DCGs are complex organelles. Their formation involves the association of a subset of luminal proteins in the trans-Golgi network. This process separates these proteins from others that remain soluble (2, 3, 12, 50). This condensate exits the trans-Golgi network in the form of an immature DCG and typically undergoes reorganization and further condensation during a maturation period. In this process, an aggregate with no apparent organization can be transformed into one with ordered contents (2, 37, 38). Upon exocytosis, the cores decondense and the cargo is dispersed, an event shaped by the interactions between core components. The soluble polypeptide cargo of adrenal chromaffin granules is condensed on an insoluble proteoglycan core (55). Upon exocytosis, the proteoglycan core rapidly expands and, in doing so, performs an important function in cargo release (39, 44). Both the generation and function of DCGs are based on compartment-specific cargo condensation and decondensation. For these reasons, the cargo can be considered to function as an active participant rather than as a mere passenger.

DCGs arose early in eukaryotic evolution and are well developed in the ciliates. The strength of classical and molecular genetics approaches in Tetrahymena thermophila and Paramecium tetraurelia makes these organisms attractive model systems for mechanistic analysis of regulated secretion. In both organisms, DCGs have received attention at the molecular level. Whereas DCG contents in many multicellular organisms were initially identified by their physiological activities (e.g., neuropeptides), the largely unknown functions of ciliate DCGs precluded this approach, and contents were identified based on their abundances in stimulated-cell supernatants (8, 35, 48, 53) or by screening for gene products essential for exocytosis (7). DCG cargo in T. thermophila consists principally of a family of granule lattice (Grl) proteins. These are derived by proteolytically processing soluble precursors. During this transition, they form an insoluble lattice (52). Upon exocytosis, the lattice remains intact but undergoes expansion to propel the DCG cargo from the cell (25). The Grl family is orthologous to the tmp protein family (trichocyst matrix proteins) in P. tetraurelia (34). The disruption or silencing of any one of the core protein genes blocks normal core formation, and each core protein can thus be considered an essential structural element of the granule cores (8, 46) (D. C. Chilcoat and A. P. Turkewitz, unpublished data).

If Grls/tmps constitute a core structure that remains insoluble, what are the molecules that are being released and how are they targeted to DCGs? Although there is no direct evidence, behavioral observations hint that ciliate DCGs release soluble cargo. For example, some predatory ciliates use regulated exocytosis to immobilize prey: the paralysis of the hapless victim implies the delivery of toxins (26). DCGs in other ciliates appear to serve defensive functions (24, 30).

Here we have addressed the task of identifying novel cargo proteins that may play roles distinct from those of the known components, taking advantage of the ability of DGCs in T. thermophila to undergo synchronous exocytosis. We have cloned genes whose transcription is stimulated by exocytosis and have characterized the product of IGR1 (induced during granule regeneration 1).

All reagents were from Sigma Chemical Co. (St. Louis, Mo.) unless otherwise noted.

Cells and cell culture. Cells were grown at 30°C with moderate agitation in a mixture containing 1% Proteose Peptone, 0.2% dextrose, and 0.1% yeast extract (all from Difco Laboratories, Detroit, Mich.) and 0.003% ferric EDTA. T. thermophila strains are designated by their micronuclear diploid genotype, followed by their macronuclear phenotype in parentheses (41). The heterokaryon strains CU428.1 mpr1-1/mpr1-1 (mp-s, VII) and B2086 mpr^s/mpr^s (mp-s, II) were from Peter Bruns (Cornell University, Ithaca, N.Y.) (41). Exocytosis-deficient (exo⁻) mutant SB281 was a gift from Ed Orias (University of California, Santa Barbara, Calif.) (42). The exo⁻ strain MN173 (V) has been described previously (36). Strains harboring pVGF-derived vectors (28) were maintained in 120 μg of paromomycin sulfate/ml.

Genetic nomenclature for T. thermophila has recently been formalized and is followed in this paper (1).

Differential-display PCR. Stimulation of exocytosis with Alcian blue and subsequent recovery were performed as described previously by using a wild-type (CU428.1) and an exo⁻ (MN173) strain (22). Total RNA preparations were isolated as described previously (22) and were subsequently treated with RNase-free DNase I (Genehunter, Nashville, Tenn.) according to the manufacturer's specifications. First-strand cDNA synthesis was performed using 2 μg of total RNA in a reaction mixture containing 50 mM Tris-Cl, pH 8.3, 75 mM KCl, 3 mM MgCl₂, 10 mM dithiothreitol, 20 μM deoxynucleoside triphosphates (dNTPs), 1 μM single degenerate poly(dT) oligonucleotide primer (four total 14-mers, T₁₂MA, T₁₂MT, T₁₂MC, and T₁₂MG, where M = A, G, or C; Genehunter), and 200 U of Superscript II RNase H⁻ reverse transcriptase (Life Technologies, Rockville, Md.), which was incubated at 65°C for 5 min and then at 37°C for 1 h. One-tenth of a single reaction mixture was then used for the PCR: 50 mM KCl, 10 mM Tris-Cl (pH 9.0 at 25°C), 0.1% Triton X-100, 1.5 mM MgCl₂, 2 μM dNTPs, 21 μCi [α-³³P]ATP (2,000 Ci/mmol; NEN Life Science Products, Boston, Mass.), 2 U of Taq polymerase (Promega, Madison, Wis.), 1 μM T₁₂MN, used for the cDNA synthesis, and a 1 μM 10-mer oligonucleotide primer (A1, AGCCAGCGAA, A2, GACCGCTTGT, A3, AGGTGACCGT, A4, GGTACCCAC, and A5, GTTGCGATCC [Genehunter], and O1, CAGGCCCTTC, O2, TGCCGAGCTG, O3, AGTCAGCCAC, O4, AATCGGGCTG, O5, AGGGGTCTTG, O6, GGTCCCTGAC, O7, GAAACGGGTG, O8, GTGACGTAGG, O9, GGGTAACGCC, O10, GTGATCGCAG, O11, CAATCGCCGT, O12, TCGGCGATAG, O13, CAGCACCCAC, O14, TCTGTGCTGG, O15, TTCCGAACCC, and O19, CAAACGTCGG [Operon, Alameda, Calif.]). Reaction mixtures were subjected to the following: 94°C for 30 s, 40°C for 2 min, and 72°C for 30 s for 40 cycles and then 72°C for 5 min. Approximately one-fourth of the reaction mixture was resolved on a 6% denaturing polyacrylamide gel. Gels were first analyzed with a phosphorimager (Storm 860; Molecular Dynamics, Sunnyvale, Calif.) and then exposed to X-ray film. DNA from a particular gel position was cloned by first repeating the PCR described above (omitting the radionucleotide) using DNA recovered from an excised gel slice as the template. These PCR products were cloned either directly or following agarose gel purification using the TA cloning system (Invitrogen, Carlsbad, Calif.).

Northern blotting was performed using radioactive probes made from a representative clone of each unique sequence in order to determine the mRNA levels of each clone's corresponding gene. Total RNA (~20 μg), isolated identically to the samples used for the differential-display reactions, was probed as described previously with the following modifications: one-half of a riboprobe synthesis reaction mixture (10 mM dithiothreitol, 500 μM [each] rATP, rGTP, rCTP, 100 μM UTP, 50 μCi [α-³²P]UTP [6,000 Ci/mmol], ~1 μg of DNA template, and either SP6 or T7 RNA polymerase [reaction buffer and conditions were provided by the supplier]) was added to 5 ml of hybridization buffer (50% formamide, 5× SSPE [1× is 180 mM NaCl, 10 mM NaH₂PO₄ {pH 7.4}, 10 mM EDTA {pH 7.4}], 0.1% sodium dodecyl sulfate [SDS], 100 μg of denatured salmon sperm DNA/ml) and allowed to hybridize overnight at 55°C. Blots were washed twice in 0.1× SSPE-0.1% SDS for 15 min at 55°C and exposed to a phosphorimager screen for analysis.

cDNA and genomic cloning. cDNA clones were obtained for many of the genuine differential display amplicons via PCR. An oligonucleotide primer was designed to direct polymerization toward the 5′ end of the gene. This primer was used in conjunction with a primer complementary to the cDNA library vector. Specific reaction conditions for the PCR were empirically determined by exploring a number of annealing temperatures, DNA polymerases, and cDNA libraries as sources of templates (λgt10 was a kind gift from Tohru Takemasa [University of Tsukuba, Tsukuba, Japan], and two plasmid-based full-length libraries are described in reference 7). The PCR was performed using 50 to 100 ng of library DNA, polymerase-specific buffer, 250 μM dNTPs, and either Taq (Promega), Expand (Roche Molecular Biochemicals, Indianapolis, Ind.), or Pfu (Stratagene, La Jolla, Calif.) thermophilic DNA polymerases. Products from positive reactions were cloned with the TOPO-TA cloning system (Invitrogen) and completely sequenced. Genomic clones encompassing particular genes were obtained via PCR using primers directed to the extreme 5′ and 3& ends of the cDNA and T. thermophila (CU428.1) genomic DNA as the template.

Sequence analysis. Sequences of both partial and full-length cDNAs were compared to entries in GenBank by using BLASTX. Conceptual translations of full-length cDNAs were analyzed for general protein features using Protean (DNA Star Software, Madison, Wis.) and further compared to known protein sequences in GenBank by using the PSI-BLAST program. Alignments using structural information were performed by using the Conserved Domain Database (National Center for Biotechnology Information, Bethesda, Md.). Potential amino-terminal signal sequences were analyzed by using the SignalP, version 2.0, World Wide Web server (www.cbs.dtu.dk/services/SignalP/).

Southern blot analysis. Genomic DNA was purified from T. thermophila as described previously (15). Approximately 20 μg of DNA was digested with various restriction enzymes in accordance with the suppliers' instructions. Southern blotting was performed either under the riboprobe conditions above or by nonradioactive detection using the GENIUS system (Roche Molecular Biochemicals).

Gene disruption. A 2.8-kb genomic DNA fragment containing the IGR1 open reading frame (ORF) was obtained via PCR using oligonucleotide primers internal to an XbaI site flanking this locus. The DNA sequence encoding these primers was obtained from a product of an inverse PCR designed to yield 650 bp 5′ and ~1,500 bp 3′ of IGR1. The NEO2 cassette (described in reference 14) was inserted between the PacI site and HincII site in IGR1, a region which spans from −47 to +199 (numbers are relative to translational initiation), via ligation of BstXI adapter oligonucleotides. Cells were transformed with the Biolistic device (Bio-Rad, Hercules, Calif.) as described previously (4). Macronuclear transformants were initially selected at 120 μg of paromomycin sulfate/ml and subsequently passaged for approximately 9 weeks during which the drug concentration was increased to 500 μg/ml.

Construction and expression of fusion and truncated proteins. All genes encoding fusion proteins in this study were expressed by using vectors derived from pVGF.1 (28). Fusions to green fluorescent protein (GFP) were generated by first adding appropriate restriction sites at the extreme 5′ and 3′ ends of the ORF and ligating into the PmeI site of pVGF.1 (IGR1 P1, GGGGCAGCTGGCAAAATGAGAAAGATCA; P2, TTGTCAGCTGAAATTTCTTCTGTTGTT). DNA sequences encoding the hemagglutinin (HA) tag (YPYDVDPYA) were incorporated into the IGR1 sequence encoding the C terminus via a PCR approach using forward primer P1 (CCTGATTATGCTTGAAAAAGACAATTCTATTAGTA) and reverse primer P2 (AACATCATAAGGATAATTTCTTCTGTTGTTTCTTCTG), which, upon intramolecular ligation of the linear amplification product, create the proper coding sequence (17). This created igr1-1(HA). A similar strategy was used to generate the deletion constructs. The signal sequence-GFP fusion involved joining the first 20 amino acids encoded by IGR1 to the amino terminus of GFP; igr1-2(Δ31-183) was derived from igr1-1(HA). Upon confirmation of the predicted DNA sequences, these genes were inserted between the PmeI and XhoI sites of pVGF.1 (replacing the vector GFP coding sequence). Positive transformants harboring each construct were obtained by electroporation of conjugating pairs (14).

Protein methods and subcellular fractionation. Preparation of whole-cell detergent lysates from T. thermophila was as described previously (52); the protocol for purifying DCG contents following dibucaine stimulation is in reference 51 and essentially consists of two to four washes to separate the insoluble DCG lattices from both soluble proteins and cells. For experiments with strains harboring pVGF-derived vectors, cultures were used at a density of ≤1 × 10⁵ cells/ml to maintain robust gene expression. Protein concentrations were determined by using the bicinchoninic acid (BCA) protein assay (Pierce, Rockford, Ill.). Following standard SDS-polyacrylamide gel electrophoresis (PAGE) and gel transfer protocols, nitrocellulose membranes were stained with Ponceau-S and subjected to standard Western blotting protocols (23). The Grl1p-specific polyclonal antibody (anti-p40) (52) was used at 1:1,000; blocking and antibody incubations were done in 5% dry milk. The HA-specific monoclonal antibody (HA.11/16B12; Covance, Richmond, Calif.) was used at 1:2,000, with blocking and antibody incubations done in 3% bovine serum albumin. Detection with chemiluminescence system SuperSignal (Pierce) was performed according to the manufacturer's instructions. For radioactive detection, ~1 to 5 μCi of ¹²⁵I-protein A (ICN Biomedicals, Costa Mesa, Calif.) was incubated directly (Grl1p) or following an incubation with a rabbit anti-mouse secondary antibody (HA; Jackson Immunoresearch Laboratories, West Grove, Pa.) with membranes and subsequently exposed to a phosphorimager screen for analysis.

For equilibrium density gradient ultracentrifugation, cells were disrupted with a ball bearing homogenizer as described previously (51). Roughly 2 × 10⁷ cells were washed and mechanically lysed at a concentration of ~10⁷/ml in the presence of protease inhibitors. Following lysis, the homogenates were loaded and sedimented through a continuous gradient in which the bottom solution contained 45% Nycodenz, 10 mM HEPES, pH 7.0, 1 mM EGTA, 1 mM MgCl₂, and 0.2% gelatin and the top solution was identical except that the Nycodenz was replaced with 0.28 M sucrose. Centrifugation was for 24 h at 30,000 rpm in an SW-41 rotor (Beckman, Fullerton, Calif.). One-milliliter fractions were collected and subsequently assayed for Grl1p and HA-reactive species by using the Western blotting conditions described above.

Detergent lysates. Approximately 1 × 10⁶ to 2 × 10⁶ cells were washed twice in a solution containing 0.17 mM sodium citrate, 0.1 mM sodium phosphate, 0.1 mM disodium phosphate, 0.65 mM CaCl₂, and 100 μM MgCl₂ and resuspended in a volume of 0.25 ml. Cells were lysed with a buffered detergent solution containing 1% Nonidet P-40, 0.4% sodium deoxycholate, EDTA (52), and protease inhibitors (E-64, antipain, leupeptin, chymostatin A) (51). Lysates were incubated on ice for 4 h and then centrifuged at top speed in a cold Microfuge (Brinkman Instruments, Westbury, N.Y.) for 10 min. Both supernatant and pellet fractions were subsequently analyzed by Western blotting.

Selection for genes induced by exocytosis. T. thermophila replaces a full complement of DCGs, called mucocysts in this species, within 4 h following induction of global synchronous exocytosis (22). During this period of regranulation, the mRNAs for two granule content genes, GRL1 and GRL4, accumulate within 60 min to about 10 times the prestimulation levels. In several exocytosis-deficient mutants, by contrast, there is no increase in these mRNAs in response to stimulation, indicating that transcription of granule-related genes responds to exocytosis itself.

Since the expression of other granule component genes is likely to be similarly induced under these circumstances, we identified transcripts that were more abundant following the stimulation of wild-type, but not exo⁻ mutant, cell lines. We isolated total RNA from two strains, wild-type CU428.1 and exo⁻ MN173, prior to and 60 min after stimulation. In the MN173 mutant strain, DCGs cannot dock at the plasma membrane and therefore do not undergo exocytosis (36). The relative abundances of individual mRNAs in these samples were assayed by the semiquantitative, reverse transcription-based PCR method termed differential display (33). Paired promiscuous primers (the first, a 5′ 10-mer, the second, one of four poly[A]-anchored 14-mers designed with mild degeneracy) were used to generate small amplification products (amplicons) from the starting mRNA templates. We identified and isolated amplicons that appeared more abundant after exocytosis in wild-type cells, ignoring that subset whose abundance also appeared to increase in the exo⁻ mutant under identical conditions.

A total of 102 products, ranging in size from 75 to >500 bp, were cloned and sequenced. These corresponded to 86 unique clones. Three of these sequences were exact matches for genes encoding components of the DCG lattice (GRL3, GRL5, and GRL7) (53) (Table 1). Since these were known to be transcriptionally induced following exocytosis, their isolation indicated that the screen was likely to produce other genes relevant to DCG biosynthesis. Each of the remaining amplicons was used as a probe to determine the abundance of the corresponding mRNA in cells treated identically to those in the initial differential-display experiments. Thirteen of these represented genes whose transcript abundances were modulated as anticipated. A set is shown in Fig. 1A, with more detailed kinetics for a subset in Fig. 1B. We amplified cDNA clones from several T. thermophila cDNA libraries for eight of these expressed sequence tags (ESTs) by using a combination of a vector-specific and an EST-specific primer. BLAST searches revealed clear homologues for seven of the full-length clones but none for the ESTs (Table 1).

TABLE 1. Differential-display PCR isolates

FIG. 1. Isolation of genes induced during DCG biogenesis in T. thermophila. (A) Each unique amplicon identified by differential-display PCR was used as a probe to assay the abundance of the corresponding mRNA in total RNA prepared from cells before (−) (more ...)

Identification of a novel DCG marker. We pursued clones encoding proteins whose likely amino-terminal signal sequences indicated targeting to the secretory pathway (54): TKI1, GIP1, and IGR1. The proteins were expressed as GFP fusion proteins to determine in vivo localization. DCGs in T. thermophila are positioned along linearly aligned plasma membrane docking sites (25). The GFP-tagged product of IGR1 [igr1-4p(GFP)] was confined to a bright and uniform constellation of puncta along the cell surface (Fig. 2A). That these corresponded to DCGs was confirmed by treating the cells with a secretagogue to induce massive synchronous exocytosis of the entire cohort of DCGs. Secretagogue treatment of cells expressing the GFP fusion resulted in virtually complete disappearance of the intracellular fluorescence (not shown). Identical patterns were observed with cells expressing a GFP fusion to known DCG core protein Grl1p (Fig. 2B).

FIG. 2. Expression of GFP fusion proteins reveals that the IGR1 product localizes to DCGs in vivo. Confocal micrographs of T. thermophila strains harboring GFP-tagged constructs. Tangential optical sections are shown. Bar, 10 μM. (A) Cells expressing (more ...)

The observed localization of igr1-4p(GFP) is meaningful only if GFP does not itself concentrate in DCGs. We therefore created a construct in which GFP was fused at its amino terminus to the signal sequence derived from Grl1p. The expressed protein was targeted to the secretory pathway and appeared within bright, highly mobile small vesicles, as well as in a faint reticular pattern (Fig. 2C). Importantly, there was no labeling of vesicles at the cell surface. GFP expressed by itself, unlinked to a signal sequence, gives uniform staining of the cytoplasm (22). We conclude that igr1-4p(GFP) is localized to DCGs and that this depends on signals in Igr1p itself.

As expected for a gene encoding a DCG cargo protein, IGR1 expression paralleled that of GRL genes for several hours following exocytosis. The mRNA rapidly accumulated during DCG biogenesis and decreased at later times. This time course was identical to that of GRL3 mRNA, although the absolute levels of induction were different. The correspondence is not seen for genes unrelated to DCGs (Fig. 1B).

Igr1p is structurally unrelated to known ciliate DCG proteins. Conceptual translation of the 930-bp ORF predicted a 309-amino-acid protein (Fig. 3A). Comparison of IGR1 cDNA and genomic DNA sequences indicated that the gene contains no introns (not shown), and Southern blotting of restriction-digested genomic DNA using the IGR1 cDNA clone as the probe revealed a single strongly reactive band in most lanes (Fig. 3B). A faint secondary band in the EcoRI-digested genomic DNA may reflect the existence of a related gene. Except for the amino-terminal signal sequence, the predicted protein does not contain significant regions of hydrophobicity and is therefore unlikely to integrate stably into the membrane (Fig. 3C). The primary sequence indicated that the new protein has a significantly different character from those of all of the Grls. The Grls, as well as the orthologous tmps in P. tetraurelia, are highly acidic and are predicted to fold almost exclusively as coiled coils (18, 53). In contrast, Igr1p is a slightly basic protein with no preponderance of coiled coils. Thus, the novel protein is unrelated by sequence to the Grls and may therefore represent the first member of a second class of proteins identified in T. thermophila DCGs.

FIG. 3. Sequence and predicted features of Igr1p. (A) The 930-bp IGR1 ORF encodes a 309-amino-acid protein, whose sequence is shown. The predicted amino-terminal signal sequence is in italics. (B) Southern blot of T. thermophila genomic DNA digested with Eco (more ...)

We failed to detect any database matches to the amino-terminal sequence of Igr1p. However, the carboxy-terminal 108 amino acids (representing about one-third of the protein) bears significant homology to the carboxy-terminal regions of a family of proteins in P. tetraurelia called Pcmps (Paramecium calmodulin-binding membrane proteins) (BLASTp; E value, <1e-4) (5). The remaining portions of Igr1p and the Pcmps have no detectable similarity. The initial description of the Pcmps as cytosolic proteins is considered below.

The similarity to the Pcmps carboxy-terminal domain may provide a hint about the structure of Igr1p, since the Pcmp domain appears distantly related to the β/γ-crystallin family of proteins (5). The β/γ-crystallin domain consists of a compact array formed by two four-stranded antiparallel beta sheets (47). To consider whether the carboxy-terminal domain in Igr1p might have a similar fold, we aligned the Igr1p carboxy-terminal domain with a Pcmp as well as an authentic β/γ-crystallin domain. This tentative alignment was based on the positions of a small number of residues conserved among crystallins that are also present in the ciliate proteins (Fig. 4A). We then attempted to trace the Igr1b primary sequence in the β/γ-crystallin tertiary structure (Fig. 4B). Those residues that are absolutely conserved, two glycines and a serine, all fell within relatively tight turns either preceding or in the first position of a beta strand. These are positions where side chain packing would be tightly constrained. The insertions or deletions in Igr1p, relative to the β/γ-crystallin, are found within loops rather than within the core structure, an indication that the superimposition of the ciliate sequences on the structure is not unreasonable (Fig. 4A). A conspicuous feature conserved among the ciliate proteins but absent in crystallin counterparts is four cysteines. When these residues are placed in the structure, they fall into two pairs. This suggests the possibility of disulfide bond formation, which would further stabilize this domain. This structural modeling does not imply any evolutionary relatedness between the ciliate proteins and authentic β/γ-crystallins, since similar structures can arise convergently. Apparent convergence on a crystallin-like fold has already been noted in several cases (9).

FIG. 4. Tentative alignment of ciliate proteins Igr1p and Pcm3p with bovine β B2-crystallin, domain 2. (A) Residues indicated by double-height letters are absolutely conserved among β/γ-crystallins (G36, S60, and G80) and were used as (more ...)

Characterization of epitope-tagged variants. For further biochemical analysis of Igr1p and preliminary dissection of putative sorting signals, we expressed full-length Igr1p with a carboxy-terminal HA epitope tag [igr1-1p(HA)] as well as a similarly tagged construct consisting of Igr1p from which we deleted amino acids 31 to 183 [igr1-2p(Δ31-183,HA)] (Fig. 5A). The latter protein contained the signal sequence (amino acids 1 to 30) followed by carboxy-terminal amino acids 184 to 309. This carboxy terminus includes the region which, based on the structural arguments presented above, seemed likely to form an independently folded domain. We likewise created a truncated protein consisting solely of the amino-terminal region (amino acids 1 to 200). This, however, could not be stably expressed in cells.

FIG. 5. Expression and characterization of HA-tagged full-length and truncated Igr1p. (A) The HA nonapeptide was fused to the extreme carboxy termini of Igr1p and a truncated derivative. The former is igr1-1p(HA). In the latter, igr1-2p(Δ31-183,HA), the (more ...)

Western blot analysis of cell lysates prepared from strains expressing either full-length igr1-1p(HA) or igr1-2p(Δ31-183,HA) revealed strong HA-reactive species. These sometimes appeared as doublets (Fig. 5B), which may reflect limited amino-terminal proteolytic degradation; importantly, the different species behaved identically in all experiments we have done. We have never detected a larger form in cell lysates, suggesting that Igr1p, unlike the Grls, is not processed from a proform. Further evidence against processing came from expressing igr1-1p(HA) in cell line SB281, a mutant that is defective in DCG biosynthesis and in Grl1p processing, so that Grl1p accumulates as the proprotein (13, 52). Igr1p expressed in SB281 was identical in mobility to that expressed in wild-type cells (not shown). No proteins reactive to HA antibodies were seen in untransformed cells, either wild type or SB281.

GFP-tagged Igr1p appeared to be localized exclusively to DCGs in vivo. For technical reasons, we could not visualize HA-tagged Igr1p in living or fixed cells, so we examined its distribution using subcellular fractionation. Postnuclear supernatants of cell lysates were resolved by equilibrium density gradient centrifugation, and the fractions containing igr1-1p(HA) were identified by Western blotting. DCGs, as marked by Grl1p, were found near the bottom of such gradients. The igr1-1p(HA) was found in the same fractions as Grl1p (Fig. 5C, top). There was a slight difference in the distributions of these two markers: while Grl1p was more abundant in fraction 10 than 9, the opposite was true for igr1-1p(HA). Several possible explanations are discussed below.

The Igr1p carboxy-terminal domain localizes to DCGs. Signals that target proteins to DCGs are poorly defined, but small stable loops appear to be important determinants in several systems (6, 10, 11, 19, 45, 56). To ask whether the carboxy-terminal domain of Igr1p is sufficient for DCG targeting, we assayed the localization of igr1-2p(Δ31-183,HA). The truncated protein behaved identically to full-length igr1-1p(HA) on equilibrium density gradients, with its peak fraction slightly offset from that of Grl1p (Fig. 5C, bottom). One possible explanation for this difference in distribution is that the full-length and truncated proteins are present in a compartment that is distinct from, though of similar density to, DCGs. We made use of a functional assay to test the extent to which igr1-2p(Δ31-183,HA) was depleted from cells exposed to a secretagogue. Under conditions of rapid synchronous exocytosis, a large fraction of DCG contents is released within seconds of stimulation. In previous experiments, the cellular levels of Grl1p and of a second abundant DCG cargo protein, p80, decreased by 60 and 70%, respectively (52). When cells expressing igr1-2p(Δ31-183,HA) were stimulated, roughly 60% of the truncated protein was released within 15 s (Fig. 5D). Grl1p itself showed >90% depletion in this experiment. This result is representative of several in which depletion of Grl1p was always somewhat greater than that of Igr1p. An internal control was the unprocessed form of Grl1p, proGrl1p. proGrl1p is not present in mature docked DCGs and is not released from stimulated cells. We conclude that, at a minimum, the majority of the HA-tagged truncated protein is localized to DCGs.

Igr1p is not retained by the Grl-based insoluble lattice. Mature Grls are assembled into insoluble lattices that form the DCG cores. While the lattices expand during exocytosis, they remain insoluble. The tendency of expanded lattices to clump makes it straightforward to enrich for Grls and associated lattice proteins from stimulated cell supernatants by several washes using low-speed centrifugation. During this period, lattices are separated both from soluble proteins and from the bulk of the cells.

We determined the relative amounts of igr1-1p(HA) in whole-cell lysates versus enriched lattices (Fig. 6A). Grl1p was enriched ~20-fold in the insoluble lattice fraction, whereas igr1-1p(HA) was depleted ~2-fold in the same fraction. Igr1p in the lattice fraction is therefore ~40-fold depleted relative to Grl1p, in contrast to their relative abundances in the whole-cell lysates. As shown above, the majority of Igr1ps are rapidly released from cells during the stimulation period. In combination, these results indicate that Igr1p, unlike any previously identified DCG cargo proteins in ciliates, does not remain associated with the expanded cores following exocytosis.

FIG. 6. Igr1p is not stably associated with the Grl-based DCG lattice. (A) The insoluble contents of DCGs were enriched from stimulated cell supernatants as described in Materials and Methods. Grl1p and igr1-1p(HA) were detected by Western blotting of whole-cell (more ...)

A different approach to assessing the relative solubility of DCG cargo proteins is by fractionating detergent lysates of unstimulated cells. Preparation of such lysates in calcium-free buffers prevents lattice expansion. It was thus possible to ask whether Igr1p can be removed from DCG cores in the preexpanded state. More than 95% of Grl1p was found in the pellet fraction of such lysates, whereas ~30% of Igr1p was found in the supernatant (Fig. 6B). The pelletable fraction seemed likely to be associated with the insoluble Grl-based lattice. To test this hypothesis, detergent lysates were prepared from SB281 cells expressing igr1-1p(HA). As mentioned above, Grl proteins are not processed in SB281 and do not assemble into insoluble lattices. The igr1-1p(HA) was present exclusively in the supernatants of such lysates (Fig. 6C). We conclude that Igr1p can associate with the condensed lattice but that a significant fraction is readily dissociable. This may indicate the existence of two pools of Igr1p or may simply reflect slow dissociation kinetics.

Comparison of Igr1p to the Grl family of granule core proteins. The partial association of Igr1p with the condensed or expanded lattices suggested that it is not a structural component of these macromolecular assemblies and prompted the prediction that the absence of Igr1p would have no major effect on lattice structure or expansion. We disrupted all macronuclear copies of IGR1 by homologous recombination using established techniques (Fig. 7A). Southern blotting confirmed the correct targeting and replacement of the all copies of IGR1 in the polyploid macronucleus. Under these blotting conditions, the strongly hybridizing band containing the IGR1 gene is detected at ~3.0 kbp. The correct targeting of the IGR1 deletion construct was confirmed by observing a diagnostic shift in the electrophoretic migration of the strongly hybridizing band. Elimination of all endogenous copies of IGR1 was indicated by the absence of the restriction fragment seen in the wild type. An additional weakly hybridizing band is seen at 2.6 kbp (Fig. 3C). This band was unchanged in IGR1Δ strains and is consistent with the existence of an IGR1-related gene.

FIG. 7. Disruption of the macronuclear IGR1 gene. (A) The construct used for interruption of macronuclear IGR1. A 256-bp fragment including the translational start site of IGR1 was replaced with the NEO2 cassette and used to transform cells. Correct targeting (more ...)

IGR1Δ cells showed no deficiency in exocytosis. Their array of DCGs was indistinguishable from that of the wild type, and the DCGs themselves had no discernible alterations in position, dimensions, or lattice structure (Fig. 7B). When IGR1Δ strains were challenged with an array of secretagogues, they responded with robust exocytic responses that were qualitatively and quantitatively indistinguishable from those in wild-type cells (not shown). Since rapid exocytosis depends on lattice expansion, we conclude that Igr1p is not essential for the assembly of functional lattices. We cannot, however, rule out the possibility that the normal role of Igr1p is redundant with that of the putative Igr1p-related protein implied by our Southern blots. Such redundancy could account for the absence of a detectable phenotype in the IGR1Δ strain. This would be in contrast to what is found for proteins encoded by the GRL family, in which disruption of any individual gene is sufficient to interfere with normal granule formation.

We compared the proteins secreted from wild-type and IGR1Δ strains. Coomassie blue-stained SDS-PAGE gels revealed an identical pattern of polypeptides, with the exception of a very lightly stained band that was reproducibly seen in wild-type but not IGR1Δ cells (Fig. 7C). The polypeptide had an apparent molecular mass of ~34 kDa, agreeing well with the predicted molecular mass of 33.2 kDa. The low concentration of this protein in isolated lattices provides no information about its abundance in intact granules. However, the protein profile of purified intact DCGs, representing both insoluble and soluble cargo, similarly shows no substantial protein with the mobility of Igr1p (8). The relatively low abundance of the protein in DCGs, relative to that of the Grls, is inconsistent with a stoichiometric role in lattice formation. It does not, however, exclude the possibility that Igr1p could act as a lattice assembly factor.

The most significant result reported in this paper is the identification of a ciliate protein, Igr1p, that is localized to and secreted from DCGs but that is not an element of the core lattice. Igr1p does not serve an essential structural function, and its absence had no effect on core assembly or expansion. Furthermore, Igr1p was not stably associated with the cores and could be dissociated from the expanded Grl1p-containing lattices following exocytosis. These results fulfill the expectation that ciliate DCGs, like those in metazoans, release macromolecules that can diffuse in the cellular environs. Considering the architectural and molecular complexity of these secretory vesicles in ciliates, it seems likely that a variety of cargo proteins play roles including but not limited to the shaping of interactions with other microorganisms during predatory, defensive, and amatory encounters. The identification of Igr1p and other proteins, which must be copackaged with the architectural Grls, may offer new insights into ciliate biology. It should also contribute to the study of mechanisms involved in sorting to DCGs in eukaryotes, given the experimental advantages of ciliates relative to metazoan model systems.

We do not know the molecular basis for the sorting of Igr1p to DCGs, but it may involve association with proteins that constitute the structural core. In this regard, it would be interesting to know whether Igr1p is free to diffuse within the condensed DCG lattice. Our analysis of Igr1p dissociation from condensed lattices in detergent-solubilized cell extracts offers some hints. In these experiments, roughly 30% of the protein was soluble. The insolubility of the remaining 70% was likely due to association with the condensed lattice, since the insoluble fraction was negligible in mutants that fail to assemble such lattices. Two different scenarios may account for the 30% found in the supernatant, each of which is consistent with the subtle difference between the distributions of Grl1p and HA-tagged Igr1p on equilibrium density gradients. One possibility is that some Igr1p protein is localized in a compartment of the secretory pathway distinct from DCGs, though also of high density. If so it represents a minority of the protein, since the majority can be rapidly released during synchronous exocytosis of DCGs. If the former is detergent soluble while the majority associated with DCGs is insoluble, these two populations could account for the partial solubility of Igr1p in cell lysates. However, the fact that all of the GFP-tagged Igr1p protein appears localized to DCGs suggests a second scenario, namely, that the difference in density gradient distributions reflects a previously undetected heterogeneity within DCGs. At the same time, however, the localization of the GFP-tagged Igr1p cannot be considered definitive, since a non-DCG-localized pool might not be visible if it were relatively diffuse.

Igr1p within these DCGs may be heterogeneous in solubility if some of the Igr1p is bound to the DCG membrane. A precedent for such association of DCG cargo proteins with the limiting membrane is chromogranin B in neuroendocrine granules. Like Igr1p, chromogranin B has no obvious membrane interaction domain but a fraction appears membrane bound (19). Chromogranin B also participates in homo- and heterotypic protein interactions (40, 49). One can therefore imagine two overlapping fractions. The first is associated with the membrane and has limited interactions with other core proteins. The second is not membrane associated and has more-extensive interactions with other core proteins. If we model Igr1p on chromogranin B, the partial solubilization of Igr1p in detergent lysates may reflect the rapid release of the first, but not the second, fraction of the protein. This scenario is also consistent with the observation that a substantial minority of Igr1p remains cell associated after exocytosis. Igr1p that is bound to the DCG membrane may remain associated with the plasma membrane or with the DCG membrane that is rapidly recovered by exocytosis-coupled endocytosis.

Igr1p was identified because the regeneration of DCGs involves de novo synthesis of its cargo and therefore coordinated expression of the corresponding genes. This is true in many cell types including adrenal cortex cells (31, 32), pancreatic β-cells (20, 21), Xenopus laevis pituitary cells (27), and Paramecium cells (16). In T. thermophila, 86 differential-display PCR products appeared to increase in abundance during recovery from exocytosis. Nineteen of these failed to detect transcripts by hybridization and may not correspond to expressed genes. Preliminary characterization of 11 of the 86 revealed that 3 were already known in T. thermophila, 6 have likely homologues in other organisms, and 2 are novel. Expression of these genes appears linked to exocytosis, since they were not induced following stimulation of exo⁻ mutants. This linkage, however, may be indirect. For example, the increased transcription of TER1, the endoplasmic reticulum (ER) retrieval receptor gene, may reflect demands on the secretory pathway imposed by DCG protein traffic in exo⁺ strains (43). exo⁺ cells undergoing synchronous fusion of thousands of DCGs may also be subject to other homeostatic demands. Direct linkage, on the other hand, certainly accounts for transcriptional activation of DCG cargo-encoding genes. It may also account for the increased expression of a putative Na-Ca exchanger, since such a membrane-localized activity may be required for calcium depletion from immature DCGs, a step in the assembly of a functional lattice (53).

In summary, DCGs in T. thermophila contain at least two classes of luminal cargo proteins. One is structural, insoluble, and essential for lattice expansion. The second, reported in this paper, can be released in a soluble form upon exocytosis. The carboxy terminus of Igr1p shows homology to the P. tetraurelia Pcmps, previously identified as cytoplasmic constituents (5). The Pcmps, like Igr1p, appear to have conserved two pairs of cysteines within the carboxy-terminal domain, but conservation of paired cysteines is unlikely to occur in proteins resident in the reducing environment of the cytoplasm. In contrast, the secretory pathway in general and DCGs in particular have an oxidizing environment consistent with disulfide bond formation. Additionally, the deduced Pcmp polypeptide sequences also hint that they may be present within vesicles. For the two Pcmps in which the amino terminus has been reported, at least one appears to us to begin with an ER translocation signal (Pcm4p). Igr1p may therefore represent a family of proteins present as DCG cargo in ciliates. Last, the copackaging of proteins with heterogeneous physicochemical properties, a feature of DCGs in multicellular eukaryotes, appears to occur in ciliates as well.

This work was supported by NIH GM50946 to A.T. In addition, A.H. was supported by predoctoral training grant GM071836.

We thank D. Chilcoat, S. Melia, and M. Lacagna, as well as other members of the Turkewitz lab, for help and discussion, and Ted Steck for careful review of the manuscript.