Targeted inhibition of Wnt signaling with a bacterial toxin fragment 1 suppresses breast cancer tumor-initiating/chemo-resistant cells 2

BRCA1 germ-line mutations are a major cause of hereditary breast cancer and BRCA1-deficient breast cancer shares many characteristics as sporadic basal-like breast cancer (BLBC). Effective 35 therapeutic targets for BRCA1-deficient BLBC remain lacking. By utilizing a BRCA1-deficient 36 BLBC mouse model based on intraductal injection of Krt8-Cre adenovirus to inactivate Brca1 37 and Trp53 in luminal mammary epithelial cells, here we report that the Wnt receptor Frizzled 7 38 (FZD7) serves as a biomarker and therapeutic target in the resulting mammary tumor cells and is 39 particularly enriched in cancer stem cells / tumor-initiating cells (CSCs/TICs). Inhibiting FZD7- 40 mediated Wnt signaling using a nontoxic FZD-binding fragment of C. difficile toxin B (TcdB FBD ) 41 attenuates growth of BRCA1-deficient tumor organoids and xenografted tumors, without 42 damaging Wnt-sensitive tissues such as bones in vivo . Finally, FZD1/2/7-positive cells are 43 enriched in chemotherapy-resistant cells in both BLBC and luminal breast tumors treated with 44 cisplatin, and TcdB FBD synergizes strongly with cisplatin in inhibiting both tumor types. These 45 findings demonstrate the therapeutic value for targeting FZD1/2/7 in treating breast cancers and 46 establish TcdB FBD as a potential therapeutic agent targeting TICs and chemotherapy-resistant 47 cancer cells.

). Further analysis using CD24 and CD29 as markers revealed 146 that FZD7 + cells are largely considered basal MECs (~81.3%, Fig. 1E). Compared to the normal 147 mammary tissues, MMTV-PyMT tumor tissues contains similar levels of FZD7 + cells, whereas 148 both C3(1)-Tag and p53/BRCA1-deficient BL tumors, particularly the latter, contained elevated 149 levels of FZD7 + cells (~6.05% and ~10.4%, respectively, Fig. 1F and Supplementary Fig. S3C). 150 Together, these data suggest that FZD7 + cells are enriched in BLBC tumor models. 151 152 TcdB FBD inhibits FZD1/2/7-mediated Wnt signaling in human breast cancer cell lines 153 To evaluate whether FZD7 may serve as a therapeutic target for treating BLBC, we took 154 advantage of our recent finding that a bacterial toxin C. difficile toxin B (TcdB) recognizes FZD1, 155 2, and 7 as its high-affinity receptors and toxin binding inhibits FZD1/2/7-mediated Wnt 156 signaling (20). Our previous studies have also defined an FZD-binding domain in TcdB (amino 157 acid residues 1285-1804, designated as TcdB FBD , Fig. 2A) (21). Both TcdB FBD and Wnt 158 recognize the CRDs. FZD1, 2, and 7 form a sub-group with nearly identical CRDs within the 159 FZD family. The co-crystal structure of TcdB FBD in complex with CRD of FZD2 (CRD2) has 160 been solved, revealing that TcdB FBD effectively blocks Wnt signaling by targeting a region in 161 CRD that is critical for docking of the palmitate in Wnt (21). All Wnts are modified by lipidation 162 through the addition of a palmitoleic acid (PAM) to a conserved serine, which is essential for 163 their secretion and binding to FZDs. Binding of TcdB FBD prevents docking of the Wnt PAM into 164 a hydrophobic groove in CRDs. Key residues for TcdB FBD interactions are conserved in CRD1, 2, 165 and 7, but varies in other FZD members, which are the reasons for the selective high-affinity 166 binding of TcdB FBD to CRD1/2/7 (21,29). 167 We first tested the ability of TcdB FBD to inhibit Wnt signaling in a human TNBC cell line 168 MDA-MB-231 using a well-established TOPFLASH/TK-Renilla dual luciferase reporter assay. 169 It has been previously reported that FZD7 is upregulated in MDA- , and qRT-PCR 170 analysis confirmed that FZD1/2/7 are expressed at higher levels than other subgroups of FZDs in 171 this cell line ( Supplementary Fig. S4A). Wnt signaling in cells was stimulated using conditioned 172 medium containing WNT3A. Nanomolar levels of TcdB FBD inhibited WNT3A-mediated 173 signaling in a dose-dependent manner (Fig. 2B). This inhibitory effect sustained for over 72 h 174 with a single exposure to TcdB FBD in the medium (Fig. 2C). As a control, a mutant form of TcdB 175 (TcdB mu ) that could no longer bind FZD1/2/7, constructed by replacing the key CRD-binding 176 residues in TcdB FBD with the corresponding residues in C. difficile toxin A that does not use 177 FZDs as receptors ( 1595 VNFLQS changed to 1596 Fig. S4B). 181 To further evaluate the ability of TcdB FBD to inhibit Wnt signaling in vivo, we injected 182 MDA-MB-231 cells with an integrated TOPFLASH reporter subcutaneously into 183 immunodeficient athymic nude mice, which resulted in tumor growth. Recombinantly purified 184 TcdB FBD was subsequently injected intraperitoneally (i.p.) to these mice at a dose of 20 mg/kg 185 once per day for 6 times. To imaging the Wnt signaling activity, D-luciferin was injected 5 min 186 (i.p.) in vivo before tumors were dissected out, and their luminescence signals were then 187 measured ex vivo. Tumors from TcdB FBD -treated mice showed ~5-fold reduced signals compared 188 with tumors from vehicle-treated control groups and TcdB mu -treated groups (Fig. 2D). 189 We next evaluated the specificity of TcdB FBD for targeting FZD1/2/7, but not other 190 closely-related FZDs (e.g., FZD5). Two pancreatic cancer cell lines, 191 are known to express high levels of FZD5 and are sensitive to small molecule pan-Wnt signaling 192 inhibitor, LGK974 (32), which targets the O-acyltransferase Porcupine required for 193 palmitoylation of all Wnts (33). While LGK974 treatment (at a concentration of 100 nM) 194 inhibited clonogenic growth as well as the sphere formation ability of these two cell lines, 195 TcdB FBD treatment at a similar concentration (150 nM) exhibited no growth inhibitory effect on 196 them in both assays  198

TcdB FBD inhibits growth of BL mammary tumors 199
To test whether TcdB FBD could affect BL mammary tumors with FZD7 expression, we 200 first treated tumor organoids derived from the above-described murine models with TcdB FBD or 201 TcdB mu . Treatment with TcdB FBD , but not TcdB mu , reduced the size and number of organoids 202 formed from the p53/BRCA1-deficient or C3(1)-Tag BL tumor cells; in contrast, TcdB FBD 203 treatment did not affect growth of the organoids formed from MMTV-PyMT luminal tumor 204 organoids ( Fig. 3A-C). Furthermore, TcdB FBD -treated p53/BRCA1-deficient tumor organoids 205 exhibited reduced expression of Wnt signaling-related genes (e.g., Axin2, Rnf43) and EMT-206 related genes (e.g., Vim, Zeb1) compared with the control and TcdB mu -treated organoids 207 ( Supplementary Fig. S5A). 208 Next, we utilized TcdB FBD to evaluate whether inhibiting FZD7-mediated Wnt signaling 209 may offer any therapeutic benefits in vivo. Subcutaneous injection of p53/BRCA1-deficient 210 tumor organoid cells (1 x 10 4 cells) into nude mice resulted in robust tumor growth. When 211 tumors reached ~50 mm 3 , TcdB FBD was injected i.p. into the recipient mice. TcdB FBD 212 administration at a dose of 20 or 50 mg/kg with the intervals indicated in Fig. 3D, but not that of 213 TcdB mu , attenuated tumor growth, although the effect is modest (Fig. 3D). At molecular levels, 214 genes related to Wnt signaling and EMT were downregulated in tumors from TcdB FBD -treated 215 mice, compared with those in vehicle or TcdB mu -treated groups ( Supplementary Fig. S5B). 216 Consistently, expression of lymphocyte enhancer-binding factor 1 (LEF1), a representative Wnt 217 signaling effector, is reduced at protein levels ( Supplementary Fig. S5C). 218 To further evaluate whether TcdB FBD could attenuate growth of human BLBC/TNBC 219 cells, we took advantage of a recently established biobank of over 100 primary and metastatic 220 human breast cancer organoid lines (34). We chose two organoid lines from this biobank, 74T 221 and 86T, which represent a luminal and a BL breast cancer line with low and high level of a 222 BRCA1-deficiency signature (i.e., signature 3), respectively (34). The luminal organoid line 74T 223 was insensitive to TcdB FBD treatment ( Supplementary Fig. 6A). In contrast, growth of the BL 224 line 86T organoids was attenuated by TcdB FBD (Supplementary Fig. 6B). Similar to the 225 p53/BRCA1-deficient xenograft model, the inhibitory effect is rather modest. These results are 226 consistent with the idea that Wnt signaling is not a driving force but rather one of the 227 contributing factors in tumorigenesis of breast cancers. Nevertheless, these findings demonstrate 228 that selective inhibition of FZD7-mediated signaling is sufficient to exhibit an inhibitory effect 229 on growth of mouse and human BLBC cells. 230 231

TcdB FBD treatment does not affect the intestine and bones 232
We next analyzed whether inhibiting FZD7-mediated signaling by TcdB FBD at 233 therapeutically effective doses is tolerated in Wnt sensitive tissues such as the intestine and 234 bones. TcdB FBD was injected into mice at 20, 50, or 100 mg/kg twice a week for five weeks. 235 These mice showed similar weight gains comparable with the control mice (Fig. 4A). To 236 examine the potential impact on the intestine, we injected (i.p.) EdU, which incorporates into 237 replicating DNAs and marks proliferating cells, at the end of the fifth week. The intestine 238 epithelium constantly turns over and newly generated cells are produced from stem cells located 239 the bottom of the crypt region. Wnt signaling is a key pathway regulating intestinal stem cells 240 and a reduction in EdU incorporation would reflect an inhibition on stem cell activity (35,36). 241 The intestinal tissues were dissected out and EdU incorporation was measured. A shown in Fig.  242 4B, TcdB FBD treatment at 20 and 50 mg/kg levels did not reduce EdU levels in the intestinal 243 tissues. There appears to be a slight reduction at 100 mg/kg, but it did not reach statistical 244

TcdB FBD inhibits FZD7 + tumor-initiating cells in vitro and in vivo 257
FACS analysis demonstrate that ~90% of TcdB FBD -bound cells from primary 258 p53/BRCA1-deficient tumors were FZD7 + ( Supplementary Fig. S7A), confirming that TcdB FBD 259 targeted FZD7 + cells in tumor tissues. As Wnt signaling plays key roles in maintaining stem cell 260 activities, we next examined whether TcdB FBD affected the activity of TICs in the p53/BRCA1-261 deficient tumor, utilizing the mammosphere/tumorsphere assay that enriches TICs by culturing 262 cancer cells in suspension (38,39). A small percentage of these cells can survive and grow in 263 suspension as spheroids, which often correlates with their ability to form tumors in mice and thus 264 they are considered as TICs. Dissociated tumor cells from the p53/BRCA1-deficient model 265 easily formed spheroids (i.e., tumorspheres) in suspension culture ( Supplementary Fig. S7B). 266 FACS analysis revealed that ~85.7% of cells forming tumorspheres are FZD7 + , confirming that 267 most TICs express FZD7 ( Supplementary Fig. S7C). TcdB FBD treatment greatly reduced sphere 268 formation of p53/BRCA1-deficient mammary tumor cells (Fig. 5A). After TcdB FBD treatment, 269 the percentage of tumorsphere formation in the first passage was 1.2-fold reduced compared to 270 that of control, increased to 1.7-fold reduction at the second passage, and became even more 271 profound in the third passage (26.0-fold reduction) (Fig. 5B). This data indicate that TcdB FBD 272 inhibited the self-renewal potential of p53/BRCA1-deficient tumor cells. To further determine 273 the effect of TcdB FBD , limiting dilution assay was employed by serial re-plating of p53/BRCA1-274 deficient tumor cells at various cell concentrations in the sphere culture. The frequency of 275 sphere-forming cells, a surrogate for the frequency of TICs, was drastically reduced after 276 TcdB FBD treatment, dropping from 0.42% (1/239) to 0.11% (1/895) compared to control (P = 277 0.0019) (Fig. 5C). These data demonstrate that TcdB FBD targeted FZD7 + TICs and suppressed 278 their capability to generate tumorsphere in vitro. 279 To further assess the effect of TcdB FBD on TICs in vivo, we performed the limiting 280 dilution assay by transplanting ~10 -1,000 p53/BRCA1-deficient tumor organoid cells to nude 281 mice. FACS analysis confirmed that most of these tumor organoid cells were FZD7 + (~81.7%, 282 Supplementary Fig. S7D-E). TcdB FBD or TcdB mu treatment was given at 20 mg/kg twice a week 283 starting from day 5 after inoculation until day 53, when most mice had reached their endpoint 284 repopulating frequency can be calculated from a serial dilution of organoid cells (Fig. 5E). 287 TcdB FBD treatment reduced this frequency by 3.68-fold from 1/62 in control mice to 1/289.9, 288 whereas TcdB mu treatment did not change the frequency (Fig. 5E). For instance, transplanting as 289 few as 10 transplanted cells in the control or TcdB mu -treated group formed tumors 2 out of 8 290 times, whereas no tumor was detected in mice treated with TcdB FBD (Fig. 5D-E). Furthermore, 291 mice transplanted with 100 organoid cells and treated with TcdB FBD exhibited a significantly 292 longer tumor-free period than those received vehicle or TcdB mu treatment (average tumor-free 293 time 82.13 days vs. 44.37 days, P = 0.030) (Fig. 5F). Consistently, mice inoculated with 10,000, 294 1,000, or 100 p53/BRCA1-deficient tumor organoid cells and treated with TcdB FBD survived 1.2 295 days (P = 0.014), 14.84 days (P = 0.001), or 22.88 days (P = 0.017) longer than those treated 296 with vehicle or TcdB mu (Fig. 5G). Collectively, these results suggest that the inhibition of 297 TcdB FBD on tumor growth is mediated by its ability to target and suppress FZD7 + TICs in vivo. 298 299

TcdB FBD synergizes with cisplatin in treating both BL and luminal breast cancers 300
Even though TcdB FBD exhibited efficacy in inhibiting growth of BLBC via targeting FZD7 + 301 TICs, it is unlikely to eliminate tumors as a single agent. We thus examined its therapeutic value 302 in combination with a standard chemotherapy drug (e.g., cisplatin). Xenograft experiments with 303 the same initial number of tumor organoid cells from the two BLBC models (p53/BRCA1-304 deficient and C3(1)-Tag) were engrafted in nude mice. Mice were then treated with TcdB FBD 305 alone, cisplatin alone, or a combination of both agents following the schedule illustrated in Fig.  306 6A-B. TcdB FBD alone reduced tumor growth in both models and its efficacy is similar to 307 treatment with cisplatin alone (Fig. 6A-B). Combination treatment with both TcdB FBD and 308 cisplatin achieved the highest level of growth inhibition, close to complete suppression, than 309 either agent alone on tumors derived from p53/BRCA1-deficient and C3(1)-Tag models, 310 demonstrating a strong synergistic effect ( Fig. 6A-B). To further confirm the synergistic effect, 311 we also examined the sensitivity of tumor organoids of these two BL cancer models. A low 312 concentration of cisplatin (0.2 µM) did not reduce the size or number of tumor organoids, 313 whereas combining this level of cisplatin with TcdB FBD reduced both size and number and 314 organoids to a level lower than treatment with TcdB FBD alone ( Supplementary Fig. S8). These 315 results suggest that targeting FZD7-mediated signaling may suppress the resistance to cisplatin, 316 or cisplatin treatment may sensitize cells to FZD7-mediated inhibition by TcdB FBD . It is possible 317 that a subpopulation of FZD7 + TIC cells are intrinsically more resistant to chemotherapy, or 318 FZD7 is upregulated in a subpopulation of cells that developed resistance. Indeed, FACS 319 analysis of tumor cells revealed that the percentage of FZD7 + cells increased after treatment with 320 cisplatin alone in both p53/BRCA1-deficient and C3(1)-Tag models, and co-administration of 321 TcdB FBD greatly reduced FZD7 + cells (Fig. 6C). 322 In addition to the two BL cancer models, we also tested the luminal tumor MMTV-PyMT 323 model in xenograft models in vivo, which do not over-express FZD7 + and its organoids are 324 resistant to TcdB FBD in vitro (Figs. 1D, 3A). As expected, TcdB FBD alone did not affect tumor 325 growth in this model in vivo (Fig. 6D). Surprisingly, TcdB FBD was able to synergize with 326 cisplatin in this model as well and the combination greatly suppressed tumor growth than using 327 Wnt signaling is a key pathway in maintaining the stemness of cells. Besides well-341 established cancer-causing mutations in Wnt pathways, recent studies have also suggested that 342 Wnt signaling upregulation contributes to many aspects of cancer development and therapy 343 resistance, including EMT, metastasis, and resistance to chemotherapy and immunotherapy, 344 properties often attributed to CSCs/TICs. Thus, Wnt signaling represents a major therapeutic 345 target in cancer treatment (15). Broadly inhibiting Wnt signaling inevitably results in side effects, 346 diminishing the therapeutic value particularly for tumors that do not carry cancer-driving 347 mutations in Wnt pathways. Inhibiting a subset of Wnt signaling may reduce the side effect and 348 it needs to be done at the ligand-receptor level as this is the most diverged node in the pathway 349 (17), but it has been difficult to develop inhibitors that can distinguish different FZD subfamily 350 members. Furthermore, whether targeting only a subset of FZDs is going to be sufficient remains 351 to be demonstrated in vivo due to concerns on potential signaling redundancy. Here we utilized 352 the natural ability of a bacterial toxin to specifically recognize FZD1, 2, 7 subgroup and 353 developed a therapeutic protein that can block FZD1/2/7-mediated Wnt signaling. We found that 354 blocking FZD1/2/7-mediated Wnt signaling reduced growth of BLBC models in vivo and also 355 synergized strongly with the traditional chemotherapy agent cisplatin in treating both BLBC and 356 luminal breast tumor models. These findings demonstrate that inhibiting a single FZD subgroup, 357 instead of pan-inhibition of Wnt signaling, is sufficient to exhibit a strong synergistic effect in 358 vivo with the standard chemotherapy, and suggest a key role of FZD1/2/7-mediated Wnt 359 signaling in the development of chemotherapy-resistance in breast cancers. Targeted inhibition 360 of this pathway could be combined with chemotherapy to eliminate/prevent therapy-resistant 361 stem cell-like cancer cells. 362 Chemotherapy remains to be the cornerstone of the first-line treatment for many tumors 363 such as BLBC/TNBC that have no suitable targeted therapy. Conventional chemotherapy is 364 effective in controlling primary tumors, but often led to a selection and/or an induction of 365 CSCs/TICs that are more resistant to these agents, leading to eventual regeneration of chemo-366 resistant tumors and metastasis. Our findings that TcdB FBD synergizes strongly with cisplatin 367 treatment add to the wealth of literatures suggesting that Wnt signaling play a key role in the 368 development/selection of treatment-resistant cells and demonstrating that a combination of Wnt 369 signaling suppression with chemotherapy is a valid strategy to eliminate CSCs/TICs at once 370 and/or prevent the emergence of therapy-induced CSCs/TICs. 371 Synergistic effect from Wnt signaling inhibition has also been reported between OMP-372 18R5 and the chemotherapy agent taxanes on a range of tumor models (14). Future studies can 373 focus on testing whether TcdB FBD would also be effective in shrinking mammary tumors when 374 combined with other types of chemotherapeutic agents (e.g., doxorubicin) and whether combined 375 TcdB FBD and cisplatin treatment could be applied to other cancer types (e.g., serous ovarian 376 cancer, oral squamous cell carcinoma). Upregulation of Wnt signaling has been associated with 377 the resistance to many common chemotherapy agents as well as radiation therapy (12). The 378 mechanistic link between Wnt signaling and therapy resistance remains to be fully established 379 and is likely multifaceted (12). For instance, recent studies suggest that Wnt signaling enhances 380 the DNA repair pathways in ovarian cancers (40), and it has been also proposed that therapy-381 induced senescence promotes cancer cell stemness via Wnt signaling (41). In our study, it is 382 likely that cisplatin treatment leads to an epithelial cell fate change toward a mesenchymal state 383 (24), which may represent a dedifferentiation process that involves Wnt signaling. Lastly, Wnt 384 signaling has been shown to contribute to immune invasion as well as systemic inflammation in 385 the tumor microenvironment that drives cancer metastasis (42-45). Whether TcdB FBD may 386 provide a synergistic effect with immunotherapy such as PD-1/PD-L1 antibodies remains to be 387

explored. 388
Intestines and bones are two major tissues sensitive to disruptions in Wnt signaling 389 pathways and are often examined when testing Wnt signaling pathway inhibitors. The finding 390 that TcdB FBD did not affect intestinal tissues is consistent with previous findings that the pan-391 Wnt inhibitors LGK974 did not cause any damage to the intestine until a dose much higher than 392 the therapeutic dose (33). These findings are also supported by clinical observations that 393 intestinal damage is not a major side effect in clinical trials of OMP-18R5 antibody. On the other 394 hand, bone density loss is the single most important side effect observed from clinical trials of 395 OMP-18R5. Our analysis showed that bones were not affected in mice by TcdB FBD , which 396 represents a major advantage over other existing pan-Wnt signaling inhibitors and neutralizing 397 antibodies. This is possibly because both FZD7 and FZD5 are expressed in diaphyseal bone and 398 potentially play redundant roles (46). FZD5-mediated Wnt signaling in bone tissues might be 399 sufficient to maintain bone density when FZD7 is selectively inhibited by TcdB FBD , whereas the 400 neutralizing antibody OMP-18R5 inhibits both FZD1/2/7 and FZD5/8 subfamilies. 401 The specificity of TcdB FBD for FZD1/2/7 allows us to establish this agent as an effective 402 targeted therapy for FZD1/2/7 + TICs and cisplatin-resistant mammary tumor cells. The crystal 403 structure of TcdB FBD -CRD complex showed that the toxin targets the lipid docking site in CRD. 404 The sequence variations surrounding this lipid docking site across different FZDs underlies the 405 reason for the selectivity of TcdB toward FZD1/2/7 (21,29). These data suggest that this lipid 406 docking site may serve as a promising therapeutic target for developing effective and selective 407 Wnt signaling inhibitors. Although TcdB FBD is unlikely to be effective for blocking Wnt 408 signaling in cancer cells with mutations in Wnt pathway components downstream of the receptor 409 level (e.g., CTNNB1, APC), it should be able to inhibit Wnt signaling at the level of 410 ligand/receptor interaction in cancer cells (breast cancer or other cancer types) that become 411 dependent on Wnt ligands, in a way similar to Porcupine inhibitors (37), but is more selective 412 based on utilization of specific FZD receptors (e.g., FZD7) and is less toxic to normal tissues 413 (e.g., bone). These properties make this bacterial toxin-derived molecule a highly promising 414 agent for targeting FZD1/2/7-mediated Wnt signaling in vivo for cancer treatment. (residues 1285-1804) was cloned into pET28a vector, with an N-terminus HA tag. Mutated 419 TcdB FBD variants (TcdB mu ) were generated by two-step PCR and verified by DNA sequencing. 420 TcdB FBD and TcdB mu were expressed in E. coli strain BL21-Star (DE3) (Invitrogen). Bacteria 421 were cultured at 37°C in LB medium containing kanamycin. The temperature was reduced to 422 16°C when OD600 reached ~0.8. Expression was induced with 1 mM IPTG (isopropyl-b-D-423 thiogalactopyranoside) and continued at 16°C overnight. Proteins were purified using Ni2+-NTA 424 subcutaneously injected into immune-deficient athymic nude mice, when tumor length reached 526 10 mm, TcdB FBD or TcdB mu was injected at 20 mg/kg dose at indicated time. Five minutes before 527 sacrifice, 100 mg/kg D-Luciferin was given (i.p.), tumors were dissected, put into 24-plate, and 528 bioluminescent imaging were examined using a Xenogen IVIS-200 system (Xenogen). Images 529 were analyzed by quantification of total photon flux of each tumor using Living Imaging 530 Software. The p53/BRCA1-deficient tumor organoid cells were transduced by lentivirus with 531 constructs 7xTcf-FFluc and subjected to the same analysis as described for MDA-MB-231 cells. 532 533 TcdB FBD in vivo toxicity assay or EdU staining. Six weeks old female nude mice were injected 534 (i.p.) with the TcdB FBD (20, 50, or 100 mg/kg) or TcdB mu (20 mg/kg) twice a week for five 535 weeks. Mice were injected intraperitoneally (i.p.) once with EdU at the dose of 100 mg/kg body 536 weight 12 h before euthanization. Intestine tissues were cleaned with cold PBS, fixed in 4% 537 formaldehyde, and embedded in paraffin. Four-µm-thick sections were prepared, and the 538 intestine crypt proliferation was conducted using Click-iT ® EdU Alexa Fluor ® 594 Imaging Kit 539 (C10339, Invitrogen) according to manufacturer's introduction. Briefly, the sections were 540 washed twice with 3% BSA in PBS and permeabilized in 0.5% Triton ® X-100, then incubated 541 with a Click-iT™ reaction cocktail, followed by incubation in 5 μg/mL Hoechst 33342 according 542 to the manufacturer's protocol. 543 544 Micro-computed tomography (μCT). The right femur was extracted at euthanization and fixed 545 in 4% paraformaldehyde for 2 days, then transferred into 70% ethanol. A 2.6-mm span of the 546 distal femoral metaphysis was scanned on a desktop μCT (μCT-35; Scanco Medical AG) at 10-547 μm resolution using 50-kV peak tube potential and 151-ms integration time to 548 measure cancellous three-dimensional morphometric properties as previously described (48). Hepes, 50 μg/ml Primocin, 1% GlutaMAX, 1:50 B27, 1mM N-Acetylcysteine, 50 ng/ml EGF, 619 20 ng/ml FGF 2, 10 μM Y-27632, 100 ng/ml Noggin. TcdB FBD or TcdB mu (150, 300, 500, 1000 620 nM) was added to the culture medium every day for 10 days, PBS was used as a vehicle control. 621 The viability of organoids was assessed using the CellTiter-Glo luminescent cell viability assay 622 (# G7570, Promega) according to the manufacturer's protocol.