattern, and eventually differentiate into spores in fruiting bodies. Here it is noteworthy that the spatial exchange between T1 and T7 cells occurs sometime between aggregation streams and tipped aggregates. (Basically from Maeda 2005).

Figure 5.

 A likely model for explaining the spatial exchange of T1 and T7 cells around the mound stage. (A) Schematic drawing showing active exchange of the relative positions of T1 and T7 cells in cell masses from the aggregation-stream to the tipped-aggregate stage. In this connection, the chemotactic ability of prestalk cells (derived from T1 cells) is known to be much higher than that of prespore cells (derived from T7 cells) (Early et al. 1995). (B) Temporal changes of the cell number and chemotactic sensitivity of T1 and T7 cells to cAMP after starvation. As indicated from Figures 1 and 4, T1 cells double in number after starvation and progress through the M phase to reach the GDT point even under starvation conditions. However, T1 cells that have passed through the GDT point never divide during further development. Meanwhile, starving T7 cells immediately enter the differentiation phase from the GDT point to acquire the chemotactic sensitivity to cAMP. Around the mound stage, however, T7 cells re-enter the cell cycle and divide, coupled with prespore differentiation in early mounds. Therefore, it is quite likely that dividing T7 cells (prespore cells) may reduce or lose the once acquired chemotactic ability around the early mound stage, resulting in the drastic exchange of the cell’s position in cell masses. (Basically from Maeda 2005).

It is of interest to know whether a similar phenomenon is also noticed in species other than D. discoideum and also without any treatment of cells for cell synchronization. For this, the sorting behavior of D. mucoroides-7 (Dm7) cells and its relation to cell-cycle position at the onset of starvation were analyzed, using nonsynchronized Dm7 cells pulse-labeled with BrdU (Amagai & Maeda 1996). The results obtained demonstrate that Dm7 cells starved at the early G2-phase aggregate most rapidly, but are eventually sorted out to the posterior prespore zone of migrating slugs. In contrast, cells starved at the mid-to-late G2-phase exhibit slower aggregation, but are sorted out to the anterior zone (tip). This is basically similar to the sorting behavior of D. discoideum, as described later, though the existence of a GDT point in the cell cycle remains to be determined using synchronized Dm7 cells. Measurements of the cell numbers and nuclear DNA contents have provided evidence that about 80% of Dm7 cells progress in their cell cycle after the formation of multicellular structures (mounds), presumably coupling with prespore differentiation as in the case of D. discoideum (Amagai & Maeda 1996). Therefore, it is unlikely that the observed sorting behavior of Ax-2 cells synchronized by a temperature shift (cold shock) is an artifact caused by the cold shock. In other words, the cell-cycle-dependent sorting during Dictyostelium development is most likely to be a common phenomenon widely recognized in different species of slime molds.

The significance of cell-cycle progression coupled with prespore differentiation has been confirmed using proper transformants (ecmA-gal and Dp87-gal cells) and selective inhibitors of cell-cycle progression, such as nocodazole and CL-A (Araki & Maeda 1998). That is, nocodazole, an inhibitor of microtubule formation, greatly inhibits cell division around the early mound stage as well as during the vegetative growth phase, when applied to exponentially growing Ax-2 cells. Essentially the same inhibition is attained by treatment of starved Ax-2 cells with CL-A, an inhibitor of serine/threonine protein phosphatases. The nocodazole- or CL-A-treated cells exhibit abnormal morphogenesis to form a stick-like multicellular structure on non-nutrient agar. More importantly, prespore differentiation as exemplified by prespore-specific Dp87 gene expression and prespore-specific vacuole (PSV) formation is greatly suppressed, whereas the differentiation of prestalk cells (ecmA-gal cells) is scarcely affected by drug treatments, thus suggesting the importance of cell-cycle progression around the early mound stage for prespore differentiation (Araki & Maeda 1998). This is consistent with the observation that 5′-bromo-2-deoxyuridine (BrdU)-labeled Ax-2 cells pass through the cell cycle (S-phase) during the mound-tipped aggregate stage, particularly in differentiating prespore cells (Zimmerman & Weijer 1993).

It has been demonstrated by Early et al. (1995) that a subtype of prestalk cells, pstA cells, appears at the peripheral ring of early mounds in a position-dependent manner, but is then sorted out into the apical tip of a tipped aggregate. Here it is easy to note the similarity of developmental behavior between pstA cells and T1 cells. To analyze more precisely the relation of the cell-cycle position at the onset of starvation to differentiation, double transformants bearing two plasmids – (i) a stable marker, the β-glucuronidase (GUS) gene with a nuclear tag under the control of the actin15 promotor, and (ii) a differentiation-specific reporter gene, cell-type specific gene promoters fused with the β-galactosidase (Gal) gene – were synchronized by temperature shift and mixed with nontransformed Ax-2 cells, followed by analysis of their behavior during development using immunohistochemical stainings of GUS and Gal. As expected, the result has showed that there is a good correlation between T1 cells and prestalk (pstA) differentiation; T1 cells starved just after the GDT point preferentially differentiate into pstA cells but never into prespore cells, while T7 cells starved just before the GDT point predominantly differentiate into prespore cells but never into pstA cells (Araki et al. 1997). Thus it is quite likely that in the normal development of Dictyostelium cells the cell-cycle position at the onset of starvation may determine the fate of cells, through dictating the cell’s position in the cell mass.

Cell-type choice and proportion regulation in a cell mass

The idea that the developmental fate of Dictyostelium cells is not determined cell autonomously and may be specified in a position-dependent manner is principally based on the fact that the slug is analogous to a regulated embryo: when cut, both the pre-stalk and prespore fragments exhibit transdifferentiation and restore the missing cell types in normal proportions (Raper 1940; Gregg 1965) . There is no difference in the number-ratio of prestalk–prespore cells in migrating slugs derived from pure populations of T1 and T7 cells, and the ratio is essentially the same as that in slugs derived from nonsynchronized cells (Maeda et al. 1989). Taken together with this fact, the data described above provide a more likely explanation for the occurrence of position-dependent differentiation, because the cell-cycle phase at the onset of starvation would dictate only the position of a cell within a cell mass in which cell-type proportioning occurs. However, a number of mechanisms have been proposed for the differentiation of two types in Dictyostelium, including cell-type choice determined either by a cell’s position within the cell mass or by non-positional factors. This will be reviewed by A. Chatwood & C. Thompson in this special issue.

Novel and multiple functions of mitochondria in the developmental system

In eukaryotic cells, mitochondria are self-reproducing organelles with their own DNA and they play a central role in adenosine triphosphate (ATP) synthesis by respiration. Increasing evidence indicates that mitochondria have novel and critical functions as the regulatory machinery of GDT, cell-type determination, cell movement, and pattern formation in the cellular slime molds. As mentioned in Growth/differentiation transition point (GDT-point) in the cell cycle, the expression of the mitochondrial ribosomal protein S4 (mt-rps4) gene is required for the initial differentiation of Dictyostelium cells from the GDT point (Inazu et al. 1999; Chida et al. 2008). In Drosophila, the germ cell line is determined by the large subunit of mitochondrial rRNA (mtlrRNA) (Iida & Kobayashi 1998; Kobayashi & Okada 1989). mtlrRNA of Dictyostelium slugs is also required for normal phototaxis and thermotaxis of a migrating slug (Wilczynska et al. 1997). Recently, a novel mitochondrial protein (Tortoise) has been shown to be essential for directional movement of Dictyostelium cells in cAMP gradients (van Es et al. 2001). Moreover, a Dictyostelium homologue (Dd-TRAP1) of TRAP-1, mitochondrial Hsp90, is closely involved in spore differentiation as well as the prestarvation response (PSR) (Morita et al. 2002, 2004, 2005; Yamaguchi et al. 2005).

In differentiating prespore cells, the mitochondrion exerts a drastic transformation to form a vacuole (M vacuole), in which a cell-type-specific organelle named a prespore-specific vacuole (PSV) is constructed with the help of the Golgi complex (Maeda 1971; Matsuyama & Maeda 1998). The mitochondria of plant cells sequentially develop crystal dilatations, spherical intracrystal inclusions, and eventually crystalloid inclusions within their cristae, all apparently in response to a series of cytoplasmic stimuli (Robert 1969). A drastic structural change of mitochondria has also been reported in the plant egg cells of Pelargonium zonale (Kuroiwa 2010).

Quite interestingly, it has been demonstrated that DIA2, one of the proteins specifically expressed in the initial differentiation of cells from the GDT-point, changes its location from ER to M-vacuoles (mitochondria-originated vacuoles) and eventually to prespore-specific vacuoles (PSVs) in differentiating prespore cells, and that exocytosis of PSVs from prespore cells and the subsequent spore differentiation are almost completely impaired in antisense-mediated transformants (dia2AS cells; Hirata et al. 2008).

Recently, developmental significance of cyanide (CN)-resistant respiration under stressed conditions has been asserted in experiments using Dictyostelium cells (Kimura et al. 2010). ρ cells with a reduced amount (about 1/4) of mitochondrial DNA (mtDNA), derived from D. discoideum Ax-2 cells, exhibit greatly delayed differentiation and fail to construct fruiting bodies (Chida et al. 2004). Also, prestalk differentiation is significantly enhanced in ρ slugs, while prespore differentiation is markedly inhibited (Chida et al. 2004).

Mitochondrial disease and mitochondria-dependent sterility as well as a close relationship between mitochondria and programmed cell death (apoptosis) have been widely recognized as notable events. An origin of mitochondrion is believed to be an aerobic bacteria that once established a symbiosis with a host cell such as archeabacteria and has been handing over parts of its own genome to the nuclear DNA of the host cell during evolution, thus resulting in failure of existent mitochondria to self-reproduce without the help of the nuclear genome. Surprisingly, however, it is evident that mitochondria still play critical and somewhat unexpected roles in a variety of cellular events including differentiation as well as in ATP synthesis by respiration.


A critical checkpoint precisely identified in the cell cycle for growth/differentiation transition and its relevance to a series of developmental events has been realized first in Dictyostelium, using its efficacy as a model organism and good method for cell synchrony. It is now quite important to know whether a specific checkpoint like the GDT point exists in the cell cycle of other organisms and is a key element in determining the cell-cycle dependency of development. The possible absence of a G1-phase in the Dictyostelium cell cycle is not so strange, because there is little or no G1-phase in rapidly dividing cells such as animal cells at the cleavage stage. The true slime mold Physalum and Hydra are also devoid of the G1-phase. The new world of mitochondria with several essential functions beyond our imagination is slowly but surely spreading every day. I hope that the data presented in this review will offer insightful suggestions on the weight of mitochondrial functions in cell differentiation and pattern formation as well as on the significance of the cell cycle in the field of cell and developmental biology. Again, cellular and molecular events occurring around the GDT point and also PSR-coupled events are particularly important for understanding the mechanisms controlling growth and differentiation. Such an analysis will also contribute to the understanding of the basal mechanism of abnormal growth as exemplified in tumor cells.


I thank Junji Chida (Tokushima University) and Ryu Itakura (Newton Press Inc.) for their excellent cooperation in preparing the figures presented in this review. Most of our recent work presented here was basically supported by a Grant-in-Aid (16370030 and 16657020) from JSPS. This work was also funded by the Mitsubishi Foundation.