Programmed knockout mutation of liver fluke granulin, Ov-grn-1 , impedes 1 malignant transformation during chronic opisthorchiasis 2

Infection with the food-borne liver fluke Opisthorchis viverrini is the principal risk factor for 34 cholangiocarcinoma in the Mekong Basin countries of Thailand, Lao PDR, Vietnam, Myanmar 35 and Cambodia. Using a novel model of CCA, involving infection with gene-edited liver flukes in 36 the hamster during concurrent exposure to a nitrosamine, we explored the role of the fluke 37 granulin-like growth factor Ov -GRN-1 in malignancy. We produced programmed gene knockout 38 flukes ( ΔOv-grn-1) by delivery of a CRISPR/Cas9/gRNA system by electroporation. Genome 39 sequencing confirmed Cas9-catalyzed mutations in the targeted genes, which was accompanied 40 by rapid depletion of transcripts and the cognate proteins. Whereas Ov-grn-1 gene-edited 41 parasites colonized the biliary tract and developed into adult flukes, less hepatobiliary tract 42 disease manifested during chronic infection with ΔOv-grn-1 worms in comparison to hamsters 43 infected with control parasites. Specifically, immunohistochemical analysis of thin sections of 44 livers revealed markedly less periductal fibrosis surrounding the flukes and less liver fibrosis 45 globally during infection with ΔOv-grn-1 genotype worms, minimal biliary epithelial cell 46 proliferation, and markedly fewer mutations of TP53 in biliary epithelial cells. Moreover, fewer 47 hamsters developed high-grade cholangiocarcinoma when infected with the ΔOv-grn-1 flukes 48 compared to controls. The clinically-relevant, pathophysiological phenotype of the hepatobiliary 49 tract confirmed a role for this secreted growth factor in malignancy and morbidity during 50 opisthorchiasis. 51 52 53 54 55 56


Thailand 23
8 correspondence: Thewarach Laha,email,thewa_la@kku.ac.th;Alex Loukas,email,24 alex.loukas@jcu.edu.au;Paul Brindley,email,pbrindley@gwu.edu  Liver fluke infection caused by species of Opisthorchis remains a major public health problem in 73 East Asia and Eastern Europe. Infection with O. viverrini is endemic in Thailand and Laos,74 where ~10 million people are infected with the parasite. Opisthorchiasis is associated with 75 hepatobiliary diseases including cholangiocarcinoma [1], or bile duct cancer [2]. Northeast 76 Thailand reports the world's highest incidence of CCA, >80 per 100,000 in some provinces. No 77 stronger link between malignancy and a parasitic infection occurs than that between CCA and O. 78 viverrini. Indeed, the International Agency for Research on Cancer of the World Health 79 Organization classifies infection with O. viverrini as a Group 1 carcinogen, i.e. definitely 80 carcinogenic in humans [2][3][4]. 81 82 The mechanism by which opisthorchiasis induces CCA is likely multi-factorial, including 83 mechanical irritation of the biliary tract during migration and feeding of the liver fluke, 84 metabolites released by the parasite, and nitrosamines in fermented foods that are a dietary staple 85 in countries of the Mekong River basin. To survive in hostile environs, parasitic helminths 86 excrete and secrete a battery of proteins (excretory/secretory proteins [ES]) and other mediators 87 with diverse effects and roles at the host-parasite interface. This interaction has long been 88 thought to manipulate host cellular homeostasis and underwrite malignant transformation during 89 chronic opisthorchiasis, but the molecular mechanisms by which these processes occur are 90 poorly understood. Focusing on the ES contribution to carcinogenesis, we here targeted the 91 growth factor Ov-GRN-1, O. viverrini granulin, one component of the ES complement that we 92 have determined induces phenotypic hallmarks of cancer [5]. We have reported that Ov-  and other ES components including extracellular vesicles (EVs) enter cholangiocytes, the 94 epithelial cells that line the biliary tract, and drive cellular signaling that can promote 95 carcinogenesis, including cellular proliferation and migration, angiogenesis and wound healing 96 [6]. Also, we recently confirmed the role of Ov-GRN-1 in driving proliferation of bile duct 97 epithelial cells (cholangiocytes) by genetic manipulation of its expression in the liver fluke both 98 by RNAi and by CRISPR/Cas9 gene editing [7]. Moreover, we have shown that infection of 99 hamsters with gene edited infectious fluke larvae is feasible and, shown that proliferation of 100 biliary epithelia is markedly suppressed during infection with the Ov-grn-1-/-(ΔOv-grn-1) 101 flukes. 102 103 There is an established, tractable model of induction of CCA in experimentally infected 104 hamsters, where 100% of hamsters infected with metacercariae of the parasite and exposed to 105 otherwise sub-carcinogenic levels of dietary dimethyl nitrosamine ( To investigate the effect of programmed gene knock-out in O. viverrini, hamsters were infected 126 with NEJs of O. viverrini that had been subjected to CRISPR/Cas genome editing. The C 127 CRISPR/Cas systems were delivered by electroporation of plasmids encoding guide RNAs 128 specific for Ov-grn-1 or Ov-tsp-2 or a control (scramble -SCR) guide RNA and Cas9 of S. 129 pyogenes. Figure 1A summarizes the experimental approach and timelines, the findings from 130 which are presented below. 131 132 133 134

152
Changes in transcription of the targeted genes that would reflect programmed genomic changes 153 were monitored by RT-qPCR. Relative to the SCR group, the Ov-grn-1 transcript levels were 154 significantly reduced in the ΔOv-grn-1 flukes by 37.6% (P<0.05) whereas the transcription of 155 Ov-tsp-2 was not significantly changed in the ΔOv-grn-1 flukes (18%, ns) ( Figure 1B). 156 Transcription of Ov-tsp-2 was significantly reduced in the ΔOv-tsp-2 flukes (62.1%; P<0.01) 157 whereas the levels of Ov-grn-1 transcripts were not significantly changed ( Figure 1C). These 158 findings indicated on-target gene knockout at both the Ov-grn-1 and Ov-tsp-2 loci. 159 160 Effect of CRISPR/Cas-9 targeted mutation of Ov-grn-1 or Ov-tsp-2 on adult flukes 161 162 In experiment 1, feces were sampled from each hamster at both 10 and 12 weeks after infection. 163 Significant differences in fecal eggs (EPG) were not apparent among the three groups at these 164 timepoints ( Figure 2A Figure 2B). 169 170 Of the worms recovered from the livers, 10-13 from each of the three hamsters in each group 171 were evaluated by RT qPCR for gene transcript levels relative to the SRC (control) flukes. 172 Transcipt levels of both genes expressed by the control SCR parasites were generally clustered 173 around 100% ( Figure 2C, D). Transcript levels for Ov-grn-1 ranged broadly in the ΔOv-grn-1 174 flukes but were significantly reduced by 54% compared to the SCR group ( Figure 2C; P≤ 0.001). 175 This broad range, 100 -0%, in Ov-grn-1 transcript levels was similar to the findings in our 176 earlier report [7]. The outcome, where most flukes either not showing a change (~100%) or, by 177 contrast, showing a near absence of transcription (~0%) was not normally distributed, which 178 required a non-parametric statistical approach. Median values were reported for these analyses. 179 In contrast to the NEJ, which showed substantial knockdown of Ov-tsp-2 (Figure 1), the adult 180 stage ΔOv-tsp-2 flukes showed only a modest reduction in transcript levels (9%) compared to the 181 SCR group ( Figure 2D). Although transcript levels for Ov-tsp-2 were significantly reduced (P≤ 182 0.01), they generally showed minimal change with two individual flukes only exhibiting > 50% 183 reduction in expression. This marked difference, when compared to transcript levels for Ov-tsp-2 184 in NEJ --reduced by 62.1%, in addition to the reduction by 65% of numbers of worms recovered 185 at necropsy, indicated that Ov-tsp-2 gene knockout lead to a lethal phenotype and that flukes of 186 this genotype failed to survive in vivo. This contrasted with programmed knockout of Ov-grn-1 187 where the majority of flukes of the ΔOv-grn-1 genotype survived even though transcription of 188 Ov-grn-1 was not detected in four of 36 flukes and 10 of 36 exhibited Ov-grn-1 transcript levels 189 <5% of those expressed in the control group worms.    There was a broad range of EPG values among all the groups ( Figure 3A) although feces of the 230 SCR fluke infected hamsters had the highest EPG values, median =11,062 EPG, the ΔOv-grn-1 231 group, 5,267 EPG, and the ΔOv-tsp-2 group 4,530 EPG (P ≤ 0.05), reflecting the same rank 232 order in numbers of worms recovered from these groups ( the role of gene knockout on hepatobiliary disease and malignancy, the histopathological 238 assessments focused on the livers of hamsters with "moderate" EPG values, i.e. ΔOv-grn-1 EPG 239 ranging from 1,000 to 20,000 ( Figure 3A, B, dotted lines). This cutoff window excluded three 240 SCR hamsters with values of 26,000 to 44,000 EPG and four ΔOv-tsp-2-infected hamsters with ≤ 241 1,000 EPG. Figure 3B shows the "moderate EPG" values with 9,279 EPG in SCR-, 5,267 in 242 ΔOv-grn-1-and 5,434 in ΔOv-tsp-2-fluke infected hamsters with overlapping inter-quartile 243 ranges. Significant differences among the groups (within the cutoff window) were not apparent. 244 Whereas the surviving flukes did not show changes in expression of Ov-tsp-2 ( Figures 2D and  245 3D), the hamsters infected with ΔOv-tsp-2 parasites were included as a low fluke burden 246 comparator. It was likely that fewer flukes survived in the ΔOv-tsp-2 group of hamsters, 247 including those hamsters with comparable EPGs, (Figure 2A, B). 248 249 Adult worms recovered from hamster livers examined at 24 weeks after infection were evaluated 250 for targeted gene transcripts by qPCR. Transcript levels of Ov-grn-1 in adult O. viverrini from 251 ΔOv-grn-1 infected hamsters were significantly decreased (89%, P≤ 0.0001) compared to the 252 SCR control group ( Figure 3C), whereas the ΔOv-tsp-2 flukes showed a non-significant increase 253 (8%) compared to the SCR group ( Figure 3D). With experiments 1 and 2 showing the surviving 254 ΔOv-tsp-2 worms expressing Ov-tsp-2 at levels comparable to the control SCR flukes, we posit 255 that substantial Ov-tsp-2 gene edits were lethal, and worms that survived to maturity likely had 256 not undergone gene editing and/or few of the cells in the worms had been edited. Although we 257 retained the ΔOv-tsp-2 group for comparison of pathogenesis, the genotype of the flukes from 258 this group was not investigated further by targeted amplicon NGS. 259 260 Synopsis of outcomes of CRISPR/Cas9 gene editing of the liver flukes 261 262 To characterize mutations from the programmed knockouts, a region of 173 bp region flanking 263 the programmed cleavage site in Ov-grn-1 was investigated by analyzing reads from targeted 264 amplicon NGS and analysis of the reads with CRISPresso2 [14]. Substitution patterns as 265 determined by the CRISPR-sub tool [15] in reads in the knockout groups were not significantly 266 different from the cognate alleles in the control SCR group worms ( Figure S1). Also, we 267 scanned insertions and deletions (indels) and, in turn, the potential impact of indels on the open 268 reading frame. Figure 4A and Table S1 present the indel percentages of NEJs and adult flukes. 269 The ΔOv-grn-1 pooled NEJs showed 3.26% indel levels (2,723 of 80,571 reads), significantly 270 more than the 0.035% level in the SCR NEJ group (18 of 51,402 reads) (P≤ 0.05). The NEJ and 271 SCR adult flukes showed a similar indel % levels, with 0.045% in the adults (41 of 91,783 272 reads). Thirteen individual ΔOv-grn-1 adult liver flukes displayed a broad range of editing 273 efficiency in terms of indel profiles. These ranged from an apparent absence of programmed 274 mutation (no indels) to near complete gene knockout (91% indels), with a median of 3.1% 275 (MΔOv-grn-1), which was significantly higher than in the SCR group flukes (P≤ 0.01). As noted 276 for levels of transcription, however, there were apparently distinct groupings of low mutation 277 flukes, six worms, (termed LΔOv-grn-1) and six high mutation flukes (HΔOv-grn-1) observed. 278 From a total of ~1.5 million NGS reads, programmed deletions (~0.5 million) were 279 overwhelmingly more common than insertions, seven of which seven were identified (Table S1). 280 281 The ΔOv-grn-1 indel percentage showed a strong inverse correlation with the transcript level 282 ( Figure 4B), with a two-tailed non-parametric Spearman correlation co-efficient rs = -0.74 (P≤ 283 0.01). The highly mutated, HΔOv-grn-1 flukes only minimally expressed the targeted Ov-grn-1 284 gene, <11% level of transcription of control SRC liver flukes. However, the highly edited 285 genotype/highly reduced transcription phenotype contrasted with the wide range of transcription 286 in the flukes with low or moderate levels of editing, LΔOv-grn-1 and MΔOv-grn-1, with a wide 287 range of transcription from 6% to 94% of the levels of the control SRC group worms.

. Gene mutation rates among liver flukes. Panel A, CRISPR/Cas9 gene editing is highly effective in only some flukes. Insertion/deletion (indel) mutation rate ~60 base pairs either side of the CRISPR/Cas9 double stranded break from CRISPResso2 analysis of NGS data, plotted as a violin plot.
SCR NEJs and adult flukes are each from a single pooled sample, while ΔOv-grn-1 NEJs are from two biological replicate pooled samples, and ΔOv-grn-1 adult flukes from 13 individual adult flukes. The highly edited flukes are denoted as HΔOv-grn-1, flukes with low editing denoted as LΔOv-grn-1, and flukes with medium level editing denoted as MΔOv-grn-1. One sample t test for either NEJs or adults comparing ΔOv-grn-1 and SCR: * = P<0.05; ** = P<0.01. The thick solid line is the median and the inter-quartiles are black dashed lines. Note the broken Y-axis with a magnified lower portion to highlight the near zero values. B. Adult fluke indel mutation rate is inversely correlated with transcript level. The indel and transcript levels plotted for each individual ΔOv-grn-1 fluke (red circles, combining data from Figures 3C and 4A) showed that the highly mutated (>50%, HΔOv-grn-1) flukes generally showed low (<11%) Ov-grn-1 transcript levels, while low edited flukes (LΔOv-grn-1) generally showed higher transcript levels. Highly significant two-tailed non-parametric correlation determined by Spearman co-efficient, rs = -0.74 (**, P≤ 0.01). The negative linear correlation (slope -0.77) is shown as a red line with shaded 95% confidence intervals (r 2 =0.62). For context, the SCR indel% are plotted against the transcript median (blue triangle) with interquartile range error bars. C. The NEJ deletions (yellow) are shown from one pooled sample and no NEJ insertions were seen. Position -9 is highlighted with a vertical dotted line and the black horizontal square bracket (└─┘) highlights the pre-ORF Indel cluster at positions -1 to -10. The genetic sequence in proximity to this cluster is shown below the xaxis and the ORF initiator methionine sequence (ATG) is highlighted in red. The majority of Indels were 1-4 nt, but a single fluke was recorded with 13 large deletions of 31-62 bp (gray dashed oval). D. When successful Ov-grn-1 gene editing was seen, the majority occurred at a single location. The graph plots the Indel position against the frequency with each ΔOv-grn-1 adult Indel shown as individual marks: highly edited (HΔOv-grn-1, green +); medium edited (MΔOv-grn-1, purple +); low edited (LΔOv-grn-1, red x). Pooled samples are shown as ΔOv-grn-1 NEJs (orange circle), SCR NEJs (light blue triangle), and SCR adults (dark blue triangle). A vertical dotted line highlights the hotspot at nt -9 that comprised 98.5-99.9% of the observed indel% for successfully edited ΔOv-grn-1 flukes and NEJs. No LΔOv-grn-1 adults, SCR NEJs or adults showed editing at this position. This hotspot is within the pre-ORF indel cluster at positions -1 to -10 (black horizontal square bracket: └─┘). Other locations with frequently noted indels are labeled with the position number and corresponding nucleotide letter.

Evaluation of nucleotide position of indels 301 302
With respect to indel length and position, mutations were spread 29 bp upstream and 54 bp 303 downstream of the double stranded break (DSB+ (ORF nucleotide position -10 to +74). Most 304 indels were deletions of a single nucleotide, some were as long as four nt and the longest was 62 305 nt. Deletions were noted along the length of the amplicon, with several higher frequency sites 306 indicated with bubbles of greater diameter in Figure 4C. The pooled NEJ indel locations 307 reflected the high frequency adult locations ( Figure 4C, large bubbles). As noted, insertions 308 were infrequently seen but they tended to associate around these high frequency indel locations 309 in NEJs and adult flukes. Of note was the mutation cluster at -1 to -10 bp, within the 5' 310 untranslated region (UTR) of the ORF. Of the 216 indels observed in adult flukes, 48 (22%) 311 occurred in this region, and five specific sites included indels from 7 or more adult worms. 312 Figure 4D shows the frequency within each sample at each base pair position. Notably, there 313 were hotspot locations in the ORF at which both SCR and ΔOv-grn-1 NEJs and adults all 314 showed a mutation, albeit at low frequency. All four ACGT nucleotides were represented in the 315 hotspots. Whereas these hotspots occurred across all groups, they were infrequently seen, with 316 the most common ORF hotspot seen at position 9T in ~0.02% of the genomes sequenced. The 317 5'UTR cluster at nt position -1 to -10 was also a distinct grouping of mutations in this panel, 318 with no mutations at this location in SCR flukes and only a single LΔOv-grn-1 minimally edited 319 fluke in this 10 bp segment. While the cluster was of interest, the striking feature was the highly 320 mutated -9T position. This single position comprised ~99% of the total indel% for M/HΔOv-grn-1 321 flukes and ΔOv-grn-1 NEJs. Despite this single position being highly mutated and the 322 neighboring base pairs comprising the commonly ΔOv-grn-1 mutated cluster, the next highest 323 frequency position was -8C with only a small proportion of mutated cells recorded (<0.09% highlight the most severe pathogenesis from the 24 week infection ( Figure 5A, B). Liver 331 pathogenesis such as the pre-cancerous dysplasia stage were evident in many bile duct epithelia 332 and usually surrounded by fibrosis ( Figure 5C). Representative micrographs of the high grade 333 cancerous transformation from the three groups are shown in Figure 5D-F. Figure 5G and Table  334 S2 summarize the pathology findings in the three groups of hamsters. Ten of 12 (83.3%) 335 hamsters in the control SCR group were diagnosed with CCA, with high grade CCA in eight of 336 the 10 hamsters and low grade CCA in other two. Dysplasia, a precancerous lesion of CCA [16] 337 was diagnosed in the remaining two of the 12 (one mild, one moderate) hamsters in the SCR 338 group. CCA emerged in seven of 13 hamsters (six with high grade CCA) in the ΔOv-tsp-2 group. 339 Of the remaining hamsters, mild dysplasia was seen in two, proliferation in one, inflammation in 340 two, and the final hamster was free of hepatobiliary lesions. CAA was diagnosed in nine of 13 in 341 the ΔOv-grn-1 group, four with high grade CCA. Of the remaining hamsters, one showed 342 moderate dysplasia, two showed proliferation, and one was free of apparent hepatobiliary 343 lesions. Lesions less severe than dysplasia, i.e. inflammation, proliferation, or no lesions, were 344 evident in several hamsters infected with ΔOv-tsp-2 (4/13 hamsters) and ΔOv-grn-1 flukes (  Livers were harvested and a component of each lobe was prepared for sectioning or used for harvesting flukes for subsequent analysis (Fig. 1, Experiment 2)

358
There was no substantial difference in the location or subtype of CCA tumors among the 359 treatment groups. The CCA tumors observed ranged in sizes from small neoplasms with 360 multifocal distributions (10/24) to fully developed tumor masses visible to the unaided eye 361 (14/24) ( Figure 5). With respect to histological classification, 19 of 24 were of the tubular type 362 [2]. One papillary/cystic CCA was seen in the ΔOv-grn-1 group and as was the mucinous type 363 CCA in one SCR-and three ΔOv-tsp-2 O. viverrini-infected hamsters. The right lobe was the 364 frequent location of the tumor in all hamsters (20/26), with nine in the left lobe, and three in the 365 middle lobe. Occasionally, tumors had developed in more than one lobe (5/26) but usually 366 occurred in a single lobe (21/26). 367 368 The pattern of pathogenesis in "moderate EPG" animals was similar to that for the SCR animals, 369 showing dysplasia to CCA, with the most common pathogenesis of CCA in 78% of animals, and 370 100% of these were high grade ( Figure 5H). No animals were excluded from the ΔOv-grn-1 371 group and the 69% CCA rate was retained, of which 44% were high grade CCA. The ΔOv-tsp-2 372 group changed substantially and resembled the SCR group when only "moderate EPG" animals 373 were included. As with the SCR group hamsters, dysplasia was the minimum pathological lesion 374 and 78% had CCA, with most (86%) being high grade CCA. We assessed pathology in relation 375 to EPG levels ( Figure 5, panel I) and noted that low pathogenesis was observed in ΔOv-tsp-2 376 hamsters with low EPG (<1,000). Beyond the arbitrary threshold of 1,000 EPG, |no correlation 377 was apparent with level of pathology; the two SCR animals with the highest EPG recorded low 378 grade CCA and degrees of dysplasia seen across the EPG range. Although we posit that at 379 euthanasia 24 weeks after infection ΔOv-tsp-2 group hamsters were infected with fewer adult 380 stage liver flukes than the other groups, the similar pattern in hepatobiliary tract lesions in both 381 the SCR and ΔOv-tsp-2 groups with the moderate EPG phenotype indicated that the damage 382 caused by chronic opisthorchiasis was similar, at and beyond a threshold number of parasites. 383 Subsequently, we focused assessments of histopathology only on the hamsters with the 384 "moderate EPG" phenotype. 385 386 Reduced biliary tract fibrosis during infection with knockout parasites 387 388 Hepatic fibrosis was detected in Picro-Sirius Red (PSR)-stained thin tissue sections and enabled 389 investigation and quantification of peribiliary fibrotic development ( Figure 6A), using an 390 automated, visual quantitation. Fibrosis was evaluated, firstly on Ishak stage, a semi-quantitative 391 measure of the degree of fibrosis spread across the liver parenchyma [17], and secondly, on the 392 fibrotic deposition localized around the liver flukes, i.e. the amount of collagen deposition 393 surrounding bile ducts occupied by parasites at the time of necropsy. 394 395 The Ishak scores correspond to degrees of damaging liver fibrosis and the levels reflect fibrotic 396 expansion in periportal areas and the degree of bridging between portal regions, increasing up to 397 level 6 with full cirrhosis ( Figure 6B). The control (SCR) group was the most severely affected, 398 with the livers of all hamsters assigned an Ishak score of 4. This indicated that fibrosis had 399 progressed extensively with marked portal-portal and portal-central bridging. Hamsters in the 400 ΔOv-grn-1 group showed a median Ishak grade of 2 (range 2 to 4), with fibrosis in most portal 401 areas with/without short fibrous septa that had not bridged to other portal regions. Hamsters 402 infected with the ΔOv-tsp-2 liver flukes showed a median Ishak score of 3 (range 2 to 4). This is 403 defined as fibrous expansion of most portal areas with occasional bridging between them [17]. 404 The ΔOv-tsp-2 group was not significantly different to the control group, whereas the ΔOv-grn-1 405 group had significantly less fibrosis compared to either SCR (P≤ 0.0001) or ΔOv-tsp-2 (P≤ 0.05) 406 groups. 407 408 409 410

Representative images of hepatic fibrosis stained by Picro-Sirius Red with CRISPR-Cas9 edited O . viverrini from the 24-week CCA model. Fibrosis is denoted as pink/red thick bands around the bile ducts (periductal fibrosis, fb) and expands from each portal triad with fibrous septa. OV = Opisthorchis viverrini fluke, H = hepatocytes, BD = bile duct, BE = biliary epithelium. B. Global liver fibrosis plotted as a violin plot. Livers were scored for pathogenic fibrosis with an Ishak Stage
Grading scale and plotted on a violin graph that spans from zero (no fibrosis) to six (cirrhosis). C. Fibrosis proximal to flukes plotted as a violin plot. Automated ImageJ fibrosis analyzer software evaluation of the percentage of collagen deposition in images surrounding fluke-containing bile ducts. Panels B+C: median shown as thick colored line and dashed black lines mark the inter-quartile ranges. Comparing groups with Kruskal-Wallis test with Dunn's multiple comparisons against SCR: ns = not significant; **, P ≤ 0.05; ***, P ≤ 0.0001, and against ΔOv-tsp-2 group: #, P ≤ 0.05; ##, P ≤ 0.01.

411
To assess localized fibrosis specifically in the area where flukes were detected, we investigated 412 the degree of periductal fibrosis immediately proximal to live flukes in the bile duct lumen. 413 Automated analysis of the collagen deposition surrounding the bile ducts was undertaken using 414 an ImageJ driven fibrosis quantification tool of the PSR-stained collagen. Of the bile duct tissue 415 surrounding SCR and ΔOv-tsp-2 flukes, 14.55% and 14.56% of the tissue was fibrotic, whereas 416 12.66% of bile duct tissue surrounding ΔOv-grn-1 flukes was fibrotic ( Figure 6C, P ≤ 0.05). 417 418 Ov-grn-1 gene edited flukes drive less cell proliferation than control flukes 419 420 Here, we explored the in vivo effects as a consequence of programmed knockout of Ov-grn-1 421 expression, following our earlier reports which centered on proliferation of the biliary epithelium 422 and/or cultured cholangiocytes in response to in vitro exposure to Ov- [18][19][20][21]. 423 Proliferation of hamster biliary cells in situ was investigated using BrdU where the thymine 424 analogue BrdU is incorporated into cellular DNA. Visualization was performed using 425 immunohistochemistry ( Figure 7A) and evaluated quantitatively. Concerning worm survival, and 426 its corollary, the fitness cost of the programmed mutation, we examined proliferation but only in 427 the bile ducts where flukes were situated. Median proliferation in bile duct tissue surrounding 428 liver flukes in the SCR (15.0%) and ΔOv-tsp-2 (11.1%) groups were not significantly different 429 from each other, whereas only 3.1% of bile duct tissue surrounding ΔOv-grn-1 flukes had 430 incorporated BrdU (P<0.0001 vs SCR; P<0.05 vs ΔOv-tsp-2, Figure 7B). By contrast, the ΔOv-431 grn-1 group (3.1%) showed significantly less proliferation than both the SCR, 4.8-fold reduction 432 (P<0.0001) and the ΔOv-tsp-2 group, 3.6-fold reduction (P<0.05).  The fitness cost of gene knockout can be assessed from programmed gene editing, an approach 472 that is employed for the unbiased identification of essential genes in other organisms and disease 473 settings [28,29]. Our present findings confirmed the power of RNA-guided targeted mutation to 474 define essentiality and relevance of parasite proteins in infection-associated morbidity and 475 malignancy. The Ov-grn-1 gene does not appear to essential for in vivo development and 476

. Reduced cholangiocyte proliferation and p53 expression during infection with ΔOv-grn-1 genotype liver flukes. Representative images of proliferating biliary cells that incorporated BrdU from liver samples proximal to flukes in SCR, ΔOv-grn-1 and ΔOv-tsp-2 groups (A). The boxed region in the upper image is magnified in the lower panel. The brown arrow highlights the positive BrdUstained nuclei in each image and the blue arrow highlights a negative bile duct cell that did not incorporate BrdU. Violin graphs of BrdU index measured from cholangiocytes in bile ducts containing a fluke (B). Representative micrograph of p53 immunohistochemistry of biliary epithelium of DMNexposed hamsters infected with gene edited liver flukes (C). Anti-p53 antibody stained nuclei brown (brown arrows); unstained negative cells indicated using blue arrows. Black dashed box in upper wide-angle image is magnified in the lower image to aid visualization. Violin plot of p53 positive cholangiocytes as a percentage (D). Where available, 500 to 800 cells were scored from sections of each of the left, middle, and right lobes of the liver of hamsters with "moderate EPG", and marked by a grey "x". In a few samples, three ΔOv-grn-1 lobes and three
survival, which has enabled investigation here on the role of this protein in driving cell 477 proliferation, pathology and ultimately contributing to CCA in vivo. Nonetheless, the reduced 478 fecundity of DOv-grn-1 liver flukes likely reflected a fitness deficit as the result of targeted gene 479 knockout. By contrast, Ov-tsp-2 appears to be essential to parasitism, since the DOv-tsp-2 480 genotype did not survive, and sequencing of the indels across the relevant region of the genome 481 confirmed that most of the surviving flukes from hamsters had undergone minimal or no editing 482 of the Ov-tsp-2 gene. These findings build upon earlier RNA interference-mediated silencing of 483 Ov-tsp-2 gene expression and the corresponding malformation of the tegument observed in vitro 484 [30]. Although Ov-tsp-2 dsRNA-treated parasites were not used for in vivo studies, the damage 485 to the tegument was extensive and survival in vivo of worms damaged to that extent appeared to 486 be unlikely. ORF. This was unpredicted given that mutations were expected at the programmed DSB [7]. The 495 sequence of the 5'UTR of Ov-grn-1 does not exhibit identity to regulatory elements in the UTR 496 database, http://utrdb.ba.itb.cnr.it/, perhaps not unexpectedly given that few helminth parasite 497 UTR regulatory elements have been characterized [31]. Why this position was preferentially 498 mutated is unclear; however, the marked reduction of transcription of Ov-grn-1 that 499 accompanied this mutation profile, which was localized in the 5'UTR, suggests the presence of 500 an active regulatory control element. 501 502 Chronic liver fibrosis is a risk factor for liver cancer. The traditional lifestyle of people living in 503 O. viverrini-endemic areas, notably a diet enriched in nitrosamines as well as routine use of 504 alcohol, when coupled with the assault on the biliary epithelium by the attachment, feeding, 505 movement, and secretions of the liver flukes that result in repeated cycles of injury and repair 506 cycles, establishes a compelling setting for malignant transformation [32][33][34]. The secretion of 507 liver fluke granulin into the bile ducts and its ability to relentlessly drive cell proliferation and to 508 (re)heal wounds inflicted by the parasite plays a central role in this process [19]. Knockout 509 mutation of Ov-grn-1 did not substantially thwart infection, development and survival of the 510 liver fluke in vivo but infection with these DOv-grn-1 flukes failed to lead to the marked cell 511 proliferation and fibrosis in the immediate vicinity of the parasites, and consequently fewer 512 hamsters developed with high-grade CCA compared with the infections with the control SCR 513 and DOv-tsp-2 parasites. Indeed, more hamsters infected with DOv-grn-1 flukes were diagnosed 514 in the low-CCA and the proliferation categories than in the other treatment groups. Knockout 515 mutation of Ov-grn-1 attenuated virulence and impeded malignant transformation during chronic 516 opisthorchiasis. 517 518 Infected hamsters exhibited elevated rates of TP53 mutation although the level was markedly 519 less during infection with DOv-grn-1 flukes. The mutational signatures and related molecular 520 pathways characteristic of human CCAs have been reviewed in depth, and the signature profiles 521 differ between fluke associated and non-fluke associated CCAs [2]. Fluke associated CCAs 522 exhibit substantially more somatic mutations than non-fluke related CCAs [22], likely the 523 consequence of opisthorchiasis-associated chronic inflammation. In conformity with the human 524 situation, reduced inflammation and fibrosis were seen in hamsters infected with the ΔOv-grn-1 525 flukes, further emphasizing the virulence of this growth factor in chronic opisthorchiasis. 526 Inactivating mutations of TP53 are more prevalent in this subgroup, as are mutations of ARID1A, 527 ARID2, BRCA1 and BRCA2, than in non-fluke related CCAs [22,[35][36][37][38]. In addition, 528 hypermethylation has been noted for the promoter CpG islands of several other aberrantly 529 expressed genes [22]. 530 531 Mosaicism of gene knock-out is a limitation of our somatic gene-editing approach. Obviating 532 mosaicism by derivation of transgenic worms following germ line transgenesis is unlikely in the 533 near future in the context of this multicellular eukaryote, especially when overlaid on the genetic 534 complexity of a hermaphroditic platyhelminth parasite with a diploid genome. In addition, this 535 parasite has a multiple host developmental cycle, which cannot reliably be reproduced in the 536 laboratory. Nonetheless, somatic genome editing is of increasing utility in biomedicine, 537 including for treatment of hemoglobinopathies [39] and in the identification of targets for disease 538 interventions [40]. Given the role of Ov-GRN-1 as a virulence factor in opisthorchiasis, 539 interventions that target this growth factor might be beneficial. Indeed, antibodies raised against 540 recombinant Ov-GRN-1 block its ability to drive proliferation of CCA The CRISPR plasmid encoding a guide RNA (gRNA) complimentary to Ov-grn-1 exon 1 termed 568 pCas-Ov-grn-1 was constructed using the GeneArt CRISPR Nuclease Vector kit (Thermo Fisher 569 Scientific) as described [7]. The programmed cleavage site at nucleotide position 1589-1608, 5'-570 GATTCATCTACAAGTGTTGA with a CGG proto-spacer adjacent motif (PAM) which 571 determined the cleavage site located at three nucleotides upstream on Ov-grn-1 was designed 572 using the online tools, http://crispr.mit.edu/ [44] and 573 CHOPCHOP,http://chopchop.cbu.uib.no/ [45,46] using the Ov-grn-1 gene (6,287 bp, 574 GenBank FJ436341.1) as the reference. A second plasmid, termed pCas-Ov-tsp-2 was 575 constructed using the same approach; pCas-Ov-tsp-2 encodes a gRNA targeting exon 5 of the 576 Ov-tsp-2 gene (10,424 bp, GenBank JQ678707.1) [47,48]. The guide RNAs encoded by Ov-grn-577 1 and Ov-tsp-2 exhibited high on-target efficiency and little or no off-target matches to the O. 578 viverrini genome. A third construct termed pCas-Ov-scramble was also prepared and included as 579 a control to normalize analysis of gene expression and programmed gene knockout. The pCas-580 Ov-scramble construct included as the gRNA, a transcript of 20 nt, 5'-581 GCACTACCAGAGCTAACTCA which exhibits only minimal identity to the O. viverrini 582 genome and which lack a PAM [49]. A mammalian U6 promoter drives transcription of the 583 gRNAs in all three plasmids and the CMV promoter drives expression of the Streptococcus 584 pyogenes Cas9 nuclease (modified to include the eukaryotic nuclear localization signals 1 and 2). 585 To confirm the orientation and sequences of gRNA in the plasmid vector, Escherichia coli 586 competent cells (TOP10) were transformed with the plasmids, plasmid DNAs were recovered 587 from ampicillin resistant colonies using a kit (NucleoBond Xtra Midi,588 Düren, Germany), and the nucleotide sequences of each construct confirmed as correct by 589 Sanger direct cycle sequencing using a U6-specific sequencing primer. 590 591 Two hundred NEJs of O. viverrini were dispensed into an electroporation cuvette, 4 mm gap 592 (Bio-Rad, Hercules, CA) containing 20 µg pCas-Ov-grn-1, pCas-Ov-tsp-2 or pCas-Ov-scramble 593 in a total volume of 500 µl RPMI, and subjected to a single square wave pulse at 125 V for 20 594 ms (Gene Pulser Xcell, Bio-Rad). The NEJ were cultured in RPMI supplemented to 1% glucose 595 for 60 min after which they were used for infection of hamsters by stomach gavage (below). 596 597 Infection of hamsters with gene edited NEJs 598 599 Figure 1 provides a timeline of the CCA model, employed in experiment 1 and 2. In a first 600 experiment, nine male hamsters aged between 6-8 weeks were randomly divided into three 601 experimental groups (Figure 1). Each hamster was infected with 100 O. viverrini NEJs. These 602 NEJs had been transfected with pCas-Ov-grn-1 plasmid, pCas-Ov-tsp-2, or the control pCas-Ov-603 scramble, and assigned the following identifiers: delta(Δ)-gene name, ΔOv-grn-1, ΔOv-tsp-2, or 604 SCR, respectively. These NEJs had been transfected with pCas-Ov-grn-1 plasmid, pCas-Ov-tsp-605 2, or the control pCas-Ov-scramble, and assigned the following identifiers: delta(Δ)-gene name, 606 ΔOv-grn-1, ΔOv-tsp-2, or SCR, respectively. The hamsters were infected with NEJ by gastric 607 gavage The infected hamsters were maintained under a standard light cycle (12 hours dark/light) 608 with access to water and food ad libitum. At day 14 following infection, the water accessed by 609 the hamsters was replaced with drinking water supplemented with dimethylnitrosamine (DMN) 610 (synonym, N-nitrosodimethylamine) (Sigma-Aldrich, Inc., St. Louis, MO) at 12.5 ppm, which 611 was maintained until 10 weeks following infection [8][9][10]. Hamster feces were collected at weeks 612 10 and 12 after infection for fecal egg counts. The hamsters were euthanized at week 14 after 613 infection via isoflurane inhalation, livers were collected, and numbers of liver flukes counted. 614 615 In the second experiment, 45 male Syrian golden hamsters, Mesocricetus auratus, 6-8 weeks of 616 age, were randomly divided into three groups each with 15 hamsters and infected with gene 617 edited NEJs (above). These NEJs had been transfected with pCas-Ov-grn-1 plasmid, pCas-Ov-618 tsp-2, or the control pCas-Ov-scramble, and assigned the following identifiers: delta(Δ)-gene 619 name, ΔOv-grn-1, ΔOv-tsp-2, or SCR, respectively. The NEJ were introduced into the hamster 620 stomach using an orogastric tube. The infected hamsters were maintained under a standard light 621 cycle (12 hours dark/light) with access to water and food ad libitum. At day 14 following 622 infection, the water accessed by the hamsters was replaced with drinking water supplemented 623 with DMN at 12.5 ppm, which was maintained until 10 weeks following infection, as above. feces were collected and fixed in 10 ml of 10% formalin. Thereafter, the slurry of formalin-fixed 660 feces was filtered through two layers of gauze, and clarified by centrifugation at 500 g for 2 min. 661 The pellet was resuspended with 7 ml of 10% formalin, mixed with 3 ml ethyl-acetate and 662 pelleted at 500 g for 5 min. The pellet was resuspended in 10% formalin solution and examined 663 at 400´ by light microscopy. EPG was calculated as follows: (average number eggs ´ total drops 664 of fecal solution)/ gram of feces. To recover the adult liver flukes, food was withdrawn from the 665 hamsters 16 hours before euthanasia. At necropsy, terminal blood was collected from the heart 666 and allowed to clot at 4˚C for 30 min followed by centrifugation at 2,100 g for 10 min. The 667 clotted-blood serum was stored at -20˚C. Intact mature O. viverrini from the hepatobiliary tract 668 were recovered during observation of the livers using a stereo dissecting microscope and stored 669 for downstream gene-editing investigation. 670 671 Extraction of nucleic acids 672 673 Pooled NEJ or single mature worms from either experimental or control groups were 674 homogenized in RNAzol RT (Molecular Research Center, Inc., Cincinnati, OH) before dual 675 RNA and DNA extraction as described [7]. Briefly, the parasite(s) were homogenized in 676 Proliferation of biliary epithelial cells was investigated by using incorporation of BrdU. In brief, 722 the liver sections of a paraffin-embedded sample were soaked in xylene, rehydrated in graded 723 alcohol solution (100%, 90%, and 70% ethanol for 5 min each), and antigen was retrieved in 724 citrate buffer (pH 6) for 5 min in a high-pressure cooker. The tissue sections were blocked with 725 3% H2O2 in methanol for 30 min and subsequently incubated with 5% fetal bovine serum in 726 phosphate buffered saline for 30 min at room temperature (RT). The sections were incubated 727 with monoclonal mouse anti-BrdU (Abcam, catalogue no. ab8955) diluted 1:200 in PBS at 4°C 728 overnight, and then probed with goat anti-mouse IgG-HRP (Invitrogen, Thermo Fisher) diluted 729 1:1,000 in PBS for 60 min at RT. The peroxidase reaction was developed with 3, 3'-730 diaminobenzidine (DAB). Sections were counterstained with Mayer's hematoxylin for 5 min 731 before dehydrating and mounting. A positive signal was indicated by a brown color under light 732 microscopy. The image was captured by a Zeiss Axiocam microscope camera ICc5 and NIS-733 Element software (Nikon, Japan). To quantify BrdU-positive nuclei, cholangiocytes were 734 counted in 10 non-overlapping fields of 400x magnification, with a total of 1,000 biliary 735 cholangiocytes counted using the counter plug-in of ImageJ 1.52P. The cell proliferation index 736 was calculated as a percentage using the formula: positive biliary nuclei/total biliary cells 737 x100%. 738 739 Immunohistochemical staining for mutant forms of p53 740 741 To investigate levels of p53 mutation [60] in cholangiocytes, paraffin-embedded tissue sections 742 were deparaffinized and rehydrated by standard methods. Thereafter, sections were incubated 743 with monoclonal mouse anti-p53 (mutant, clone Ab-3 PAb240 catalogue no. OP29-200UG) 744 (Merck, Darmstadt, Germany) diluted 1:100 in PBS at 4°C overnight , and after thorough 745 washing, probed with goat anti-mouse IgG-HRP (Invitrogen, Carlsbad, CA) diluted 1:1,000 in 746 PBS for 60 min at 25°C. The peroxidase reaction was developed with 3,3'-DAB and sections 747 counterstained with Mayer's hematoxylin for 5 min. A human CCA cell line served as the 748 positive control for p53 positivity [61,62]. Images of high-power fields (400x magnification) of 749 the biliary epithelium were taken in five non-overlapping fields of each of the right, middle, and 750 left liver lobes using a Zeiss Axiocam fitted with a ICc5 camera and NIS-Element software 751 (Nikon). The percentage of mutant p53-positive cholangiocytes was determined by calculating 752 positive cells from 500 to 800 cholangiocytes from the right, middle, and left lobes of the liver 753 using imageJ. 754 755 PSR staining for fibrosis evaluation 756 757 Liver tissue sections where O. viverrini reside were selected for fibrosis measurements with PSR 758 (Abcam, catalogue ab150681). The thin sections were deparaffinized in xylene and rehydrated 759 through an ethanol gradient. PSR solution was applied to the sections and incubated at 25°C for 760 60 min. Excess dye was removed by washing twice in dilute acetic acid (0.5%) after which 761 sections were dehydrated through graded series of ethanol and xylene, the slides cleared with 762 100% xylene, mounted in Per-mount, and air dried overnight. Fibrosis around the bile duct 763 (periductal fibrosis, PF) proximal to the liver flukes was evaluated by two approaches: first, by 764 scoring according to accepted criteria [17,55]. Samples were blinded and fibrosis scores (0-6) 765 were graded semi-quantitatively by Ishak stage (Table 1)