Giardia duodenalis Surface Cysteine Proteases Induce Cleavage of the Intestinal Epithelial Cytoskeletal Protein Villin via Myosin Light Chain Kinase

Giardia duodenalis infections are among the most common causes of waterborne diarrhoeal disease worldwide. At the height of infection, G. duodenalis trophozoites induce multiple pathophysiological processes within intestinal epithelial cells that contribute to the development of diarrhoeal disease. To date, our understanding of pathophysiological processes in giardiasis remains incompletely understood. The present study reveals a previously unappreciated role for G. duodenalis cathepsin cysteine proteases in intestinal epithelial pathophysiological processes that occur during giardiasis. Experiments first established that Giardia trophozoites indeed produce cathepsin B and L in strain-dependent fashion. Co-incubation of G. duodenalis with human enterocytes enhanced cathepsin production by Assemblage A (NF and S2 isolates) trophozoites, but not when epithelial cells were exposed to Assemblage B (GSM isolate) trophozoites. Direct contact between G. duodenalis parasites and human intestinal epithelial monolayers resulted in the degradation and redistribution of the intestinal epithelial cytoskeletal protein villin; these effects were abolished when parasite cathepsin cysteine proteases were inhibited. Interestingly, inhibition of parasite proteases did not prevent degradation of the intestinal tight junction-associated protein zonula occludens 1 (ZO-1), suggesting that G. duodenalis induces multiple pathophysiological processes within intestinal epithelial cells. Finally, this study demonstrates that G. duodenalis-mediated disruption of villin is, at least, in part dependent on activation of myosin light chain kinase (MLCK). Taken together, this study indicates a novel role for parasite cathepsin cysteine proteases in the pathophysiology of G. duodenalis infections.

Villin is a unique cytoskeletal protein expressed in gastrointestinal, renal, and urogenital epithelial cells. Within the gut, villin is primarily expressed in the microvilli of IEC where it provides an essential function in regulating the organization of epithelial brush border microfilaments during periods of physiological stress [41]. This protein is capable of polymerizing and depolymerizing actin, via its ability to cap, sever, nucleate and bundle actin filaments [42]. Giardia is known to shorten epithelial brush border microvilli, thereby contributing to malabsorptive diarrhea [38,43,44]. Research has also demonstrated that villin is a pro-survival factor, as its overexpression results in delayed epithelial apoptosis in vitro, and its deletion in vivo enhances susceptibility to dextran sodium sulfate (DSS)induced colitis due to increased rates of IEC apoptosis [45]. In its pathogenic cascade, Giardia has been found to cause epithelial apoptosis [38,46]. In addition, villin has been shown to participate in IEC migration and wound healing [42,[47][48][49][50]. Collectively, these results demonstrate that villin is a critical homeostatic protein with multiple functions that extend beyond the maintenance of IEC microvilli. The multifunctional purpose of this protein, in addition to its close relationship with F-actin, warrants further research on its interaction with G. duodenalis. Indeed, previous research has demonstrated that G. duodenalis trophozoites disrupt intestinal epithelial F-actin [39]. Moreover, remodelling of intestinal epithelial villin has been observed during late stages of G. duodenalis infection in vivo via CD4+ and CD8+ immune responses [51]. However, it remains to be determined whether G. duodenalis parasite products directly target villin. The objectives of this study were to characterize cysteine protease activity in G. duodenalis trophozoites during co-incubations with IECs, and to determine whether there is a role for these proteases in intestinal epithelial cytoskeletal villin breakdown and disruption.

Giardia duodenalis trophozoite isolation
Confluent tubes of G. duodenalis trophozoites were collected by cold shock on ice for 30 min. Following this, 15ml tubes were pooled into 50 ml polypropylene tubes, centrifuged at 500g for 10 min at 4°C, and resulting supernatants were aspirated. The pellets were resuspended in 10ml of chilled sterile 1X phosphate buffered saline (PBS) (Sigma-Aldrich) and centrifuged at 500g for 10 min at 4°C. The resulting pellet was resuspended in Caco-2 growth media, the trophozoites enumerated using a hemocytometer, and their concentration was adjusted to a multiplicity of infection (MOI) of 10 parasites per 1 host cell (MOI of 10:1).

Giardia duodenalis trophozoite DNA extraction
In preparation for DNA extraction, trophozoites were harvested by placing culture tubes on ice for 10 min. Tubes were then gently inverted to mix the contents, and 2 ml of each suspension were pipetted into 15 ml centrifuge tubes (Falcon), and PBS pH 7.4 was added to a final volume of 10 ml. Tubes were centrifuged at 1,000g for 5 min. The supernatants were decanted and the pellets were washed twice more by repeating the centrifugation. Trophozoites were suspended in a final volume of 1 ml PBS. Total DNA was extracted from the trophozoites using the DNeasy Tissue Kit (Qiagen Inc., Mississauga, ON), using a modified protocol. Two hundred microliters of the suspended trophozoites were transferred to 1.5 ml microcentrifuge tubes, and lysed overnight at 56°C using 180 μl of lysis buffer and 20 μl of proteinase K (20 mg/ml) supplied with the DNeasy Tissue Kit. The manufacturer's instructions were then followed to purify the DNA. Nucleic acid was eluted with 100 μl of elution buffer. Positive control, G. duodenalis cysts (Waterborne, Inc., New Orleans, LA), and negative control, DNase-free water (Sigma-Aldrich Canada Co., Oakville, ON), were also extracted in parallel.

PCR Protocols
A nested-PCR was performed to amplify fragments of the Giardia 16S rRNA gene as described [62]. PCR amplification of fragments of the glutamate dehydrogenase (gdh) gene for Giardia, and a restriction fragment length polymorphism (RFLP) assay for genotyping, was performed as described [63].

DNA Sequencing
DNA sequencing of the products of both 16S rRNA and gdh PCR was performed at the McGill University and Genome Quebec Innovation Centre in Montreal, QC, using a 3730xl DNA Analyser (Applied Biosystems, Foster City, CA). PCR products were purified and bi-directional sequencing was performed using the same primers as the original amplifications. DNA sequences were assembled, edited and aligned using SeqScape v2.5 (ABI). Resulting consensus sequences were then aligned with representative sequence data from G. duodenalis Assemblages and trimmed to identical lengths of 189 bp or 367 bp for 16S rRNA and gdh genes, respectively, using Bioedit v7.1.3.0 [64]. Consensus sequences were compared to reference sequences in GenBank using NCBI standard nucleotide BLAST (blastn). Phylogenetic trees were generated using Molecular Evolutionary Genetics Analysis (MEGA v5.2.1) (http://www. megasoftware.net/) with a Kimura two parameter model neighbour-joining analysis, with 100 bootstraps and pair-wise deletion.

Giardia duodenalis trophozoite viability
Motility of G. duodenalis trophozoites was used to assess parasite viability after 2-or 24-hour co-incubation. Subsequently, Caco-2 cell supernatants were collected, vortexed, and 10 μL of cell supernatant was analyzed on a hemocytometer. The ratio of motile, or swimming, trophozoites to total trophozoite counts was assessed as a marker of viability [65].

Giardia duodenalis modulation of intestinal epithelial villin
G. duodenalis trophozoites (NF, S2, or GS/M) were co-incubated with confluent Caco-2 monolayers at an MOI of 10:1 for 2 or 24 hours. Identical experiments were performed, except trophozoites were initially pre-treated with E-64d (10 μM), Ca-074Me (10 μM), or vehicle control (dimethyl sulfoxide; DMSO) 3 hours prior (see below) to co-incubation with in vitro Caco-2 monolayers. Similarly, Caco-2 monolayers were pre-treated with one of E-64d (10 μM), Ca-074Me (10 μM), ML-9 (40 μM), Z-DEVD-FMK (50 μM), or vehicle control (DMSO) prior to co-incubation with G. duodenalis trophozoites for 2 or 24 hours. In complimentary experiments, G. duodenalis NF trophozoites were co-incubated with Caco-2 monolayers grown to confluence on the bottom of 12-well plates; in this instance, G. duodenalis trophozoites were seeded into the top compartment of 0.4 μm transwells to prevent direct contact between in vitro monolayers and parasites. Incubation conditions were maintained at 37°C, 5% CO 2 , and 96% humidity for experimental duration. At the end of the incubation period, cell supernatants were collected and processed for cathepsin activity assays (see below), while monolayers were washed once with PBS and subsequently processed for Western blotting analysis (see below).
In parallel experiments, Caco-2 monolayer cellular lysates were co-incubated with G. duodenalis NF trophozoite sonicates for 2 hours. Briefly, culture media from confluent Caco-2 monolayers was removed, and cells were washed once with PBS. Following this, the PBS was aspirated, the Caco-2 monolayers were collected in 200μL protease inhibitor-free radioimmunoprecipitation assay (RIPA) buffer (1% Igepal, 0.1% SDS, and 0.5% sodium deoxycholate in PBS) in microcentrifuge tubes, and sonicated at level 3 for 5 seconds (550 Versonic Dismembranator, Fisher Scientific). Caco-2 lysates protein concentrations were adjusted to 3mg/ml using the Bradford assay method (BioRad Laboratories, Hercules, CA). At the same time, G. duodenalis NF trophozoites were adjusted to a concentration of 1x10 7 trophozoites/ml and sonicated three times at level 4 for 30 seconds on ice (550 Versonic Dismembranator, Fisher Scientific). The resulting Caco-2 lysates and G. duodenalis trophozoite sonicates were co-incubated for 2 hours in the presence of 10 mM DTT and in the presence or absence of E-64 (200 μM) or Ca-074Me (200 μM). After 2 hours, samples were processed for Western blotting analysis (see below).

Whole cell protein extraction for Western blotting
At the end of the co-incubation for 2 or 24 hours, trophozoites were removed from the co-culture via three ice-cold PBS washes. Caco-2 monolayers were subsequently collected in RIPA buffer supplemented with a protease inhibitor cocktail tablet (Complete-Mini, Roche Diagnostics, Laval, QC). After a 30 min incubation at 4°C, cellular lysates were sonicated at level 3 for 5 seconds and centrifuged at 10,000g for 10 min at 4°C. The resulting supernatant was collected and protein concentrations determined using the Bradford assay method [66]. Protein samples were then normalized to 1.0 mg/ml and then combined at a 1:1 ratio with 2X electrophoresis buffer (100 mM Tris-HCl, pH 6.8, 4% sodium dodecyl sulfate (SDS), 0.2% bromophenol blue, 20% glycerol, 200 mM β-mercaptoethanol) to further dilute the samples to a final concentration of 0.5 mg/ml. The samples were then denatured at 95°C for 3 min and stored at-20°C until further analysis.

Western blotting
Protein samples were separated via SDS-PAGE (7-12%) and transferred to nitrocellulose membranes (Whatman, Buckinghamshire, England) over 1 hour at 100V. The membranes were blocked using 5% fat-free milk solution in 1X Tris-buffered saline + 0.1% Tween (TBS-T) for 1 hour. Primary antibodies were diluted in the same 5% milk solution and incubated with the membranes overnight at 4°C. After three 15-min washes with TBS-T, HRP-conjugated secondary antibodies, also diluted in 5% milk solution, were added to the membranes for 1 hour at room temperature. The membranes were washed three times with TBS-T for 15 min each and then visualized using ECL-plus chemifluorescence detection system (GE Healthcare, Pittsburgh, PA). ECL-plus was added the membranes for 5 min. The membranes were visualized on ECL film (GE Healthcare). The films were scanned for densitometric analysis using the software ImageJ (http://rsbweb.nih.gov/ij/). The membranes were stripped using 0.5M acetic acid (45 min incubation) and 0.2 M sodium hydroxide (5 min incubation), and re-probed with a GAPDH antibody to ensure equal loading of the gels. In all instances, GAPDH was used as the loading control.

Inhibition of G. duodenalis cysteine proteases
Previous research has demonstrated that concentrations of 10μM E-64d or Ca-074Me do not affect G. duodenalis trophozoite viability, but significantly reduce cathepsin-like cysteine protease activity [34,67]. Therefore, experiments involving G. duodenalis trophozoite and Caco-2 monolayer co-incubations were repeated with the administration of either 10μM E-64d, 10 μM Ca-074Me, or vehicle control (DMSO) for 2 or 24 hours. Following this, samples were collected and processed for cathepsin activity assays and Western blotting analysis. In separate experiments, confluent tubes of G. duodenalis trophozoites were treated with E64d (10μM), Ca-074Me (10μM), or vehicle control (DMSO) for 3 hours. Following this, G. duodenalis trophozoites were harvested by cold shock on ice for 30 min and then co-incubated with Caco-2 monolayers for 2 or 24 hours. Samples were again collected for cathepsin activity assays and Western blotting analysis.

Sample extraction for cathepsin activity assays
After co-incubation of G. duodenalis trophozoites with Caco-2 monolayers (as described above), supernatants, parasites, and IECs were collected and processed for cathepsin activity assays. Cell supernatants were collected, centrifuged at 500g, 4°C for 10 min and stored for further use. The supernatant pellet, representing G. duodenalis trophozoites, was resuspended in ice-cold PBS, centrifuged at 500g, 4°C for 10 min, and the trophozoite pellet re-suspended in RIPA buffer; this fraction was then sonicated at level 4 for 30 seconds on ice and centrifuged at 10,000g, 4°C for 10 min. A Bradford assay was performed and samples normalized to 1.0 mg/ ml. Adherent G. duodenalis trophozoites were removed from Caco-2 monolayers by modifying a previously described protocol [68]. In short, a sterile 10μM formononetin solution was made in the Caco-2 growth medium, added to Caco-2 monolayers, and then allowed to incubate at 37°C, 5% CO 2 for 60 min. Following this, the formononetin solution was aspirated, and monolayers were washed with ice-cold PBS three times. Monolayers were then collected and lysed in RIPA buffer not containing a protease inhibitor table, sonicated on level 3 for 5 seconds, and then centrifuged for 10 min at 10,000g at 4°C. A Bradford assay was performed on collected Caco-2 lysates and samples were then normalized to 3.0 mg/ml. All samples were stored at-70°C until further use.

Cathepsin activity assays
Assessment of cathepsin cysteine protease activity was performed via recording the liberation of 7-aminomehtylcoumarin (AMC) from fluorogenic substrates, whereby cathepsin protease activity correlates to an increase in detectable relative light units (RFUs) over time [53,54]. To assess supernatant cathepsin cysteine protease activity, samples were thawed and incubated at a 1:2 ratio with cathepsin assay buffer (100 mM sodium acetate, 10 mM DTT, 0.1% Triton X-100, 1 mM EDTA, 0.5% DMSO, 200 μM ZFR-AMC or ZRR-AMC). Assessment of intra-trophozoite or intracellular Caco-2 cathepsin cysteine protease activity involved incubating samples at a 1:3 ratio with cathepsin assay buffer. All samples were incubated in 96-well clear, flatbottom plates and at 37°C for 5 min and subsequently measured kinetically using a microplate reader (SpectraMax M2 e , Molecular Devices, Sunnyvale, CA) at 37°C with excitation and emission wavelengths of 354nm and 445nm, respectively. Measurements were recorded every 30 seconds for 5 min. For all experiments, cathepsin assay buffer was adjusted to a pH of 7.2 to mimic the luminal pH of the upper small intestine.

Immunofluorescence
Caco-2 monolayers grown to confluence in Lab-Tek chamber slides (Nalgene Nunc International, Naperville, IL) (8-10 days) were co-incubated for 2 or 24 hours with 1.0x10 7 G. duodenalis NF trophozoites pretreated for 3 hours with E64d (10uM) or vehicle control (dimethyl sulfoxide; DMSO) at 37°C, 5% CO 2 . In another set of experiments, Caco-2 monolayers were pre-treated with ML-9 (40μM) for 30 min prior to co-incubation with G. duodenalis NF trophozoites for 2 or 24 hours. At the end of the co-incubation, the infection medium was aspirated off out of each chamber, followed by two ice-cold PBS washes. To fix/permeabilize the cells, ice-cold methanol was then added to each well and the chamber slides were subsequently incubated at 4°C for 30 min. After two more PBS washes, the cells were blocked with heat inactivated-fetal bovine serum (HI-FBS; VWR) for 15 min at room temperature. Primary antibodies were prepared in a solution of 2% FBS-PBS at appropriate concentrations that was subsequently administered to cells for 1 hour at 37°C. Following two PBS washes, the cells were incubated with fluorescent secondary antibodies (also prepared in 2% FBS-PBS) for 1 hour at 37°C. The cells were subsequently washed twice with PBS. Nuclei were counterstained with 1μM Hoechst fluorescent staining 33258 (Invitrogen Life Technologies) for 30 min at 37°C. The cells were washed once with PBS and the slides were mounted with Aqua Poly/Mount (Polyscience Inc.; Warrington, PA). Micrographs were obtained using a Leica DMR Microscope, with a Retiga 2000R (Q Imaging, BC) at 400x. All images were collected with the same gain and exposure time lengths. Micrographs presented in the Results section are representative images of 2 replicate monolayers from three separate experiments.

Statistical analysis
All data are representative of at least three separate experiments and expressed as means ± SEM, where applicable. All statistical analyses were performed using the software, Graphpad Prism 6, which ensures normality of data prior to analysis. Comparisons between groups were made using a Student t-test or one-way ANOVA, followed by Tukey's test for multiple comparison analysis. Statistical significance was established at p<0.05.

Results
Giardia duodenalis NF and S2 trophozoites are Assemblage A isolates PCR-RFLP of the glutamate dehydrogenase (gdh) gene revealed identical banding patterns for both NF and S2 G. dudoenalis isolates, matching the expected banding pattern for G. duodenalis Assemblage A (Fig 1). DNA sequence data on both 16S rRNA and gdh PCR products

Giardia duodenalis trophozoites contain and release catB and L cysteine proteases
Previous research has demonstrated that G. duodenalis trophozoites contain, as well as release, cathepsin cysteine proteases into cell supernatants [28][29][30][31]34]. However, the kinetics of cathepsin cysteine protease activity within multiple G. duodenalis trophozoites, IECs, or cell supernatants during co-incubation of parasites and IECs remain unknown. Therefore, cathepsin cysteine protease activity levels were assessed within parasites, IECs, and cell supernatants, following co-incubation. Cathepsin cysteine protease activity was determined by calculating the slope from the RFU versus time, as illustrated in Fig 3A. Figs 3B, 3C, 4B, 4C, 5B and 5C illustrate these calculated slopes and are shown as histograms. As demonstrated by an increase in the number of detectable RFUs over time, hydrolysis of the catB and L fluorogenic substrate ZFR-AMC occurred in G. duodenalis NF trophozoite sonicates alone (Fig 3A), or cell supernatants from co-incubation (Fig 4A) with Caco-2 monolayers; in contrast, hydrolysis was not observed inside the Caco-2 cells incubated in the presence or absence of G. duodenalis NF trophozoites (Fig 5A).   Slope values were determined when G. duodenalis trophozoites (NF, S2, or GS/M) were coincubated in Caco-2 growth media in the presence or absence of Caco-2 monolayers. Hydrolysis of the catB/L substrate ZFR-AMC ( Fig 3B) and catB-selective substrate ZRR-AMC (Fig 3C) was observed in sonicates from tested G. duodenalis isolates (NF, S2, or GS/M) and values were not statistically significant from each other (Fig 3B and 3C). A significant increase in ZFR-AMC ( Fig 4B) and ZRR-AMC (Fig 4C) hydrolysis was detected in cell supernatants collected from G. duodenalis trophozoites incubated Caco-2 growth media in the presence or absence of Caco-2 monolayers, when compared against control groups. Interestingly, co-incubation of G. duodenalis NF or S2 trophozoites with Caco-2 monolayers significantly increased cell supernatant hydrolysis of ZFR-AMC, compared against the same isolate incubated in the absence of Caco-2 cells (Fig 4B). This trend was not observed when hydrolysis of ZRR-AMC was analyzed (Fig 4C). Hydrolysis of ZFR-AMC ( Fig 5B) and ZRR-AMC (Fig 5C) was not significantly increased in Caco-2 monolayers incubated in the presence or absence of G. duodenalis trophozoites. Therefore, G. duodenalis trophozoites do not appear to increase catB/L activity inside Caco-2 monolayers. These results demonstrate that catB/L cysteine protease activities are active at similar levels within G. duodenalis trophozoites, and that parasites release cysteine proteases into cell supernatants. However, this activity was further increased in two Assemblage A isolates (NF and S2) following their exposure to Caco-2 monolayers, while this further increase could not be detected when cells were co-incubated with the Assemblage B GS/M. G. duodenalis NF trophozoites were used for the rest of the study to assess whether and how parasite catB/L proteases may affect host IECs. Initial experiments sought to determine whether catB/L activity could be inhibited within G. duodenalis trophozoites via the broadspectrum clan CA cysteine protease inhibitor E64d (10 μM) or the catB-specific inhibitor Ca-074Me (10 μM). Parasite viability was not affected by 3-hour treatment with E-64d or Ca-074Me; compared against control G. duodenalis trophozoites, no significant difference in the proportion of motile to non-motile trophozoites in groups treated with E64d, Ca-074Me, or vehicle control (DMSO) was observed ( Fig 6A). Next, Caco-2 monolayers were co-incubated with G. duodenalis NF trophozoites pretreated with E64d for 2 or 24 hours; this treatment significantly reduced the hydrolysis of ZFR-AMC within G. duodenalis sonicates (Fig 6B) and cell supernatants (Fig 6C), following their 2-hour co-incubation with Caco-2 monolayers. Similarly, hydrolysis of ZFR-AMC was significantly reduced within G duodenalis sonicates (Fig 6D) and cell supernatants (Fig 6E) when parasites were pre-treated with E-64d or Ca-074Me for 3-hours and subsequently co-incubated with Caco-2 monolayers for 24 hours. Collectively, these results demonstrate that G. duodenalis catB and L proteases are sensitive to inhibition using commercial protease inhibitors.

Giardia duodenalis cathepsin cysteine proteases promote villin breakdown in a contact-dependent manner
Recent research has focused on the ability of G. duodenalis to disrupt tight junctional proteins [37,38,40,69]. Less research has focused on this parasite's ability to affect intestinal epithelial cytoskeletal proteins. G. duodenalis disrupts cytoskeletal actin filaments in a myosin light chain kinase (MLCK)-dependent manner [39]. Moreover, host CD4+ and CD8+ T-lymphocytes are responsible for villin cleavage during late-stage G. duodenalis infection in vivo [51]. It is not known whether G. duodenalis parasite products are capable of directly targeting intestinal epithelial cytoskeletal proteins such as villin. As previous findings have shown that Entamoeba histolytica proteolytically degrades villin using cysteine proteases [70], we hypothesized that G. duodenalis cathepsin-like cysteine proteases was able to degrade and disrupt villin.
Co-incubation of G. duodenalis NF trophozoites with Caco-2 monolayers for 2 hours yielded a~45kDa cleavage product as determined by Western blotting (Fig 7A and 7B). Fulllength villin (90kDa) was not altered, as determined via densitometry (Fig 7B). Similar results were observed when parasites and Caco-2 monolayers were co-incubated for 24 hours as illustrated from Western blotting and densitometric analyses (Fig 7C and 7D). Densitometry also demonstrated that E-64d pre-treatment of G. duodenalis NF trophozoites prior to co- incubation with monolayers resulted in decreased detection of the villin cleavage product at both 2 ( Fig 7B) and 24 (Fig 7D) hours. Follow-up experiments were performed to determine whether co-incubation of parasite sonicates with Caco-2 cellular lysates for 2 hours also resulted in villin cleavage. As demonstrated via Western blotting and densitometry, co-incubation of G. duodenalis trophozoite sonicates with Caco-2 cellular lysates resulted in increased detection of villin cleavage fragments (Fig 8A and 8B). Importantly, this was significantly reversed when experiments were performed in the presence of E-64d (Fig 8A and 8B). These results suggest that G. duodenalis parasite products are capable of cleaving villin within intestinal epithelial cellular lysates. Similarly, immunofluorescent staining indicated that co-incubation of G. duodenalis trophozoites with Caco-2 monolayers resulted in redistribution of villin protein at 2 ( Fig 9A) and 24 (Fig 9B) hours. Importantly, pre-treatment of G. duodenalis trophozoites with E-64d, at least partially, prevented redistribution of villin within IECs at 2 and 24 hours (Fig 9A and 9B).
To determine whether secreted products were sufficient to cause a villin break down, parasites and Caco-2 monolayers were separated by 0.4μm transwells and co-incubated for 24 hours; interestingly, Western blotting and subsequent densitometry indicated that villin breakdown was not seen in these groups (Fig 10A and 10B). Together, these results demonstrate that G. duodenalis clan CA cysteine proteases are involved in the cleavage of cytoskeletal villin in Caco-2 monolayers. Moreover, these data suggest that a parasite surface clan CA cysteine protease contributes to villin cleavage and redistribution within Caco-2 monolayers.  Giardia duodenalis clan CA cysteine proteases do not contribute to ZO-1 breakdown Previous research has demonstrated that G. duodenalis trophozoites promote the breakdown and redistribution of the tight junction-associated protein zonula occludens 1 (ZO-1) [37][38][39]69]. Therefore, we decided to assess whether G. duodenalis clan CA cysteine proteases affect ZO-1 in a manner similar to villin. As determined via Western blotting and corresponding densitometry, co-incubation of G. duodenalis trophozoites with Caco-2 monolayers resulted in loss of full-length ZO-1 protein at 2 (Fig 11A and 11B) and 24 (Fig 11C and 11D) hours. However, pre-treatment of G. duodenalis trophozoites with E-64d, and subsequent incubation with Caco-2 monolayers, did not affect degradation of ZO-1 at 2 (Fig 11A and 11B) or 24 hours ( Fig  11C and 11D). These observations suggest that degradation and redistribution of villin and ZO-1 may occur via separate mechanisms.

Giardia duodenalis-induced villin breakdown is partially dependent on myosin light chain kinase
As Giardia is known to disrupt tight junctional proteins by activating MLCK and caspase-3 [38,39]), we decided to assess whether G. duodenalis clan CA cysteine proteases induce villin disruption via activation of MLCK or caspase-3. Therefore, experiments were performed whereby Caco-2 monolayers were pre-treated with the MLCK-selective inhibitor ML-9 or the caspase-3-specific inhibitor Z-DEVD-FMK prior to co-incubation with G. duodenalis trophozoites for 2 or 24 hours. Pre-treatment of Caco-2 monolayers with ML-9 prior to coincubation with G. duodenalis trophozoites for 2 hours did not affect villin cleavage, as determined via Western blotting analysis (Fig 12A and 12B). Interestingly, villin degradation was significantly reduced in Caco-2 monolayers pre-treated with ML-9 prior to co-incubation with G. duodenalis trophozoites for 24 hours (Fig 12C and 12D). No significant difference in villin cleavage was observed when Caco-2 monolayers were pre-treated with Z-DEVD-FMK and coincubated with G. duodenalis trophozoites for 2 (Fig 13A and 13B) or 24 hours (Fig 13C and  13D). Collectively, these results demonstrate that G. duodenalis-mediated villin degradation and redistribution is, at least partially, dependent on MLCK activation.

Discussion
Results from this study reveal a previously unrecognized role for parasite surface clan CA cysteine proteases in the pathophysiology of G. duodenalis infections, and implicate them as potential virulence factors. CatB/L activity was observed within G. duodenalis trophozoites and in parasite supernatants following exposure to host IECs. This activity was inhibited when parasites were pre-treated with the broad-spectrum clan CA, cysteine protease inhibitor E-64d [52] or the catB-specific inhibitor Ca-074Me [55]. Co-incubation of the parasite with IEC's increased the production of catB and L activity for Assemblage A, but not for the Assemblage B isolate. However, no cathepsin activity was observed within the Caco-2 cells following exposure to G. duodenalis trophozoites from any isolate, ruling out any translocation of the cysteine proteases into the cells from the supernatant. G. duodenalis trophozoites degraded and redistributed the intestinal epithelial cytoskeletal protein villin; this required direct contact between the parasite and IEC, and was prevented when parasites were pre-treated with E-64d prior to incubation with intestinal monolayers. In addition, this study elucidated that degradation and redistribution of the tight junction protein ZO-1 was not dependent on factors sensitive to inhibition with E-64d, thereby suggesting that G. duodenalis may induce multiple pathophysiological responses within host IECs, some mediated by parasite clan CA cysteine proteases, some not. Finally, this study demonstrates that degradation and disruption of villin is partially dependent upon MLCK activation. At early time points, MLCK inhibition in intestinal monolayers exposed to G. duodenalis trophozoites did not prevent villin degradation; however, MLCK inhibition prevented G. duodenalis-induced villin degradation at later time points. Collectively, these results suggest that G. duodenalis surface clan CA cysteine proteases induce villin degradation and redistribution within IECs and that this effect, at least in part, is mediated by MLCK activation within IECs.
To elucidate the role of parasite cathepsin cysteine proteases in the pathophysiology of G. duodenalis infections, we used a well-established in vitro cell culture model [18,19,34,57,71,72]. Furthermore, our experiments used inhibitors, which have previously been shown to successfully attenuate catB and L cysteine proteases in mammalian and protozoan parasites [24,52,55,73], including G. duodenalis [34,67]. Results from this study corroborate previous observations that G. duodenalis trophozoites promote the disruption of intestinal epithelial barrier proteins in vitro [37-40, 44, 69], and in vivo [46], and disrupt cytoskeletal protein villin [51]. These studies implicated host lymphocytes in the disruption of the intestinal epithelial cytoskeleton. In contrast, our study demonstrates that parasite clan CA cysteine proteases can directly disrupt intestinal epithelial villin.
Within the gastrointestinal tract, the cytoskeletal protein villin is exclusively expressed by IECs, and its primary function is to maintain microvillous brush border integrity during periods of stress via its ability to promote the assembly and disassembly of the actin cytoskeleton (reviewed in [49]). Pathogen-mediated actin cytoskeleton disruptions can aid in the establishment of infection [74][75][76][77][78]. Interestingly, inhibition of contact between parasites and IECs via chemical disruption of G. duodenalis lipid rafts prevents actin cytoskeletal remodelling [71]. Consistent with these observations, results from the present study demonstrate that G. duodenalis trophozoites were unable to induce villin degradation and redistribution within IECs when the two were separated by transwells. Collectively, these results suggest that intestinal epithelial actin cytoskeletal remodelling during giardiasis is dependent at least in part on contact between G. duodenalis trophozoites and host IECs. It remains to be elucidated whether actin cytoskeletal remodelling via villin activation is necessary for establishment of G. duodenalis infection.
Disruption of villin and the related protein ezrin occurs during the late stages of G. duodenalis infection in vivo, and is dependent on CD4+ and CD8+ lymphocytes [51]. Data from the present study indicate that early disruption of villin within IECs is dependent on clan CA cysteine proteases. Together, these results suggest that G. duodenalis uses more than one mechanism to disrupt villin and the intestinal epithelial brush border. A variety of invasive gastrointestinal pathogens are reliant on the intestinal epithelial expression of villin and its ability to remodel the actin cytoskeleton to facilitate entry and dissemination through host tissues [79][80][81]. Furthermore, although Salmonella sp. and enteropathogenic Escherichia coli induce villin redistribution, this occurs in the absence of villin proteolysis [82,83]. Therefore, G. duodenalismediated disruption of villin may significantly affect the ability of other gastrointestinal pathogens to colonize and cause disease. Moreover, G. duodenalis infections have been reported to occur simultaneously with other pro-inflammatory gastrointestinal pathogens [84][85][86][87]. A recent report examining giardiasis in Tanzanian children found that individuals infected with G. duodenalis had reduced incidence of diarrhoeal disease and fever, and lower serum inflammatory scores when compared to individuals not infected with G. duodenalis [88]. More research is warranted to assess whether G. duodenalis infections may prevent other pathogens from establishing and/or inducing disease within their host.
Several parasites contain and produce cathepsin cysteine proteases (reviewed in [22,23]). The G. duodenalis genome contains genes for multiple cathepsin B and L cysteine proteases, but their functions have only begun to be described [26]. Results illustrated herein indicate that G. duodenalis cathepsin cysteine proteases are involved in the degradation and redistribution of the intestinal epithelial cytoskeletal protein villin, in keeping with the observation that Entamoeba histolytica cysteine proteases promote villin breakdown [70]. The role of villin breakdown in the establishment of infection by enteric parasites requires further investigation. Prior to this study, G. duodenalis cathepsin cysteine proteases had been known to play a role in trophozoite encystation and excystation [33], and degrading the potent neutrophil chemoattractant interleukin-8 (CXCL8) [34]. Importantly, G. duodenalis-mediated attenuation of CXCL8 did not require direct contact between parasites and host IEC monolayers [34]. These results suggest G. duodenalis trophozoites possess multiple types of cathepsin cysteine proteases, and each may have a different role in disease pathogenesis. Indeed, single-celled parasites can possess both surface-associated and secreted cathepsin cysteine proteases which may have unique roles in pathogenesis [89][90][91][92].
In conclusion, our data reveal a novel role for G. duodenalis cathepsin cysteine proteases in the degradation and redistribution of the intestinal epithelial cytoskeletal protein villin. These results support previous observations that disruption of the intestinal epithelial cytoskeleton requires intimate contact between parasites and intestinal epithelial cells. The present findings also suggest that Giardia cysteine proteases disrupt villin at least in part by engaging epithelial MLCK. Our findings are the first to indicate that G. duodenalis cathepsin cysteine proteases represent potential parasite virulence factors capable of inducing pathophysiological responses within host intestinal epithelial cells.