ABSTRACT
The precise subcellular localization of the components of the cyclic AMP (cAMP) signaling pathways is a crucial aspect of eukaryotic intracellular signaling. In the human pathogen Trypanosoma brucei, the strict control of cAMP levels by cAMP-specific phosphodiesterases is essential for parasite survival, both in cell culture and in the infected host. Among the five cyclic nucleotide phosphodiesterases identified in this organism, two closely related isoenzymes, T. brucei PDEB1 (TbrPDEB1) (PDEB1) and TbrPDEB2 (PDEB2) are predominantly responsible for the maintenance of cAMP levels. Despite their close sequence similarity, they are distinctly localized in the cell. PDEB1 is mostly located in the flagellum, where it forms an integral part of the flagellar skeleton. PDEB2 is mainly located in the cell body, and only a minor part of the protein localizes to the flagellum. The current study, using transfection of procyclic trypanosomes with green fluorescent protein (GFP) reporters, demonstrates that the N termini of the two enzymes are essential for determining their final subcellular localization. The first 70 amino acids of PDEB1 are sufficient to specifically direct a GFP reporter to the flagellum and to lead to its detergent-resistant integration into the flagellar skeleton. In contrast, the analogous region of PDEB2 causes the GFP reporter to reside predominantly in the cell body. Mutagenesis of selected residues in the N-terminal region of PDEB2 demonstrated that single amino acid changes are sufficient to redirect the reporter from a cell body location to stable integration into the flagellar skeleton.
In eukaryotes, the ubiquitous second messenger cyclic AMP (cAMP) is generated from ATP by membrane-integral or by cytoplasmic, CO2-regulated cyclases (35, 44). The cAMP signal is processed by a small group of receiver proteins, including the regulatory subunit of protein kinase A (28), cAMP-gated ion channels (4), and the guanine-nucleotide-exchange proteins EPAC1 and EPAC2 (39). The cAMP signal is terminated by the action of a family of cyclic nucleotide-specific phosphodiesterases (PDEs) (9). This paradigm is rather straightforward, involves a limited number of players, and is generally well understood, at least in mammalian cells. However, much less is known about how individual cAMP signals are temporally and spatially controlled. Since most eukaryotic adenylyl cyclases are integral membrane proteins, often restricted to specific membrane subdomains (10), cAMP signaling is usually initiated at the cell membrane (40). However, diffusion of cAMP away from its site of generation is rapid, with diffusion coefficients being about 400 μm2/s (8, 15, 29), translating into diffusion velocities of 30 to 40 μm/s. As a consequence, the signal would reach the center of the cell with a diameter of 3 μm within less than 50 ms and would rapidly saturate the entire cell. While regulation through fluctuating cellular levels of cAMP represents a valid paradigm of cAMP signaling, it has become clear that other, more localized modes of cAMP signaling must also exist. Several groups have shown that the cAMP response of a given cell can differ depending on what set of receptors activates the cyclase response (14, 30, 41, 42). Similarly, the cAMP response of endothelial cells depends on the subcellular site where the cAMP is produced. They tighten their barrier function when cAMP is produced by membrane-bound adenylyl cyclases but become more permeable when cAMP is produced in the cytoplasm (17, 45). The distinct subcellular localization of cAMP signals was experimentally demonstrated using an array of techniques (29, 40, 55, 56).
Physically tethered PDEs might serve to confine newly synthesized cAMP to defined microdomains. Only cAMP-binding proteins that are localized within or extend into such microdomains would be able to receive the cAMP signal (17, 49). cAMP concentrations within such domains might rise and fall rapidly, reaching peak concentrations much more rapidly and locally far beyond the steady-state cAMP levels measured in whole-cell extracts. Such spatially organized, tethered PDEs can generate local sinks into which cAMP disappears (1, 23). This paradigm would allow the simultaneous presence of numerous local cAMP concentration gradients within a single cell, allowing great flexibility in signal generation and intracellular signal transmission. This concept is based on the distinct subcellular localization and physical association of PDEs with subcellular structures and on the existence of localized subcellular cAMP pools, for which there is extensive experimental support (3, 5, 13, 50, 52). Interestingly, PDEs localized in different subcellular regions may still be able to compensate for each other. Ablation of the cilium-specific PDE1C from the olfactory neurons in the mouse did not prolong response termination, as long as the cytoplasmic PDE4 in the cell body was still present (11).
The unicellular eukaryote Trypanosoma brucei is the causative agent of human sleeping sickness in sub-Saharan Africa. It belongs to the large order of the kinetoplastida, which includes many medically and economically important pathogens of humans, their livestock, and their crops worldwide (27). Trypanosomes are very small cells (about 15 by 3 μm in diameter) that carry a single flagellum (10 by 0.5 μm). The volume of a procyclic trypanosome of strain 427 is (9.6 ± 0.8) × 10−14 liter (Markus Engstler, personal communication), with the flagellum representing about 15% of this. A signaling threshold concentration of 1 μM cAMP corresponds to just about 30,000 molecules of cAMP per cell. Given a diffusion coefficient of 400 μm2/s (29), unrestricted diffusion of cAMP would swamp the cell within 50 ms. Obviously, temporal and spatial control of cAMP signaling is crucial for T. brucei. Strategically located, physically tethered PDEs might thus play an important role in the architecture of the cAMP signaling pathways in T. brucei.
The genomes of T. brucei and of other kinetoplastids, such as T. vivax, T. cruzi, Leishmania major, L. infantum, and L. braziliensis, all code for the same set of five cyclic nucleotide-specific PDEs (25, 53). In T. brucei, the genes for T. brucei PDEB1 (TbrPDEB1; subsequently termed PDEB1) and TbrPDEB2 (PDEB2) are tandemly arranged on chromosome 9 and code for two very similar cAMP-specific PDEs, each with two GAF (mammalian cyclic GMP-dependent PDEs, Anabaena adenylyl cyclases, Escherichia coliFhlA) domains (21) in their N-terminal regions (38, 57). These two PDEs were also studied experimentally in T. cruzi (12) and L. major (24, 52), and orthologues are present in all kinetoplastid genomes available so far. Despite their high overall sequence similarity, PDEB1 and PDEB2 exhibit distinct subcellular localizations (31). PDEB1 is predominantly found in the flagellum, where it is stably associated with cytoskeletal components that are resistant to detergent extraction. In contrast, PDEB2 is mostly localized in the cell body, from where it is fully extractable by nonionic detergents. However, a minor fraction of PDEB2 also associates with the flagellar skeleton in a Triton-resistant manner, most likely through interaction with PDEB1. Earlier work has shown that both PDEB1and PDEB2 are essential enzymes in bloodstream-form T. brucei (31), while TbPDEA, TbPDEC, and TbPDED play minor roles (20; S. Kunz, unpublished data).
MATERIALS AND METHODS
Sequences analysis.Sequences for PDEB1 and PDEB2 of T. brucei (GeneDB nomenclature Tb09.160.3590 and Tb09.160.3630, respectively), T. vivax (tviv693g06.q1k_8 and tviv693g06.q1k_12, respectively), T. cruzi (Tc00.1047053508277.100 and Tc00.1047053508277.110, respectively), L. infantum (LinJ15_V3.1550 and LinJ15_V3.1540, respectively), L. braziliensis (LbrM15.1080 and LbrM15.0880, respectively), and L. major (NCBI accession numbers AAR88146 and AAR88145 , respectively) were aligned by use of the Clustal X program. GAF-A domains correspond to amino acids 234 to 383 of PDEB1, while GAF-B domains correspond to amino acids 407 to 554. The catalytic domains were defined by the crystal structure of LmjPDEB1 (52) and correspond to amino acids 586 to 908 of PDEB1. From the individual sequence sets, average distance trees were constructed in the Jalview program and unrooted trees were displayed using the NJPlot program.
Trypanosome culture.Procyclic T. brucei (stock 427) was cultured in SDM-79 medium containing 5% fetal calf serum (FCS) (7). Bloodstream-form trypanosomes were cultivated in HMI-9 medium (22) supplemented with 10% FCS.
In situ tagging of PDEB1 in bloodstream forms.For C-terminal in situ tagging (32), the last 713 bp of the open reading frame of PDEB1 (without the stop codon; bp 2074 to 2787) and the first 741 bp of its 3′ untranslated region (UTR) were PCR amplified from genomic DNA using an Expand high-fidelity system (Roche Applied Sciences, Rotkreuz, Switzerland) and the following primers (Xho, BamHI, and XbaI sites are underlined): 2C-com-pMOTag-f (ATCTCGAGGAGAAGTTAACCGAGCTTGAG), 2C-com-pMOTag-r (ATCTCGAGACGAGTACTGCTGTTGTTGC), 2C-UTR-pMOTag-f (ATGGATCCGTGAAATTGAAGAAGTCAGTTGA), and 2C-UTR-pMOTag-r (ATTCTAGAATTTACCTGTTGCGGTTCCCA).
The PCR products were finally ligated into the vector pMOTag4YH (YFP-HA Tag [32]). The final construct was digested with NotI and ApaI, ethanol precipitated, and transfected, as described previously (31, 54). Transfectants were selected with 1 μg/ml hygromycin. The correct integration of the tagging cassette was verified by Southern and Western blotting analyses.
In situ tagging of PDEB2.The construct for C-terminal in situ tagging of PDEB2 with a 3× c-Myc tag was described previously (31). The construct was digested with KpnI and NotI, ethanol precipitated, and transfected into 221 (NYSM) bloodstream-form cell lines (54). Transfectants were selected with 1 μg/ml hygromycin, and the correct integration of the tagging cassette was verified by Southern and Western blotting analyses. Since in strain 427, used in these experiments, one allele of PDEB2 has undergone gene conversion with PDEB1, which results in the replacement of its GAF-A domain (26), clones that carried the tag in the unaltered allele were selected, using PCR screening with a diagnostic primer pair (26).
Truncation constructs.Initially, truncated versions of PDEB1 and PDEB2 carrying a green fluorescent protein (GFP) tag were produced by our in situ tagging methodology, using the pMOTag vector series (31). The resulting fusion proteins were present at very low levels only and were barely detectable by immunoblotting and were present in insufficient amounts for fluorescence microscopy. Ectopic expression from the vector pGaprone-GFPΔLIIβ (18) was therefore chosen as an alternative. This vector mediates the expression of the ectopic gene by a procyclin promoter and a procyclin 3′ UTR containing the LII deletion to maximize RNA stability (18). It allows both transient expression and stable integration into one of the procyclin loci, and it enables the necessary strong expression level. Truncated versions of PDEB1 fused to GFP were generated by PCR (Expand high-fidelity PCR system; Roche) using the common forward primer 2C-N-GFP-f (5′-ATCTCGAGATGTTCATGAACAAGCCCTTTG-3′) (the XhoI site is underlined and the start codon is in boldface) and the following specific reverse primers (the AgeI, SalI, and XhoI sites are underlined): B1-70rev (5′-CGACCGGTAATCCGCTTTGATCGAGAACCT-3′; bp 1 to 210 encoding amino acids 1 to 70), B1 1-114GFP-r (5′-ATGTCGACGGCACCCGGTACACTTATC-3′; bp 1 to 342 encoding amino acids 1 to 114), B1 1-212GFP-r (5′-CGGTCGACAGTTGATTCTCGCTTCCTC-3′; bp 1 to 637 encoding amino acids 1 to 212), and B1endGAFGaprone (5′-ATACCGGTCGGCATTGCAGAATGAAGTT-3′; bp 1 to 1980 encoding amino acids 1 to 660).
For constructing the B1(Δ1–114)::GFP and B1(Δ1–70)::GFP cell lines, the B1endGAFGaprone primer was used together with the two specific forward primers B1Δ114 (5′-TACTCGAGATGCCGGGGATTCGTACATA-3′; bp 343 to 1980 encoding amino acids 115 to 660) and B1Δ70 (5′-TACTCGAGATGACTCCAACAAGCAATGCAACAC-3′; bp 211 to 1980 encoding amino acids 71 to 660). The XhoI sites are underlined, and the newly introduced start codons are in boldface.
N-terminally shortened versions of PDEB2 were generated by PCR using the common forward primer 2B-N-GFP-f (5′-TACTCGAGATGACACACAACGGTGGTC-3′) and the following specific reverse primers (the AgeI and SalI sites are underlined): B2-70rew (5′-ATACCGGTGTAACGGGAGGCTGGATCA-3′; bp 1 to 210 encoding amino acids 1 to 70) and B2 1-212GFP-r (5′-CGACCGGTTTCAGTTTCTCTGCCTGAATG-3′; bp 1 to 637 encoding amino acids 1 to 212). The PCR products were cloned into the pCR2.1-TOPO vector, sequenced, and cloned into the matching sites of the pGaproneΔLII GFP vector (19). Linearization of the finished plasmids was done with MamI. The linearized vectors were ethanol precipitated, and 10 μg final DNA was transfected into mid-log-phase procyclic trypanosomes (∼5 × 106 cells/ml) with a Bio RTX electroporator (1.4 kV, 25 Ω). Transfectants were selected with 15 to 25 μg/ml G418 for 10 to 14 days.
Site-directed mutagenesis.Site-directed mutagenesis was performed on the open reading frames of nucleotides 1 to 210 of PDEB1 (mutants B1_M1 to B1_M5) and PDEB2 (mutants B2_M6 and B2_M7) in the pCR2.1-TOPO vector using Pfu DNA polymerase (Promega, Duebendorf, Switzerland), followed by digestion of the template with DpnI.
Primer sequences for each of the mutations in B1(1–70)::GFP were as follows (mutated nucleotides are in boldface, and numbering refers to the open reading frame of PDEB1).
Introduction of the T28P/E29P double mutation in mutant B1_M1 was performed in two steps. To first introduce T28P, the following primer pair was used: forward primer 5′-G70CGTTTGCCATCCCTGAAGCAATCCTCGCT99-3′ and reverse primer 5′-A99GCGAGGATTCTTCAGGGATGGCAAACGC70-3′. To introduce E29P in a second step, the primer pair was as follows: forward primer 5′-A68GGCGTTTGCCATCCCTCCAGCAATCCTCGCT99-3′ and reverse primer 5′-A99GCGAGGATTGCTGGAGGGATGGCAAACGCCT68-3′.
For mutant B1_M2 (A51P/A52P), forward primer 5′-A141AGTGGACTGCCACCCCTTATCAAACG167-3′ and reverse primer 5′-C167GTTTGATAAGGGGTGGCAGTCCACTT141-3′ were used; for mutant B1_M3 (P58A), forward primer 5′-G154CCCTTATCAAACGTATTGCTTATGATATCCTTGTTGAGG-3′ and reverse primer 5′-C193CTCAACAAGGATATCATAAGCAATACGTTTGATAAGGGC154-3′ were used; for mutant B1_M4 (T43A), forward primer 5′-A115AACGCAGCTTTGCGTCCTCCGAAAAAAG143-3′ and reverse primer 5′- C143TTTTTTCGGAGGACGCAAAGCTGCGTTT115-3′ were used; and for mutant B1_M5 (C22L), forward primer 5′-C46ACGAGTCGGAGCACCTTCTTGAGGCGTTTGCCATC81-3′ and reverse primer 5′-G81ATGGCAAACGCCTCAAGAAGGTGCTCCGACTCGTG46-3′ were used.
The primer sequences for the two mutations in B2(1-70)::GFP were as follows (numbering refers to the open reading frame of PDEB2): for mutant B2_M6 (L10C), forward primer 5′-T21GGTCGTCATCTGTGTGAGGCGGTTACGCT50-3′ and reverse primer 5′-A50GCGTAACCGCCTCACACAGATGACGACCA21-3′, and for mutant B2_M7 (Y45P), forward primer 5′-G126TTATTCGAGAAGCCTCAAGATATCCTCGTGG157-3′ and reverse primer 5′-C157CACGAGGATATCTTGAGGCTTCTCGAATAAC126-3′.
All mutations were verified by DNA sequencing. The fragments were isolated by an XhoI/AgeI double digestion and were subcloned into the pGaproneΔLII vector (18). Ten micrograms of circular DNA was transiently transfected into 1.2 × 108 procyclic trypanosomes. Transfection efficiencies varied between 5 and 20%. Cells were analyzed 24 h after transfection.
Triton X-100 fractionation.Trypanosomes (8 × 107) were washed once in phosphate-buffered saline (PBS) and then lysed on ice or at 37°C for 10 min in HME buffer (100 mM HEPES, pH 6.9, 1 mM MgSO4, 1 mM EGTA [47]) containing 0.5% to 1% Triton X-100 and supplemented with protease inhibitor (complete mini, EDTA free; Roche). The extracts were centrifuged at 13,000 rpm for 10 min at 4°C. Supernatants and pellets were analyzed by Western blotting. Gels were transferred to Immobilon-P membranes and probed with a mouse anti-GFP antibody (diluted 1:1,000; Boehringer Mannheim). Control antibodies were polyclonal rabbit anti-BiP (endoplasmic reticulum staining, as a marker for Triton soluble proteins, diluted 1:50,000; a gift of Jay Bangs, University of Wisconsin, Madison, WI [2]) and polyclonal rat anti-paraflagellar rod (anti-PFR; as a marker for Triton-insoluble flagellar proteins, diluted 1:30,000 [46]).
Fluorescence microscopy.Transfected clones were washed in PBS once and resuspended in cold PBS at a density of 5 × 106 cells/ml. All following steps were performed in a moist chamber at room temperature. The cells were allowed to adhere to poly-lysine-coated 21-well slides (Erie Scientific Company) for 20 min. After fixation with cold methanol for 5 min or with 4% paraformaldehyde for 10 min, they were permeabilized with Triton X-100 (1%) for 1 min.
For the preparation of cytoskeletons, cells were extracted once with cold HME buffer or MME buffer (100 mM morpholineethanesulfonic acid, pH 6.9, 1 mM MgSO4, 1 mM EGTA [45]) containing 0.5% Triton X-100 for 5 min and were then fixed with 4% paraformaldehyde.
Slides were rehydrated in PBS for 5 min before they were blocked with PBS–2% (wt/vol) bovine serum albumin for 1 h or directly mounted with Vectashield mounting medium for direct GFP detection. Primary antibodies were a polyclonal rat anti-PFR antibody (dilution, 1:1,000 [46]) for detection of PFR and monoclonal rat antibody Y1/2, which is specific for tyrosinolated tubulin (48) for detection of the basal bodies. After three washing steps in PBS for 5 min each time, the cells were stained for 1 h with the secondary antibody (Alexa Fluor goat anti-rat 595; dilution, 1:750; Molecular Probes, Invitrogen, Basel, Switzerland) in PBS–2% (wt/vol) bovine serum albumin. Slides were washed in PBS three times for 5 min each time, coverslips were mounted with Vectashield mounting medium containing 4′,6-diamidino-2-phenylindole (DAPI; Vector Laboratories, Geneva, Switzerland), and the slides were analyzed with a Leica DM6000B fluorescence microscope.
Coimmunoprecipitation.The cell lines used were described previously (31). In line 2C-A5, one allele of PDEB1 is tagged with 3× hemagglutinin (HA), and line 2C-A5(2B4M)1 contains one allele of PDEB1 tagged with 3× HA and one allele of PDEB2 tagged with 3× c-Myc. Trypanosomes (3.3 × 108 cells) were washed once in ice-cold PBS and then solubilized in 0.5 ml RIPA buffer (50 mM Tris HCl, pH 8.8, 150 mM NaCl, 1% NP-40, 0.5% deoxycholate, 0.1% SDS, Roche Complete mini, EDTA free, as a protease inhibitor). Lysates were incubated on ice for 15 min with occasional vortexing and were finally centrifuged at 13,000 rpm for 15 min. The supernatant was incubated with 60 μl of anti-c-Myc agarose conjugate (Sigma, Buchs, Switzerland) or 60 μl of anti-HA affinity matrix (Roche) for 2 h at 4°C. Beads were washed three times in 1:4-diluted RIPA buffer before elution of the bound material by boiling in 60 μl of 2× Laemmli sample buffer and processing for SDS-PAGE (10% gel) and Western blotting. After gel electrophoresis, the proteins (5 × 106 cell equivalents) were transferred onto Immobilon-P membranes and immunostained with rat anti-HA 3F10 antibody (diluted 1:1,000; Roche), mouse anti-c-Myc 9E10 antibody (diluted 1:1,000; Santa Cruz), or anti-PFR antiserum.
Sucrose gradient sedimentation.2CA5(2B4M)1 procyclic trypanosomes (PDEB1 tagged with 3× HA, PDEB2 tagged with 3× c-Myc) were lysed in ice-cold RIPA buffer supplemented with complete protease inhibitor cocktail (Roche), as described above.
Cell lysates (0.5 ml) were overlaid onto a 5 to 20% sucrose gradient (12-ml gradient volume; sucrose dissolved in 50 mM Tris HCl, 150 mM NaCl, pH 8.8). The sucrose gradients were centrifuged at 36,000 rpm for 18 to 24 h at 4°C. After the run, 0.5-ml fractions were collected and their sucrose contents were determined by refractometry. Each fraction was analyzed by Western blotting for the presence of the tagged PDEB1 and PDEB2 using standard procedures and antibody concentrations, as described above. As markers for the sucrose gradient sedimentation, 0.5 ml of bovine serum albumin, catalase, aldolase, and ferrodoxin was loaded individually onto parallel sucrose gradients. The individual fractions of the marker gradients were analyzed for their protein content using the Bradford protein assay (Bio-Rad).
RESULTS
PDEB1 and PDEB2 can functionally complement each other.Earlier work with procyclic T. brucei had shown that PDEB1 and PDEB2 exhibit distinct subcellular localizations. PDEB1 was predominantly located in the flagellum, where it remained tightly associated to a skeletal component, most likely the PFR structure (31). In contrast, PDEB2 is mostly localized in the cell body and was extractable by nonionic detergent. A minor part of PDEB2 also associates with the flagellum and is not extractable by nonionic detergent. To determine if this distinct subcellular localization is a general phenomenon in T. brucei or if it is restricted to the previously analyzed procyclic forms, the subcellular localization of PDEB1 and PDEB2 was examined in bloodstream forms, using in situ tagging (Fig. 1A). A distinct localization of the two enzymes was again observed, very reminiscent of what had earlier been found in procyclic forms. PDEB1 is stably associated with the flagellar skeleton, whereas PDEB2 is mostly located in the cell body, while a minor fraction colocalizes with PDEB1 in the flagellum. These findings demonstrate that the specific subcellular locations of the two PDEs are very similar between bloodstream and procyclic forms. They also confirm that a minor fraction of PDEB2 is reproducibly found in the flagellum, where it is resistant to extraction with nonionic detergent. Nevertheless, the proteins can be solubilized from the flagellar skeleton with RIPA buffer (see Materials and Methods), which contains 0.1% sodium dodecyl sulfate. Sucrose gradient sedimentation of RIPA buffer-solubilized proteins (Fig. 1B) demonstrated that PDEB1-3× HA and PDEB2-3× c-Myc cosediment and are found in the same fractions as the marker protein catalase (232 kDa). As the monomeric molecular masses of PDEB1-3× HA and PDEB2-3× c-Myc are 107 kDa (103.6 kDa for PDEB1 plus 3.4 kDa for 3× HA) and 108.6 kDa (103.2 for PDEB2 plus 5.4 kDa for 3× c-Myc), respectively, these data suggested that the two proteins might form heteromers that allow the transport of PDEB2 to the flagellum and its integration into the skeletal structure via a signal located on PDEB1. To confirm the existence of heteromers, coimmunoprecipitation was done with lysates of strain 2C-A5(2B4M)1, which expresses HA-tagged PDEB1 and 3× c-Myc-tagged PDEB2. These experiments demonstrated that PDEB1 and PDEB2 do coimmunoprecipitate, despite the presence of 0.1% SDS in the extraction buffer (Fig. 1C). These results were further confirmed by mass spectrometry analysis of the immunoprecipitated proteins (data not shown). In conjunction, these data indicate that PDEB1 and PDEB2 can form heteromers that are sufficiently stable to survive the harsh conditions of the RIPA buffer used to solubilize the cells. They suggest that the limited amount of PDEB2 that is found in the flagellar skeleton might be due to stable PDEB1/PDEB2 heteromers that are targeted via a putative signal located on PDEB1.
(A) Localization of PDEB1 and PDEB2 in bloodstream forms. Bar, 5 μm. (B) Sucrose gradient sedimentation. Fractions were analyzed by Western blotting for PDE1 (HA tag) and PDE2 (c-Myc tag). Sedimentation markers are ferredoxin (450 kDa), catalase (232 kDa), aldolase (158 kDa), and hemoglobin (64 kDa). Arrow, direction of sedimentation. (C) Coimmunoprecipitation of HA-tagged PDEB1 and 3× c-Myc-tagged PDEB2. (Left panel) Lane 1, cell line 2C-A5 (PDEB1, HA tagged) immunoprecipitated with anti-HA beads; lane 2, cell line 2C-A5 (PDEB1, HA tagged) immunoprecipitated with anti-c-Myc beads; lane 3, cell line 2C-A5(2B4M)1 immunoprecipitated with anti-HA beads; lane 4, cell line 2C-A5(2B4M)1 immunoprecipitated with anti-c-Myc beads. Immunoprecipitates were analyzed by immunoblotting with anti-HA antibody (α-HA). (Right panel) As described for the left panel, but the immunoprecipitates were revealed with anti-c-Myc antibody (α-cMyc).
Despite this clearly distinct subcellular localization of the two enzymes, the homozygous deletion of either PDEB1 or PDEB2 does not produce a discernible phenotype in cultured procyclic or bloodstream forms (M. Baresic and A. Schmid, unpublished data). Nevertheless, their combined deletion is lethal for both procyclic forms (A. Schmid, unpublished) and bloodstream forms (57), as is a combined RNA interference against the two bloodstream forms (31). These results suggest that PDEB1 can fully complement PDEB2, and vice versa, despite their different intracellular localizations. This location-independent complementation between two PDEs with distinct subcellular localizations is very reminiscent of what has recently been described for PDE1C and PDE4A in murine olfactory neurons (11).
Identification of localization signal(s) in PDEB1.Amino acid sequence alignments between PDEB1 and PDEB2 demonstrated a strikingly uneven pattern of sequence conservation. The N-terminal 212 amino acids exhibit only 25% sequence identity between the two enzymes. In contrast, the subsequent regions containing the GAF-A and GAF-B domains are 96% identical, and the two catalytic domains are 88% identical. Multiple-sequence alignments were performed with the four regions (the N termini, the GAF-A and the GAF-B domains, and the catalytic domains) of the PDEB1 and PDEB2 orthologues of six kinetoplastid genomes (T. brucei, T. vivax, T. cruzi, L. major, L. infantum, and L. brasiliensis). The GAF-A, GAF-B, and catalytic domains all cluster strictly according to species. In contrast, the N-terminal 70 amino acids cluster according to the isoenzyme type; i.e., the N termini of the B1 homologues form one cluster, while those of the B2 enzymes form another (Fig. 2). The same pattern was observed when the first 212 amino acids instead of the first 70 amino acids were analyzed (data not shown). In conjunction, these data demonstrate that the N-terminal region is the distinctive feature between the two enzymes, and they suggest that this might be involved in the distinct subcellular targeting of PDEB1 and PDEB2.
Clustering of various domains of PDEB1 and PDEB2. The N-terminal domains (amino acids [aa] 1 to 70) of all PDEB1 and PDEB2 orthologues cluster according to isoenzyme type. The other three domains (the GAF-A, GAF-B, and catalytic [cat] domain) cluster according to species. For domain definitions, see Materials and Methods. Dark boxes, B1 members; white boxes, B2 members; Tb, Trypanosoma brucei; Tv, T. vivax; Tc, T. cruzi; Lm, Leishmania major; Li, L. infantum; Lb, L. braziliensis.
To experimentally explore this hypothesis, the N-terminal 212 amino acids of either PDEB1 and PDEB2 were fused to a GFP reporter, creating constructs (B1(1–212)::GFP and B2(1–212)::GFP, respectively; Fig. 3). The constructs were integrated into the procyclin locus, and stably transfected lines were generated. All cell lines in this and subsequent experiments were analyzed by the use of three complementary criteria: fluorescence of whole cells, fluorescence of Triton X-100-extracted cytoskeletons, and immunoblotting of supernatants (containing soluble and membrane proteins) and pellets (detergent-insoluble proteins) of Triton X-100-extracted cells. The data given in Fig. 4A demonstrate that the N-terminal 212 amino acids of PDEB1 are sufficient to locate the B1(1–212)::GFP reporter to the flagellum and to stably integrate it into the flagellar skeleton. In contrast, B2(1–212)::GFP accumulates in the cell body (Fig. 4B, row 1). The B2(1–212)::GFP reporter is almost quantitatively solubilized by Triton, with only a few spots in the cell body remaining (Fig. 4B, row 2). This differential localization was further corroborated by the analysis of immunoblots from such detergent extracts (Fig. 4B, row 3). Control antibodies for characterizing the Triton-soluble and -insoluble fractions were anti-BiP (3) and anti-PFR (46), respectively. In conjunction, the data given in Fig. 4A and B indicate that the N-terminal 212 amino acids of PDEB1 contain a signal(s) that mediates the integration of the reporter into the flagellar skeleton in a form that renders it resistant to extraction by Triton X-100. Importantly, they also demonstrate that neither the GAF domains nor other parts of PDEB1 are required for flagellar targeting and integration.
Schematic representation of constructs. FL, full-length PDEB1 or PDEB2; open circle, C-terminal c-Myc or HA tag; B1(1–212)::GFP, B1(1–114)::GFP, and B1(1–70)::GFP, GFP fusions carrying the N-terminal 212, 114, and 70 amino acids of PDEB1, respectively; B2(1–212)::GFP, GFP fusion with the N-terminal 212 amino acids of PDEB2; B1(Δ1–114)::GFP and B1(Δ1–70)::GFP, derivatives of PDEB1 where the catalytic domain was replaced by GFP and the N-terminal 114 or 70 amino acids was deleted, respectively; rectangles, GAF domains A and B and catalytic domain; oval symbols, GFP protein. The scale at the bottom of the figure represents the amino acid numbering of PDEB1.
The N terminus of PDEB1, but not that of PDEB2, confers integration into the flagellar skeleton. (A) Amino acids 1 to 212 of PDEB1 lead to stable integration of the GFP reporter into the flagellar skeleton; (B) amino acids 1 to 212 of PDEB2 do not; (C) amino acids 1 to 114 are sufficient for flagellar integration of the reporter; (D) a reporter that contains the entire N terminus of PDEB1 but that lacks the first 114 amino acids is not integrated into the flagellar skeleton; (E) amino acids 1 to 70 of PDEB1 are sufficient for stable integration of the reporter into the flagellar skeleton; (F) a reporter containing the entire N terminus of PDEB1 but lacking the first 70 amino acids is not integrated into the flagellar skeleton. For a detailed description of the reporter constructs, see the legend to Fig. 3. Row 1, whole cells; row 2, cytoskeletons; green, fluorescence of GFP reporter; red, anti-PFR antibody; yellow, regions of overlap of GFP and PFR staining; blue, DAPI staining of nuclei and kinetoplasts; bar, 10 μm; row 3, Western blot analysis of Triton X-100-fractionated cells; S, detergent-soluble fraction; P, detergent-insoluble fraction; GFP, reporter; PFR, marker for detergent-insoluble proteins; BiP, marker for detergent-soluble proteins.
The N-terminal region of PDEB1 was then further truncated to 114 and 70 amino acids [B1(1–114)::GFP and B1(1–70)::GFP, respectively; Fig. 3], and the same series of experiments was performed. The data given in Fig. 4C and E demonstrate that these truncated sequences are still sufficient to mediate the Triton-resistant integration of the reporter into the flagellar skeleton. In addition, Triton X-100 fractionations done at 37°C demonstrated that the B1(1–70)::GFP reporter remained detergent insoluble even at this elevated temperature (data not shown). This observation indicates that the reporter is, in fact, integrated into the flagellar skeleton and is not merely associated with Triton-insoluble lipid rafts of the flagellar membrane (51). This established that the N-terminal 70 amino acids of PDEB1 contain all the information required for the transport to and the faithful integration into the flagellar skeleton. To corroborate this finding, two additional constructs were used that correspond to full-size PDEB1 in which the catalytic domain was replaced by GFP and full-size PDEB1 from which the 114 or 70 N-terminal amino acids were missing (constructs B1Δ114::GFP and B1Δ70::GFP, respectively; Fig. 3). Both of these remained in the cell body, mostly in a Triton-extractable form (Fig. 4D, row 3, and Fig. 4F, row 3). This further confirms that the very N-terminal amino acid sequence is crucial for flagellar localization and Triton-resistant integration of PDEB1 and that the GAF domains are not required for flagellar targeting or integration.
Delimitation of the signal in amino acids 1 to 70 of PDEB1.Analysis of the first 70 amino acids of all kinetoplastid homologues of PDEB1 and PDEB2 revealed that these stretches are clustering according to their belonging to the B1 or the B2 type of PDEs rather than according to the species from which they originate (Fig. 2). In these N-terminal regions of both the B1 group and the B2 group, two or three helical regions are predicted (Fig. 5). Though several amino acids are absolutely conserved in all sequences analyzed, some differences between the two groups were discerned. To further delimit the localization/integration signal in PDEB1, selected amino acids were mutated, as detailed in Table 1 and Fig. 5.
Positions of mutations in the sequence alignment of the N-terminal 70 amino acids of PDEB1 and PDEB2. Alignment of N termini of PDEB1s (A) and of PDEB2s (B) of different kinetoplastid species. Gray boxes, predicted helices (http://zhanglab.ccmb.med.umich.edu/I-TASSER [43]); black bar in panel A, a predicted conserved casein II kinase phosphorylation site. The extent of sequence conservation is indicated below the alignments. (C) Definition of mutants. Gray boxes, predicted helices after introduction of the mutations. Mutations are given in the circles. Numbering is according to the PDEB1 sequence. For the organism abbreviations, see the legend to Fig. 2. wt, wild type. Clustal consensus symbols: asterisk, identical; colon, conserved; period, semiconserved.
Mutations introduced into the N termini of PDEB1 and PDEB2
The appropriate constructs were transiently transfected into procyclic trypanosomes, and their subcellular localization in whole cells and in Triton-extracted cytoskeletons was analyzed 18 to 24 h after transfection (Fig. 6). In whole cells, the wild-type construct [B1(1-70)::GFP; see Fig. 5 for its predicted helical structure] localizes along the flagellum. An accumulation of fluorescence is also seen around the flagellar base (Fig. 6A). In detergent-extracted cytoskeletons, the construct remains tightly associated with the flagellum. The accumulation of fluorescence around the flagellar base is no longer detectable, indicating that the latter material is detergent soluble (Fig. 6G). The wild-type reporter is detected along the entire flagellum, though its concentration along the innermost segment, immediately adjacent to the basal bodies, might be decreased (Fig. 6H). In dividing cells, the reporter is located both on the preexisting flagellum and on the nascent flagellum (Fig. 6I). In mutant B1_M1, the first helical region is predicted to be disrupted by the replacement of T28E29 by two prolines (Fig. 3). In whole cells, the corresponding reporter is confined to the cell body, where it accumulates at the base of the flagellum. It is entirely absent from the flagellum (Fig. 6B) and is fully Triton extractable (Fig. 6K). In mutants B1_M2, B1_M3, and B1_M5, no alterations in the helical structures are predicted, and all three mutants exhibit wild-type localization (as a representative example, see B1_M3 in Fig. 6L). As expected, the wild-type B2 reporter B2(1-70)::GFP localizes to the cell body, is absent from the flagellum (Fig. 6D), and is fully Triton soluble (Fig. 6L). In contrast, both mutant B2_M6 and mutant B2_M7 localize not only to the cell body but also to the flagellum (Fig. 6E and F). However, they both remain Triton soluble.
Fluorescence of cytoskeletons of cells transiently transfected with individually mutated reporters. (A to F) Whole cells; (G to M) Triton-extracted cytoskeletons. The constructs used for transfection are given for each panel. Green, GFP reporter; red, Y1/2 antibody against tyrosinated tubulin which stains the growing tip of the cytoskeleton and the basal bodies (48); blue, DAPI staining visualizing cell nuclei and kinetoplasts. (A to G and I to M) Bars, 5 μm; (H) bar, 1 μm.
DISCUSSION
Strict control of intracellular cAMP is crucial for the survival of T. brucei. As shown in other eukaryotes (1, 23, 49), strict compartmentalization of cAMP signaling and the separate regulation of individual cAMP pools are probably important in T. brucei as well. Earlier work from the Institute of Cell Biology has shown that two of the five PDEs that are encoded in the T. brucei genome, PDEB1 and PDEB2, are the major players for the control of intracellular cAMP (31). PDEB1 is predominantly located along the flagellum, where it most likely is associated with the PFR. In contrast, PDEB2 is mostly located throughout the cell body, with a minor fraction colocalizing with PDEB1 in the flagellum (31). Upon Triton X-100 fractionation, PDEB2 from the cell body is quantitatively solubilized. In contrast, PDEB1 and the fraction of PDEB2 that is localized in the flagellum cannot be solubilized by Triton. In fact, the flagellum-localized PDEs become solubilized only under conditions where the PFR structure starts dissociating (unpublished data), indicating that they are stably integrated into the flagellar skeleton. This is in good agreement with the observation that both PDEs are present in the flagellar proteome after Triton extraction and after extensive washing of the flagellar skeleton with 1 M NaCl (6). While this observation was made with bloodstream forms, a more recent paper from the same group did not detect the two PDEs in the flagellar proteome obtained from procyclic cells under similar conditions (36). It is currently unclear if this difference merely reflects variations in the methodology or if it points to biological differences between the flagella of the two life cycle stages (36). Flagellar PDE1 and PDE2 can be solubilized from the flagellar skeleton by RIPA buffer (see Materials and Methods), which contains 0.1% SDS. Despite these harsh conditions, the two proteins can be coimmunoprecipitated, indicating either that they form detergent-stable heteromers or that they are tightly interacting with a common scaffold protein. However, no such hypothetical protein was detected when the immunoprecipitates were analyzed by gel electrophoresis and mass spectrometry. Similarly, sucrose gradient sedimentation indicated that the two PDEs sediment as homo- or heterodimers, again favoring the former hypothesis. This is further supported by the recent demonstration, by X-ray crystallography, that human PDE2A, an enzyme with an architecture that is closely similar to the architectures of trypanosomal PDEB1 and PDEB2, forms a parallel dimer, with the two GAF domains and the catalytic domains providing the contact surfaces (34).
Sequence alignments between PDEB1 and PDEB2 demonstrated that the overall degree of sequence identity is >75% but that the N-terminal 212 amino acids share only 25% sequence identity. This immediately suggested that the N-terminal region might be responsible for the different subcellular localizations of the two isoenzymes. Further analysis of distinct domains (the N-terminal 212 or 70 amino acids, GAF-A domain, GAF-B domain, and catalytic domain) of PDEB1 and PDEB2 from six different kinetoplastid species further strengthened this view. The GAF-A, GAF-B, and catalytic domains all cluster according to species; i.e., the corresponding domains of PDEB1 and PDEB2 of each species are closest neighbors. In contrast, the N termini (either the first 70 or the first 212 amino acids) cluster according to isoenzyme type, suggesting different functions of the two N-terminal regions. GFP reporter constructs fused to sequentially truncated N termini of PDEB1 revealed that the first 70 amino acids are both necessary and sufficient for directing the reporter to the flagellum and for mediating its stable (i.e., Triton-resistant) integration into the flagellar skeleton. In contrast, reporters carrying the first 212 amino acids of PDEB2 remain in the cell body. These observations established that the first 70 amino acids of PDEB1 contain the targeting and flagellar integration signals, and no other parts of the PDEB1 enzyme are apparently required to direct localization. For further confirmation, two additional reporters were used that corresponded to full-size PDEB1 whose catalytic domain had been replaced by GFP and full-size PDEB1 whose first 114 or 70 amino acids had been deleted. Both reporters remained Triton soluble, further underlining the crucial role of the N-terminal 70 amino acids for a stable integration into the flagellar skeleton. Thus, the Triton-soluble cell body localization may be the default state of PDEs, unless they contain an N-terminal signal that confers their transport to the flagellum and their stable integration into its skeleton.
Given the fact that the N terminus of PDEB1 mediates flagellar localization and stable integration into its skeleton while the corresponding region of PDEB2 does not, the question remained as to why a minor fraction of PDEB2 is found in stable association with the flagellum. Coimmunoprecipitation and sedimentation analyses suggested that PDEB1 and PDEB2 can form stable heteromers. These could conceivably be pulled to the flagellum and integrated into its skeleton via the N terminus of the PDEB1 partner.
The N termini of PDEB1 and PDEB2 entirely lack recognizable motifs or predicted structures that could be correlated to their function. Attempts to identify flagellum-specific motifs by structure- or motif-based homology searches of sequence databases, including the trypanosome flagellar proteomes, were negative. Structural prediction of the N termini did not produce any models with high confidence (the C scores of I-Tasser models varied between −1.8 and −2.5), with the exception of revealing two or three helical regions in either sequence that were robustly predicted. Point mutations were then introduced to query if the predicted α-helical regions might represent or contribute to the signal required for flagellar localization and stable integration into its skeleton. The selected point mutations were introduced into the B1(1-70)::GFP and the B2(1-70)::GFP reporter constructs. The wild-type B1(1-70)::GFP construct is localized along the entire flagellum. In dividing cells, the reporter is present both in the preexisting flagellum and in the newly forming flagellum. The very first segment of the flagellum, immediately adjacent to the basal body, always showed much weaker fluorescence than the remainder of the flagellum, suggesting an outward gradient of PDEB1 density along the flagellum. In mutant B1_M1, the first helical region is disrupted by the introduction of two successive proline residues. Interestingly, this is the only B1 mutant of the entire set analyzed that behaved differently from the wild type. The B1_M1 construct remains in the cell body, is completely absent from the flagellum, and is fully extractable by Triton X-100. Within the cell body, B1_M1 accumulates toward the base of the flagellum, but does not enter it. This may suggest that the mutant protein is still directed correctly toward the flagellar entry but cannot interact with the flagellar transport machinery and thus is unable to enter the flagellum. An uninterrupted helix 1 may be required for this crucial interaction with the flagellar transport. The subcellular localization of the B1_M1 reporter is distinctly different from that of the GFP control, indicating that the observed accumulation around the flagellar base is mediated by its N-terminal 70 amino acids.
In the N terminus of PDEB1, a short motif (H20L21C22) within the predicted helix 1 shows some similarity to a flagellar translocation signal, HLA, that was described and experimentally verified in the PFR protein (16). The motif in PDEB1 is HLC, not HLA, but alanine and cysteine are structurally similar amino acids, both being small and of similar hydrophobicities (hydropathy indices, 1.8 and 2.5 for alanine and cysteine, respectively). In PDEB2, this motif is changed to HLL; i.e., its last amino acid is larger and more hydrophobic. To investigate a potential role of this motif in flagellar localization, the two sequences were interconverted, replacing C22 with L22 in PDEB1 (mutant B1_M5) and L22 with C22 in PDEB2 (mutant B2_M6). Mutant B1_M5 behaved exactly as wild-type PDEB1. Mutant B2_M6 largely remained in the cell body, as does the wild-type B2(1-70)::GFP construct, but it also exhibited a marked flagellar staining. However, all of B2_M6 remained fully Triton soluble, indicating that even the fraction that is localized to the flagellum does not become stably integrated into the flagellar skeleton.
The altered subcellular localization of B2_M6 might also be due to the predicted disruption of helix 2. A similar disruption of helix 2 is also predicted for B2_M7, which similarly directs the reporter to the flagellum in whole cells. As in B2_M6, all of the B2_M7 reporter also remains Triton soluble. Interestingly, B2_M7 also accumulates at the flagellar base, as seen with the B1(1-70)::GFP wild type and with mutant B1_M1. The disruption of helix 2 that is predicted for both B2_M6 and B2_M7 may interfere with their stable integration into the flagellar skeleton. In conjunction, our data demonstrate that the N-terminal 70 amino acids of PDEB1 contain all signals required for flagellar localization and for the stable integration into the flagellar skeleton. The detailed nature of this signal(s) remains to be understood.
The observation of a stable integration of PDEs into the flagellar skeleton demonstrates that the flagellar skeleton may represent a scaffold for spatially organizing the PDEs and cAMP signaling. This is in good agreement with the current paradigm of cyclic nucleotide signaling in eukaryotic cells at large. It has become obvious that the PDEs are mostly bound to scaffolding proteins which assemble the major components of different signaling pathways (1, 17, 23, 49, 55). The close physical proximity of all components is important for many aspects of signaling, such as controlling the diffusion range of cyclic nucleotides, mediating cross talk between different signaling pathways, and permitting feedback to input components, such as the cyclases and the receptors that control their activity. The observation that the flagellar scaffold serves as a solid-state support for enzymes involved in signaling or energy metabolism is not novel (33, 37), but it further underlines the importance of this structure to organize intracellular processes.
The role of cAMP signaling in the flagellum remains to be explored. Regulation of flagellar motility is an obvious option, and unpublished data from two different approaches suggest that this is the case in procylic-form but not in bloodstream-form trypanosomes (M. Oberholzer, unpublished data; A. Schmid and T. Seebeck, unpublished data). Given the fact that the flagellum may be important for environmental sensing and for the interaction with host tissue, it is also conceivable that receptor-type adenylyl cyclases in the flagellar membrane, in combination with PDEs in the flagellar skeleton, may represent a sensory circuit of the trypanosomes. While these questions constitute the immediate next steps toward understanding the mechanics of PFR structure and cAMP signaling, the trypanosomal flagellum might also represent an attractive model system for studying the interaction of eukaryotic signaling pathways with scaffolding structures in general.
ACKNOWLEDGMENTS
We are greatly indebted to Xuan Lan Vu for her relentless technical support, to Jay Bangs (University of Wisconsin) for his BiP antibody, to Isabel Roditi (Institute of Cell Biology) for the pGaprone-GFP vector, to Gaby Schumann (Institute of Cell Biology) for her patient help with transfection problems, and to Mani Heller (Department of Clinical Research) for mass spectrometry analysis.
This work was funded by grant 3100A-109245 from the Swiss National Science Foundation.
FOOTNOTES
- Received 12 May 2010.
- Accepted 22 July 2010.
- Copyright © 2010, American Society for Microbiology
REFERENCES
