Identification of Equine Lactadherin-Derived Peptides that Inhibit Rotavirus Infection via Integrin Receptor Competition

Background: Results: A 20 aa peptide (namely pDGE-RGD) derived from equine lactahderin inhibits human rotavirus infectivity. Conclusion: pDGE-RGD interacts specifically with α2β1 integrin, hindering the rotavirus cell-attachment process. ABSTRACT Human rotavirus is the leading cause of severe gastroenteritis in infants and children under the age of five years in both developed and developing Human lactadherin, a milk-fat globule membrane (MFGM) glycoprotein, inhibits human rotavirus infection in vitro, whereas bovine lactadherin is not active. Moreover, it against symptomatic rotavirus infections. To explore the potential antiviral activity of lactadherin sourced by equines we undertook a proteomic analysis of MFGM proteins from donkey milk and elucidated its amino acid sequence. Alignment of the human, bovine and donkey lactadherin sequences revealed the presence of an Asp-Gly-Glu (DGE) α 2 β 1 integrin-binding motif in the N-terminal domain of donkey sequence only. Since integrin α 2 β 1 plays a critical role during early steps of rotavirus host cell adhesion, we tested a mini-library of donkey lactadherin-derived peptides containing DGE sequence for anti-rotavirus activity. A 20 amino acid peptide containing both DGE and RGD motifs (named pDGE-RGD) showed the greatest activity, and its mechanism of antiviral action was characterised: pDGE-RGD binds to integrin α 2 β 1 by means of DGE motif and inhibits rotavirus attachment to the cell surface. These findings suggest the potential anti-rotavirus activity of equine lactadherin and support the feasibility of


developing an anti-rotavirus peptide that acts by hindering virus-receptor binding.
Human rotavirus is a double-stranded RNA virus belonging to the Reoviridae family. It is transmitted by the faecal-oral route, causing severe gastroenteritis in infants and children under the age of five years in both developed and developing countries, although mortality occurs mainly in developing countries. As a result of rotavirus infection, 114 million episodes of diarrhea, 25 million clinic visits, 2.4 million hospital admissions, and more than 500,000 deaths in children up to age 5 occur worldwide annually (1). Groups at increased risk for rotavirus infection include children who are hospitalized or in community care centers and undernourished and/or immunodeficient children (2). Moreover, since exclusive breastfeeding was found to be associated with a lower incidence of rotavirus gastroenteritis (3)(4)(5)(6), nonexclusively breastfed children are considered an additional group more vulnerable to rotavirus infections. The mature virion is a triple-layered particle of about 100 nm in diameter: the most external layer is comprised of two viral proteins (VP), VP7 (~34 kDa) and VP4 (87 kDa) (7,8) with VP4 being the major determinant of tropism and receptor binding (9)(10)(11)(12). Trimeric spikes of VP4 are anchored into the intermediate VP6 layer, while the trimeric calcium-binding protein VP7 covers the virion surface, locking VP4 spikes into place. The proteolytic cleavage of VP4 by trypsin is essential for optimum rotavirus infectivity and produces two subunits, VP5* (60 kDa) and VP8* (28 kDa), which remain associated with the virion (13)(14)(15). Initial cell attachment by rotaviruses is mediated by VP8* binding to host cell glycans (16). Infection of permissive cells by many rotaviruses, including human (e.g. Wa and K8), monkey (RRV and SA11), and bovine (NCDV) strains, also depends on virus binding to particular integrins, a family of cell surface proteins that recognize extracellular matrix proteins (e.g., collagen), cell surface ligands (e.g., Vascular cell adhesion molecule-1) (17), growth factors (e.g., Fibroblast growth factor-1)( 18), and viral proteins (e.g., rotavirus). VP5* recognition of the collagen-binding α2β1 integrin is a key event in rotavirus binding and entry into cells which is followed by the interaction of VP7 with integrins αxβ2, α4β1 and αvβ3 (9,(19)(20)(21)(22)(23)(24). The VP5* of almost all group A rotaviruses contain the Asp-Gly-Glu (DGE) sequence at aa 308-310, a motif that has been implicated in α2β1 recognition by type I collagen (17). Mutation of the putative α2β1 ligand sequence DGE abrogates binding of truncated VP5* to the integrin α2 subunit I domain (α2I) and VP5* competition with RRV cell binding and infectivity (9,25). In addition, DGE-containing peptides such as Asp-Gly-Glu-Ala (DGEA) specifically inhibit rotaviruscell binding and infection mediated by α2β1 (9,20,21,25). Binding by infectious monkey (SA11, RRV) and human (Wa) rotaviruses to recombinant α2β1 expressed on K562 cells was specifically inhibited by DG-containing peptides and a function-blocking antibody to the α2I domain (9,21,23). Therefore, the interaction of rotavirus with α2β1 integrin can be considered a target for the development of antiviral agents aimed at preventing or reducing rotavirus infection. Bioactive components in milk are an important research focus (26). In vitro, lactadherin, a milk-fat globule membrane (MFGM) glycoprotein, inhibits rotavirus infection when sourced from humans, whereas bovine lactadherin is not active against human rotavirus (27). Moreover, it has been reported that human milk lactadherin protects breastfed infants against symptomatic rotavirus infections (6). Lactadherins isolated from different animal species are characterised by two N-terminal epidermal growth factor (EGF)-like domains, the second of which contains an Arg-Gly-Asp (RGD) integrin-binding motif that engages αvβ3/5 integrins and two C-terminal Discoidin/F5/8C domain repeats (28). Bovine lactadherin consists of 427 amino acid residues, including an N-terminal signal sequence of 18 amino acids that is cleaved during intra-cellular processing. In contrast, human lactadherin is a protein of 387 amino acids including an Nterminal signal sequence of 23 residues. When separated by two-dimensional electrophoresis (2DE), each of these proteins shows an apparent molecular mass above the theoretical value, with the difference due to glycosylation (29-31). MFGM-associated proteins from water buffalo, sheep and horse also have been analysed (32-34). Comparative species-to-species structural studies among MFGM proteins showed high variation especially for lactadherin (35). It is proposed that these differences could explain the divergent effects of bovine and human lactadherin on human rotavirus infection. In order to test if the lack of rotavirus inhibition by bovine lactadherin is atypical or represents a shared property amongst lactadherins of related domesticated animals we undertook a proteomic analysis of MFGM proteins from equine milk and elucidated the amino acid sequence of donkey lactadherin. Alignment of the human, bovine and donkey lactadherin sequences revealed the presence of a DGE integrin-binding sequence motif in the donkey sequence only, suggesting a potential antirotavirus activity. To test this hypothesis, a mini-library of donkey lactadherin-derived peptides containing DGE sequence was tested for anti-rotavirus activity. A 20 amino acid peptide containing both DGE and RGD sequence showed the greatest activity, and its mechanism of antiviral action was characterised. These findings shed light on the mechanisms of equine lactadherin inhibition of rotavirus infection and support the feasibility of developing an anti-rotavirus peptide that acts through inhibition of virus-receptor binding.

EXPERIMENTAL PROCEDURES
Preparation of fat globules from donkey milk -Milk was collected from several donkeys at a single farm. Protease inhibitors (Complete; Roche Diagnostics, Basle, Switzerland) were added and the milk immediately stored at -20°C. Thawed milk was centrifuged at 5000×g for 30 min at 10°C and the pellet discarded. The cream layer and skimmed milk were centrifuged at 189,000×g for 70 min at 6°C. Fat globules were recovered in the supernatant and washed three times with 0.9% (w/v) NaCl.

Sample protein preparation and 2-DE -
Washed fat globules were incubated at 4°C overnight in 20 mM Tris-HCl pH 8.6 containing 1% (w/v) ASB-14, 1% (v/v) Triton X-100, 7 M urea and 2 M thiourea to extract the proteins associated with fat globule membranes. After centrifugation at 18,400×g for 10 min at 10°C the floating cream layer was discarded. Proteins were precipitated from the supernatant with methanol and chloroform, as described previously (36). Pellets containing proteins were solubilized in 20 mM Tris-HCl pH8.6 containing 7 M urea, 2 M thiourea, 1% (w/v) ASB-14, 1% (v/v) DTT and 0.5% (v/v) IPG buffer. Total protein was quantified using the 2-D Quant kit (GE Healthcare). Extracted proteins (200 µg) were loaded onto 13 cm pH 3-10 NLIPG Strips (GE Healthcare Bio-Sciences Corp, USA). IEF was carried out on an IPGphor unit (GE Healthcare) at 20°C and 8000 V for a total of 70,000 Vhr. Strips were incubated at room temperature in 50mM Tris-HCl pH 8.6 containing 6 M urea, 30% (v/v) glycerol, 2% (w/v) SDS, enriched with 2% (w/v) DTT for 20 min and afterwards with 4.5% (w/v) iodoacetamide for 20 min. SDS polyacrylamide gel electrophoresis was carried out on homogeneous running gels with 11.7% total acrylamide concentration and a 2.6% grade of cross-linking (Ettan DALT II system, GE Healthcare) at 400 V and 50 mA per gel for approx. 3 h. Gels were stained using the Processor Plus (GE Healthcare) with Blue Coomassie Colloidal stain (37) and scanned with a GS-800 Densitometer (Biorad Laboratories Inc., Hercules, CA, USA).
Enzymatic digestion of proteins -In-gel Multiple Enzymatic Digestion on selected 2-DE spots was performed according to the published method (38). Briefly, excised spots were destained for 2 h in 40% (v/v) ethanol containing 50 mM NH 4 HCO 3 , dehydrated 3 times (10 min each) in acetonitrile and dried in a vacuum concentrator for 30 min. Rehydration was performed at room temperature for 5 min with 5 µl of 75 ng/µl trypsin in 25 mM NH 4 HCO 3 (Promega Corp., Madison, WI), 5 µl of 75 ng/µl Asp-N solution (Sigma), or 5 µl of 75 ng/µl of chymotrypsin in 25 mM NH 4 HCO 3 (Sigma). Digestion continued at 37°C. Asp-N peptides were collected after 2 h of digestion and submitted to mass spectrometry (MS). Trypsin and chymotrypsin digestions were performed overnight after the addition of 15 µl of 25 mM NH 4 HCO 3 /10% (v/v) acetonitrile. These supernatants were collected and submitted for MS analysis on the following day Mass spectrometric analysis -For MALDI-TOF mass spectrometry, 0.5 µl of each peptide mixture was applied to a target disk and allowed to air-dry. Subsequently, 0.5 µl of matrix solution (1% (w/v) α-cyano-4hydroxycinnamic acid in 30% (v/v) acetonitrile and 0.1% (v/v) trifluoroacetic acid were applied to the dried sample and dried under vacuum. Spectra were collected on a Ultraflex II MALDI-TOF-TOF (Bruker Daltonik GmbH, Germany). The reported mass spectra are the result of signals of approximately 800 laser shots. The spectra were externally calibrated using the standard calibration mixture from Bruker Daltonik (GmbH, Germany). Manual/visual estimation of the mass spectra was performed using Flex Analysis software (Bruker Daltonik GmbH, Germany). For LC-MS/MS experiments an Agilent's 1100 Series liquid chromatography and LC/MSD XCT series ion trap mass spectrometer equipped with nano-ESI source were used. Peptide mixtures were separated on a RP C18 column (Zorbax 300SB-C18 3.5 µm 150x0.075 mm) (CPS) equilibrated with 0.1% (v/v) formic acid in water. Peptides were eluted over 55 min with a linear gradient of 5%-70% (v/v) of 0.5% water/0.1% formic acid in acetonitrile. The flow rate was 300 nl/min. The mass spectrometer was operated in positive ion mode. Data were processed using DataAnalysis software (Agilent Technologies, Santa Clara, U.S.).

Spectra interpretation -For Peptide Mass
Fingerprinting (PMF) identification, the MS-Fit software package in ProteinProspector version 5.10.9 (http://prospector.ucsf.edu/) was used for spectra interpretation (39). Data were searched against the Horse database UniProtKB.2012.03.21. The parameters used for the search were: Scarbamidomethyl derivate on cysteine as fixed modification, oxidation on methionine as variable modification and two missed cleavage sites for trypsin digestion. The peptide mass tolerance was 30 ppm. The LC-MS/MS and MALDI-TOF/TOF spectra were analyzed using the MASCOT MS/MS Ions Search engine (http://www.matrixscience.com/) against the UniProtKB.2012.03.21database.The parameters used for the search were: S-carbamidomethyl derivate on cysteine as fixed modification, oxidation on methionine as variable modification, two missed cleavage sites for each enzymatic digestionand default selection of charge states ions (+2, +3). The peptide mass tolerance was set to 0.6 Da and the fragment mass tolerance to 0.8 Da. Protein hits were validated if the protein scores were above the MASCOT default significance threshold (p≤0.05), and a minimum of two unique peptides matched.
N-terminal sequencing -Spots were passively eluted from gels as described elsewhere (40) and the membrane analyzed by direct micro sequencing on a Procise 492 protein sequencer (Applied Biosystems, CA, USA). All chemicals were purchased from Applied Biosystems. Pasching, Austria) supplemented as above. The derivation of CHO K1 cell transfectants expressing human α2 integrin subunit (CHO α2) or the PBJ-1 empty vector (CHO K1 PBJ-1) has been described previously (41)(42)(43)(44)(45). CHO cell transfectants were grown as for the parental cell line, except that G418 sulfate (Sigma) at 0.1 µg/ml to 0.8 µg/ml was included in the growth medium. Human rotavirus strain Wa was obtained from the ATCC; the origin of porcine rotavirus P9 G3 strain CRW-8 has been described previously (46). Wa and CRW-8 infectivity was activated with 5 µg/ml of porcine pancreatic trypsin type IX (Sigma) for 30 min at 37°C and propagated in MA104 cells in MEM containing 0.5 µg/ml trypsin as described previously (47). were purchased from Proteogenix. MALDI-TOF mass spectrometry analysis confirmed that pDGE-RGD, the peptide of major interest for this study, was soluble and monomeric (data not shown). Rotavirus infectivity assays -Peptide blockade of virus infectivity was determined by focus reduction assay using confluent MA104 cell monolayers in 96-well trays, as described previously (21). Unless otherwise stated, cells were treated for 1 h at 37°C with peptides at 0.01 µM to 500 µM in serum-free medium prior to virus addition. Rotavirus infection was performed at a multiplicity of infection (MOI) of 0.02, unless otherwise indicated, for 1 h at 37°C, in presence of the peptide unless otherwise stated. Infected cells were washed with serum-free medium and incubated in this medium at 37°C in a humidified incubator in 5% (vol/vol) CO 2 -95% (vol/vol) air. In some experiments (as indicated), peptide pDGE-RGD was included with these washed, infected cells for the duration of their incubation. After 16 h of incubation, infected cells were fixed with cold acetone-methanol (50:50), and viral titers determined by indirect immunostaining, by using UltraTech HRP Streptavidin-Biotin Detection System (Beckman Coulter) according to the manufacturer's instructions. Peptide blockade of viral infectivity is expressed as mean % ± SD. Where possible, anti-viral effective concentration (IC 50 ) values were calculated by regression analysis using the dose-response curves generated from the experimental data, using PRISM 4 (GraphPad Software, San Diego, CA, U.S.A.). IC 50 values were compared using the sum-of-squares F test.

Antibodies -
Assay of virus yield -To test the ability of pDGE-RGD to inhibit multiple cycles of viral replication, confluent MA104 cells in 24-well trays were infected with trypsin-activated Wa rotavirus (MOI=0.02) for 1 h at 37°C and washed as above. Cells were incubated in medium supplemented with 0.5 µg/ml porcine trypsin and 30, 100 or 300 µM pDGE-RGD, with fresh pDGE-RGD added to peptide-treated wells at 16 h after infection and at later fixed intervals, and an equal volume of medium added to untreated controls at these times. Infected cells and cell supernatants were harvested at 48 h post infection and virus titers determined by indirect immunostaining of MA104 cell monolayers inoculated with serial dilutions of the samples. The assay was performed in triplicate.
Rotavirus-cell binding assay -Confluent MA104 cell monolayers in 24-well trays were washed, incubated with peptides or antibodies for 1 h at 37°C and cooled on ice for 20 min in the continuing presence of peptides or antibodies. Trypsin-activated virus, which had been cooled to 4°C, was allowed to attach to cells on ice for 1 h (MOI=3) in presence of peptides or antibodies. Cells were washed with cold MEM; cold MEM was added. Cells were subjected to two rounds of freeze-thawing, incubated at 37°C for 30 min with 10 µg/ml porcine trypsin to release bound virus, and the lysate clarified by low speed centrifugation for 10 min. Cell-bound virus titers were determined by indirect immunostaining as above.
Rotavirus-cell entry assay -Confluent MA104 cell monolayers in 96-well trays were washed twice with MEM and cooled on ice for 20 min. Trypsin-activated virus, which had been cooled to 4°C, was adsorbed to cells on ice for 1 h at 4°C (MOI=0.02). Peptide pDGE-RGD was added to washed cells at 0.01 µM to 500 µM for 1 h at 37°C. Rotavirus entry was stopped by removal of surface-bound virions with two quick washes with 3 mM EGTA in PBS, which removes the virion outer capsid (48,49). Cells were incubated in MEM for 16 h at 37°C, fixed, stained and titers determined as described above for the infectivity assay.
Peptide-cell binding assay -Confluent CHO K1 and CHO α2 cells in 96-well trays were washed twice with DMEM containing 1% (w/v) BSA and incubated for 1 h at 37°C with 0.3 µM, 3 µM or 30 µM Biotin-pDGE, Biotin-pAGE or Biotin-pDGE-RGD peptides in the same DMEM diluent. Washed cells were incubated with streptavidin-HRP (Beckman Coulter) for 10 min at room temperature, washed twice with PBS and reacted with ABTS reagent (Thermoscientific, IL, U.S.A.) for 30 min at room temperature The reaction was stopped with PBS containing 5% (w/v) SDS and the absorbance determined at λ=405 nm.
Cell viability assay -The cytotoxicity of peptide pDGE-RGD was assessed in MA104 and CaCo2 cells seeded in 96-well plates at a density of 1.2×10 4 cells/well. After 24 h of culture, cells were treated with peptide at 0.0044 µM to 880 µM. Cell viability was determined 48 h later with the CellTiter 96 Proliferation Assay kit (Promega, WI, USA) according to the manufacturer's instructions. Absorbances were measured using a model 680 microplate reader (Biorad) at 490 nm. Viability in treated cells was expressed as a percentage relative to cells incubated with culture medium alone. The 50% cytotoxic concentration (CC 50 ) was determined using logarithmic viability curves. Where possible, a selectivity index (SI) was calculated by dividing the CC 50 by the IC 50 value.
Statistical analysis -One-way analysis of variance (ANOVA), in some cases followed by the Bonferroni test, was used to assess the statistical significance of differences in virus titers and absorbance data. Significance was set at the 95% level.

RESULTS
Proteomic analysis and protein sequencing indicate the presence of two putative integrin ligand motifs in equine lactadherin -The 2-DE map of donkey milk fat globule membrane proteins was determined ( Figure 1). The 2DE map showed a pattern very similar to that previously published for horse MFGM proteins, where lactadherin was the major protein and was identified in several spots at different molecular masses (34). The spots 1 to 4, corresponding to the molecular masses of horse lactadherin, were excised and the proteins identified by MALDI-TOF MS and PMF as the horse milk fat globule-EGF factor 8 splice variant, also known as horse lactadherin (Table  1). To understand the nature of the difference between the ~45 kDa and ~50 kDa proteins, their N-terminal amino acid sequences were determined ( Table 2). The sequences of proteins 1 and 2 matched the N-terminal amino acid sequence of horse lactadherin (UniProt F7B0S3) with an identity of ≥95%. Proteins 3 and 4 corresponded to the same horse lactadherin sequence, commencing from the amino acid residue at position 44. As the 45 kDa form was most similar to the human counterpart, its amino acid sequence was examined further. The results of multiple enzymatic digestions of this 45 kDa protein with trypsin, Asp-N and chymotrypsin, followed by MALDI-TOF/TOF and nano LC−ESI MS/MS analyses (Table 3), together with the N-terminal amino acid sequence, gave an (almost) complete coverage of the sequence of this form of donkey lactadherin ( Figure 2). The sequence was submitted to UniProt database and the accession number C0HJR4 was assigned. The sequence alignment of this equine (EQ) lactadherin with the bovine (BO) and human (HU) lactadherins revealed some very interesting features in the N-terminal domains ( Figure 3). An RGD motif is conserved in the three proteins, being present at amino acid residues 23 to 25 in equine and human lactadherin, and at residues 67 to 69 in bovine lactadherin. RGD is an integrin-binding motif that engages α v β 3 , α v β 5, α 3 β 1 and α 5 β 1 integrins to facilitate cell adhesion. Surprisingly, the equine lacthaderin was shown to contain an additional integrin binding motif, the DGE sequence that can specifically bind the α 2 β 1 integrin, which is present at residues 12 to 14.
Functional roles of the equine lactadherin RGD and DGE sequences in inhibition of rotavirus infection -In order to assess the importance of the RGD sequence (shared by human, bovine and equine lactadherin) and the DGE motif (present in equine lactadherin only) for inhibition of rotaviral infection, the effect on Wa infection of cellular treatment with the 8 amino acid peptides pRGD-Hu, pRGD-Bo, pRGD-Eq, and pDGE was determined. These human, bovine and equine RGD-peptides inhibited Wa infection of MA104 cells in a dose-dependent fashion to maxima at 500 µM of 57%, 50% and 47% respectively ( Figure  4A). In contrast, MA104 cell treatment with an 8 amino acid RGE peptide harboring a single amino acid substitution (D→E) known to inhibit the ability of RGD triplet to bind integrin (pRGE), had no effect on virus infectivity even at high concentrations ( Figure  4A). Notably, the DGE-containing peptide pDGE derived from equine lactadherin inhibited Wa infectivity to maximum of 84% ( Figure 4B), making it possible to assess its IC 50 at 17 µM (Table 4). As expected, an AGE peptide (pAGE), despite its ability to inhibit rotaviral infectivity to maximum of 80%, showed a significantly higher IC 50 (257 µM) than pDGE (p<0.0001; Table 4). As positive control, we tested the Type I collagen-derived peptide Asp-Gly-Glu-Ala (pDGEA); this α 2 β 1 integrin ligand showed an antiviral activity comparable to the one previously reported (21), as depicted in Figure 4B.
A 20 amino acid region of equine lactadherin has elevated specific anti-rotaviral activity in epithelial cell cultures that depends on its RGD and DGE sequences -Interestingly, the 20 amino acid peptide pDGE-RGD, corresponding to amino acids 10 to 29 of equine lactadherin and harboring both the RGD and DGE motifs, showed the highest anti-rotaviral efficacy (IC 50 =4.4 µM; Table 4). pDGE-RGD was therefore selected as the lead compound for further studies. Of note, the pDGE-RGD peptide maintained a significant (p<0.05) inhibitory activity against high MOIs of Wa ( Figure 4C). The stage of the rotavirus replication cycle inhibited by peptide pDGE-RGD was examined. Peptide addition 1 h before or immediately after infection did not inhibit Wa infectivity. The greatest infectivity inhibition was achieved when the peptide was added during infection ( Figure 5), as performed in the experiments described above and the following studies. To assess the role played by each motif in inhibition of rotavirus infectivity, the antiviral activity of several mutant peptides derived from pRGD-DGE that harbored mutations in DGE (D→A, pAGE-RGD), RGD (D→E, pDGE-RGE) or both triplets (pAGE-RGE) was tested. These peptides failed to show any antiviral activity even at high concentrations, demonstrating the importance of these motifs for anti-rotavirus activity ( Table  4). In contrast, peptides with single amino acid substitutions outside these functional groups, namely pDGE-RGDmut1, pDGE-RGDmut2 and pDGE-RGDmut3, retained the ability to inhibit Wa infection (Table 4), at levels comparable to those of pDGE-RGD. No significant difference was seen between the IC 50 of pDGE-RGD and these peptides with amino acid substitutions outside the DGE and RGD functional groups (p>0.05). The antiviral efficacy of pDGE-RGD and its AGE-and RGE-mutants was also evaluated in the Caco-2 intestinal cell line (Table 5). While pDGE-RGD showed antiviral activity in Caco-2 cells (IC 50 =8.8 µM) comparable to that in MA104 cells, pDGE-RGE, pAGE-RGD, and pAGE-RGE exhibited a significant loss of activity (p<0.0001). Interestingly, these Caco-2 cell studies identified significant differences in antiviral activity between these three peptides (0.002<p<0.0194), with pDGE-RGE showing a lower IC 50 (46 µM) than pAGE-RGD (IC 50 =101 µM) and pAGE-RGE (IC 50 =152 µM). This suggests that the DGE sequence plays a major role in the process leading to rotavirus inhibition. In order to rule out any non-specific effect due to loss of cell viability, MTS cell viability assays were performed after incubation of Wa-infected and control MA104 and Caco-2 cell cultures for 72 h with a range of peptide concentrations. The tested peptides did not impair cell viability at any concentration tested on MA104 cells, showing a 50% cytotoxic concentration (CC 50 ) of >880 µM (Tables 4 and  5). In Caco-2 cells, the pDGE-RGD CC 50 was 945 µM. CC 50 values were used to calculate the selectivity index of pDGE-RGD, which was >100 for both cell lines (Tables 4 and 5).
Infectious rotavirus yield is reduced by the 20 amino acid lactadherin peptide -The antiviral activity of pDGE-RGD was evaluated in a virus yield assay, a more a stringent test that allows multiple cycles of viral replication to occur before measuring the titer of infectious virus produced. Wa rotavirus yield was reduced in a dose-dependent manner in cells treated with 100 µM or 300 µM pDGE-RGD (p<0.05), showing 50% and 84% inhibition, respectively ( Figure 6).
The 20 amino acid lactadherin peptide binds α 2 β 1 integrin in a DGE-dependent manner -The results described above suggested that pDGE-RGD may inhibit rotavirus infectivity by binding the α 2 β 1 integrin, thereby blocking viral binding to this cellular receptor. The ability of equine lactadherin peptides to specifically bind α 2 β 1 was tested by treating CHO cells expressing surface α 2 β 1 with biotinylated pDGE-RGD, pDGE and pAGE peptides. Peptides pDGE-RGD and pDGE bound to the surface of α 2 β 1 -expressing CHO cells in a dosedependent manner, whereas pAGE did not show this ability ( Figure 7A). Peptide pDGE-RGD bound to a greater extent than pDGE, as pDGE-RGD binding was significant greater than that of pDGE at 3 µM and 30 µM (0.01<p<0.02). In contrast, pDGE binding to the CHO parental cell line (CHO-K1) ( Figure  7B) was reduced compared to α 2 β 1 -expressing CHO cells at all concentrations tested (p<0.0006). No difference was detected between pDGE and pAGE binding to CHO-K1 cells (p>0.05). Interestingly, substantial binding of pDGE-RGD to CHO-K1 cells was evident, indicating binding to cell surface component(s) other than α 2 β 1. Consistently, at 30 μM pDGE-RGD binding to the surface of α 2 β 1 -expressing CHO cells was significantly greater than to the parental cell line (p<0.05).
The 20 amino acid lactadherin peptide preferentially inhibits rotavirus-cell attachment rather than entry, in a DGE-dependent manner -In order to evaluate the ability of pDGE-RGD to inhibit rotavirus-cell attachment we measured the titer of infectious Wa rotavirus particles attached to pDGE-RGD-treated and untreated MA104 cells. pDGE-RGD significantly reduced Wa binding by a maximum of 67% (p<0.0001), while pAGE-RGE did not show any inhibitory activity (p>0.05; Figure 8A). As found previously (9), the AK7 MAb to integrin α 2 β 1 reduced viruscell binding to maximum of 51% (p<0.05), whereas MAb LM609 to integrin α v β 3 had no significant effect (p>0.05) effect. In a control experiment, we tested the antiviral efficacy of pDGE-RGD against the integrin-independent rotavirus strain CRW-8. As expected, pDGE-RGD did not inhibit CRW-8 infectivity at any concentration tested ( Figure 8B). The effect of pDGE-RGD on viral entry also was investigated. This peptide significantly inhibited Wa rotavirus-cell entry only when added to cells at very high concentations i.e. 300 µM or 500 µM (p<0.05; Figure 8C). In order to rule out any virucidal effect of pDGE-RGD on rotavirus, Wa was exposed for 1 h to a pDGE-RGD concentration (100 µM) able to inhibit to 80% of Wa infectivity. The infectious titer of the pDGE-RGD-treated viral suspension was indistinguishable for that of the untreated control (p>0.05; Figure 8D).

Many human and animal rotaviruses contain the
Asp-Gly-Glu (DGE) sequence in VP8*, which is necessary for their usage of the α2β1 integrin to aid infection. In this study we have identified the putative integrin ligand motifs RGD and DGE in the N-terminal domain of donkey lactadherin. While the RGD motif is common to human, bovine and donkey lactadherins, the DGE motif is found only in the donkey sequence. Direct evaluation of the antiviral activity of donkey lactadherin would be of interest but has proven to be difficult. Difficulties arise when experiments to purify donkey lactadherin with the same methods used for the human and bovine proteins are attempted, since human and bovine milk contain a much higher amount of fat (approx 3.5%) than donkey milk (approx. 0.3% ) from which only a very small amount of lactadherin could be recovered. For this reason we turned to the recombinant protein. We achieved bacterial expression of full-length donkey lactadherin, but the recombinant His-tagged protein was insoluble and could not be refolded (unpublished observations). This is consistent with the insolubility of milk-derived lactadherins, which are associated with the milk fat globule membrane (50). In any case, our main focus is on the development of rotavirus inhibitors based on derivative peptides, rather than analysing the full-length protein: one main reason is that it is very unlikely that lactadherin molecules pass through the digestive tract intact -at least not in a relevant amount (e.g. we have easily digested donkey lactadherin with trypsin to solve their primary structure). Moreover, the ability of the DGE integrin binding domain to inhibit rotavirus replication can be addressed much more clearly working with small peptides rather than intact proteins, where the tertiary and eventually quaternary structure of the molecule could mask the biological activity of the inner part of some protein segment. To this end, we have demonstrated that peptides derived from donkey lactadherin, which contain the intact RGD and DGE sequences, specifically and effectively inhibit rotavirus infection at low µM levels. We selected a 20 amino acid peptide derived from donkey lactadherin harboring the RGD and a DGE motif (pDGE-RGD) depending on its favourable SI and its ability to limit the production of viral progeny resulting from multiple cycles of viral replication.
The differences between MA104 and Caco-2 cells in the effectiveness of pDGE-RGD in preventing rotavirus cell infection parallel the data documented in a previous work (21), showing that a DGEA peptide was a less effective inhibitor of virus binding in Caco-2 cells than in MA104 cells. The higher surface expression of α2β1 on Caco-2 cells than on MA104 cells (51) could explain these findings, as higher levels of peptide would be needed to block all available integrin sites that could bind virus.
During our investigation of the functional roles of the equine lactadherin RGD and DGE sequences in inhibition of rotavirus infection, we found that pAGE-RGD, pDGE-RGE and pAGE-RGE mutant peptides lack antiviral activity on MA104 cells. Interestingly, we were able to discriminate a different contribution from the two motifs using Caco2 cells, with the DGE triplet being more critical than RGD in limiting Wa infectivity (IC 50 of pAGE-RGD > IC 50 of pDGE-RGE). These data -together with the fact that peptides harboring single amino acid substitutions outside these functional motifs retain their antiviral activity -suggest that both triplets are important for the inhibition of rotavirus infectivity. These specific substitutions (i.e. D→A for DGE and D→E for RGD) were chosen because an aspartate (D) or glutamate (E) residue is a critical feature of all integrin recognition sites (52). The amino acid substitution in the DGE motif (i.e. D→A) is known to alter the integrin α2β1 binding ability of VP5* (21, 24, 32), with D308 in VP5* being the major requirement for α2I binding by rotavirus. Overall, these results suggest that the antiviral activity of pDGE-RGD is dependent on its ability to interact with α2β1 integrin. This notion is supported further by the finding that infection by integrin-independent virus CRW-8 was unaffected by pDGE-RGD. The demonstration of the antiviral efficacy of pDGE (an 8 aa peptide hardoring only the DGE triplet) extends data previously reported (21), which showed that individual peptide polymers containing the collagen-binding motif DGEA inhibit rotavirus infectivity by more than 60% in Caco-2 and MA104 cells and provided additional evidence of the importance of α2β1 in the rotavirus cell attachment and entry process. We also tested the antiviral efficacy of DGEA peptide in our experimental systems, confirming the efficacy previously reported against the Wa strain (21) and showing that pDGE has a greater inhibitory capacity. Notably, an 8 aa mutant peptide (pAGE) showed a significant antirotaviral activity only at high concentrations. Taken together, these results confirm that the antiviral activity of pDGE-RGD peptide is mostly dependent on the integrity of the DGE triplet. Nevertheless, pDGE shows a less effective antiviral activity than pDGE-RGD, suggesting that a second region of this peptide provides an additional contribution in inhibiting rotavirus infectivity. To this end, we showed that 8 aa RGD peptides (namely pRGD-Hu, pRGD-Bo and pRGD-Eq derived from human, bovine or equine lactadherin respectively) are endowed with a slight antiviral efficacy. Published studies provide evidence that rotavirus infectivity is not inhibited by RGD peptide, and functionblocking antibodies to RGD-binding integrins do not inhibit rotavirus infection (20,23,53). No evidence for Wa binding to integrins via RGD is available. We speculate that the slight antiviral efficacy of RGD peptides may be due to an indirect effect on rotavirus cell binding, perhaps by anchoring pDGE-RGD peptide to the cell surface and sterically blocking Wa virus access to receptors. The results of the virucidal assay rule out the possiblity of a direct effect of peptide treatment on rotavirus particles, while the time of addition assays suggest that pDGE-RGD might function by hindering the attachment of the virus to the host cell. The results of rotavirus cell-binding experiments confirm this conclusion: the titer of infective particles attached on the surface of pDGE-RGD treated cells is significantly reduced compared with the untreated control. Moreover, pAGE-RGE peptide does not reduce the number of infectious particles attached to the cell surface, showing that this ability is strictly dependent on the integrity of the DGE or RGD sequences. Consistent with these considerations, we demonstrated that both pDGE and pDGE-RGD bind α 2 β 1 integrin in a DGE-dependent manner, since pAGE does not interact with α2β1-expressing CHO-K1 cells.
The pDGE and pDGE-RGD peptides interact in a dose-dependent manner with α2β1 expressed on the surface of CHO-K1 cells. Importantly, pDGE-RGD bound the cellular surface to a greater extent than pDGE, which parallels the higher antiviral activity of the longer peptide. Moreover, we show that pDGE-RGD (unlike pDGE) is able to interact with the surface of CHO-K1 parental cell line that does not express α2β1 on the cellular surface. This result suggests that pDGE-RGD interacts wih a second surface determinant, a clue that could explain the greater antiviral activity of pDGE-RGD compared to pDGE. Of the known RGDbinding integrins, CHO-K1 cells express α5β1 and αvβ5, but not α3β1 or αvβ3 (54,55). It is likely that pDGE-RGD binds α5β1 or αvβ5 on CHO-K1 cells and α2β1-expressing CHO-K1 cells, as well as binding α2β1 on α2β1expressing CHO-K1 cells. The close proximity of the RGD and DGE motifs in pDGE-RGD is expected to preclude binding to both α2β1 and an RGD-recognising integrin simultaneously by a single peptide molecule. However, peptide anchorage to an RGD integrin like α5β1, which is abundant on CHO-K1 cells, could prevent virus access to α2β1 and possibly other receptors. Interestingly, rotavirus-cell entry experiments showed that pDGE-RGD is able to inhibit rotavirus infectivity when added after virus-cell attachment, even if only slightly and at high concentrations. One possibility is that high concentrations of pDGE-RGD could interfere with virus-cell attachment, even after it took place, by mediating detachment of part of the viral population from the cell surface. It is also possible that integrin-bound pDGE-RGD is endocytosed and inhibits virion conversion into the transcriptionally-active, double-layered particle, which for Wa has been proposed to occur during exit from late endosomes (56). Interestingly, Wa VP8* binds the GM1 ganglioside, which can associate with integrins such as α5β1 and α2β1 in membrane microdomains (57,58) and has been proposed to classify cargo for uptake and trafficking in late endosomes (59). Binding of pDGE-RGD to β1 integrins internalised in endosomes also might interfere with Wa access to the cytoplasm, which is essential for replication (60).
Overall, the information obtained in the present study indicates that a specific 20 aa peptide from the N-terminal domain of donkey lactadherin is able to block rotavirus celladhesion and infectivity in a DGE-dependent manner. Previously reported data (21) show that polymerization increases the effectiveness of peptide polymers containing the collagenbinding motif DGEA as rotavirus inhibitors by up to 10-fold, suggesting that it might be possible to develop more effective inhibitors of rotavirus-integrin interactions. In our experimental system, we also tested the in vitro antiviral activity of a dendrimeric peptide harboring multiple DGE motifs; this molecule displays an IC 50 (3.6 µM) that is not significantly different from the pDGE-RGD IC 50 (p>0.05; data not shown). Currently, no antivirals are commercially available for rotavirus. The discovery of the pDGE-RGD peptide may prove useful in the development of inhibitors of receptor recognition by rotavirus or other viruses that use α2β1 integrin during cell attachment and entry like human echovirus 1 (61,62).