Structural Analyses of a Dominant Cryptosporidium parvum Epitope Presented by H-2Kb Offer New Options To Combat Cryptosporidiosis

ABSTRACT Cryptosporidium parvum has gained much attention as a major cause of diarrhea in the world, particularly in those with compromised immune systems. The data currently available on how the immune system recognizes C. parvum are growing rapidly, but we lack data on the interactions among host major histocompatibility complex (MHC) diversity and parasitic T-cell epitopes. To identify antigenic epitopes in a murine model, we performed systematic profiling of H-2Kb-restricted peptides by screening the dominant Cryptosporidium antigens. The results revealed that the glycoprotein-derived epitope Gp40/15-SVF9 induced an immunodominant response in C. parvum-recovered C57BL/6 mice, and injection of the cytotoxic-T-lymphocyte (CTL) peptide with the adjuvant activated peptide-specific CD8+ T cells. Notably, the SVF9 epitope was highly conserved across Cryptosporidium hominis, C. parvum, and many other Cryptosporidium species. SVF9 also formed stable peptide-MHC class I (MHC I) complexes with HLA-A*0201, suggesting cross-reactivity between H-2Kb and human MHC I specificities. Crystal structure analyses revealed that the interactions of peptide-MHC surface residues of H-2Kb and HLA-A*0201 are highly conserved. The hydrogen bonds of H-2Kb–SVF9 are similar to those of a dominant epitope presented by HLA-A*0201, which can be recognized by a public human T-cell receptor (TCR). Notably, we found double conformations in position 4 (P4), 5 (P5) of the SVF9 peptide, which showed high flexibility, and multiple peptide conformations generated more molecular surfaces that can potentially be recognized by TCRs. Our findings demonstrate that an immunodominant C. parvum epitope and its homologs from different Cryptosporidium species and subtypes can benefit vaccine development to combat cryptosporidiosis.

responses in protective immunity against C. parvum infection but also advance our understanding of T-cell immune responses toward Cryptosporidium and, by extension, assist in vaccine development against cryptosporidiosis.

RESULTS
Epitope prediction for novel C. parvum proteins. A computer-based program was used to identify the potential mouse H-2 allele-specific peptides from C. parvum (32). The results of in silico prediction indicated that H-2K b might bind to more C. parvum epitopes than other mouse MHC I alleles. Interestingly, H-2K b -restricted epitopes in C57BL/6 mice were found to partly overlap peptides of human HLA-A*0201, one of the most prevalent HLAs in the global population (33). According to data from the Allele Frequency Net Database (http://www.allelefrequencies.net) (34), the HLA-A*0201 allele is widely distributed in South America, North America, Europe, Africa, and other countries and regions. The geographic distributions of HLA-A*0201 and the worldwide distribution of C. parvum outbreaks and morbidity are shown in Table S1 in the supplemental material. Next, we screened the main surface protein sequences, including Gp40/15, Cp23, Cp15, Gp900, and CSL, from C. parvum using NetMHCpan4.0, from which the algorithm generated peptide predictions for H-2K b and HLA-A*0201 alleles. Altogether, 202 8to 11-mer peptides were predicted to fit the H-2K b or HLA-A*0201 motifs, including 107 H-2K b -restricted peptides and 110 HLA-A*0201-restricted peptides. They had theoretical 50% inhibitory concentration (IC 50 ) values ranging from 2.9 to 1179.35 nM, with the levels of peptide binding to the MHC ranked as SB (strongly binding peptides, which is defined as a percentage rank of affinity (%rank) of less than 0.5) (%Rank , 0.500) or WB (weakly binding peptides) (0.500 , %Rank , 2.000) (Table S2). From the in silico analysis, 15 unique peptide sequences (8 9-mers, 5 10-mers, and 2 11-mers) were selected according to the following criteria: (i) peptides predicted to bind both H-2K b and HLA-A*0201 and (ii) strong binders with IC 50 values of ,1,000 nM (Table S2). Although H-2K b predominantly prefers octamers, most MHC molecules have a strong preference for 9-mer ligands (35). Thus, eight nonamerpromiscuous H-2K b and HLA-A2 binders were prioritized for the following experimental analysis, none of which have previously been reported.
The Gp40/15-derived epitopes induced a significant CD8 + T-cell response in C57BL/6 mice. To examine the T-cell responses specific for C. parvum surface antigens, C57BL/6 mice were vaccinated using a strategy of priming with attenuated oocysts and boosting with synthetic peptides (36,37). Individual peptides were tested, and Tcell-derived IFN-g was quantified using an ELISpot assay (38) (Fig. 1). Five peptides (Gp40/15-SVF9, Gp900-MIY9, CSL-MIW9, Gp900-KML, and Gp40/15-AIF9) elicited relatively strong responses compared to those in spleen cells without peptides restimulated with spot magnitudes of 1,042, 476, 309, 415, and 1,085 input cells, respectively (P , 0.01), whereas the remaining peptides elicited only weak responses ( Fig. 1A and Table 1). Two peptides (Gp40/15-SVF9 and Gp40/15-AIF9 [referred to as SVF9 and AIF9 here]) consistently elicited robust IFN-g responses in a proportion of the total T cells. Given the importance of IFN-g production in the ELISpot assay, we further investigated SVF9-and AIF9-specific IFN-g production by CD8 1 T cells via ELISpot and intracellular cytokine staining assays following three sets of immunizations with the SVF9 or AIF9 peptide vaccines. The results showed that the SVF9 and AIF9 peptides induced significantly higher numbers of IFN-g spots in both total splenocyte cultures and CD8 1 T-cell cultures, unlike the splenocytes isolated from the mouse groups immunized with Freund's adjuvant or the naive negative control (P , 0.01) (Fig. 1B). Cells stimulated with no peptide or cells from naive mice stimulated with either peptide did not respond to either SVF9 or AIF9. The values of the total IFN-g-positive (IFN-g 1 ) cells and CD8 1 T cells are indicated by representative ELISpot wells (Fig. 1C).
Since the Gp40/15-derived peptides SVF9 and AIF9 induced a high level of CD8 1 T-cell responses after receiving both C. parvum oocysts and peptide vaccines, we wondered whether oocyst infection alone could trigger a similar immunodominant response. Thus, we quantified the responses in the spleen after one or two immunizations with 1 Â 10 6 C. parvum oocysts ( Fig. 2A). After a single immunization, SVF9-specific CD8 1 T-cell responses reached the highest magnitude on day 7 ( Fig. 2A and B). The responses were slightly decreased on day 14 as a contraction of the response occurred, but more robust responses were observed after two immunizations with oocysts. Compared with SVF9 peptides, AIF9-specific CD8 1 T-cell responses were relatively weak and showed no significant differences in primary and secondary responses at 14 and 28 days ( Fig. 2A and B). Priming with viable C. parvum oocysts alone was sufficient to provide protective immunity to C57BL/6 mice, and the parasite load was significantly decreased with the second challenge with oocysts administered to IFN-g knockout (GKO) mice (Fig. 2C). To further characterize these SVF9-specific CD8 1 T cells, CD8 1 T cells from the oocyst-immunized mice were stained with the H-2K b -SVF9 multimer, and the frequencies of SVF9-specific CD8 1 T cells were significantly higher than those in naive mice ( Fig. 2D and E). Together, these results indicated that immunization FIG 1 Identification of C. parvum-specific CD8 1 T-cell epitopes in C57BL/6 mice immunized with oocysts followed by peptide. (A) IFN-g ELISpot analysis of splenocytes obtained 7 days after C. parvum oocyst and peptide immunization of mice (n = 6). Seven days after the last immunization, cells were restimulated with 8 peptides in vitro. (B) Splenocytes were isolated with a mouse CD8 1 T-cell isolation kit (CD8 1 T cell) or without depletion (Total splenocytes). The peptide-immunized groups were immunized with 50 mg (100 mL) of either peptide packaged with Freund's adjuvant in a total volume of 200 mL per mouse. The adjuvant controls received 100 mL of PBS and 100 mL of Freund's adjuvant, and the naive controls were each injected with 200 mL of PBS. (C) Representative ELISpot wells and splenocytes of CD8 1 T cells isolated from mice immunized with oocysts and peptides. Frequencies of IFN-g 1 cells/total CD8 1 T cells are indicated. Data represent means 6 standard deviations (SD), with significance determined using Student's t test (**, P , 0.01; ***, P , 0.001).
T-Cell Epitopes in Cryptosporidium mBio with C. parvum oocysts recruits SVF9-specific CD8 1 T cells to undergo significant peptide-specific T cell responses. These findings confirm that the SVF9 peptide acts as a CTL epitope and is a potential vaccine candidate for C. parvum. Vaccination with SVF9 peptides in Freund's adjuvant elevates protective CTL responses in the small intestine and spleen. SVF9-specific CD8 1 T-cell responses are considerably more abundant following in vitro restimulation in oocyst-immunized mice. We therefore tested whether vaccination with the SVF9 peptide in Freund's adjuvant was able to induce a protective CTL response and the benefits of parasite elimination in C57BL/6 mice. As shown in Fig. 3, mice were subcutaneously vaccinated with the SVF9 peptide admixed in complete Freund's adjuvant (CFA) or incomplete Freund's adjuvant (IFA) at 2week intervals before challenge with 2 Â 10 6 oocysts during the immunization protocol (Fig. 3A). Priming with the SVF9 peptide induced higher frequencies of IFN-g secretion than those of CD8 1 T cells from both the spleen and small intestine of oocyst-infected mice (P , 0.001), and cells from naive or adjuvant-immunized control mice did not respond to SVF9. The antigen-specific response of intraepithelial lymphocytes (IELs) was generally stronger than that of splenocytes, probably due to local site differences in C. parvum ( Fig. 3B and C). Using the highly immunogenic SVF9 CTL peptide in Freund's adjuvant as a model peptidebased vaccine, we investigated whether SVF9 peptide vaccination could prevent C. parvum infection. Next, the number of oocysts per gram (OPGs) was estimated at 1 to 28 days postinfection (dpi). As shown in Fig. S1, all mice inoculated with oocysts became infected. Oocysts were first detected in the feces at 4 dpi by light microscopy and peaked at days 6 to 8 in C57BL/6 mice, with a second level peaking at 12 dpi. Vaccination with the SVF9 CTL peptide could partially attenuate the severity of C. parvum infection in C57BL/6 mice albeit at a low level (Fig. S1). The oocyst shedding intensity in peptide-vaccinated mice was significantly lower at the peak of the oocyst excretion curve (7 dpi) than that in naive control mice (t = 10.099; P = 0.000). No oocysts were found in the SVF9-immunized group at 15 dpi, suggesting a critical role of T lymphocytes in parasite clearance (11).
The conserved SVF9 peptide is stably presented by the H-2K b and HLA-A*0201 molecules. As the SVF9 peptide is derived from the Gp60 protein, which is relatively less conserved and more likely to be mutated than other Cryptosporidium proteins, we assessed the level of homologous peptides from Cryptosporidium spp. (Table 2). Interestingly, the SVF9 epitope is highly conserved in C. hominis, C. parvum, C. tyzzeri, and C. felis, which are responsible for over 90% of infections in humans. To evaluate its ability to bind to mouse H-2K b and human HLA-A*0201 heavy chains, the SVF9 peptide was refolded with both H-2K b and HLA-A*0201 using an in vitro refolding method and purified by reverse-phase high-performance liquid chromatography (HPLC) (SciLight Biotechnology). The chromatographic results showed that the SVF9 peptide formed stable complexes with both H-2K b and HLA-A*0201 (Fig. 4A), but the AIF9 peptide formed low-stability complexes with H-2K b and could be collected by gel filtration, but a For the ELISpot assays, 11 indicates that CD8 1 T cells from immunized mice elicited a relatively strong response compared to that for spleen cells without restimulated peptides, and 2 indicates that the peptides elicited only a weak response and no significant difference between spleen cells with and those without restimulated peptides.
In the in vitro refolding assay, an elution peak representing the pMHC complex in the gel filtration experiments indicated that the peptide was able to bind to the MHC molecule; 11 indicates that the complex was stably eluted under anion-exchange conditions, 1 indicates that the refolding efficiency was 50 to 60% of that of 11 and that the complex was not stable under anion-exchange conditions, and 2 indicates that there was no elution peak 2 for the pMHC complex or that it had a much lower peak 2 (less than 80% of peak 1) at 78 to 84 mL in the measurement by Superdex 200 16/600 HiLoad size-exclusion chromatography. The anchor residues of the peptides are shown in bold. All of the peptides came from the C. parvum Iowa II CSL zinc finger protein and the Gp40/15 or Gp900 glycoprotein. T-Cell Epitopes in Cryptosporidium mBio the complexes dissociated under Resource-Q anion-exchange chromatography conditions at between 12 and 16 mS/cm (Fig. 4B). The AIF9 peptide could not bind with HLA-A*0201, and no standard chromatographic peaks appeared at 78 to 84 mL by size exclusion chromatography, such as the no-peptide negative control (Fig. 4C). Furthermore, we refolded and crystallized the H-2K b -SVF9 complex and solved its structure. H-2K b complexed with SVF9 was crystallized in the P1211 space group with a high resolution of 2.3 Å (Table 3). Within one asymmetric unit, there are two H-2K b molecules, referred to below as M1 and M2. The electron density map was clear for the peptide, indicating a stable and rigid conformation of the SVF9 peptide in the H-2K b cleft ( Fig. 5A and B). The comparison showed that the carbon backbones of the SVF9 peptide in M1 and M2 are not coincident, and the distances between the P4 and P5 atoms of the peptides could reach 1.6 Å and 1.2 Å, respectively (Fig. 5C). By comparing all of the other nonapeptides presented by mouse H-2K b molecules (Protein Data Bank [PDB] accession no. 1G7P, 1FZO, 1WBZ, and 2VAB), the M1 peptide conformation was found to be almost identical to that of the molecule under PDB accession no. 2VAB, a dominant CD8 1 T-cell epitope derived from Sendai virus nucleoprotein (39,40) (Fig. 5D).
Although the conformations of the SVF9 peptide in M1 and M2 are different, the accommodation of residues by the 6 pockets is roughly the same. The major proportions of P1, P2, P3, P6, and P9 in the SVF9 peptide are accommodated by the A, B, D, C, and F pockets, respectively. The SVF9 peptide bound to the H-2K b cleft via canonical primary anchors of small hydrophobic residues, P2 Val and P9 Leu, characteristics similar to those of H-2K b and HLA-A*0201. The complex structure of H-2K b exhibits a high degree of similarity with the octamer, as a secondary anchor residue (F or Y at position 6) is deeply buried in central pocket C of the H-2K b binding groove. In M1, there are 16 hydrogen bonds between the peptide-binding groove (PBG) and the SVF9 peptide ( Fig. 6A), but in M2, there are 19 hydrogen bonds and one salt bridge (Fig. 6B). Moreover, among the water molecules involved in the hydrogen bond network between the PBG and the SVF9 peptide, 5 water molecules form 5 direct hydrogen bonds with the SVF9 peptide in M1, but no water molecules are involved in M2 ( Fig. 6A and B). The main differences are found mainly in the middle of the groove. The P4 Ala and P5 Ile residues exhibit double conformations in the M2 molecule. The double main chains of P4 Ala form two hydrogen bonds with Asn70 and Arg155, but no interaction was found for the P4 Ala residue of M1 (Fig. 6B). In addition to hydrogen bonds, the interactions formed by van der Waals forces (VDWs) are also different in the two H-2K b H chains. The total numbers of VDWs are 147 and 160 in M1 and M2, respectively ( Table 4). The water molecules of M1 and more hydrogen bonds of M2 could fit into the different PBG conformations to stabilize the distinct peptide conformations. By comparing the peptide-MHC interactions of H-2K b and HLA-A*0201-YLQ-TCR (PDB accession no. 7RTR) (41), the peptide conformations of P1, P2, P3, and P9 are similar to that of HLA-A*0201, and the residues of peptide-MHC interactions are highly conserved, except for Arg155/Gln155 and Asn70/His70 (Fig. 6C). The hydrogen bonds of P1 Ser, P2 Val, P3 Phe, and P9 Leu (the N and C termini of PBG) of the SVF9 peptide are similar to those of a dominant severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2)derived epitope presented by HLA-A*0201, which can be recognized by a public human TCR (41) (Fig. 6D and E). We believe that the conserved SVF9 peptide can be stably presented by both H-2K b and HLA-A*0201 molecules.
Structure-based epitopes predicted in important Cryptosporidium species. It is noteworthy that some peptides that appear to have the correct peptide-binding motif T-Cell Epitopes in Cryptosporidium mBio still bind poorly. For example, peptide CSL-ALY9 (ALYDAYCIL), which has anchor residues that are present in high-affinity peptides (Leu at position 2 and Leu at position 9) (Table 1), binds very poorly to H-2K b and HLA-A*0201 by gel filtration and anionexchange chromatography ( Fig. 4C and Table 1). Furthermore, some peptides have T-Cell Epitopes in Cryptosporidium mBio been discovered to bind strongly to class I MHC molecules but cannot yet be recognized by CTLs, such as Gp900-VIM9 (VIMNPLFSL) ( Table 1 and Fig. 1). These observations indicate that the rules that predict binding affinity will be more critical.
According to the crystal structure of H-2K b with nonapeptides, the B, C, and F pockets accommodated the side chains of P2, P6, and P9, respectively, so they are critical for determining the peptide-binding motif of H-2K b . The B pocket is composed of Tyr7, Tyr22, Glu24, Val34, Tyr45, Glu63, Lys66, and Ala67 and shows obvious negatively charged polarity (Fig. 7A). There are two negatively charged residues (Glu24 and Glu63) in the B pocket, and Glu63 forms hydrogen bonds with the main-chain atoms of P2 V in the M2 molecule. The side chain of P2 V stretches to the a1 helix and is inserted into the B pocket, and it is surrounded by uncharged amino acids, except for Glu24 and Glu45 ( Fig. 7A and B). Based on our ELISpot and in vitro refolding assays, P2 I/V/M is suitable for binding the B pocket, and P2 R/A/G/S could also form stable complexes that survived the strong ionic buffer during anion exchange (42,43). The C pocket is composed of Val9, Asn70, Ser73, Phe74, Ser99, Gln114, and Tyr116. The main chain of P6 Phe forms two hydrogens with Asn70 and Ser73 of the a1 domain and shows hydrophilic and neutral properties ( Fig. 7C and D). P6 F/Y/R/P/L/N/S is suitable for binding the C pocket based on our results and resolved structures, as is P6 I/V/K/H, with similar properties (42,43). The F pocket is composed of the conserved residues Tyr84, Tyr123, Thr143, Lys146, and Trp147 as well as the less conserved residues Asp77, Thr80, Leu81, Ile95, and Ile124 ( Fig. 7E and F). Our results and those from previous studies confirmed that P9 L/M/V/I/A is the most suitable anchor residue for the F pocket of H-2K b . In addition, the upward-stretching side chains of P5(I) interact with the TCRs and form many VDWs with the positively charged amino acid Arg155, so P5 of the nonapeptide strongly prefers an amino acid with a large benzene ring or a long side chain. Thus, on the grounds of the results of our structural and functional studies, Based on the preliminary motif described above, we screened the main surface and apical complex protein sequences of C. parvum, C. hominis, and C. felis. There were 49 9-mer epitope peptides that were predicted to fit the motif, including Gp40/15, Cp15, Cp23, CSL, Gp900, and 70-kDa heat shock protein (HSP70). Forty-two of them were completely conserved in C. parvum and C. hominis, and 12 were completely conserved in all three species (Fig. 8; Table S3). HSP70, Cp23, Cp15, and Gp40/15 origin peptides are more conserved  than those of CSL and Gp900. HSP70 is highly conserved not only in C. hominis and C. parvum but also in many other Cryptosporidium species such as C. meleagridis, C. felis, C. canis, C. ubiquitum, C. cuniculus, C. viatorum, C. bovis, and C. xiaoi. None of the other peptides that we predicted here have been deposited in the Immune Epitope Database (IEDB).

DISCUSSION
Vaccination with C. parvum oocysts and Gp40/15-SVF9 synthetic peptides elicits peptide-specific CTL responses. Adaptive immune responses have been implicated as being important mechanisms of parasite-induced protection, and CD4 1 Th1 cells and CD8 1 cytotoxic T lymphocytes (CTLs) are two major subsets of cells mediating protective immunity (10,17,(44)(45)(46)(47)(48)(49)(50)(51). As described in detail in Results, splenocytes isolated from C. parvum-infected and peptide-immunized mice secreted IFN-g following ex vivo restimulation with soluble peptides (Fig. 1A). Two peptides (SVF9 and AIF9) induced significantly higher numbers of IFN-g spots in both total splenocyte cultures and CD8 1 T-cell cultures (Fig. 1B). The peptides induced significantly higher levels of IFN-g production in total splenocytes than in CD8 1 T-cell cultures, natural killer (NK) cells, dendritic (DC) cells, or CD4 1 T cells, as well as CD8 1 T cells as the most likely explanation (44,52,53). Vaccination with C. parvum CTL peptides in adjuvant elevated SVF9-specific CD8 1 T-cell responses but did not ensure long-term effectiveness in pro- is the observed intensity and ,I(hkl)> is the average intensity from multiple measurements. c CC1/2, percentage of correlation between intensities from random half-datasets. d R = R hkl kF obs j 2 k jF calc kR hkl jF obs j, where R free is calculated for a randomly chosen 5% of reflections and R work is calculated for the remaining 95% of reflections used for structure refinement. e RMSD, root mean square deviation. viding adequate protective immunity ( Fig. 3; see also Fig. S1 in the supplemental material). Using the natural Gp40/15 protein-flanking residues to extend the minimal CTL peptide to a 30-amino-acid (aa)-long peptide or supplementation with a CD4 1 Th peptide may rescue the function of SVF9-induced CD8 1 T cells and enhance the CTL response to C. parvum (54). Far from perfect, by utilizing IFN-g as a readout of our assays, we cannot formally exclude other cytokines or cellular markers that may serve as additional signatures of protection.
The SVF9 and AIF9 peptides overlapped and were derived from amino acids 313 to 321 and 316 to 324 of the Gp40/15 protein, respectively. Although similar in terms of their peptide sequences, with Val or Ile at position 2 and Leu at position 9, the AIF9specific CD8 1 T-cell response was relatively weak due to the different conformations at position 5 (P5). According to the crystal structure of H-2K b with the SVF9 peptide, P5 of the nonapeptide strongly preferred an amino acid with a large benzene ring or a long side chain, whereas the AIF9 peptide has a P5(A) without a side chain. To determine the primary role of the P5 residue, P5(I) of SVF9 was replaced by alanine (A), and the results confirmed our hypothesis. The substitution of the P5 residue cannot affect peptide binding as the P5 Ala mutant peptide could efficiently form a complex with H-2K b . The standard chromatographic peak appeared in the fractions at elution volumes of 78 to 84 mL (marked with red asterisks), similar to the wild-type SVF9 peptide (Fig. S2A). However, restimulation with the P5 Ala mutant peptide resulted in relatively weak IFNg production by CD8 1 T cells in an intracellular cytokine staining assay (Fig. S2B).
Structural analyses of human public TCRs recognizing the dominant SVF9 epitope presented by H-2K b . Based on our analysis and previous studies, the peptide-binding motif of H-2K b partly overlaps HLA-A*0201, one of the most prevalent HLAs in the T-Cell Epitopes in Cryptosporidium mBio global population, which also prefers L or M at position 2 and I or V at the C terminus (33). According to data from the Allele Frequency Net Database (http://www.allelefrequencies .net) (34), the HLA-A*0201 allele is widely distributed in South America, North America, Europe, Africa, and many other countries and regions. For example, in Bolivia and Peru in South America, the gene frequency is 0.6667, and 85.7% of individuals have the allele. Although the distributions of C. parvum subtype families in humans are distinct in different areas and under different socioeconomic conditions (1), C. parvum and C. hominis are responsible for over 90% of infections in humans, and they are responsible for almost all cryptosporidiosis outbreaks investigated (55). Can the dominant SVF9 epitope be recognized by human T cells? The molecular recognition of the TCR and pMHC is at the heart of immunology. Although the TCR is specific for the pMHC, it is also "cross-reactive" with many MHC molecules (56). Protein-protein interactions are governed mainly by van der Waals interactions, hydrogen bonds, and salt bridges, similar to TCRs and MHC molecules (56,57). With a model of the structure of the H-2K b human public TCR presenting the SVF9 peptide, based on modeling from the HLA-A*0201-public TCR complex (PDB accession no. 7RTR), we found that the Va domain of the TCR lies mainly over the a2 helix of MHC I, and the Vb domain lies mainly over the respective a1 MHC helices (Fig. 9A). Of note, the contacts of complementarity determining region (CDR), such as CDR1a, CDR2a, and CDR2b among the TCR-pMHC (H-2K b and HLA-A*0201) complexes were similar (Table S4). The conserved contacts between the TCR V regions and MHC helices are flat landscapes that are a secondary consequence of CDR3peptide interactions (56). The most variable regions of the TCR (CDR3) are positioned in the center of the binding interface where they contact the peptides, and the P5 residues of T-Cell Epitopes in Cryptosporidium mBio the peptides are surrounded by the CDR3a and CDR3b loops. The spatial conformation of P5(I) was found to be almost identical to that of P5(R) of HLA-A*0201-TCR (PDB accession no. 7RTR), and Asp109b and Asp110a in the conserved motif of the public TCR seem to exhibit a preference for recognizing the dominant SVF9 epitope (41) (Fig. 9B). This hypothesis has also been supported by functional studies with ELISpot assays and ICS ( Fig. 1 and 3). Obviously, the conformations of the prominently exposed P5(R) in the structure under PDB accession no. 7RTR, P5(I) in SVF9, P5(K) in CSL-MIW9, P5(Y) in Gp900-MIY9, and P5(K) in Gp900-KML all have relatively long side chains (Table 1), which can interact with polymorphic TCRs. However, the centers of the negative peptides are likely to be similar to those (PDB accession no. 1QEW, 1QR1, and 3MR9) presented by HLA-A*0201. Similarly, P5(P) in Gp900-VIM9, P5(V) in Gp900-SVS9, and P5(A) in CSL-ALY9 may have the same location and do not assume specific conformations to make stabilizing contacts with the TCRs (58) (Fig. S3).
Another interesting phenomenon is that the peptide in the M2 molecule of H-2K b -SVF9 exhibits double conformations (Fig. 5). Does the flexibility of P4 and P5 in the center of the SVF9 peptide increase immunogenicity? Perhaps in the context of a peptide that binds well, increased flexibility is more immunogenic, but in the context of a poorly binding peptide, increased flexibility does not increase immunogenicity (58). Notably, SVF9 could be an immunogenic epitope based on its ability to induce peptide-specific CD8 1 T-cell responses upon in vivo functional validation ( Fig. 1 to 3). The amino acid sequences of the SVF9 epitope are highly conserved in C. hominis, C. parvum, C. felis, and C. tyzzeri (Table 2), which will be useful for designing cross-protective vaccines against these species. In brief, we provide functional and structural analyses of the dominant C. parvum epitope presented by H-2K b and HLA-A*0201. Further work is needed to improve the immunogenicity of Gp40/15-based vaccines as well as other Cryptosporidium proteins involved in the infection and invasion of host cells, such as the rhoptry proteins and the MEDLE family of secretory proteins (59)(60)(61). Developing cross-protective vaccines for use in livestock and reducing oocyst shedding would offer new options to combat cryptosporidiosis and have a significant impact on public health.

MATERIALS AND METHODS
Animals and materials. C57BL/6 female mice were purchased from Beijing Vital River Laboratory Animal Technology Co., Ltd., People's Republic of China. The animal handling and experimental procedures were carried out in compliance with the recommendations of the guide for the care and use of laboratory animals of the Ministry of Health, People's Republic of China (62). The experimental protocol was  Prediction, synthesis, and treatment of epitope peptides. To identify the potential H-2K b allelespecific peptides from C. parvum, a computer-based program was used via an updated version of the NetMHCpan4.0 server (https://services.healthtech.dtu.dk/service.php?NetMHC-4.0) (32). The peptides were synthesized and purified to 90% purity by reverse-phase high-performance liquid chromatography (HPLC) with mass spectrometric analysis (SciLight Biotechnology, Beijing, People's Republic of China). After synthesis, the peptides were stored in lyophilized aliquots at 280°C and dissolved later in dimethyl sulfoxide. Peptides were diluted in phosphate-buffered saline (PBS) and tested individually at 10 mg/mL for the ELISpot assays and T-cell stimulations.
Macromolecule production and in vitro refolding assays. The recombinant proteins were expressed as inclusion bodies and purified as described previously (63). The pMHC I complexes (H-2K bmouse b2m (mb2m)-peptide and HLA-A*0201-human b2m (hb2m)-peptide) were prepared by refolding assays as described previously, with some modifications (64). Briefly, the MHC I heavy chain and b2m inclusion bodies were separately dissolved in a solution containing 50 mM Tris-HCl (pH 8.0) and 6 M Guanidine-HCl. The MHC I heavy chain (H-2K b or HLA-A*0201), mb2m or hb2m, and the peptide, at a 1:1:3 molar ratio, were refolded in refolding buffer (100 mM Tris-HCl [pH 8.0], 2 mM EDTA, 400 mM L-arginine HCl, 0.5 mM oxidized glutathione, and 5 mM reduced glutathione) at 277 K by gradual dilution. After 8 to 12 h of incubation at 4°C, the soluble portions of the complexes were concentrated and then purified by size exclusion chromatography on a Superdex 200 16/600 HiLoad column and by Resource-Q anion-exchange chromatography (GE Healthcare, Stockholm, Sweden).
Immunizations. Female C57BL/6 mice (6 to 8 weeks old; 6 per group) were housed under specificpathogen-free conditions at 25°C with autoclaved food and water provided ad libitum. Next, the mice   Table S4.
T-Cell Epitopes in Cryptosporidium mBio were divided into the following groups: the peptide-immunized groups, the oocyst infection group, the Freund's adjuvant control group, and the naive control group. C. parvum oocysts collected from infected calves were propagated in C57BL/6 interferon gamma knockout (GKO) mice (stock no. 002287; Jackson Laboratories, USA), which were kindly provided by Yurong Yang (Henan Agricultural University, Zhengzhou, People's Republic of China). C57BL/6 mice were subcutaneously immunized three times with the mixed peptide at 10-to 14-day intervals. The mixed-peptide vaccine administered to each mouse consisted of 50 mg of polypeptides dissolved in PBS and emulsified in complete Freund's adjuvant (CFA) for the primary immunization or incomplete Freund's adjuvant (IFA) for the booster immunization in a total volume of 200 mL. The Freund's adjuvant-immunized group was given 100 mL Freud's adjuvant and 100 mL PBS per mouse. Each vaccine solution was emulsified before administration. Two weeks after the last immunization, mice were challenged with 2 Â 10 6 oocysts for the quantitation of fecal parasite loads, four mice from each group were sacrificed, and their splenocytes and small intestine intraepithelial lymphocytes (IELs) were isolated for ELISpot assays and intracellular cytokine staining. ELISpot assay. The identification of the peptide-specific CTL epitopes was performed using the IFNg ELISpot assay (product code 3321-4HPT-2; Mabtech, BD Biosciences) according to the manufacturer's instructions. Single-cell suspensions were made from the spleens of the peptide-immunized, adjuvantimmunized, and naive mice. Erythrocytes were lysed by ammonium chloride treatment. The lymphocytes were seeded in ELISpot plates at 1 Â 10 6 cells/well and stimulated with each peptide at a final concentration of 10 mg/mL for 12-48 h at 37°C. Cells with nonspecific concanavalin A (ConA) (10 mg/mL) (Sigma) stimulation served as a positive control, and naive mice served as a negative control. All antibodies and reagents used for the ELISpot assay were obtained from Mabtech. An automated ELISpot reader (AID, Germany) was used to count the spots using RAWspot technology for multiplexing at the single-cell level. Spot-forming cells (SFCs) were adjusted by subtracting the average negative values and expressed as SFCs per 10 6 splenocytes. A positive response was defined as having at least 100 SFCs/10 6 input cells. Cells were plated in at least 3 replicate wells under each condition. The results were quantified as SFCs per 10 6 murine splenocytes (65).
Intracellular cytokine staining. Single-cell suspensions of murine IEL cells were isolated as previously described (66). Briefly, the intestines were removed, and the feces in the intestines were cleared by holding the intestines with forceps and flushing them with a syringe filled with calcium-and magnesium-free 1Â Hanks' buffered salt solution (HBSS) (Gibco). The pieces of the intestine were incubated in 15 mL of digestion solution (HBSS containing 5 mM EDTA and 1 mM dithiothreitol [DTT]) at 37°C for 20 min under slow rotation. Next, the cells were further purified on a discontinuous Percoll gradient. A total of 2 Â 10 6 IELs or splenocytes were prepared in 24-well plates and incubated with a final concentration of 10 mg/mL of peptides and 21 mg/mL of anti-CD28 (clone 37.51) in the presence of brefeldin A (eBioscience) for 5 to 6 h at 37°C with 5% CO 2 . Following stimulation, cells were first stained with BD fixable viability dye 510 and surface stained for 30 min with CD4 (clone GK1.5), CD3 (clone 17A2), and CD8 (clone 53-6.7) (BioLegend) antibodies. Next, the cells were fixed and permeabilized using BD Cytofix/ Cytoperm solution (BD Biosciences) and intracellularly stained with anti-IFN-g-phycoerythrin (PE) (clone XMG1.2) or mouse isotype IgG1-PE (clone MOPC-21) (all from BioLegend) for an additional 30 min. Samples were washed, resuspended in PBS, and acquired on a BD Fortessa instrument (BD Biosciences). Data were analyzed using FlowJo 10.8 (TreeStar, Inc.) and graphically represented using GraphPad Prism V8 (GraphPad Software).
Quantitation of fecal parasite loads. C57BL/6 mice were infected by gavage with 2 Â 10 6 C. parvum oocysts followed by peptide immunizations with CFA or IFA. Fecal samples were collected at 1 to 28 days postinfection. The samples were analyzed for parasite loads as oocysts per gram (OPGs) using a hemocytometer as previously described (67,68). Briefly, Cryptosporidium oocysts were isolated from fecal samples using Sheather's sugar flotation method, and sampling for each group was repeated three times to calculate an average value. The number of OPGs was estimated according to the number of oocysts counted and a curve diagram of oocyst excretion.
Crystallization and data collection. The SVF9 peptide of C. parvum (GenBank accession no. AAL07532.1) could form a stable complex with H-2K b following in vitro refolding and was selected for crystallization with the H-2K b heavy chain and mb2m. The peptide-H-2K b (pH-2K b ) complexes were ultimately concentrated to 10 mg/mL in a buffer containing 20 mM Tris (pH 8.0) and 50 mM NaCl for crystallization. The sample was mixed with reservoir buffer at a 1:1 ratio and crystallized by the sitting-drop vapor diffusion technique at 291 K. An index kit (Hampton Research, Riverside, CA) was used to screen for optimal crystal growth conditions. After several days, crystals of H-2K b complexed with the SVF9 peptide and mb2m were obtained with Index64 solution [12% (wt/vol) polyethylene glycol 3350, 0.05 M cobalt(II) chloride hexahydrate, 0.05 M nickel(II) chloride hexahydrate, 0.05 M cadmium chloride hydrate, 0.1 M HEPES]. Diffraction data were collected at a resolution of 2.3 Å at the Shanghai Synchrotron Radiation Facility (SSRF) (Shanghai, People's Republic of China) using Beamline BL18U at a wavelength of 1.5418 Å (69). The crystals were first soaked in reservoir solution containing 25% glycerol as a cryoprotectant and then flash-cooled in a stream of gaseous nitrogen at 100 K. The collected intensities were indexed, integrated, corrected for absorption, scaled, and merged using the HKL3000 package (70).
Structure determination and refinement. The crystal of pH-2K b belongs to the P1211 space group, and the structure was solved by molecular replacement using Molrep and Phaser in the CCP4 package, using the mouse H-2K b structure (Protein Data Bank [PDB] accession no. 1VAC) as the search model (71)(72)(73). Extensive model building was performed manually with Coot (74), and restrained refinement was conducted using REFMAC5 (75). Additional rounds of refinement were carried out using the Phenix refine program implemented in the Phenix package together with isotropic atomic displacement parameter refinement and bulk solvent modeling (76). The stereochemical quality of the final model was assessed with the PROCHECK program (77).
Data availability. The crystal structures have been deposited in the Protein Data Bank (https://www .rcsb.org) under accession no. 7WCY.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. FIG S1, TIF