Targeted CRISPR screens reveal genes essential for Cryptosporidium survival in the host intestine

Abstract

The Cryptosporidium parasite is one of the leading causes of diarrheal morbidity and mortality in children, and adolescent infections are associated with chronic malnutrition. There are no vaccines available for protection and only one drug approved for treatment that has limited efficacy. A major barrier to developing new therapeutics is a lack of foundational knowledge of Cryptosporidium biology, including which parasite genes are essential for survival and virulence. Here, we iteratively improve the tools for genetically manipulating Cryptosporidium and develop a targeted CRISPR-based screening method to rapidly assess how the loss of individual parasite genes influence survival in vivo. Using this method, we examine the parasite's pyrimidine salvage pathway and a set of leading Cryptosporidium vaccine candidates. From this latter group, using inducible knockout, we determined the parasite gene known as Cp23 to be essential for survival in vivo. Parasites deficient in Cp23 were able to replicate within and emerge from infected epithelial cells, yet unable to initiate gliding motility which is required for the reinfection of neighbouring cells. The targeted screening method presented here is highly versatile and will enable researchers to more rapidly expand the knowledge base for Cryptosporidium infection biology, paving the way for new therapeutics.

Fig. 1. In vivo CRISPR screen of the Cryptosporidium pyrimidine salvage pathway.

a Schematic illustrates how a knockout vector is generated. The first Golden Gate reaction occurs between a Cas9 expression plasmid and a 300 bp unique segment (containing the two 50 bp homology arms (one of which serves dual function as the gRNA), a unique DNA barcode, BsmBI restriction enzyme sites and the tracrRNA). The second Golden Gate reaction, using the BsmBI restriction enzyme sites, inserts a variable selection/reporter cassette to generate a complete knockout vector. The knockout vector then contains all the machinery to disrupt a gene of interest by inserting a variable selection cassette and barcode at the genomic locus. b A knockout vector targeting thymidine kinase (cgd5_4440) was generated and transfected into C. parvum sporozoites that were used to infect Ifnγ^–/– mice under paromomycin selection. Faecal samples were collected and the luminescence in faecal material was monitored. Data shows the mean faecal luminescence from a pooled cage sample ( ± SEM of 2 technical replicates), n = 4 mice. c Alignment of reads from whole genome sequencing of the thymidine kinase knockout strain to the C. parvum IOWAII genome at the site of insertion. Note the complete lack of alignment to the PAM site, which is removed by the homologous repair event. d Overview of CRISPR screening method. Following construction of KO vectors (detailed in 1a), sporozoites are transfected with gene specific KO vectors and used to infect mice. Specific barcodes are then amplified via high fidelity PCR and used to calculate fold enrichment (log2[%barcode(output) / %barcodes(input)]) which measures the relative fitness contribution of each gene. e Mouse faecal material was collected, and luminescence was monitored in the pooled sample from the pyrimidine salvage pathway CRISPR screens at 1 and 2 KO vectors per gene. Data shows the mean faecal luminescence from a pooled cage sample ( ± SEM of 2 technical replicates), n = 4 mice. f Rank ordered fold enrichment scores from the 1 and 2 KO vectors per gene pyrimidine salvage CRISPR screens. The colour indicates the relative fitness contribution, with dark purple showing high fitness conferring and dark green showing low fitness conferring. g Comparison of the fold enrichment scores between the 1 and 2 KO vectors per gene pyrimidine salvage CRISPR screens. Confidence refers to the inverse of the 95% confidence interval when comparing the log2 fold change scores from each screen (see methods).

Fig. 2. Validating screen results with diCRE mediated excision.

a, b TK-diCRE (shown in green) and ribonucleotide reductase (RNR) -diCRE (shown in purple) parasites were generated and used to infect an HCT8 monolayer in the presence or absence of rapamycin. Genomic DNA was extracted at 6, 24, and 48 h, and diagnostic PCRs confirmed the level of excision at the given time points. Data shown is representative of 2 biological replicates. c HCT8 cell monolayers infected with TK-diCRE and RNR-diCRE parasites in the presence or absence of rapamycin. At 6, 24, and 48 h the monolayers were fixed, stained, and the number of parasites per host nuclei was quantified. Box plot whiskers show the minimum and maximum values, the box shows the 25th to 75th percentile and the line shows the median. Representative images are shown, with nuclei in blue (Hoechst), and parasites in green (Vicia villosa lectin). Scale bar = 30 μm. Data shown is representative of 2 biological replicates. Significance was determined using a two-tailed unpaired t test. d TK-diCRE or RNR-diCRE parasites were used to infect Ifnγ^–/– mice, which were either treated with rapamycin or vehicle (DMSO) in their drinking water starting at day 2 post infection. Data shows the mean faecal luminescence from a pooled cage sample ( ± SEM of 2 technical replicates), n = 2 mice per condition. Note that a cell culture dish in the figure indicates an in vitro experiment, while a mouse silhouette indicates an in vivo experiment.

Fig. 3. In vivo CRISPR screen of Cryptosporidium vaccine candidates.

a Mouse faecal material was collected and luminescence was monitored during infection. Each replicate was conducted with 2 KO vectors per gene. Data shows the mean faecal luminescence from a pooled cage sample ( ± SEM of 2 technical replicates), n = 5 Ifnγ^–/– mice for screen 1 and n = 3 Ifnγ^–/– mice for screen 2. b Rank ordered fold enrichment scores from the vaccine candidate CRISPR screens. The colour indicates the relative fitness contribution of a gene, with dark purple being high fitness conferring and dark green being low fitness conferring. c Comparison of the fold enrichment scores between the replicate CRISPR screens for each gene. Confidence refers to the inverse of the 95% confidence interval when comparing the log2 fold change scores from each screen (see “Methods”). d Mouse faecal material was collected and luminescence was monitored during infection. Data shows the mean faecal luminescence from a pooled cage sample ( ± SEM of 2 technical replicates), n = 5 Ifnγ^–/– mice. Barcodes from KO parasites could be easily monitored over time (e) and within individual mice (f).

Fig. 4. Immunodominant antigen 23 is essential and required for reinvasion of host cells.

a, b HCT8 cell monolayers infected with Cp23-diCRE in the presence or absence of rapamycin. At 6, 24, and 48 h, gDNA was extracted and diagnostic PCRs confirmed the level of excision at the given timepoint. Data shown is representative of 2 biological replicates. c HCT8 cell monolayers infected with Cp23-diCRE in the presence or absence of rapamycin. At 6, 24, and 48 h, the monolayer was fixed and stained and the parasite per host nuclei was quantified. Box plot whiskers show the minimum and maximum values, the box shows the 25th to 75th percentile and the line shows the median. Representative images are shown, with nuclei in blue (Hoechst), and parasites in green (Vicia villosa lectin). Scale bar = 30 μm. Data shown is representative of two biological replicates. Significance was determined using a two-tailed unpaired t test. d. Ifnγ^–/– mice were infected with 50,000 Cp23-diCRE parasites and treated with rapamycin or DMSO control in their drinking water at 2 days post infection. Data shows the mean faecal luminescence from a pooled cage sample ( ± SEM of 2 technical replicates), n = 2 mice per condition. e Super resolution microscopy at 24 h post infection in vitro with Cp23-diCRE parasites in the presence or absence of rapamycin; green (helix pomatia agglutinin (HPA)), parasite; magenta, (sytox), nuclei; red (Cp23 Ab), Cp23. Scale bar = 2 μm. Data shown is representative of two biological replicates. f Expansion microscopy of the C. parvum asexual stages: sporozoite, merozoite, trophozoite and meront, and the sexual stages: macrogamont (female) and microgamete (male); green (helix pomatia agglutinin (HPA)); blue (N-hydroxysuccinimide (NHS ester); magenta (sytox); red (αCp23). Scale bar = 5 μm. Median expansion factor of 4.5. g Transmission electron microscopy of an excysted and unexcysted C. parvum sporozoite coupled with immunogold labelling with Cp23 antibody. Scale bar = 300 nm. h Permeabilisation assay of C. parvum sporozoites. The permeabilised condition used 0.1% Triton X-100 and the non-permeabilised condition used PBS. Grey (tryptophan synthase (αTrpB); green (Vicia villosa lectin); red (αCp23); blue (Hoechst). All images were taken using the same settings and exposure. Scale bar = 5 μm. Data shown is representative of two biological replicates. i HCT8 cell monolayers infected with either Cp23-diCRE or wildtype parasites in the presence or absence of rapamycin, and at 22 h post infection life stages were quantified. Data shows 2 biological replicates ± sd. n = 438 Cp23-diCRE - RAP, n = 173 Cp23-diCRE + RAP, n = 213 wildtype—RAP, n = 203 wildtype + RAP. j, k Live imaging of the reinvasion event was carried out at 18 h post infection of an HCT8 monolayer with Cp23-diCRE parasites in the presence or absence of rapamycin. All live microscopy data shown is from 2 biological replicates with n = 9 egress events recorded for each condition. Gliding distance was only measured for merozoites with clear movement following egress (n = 18 for DMSO and n = 5 for RAP). Movement of individual merozoites was tracked using Fiji software. Scatter plots show the mean ± sd with significance determined using a two-tailed unpaired t-test. Representative images are shown in (k) with movies in supplementary data. Scale bar = 8 μm. Note that a cell culture dish in the figure indicates an in vitro experiment, while a mouse silhouette indicates an in vivo experiment.