The most abundant cyst wall proteins of Acanthamoeba castellanii are three sets of lectins that bind cellulose and chitin and localize to distinct structures in cyst walls

Acanthamoeba castellanii, cause of keratitis and blindness, is an emerging pathogen because of its association with contact lens use. The cyst wall contributes to pathogenesis as cysts are resistant to sterilizing reagents in lens solutions and to antibiotics applied to the eye. We used transmission electron microscopy, as well as structured illumination microscopy and probes for cellulose and chitin, to show that purified cyst walls of A. castellanii retain an outer ectocyst layer, an inner endocyst layer, and conical ostioles that connect the layers. Mass spectrometry showed candidate cyst wall proteins are dominated by three families of lectins (named here Luke, Leo, and Jonah), each of which binds to microcrystalline cellulose and to a lesser degree chitin. A Jonah lectin, which has one choice-of-anchor A (CAA) domain, localizes to the ectocyst layer of mature cyst walls. Luke lectins, which have two or three carbohydrate-binding modules (CBM49), localize to the endocyst layer and ostioles. A Leo lectin, which has two domains with eight Cys residues each (8-Cys), also localizes to the endocyst layer and ostioles. In summary, the most abundant A. castellanii cyst wall proteins are three sets of lectins, which have carbohydrate-binding modules that are conserved (CBM49s of Luke), newly characterized (CAA of Jonah), or unique to Acanthamoebae (8-Cys of Leo). Despite their lack of common ancestry, Luke and Leo lectins both localize to the endocyst layer and ostioles, while the Jonah lectin localizes to the ectocyst layer. IMPORTANCE Fifty years ago, investigators identified cellulose in the Acanthamoeba cyst wall, which has an outer ectocyst layer, an inner endocyst layer, and conical ostioles that connect the layers. Here we show cyst walls also contain chitin and three large sets of cellulose- and chitin-binding lectins, which have distinct localizations. The Acanthamoeba cyst wall, therefore, is more complicated than cyst walls of Entamoeba and Giardia (causes of dysentery and diarrhea, respectively), which have a single layer, a single glycopolymer, and small sets of one or two lectins. In contrast, the Acanthamoeba cyst wall is far simpler than the walls of fungi and plants, which have multiple layers, numerous glycopolymers, and hundreds of proteins. In addition to providing a better understanding of the cell biology and biochemistry of the Acanthamoeba cyst wall, these studies may lead to diagnostic antibodies that bind to cyst wall proteins and/or therapeutics that target chitin.


Introduction
promoter in cyst walls of stably transfected protists (32, 33). We also determined whether 84 each CWP, which was made as a maltose-binding protein (MBP) in the periplasm of bacteria, 85 binds to microcrystalline cellulose and/or chitin beads (34).

87
TEM and SIM showed purified A. castellanii cyst walls contained distinct endocyst and 88 ectocyst layers, as well as ostioles. Cyst wall preparations were made by sonicating cysts 89 and separating walls from cellular contents by density centrifugation and retention on a 90 membrane with 8 µm-diameter pores. For TEM, intact cysts and purified cyst walls were 91 frozen under high pressure, and fixatives were infiltrated at low temperature (35). Purified cyst 92 walls had intact ectocyst and endocyst layers, as well as conical ostioles that link them ( Fig. 1)

93
(13). The purified walls were missing amorphous material that fills the space between the 94 inner aspect of the cyst wall and the plasma membrane of the trophozoite inside. 95 For SIM, we used probes that bind chitin (wheat germ agglutinin, WGA) and β-1,3 and 96 β-1,4 polysaccharides (calcofluor white, CFW) in the walls of fungi and cysts of Entamoeba 97 (36-39). CFW, a fluorescent brightener, has also been used to diagnose Acanthamoeba cysts 98 in eye infections (40). In addition, we made a glutathione-S-transferase (GST) fusion-protein, 99 which contains the N-terminal CBM49 of a candidate CWP of A. castellanii ( Fig. S1 and Excel 100 file S1) (41). The GST-AcCBM49 expression construct was designed to replicate that used to 101 determine the carbohydrate binding properties of SlCBM49, which is a C-terminal SlCBM49 from tomato. A fourth conserved Trp is present in the CBM49 of D. discoideum 129 cellulose-binding proteins (46). The other CBM49s (middle and/or C-terminal) of Luke lectins 130 had two conserved Trp residues. Luke lectins were acidic (pI 5 to 6) and had formula weights  (Table 1 and Excel file S1). In general, Luke lectins with two 136 CBM49s had more unique peptides than Luke lectins with three CBM49s. One to four unique 137 peptides were derived from three CBM49-metalloprotease fusion-proteins, which consisted of 138 an N-terminal signal peptide followed by a single CBM49 with four conserved Trp residues and 139 a metalloprotease (ADAM/reprolysin subtype) with a conserved catalytic domain       These results showed that ~500 bp of the 5' UTR was sufficient to cause encystation-specific three CBM49s localized to the same place. 243 We think the timing of expression and locations of the GFP-tagged CWPs in cyst walls 244 were accurate for the following reasons. First, RT-PCR showed that mRNAs of representative 245 Luke, Leo, and Jonah lectins, as well as cellulose synthase (ACA1_349650), were absent or 246 nearly absent from trophozoites but were present during the first three days of encystation 247 (Fig. S3). In contrast, glyceraldehyde 3-phosphate dehydrogenase (GAPDH), which catalyzes to Western blots of proteins from cysts but not from trophozoites (Fig. S4). We were unable to 252 generate rabbit antibodies to the Luke lectin, using methods that worked to make antibodies to     (44). The Luke lectin binds cellulose and chitin, while the D. 310 discoideum proteins with a single CBM49 bind cellulose (46). Chitin-binding by DdCBM49 or 311 SlCBM49 was not tested, because this glycopolymer is not present in D. discoideum and 312 tomato walls. Demonstration that CBM49s of the Luke lectin also bind chitin fibrils is new, but 313 is consistent with recent studies showing CBMs may bind more than one glycopolymer (67).

314
The metalloprotease fused to an N-terminal CBM49 of A. castellanii is absent in D. 315 discoideum, while D. discoideum adds two CBM49s to a cysteine proteinase, which lacks 316 these domains in A. castellanii. The CBM49 may act to localize the metalloproteases to the A.      Glycosylphosphatidylinositol anchors were searched for using big-PI (47). AmoebaDB, which 465 contains sequence information from the Neff strain and ten other Acanthamoeba strains, was 466 used to identify genome sequences, predict introns, and identify paralogous proteins (28, 29).

467
The NR database at the NCBI was used to identify homologs of candidate CWPs in other  Table 1). The GFP tag was placed at the C-   Coomassie stained gels were run in parallel for loading control.         (Fig. 3). A Leo(TKH) lectin has a signal peptide, two domains containing eight Cys residues each (red Cs), and a long Thr-, Lys-, and His-rich spacer (brown). A Jonah(3) lectin has three CAA domains (green), hydrophobic regions (tan), and short Ser-and Pro-rich spacers (light blue).
FIG S3 RT-PCR shows mRNAs of representative Luke, Leo, and Jonah lectins, as well as those of cellulose synthase, are encystation-specific. DNA and total RNA were extracted from trophozoites and organisms encysting for one to three days. RT-PCRs were performed with primers specific for segments of each CWP mRNA, as well as primers specific for segments of mRNAs for GAPDH and cellulose synthase (see Supplemental Table 1). PCR with DNA was used as a positive control, while omission of reverse-transcriptase (-RT) was used as a negative control. Messenger RNAs encoding CWPs and cellulose synthase were absent or nearly absent in trophozoites but were easily detectable in encysting organisms. In contrast, mRNAs for GAPDH were expressed by both trophozoites and encysting organisms (33).

FIG S4
Western blots with rabbit antibodies to peptides of Jonah and Leo lectins show each CWP is absent in trophozoites but is easily detected in mature cysts. A. Coomassie blue stain of proteins of lysed trophozoites and cysts, as well as molecular weight standards (M). B. Western blotting showed rabbit antibodies to a 50-amino acid peptide of a representative Jonah(1) lectin (underlined in Fig. 3) bound to a cyst protein of the predicted size (red underline) and to an MBP-Jonah(1) fusion-protein made in the periplasm of bacteria. The antibody also bound to degradation products of Jonah (1) lectin. In contrast, the anti-Jonah(1) antibody did not bind to either trophozoites or MBP alone (negative controls). C. Rabbit antibodies to a 16-amino acid peptide of a representative Leo lectin also bound to cyst proteins and to an MBP-Leo fusion but not to trophozoite proteins or to MBP alone. In addition to Leo of the predicted size (red underline), anti-Leo antibodies to a higher molecular weight form, which may be a dimer. These results confirmed encystation-specific expression of Jonah(1) and Leo lectins (Figs. 4 and S3). None of the rabbit antibodies the CWP peptides were useful for labeling cyst walls for widefield microscopy or SIM.

FIG S5.
Control GFP constructs localize to the cytosol (CSP21-GFP) and secretory vesicles (GFP with an N-terminal signal peptide, SP-GFP) of mature cysts. A. The 21-kDa cyst-specific protein (CSP21) fused to GFP was absent in trophozoites (not shown) but formed punctate structures in the cytosol of cysts (22). B. GFP with an N-terminal signal peptide from Luke lectin and expressed under a GAPDH promoter localized to secretory vesicles of mature cysts (33). These controls make it unlikely that localizations of CWP-GFP constructs to the cyst wall were artifacts (Fig. 4).

FIG S6
Contrasting uses of CBM49 by A. castellanii and D. discoideum. CBM49, which was first shown to be a cellulose-binding domain at the C-terminus of tomato cellulase, is repeated two or three times in Luke lectins of A. castellanii and is also present at the N-terminus of a metalloprotease. In contrast, CBM49 is present in a single copy in the majority of D. discoideum proteins and as three copies in rare proteins. CBM49 is also present in two copies in a D. discoideum cysteine protease and as a single copy in a GH5 glycoside hydrolase. A. castellanii Leo lectins and E. histolytica Jacob lectins have common structures, even though they share no common ancestry (convergent evolution). Abundant cyst wall proteins of A. castellanii (Leo lectins) and E. histolytica (Jacob lectins) have unique 8-Cys or 6-Cys domains, respectively, that bind cellulose or chitin. In each protist, some of the lectins lack spacers, while others have spacers rich in Thr, Lys, and His (A. castellanii) or Ser and Thr (E. histolytica).

FIG S8
Contrasting use of choice of anchor A (CAA) domains in A. castellanii, oomycetes, and bacteria. Jonah lectins, which are abundant in cyst walls of A. castellanii, have one or three CAA domains. The former are preceded by Thr-, Lys-, and Cys-rich sequences (gray), while the latter are separated by Ser-and Pro-rich spacers (blue) and hydrophobic domains (tan). Predicted proteins of oomycetes (Pyromyces or Neocallimastix) have three to five CAA domains, while the spore coat protein of Bacilllus has a single CAA domain attached to a collagen-binding domain, which is absent in A. castellanii.
Excel file S1 This file lists the most abundant proteins identified by mass spectrometry of cyst walls purified on the Percoll gradient and retained on a membrane containing 8-µm pores. Proteins with <7 unique peptides have been removed, because they are dominated by cytosolic contaminants. Luke, Leo, and Jonah lectins have been highlighted in orange. Other candidate CWPs are marked in yellow.
Excel file S2 Complete list of proteins identified by mass spectrometry of cyst walls. This list includes a preparation that was heavily contaminated with cytosolic proteins, because the Percoll gradient and porous membrane were omitted during their purification. Only proteins with at least two unique peptides are included.