Comparative Cell Biology and Evolution of Annexins in Diplomonads

Annexins are proteins that associate with phospholipids in a Ca2+-dependent fashion. These proteins have been intensely studied in animals and plants because of their importance in diverse cellular processes, yet very little is known about annexins in single-celled eukaryotes, which represent the largest diversity of organisms. The human intestinal parasite Giardia intestinalis is known to have more annexins than humans, and they contribute to its pathogenic potential. In this study, we investigated the annexin complement in the salmon pathogen Spironucleus salmonicida, a relative of G. intestinalis. We found that S. salmonicida has a large repertoire of annexins and that the gene family has expanded separately across diplomonads, with members showing sequence diversity similar to that seen across kingdom-level groups such as plants and animals. S. salmonicida annexins are prominent components of the cytoskeleton and membrane. Two annexins are associated with a previously unrecognized structure in the anterior of the cell.

The annexins have a patchy phylogenetic distribution in eukaryotes, but homologs have been detected in the genomes of animals (group A and B), fungi (group C), plants (group D), and some protists of the Excavata and SAR supergroups (group E) (3). Annexin diversity has been well studied in animals and plants, but comparatively little is known about the evolution, diversity, and characteristics of annexin homologs from protists. The annexin homologs in the diplomonad Giardia intestinalis, also designated alpha-giardins, are the best-studied annexins assigned to group E (4). The alphagiardins were originally discovered as prominent proteins of the structurally complex Giardia intestinalis cytoskeleton and found to be immunodominant during human G. intestinalis infection (5,6). Annexins in parasites have been suggested to be good vaccine candidates (7), and immunization with a live, recombinant alpha-1 giardin vaccine formulation provided protection from a Giardia challenge in a murine model (8). Three distinct crystal structures of alpha-giardins (alpha-1, -11, and -14 giardins) have been solved, and this has revealed that they contain the typical all alpha-helical annexin fold but with unusual calcium coordination schemes not known in other annexins (9)(10)(11). These features give alpha-giardins unique properties; the alpha-1 giardin response to the Ca 2ϩ concentration is highly unusual, with abrogated binding to acidic phospholipids at high concentrations, a feature that might be coupled to environmental sensing early during host colonization (11). Sequencing of the first G. intestinalis genome revealed the presence of 21 alpha-giardin genes that showed clear evidence of repeated duplications and divergence (4,12). Annexin homologs from Spironucleus, another parasitic diplomonad lineage, were reported to be more similar to other annexins than to alpha-giardins (13). Nothing is known about the biological role of annexins in Spironucleus, but recently, the publication of the Spironucleus salmonicida genome and the development of an efficient transfection system have enabled comprehensive molecular studies to be performed (14,15). The large evolutionary distance between Giardia and Spironucleus and large ultrastructural differences present an interesting system in which to explore the diversity and function of a large gene family spread across several major branches of the eukaryotic tree.
In this study, we took an evolutionary cell biology approach (16) to study the annexin homologs of S. salmonicida, combining experimental and phylogenetic methods. We demonstrate the distinct profiles of Ca 2ϩ association with membrane lipids of two of the S. salmonicida annexins. The annexins constitute expanded gene families in different diplomonad lineages, and the 14 distinct annexins of S. salmonicida show subcellular localization to the cytoskeleton, membrane, and cytosol. Proximity labeling with engineered ascorbate peroxidase (APEX) fusions and 3,3=-diaminobenzidine (DAB) as the substrate coupled to transmission electron microscopy (TEM) revealed the fine ultrastructural details of the localization of a subset of the annexins, including two that demarcate a novel cellular structure possibly involved in cell adherence. This study, combined with earlier results, shows that the annexin family is expanded in diplomonads and that many of these group E annexins interact with different cytoskeletal structures or the membrane.

RESULTS
Identification of annexin genes in the S. salmonicida genome. The presence of annexin homologs in the draft genome of S. salmonicida was investigated by reciprocal BLASTP searches employing alpha-giardins (giardial annexin homologs) and human annexin V as queries. This search identified 16 full-length annexin homologs, 14 of which were encoded by distinct genes (ANX1 to -14; Table 1). Two genes, ANX3 and -9, were each present in two identical copies. The S. salmonicida annexins displayed 23 to 41% amino acid sequence identity to alpha-giardins, as well as annexins from nondiplomonad eukaryotes. Only 3 of the 14 S. salmonicida annexins showed the highest identity to an alpha-giardin.
Primary structure of S. salmonicida annexins. We performed comparative analyses of S. salmonicida annexins to investigate the presence or absence of signature motifs typically found in annexins and/or alpha-giardins. A discriminative motif search of S. salmonicida annexins versus Giardia alpha-giardins with the MEME tool (45) revealed a motif that is identical or highly similar to AB loops (type II Ca 2ϩ coordination sites) in the former. Intact AB loops are unusual in alpha-giardins, with examples found only in alpha-14 giardin (17). An alignment of S. salmonicida annexins shows that they have at least one canonical AB loop across the four endonexin repeats (G-X-G-T-D-38X-D/E), except that annexin 13 appears to have lost the type II site in all four domains (Fig. 1). Annexins 3, 9, and 12 have canonical AB loops in all four repeats.
Annexins in animals are known to display divergent N-terminal sequences (before the first annexin repeat) that determine their interactions with other proteins (18). The N-terminal sequences of S. salmonicida annexins are mostly short (3 to 176 amino acids [aa]), with 10 extensions being shorter than 18 aa. This unusual feature has previously been observed in alpha-giardins and plant annexins (18). The clear exception is annexin 13, which is substantially larger than the other annexins in S. salmonicida and has an additional domain with low similarity to DNA sulfur modification protein DndD, which has been described as a putative structural maintenance of chromosomes protein-like ATPase (19).
The N-terminal domain of vertebrate annexins is known to be modified by phosphorylation, transglutamination, myristoylation, and S-glutathionylation (20). Lipid modification is also a factor in the localization of alpha-19 giardin in G. intestinalis (21). We employed the GPS-Lipid tool (http://lipid.biocuckoo.org/) (22) to predict potential modification positions in the annexins (see Fig. S1 in the supplemental material). Annexins 3 and 12 were each predicted to have potential palmitoylation sites and one myristoylation site resembling the alpha-19 giardin dual-acylation signal (21). Single and dual sites for geranylgeranyl and farnesylation in the C-terminal region were predicted for annexins 6 and 10, respectively.
To conclude, the primary sequence analyses of Spironucleus annexins indicate that, despite being highly divergent, they have retained Ca 2ϩ coordination via AB loops. Some members might be targets for protein modifications like lipid acylation to aid in subcellular localization.  regulated by the Ca 2ϩ concentration. As a first step to verify that S. salmonicida annexin homologs are bona fide annexins, we fractionated annexin-expressing transfectant cells to determine if they are associated with the cell membranes. We established transfectant cell lines expressing C-terminally epitope-tagged annexins from their native promoters (except annexin 7, which was expressed from the annexin 3 promoter). The presence of each annexin (except annexins 6, 8, and 11) in the cytosolic and membranous fraction of the transfectants was investigated by detergent-based selective fractionation with the Mem-PER reagent and Western blotting. Annexins 4, 5, 9, and 12 were distributed in both the cytosolic and membrane fractions, while much weaker membrane associations of annexins 2, 3, 10, and 13 were found. Annexin 1 and 8 signals were below the detection level in either fraction ( Fig. 2A; see Fig. S2A in the supplemental material). Next, we investigated the membrane-binding characteristics of two S. salmonicida annexin homologs, one being annexin 3 with four canonical AB loops and the second being annexin 5 with a single intact AB loop in the fourth endonexin repeat. Annexins 3 and 5, expressed as glutathione S-transferase (GST) fusions, were purified by glutathione affinity purification and incubated on membrane lipid strips that contain 15 different biological lipids. The membrane lipid strips were queried with an anti-GST antibody conjugated to horseradish peroxidase (HRP), followed by colorimetric detection of bound protein. The assay was performed in duplicate in the presence or absence of Ca 2ϩ (by EGTA chelation), and GST alone served as a negative control. We employed alpha-14 giardin and the GST-tagged phospholipase C-␦1 (PLC-␦1) domain as positive controls (10). Annexin 3 was found to associate to cardiolipin and phosphatidylserine (PS), with some minor binding also to phosphatidylglycerol (PG), all in a Ca 2ϩdependent fashion ( Less prominent binding to sulfatide and PG was also observed.
In summary, two divergent S. salmonicida annexin homologs were shown to be bona fide annexins with distinct phospholipid-binding profiles. Several other annexins were further recovered in the membrane fraction by cell fractionation.

Phylogeny of annexins.
We performed a maximum-likelihood phylogenetic analysis to investigate the evolutionary history of annexin homologs in eukaryotes. We included annexins from S. salmonicida and those of other hexamitid flagellates (S. vortens, S. barkhanus, Trepomonas sp. strain PC1), alpha-giardins from Giardia intestinalis and Giardia muris, and annexin homologs representing the diversity of eukaryotes (Fig. 3). Annexin homologs were found in all of the five currently recognized eukaryotic groups (23), indicating that the protein has a wide distribution in eukaryotes.
All diplomonad sequences form a cluster with strikingly large sequence diversity compared to that of other annexins. The hexamitid flagellate (S. salmonicida, S. barkhanus, S. vortens, and Trepomonas) annexins are found in a large cluster with alphagiardins nested within. However, the position of the root within the diplomonad cluster is likely a result of long-branch attraction; the hexamitid annexins and alpha-giardins are possibly sister clades, indicating that the annexin family has expanded independently in the two groups of diplomonads. Accordingly, there are no orthologous annexins between Giardiinae and Hexamitinae in the phylogeny (Fig. 3). The topology of the alpha-giardins agrees with the previous study of alpha-giardins of G. intestinalis (4), except in the position of the alpha-3 and alpha-15 giardins. In almost all cases, the G. muris homologs branch together with the G. intestinalis sequences for a specific alpha-giardin, suggesting functional conservation. G. muris alpha-giardins were present at chromosomal loci similar to those in G. intestinalis, but some individual genes were found to be missing or rearranged in the genome, indicative of lineage-specific duplications or gene losses (see Fig. S4A to D in the supplemental material). In a  recurring theme, the S. salmonicida annexins showed clusters on chromosomes 3, 4, and 6 (see Fig. S5A to D). The clusters on chromosomes 3 and 6 appear to represent loci of gene duplication based on the phylogenetic position of the genes involved (Fig. 3). The closely related species S. salmonicida and S. barkhanus share homologs for most genes, similar to the situation in G. intestinalis and G. muris. Across Hexamitinae, there are both orthologs and paralogs indicative of multiple gene duplications and losses (Fig. 3).
Expression of S. salmonicida annexins. In our earlier characterization of the S. salmonicida genome, we studied gene expression during growth in vitro by RNA sequencing (RNA-Seq) (15). This and later analyses showed that all of the 14 annexin genes are expressed in S. salmonicida trophozoites (Table 1). Western blotting was used to examine annexin expression at the protein level in stably transfected S. salmonicida trophozoites. Most of the annexins investigated (1 to 6, 8, 9, and 11 to 14), except annexins 7 and 10, were clearly detected as discrete proteins species with molecular weights consistent with in silico predicted values (see Fig. S6 in the supplemental material). Annexin 10 was detected by Western blotting but migrated as a slightly larger polypeptide than the in silico predicted size. The expression of annexin 7 was detected only when the gene was placed under the control of an exogenous promoter (see Fig. S6). The levels of protein expression by Western blotting did not correlate well with levels inferred by RNA-Seq (Table 1).

Immunolocalization of epitope-tagged annexins.
In light of the clear association of alpha-giardins with the cytoskeleton in Giardia (4), we were interested in obtaining localization data for the S. salmonicida annexins. We used the abovedescribed transfectant cell lines expressing C-terminally epitope-tagged constructs of all of the annexins in S. salmonicida and imaged them by either confocal laser scanning or structured illumination superresolution microscopy. The tagged annexins displayed a variety of localizations within the cell, and some were very specific in their localization (Fig. 4). Annexin 1 localized to the plasma membrane with a spotty appearance, and significant amounts of label also decorated the internal parts of the recurrent flagella or striated lamina around the recurrent flagella (Fig. 4A). Annexins 2 and 9 were found in numerous cytoplasmic foci ( Fig. 4B and G).  (Fig. 4D). Annexins 5 and 13 localized to a previously undescribed structure in the anterior part of the cell, and some label was also found in the plasma membrane ( Fig. 4E and I). In some cells, a smaller labeled structure is found in the middle portion of the cell body (see Fig. S8). In dividing cells, this smaller structure is translocated to the opposite side of the original structure to the anterior position of the newly forming daughter cell. Annexin 6 uniquely localize to eight, often paired, foci positioned in the anterior part of the cell in most trophozoites (Fig. 4F). Two of the foci are usually located between the nuclei in a position posterior to the other six foci. In cells about to divide, the number of foci is doubled. Annexin 10 seems to be localized to the plasma membrane, but the latter also displayed two fibrous signals that stretch through the cell body along the axis of the recurrent flagella (Fig. 4H). Annexin 7 was not expressed from its own promoter under laboratory conditions, and only upon expression from the annexin 3 promoter could we detect the tagged protein in the cytoplasm (Fig. 4L). Likewise, a mostly cytosolic localization of annexins 8, 11, and 14 was also seen (Fig. 4M, N, and J, respectively).

Ultrastructural localization of annexins with APEX. The function of annexins
could be easier to understand if they could be coupled to specific ultrastructural elements in the cell. The possibilities of using genetically encoded proximity labeling tags with broad functionality have recently been realized in several variants (24,25). We have adapted the APEX system (25) for use in diplomonads with DAB as a substrate to reveal the ultrastructural localization of a fusion partner. We constructed C- terminal  FIG 4 S. salmonicida annexin localization. Localization of 3xHA (A, B, D, G, H, and J to N) and V5 (C, E, F, and I)  APEX-V5 fusions of annexins 3 to 6, 10, and 13 to try to obtain a transect of the diverse set of localizations seen by immunolocalization (Fig. 4). We were unable to generate annexin 4 transfectants expressing the APEX-V5 fusion. For the remaining constructs, we studied the subcellular localization by immunofluorescent staining with the V5 epitope. Annexins 3, 5, 6, and 13 showed localizations identical to those of the hemagglutinin (HA)-tagged constructs (Fig. 4). However, annexin 10 was only sporadically expressed and cells with expression did not show localization to the recurrentflagellar region as in the HA-tagged construct.
Transfectant cells grown in hemin-supplemented LYI (liver digest, yeast extract, and iron) growth medium were washed and promptly fixed in glutaraldehyde in cacodylate buffer and reacted with DAB. At that stage, cells were viewed by phase-contrast microscopy (Fig. 5A, E, I, M, and Q). Annexin 3 displayed prominent dark flagella, with the internal part of the recurrent flagella clearly observable (Fig. 5A). Annexin 5 showed a strong curving label in the anterior part of the cell (Fig. 5E). Cells in the stages leading to cell division displayed a smaller label situated in the middle of the cell that progressively seemed to move closer to the anterior as cells grew in size before cell division (Fig. 5E). Annexin 6 exhibited six dark foci in the anterior of the cell, with cells close to cell division showing double the number of foci (Fig. 5I). The foci were often arranged in pairs. Annexin 13 showed labeling similar to that of annexin 5 but with much less intense staining (Fig. 5M).
Light microscopy revealed many details of the subcellular localization of the four annexins investigated, but we were interested in learning the ultrastructural fine details of their localization. Samples were reacted with osmium to make DAB precipitates visible and then prepared for TEM. For annexin 3, we observed increased contrast in the flagella, especially around the outer microtubule doublets of the 9 ϩ 2 axoneme (Fig. 5B to D; see Fig. S9A and B). Biradial staining was also noted in the striated lamina running the length of the recurrent flagella (Fig. 5B to D; see Fig. S9A and C). The annexin 5 transfectants exhibited strong staining in a large area in the anterior part of the cells (Fig. 5F and G; see Fig. S9D to F). DAB precipitates were also noted on either side of the cell membrane in the flagella, as well as the cell body (Fig. 5H). TEM of annexin 6-APEX revealed a signal in the vicinity of the basal bodies with prominent DAB staining in and around the centrioles (Fig. 5J to L). Subtle DAB staining of annexin 13-APEX was observed in the anterior of the cell in a location similar to that of annexin 5 ( Fig. 5N to P). Wild-type cells reacted with DAB showed a staining pattern similar to that of control cells where DAB had been omitted (Fig. 5Q to T). On some occasions, a DAB reaction product could be seen in the interior of microtubules.

DISCUSSION
In this study, we investigated members of the annexin E gene family in diplomonads, with an emphasis on S. salmonicida, by both phylogenetic analyses and experimental approaches. We also identified alpha-giardins from a second Giardia species, Giardia muris. The partition between alpha-giardins and annexins from Spironucleus and Trepomonas in the phylogenetic analysis suggests that either a single annexin gene was present in the ancestral diplomonad or the family has a high rate of gene turnover within diplomonads. The diversity of the annexin sequences observed in diplomonads is similar to, or even larger than, the diversity of annexin sequences from different eukaryotic supergroups (Fig. 3). It would be very interesting to investigate the occurrence of annexins in free-living relatives of diplomonads, especially since no annexin homologs from other members of the Excavata supergroup are known. The human genome encodes 12 annexins, whereas G. intestinalis and S. salmonicida have 19 and 14 different annexins, respectively. It is still not known why the annexins have been expanded in the diplomonads, but it is tempting to speculate that the less complex actin cytoskeletons of parasitic diplomonads have led to the expansion of annexins to replace this loss (26).
The Spironucleus annexins appear to be less derived than alpha-giardins, and most harbor at least one type II Ca 2ϩ coordination site. In accordance with this observation, we showed Ca 2ϩ -dependent interaction between two S. salmonicida annexins and acidic phospholipids, establishing that they are bona fide annexins. Moreover, the annexins displayed different binding preferences, in line with their low sequence identity and nonoverlapping cellular localizations. There might be a correlation between the number of intact type II sites and the strength of membrane association since annexins 3, 9, and 12 all have four canonical sites and are found in the membrane fraction. At the same time, the annexin with the most prominent membrane association, annexin 4, has only a single conserved repeat and presumably associates by a separate mechanism. Several annexins showed a potential for posttranslational modifications via acylation, a property previously described in the N-terminal alpha-19 giardin (21). Consequently, annexins 3 and 12 also localize to the flagellum, like alpha-19 giardin (see Fig. S7 in the supplemental material). The C terminus of annexin 10 carries an extension of 14 aa with two paired cysteine residues in the second-and third-last residues and shows a slight shift in molecular size from calculated values during SDS-PAGE (Table 1; see Fig. S6 in the supplemental material). The large APEX-V5 tag led to mislocalization of annexin 10, while the small 3xHA epitope tag appears to be tolerated. We speculate that this phenotype is due to the inability of annexin 10 to become acylated and attain its intended cellular location. Unfortunately, without TEM data, we were unable to determine if annexin 10 localizes to the recurrent flagella, internal membranes, or striated lamina (27).
Many of the strongly expressed annexins in S. salmonicida localize to the membrane or cytoskeleton, while more weakly expressed annexins often localize to the cytosol (Fig. 4). It is possible that these weakly expressed annexins play a more prominent role during other parts of the poorly described S. salmonicida life cycle. Two of the annexins, 5 and 13, were found to localize to a previously undescribed cellular organelle in the anterior end of the cell above the nuclei (Fig. 4E and I). The structure labeled by these two proteins is homogeneous and electrolucent in TEM of wild-type cells and corresponds to the region of the cell that appears to mediate attachment to various substrates. In the related flagellate Spironucleus torosa, which infects gadoid fish, this area of the cell has been seen to mediate intimate contact with the epithelium of the host (28). The S. salmonicida genome lacks signature proteins (beta-, delta-, and gamma-giardins; SALP-1; median body protein) of the Giardia ventral disc. On the other hand, at least three alpha-giardins have been found associated with this attachment organelle (29). It is possible that S. salmonicida has evolved an attachment organelle by co-opting homologous proteins such as those found in the Giardia ventral disc.
The localization of annexin 3 was also studied by APEX proximity labeling. Annexin 3 localizes prominently to the flagella, and DAB staining also labels the striated lamina that supports the internal portion of the recurrent flagella. Annexin 3 interacted with cardiolipin, PS, and PG, the first two of which are also binding partners of alpha-14 giardin, which is found in giardial flagella (10). Besides membrane interaction, alpha-14 giardin is known to interact with tubulin, has the capacity to self-assemble into multimeric structures, and forms local slubs in the flagella at the proximal and distal ends of the axoneme (30). Analogously, annexins 3 and 12 colocalize with tubulin in S. salmonicida by confocal microscopy, and proximity labeling suggests a close association with tubulin (see Fig. S7 in the supplemental material). Superresolution microscopy revealed that annexin 3 is also unevenly distributed in the flagella, reminiscent of the slubs observed in Giardia (30). The development of proximity labeling techniques using the APEX peroxidase and other substrates offers a promising avenue by which to investigate interaction partners. Future studies will be directed at unraveling the makeup of the novel anterior organelle and its putative role in attachment. Additionally, we will attempt to find interaction partners of annexins in S. salmonicida to learn more about their cellular roles.
The alpha-giardins in Giardia were the first annexins from a protozoan eukaryote to be described (31)(32)(33). Our study and previous studies of alpha-giardins (4,6,17,21,30,32,34,35) have provided insights into the functional roles of members of the annexin protein family in diplomonads in particular and eukaryotic cells in general. Furthermore, our bioinformatic search identified annexin homologs from a wide range of microbial eukaryotes representing all five major groups (Fig. 3) (23). Most of these are unicellular and should probably be classified as group E annexins according to the current annexin nomenclature (36). This is problematic since unicellular eukaryotes are not a monophyletic group and many of the identified annexins of protists (Fig. 3) are more closely related to, and probably share more functional features with, annexins in the A, B, C, or D group than the group E annexins studied here. We think that the annexin classification system needs to be reevaluated in light of the current knowledge of annexin distribution in eukaryotes and eukaryote phylogeny.
The phylogenetic analysis suggests that the annexin protein family probably was present early in eukaryote evolution (Fig. 3). However, annexins are not specific for eukaryotes. Thirty-four putative bacterial annexin proteins have been identified in 17 different bacterial species (37). Unfortunately, the bacterial annexin domains are divergent from the eukaryotic annexins and do not show affinity to any specific eukaryotic group. Therefore, the evolutionary origin of annexins in eukaryotes is impossible to pinpoint on the basis of the current data.
Additional studies of annexins from diverse protists would provide a deeper understanding of different functions of annexins. For example, annexin homologs are present in unicellular sister groups of animals such as Choanoflagellida and Ichthyosporea. Phylogenetic analysis (Fig. 3) suggests that multiple paralogs of annexins were indeed present in the last common ancestor of these protist groups and animals. Functional studies of annexins of these protists, similar to our study, would be very useful to understand the function of annexins of animals and humans. We believe that a combination of genomic and phylogenetic analyses and experimental approaches such as those used here in studies of new microbes from different parts of the tree of life will further reveal the function and evolutionary history of annexins in eukaryotes.
Epitope tagging vector construction and transfection. The coding sequences and 100 to 400 bp of the putative promoters of the 14 annexin genes were PCR amplified from genomic DNA of S. salmonicida (see Table S1 in the supplemental material). The PCR products were column purified, digested with appropriate restriction enzymes, and ligated into vector pSpiro-PAC-3xHA-C (14) or pSpiro-PAC-APEX-V5. The pSpiro-PAC-APEX-V5 vector was created by amplifying the pea APEX gene from pcDNA3 Connexin43-APEX (Addgene plasmid 44439) (25) with the APEX-F and APEX-V5-R primers and inserting it in place of the 3xHA tag at the NotI and ApaI sites of pSpiro-PAC-3xHA-C. The finished constructs were sequenced to confirm the sequence of the final construct. Transfection of S. salmonicida followed the method described previously (14).
Mem-PER extraction. Cells were membrane fractionated with the Mem-PER Eukaryotic Membrane protein extraction kit (Pierce 89826) by using the protocol for mammalian cells. Detergents were removed with the SDS-PAGE sample preparation kit (Thermo Scientific 89888) according to the manufacturer's recommendation. The protein concentrations were measured and normalized with Qubit. SDS-PAGE and Western blotting were performed as described below. Loading controls were generated with the TGX stain-free system (Bio-Rad) according to the manufacturer's recommendation.
The lipid-binding capacities of annexins 3 and 5, alpha-14 giardin, and GST were tested with membrane lipid strips (P-6002; Echelon Research Laboratories, Salt Lake City, UT). The membranes were blocked with 3% fatty-acid-free bovine serum albumin (A8806; Sigma) in phosphate-buffered saline-0.1% Tween 20 (pH 7.4) for 1 h. GST-tagged proteins (0.5 g/ml annexin 3 and alpha-14 giardin, 2 g/ml annexin 5 and GST) in blocking buffer were incubated with the membrane for 1 h. The PI(4,5)P 2 Grip binding protein (G-4501; Echelon Research Laboratories, Salt Lake City, UT), derived from the N-terminally GST-tagged recombinant PLC-␦1 PH domain, was used as a positive control and diluted to 0.5 g/ml according to the manufacturer's instructions. The membranes were incubated for 1 h with an HRPcoupled anti-GST antibody (K-SEC2; Echelon Research Laboratories, Salt Lake City, UT), and the bound protein was detected with the K-TMBP substrate (Echelon Research Laboratories, Salt Lake City, UT). Lipid-protein interactions were quantified by measuring the intensity of color compared to global background staining with the Quantity One v 4.6.6 software (Bio-Rad).