Evidence that differentiation-inducing factor-1 controls chemotaxis and cell differentiation, at least in part, via mitochondria in D. discoideum

ABSTRACT Differentiation-inducing factor-1 [1-(3,5-dichloro-2,6-dihydroxy-4-methoxyphenyl)hexan-1-one (DIF-1)] is an important regulator of cell differentiation and chemotaxis in the development of the cellular slime mold Dictyostelium discoideum. However, the entire signaling pathways downstream of DIF-1 remain to be elucidated. To characterize DIF-1 and its potential receptor(s), we synthesized two fluorescent derivatives of DIF-1, boron-dipyrromethene (BODIPY)-conjugated DIF-1 (DIF-1-BODIPY) and nitrobenzoxadiazole (NBD)-conjugated DIF-1 (DIF-1-NBD), and investigated their biological activities and cellular localization. DIF-1-BODIPY (5 µM) and DIF-1 (2 nM) induced stalk cell differentiation in the DIF-deficient strain HM44 in the presence of cyclic adenosine monosphosphate (cAMP), whereas DIF-1-NBD (5 µM) hardly induced stalk cell differentiation under the same conditions. Microscopic analyses revealed that the biologically active derivative, DIF-1-BODIPY, was incorporated by stalk cells at late stages of differentiation and was localized to mitochondria. The mitochondrial uncouplers carbonyl cyanide m-chlorophenylhydrazone (CCCP), at 25–50 nM, and dinitrophenol (DNP), at 2.5–5 µM, induced partial stalk cell differentiation in HM44 in the presence of cAMP. DIF-1-BODIPY (1–2 µM) and DIF-1 (10 nM), as well as CCCP and DNP, suppressed chemotaxis in the wild-type strain Ax2 in shallow cAMP gradients. These results suggest that DIF-1-BODIPY and DIF-1 induce stalk cell differentiation and modulate chemotaxis, at least in part, by disturbing mitochondrial activity.


INTRODUCTION
The vegetative amebae of the cellular slime mold Dictyostelium discoideum feed on bacteria. Starvation initiates morphogenesis: cells gather to form a slug-shaped multicellular aggregate and differentiate into two distinct types ( prespore and prestalk), which eventually form a fruiting body consisting of spores and a multicellular stalk. Because of the simple pattern of its life cycle (cell differentiation and morphogenesis), D. discoideum is an excellent model in cell and developmental biology (Annesley and Fisher, 2009) (http://dictybase.org/).
To elucidate the mechanisms underlying the effects of DIF-1 (and possibly DIF-2), we synthesized two fluorescent derivatives of DIF-1, boron-dipyrromethene (BODIPY)-conjugated DIF-1 (DIF-1-BODIPY) and nitrobenzoxadiazole (NBD)-conjugated DIF-1 (DIF-1-NBD) (Fig. 1B,C), and investigated their localization and function in D. discoideum cells. We show that DIF-1-BODIPY, but not DIF-1-NBD, is bioactive and appears to function similarly to DIF-1: this derivative induces stalk cell formation in vitro in the presence of cAMP in HM44 (a DIF-deficient strain) (Kopachik et al., 1983) and suppresses chemotaxis of cells of the wild-type strain Ax2 in shallow cAMP gradients. We also show that DIF-1-BODIPY is undetectable inside the cells during an early stage of development but is localized to intracellular organelles, mainly mitochondria, during a later developmental stage. We examined the effects of DIF-1, DIF-1-BODIPY, and the mitochondrial uncouplers dinitrophenol (DNP) and carbonyl cyanide m-chlorophenylhydrazone (CCCP), and the results suggest that DIF-1 (and DIF-1-BODIPY) induces stalk cell differentiation and modulates chemotaxis, at least in part, via mitochondria.

Synthesis of fluorescent derivatives of DIF-1 and assay of stalk cell induction
The synthetic schemes of DIF-1-BODIPY and DIF-1-NBD are shown in Fig. 1B,C. We also synthesized the control compound butyl-BODIPY (Bu-BODIPY) (Kubohara et al., 2013). The effects of DIF-1, DIF-2, and the fluorescent compounds on in vitro stalk cell differentiation in the DIF-deficient strain HM44 are shown in Fig. 2. Even in the presence of cAMP, HM44 cells cannot differentiate into stalk cells in vitro unless exogenous DIF is supplied; therefore, HM44 cells are suitable for the assay of stalk cell induction by DIF-like molecules (Kopachik et al., 1983;Kubohara et al., 1993;Kubohara and Okamoto, 1994). As expected, DIF-1 or DIF-2 (2 nM) induced stalk cell formation in HM44 in the presence of cAMP; DIF-1-BODIPY (0.1-5 µM) dose-dependently induced stalk cell formation in up to 60%-80% of the cells under the same conditions (Fig. 2). By contrast, neither Bu-BODIPY (5 µM) nor DIF-1-NBD (0.1-5 µM) induced any stalk cell formation (Fig. 2). Cellular localization of DIF-1-BODIPY during in vitro stalk cell differentiation We next compared the cellular localization of DIF-1-BODIPY and DIF-1-NBD in HM44 cells. After 1-h starvation (incubation), cells were ameboid and were hardly stained with DIF-1-BODIPY or DIF-1-NBD (Fig. 3A), whereas cells fixed with formalin after starvation were stained well with the bioactive derivative DIF-1-BODIPY, but not with the nonbioactive derivative DIF-1-NBD (Fig. 3B).
We then compared cellular localization of DIF-1-BODIPY and the nonbioactive control compound Bu-BODIPY during in vitro differentiation of HM44 cells. After 1-h starvation (incubation), cells were hardly stained with DIF-1-BODIPY (Fig. 4A). After 20-h incubation with cAMP and DIF-1-BODIPY, cells were still ameboid; some of them had formed aggregates, in which some cells were stained with DIF-1-BODIPY, and there was heterogeneity among the cells (Fig. 4C). At 28 h, cells had begun to differentiate into stalk cells; one or more autophagic vacuoles had formed in each cell, each cell had formed a cell wall, and many cells were stained with DIF-1-BODIPY to a variable extent (Fig. 4E). At 48 h, most cells had differentiated into stalk cells and were stained with DIF-1-BODIPY; the signal was stronger in cytoplasmic regions than in autophagic vacuoles (Fig. 4G). However, cells fixed with formalin were stained with DIF-1-BODIPY at each time point (Fig. 4B,D,F,G). These observations suggest that DIF-1-BODIPY (and possibly DIF-1) is unable to penetrate into the cells or is pumped out from the cells during the early phase, but not during later phases, of cell differentiation. By contrast, Bu-BODIPY neither induced stalk cell formation nor was detected in the cells at any time point, even if they were fixed with formalin ( Fig. 4). Taken together, these results indicate that DIF-1-BODIPY can be used to probe cellular uptake and localization of DIF-1.

Effects of CCCP and DNP on stalk cell differentiation
We have recently shown that DIF-1 and its derivatives act as mitochondrial uncouplers in mammalian cells (Kubohara et al., 2013. To determine whether DIF-1 (and DIF-1-BODIPY) induces stalk cell differentiation by affecting mitochondria in D. discoideum, we examined the effects of CCCP and DNP on stalk cell formation in HM44. As shown in Fig. 6, CCCP (25-50 nM) or DNP (2.5-5 µM) weakly but significantly induced stalk cell formation in the presence of cAMP; at higher concentrations, both uncouplers were toxic to the cells (data not shown). The stalkinducing activities of CCCP and DNP did not exceed ∼10% and ∼20%, respectively (Fig. 6A); neither CCCP nor DNP showed additive effects with DIF-1 at a low concentration (0.4 nM). These results suggest that DIF-1 induces stalk cell differentiation partly by uncoupling mitochondrial activity but also via another pathway.

Effects of CCCP and DNP on chemotactic cell movement
To demonstrate that DIF-1 might function by disturbing mitochondrial activity, we examined the effects of CCCP and DNP on chemotactic cell movement toward cAMP (Fig. 7B). As expected, CCCP (25-50 nM), DNP (5 µM) and DIF-1 (10 nM) significantly suppressed chemotaxis of Ax2 cells in shallow cAMP gradients but hardly affected chemotaxis of gbpBcells (Fig. 7B).
These results indicate that all three compounds suppress chemotaxis via a GbpB-dependent pathway and that DIF-1 (and possibly DIF-1-BODIPY) might suppress chemotaxis in shallow cAMP gradients by uncoupling mitochondrial activity.

Localization of DIF-1-BODIPY in aggregating Ax2 cells
Finally, we localized DIF-1-BODIPY in aggregating (chemotacting) Ax2 cells under submerged conditions without exogenous cAMP; under these conditions, cells formed streaming aggregates (Fig. S1). After 3-h incubation, we still observed single ameboid cells; living cells were not stained with DIF-1-BODIPY, although formalin-fixed cells were clearly stained (Fig. S1A). At 15 h, cells formed aggregates, in which a small fraction of the cells was clearly stained with DIF-1-BODIPY; formalin-fixed cells were strongly stained (Fig. S1B).

DISCUSSION Biological activities of DIF-1-BODIPY
In this study, we designed and synthesized the fluorescent DIF derivative DIF-1-BODIPY (Fig. 1B) and found that DIF-1-BODIPY (0.1-5 µM) induced stalk cell differentiation in the presence of cAMP in the DIF-deficient strain HM44 (Fig. 2). DIF-1-BODIPY (1-2 µM) also suppressed chemotaxis in shallow cAMP gradients in Ax2 cells (Fig. 7). Although we do not exclude the possibility that DIF-1-BODIPY at several micromolars might affect cellular functions nonspecifically, the present results indicate that DIF-1-BODIPY can mimic the effects of DIF-1 in D. discoideum.

Subcellular localization of DIF-1-BODIPY to mitochondria
The hydrophobic indices of DIF-1 [ClogP (CP), 4.21] and DIF-1-BODIPY (CP, 5.85) (Fig. 1A,B) indicate that both compounds are likely to penetrate the cell membrane. However, we found that DIF- Cells were then stained for 0.5 h with Hoechst and (C) 5 µM DIF-1-BODIPY or (D) Bu-BODIPY, then washed and observed as above. Three-dimensional images were constructed from z-stacked two-dimensional (2D) images; two representative 2D projections of the 3D images are shown. Nonlinear adjustment was performed on 3D images to obtain clear high-contrast images. Note that DIF-1-BODIPY and MitoTracker co-localized to mitochondria.
1-BODIPY was absent in cells at early stages of development but gradually penetrated into or was taken up by cells differentiating to stalk cells (Fig. 4). DIF-1-BODIPY localized to mitochondria (Fig. 5). During early, but not late, stages of development, DIF-1-BODIPY (and thus DIF-1) might be pumped out of cells (Fig. 8). The molecular size of DIF-1-BODIPY is larger than that of DIF-1 and the hydrophobic indices of the two compounds are different (Fig. 1A,B). However, because Bu-BODIPY was not detected in any organelles (Figs 4 and 5), the mitochondrial localization of DIF-1-BODIPY was not caused by the BODIPY moiety but likely reflects localization of DIF-1.
DIFs and their derivatives possess anti-tumor activities when tested on mammalian tumor cells, and derivatives of DIF-3 are more potent anti-tumor agents than those of DIF-1 (Asahi et al., 1995;Gokan et al., 2005;Kubohara, 1999;Takahashi-Yanaga et al., 2003;Akaishi et al., 2004;Kubohara et al., 2015;Oladimeji et al., 2015). The fluorescent DIF-3 derivative BODIPY-DIF-3 penetrates the cell membrane, localizes to mitochondria, and suppresses cell growth in some of the tumor cell lines tested (Kubohara et al., 2013. Bioactive DIF derivatives and the mitochondrial uncoupler CCCP promote oxygen consumption in mitochondria isolated from mouse liver; the anti-tumor activity of DIF derivatives might result, at least in part, from their uncoupling activity in the mitochondria of tumor cells (Kubohara et al., 2013. In D. discoideum, DIF-1 can disturb mitochondrial membrane potential and respiration, suggesting that it might function as an uncoupler (Shaulsky and Loomis, 1995), although the effective concentration range of DIF-1 (0.1-1 µM) rather exceeded its putative physiological concentrations (at most ∼0.1 µM; Kay, 1998). However, DIF-1 at 0.1 µM was later shown to affect mitochondrial membrane potential in Ax2 cells (Arnoult et al., 2001) and to promote mitochondrial oxygen consumption and induce ATP depletion in an autophagy mutant strain (Laporte et al., 2007;Giusti et al., 2009), suggesting that it can act as an uncoupler at physiological concentrations. In the present study, we have shown that the mitochondrial uncouplers CCCP and DNP induce partial stalk differentiation of HM44 cells (Fig. 6) and that the uncouplers and DIF-1 suppress chemotaxis in Ax2 cells in shallow cAMP gradients (Fig. 7B). Taken together, these data suggest that DIF-1 might function, at least in part, via mitochondria ( possibly as an uncoupler) in D. discoideum. Unexpectedly, however, neither CCCP (25-50 nM) nor DNP (2.5-5 µM) showed an additive effect with 0.4 nM DIF-1 on stalk cell formation (Fig. 6). Although the cause of the absence of such effect is unknown, DIF-1 might function via multiple signaling cascades, only one of which may be mimicked by CCCP and DNP; DIF-1 at 0.4 nM might be sufficient to saturate this pathway. Alternatively, as CCCP >50 nM and DNP >5 µM were toxic to the cells (data not shown), their toxicity might cancel their stalkinducing activity in the presence of DIF-1.

Biological activity and cellular localization of DIF-1-NBD
In the present study, we have also synthesized DIF-1-NBD, a fluorescent amide derivative of DIF-1 (Fig. 1C). DIF-1-NBD was expected to be a good probe for DIF-1 because some amide derivatives of DIF-1 are excellent inducers of stalk cell differentiation in HM44 cells (Kikuchi et al., 2008), and also because the molecular size of DIF-1-NBD is much smaller than that of DIF-1-BODIPY and its CP value (3.51) suggests that DIF-1-NBD can penetrate the cell membrane (Fig. 1A). Unfortunately, however, DIF-1-NBD (5 µM) neither induced stalk cell differentiation (Fig. 2) nor appeared to localize to any parts of HM44 cells (Fig. 3). Although we cannot exclude that the localization of DIF-1-BODIPY in mitochondria reflects its non-specific binding because of its high concentration, the absence of cell staining or activity of DIF-1-NBD (5 µM) and another non-bioactive compound Bu-BODIPY (5 µM) (Figs 2-4) indicates that it is likely that the biological activities and cellular localization of DIF-1-BODIPY (5 µM) reflect those of DIF-1 at nanomolar concentrations.

Cell lines and cell culture
The D. discoideum DIF-deficient strain HM44 (Kopachik et al., 1983) was used for in vitro stalk cell induction assay. The axenic strain Ax2 and the gbpB null strain gbpBderived from Ax2 (Bosgraaf et al., 2002a,b;Goldberg et al., 2002) were used for chemotaxis assay. These strains were obtained from the National BioResource Project (NBRP Nenkin, Tsukuba, Japan). HM44 cells were grown in association with Klebsiella aerogenes on a modified SM agar plate (Inouye, 1988) at 21°C, whereas Ax2 and gbpBcells were grown axenically at 21°C in HL-5 liquid medium (Sussman, 1987). Growing cells were collected by centrifugation (500× g, 3 min).

Low-magnification phase-contrast and fluorescence microscopy
Starved HM44 or Ax2 cells were incubated for the indicated times with 1.5 ml of the stalk salt solution [2 mM NaCl, 10 mM KCl, 1 mM CaCl 2 , 50 µg ml −1 penicillin, 100 µg ml −1 streptomycin sulfate and 10 mM 2morpholinoethanesulfonic acid-KOH (MES-KOH) pH 6.2 containing various additives (DIF-1, DIF-1-BODIPY, Bu-BODIPY, DIF-1-NBD, and/ or cAMP) in 35-mm tissue culture dishes (Becton Dickinson, Franklin Lakes, NJ, USA) (5×10 5 to 10 6 cells/dish). The cells were washed three times with the stalk salt solution and submerged in 1.5 ml of the same solution. The cells were observed at room temperature with a Leica DM IRB fluorescence microscope (Leica, Wetzlar, Germany), and digitized images were analyzed with the Leica Application Suite (version 3.3.0).
Alternatively, cells were incubated for the indicated times with additives (DIF-1 and/or cAMP) and fixed for 15-20 min at room temperature in 2 ml of 3.7% (v/v) formaldehyde in PBS(-) (20 mM phosphate buffered saline, pH 7.4), washed three times in PBS(-), stained for 30 min with DIF-1-BODIPY, Bu-BODIPY, or DIF-1-NBD, washed three times with PBS(-), and observed at room temperature under the same microscope.
Multi-color imaging of formalin-fixed cells Starved HM44 cells were incubated at 21°C for 1 h with 1.5 ml of the stalk salt solution containing MitoTracker (0.2 µM) in 35-mm µ-Dishes (ib81156; ibidi, Martinsried, Germany) (5×10 5 cells/dish). Alternatively, starved HM44 cells were incubated for 20 h with 1.5 ml of the stalk salt solution containing 5 mM cAMP in 35-mm plastic dishes (5×10 5 cells/dish) and then for 1 h with the same solution containing MitoTracker (0.2 µM). Cells were fixed for 15-20 min at room temperature with 2 ml of 3.7% (v/v) formaldehyde in PBS(-), washed three times with PBS(-), and stained for 30 min with Hoechst 33342 (1 µg ml −1 ) and DIF-1-BODIPY (5 µM) or Bu-BODIPY (5 µM). Cells were washed three times with PBS(-), submerged in 1.5 ml of PBS(-) and observed at room temperature with a Keyence BZ-9000 fluorescence microscope (Keyence, Osaka, Japan) equipped with an oil immersion 100× lens (CFI Plan Apo VC100XH) (Keyence) and multi-filters that can distinguish up to four fluorescent probes simultaneously. The original photos were deconvoluted using the Keyence BZ analyzer software to reduce 'haze'. Z-stack sections were collected at 0.4-µm intervals, and the deconvoluted images were compiled into threedimensional (3D) images. All color images are presented in pseudo-colors.

Small-population assay of chemotaxis
Chemotaxis toward cAMP was assessed by using Ax2 and gbpBstrains in the presence of additives as described previously Kubohara, 2009, 2016;Kuwayama et al., 2011).

Hydrophobic index
To estimate the membrane permeability of each compound, its hydrophobic index (CP) (Fig. 1) was calculated by using ChemDraw10.0 software (Cambridgesoft, Cambridge, MA, USA).

Statistical analysis
Unpaired Welch's t-test (one-tailed) was used. P<0.05 was considered to indicate significant differences.