Real-time measurements of cAMP production in live Dictyostelium cells

The ubiquitous second messenger cAMP regulates a wide range of cellular processes, including differentiation, proliferation and cell motility. Cyclic AMP is also essential for the survival of the social amoebae Dictyostelium discoideum (Manahan et al., 2004). Upon starvation, these cells spontaneously aggregate and enter a developmental program that leads to the formation of a multicellular structure that can withstand harsh environmental conditions. A few hours after the initiation of starvation, cells start to synthesize cAMP and whereas part of the cAMP produced remains inside the cells to activate downstream effectors, part of the cAMP is secreted, functioning as a chemoattractant. The intracellular cAMP binds the regulatory subunit of PKA and releases its catalytic subunit, which regulates gene expression and is essential for development (Aubry and Firtel, 1999). Cells lacking the catalytic subunit of PKA fail to enter development (Anjard et al., 1993; Mann and Firtel, 1993). The secreted cAMP binds and stimulates specific G-protein-coupled receptors (named cARs for cAMP receptors), leading to the activation of a variety of effectors that control chemotaxis, gene expression and signal relay, via the synthesis and secretion of additional cAMP (Manahan et al., 2004).

Because cAMP is essential for the survival of Dictyostelium, its synthesis, degradation, and detection are exquisitely regulated. The synthesis of cAMP is mediated by adenylyl cyclases, which convert ATP into cAMP. Dictyostelium cells express distinct adenylyl cyclases during development (Kriebel and Parent, 2004). The enzyme responsible for cAMP production in early development is ACA (adenylyl cyclase expressed during aggregation), which is activated through cAR1 in a PI3K-dependent fashion. Cells lacking ACA fail to enter development and remain as smooth monolayers when starved (Pitt et al., 1992). We previously established that ACA is distributed in two distinct pools in polarized cells. One is restricted to the plasma membrane; the other is localized on highly dynamic intracellular vesicles that coalesce at the back of cells and is required for cells to migrate in a head-to-tail fashion and stream during chemotaxis (Kriebel et al., 2003; Kriebel et al., 2008). Another adenylyl cyclase expressed during aggregation is ACB (adenylyl cyclase B). In contrast to ACA, ACB appears to be constitutively active and is proposed to be required for terminal differentiation and is unresponsive to G-protein activation (Kim et al., 1998; Meima and Schaap, 1999; Anjard et al., 2001). Intracellular cAMP levels are also regulated by phosphodiesterases (PDEs), which degrade cAMP into 5′-adenosine monophosphate. Dictyostelium cells express four distinct PDEs with activity against intracellular cyclic nucleotides, with RegA (also referred to as PDE2) being the main high-affinity cAMP-specific PDE (Shaulsky et al., 1998; Bader et al., 2007). Cells lacking RegA behave similarly to cells lacking the regulatory subunit of PKA and exhibit constitutive PKA catalytic activity. Both cell lines show precocious development and aberrant terminal differentiation (Shaulsky et al., 1998).

Studies performed at the single cell level in Dictyostelium have shown the spatio-temporal dynamics of several chemotactic signaling components (Van Haastert and Devreotes, 2004; Bagorda et al., 2006). However, no studies have reported intracellular cAMP dynamics at the single cell level during chemotaxis. In the past few years the development of fluorescence resonance energy transfer (FRET)-based sensors has allowed the investigation of intracellular cAMP production in single live mammalian cells (Willoughby and Cooper, 2008). These types of studies have postulated new biological concepts, such as intracellular cAMP compartmentalization in cardiac myocytes (Zaccolo and Pozzan, 2002) and waves of intracellular PKA activation in retinal neurons (Dunn et al., 2006). Based on these sensors, we have expressed a functional FRET-based cAMP probe in Dictyostelium cells. We aimed to determine (1) the intracellular cAMP dynamics following chemoattractant stimulation in single live Dictyostelium cells; (2) the specific contribution of ACA, ACB and RegA in this response; and (3) whether the distinct ACA distribution at the back of cells leads to a spatially restricted intracellular cAMP pool in chemotaxing cells.

The tetrameric PKA-based sensor is the first FRET probe developed to detect cAMP levels in live cells and it has been extensively used to monitor the intracellular cAMP dynamics in mammalian systems (Adams et al., 1991). It consisted of catalytic (C) and regulatory (R) subunits of PKA, chemically labeled with fluorescein and Rhodamine, respectively. In the absence of cAMP, both subunits form a tetrameric holoenzyme complex, R2C2, in which a high FRET signal between fluorescein and Rhodamine is observed. When cAMP binds to the regulatory subunits, the catalytic subunits dissociate, leading to a loss of FRET. Various modifications of this probe were later generated, where the donor came from the regulatory RII β-subunit of mammalian PKA tagged with CFP and the acceptor came from the catalytic subunit tagged with YFP (Fig. 1A) (Zaccolo and Pozzan, 2002). Here again, in the presence of low cAMP concentrations the probe retains its tetrameric structure and its FRET conformation. When cAMP levels increase and bind to the regulatory subunit, it dissociates from the catalytic subunit and the FRET signal is lost. To measure cAMP changes with this FRET-based sensor it is necessary to have the same level of expression of the donor and acceptor. We engineered the two constructs in Dictyostelium expression plasmids, co-transfected them in wild-type (WT) cells and assessed the expression level of the two plasmids using western analysis. During Dictyostelium development, a process during which cells express the various chemotactic signaling components and become polarized and motile, we observed a dramatic decrease in the expression levels of the CFP-RII β-subunit (Fig. 1B,C). We do not know why the CFP signal decreases during development. We speculate that overexpression of the active PKA catalytic subunit (the acceptor) downregulates the expression of the PKA regulatory subunit. Indeed, we observed that transfected cells were defective in development, because they were not able to form fully differentiated fruiting bodies (data not shown). Since the development of Dictyostelium is a PKA-dependent process, the lack of completion of the process suggests that PKA overexpression is altering the physiology of the cells. In any case, the uneven expression of the FRET acceptor and donor makes it impossible for us to use the tetrameric PKA sensor to detect intracellular cAMP dynamics in Dictyostelium cells.

We next constructed a FRET monomeric sensor based on the probe originally designed by Nikolaev and collaborators (Nikolaev et al., 2004). We cloned the cAMP-binding domain B from the mammalian PKA RII β-subunit between eCFP and eYFP in a Dictyostelium expression vector (see Materials and Methods for details). When cAMP levels are low, the probe retains a molecular conformation that allows FRET between the eCFP and eYFP (Fig. 2A). When cAMP levels increase and bind the B-domain, a conformational change of the probe ensues leading to a decrease of the FRET signal. The B-domain cAMP sensor, expressed in WT Dictyostelium cells, showed a uniform cytoplasmic distribution and a comparable level of the CFP and YFP signal (Fig. 2B). Since the probe binds cAMP, we worried that it might act as a sink for intracellular cAMP and alter development. We therefore carefully assessed the phenotype of cells expressing the B-domain. We first measured the ability of the cells to chemotax towards a point source of external cAMP. As shown in Fig. 2C, the cells transfected with the B-domain sensor were able to migrate directionally and align in a head to tail fashion during chemotaxis similarly to WT cells. In addition, we found that the transfected cells exhibited normal development when plated on non-nutrient agar showing characteristic aggregation streams and fruiting body formation (Fig. 2D). Taken together, these findings show that, in contrast to the tetrameric cAMP probe, the B-domain cAMP sensor is not toxic to cells, and is therefore a good method to study cAMP dynamics in Dictyostelium.

We next set out to visualize cAMP dynamics by measuring FRET efficiency changes in chemotaxis-competent cells. Confocal images were recorded as required by the acceptor-sensitized fluorescence technique and processed using the pFRET software (see Materials and Methods), which calculates FRET efficiency values. Fig. 3 shows a montage of images depicting the change in FRET signal observed as a function of time following a uniform saturating dose of chemoattractant (10 μM) in a representative cell (also see supplementary material Movie 1). The graph shows that upon receptor stimulation, there was a robust decrease in FRET efficiency (20.4±1.4%, n=10), which remained sustained during the 9 minute time course.

To determine whether the FRET efficiency decrease is specific for cAMP produced under receptor stimulation, we generated mutated versions of the B-domain probe lacking the ability to bind cAMP (supplementary material Fig. S1) (DiPilato et al., 2004; Lopez De Jesus et al., 2006). The two mutated versions of the B-domain probe were expressed to similar levels as the WT probe in WT cells and did not alter Dictyostelium development (data not shown). As depicted in Fig. 4A, chemoattractant stimulation of chemotaxis-competent cells transfected with either of the two mutated versions of the probe did not give rise to a change in FRET efficiency, strongly suggesting that the WT version of the FRET sensor specifically detects cAMP changes. To determine which AC pool the FRET efficiency decrease was specific for, we transfected the B-domain probe in aca^– and acb^– cells. We found that the decrease in FRET efficiency was highly specific for the cAMP produced by ACA, because aca^– cells showed no significant decrease in FRET efficiency, whereas the acb^– cells behave similarly to WT cells (Fig. 4B). These findings also establish that the large change in FRET response we observed in WT cells was not due to changes in cell shape. Indeed, aca^– cells gave a normal cringe response when stimulated with chemoattractants, yet they showed no significant decrease in FRET efficiency. Together, these findings show that the B-domain cAMP probe specifically and robustly detects cAMP derived from ACA in chemotaxis-competent cells. Moreover, the results with aca^– cells clearly show that the FRET response is not affected by the exogenous cAMP stimulus nor does it respond to changes in intracellular cGMP levels, which occur normally in aca^– cells (Kuwayama and Van Haastert, 1998).

Fig. 3 shows that the intracellular levels of cAMP detected by the B-domain sensor are sustained during the entire time course of the experiment. This finding contrasts with previous biochemical studies, which show that once synthesized intracellular cAMP is rapidly cleared through intracellular degradation or secretion (Dinauer et al., 1980). We reasoned that the sustained FRET response observed under saturating stimulation conditions (10 μM) is dependent on the affinity of the B-domain probe for cAMP. A high-affinity probe would quickly saturate and not represent a true readout of the cAMP levels in cells. We therefore measured the affinity of the probe by performing in vitro fluorometric measurements of FRET responses on lysates from WT cells expressing the B-domain probe. This assay shows that the probe has an EC₅₀ value <600 nM (Fig. 5A). To gain further insight into the ability of the probe to dissociate cAMP, we determined the capacity of the probe to return to its full FRET conformation upon the removal of cAMP. We found that the exogenous addition of the phosphodiesterase PdsA (Lacombe et al., 1986; Garcia et al., 2009) completely and rapidly reversed the loss of FRET caused by the addition 600 nM cAMP (Fig. 5B). These findings establish that the dissociation kinetic of our B-domain probe is fast enough to sense the rapid dynamic changes in intracellular cAMP.

Based on the estimation that intracellular cAMP levels can easily reach μM ranges (Roos et al., 1977), we concluded that our probe saturated rapidly, giving rise to a sustained FRET response. Indeed, when the cells were stimulated with a subsaturating dose of cAMP (1 nM) we observed a transient profile of cAMP accumulation, which reached a peak about 2-3 minutes after receptor stimulation and returned close to basal levels by 5-8 minutes (Fig. 5C). Based on previously published work, we attempted to generate a B-domain sensor with lower affinity for cAMP. Unfortunately, our efforts gave rise to a probe exhibiting an affinity that was too low (for details, see supplementary material Fig. S2). Taken together, our findings establish that the relative high affinity of our B-domain probe for cAMP allows the accurate detection of cAMP levels under subsaturating receptor stimulation conditions, in accordance with previously published biochemical measurements.

To investigate the contribution of intracellular phosphodiesterase activity in the transient accumulation of cAMP we observed following subsaturating stimulation, we expressed the B-domain sensor in cells lacking RegA (Shaulsky et al., 1998; Bader et al., 2007). As shown in Fig. 5D, stimulation of these cells with either a saturating or a subsaturating concentration of chemoattractant, gave rise to a decrease in FRET efficiency that remained constant during the entire experimental time course. Moreover, when we compared the chemoattractant-mediated peak FRET response of WT and regA^– cells, we found that the relative FRET decrease was lower in cells lacking RegA (Fig. 5E). We measured significantly lower basal raw FRET values in regA^– cells compared with WT cells (18.6±0.5, n=25, versus 22.9±1.7, respectively; n=18; P=0.008), which could be due to higher basal cAMP levels in regA^– cells. Nevertheless, because both degradation and secretion regulate intracellular cAMP levels, we were surprised to find that intracellular cAMP levels remain high for up to 10 minutes after a subsaturating stimulation with chemoattractants in regA^– cells. This raised the possibility that regA^– cells cannot secrete cAMP. To examine this further, we subjected chemotaxis-competent regA^– cells to a micropipette filled with cAMP and monitored their ability to form stream during chemotaxis – a process that requires cAMP secretion (Kriebel et al., 2003). As depicted in supplementary material Movie 2, these cells responded to the micropipette by migrating towards its tip in a head-to-tail fashion to form streams. In contrast to WT cells, which formed long streams that resulted in large aggregation centers, regA^– cells formed smaller aggregation centers and correspondingly shorter streams. We also used the EZ-Taxiscan chemotaxis chamber to assess chemotaxis and streaming. Using microfluidics, this system generates stable shallow chemoattractant gradients in a small, restricted area. Under these conditions, cells lacking RegA showed robust chemotaxis and signal relay as they readily formed streams during migration and aggregated into small clusters (see supplementary material Movie 3). Furthermore, as observed in WT cells, we found that ACA-YFP was enriched at the back of polarized regA^– cells (data not shown). These findings establish that cells lacking RegA have the ability to secrete cAMP and relay signals to neighboring cells. It therefore appears that the transient intracellular cAMP accumulation observed under subsaturating conditions (measured by a return of FRET response to basal values) depends on the intracellular PDE activity of RegA.

We next wanted to determine the spatio-temporal dynamics of intracellular cAMP in actively chemotaxing cells. We exposed chemotaxis-competent WT cells expressing the monomeric sensor to a micropipette filled with cAMP and monitored the FRET signal during chemotaxis. Fig. 6 depicts a montage of FRET images taken as cells are migrating (also see supplementary material Movie 4). Although we clearly observed variations in cytosolic cAMP levels in chemotaxing cells, careful analyses of the change in FRET efficiency in the migrating cells did not reveal intracellular compartmentalization of cAMP. Similarly, we did not observe compartmentalization in cells spontaneously migrating in streams (data not shown). From these findings, we conclude that the monomeric cAMP sensor does not reveal cAMP compartmentalization in cells chemotaxing to an external source of chemoattractant or in cells responding to natural cAMP waves.

We aimed to measure intracellular cAMP dynamics in response to chemoattractant stimulation in Dictyostelium cells. Although cAMP has a crucial role throughout the developmental program of Dictyostelium, we focused on measuring cAMP fluctuations in single cells that were synchronized to the chemotaxis-competent stage. This allowed us to limit changes in responses that are dependent on aberrant development (particularly in mutant cell lines) and to focus on a homogenous cell population. Our findings establish that, following chemoattractant stimulation, the intracellular cAMP produced is synthesized by the adenylyl cyclase ACA and, as expected, the kinetics of the intracellular cAMP levels match the kinetics of ACA activation, which peak 1 minute after chemoattractant addition and slowly return to basal levels after ∼5 minutes (Franca-Koh et al., 2006). Our findings also establish that the intracellular cAMP produced is degraded by the phosphodiesterase RegA. In cells lacking RegA, the cAMP produced following a subsaturating stimulation of chemoattractant does not decline and remains elevated. Using a different FRET probe that possesses a lower affinity for cAMP, the group of Satoshi Sawai reported similar transient cAMP fluctuations in WT cells (Gregor and Sawai, http://dictybase.org/DictyAnnualConference/dicty2007.pdf; Gregor, Fujimoto and Sawai, http://dictybase.org/DictyAnnualConference/dicty2008.pdf). Intriguingly, they also observed oscillations with ∼3 minute period, under persistent chemoattractant stimulations. We have looked extensively for spatio-temporal changes in intracellular cAMP levels under a variety of conditions during chemotaxis and have not observed oscillatory fluctuations in intracellular cAMP levels using our probe. We suspect that the relatively lower affinity of their probe for cAMP (K_d=2 μM) allowed them to observe these oscillations.

Interestingly, it has been shown that, in chemotaxis-competent cells, intracellular cAMP is rapidly degraded and secreted (Dinauer et al., 1980). However, in regA^– cells, cAMP levels remain high for up to 10 minutes after a subsaturating stimulation with chemoattractants. We have shown that these sustained intracellular cAMP levels are not due to the inability of regA^– cells to secrete cAMP, because these cells can relay signals to neighboring cells, albeit abnormally. Indeed, Sawai and colleagues have shown that although regA^– cells have the ability to propagate chemoattractant waves, they do so less regularly and over shorter distances than seen in WT cells (Sawai et al., 2005). Although it is possible that the relative high affinity of our probe for cAMP does not allow us to measure a transient cAMP response in cells lacking RegA, we do not think that this is the case. First, although we measured higher basal cAMP levels in regA^– cells, our probe retains a high basal FRET value in these cells that can further respond to cAMP fluxes. Second, in these cells the extent of ACA activation following a subsaturating stimulation gives rise to a small activity peak at 30 seconds, similarly to what we observe in WT cells (supplementary material Fig. S3). Nevertheless, in contrast to WT cells, we never measured a return to basal intracellular cAMP levels in the chemotaxis-competent regA^– cells. To reconcile these findings, we propose that the two cellular pools of ACA generate two distinct pools of cAMP. In this scheme, the plasma-membrane-bound ACA would give rise to intracellular cAMP that is poised to activate PKA and is subject to degradation by RegA. Alternatively, the vesicular pool of ACA that is enriched at the back of cells (Kriebel et al., 2008) could produce cAMP that is destined for secretion and signal relay. Indeed, our probe is exclusively found in the cytoplasm of cells. Moreover, using electron microscopy, we have shown that the vesicular pool of ACA is mainly found in multivesicular bodies (MVBs) with the catalytic site of ACA facing the cytoplasmic side of the small vesicles released, which are called exosomes (Kriebel et al., 2008). Consequently, we propose that our monomeric cAMP FRET sensor solely detects the intracellular pool of cAMP synthesized from the plasma-membrane-bound ACA, which builds up in the absence of RegA, resulting in a prolonged FRET response. Our model would also explain why we do not measure cytosolic cAMP compartmentalization in chemotaxing and streaming cells. Since cells with aberrant intracellular cAMP dynamics, such as regA^– and pkareg^– cells, exhibit aberrant external cAMP wave propagation (Sawai et al., 2005), we envision that internal cAMP levels ultimately regulate cAMP secretion. This could occur via the cAMP effectors PKA or EPAC, similarly to that shown for various exocytosis events in mammalian cells (Szaszak et al., 2008). In particular, PKA has been shown to control the exocytosis of chemokine-containing storage granules in endothelial cells (Oynebraten et al., 2005). We foresee that compartmentalized ACA activity would exquisitely regulate differentiation and cell-cell communication events during starvation. Additional experiments are currently under way to specifically test this hypothesis.

The RII β-CFP and the CAT-YFP plasmids were generated by amplification of the RII-CFP and C-YFP plasmids (Zaccolo and Pozzan, 2002). The amplification products were cloned into the unique BglII site of the Dictyostelium expression vector CV5, a p88d1-base plasmid with the Actin15/2H3 expression cassette (Johnson et al., 1991; Hughes et al., 1994), which provides stable protein expression throughout development (data not shown). The two plasmids were co-electroporated into Dictyostelium cells, as previously described (Howard et al., 1988), using a Bio-Rad gene pulser. Cells stably co-expressing the two plasmids were selected in medium containing 20 μg/ml G418. The DNA construct for the FRET B-domain sensor was generated through several cloning steps. First, the B-domain was obtained by PCR amplification of the mammalian cAMP-binding domain B (M264 to A403) using the RII-CFP plasmid as template and cloned into pCR 2.1 TOPO TA Cloning Vector (Invitrogen). Next the ECFP gene was amplified from Living Colors pECFP plasmid (BD Biosciences, NJ) and cloned at the 3′-end of the B-domain cDNA. The resulting B-domain-ECFP fusion was cloned into the pDXA-YFP-MCS1 vector (Knetsch et al., 2002) using the unique XhoI site, such that the YFP gene was at the 5′-end of the B-domain and finally cloned into the CV5 expression plasmid. Point mutations were introduced into the B-domain sensor construct using a QuikChange II XL-site-directed mutagenesis kit (Stratagene). The various expression plasmids have been deposited in the Dictyostelium Stock Center and are available publicly. Each sensor was electroporated in Dictyostelium cells and positive cells were selected in medium containing 20 μg/ml G418.

WT (AX₃), B-domain WT, B-domain aca^–, B-domain acb^– and B-domain regA^– cells were grown in shaking cultures in HL5 medium, harvested by centrifugation, and differentiated to the chemotaxis-competent stage by resuspension in DB (5 mM Na₂PO₄, 5 mM NaH₂PO₄, pH 6.2, 2 mM MgSO₄, and 200 μM CaCl₂) and providing exogenous pulses of 75 nM cAMP as previously described (Kriebel et al., 2003).

Chemotaxis was measured by placing cells in a gradient of chemoattractant generated by a micropipette as previously described (Kriebel et al., 2003). The EZ-Taxiscan experiments were performed as described by the manufacturer on non-coated surfaces (Effector Cell Institute, Tokyo, Japan). Development on non-nutrient agar was performed as previously described (Devreotes et al., 1987). All images in a series were processed identically using IVision and/or Adobe Photoshop software.

A series of single-plane time-lapse images were collected with a Zeiss Axiovert 200M microscope fitted with the Zeiss LSCM 510 Meta confocal system (Carl Zeiss, Thornton NY), using a 63× Plan-Apochromat objective (Carl Zeiss). To measure FRET by the donor-sensitized acceptor fluorescence technique, three images were acquired for each set of measurements: (1) YFP fluorescence (514 nm excitation, 530 nm LP emission filter; (2) CFP fluorescence (458 nm excitation, 475-505 nm BP emission filter); (3) FRET (458 nm, 530 nm LP emission filter). The last is an uncorrected FRET image, which contains the `bleed-through' of the CFP emission into the YFP detection channel and the cross-excitation of YFP by the CFP excitation laser. To obtain corrected FRET images, we acquired reference images from single-labeled CFP- and YFP-expressing cells (using the same set of acquisition parameters), as required by the pFRET Software (CircuSoft Instrumentation LLC, Hockessin, DE) that we subsequently used to analyze our FRET results (Wallrabe and Periasamy, 2005).

By loading each Tiff file from every set of experiment in the pFRET software, we obtained FRET efficiency images and, for each of them, the average of FRET efficiency values. Those values were then normalized using, as reference value, the average of the three images preceding the stimulation. The reference value was then set to 100%.

B-domain WT cells were starved for 2 hours in DB, centrifuged, washed twice and finally resuspended at 8×10⁷ cells/ml in ice cold 10 mM Tris-HCl, pH 7.5, 0.2 mM EGTA, 200 mM sucrose. Cells were rapidly filter-lysed using 5 μm nucleopore filters. The resulting lysates were centrifuged at 9000 r.p.m. for 20 minutes at 4°C and the resulting supernatant used for analysis. Fluorescent emission spectra of total cell lysates from B-domain WT cells were measured (excitation 425 nm, emission range 460-550 nm) with a luminescence spectrometer (Aminco Bowman Series 2, Thermo Electron) before and after adding various concentration of cAMP. In a subset of experiments, partly purified PdsA (Garcia et al., 2009) was added immediately after stimulation with cAMP. The ratio of the CFP emission peak over YFP emission peak was calculated and expressed as a percentage of the maximum ratio, obtained with the saturating cAMP concentration.