Amts in most microorganisms, fungi and plants function to import ammonia in order to obtain nitrogen as a nutrient source. In contrast, Dictyostelium cells generate excess ammonia due to the extensive use of protein catabolism during feeding on other microbes or on an enriched broth. The Amt null strains and the parental wild-type strain, Ax4, were used to test if one or more of the Amts function in ammonia release. Excreted ammonia, trapped within buffer soaked supporting pads, as well as intracellular ammonia levels were examined during the developmental program from 12 to 18 hours post-starvation. This is the time period when AmtA and AmtC function in the choice between slug formation versus culmination 2224.

Developing cells excreted ammonia in a linear manner with regard to time after the initiation of development. All strains disrupted in at least one amt gene excreted less ammonia and had a reduced rate of ammonia release relative to the parental strain (Figure 1). Ammonia release from the various strains during the first 12 hours of development (not shown) was consistent with that presented in Figure 1, suggesting a relatively constant rate of ammonia excretion from initiation to 18 hours of development that is dependent on the presence or absence of Amts. Similar kinetics of ammonia release were previously observed for the NC4 strain, a wild isolate 19. While the rates for two of the double null strains were lower than the respective single nulls, this was not the case for amtB-/amtC-. No compensatory effect was found at the mRNA level to account for this (not shown).

In contrast to alterations in amounts of ammonia released from the Amt mutant strains, the intracellular levels of ammonia for the single and double null strains were no different from that in the parental strain. In cells growing on bacteria, all strains contained approximately 0.05 micromoles of ammonia per 6 × 10⁷ cells. By three to five hours post-starvation the intracellular levels in all strains had dropped to about 0.025 micromoles, or half the levels found in growing cells. This level was maintained for at least 18 hours post-starvation in all strains.

Using the Act15 promoter, fusion constructs of the Amts with C-terminus YFP and RFP tags were made and transformed into the relevant null strains. To test the functionality of the fusion proteins, excreted ammonia from the transformed null strains was compared to that of the untransformed parental and null strains. For each of the three null strains, ammonia excretion was increased when the strains were transformed with the plasmid encoding the appropriate RFP fusion protein (Figure 2). For the amtA and amtB null strains, excretion levels were equivalent to or slightly less than that of the parental strain possessing the non-disrupted, endogenous amt gene. Expression of AmtC-RFP in amtC- cells resulted in ammonia excretion levels that were intermediate to those of the parental and amtC null strains. The respective YFP fusions also restored normal ammonia excretion levels when expressed in the amtA and amtB null strains but not in the amtC null (not shown).

The fusion constructs of the three Amts were transformed into Ax4, and the same localization patterns were seen in growing cells of both the wild type and null strains. To our knowledge, for all previous species examined, the Amt/MEP/Rh proteins have been found exclusively on the plasma membrane (see reviews 1726) with the sole exception being RhgA, one of two Rh homologues in Dictyostelium 27. As expected, AmtA-RFP was localized on the plasma membrane, and surprisingly, more strongly on subcellular organelles (Figure 3A). This also was true for AmtA-YFP, and when the two fusion proteins were co-expressed to confirm their interchangeability, colocalization was essentially 100% (Figure 3B). Because the internal organelle fluorescence levels were so high, resolving them photographically frequently required lowering the exposure time to such an extent that the plasma membrane localization was no longer visible. Expression of YFP and RFP without being fused to an Amt resulted in cytosolic localization for both proteins.

Well defined subcellular membrane and organellar markers exist for Dictyostelium. Colocalization of a fluorescently tagged protein expressed by the Act15 promoter with the markers has been used extensively to define subcellular localization 28293031. To begin defining AmtA subcellular localization, the AmtA-RFP construct was transformed into a VatM-GFP strain. VatM is the Dictyostelium homologue to the V₀ domain of H⁺ V-ATPases and is responsible for translocation of protons across various membranes. In Dictyostelium, VatM is localized strongly on the membranes of the contractile vacuolar system, and 10-fold less densely on the membranes of phagosomes and most endolysosomes (post-lysosomes lose their proton pumps) 32. Figure 3C shows representative images from three focal planes in a Z stack of a gently flattened, vegetative cell, co-labeled with AmtA-RFP and VatM-GFP. Near the bottom of the cell (0.00 μm), the characteristic contractile vacuole pattern of VatM-GFP was present, and no AmtA-RFP appeared to be colocalized with it. More internally (0.76–1.51 μm), colocalization became apparent with numerous yellow circular membranes, while some VatM-GFP localized alone. One circular organelle at the bottom of Figure 3C was labeled with only AmtA-RFP and may be a late endosome/post-lysosome that has already lost VatM-GFP.

To confirm the absence of AmtA in the contractile vacuolar system, AmtA-RFP was transformed into a strain expressing dajumin-GFP 33, which exclusively labels contractile vacuolar membranes. The images in Figure 3D are from three representative focal planes in a Z-stack showing that AmtA-RFP and dajumin-GFP were mutually exclusive from one another throughout the cell, confirming that AmtA was not localized on contractile vacuolar membranes.

Having excluded the contractile vacuolar system, we expected the colocalized membranes in the VatM-GFP/AmtA-RFP strain to be vesicles in the endolysosomal system. To confirm this, AmtA-YFP cells were incubated with TRITC-dextran for at least one hour to label the lumen of endosomes. The results in Figure 3E indicate that many of the AmtA-YFP labeled organelles were endolysosomes. To determine if AmtA also was localized to phagosomes, AmtA-YFP cells were incubated with heat-killed yeast cells that had been stained with propidium iodide (Figure 3F). The cells were prodigious feeders and seemingly every AmtA-YFP labeled membrane within the feeding cells was recruited to the phagosomes.

AmtB fused to YFP or RFP was localized on the plasma membrane (Figures 4A and 4B). While AmtB-YFP labeled numerous internal vesicle membranes, AmtB-RFP was less stable, with no recognizable internal organelle membrane localization. Even so, both AmtB fusion strains demonstrated complete recovery of ammonia export levels in our assay, suggesting that plasma membrane localization may be sufficient for that process.

The AmtB-YFP strain was used to study the intracellular localization of AmtB. Because AmtB-YFP and the strains with GFP-labeled organelles both used G418 resistance, determining if AmtB was localized on the contractile vacuole was done with the vital dye FM4–64. Pexophagy studies with yeast have used FM4–64 to delineate vacuolar membranes for 3–9 hours 34 and pulse experiments with Amoeba proteus have demonstrated the stability of FM4–64 on contractile vacuoles during contraction cycles for at least 4 hours after removal of the dye 35. In Dictyostelium, FM4–64 initially stains the plasma membrane, moves onto the contractile vacuolar membranes over the course of the next 15–20 minutes, and remains stable on the vacuolar membranes for numerous contraction cycles 3637. Figure 4C is three frames from a time series of AmtB-YFP cells stained with FM4–64, with photography commencing approximately one minute after addition of the dye. A strong colocalization signal was present almost immediately on the plasma membrane and colocalization appeared on numerous vacuoles within the cells by the end of the time series. This experiment was repeated in two independently transformed strains of Ax4 and an amtB- strain. The pattern was consistent in all of the strains, strongly suggesting that AmtB was localized on the membrane of contractile vacuoles.

To determine if AmtB was localized on endolysosome membranes, endosomes were loaded with TRITC-dextran as described for AmtA. Figure 4D clearly shows numerous TRITC filled vesicles with YFP labeled membranes, confirming the presence of AmtB within the endolysosomal system. Incubating the cells with propidium iodide stained yeast demonstrated that AmtB also was localized on phagosomes (Figure 4E).

Amt/MEP proteins characterized in other species have been determined to have 11 transmembrane helices with the amino terminus exterior to the cell and the carboxy-terminus within the cytosol 121314. To examine whether Dictyostelium ammonium transporters were oriented similarly, cells expressing the C-terminal tagged transporters were incubated with proteinase K and the cell membrane fluorescence recorded. For both the AmtA and AmtB strains, fluorescence remained on the plasma membrane after protease treatment, in agreement with the expected topology (not shown). To more directly establish topology within the intracellular membranes, endosomes of YFP strains were loaded with TRITC-dextran, concentrated after cell lysis, and treated with or without protease. As shown in Figure 5, protease treatment removed the green rings, while the vesicles remained intact as revealed by the red "balls". This confirms an orientation for the transporters with a cytosolic C-terminal YFP.

Expression of AmtC-YFP resulted in strong internal organellar membrane labeling, but visualizing the nuclei with Hoechst stain in fixed cells demonstrated that the majority of the rings surrounded nuclei (Figure 6A). This finding suggests that the protein was not successfully exporting from the ER, perhaps due to improper folding or assembly. While it is possible that AmtC normally localizes to the ER, it seemed unlikely given that the ammonia export assay showed that AmtC-YFP had export activity no higher than the amtC null strain and that AmtC-RFP had weak plasma membrane localization (Figure 6B). In addition, the ammonia export assay demonstrated that the AmtC-RFP strain had a significant increase in ammonia export relative to the null strain (Figure 2). Even so, most AmtC-RFP fluorescence was within organelles, suggesting significant misfolding or improper assembly in the ER and degradation in endolysosomes, where it colocalized with FITC-dextran (Figure 6B).

The AmtC-RFP strain was transformed with either the AmtA-YFP or the AmtB-YFP fusion construct, hoping at a minimum to obtain a colocalization signal on the plasma membrane. When AmtC-RFP and AmtA-YFP were co-expressed, colocalization was apparent on the plasma membrane, with AmtC-RFP having a consistent and stronger signal than it did when expressed alone. Additionally, AmtC-RFP fluorescence levels on the plasma membrane frequently were stronger than that of AmtA-YFP. Surprisingly, some colocalization of AmtC and AmtA on vesicle membranes also was observed (Figure 6C), although there was considerable variation between cells, as might be expected given the unavoidable variability in plasmid copy numbers of the co-transformed constructs. When the cells were incubated with live yeast, both Amts were observed on phagosome membranes (Figure 6D, arrows) in addition to the membranes of other vesicles.

In comparison to AmtC-RFP/AmtA-YFP, colocalization of AmtC-RFP with AmtB-YFP appeared complete in non-phagocytosing cells (Figure 6E). When cells were incubated with live yeast to induce phagocytosis, there was more variation, with AmtC and AmtB being present either separately or colocalized on phagosome membranes, at times within the same cell (Figure 6F).

The results of co-expression of AmtC with either AmtA or AmtB indicate a stabilization of AmtC relative to when it was expressed alone and successful targeting of AmtC to various organelles, including phagosomes. Stabilization might arise by formation of heterotrimers with the respective co-transformed transporter. However, given the variable relative expression levels during colocalization and the apparent localization of AmtC alone on some membranes during co-expression, AmtC may be stabilized by unknown means that produced independently targeted homotrimers or anomalous heterotrimer formation that resulted in AmtC monomers being targeted to internal organelles where they would not normally be found.

Strains of Dictyostelium were maintained as axenic cultures in HL5 medium 44 in flasks or petri dishes and on SM plates with Klebsiella pneumoniae as a bacterial food source 45. Dictyostelium discoideum strain Ax4 was the wild-type strain used in all experiments and was the parental strain for the amt null strains. The amt null parental strains transformed with fluorescent fusion constructs were blasticidin resistant and were: BS155 (amtA-) 24; BS167 (amtB-) (JHK, unpublished); and BS154 (amtC-) 25. The double null strains used were: BS165 (amtC-/amtA-) 24; BS169 (amtB-/amtA-) (YX, unpublished); and BS168 (amtB-/amtC-) (YX, unpublished). Parental strains with GFP-labeled organelles were G418 resistant and were: Ax2-VatM-GFP (DBS0235537) and Ax2-dajumin-GFP (DBS0236185). These were obtained from the Dictyostelium Stock Center 46. All localization experiments were done on vegetative, exponentially growing cells.

The coding region of each ammonium transporter was amplified from a cDNA clone (ddv30m16 for amtA and ddc19i14 for amtB) or reverse transcribed from mRNA (amtC) with primers containing in-frame artificial BamHI restriction sites and were ligated into the T-Easy Vector (Promega). The cDNA clones were obtained from the Dictyostelium cDNA project 4748. After sequencing to obtain clones without mutations, the coding regions were released by restriction digest with BamHI and were ligated into the BamHI sites of pDXA-mcs-YFP, pDXA-YFP-mcs 49, pDXA-GFP2 and pTX-GFP 50. Positive colonies were identified by PCR and appropriate restriction digests to check orientation. The resulting plasmids contained the amt coding regions fused to YFP or GFP at the N or C terminus, as well as a G418 resistance cassette. After introduction into Ax4 cells by electroporation 5152, only the C-terminal fusion constructs in the pDXA-mcs-YFP backbone fluoresced. This result paralleled earlier work with N- and C-terminal FLAG tag vectors, in which only the C-terminal constructs were successfully stained (JHK/CTD, unpublished), suggesting that alteration to the N-terminus prevents proper protein folding and/or membrane insertion. Additionally, only the pDXA-mcs-YFP vector contained a flexible linker to separate the fluorophore from the transporter, and this may have allowed protein maturation to progress.

Based on this information, a C-terminal fusion vector with hygromycin resistance was constructed using mRFPmars as the fluorophore, pDXA-mcs-RFPmars-Hygro. The coding region for mRFPmars 53 was amplified from 339-3: mRFPmars in pBsrH (ID 1), which was obtained from the Dictyostelium Stock Center 46, using the primers RFP-1 (atgcatCA(GGAGGATCA)₃GGA

ATG

TAA

All of the C-terminal YFP and RFP fusion constructs were transformed into Ax4 and the respective null strains by electroporation. The RFP fusion constructs additionally were transformed into the strains Ax2-VatM-GFP and Ax2-dajumin-GFP.

Cells were grown in association with bacteria that were removed by low speed centrifugation as described 45, except that sodium phosphate replaced the potassium phosphate in the standard PDF buffer because the potassium ions interfered with ammonium detection. Cell numbers were determined by direct counting. Within 60 mm Petri dishes, 6 × 10⁷ cells were plated for development on filters (Millipore HABP04700) atop cellulose pads saturated with 1.9 ml Na-PDF. Dishes were placed within a humidity chamber with overhead light during development. Because the pH of the Na-PDF buffer was 6.5, most ammonia released was retained in the pads as ammonium ions. At appropriate time points, filters were removed from the pads, and 4 ml of 0.1× ISA (0.25 M Mg₂Acetate, 0.5 M acetic acid), the buffer recommended for use with the ammonium electrode, was added. After gently rotating the dishes for 20 minutes, the pads were removed and the ammonium concentration of the buffer was determined by using an ammonium-specific electrode with a sensitivity and linear range of measurements between 0.01 to 100 mM (Orion 93-18, Thermo Electron Corporation, Beverly, MA). The number of cells and volumes used resulted in ammonium concentrations (0.1–0.5 mM) well within the linear range of the electrode. In some instances, the cells were washed from the filters using Na-PDF, pelleted, and resuspended in 2 ml 0.1× ISA. The cells were lysed with glass beads and vigorous vortexing with completion of lysis confirmed microscopically. Intracellular ammonium levels (in the range of 0.01–0.1 mM) were determined using the ammonium electrode, and the protein concentration of each sample was determined using a Coomassie Plus assay kit (Pierce, Rockford IL). Inclusion of detergent (0.5% NP-40) during lysis did not change the amount of detected intracellular ammonium.

Exponentially growing cells of the AmtA-YFP and AmtB-YFP strains were transferred from HL5 to a low fluorescence axenic medium (LoFlo) 54 and 1 × 10⁵ cells in 300 μl were placed in the well of an 8-well Lab-tek chambered cover glass (Nunc 155411) and allowed to settle for 30–60 minutes. The cover glass was placed on the microscope stage and after photographing a group of cells with obvious cell membrane localization, 100 μg of proteinase K was added to the well without disturbing the cells. The cells were observed and photographed 10 minutes later to determine if fluorescence remained on the plasma membrane.

For membrane topology determinations of organelles, exponentially growing cells (1 × 10⁸) of AmtA-YFP and AmtB-YFP were incubated with TRITC-dextran (0.4 mg/ml) in 15 ml of 50:50 HL5:LoFlo for 2 hours, pelleted by centrifugation, and washed 3× in ice-cold homogenization buffer (HB) (20 mM HEPES, 0.5 mM EGTA, 25 mM sucrose, pH 7.6). The cells were lysed by suspending in 1 ml of HB, drawing into a syringe, and passing through a stack of two 5.0 μm filters (Millipore SVLP02500) using a Swinnex Filter Holder (Millipore SX0002500). The lysate was centrifuged at 2,000 g for 5 minutes at 4°C to remove nuclei and unlysed cells, and the resulting supernatant was centrifuged at 37,000 g for 10 minutes at 4°C to pellet the remaining cellular organelles, including intact TRITC-dextran loaded endosomes. The pellet was resuspended in 100 μl of ice-cold HB and divided between two tubes, one of which contained 50 μg of proteinase K. After incubating on ice for 10 minutes, the concentrated endosomes were quickly warmed to room temperature, mixed with an equal volume of molten 4% low-melt agarose in HB and quickly compressed between two cover glasses separated by #2 bridges. Microscopic examination commenced immediately.

In order to determine which organelles the ammonium transporters were localized on, the RFP fusion constructs were transformed into Ax2-VatM-gfp, which is localized on the membranes of the contractile vacuolar system, endolysosomes and phagosomes 32, and into Ax2-dajumin-gfp, which is localized only on contractile vacuole membranes 33. Colocalization of the ammonium transporters with one another was examined by transforming a second fusion construct into an existing fluorescing strain (Ax4 or relevant null) resulting in co-transformed strains expressing: AmtA-RFP/AmtA-YFP, AmtC-RFP/AmtA-YFP, and AmtC-RFP/AmtB-YFP.

Specific testing of membrane localization in the endolysosomal system was done by loading the endosomes with fluorescent dextrans 555657. FITC-dextran (70 kDa; Sigma FD70S) was used in the RFP strains and TRITC-dextran (65–76 kDa; Sigma T1162) in the YFP strains. Briefly, 1 × 10⁷ cells were washed three times in DB (5 mM Na₂HPO₄, 5 mM KH₂PO₄, 1 mM CaCl₂, 2 mM MgCl₂, pH 6.5.) and resuspended in 1 ml of DB containing 1–2 mg/ml of the fluorescent dextran in a 2 ml microcentrifuge tube. The cells were incubated with rocking in the dark for 1–2 hours. An equal volume of 0.4% Trypan Blue (Sigma T8154) was added to quench any unincorporated fluorescence, and the cells were washed four times in DB prior to depositing 1 × 10⁵ cells in 300 μl into a well of a chambered cover glass for photography.

FM 4–64 (Molecular Probes) was used to test contractile vacuolar colocalization in the AmtB-YFP strain 36. Cells were allowed to attach in the well of a chambered cover glass for 30 or more minutes in 250 μl of DB. After placement on the microscope stage, 250 μl of the steryl dye FM4–64 (2 μg/ml) was added to the well, resulting in a 1 μg/ml final concentration. Time-lapse photography commenced almost immediately, with photographs taken every 2 minutes for 10 cycles.

Phagocytosis experiments to test phagosome colocalization were done with YFP strains using heat killed yeast (80°C for 5–10 minutes) that were subsequently non-specifically stained with 5 mM propidium iodide in PBS for 15 minutes 58. The stained yeast were washed three times in PBS, resuspended in DB, and stored frozen at -20°C. Dictyostelium cells (2 × 10⁶) were suspended in 1 ml DB with 1.2 × 10⁷ thawed yeast cells 59 and were rocked in the dark for 1 hour. Cells (2 × 10⁵) and yeast were allowed to settle on cover slips for a minimum of 30 minutes and were gently flattened with agar overlays 60 immediately prior to photographing. Similar experiments were carried out with strains expressing two Amts using live yeast.

All images except Figures 4B and 6A and 6B were photographed on a Zeiss LSM510 inverted confocal microscope using a Plan-Apochromatic 63×/1.4 oil DIC or a Plan-Neofluar 100×/1.3 oil DIC lens. Excitation of GFP, YFP and FITC was with the 488 band of the argon laser using a 515–530 or 515–550 nm emission filter. The same 488 nm argon laser was used for FM4–64 but using a 650 nm long-pass emission filter. RFP, TRITC and propidium iodide excitation was with the HeNe1 543 nm laser using a 560 nm long-pass emission filter. Figures 4B and 6A and 6B were photographed on a Nikon Diaphot-TMD inverted compound microscope with an Olympus 100× oil immersion lens using a Q-Imaging Retiga 1300 CCD camera with FITC (Chroma 41001 (Ex. 460–500/BP510–560)), rhodamine (Nikon DM580 (Ex. 510–560/LP590)) and DAPI (Nikon UV2A (Ex. 330–380/LP420)) filters. Confocal images used in the figures were exported from the Zeiss LSM Image Browser, imported into Simple PCI (Hamamatsu Corp.) to assemble exportable panels, and lastly imported into Microsoft Word for figure assembly. Minimal to no alterations were made of the original images beyond cropping as necessary.