E3 ubiquitin-ligase Hakai induces LRP4 degradation and regulates Wnt/β-catenin signalling in colorectal cancer cells

35 The epithelial-mesenchymal transition (EMT) is closely linked to the acquisition of cancer stem cell (CSC) 36 properties, which contribute to treatment resistance and metastasis. This study investigates the role of 37 the E3 ubiquitin-ligase Hakai, the first identified post-translational regulator of E-cadherin stability, in 38 promoting CSC traits in colorectal cancer (CRC). To examine Hakai’s involvement in CSC regulation, we 39 used an inducible shRNA in a HT29 cells. Under conditions that promote CSC characteristics, we silenced 40 Hakai and evaluated tumoursphere formation and CSC marker expression. Proteomic and bioinformatic 41 analyses were performed to identify Hakai-regulated proteins in tumoursphere cultures. Additionally, 42 Western blot, RT-qPCR, co-immunoprecipitation, immunofluorescence and TOPFlash assays were 43 employed to study CSC-related protein regulation in response to Hakai expression. Furthermore, we 44 assessed the impact of Hakin-1, the pharmacological inhibitor specifically targeting Hakai’s HYB domain 45 responsible for its E3 ubiquitin-ligase activity, on tumoursphere properties. Hakai silencing significantly 46 reduced tumoursphere size and number accompanied by decreased expression of CSC markers and Wnt 47 target genes. CSC-related proteins regulated by Hakai were identified, including LRP4, a negative regulator 48 of Wnt/β-catenin pathway. Hakai interacts with LRP4, promoting its ubiquitination and degradation. 49 Moreover, Hakai overexpression induces hyperactivation of Wnt/β-catenin sand disrupts LRP4’s inhibitory 50 effect. Treatment with Hakin-1 effectively inhibited self-renewal and promoted differentiation within 51 tumourspheres. These findings suggest that Hakai promotes CSCs properties by hyperactivation of the 52 Wnt/β-catenin pathway via LRP4-mediated modulation. Additionally, Hakin-1 emerges as a promising 53 therapeutic agent targeting CSCs by enhancing differentiation and attenuating Wnt/β-catenin activity, 54 highlighting Hakai as a potential target for improving CSC treatment. 55 56

E3 ubiquitin-ligase Hakai; LRP4; Wnt/β-catenin pathway; cancer stem cell; colon cancer. 57 58 59 60 61 62 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 20, 2025. ; https://doi.org/10.1101/2025.11.20.689197doi: bioRxiv preprint 3

63 Colorectal cancer (CRC) is one of the most prevalent and lethal malignancies worldwide, characterized by 64 high heterogeneity and frequent resistance to therapy. A growing body of evidence indicates that cancer 65 stem cells (CSCs), a subpopulation of tumour cells with the ability to self-renew, differentiate, and tumour-66 initiating potential, play a critical role in tumour progression, metastasis, and resistance to therapy [1,2] 67 Therefore, understanding the molecular mechanisms that regulate CSC properties is essential for 68 developing more effective therapeutic strategies against CRC. Wnt/β-catenin signalling plays a critical role 69 during homeostasis and in various diseases, including cancer. Indeed, over 90% of colorectal cancers 70 involve hyperactivation of Wnt/β-catenin signalling [3,4]. The Wnt canonical pathway is dependent on β-71 catenin and TCF/LEF transcription factors, and primarily controls cell differentiation and proliferation. In 72 the absence of Wnt ligands, the free pool of β-catenin is phosphorylated and incorporated into the β-73 catenin destruction complex (composed of APC, AXIN1/2, CK1, and GSK3β), leading to its degradation by 74 proteasome. In the presence of Wnt ligands, the Wnt pathway is activated through Wnt binding to Frizzled 75 (FZD) and LRP5/6 receptors. This binding leads to the activation of the cytosolic effector protein DVL, 76 leading to inhibition of the destruction complex and thereby prevents β-catenin degradation. 77 Subsequently, β-catenin accumulates in the cytoplasm and the excess translocates to the nucleus. Here, 78 β-catenin activates TCF/LEF transcription factors, promoting the expression of Wnt/β-catenin target 79 genes including C-MYC, CCND1, AXIN2, MMP7 or LGR5 [5,6]. The activation of Wnt/β-catenin target genes 80 plays a crucial role in the initiation and maintenance of colorectal tumours and cancer stem cells (CSCs), 81 making the Wnt/β-catenin pathway an attractive therapeutic target for CSC eradication [7]. On the other 82 hand, LRP4 (also known as MEGF7), a member of the low-density lipoprotein receptor family, acts 83 predominantly as a negative regulator of the Wnt/β-catenin pathway, often antagonizing LRP6-mediated 84 signalling [8]. LRP4 function has been implicated in various neurological and developmental disorders. It 85 plays a critical role in embryonic development, with loss-of-function mutations associated to 86 malformation of limb, teeth, and kidneys [9–11]. It modulates both Wnt and BMP pathways, notably 87 through interactions with antagonists like WISE, DKK1, and SOST [12,13]. Beyond Wnt, LRP4 influences 88 other pathways, such as the PI3K/AKT pathway in epithelial–mesenchymal transition (EMT) [14]. 89 Importantly, while emerging data hint at its role in cancer biology, LRP4’s precise contribution to Wnt/β-90 catenin signalling in colorectal cancer and its specific effects on CSC are poorly understood. 91 Ubiquitination is the second most prevalent post-translational modification after phosphorylation. This 92 process involves the ATP-dependent covalent attachment of ubiquitin, a conserved protein composed of 93 76 amino acids, to a substrate protein through a cascade of enzymatic reactions involving the E1 activating 94 enzymes, E2 conjugating enzymes, and E3 ubiquitin-ligase enzymes. E3 ubiquitin-ligases are pivotal 95 regulators of protein stability and function, orchestrating the final step of the ubiquitination cascade that 96 targets specific substrates for proteasomal degradation or modulates their activity, localization, and 97 interactions [15,16]. Hakai was initially identified as an E3 ubiquitin-ligase that targets E-cadherin for 98 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 20, 2025. ; https://doi.org/10.1101/2025.11.20.689197doi: bioRxiv preprint 4 ubiquitination, leading to its endocytosis and lysosomal degradation, a process that disrupts adherens 99 junctions, facilitates cell–cell dissociation, and promotes EMT [17]. This classic function positioned Hakai 100 as a critical regulator of epithelial plasticity [18]. Subsequent studies by our group and others have 101 expanded its oncogenic profile, linking Hakai to enhanced tumour cell invasion, migration, and cellular 102 plasticity across various cancer types [19–21]. Structurally, Hakai contains a RING finger domain typical of 103 E3 ligases and a conserved Hakai HYB (Hakai pTyr-binding) domain, which mediates the recognition of 104 phosphorylated substrates such as E-cadherin. Notably, the development of Hakin-1, a selective small-105 molecule inhibitor of Hakai, has opened new avenues for therapeutic intervention by targeting its E3 106 ligase activity [22]. 107 The Wnt/β-catenin pathway is one of the key activators of the EMT, and it has been shown to play a role 108 in the formation and maintenance of CSCs across various carcinomas [18,23]. It plays a pivotal role in the 109 development and regeneration of the small intestinal epithelium, including the differentiation of Paneth 110 cells at the crypt base [24]. In the intestinal and colonic epithelium, activation of Wnt target genes 111 promotes the expansion of the stem cell compartment, whereas attenuation of Wnt/β-catenin signalling 112 facilitates cellular differentiation [25] A functional relationship has been established between EMT and 113 the acquisition of CSC properties, which is closely linked to therapy resistance, metastasis, and tumour 114 relapse. Despite these advances, the role of Hakai in regulating CSC properties and its potential functional 115 interplay with Wnt/β-catenin pathway remain largely unexplored. 116 In the present study, by using a colon cancer tumoursphere culture grown under conditions that promote 117 the induction of CSC-like properties, we demonstrate that silencing Hakai reduces both the size and 118 number of tumourspheres, accompanied by a decrease in CSC markers. Through proteomic and 119 computational analyses, we identify several CSC-related proteins, notably LRP4. We show that Hakai 120 interacts with LRP4, promoting its ubiquitination and subsequent degradation. Furthermore, Hakai 121 overexpression activates Wnt/β-catenin signalling and cooperates with LRP6 to overcome the 122 antagonistic effect of LRP4. Together, our findings uncover a novel regulatory mechanism by which Hakai 123 modulates Wnt signalling via LRP4 degradation, with potential implications for CSC maintenance and 124 colon cancer progression. Moreover, treatment with Hakin-1, a pharmacological inhibitor of Hakai, 125 impairs tumoursphere formation by reducing stemness-associated features, disrupting 3D structural 126 integrity, and promoting cellular differentiation, further supporting the therapeutic potential of targeting 127 Hakai in colorectal cancer. 128 129 130 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 20, 2025. ; https://doi.org/10.1101/2025.11.20.689197doi: bioRxiv preprint 5

131 Reagents and antibodies 132 The primary antibody used for Western blot and immunofluorescence assays were as follows: Hakai 133 (Invitrogen, 36-2800), LRP4 (Abcam, Ab174637), GAPDH (Invitrogen, 39-8600), LGR5 (Abcam, ab75850 134 and Invitrogen, MA5-25644), NANOG (Santa Cruz Biotechnology, sc-374001), Tubulin (Sigma-Aldrich, 135 T9026), Vinculin (Cell Signalling Technology, E1E9V), MUC2 (Cell Signalling Technology, CCP58), LC3 I/II 136 (Cell Signalling Technology, 4108), LRP6 (Cell Signalling, C47E12), V5 (Genscript, AØ1724 and Sigma, 137 V8137), HA (Roche, 12CA5), E-cadherin (BD Bioscience, 610182 and Abcam, ab11512) and PRPS2 (Abcam, 138 Ab96222). As secondary antibodies anti-mouse IgG (GE healthcare, NA934), anti-rabbit IgG (GE 139 healthcare, NA931), anti-mouse-IgG-488 (Life Technologies, A11001), anti-rat-IgG-488 (Life Technologies, 140 A11006), anti-mouse-IgG-594 (Life Technologies, A11005), anti-rat-IgG-568 (Life Technologies, A11077). 141 Proteasome inhibitor MG132 (Sigma-Aldrich) was added for 6 h using 10 µM and 30 µM. Lysosome 142 degradation inhibitor Chloroquine (Sigma-Aldrich), was employed for 24 h at 50 µM and for 6 h at 100 143 µM. Autophagy inhibitor 3-Methyladenine (Sigma-Aldrich) was added for 24h at 5 mM and 10 mM. 144 Compound Hakin-1 [4-(5-{[2-(4-nitrophenyl)-2-oxoethyl]thio}-1H-tetrazol-1-yl)benzoic acid] were 145 obtained from ChemBridge Corporation. 146 147 Cell culture 148 Human colon cancer cell lines HCT116 (ATCC®#CCL-247TM) were obtained from the ATCC and were grown 149 in DMEM. HEK293/T were cultured in Dulbecco's Modified Eagle's Medium (DMEM) - High Glucose 150 (Gibco). Human colon adenocarcinoma HT29 cell line was purchased from Sigma-Aldrich and cultured in 151 Mc Coy's Modified Medium (Gibco). HT29 cells with a doxycycline-inducible lentiviral system for sh CBLL1 152 was previously established by our group [26]. The doxycycline-inducible lentiviral system for CBLL1 153 (SMARTvector Inducible Lentiviral shCBLL1) was obtained from Dharmacon (Horizon Perkin Elmer Group). 154 Lentiviral particles encoding shRNA for CBLL1 or a non-targeting control (sh CONTROL) were generated 155 and propagated following established protocols for transduction into cancer cells. HT29 sh CONTROL and 156 HT29 shCBLL1 cells were cultured in McCoy’s 5A Medium with doxycycline (1 μg/mL; Sigma- Aldrich) to 157 induce shRNA expression. Following induction, the efficiency of Hakai ( CBLL1 gene ) knockdown was 158 verified by Western blot analysis. All culture medium were supplemented with 1% penicillin-streptomycin 159 (Gibco, 5000 U/ml) and 10% of Fetal Bovine Serum (FBS, Corning) for monolayer cultures. Cells were 160 cultured at 37 °C in a humidified incubator with a CO 2 concentration of 5%. All cell lines were 161 authenticated, used from early passages and monthly tested for mycoplasma contamination. 162 163 Tumourspheres formation assays 164 Cells were placed into ultra-low attachment 6-well plates (Costar 6-well Clear Flat Bottom Ultra-Low 165 Attachment Multiple Well Plates, Corning) at a density of 10 x 104 cells per well in a Stem Cell Medium (20 166 ng/ml human EGF, 10 ng/ml human b-hFGF, 1x B27 and 1x GlutaMax in DMEM/F-12 Media), and then 167 incubated at 37 °C with 5% CO 2 from 3 to 5 days. Flat bottom plates were used to avoid the formation of 168 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 20, 2025. ; https://doi.org/10.1101/2025.11.20.689197doi: bioRxiv preprint 6 cell aggregates instead of tumourspheres structures. Tumourspheres formation was monitored and 169 characterized by phase-contrast microscopy using an Eclipse-Ti microscope (Nikon). After the incubation 170 period, images of tumourspheres formed in each well were captured. To analyze tumoursphere size, area 171 of at least 15 tumourspheres per experiment was measured using the Freehand Tool in ImageJ software 172 (National Institutes of Health, USA). To quantify the number of tumourspheres, self-renewal assays were 173 conducted as follows: cells were seeded at a density of 500 cells/mL in 24-well low-attachment plates, as 174 previously described. After 5 days of culture, tumourspheres were analyzed by microscopy, and the 175 number of tumourspheres was counted. The size and number of tumourspheres were graphically 176 represented using GraphPad software. 177 178 Western blot analysis 179 Cellular proteins were extracted using lysis buffer containing 1% Triton X-100, 20 mM Tris-HCl (pH 7.5), 180 and 150 mM NaCl, supplemented with protease inhibitors: 10 μg/ml aprotinin, 10 μg/ml leupeptin, and 1 181 mM phenylmethylsulfonyl fluoride (PMSF). To prevent the activity of deubiquitinating enzymes during 182 ubiquitination and co-immunoprecipitation assays, 10 mM N-ethylmaleimide (Sigma-Aldrich) was added. 183 Protein concentration was determined using the Pierce™ BCA Protein Assay Kit (Thermo Fisher Scientific), 184 following the manufacturer’s instructions. Equal amounts of protein were resolved by SDS-polyacrylamide 185 gel electrophoresis (SDS-PAGE) and transferred onto polyvinylidene difluoride (PVDF) membranes 186 (Millipore). Membranes were blocked for 1 hour at RT under agitation in blocking buffer [5% non-fat dry 187 milk (Sigma-Aldrich) in TBS-T (20 mM Tris base, 150 mM NaCl, 0.1% Tween-20, pH 7.4)]. After blocking, 188 membranes were cropped and incubated with specific primary antibodies to the proteins of interest, 189 followed by incubation with appropriate HRP-conjugated secondary antibodies. Signal detection was 190 performed using the Luminata™ Crescendo Western HRP Substrate (Millipore) and visualized with the 191 Amersham Imager 600 system (GE Healthcare). 192 193 RNA extraction and quantitative real-time PCR 194 Total RNA was extracted using TriPure Isolation reagent (Roche) and converted to cDNA using the NZY 195 First-Strand cDNA Synthesis Kit (NZYTech). qRT-PCR was performed using LightCycler® 480 SYBR Green I 196 Master (Roche) on an LightCycler® 480 Instrument (Roche). Comparative CT method (∆∆CT method) was 197 performed to analyse qPCR data using β-ACTIN as a housekeeping control. Primer sequences are detailed 198 in Table S1. 199 200 Proteomic analysis 201 Equal amounts of each sample were reduced with 10 mM dithiothreitol (DTT) for 1 hour at 37 °C under 202 agitation. Subsequently, alkylation with 50 mM iodoacetamide (IAA) was performed in the dark for 45 203 minutes at RT. Samples were digested with sequencing grade-modified trypsin (Promega) at 1:30 enzyme-204 to substrate ratio. After 17 hours of digestion at 37 °C under agitation, samples were acidified with 10% 205 trifluoroacetic acid to ~pH 3 to stop the reaction and then, samples were concentrated using a under 206 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 20, 2025. ; https://doi.org/10.1101/2025.11.20.689197doi: bioRxiv preprint 7 speed-vacuum (SpeedVac, Thermo Fisher Scientific) and sonicated for 3 minutes. Subsequently, the 207 digested peptides were desalted using in-house made stage tips (3M Empore SPE-C18 disk, 47 mm, Sigma-208 Aldrich). 209 210 The dried eluates were reconstituted in water containing 0.1% FA for direct liquid chromatography-mass 211 spectrometry (LC-MS) analysis. 200 ng of peptide mixture were loaded in a nanoElute (Bruker Daltonics) 212 nano-flow LC that was coupled to a high resolution timsTOF Pro (Bruker Daltonics) with a CaptiveSpray 213 ion source (Bruker Daltonics). Liquid chromatography was conducted at 50 °C with a constant flow of 400 214 nL/minute on a reversed-phase column (15 cm x 75 m i.d.) with a pulled emitter tip, packed with 1.9 m 215 C18-coated porous silica beads (Dr. Maisch, Ammerbuch-Entringen, Germany). Chromatographic 216 separation was performed using a linear gradient of 5-35% Buffer B (100% acetonitrile and 0.1% FA) over 217 a period of 30 minutes, followed by an increase to 95% within 2 minutes and maintaining at 95% for 8 218 minutes. To prevent the sample carryover, two blank injections were performed after each sample run to 219 prevent any residual material. This involved a 30 minutes short gradient total run time (transitioning from 220 2% to 95% of Solvent B over 5 minutes; then from 95% to 5% B during another 5 minutes; maintaining 5% 221 B for 5 minutes, reaching 95% at 25 min, and finally holding at 95% for 5 minutes to wash the column). 222 Peptides underwent electrospray ionization (ESI) and were analyzed using data-dependent acquisition 223 (DDA) mode with parallel accumulation–serial fragmentation (PASEF) enabled. 224 225 Raw data files were analyzed to obtain protein identifications using MaxQuant and LFQ Analyst was used 226 for the analysis of Label-Free Quantitative (LFQ) proteomic data sets. Statistical analyses were carried out 227 with an in-house R script using the “MaxQuant “proteinGroup.txt” as the primary input file in conjunction 228 with an experimental design text file, which describes conditions and replicates. Contaminant proteins, 229 reverse sequences, proteins identified “only by site” or with “only by a single peptide” were removed. 230 LFQ intensities data were converted to a log2 scale, and replicates were grouped by conditions based on 231 the information provided in the “experimental design.txt” file. Missing values were imputed using the 232 “Missing not At Random” (MNAR) method, which uses random draws from a Gaussian distribution left-233 shifted by 1.8 SD (standard deviation) with a width of 0.3. To determine significantly regulated proteins 234 between conditions, a cutoff of an adjusted p-value cutoff of 0.05 (Benjamini−Hochberg method) along 235 with a fold change of 1.5 and a requirement 74 of at least two peptides was applied. 236 237 Plasmid transfection 238 Cell lines were transfected with plasmids pcDNA3.1, pcDNA-FLAG-Hakai and pBSSR-HA-Ubiquitin and 239 which were kindly provided by Yasuyuki Fujita (Hokkaido University, Japan). pcDNA-LRP4 were a kindly 240 gift of Tatsuo Suzuki (Shinshu University School of Medicine, Japan). Plasmids for TOPFlash, FOPFlash, TK-241 Renilla, Myc-LRP6, MESD and Myc-LGR5 were kindly provided by Madelon Maurice (Utrecht, 242 Netherlands). The transfection experiments were performed using FuGENE® 6 Transfection Reagent 243 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 20, 2025. ; https://doi.org/10.1101/2025.11.20.689197doi: bioRxiv preprint 8 (Promega) and Opti-MEM™ Reduced Serum Medium (Thermo Fisher), following the manufacturer’s 244 protocol. 245 246 Immunoprecipitation and ubiquitination assays 247 For immunoprecipitation experiments, cells were lysed using lysis buffer containing 1% Triton X-100, 20 248 mM Tris-HCl (pH 7.5), and 150 mM NaCl, supplemented with 10 μg/ml leupeptin, 10 μg/ml aprotinin, 1 249 mM PMSF and 10 mM N-ethylmaleimide for 30 minutes at 4 °C. After centrifugation at 18,000×g for 10 250 minutes, supernatants were immunoprecipitated with 2.5 μg of anti-LRP4 antibody or mouse IgG (Santa 251 Cruz Biotechnology, USA) at 4 °C for 2 hours, bound to 60 µL protein G PLUS-Agarose beads (Santa Cruz 252 Biotechnology). Immunoprecipitated proteins and input control protein lysates were then loaded into 253 SDS-PAGE gels to perform Western blot analysis with the indicated antibodies. 254 255 TOPFlash luciferase reporter assay 256 HEK293T or HCT116 cells were seeded into 24-well plates and grown overnight. The next day, cells were 257 transfected with 30 ng of the reporter plasmid TOPFlash or FOPFlash, 5 ng Thymidine Kinase (TK)-Renilla 258 and 50-150 ng of each of the indicated plasmids. Plasmids concentrations were 150 ng pcDNA-V5-Hakai, 259 50 ng pcDNA-LRP4, 100 ng MYC-LRP6 + 25 ng MESD and 100 ng MYC-LGR5, respectively. All transfections 260 were performed with FuGENE® 6 Transfection Reagent according to the manufacturer’s protocol. Six 261 hours post transfection, cells were incubated with control L-cell conditioned medium (LCM) or WNT3A-262 conditioned medium (WCM) overnight. WNT3a-conditioned medium (CM) were produced as described 263 [27]. After 24 hours, total cell lysates were extracted with Passive Lysis Buffer (Promega) for 20 minutes 264 at room temperature and levels of Renilla and Firefly luciferase was measured using the Dual-Luciferase 265 kit (Promega) according to manufacturer’s instructions using a Centro LB960 luminometer (Berthold). 266 267 Immunofluorescence assays 268 For immunofluorescence experiments of cells, HEK293T cells were seeded on laminin-coated glass 269 coverslips in 24-well plates at a density of 0.25 cm 2. After overnight incubation, cells were transfected 270 with 150 ng PCDNA-V5-Hakai, 100 ng PCDNA-LRP4 and/or empty vector PCDNA3.1 during 24 hours. 271 Transfection was carried out using FuGENE® 6 Transfection Reagent and with a DNA ratio of 5:1. The next 272 day, cells were fixed in 4% PFA for 30 minutes at room temperature. After the incubation, the reaction 273 was quenched in 0.05 M NH 4Cl for 15 minutes. Cells were blocked in PBS containing 2% BSA and 0.1% 274 Saponin. Primary and secondary antibody incubations were performed in blocking buffer for 1 hour and 275 45 minutes at room temperature, respectively . Cells were mounted in ProLong Gold (Life Technologies) 276 and left drying overnight in the dark before being analysed using an LSM700 confocal microscope. 277 To perform immunofluorescence assays of tumourspheres, tumourspheres culture were collected and 278 centrifuged. After washing with cold PBS, tumourspheres of each ultra-low attachment 6-well were 279 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 20, 2025. ; https://doi.org/10.1101/2025.11.20.689197doi: bioRxiv preprint 9 seeded in one well of an 8- well chamber (Millicell EZ SLIDE 8-well glass, Millipore) and they were fixed in 280 200 μl of 4% PFA in PBS for 40 minutes at room temperature, then permeabilized and blocked with PBD 281 0.2 T buffer [1% BSA, 1% DMSO, 0.2% Triton X-100 in PBS] for 1 hour. Primary antibodies were incubated 282 overnight at 4°C, followed by incubation with secondary antibodies for 3 hours in the dark. Nuclei were 283 stained with Hoechst 33342 (Life Technologies) 1:5000 diluted in PBS for 5 minutes and tumourspheres 284 were mounted for microscopy using ProLong Gold antifade reagent (Life Technologies). Images were 285 acquired with a Nikon A1R confocal microscope and analysed using Python with NumPy and Matplotlib 286 to measure fluorophore intensities. Results were quantified from five pictures in three replicate 287 experiments, expressed as mean ± SEM, and analysed by t-test for statistical significance. 288 289 Statistical analysis 290

are presented as mean ± SEM, as indicated in the figures. Statistical analysis and graphical 291 representations were performed using GraphPad Prism (Version 8, GraphPad Software). The Shapiro-Wilk 292 test was applied to assess the normality of the data and determine whether they followed a Gaussian 293 distribution. Statistical significance was evaluated using a t-test for comparisons between two groups, or 294 ANOVA for comparisons involving three or more groups. If the ANOVA revealed significant differences, 295 post-hoc tests were conducted to identify which groups differed significantly. Dunnett’s post-hoc test was 296 used when comparing multiple treatment groups to a single control. The results are presented in the 297 figures as *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001. 298 299 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 20, 2025. ; https://doi.org/10.1101/2025.11.20.689197doi: bioRxiv preprint 10

300 Hakai-silencing reduces colon cancer tumourspheres and modulates self-renewal and differentiation-301 related properties 302 To investigate the mechanistic insights of Hakai in modulating colon CSCs, we used HT29 tumourspheres, 303 a three-dimensional culture model derived from the human colon adenocarcinoma cell line HT29. This 304 model has been extensively used to study CSCs within the context of colorectal cancer [28]. HT29 305 tumourspheres serve as a valuable tool for advancing our understanding of CSC biology and developing 306 novel therapeutic approaches for targeting CSCs in colorectal cancer and other tumour types. Hakai-307 silencing was induced in an HT29 monolayer culture by using a previously reported inducible viral-308 transduced system by doxycycline [26]. After 72 hours, HT29 tumourspheres were induced for 5 days in 309 ultra-low attachment plates with a specific medium to enrich CSC growth, as shown in the schematic 310 workflow ( Fig. 1A ). This assay showed that Hakai-silencing reduces the number and size of the 311 tumourspheres ( Fig. 1B-D), further indicating the impact of Hakai on cancer cells self-renewal capacity 312 and proliferation. In accordance with previous reports [29], Hakai-silencing was accompanied by the 313 downregulation of stem cell markers and Wnt/β-catenin targets genes (Fig. 1E-F) including LGR5, the best-314 established CSC biomarker for colorectal, and NANOG transcription factor, a universal CSC marker. These 315 findings highlight the contribution of Hakai to the development of CSC tumourspheres, reinforcing its 316 potential relevance as a therapeutic target in colorectal cancer. 317 318 319 Proteomic and bioinformatic analysis identifies novel cancer-related candidates regulated by Hakai 320 In order to investigate the extent to which Hakai-silencing may control altered protein expression in CSC, 321 we decided to follow two different strategies: a proteomic study comparing Hakai-silencing in colon 322 cancer tumourspheres to control conditions, and consulting the UbiBrowser database that is able to 323 predict the proteome-wide human Hakai-substrate interaction network [30]. For the proteomic analysis 324 we used nano-LC-MS/MS analysis coupled to a timsTOF Pro mass spectrometer to identify novel Hakai-325 regulated proteins in colon CSC tumourspheres. The schematic workflow of the process is represented 326 (Fig. 2A). The confirmation of Hakai-silencing was shown at the protein level in three biological replicates 327 per experimental group analysed in the proteomic study ( Fig. 2B). Mass spectrometry analysis identified 328 a total of 3,602 proteins with ≥ 2 peptides. A total of 103 proteins (2.85%) were specifically enriched in 329 Hakai-silenced HT29 tumourspheres (+ Dox), whereas 118 proteins (3.28%) were specifically enriched in 330 control HT29 tumourspheres (- Dox) (Fig. 2C). The principal component analysis (PCA) and Volcano plot of 331 differentially proteins identified are shown ( Fig. 2D-E). Notably, the sum of the principal component PC1 332 (PC1, X-axis) and PC2 (PC2, Y-axis) captured the majority of the variance, approximately 80%, providing a 333 good representation of the variance in the samples ( Fig. 2D). The distinct clustering observed suggests 334 significant differences in the proteome between the two conditions. We conducted a hierarchical 335 clustering analysis of the differentially protein expression from three replicates comparing Hakai-silenced 336 to control tumourspheres, which indicates protein expression was similar in each group ( Fig. 2E ). As 337 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 20, 2025. ; https://doi.org/10.1101/2025.11.20.689197doi: bioRxiv preprint 11 represented by black dots in the Volcano plot analysis ( Fig. 2F ), eight proteins were significantly 338 upregulated (Fold change ≥ 1.5 and p-value  0.05) in Hakai-silenced tumourspheres, while three proteins 339 were significantly downregulated. The boxplot represents the regulated proteins in Hakai-silenced 340 tumourspheres compared to control tumourspheres (Fig. 2G), including MUC2, DDX55, CLMN, C19ORF43, 341 RABGGTA, SLC35F2, UBXN7, and UFSP2 that were significantly enriched in Hakai-silenced tumourspheres, 342 whereas NIT1, PRPS2, and WDR26 were significantly depleted. The complete dataset from the proteomic 343 analysis is shown in Table 1, including the biological functions of the identified proteins obtained from 344 Uniprot. Interestingly, several identified proteins are cancer-related proteins involved in CSC or 345 differentiation. However, due to the characteristics of the ubiquitination cascade, which involves transient 346 interactions characterized by rapid kinetics and weak target affinities, in general, E3 ubiquitin-ligase 347 substrates were not identified, making it challenging to identify and study their specific substrates. 348 Therefore, we used the UbiBrowser platform, which allowed us to determine the predicted Hakai-349 substrate interaction network [30]. By using UbiBrowser 2.0 [31], 116 proteins were identified as potential 350 predicted substrates for Hakai-mediated ubiquitination ( Table S2) [32]. We shortlisted the top 20 351 predicted novel substrates that could potentially be ubiquitinated by Hakai protein (in blue). Moreover, 4 352 already known substrates were identified, shown in red ( Fig. 2H). Thus far, very few substrates for Hakai 353 were biochemically or functionally reported, including E-cadherin, Cortactin, DOK1 or Ajuba and specific 354 substrates of Hakai in CSC conditions remain unknown [17,20,33]. 355 356 357 Gene Name Protein names Protein IDs Biological function involved from Uniprot database Fold change (Log2) Adjusted p-value Mol. weight [kDa] Peptide counts (all) Peptide counts (unique) Number of missing values imputed NIT1 Nitrilase homolog 1 Q86X76 Catalyzes the hydrolysis of the amide bond in N-(4-oxoglutarate)-L-cysteinylglycine (deaminated glutathione), a metabolite repair reaction to dispose of the harmful deaminated glutathione. Plays a role in cell growth and apoptosis: loss of expression promotes cell growth, resistance to DNA damage stress and increased incidence to NMBA-in duced tumors. Has tumor suppressor properties that enhances the apoptotic responsiveness in cancer cells; this effect is additive to the tumor suppressor activity of FHIT. It is also a negative regulator of primary T-cells. 2.99 0.00104 35.896 3 3 3 PRPS2 Ribose-phosphate pyrophosphokinase 2 P11908 Catalyzes the synthesis of phosphoribosylpyrophosphate (PRPP) that is essential for nucleotide synthesis. 1.99 0.0151 34.769 8 2 3 WDR26 WD repeat- containing protein 26 Q9H7D7 G-beta-like protein involved in cell signal transduction. Acts as a negative regulator in MAPK signaling pathway. Functions as a scaffolding protein to promote G beta:gamma- mediated PLCB2 plasma membrane translocation and subsequent activation in leukocyte s. Core component of the CTLH E3 ubiquitin-protein ligase complex that selectively accepts ubiquitin from UBE2H and mediates ubiquitination and subsequent proteasomal degradation of the transcription factor HBP1. Acts as a negative regulator of the canonical Wnt signaling pathway through preventing ubiquitination of beta- catenin CTNNB1 by the beta- catenin destruction complex, thus negatively regulating CTNNB1 degradation. Serves as a scaffold to coordinate PI3K/AKT pathway- driven cell growth and migration. Protects cells from oxidative stress-induced apoptosis via the down-regulation of AP- 1 transcriptional activity as well as by inhibiting cytochrome c release from mitochondria. Also it protects cells by promoting hypoxia-mediated autophagy and mitophagy (By similarity). 1.79 0.00241 72.123 2 2 3 DDX55 ATP-dependent RNA helicase DDX55 Q8NHQ9 Probable ATP-binding RNA helicase. -0.926 0.0476 68.546 5 5 0 UFSP2 Ufm1-specific protease 2 Q9NUQ7 Thiol-dependent isopeptidase that recognizes and hydrolyzes the peptide bond at the C-terminal Gly of UFM1, a ubiquitin- like modifier protein bound to a number of target proteins. Does not hydrolyze SUMO1 or ISG15 ubiquitin-like proteins. Through TRIP4 deufmylation may regulate intracellular nuclear receptors transactivation and thereby regulate cell proliferation and differentiation. -0.942 0.0433 53.261 4 4 2 CLMN Calmin Q96JQ2 Showing features for region, domain. -1.33 0.0279 111.65 6 6 0 SLC35F2 Solute carrier family 35 member F2 Q8IXU6 Putative solute transporter. -1.87 0.00241 41.211 2 2 3 MUC2 Mucin-2 Q02817 Coats the epithelia of the intestines, airways, and other mucus membrane-containing organs. Thought to provide a protective, lubricating barrier against particles and infectious agents at mucosal surfaces. Major constituent of both the inner and outer mucus layers of the colon and may play a role in excluding bacteria from the inner mucus layer. -1.95 0.0279 540.29 5 5 3 UBXN7 UBX domain- containing protein 7 O94888 Ubiquitin-binding adapter that links a subset of NEDD8- associated cullin ring ligases (CRLs) to the segregase VCP/p97, to regulate turnover of their ubiquitination substrates. -2.27 0.0433 54.862 3 3 4 C19orf43 Uncharacterized protein Q9BQ61 Exoribonuclease that is part of the telomerase RNA 3' end processing complex and which has the ability to all four unpaired RNA nucleotides from 5' end or 3' end with higher efficiency for purine bases. -2,33 0.0279 18.419 2 2 3 RABGGTA Geranylgeranyl transferase type-2 subunit alpha Q92696 Catalyzes the transfer of a geranylgeranyl moiety from geranylgeranyl diphosphate to both cysteines of Rab proteins with the C-terminal sequence -XXCC, -XCXC and -CCXX, such as RAB1A, RAB3A, RAB5A and RAB7A. -2.37 0.00804 65.071 2 2 2 358 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 20, 2025. ; https://doi.org/10.1101/2025.11.20.689197doi: bioRxiv preprint 12 Table 1. Identification of Hakai regulated proteins in Hakai-silenced HT29 colon cancer tumourspheres. 359 Statistically significant identified proteins by iTRAQ analysis of Hakai-silenced HT29 colon cancer 360 tumourspheres compared to control tumourspheres. The selected unique proteins found identified with 361 fold change >1.5 and a cut-off of adjusted p-value < 0.05. 362 363 Hakai interacts with LRP4 thereby inducing its ubiquitination and degradation 364 The predicted novel Hakai substrates identified by UbiBrowser included several cancer-related proteins. 365 Given that previous studies have reported Hakai is involved in EMT and that our results support Hakai 366 implication in CSC maintenance (Fig. 1), we directed our focus towards proteins involved in CSC regulation. 367 Interestingly, our top four protein, LRP4 is a member of the low-density lipoprotein receptor (LRP) and 368 can act as negative regulator of Wnt signalling [10]. LRP4 is primarily known in neuromuscular junction 369 development and bone formation [10,12], however the significance in CSC is not well understood. To 370 investigate the potential link between Hakai and LRP4, FLAG-Hakai was transiently transfected into 371 HEK293 and colon cancer HCT116 cells, showing that Hakai overexpression reduced LRP4 protein levels in 372 both cell lines ( Fig. 3A ). Next, immunofluorescence analyses were conducted using HEK293 cells 373 transiently overexpressing V5-Hakai in presence or absence of overexpressed LRP4. Hakai overexpression 374 alone was detected mainly in the nucleus but also in the cytosol, consistent with previous reports [34]. 375 LRP4 was located at cell membrane, however, when both proteins were overexpressed, LRP4 expression 376 was decreased (indicated by arrows) ( Fig. 3B ), indicating that Hakai may regulate LRP4 expression. As 377 previously mentioned, the absence of co-localization of both proteins may be attributed to transient 378 interactions between Hakai and LRP4. Given that Hakai overexpression downregulates LRP4 protein 379 levels, and considering that LRP4 is a predicted new substrate for Hakai by UbiBrowser, we investigated 380 whether we could detect a protein–protein interaction by immunoprecipitation. As co-381 immunoprecipitation using endogenous proteins was challenging, FLAG-Hakai, LRP4, HA-ubiquitin were 382 transiently transfected in HEK293T cells and LRP4 antibody was used for immunoprecipitation. As 383 expected, co-immunoprecipitation was observed for Hakai and LRP4 when overexpressed, further 384 confirming Hakai and LRP4 interaction ( Fig. 3C). As previous studies reported Hakai as an E3 ubiquitin-385 ligase that induces ubiquitination and subsequent degradation of its substrate, we predicted that LRP4 386 could act as a substrate for Hakai. To prove this hypothesis, we carried out an ubiquitination assay 387 following the immunoprecipitation of LRP4. Western blot analysis showed that Hakai overexpression 388 induces LRP4 ubiquitination, shown as a HA-Ub smear in LRP4 immunoprecipitates ( Fig. 3D ). Taken 389 together, these results show that Hakai interacts with LRP4 and induces its ubiquitination. Then, we 390 further investigated the mechanism by which LRP4 degradation was achieved by Hakai. We analysed the 391 effect on LRP4 protein expression in presence or absence of the proteasome inhibitor MG132. We 392 observed that LRP4 protein levels were increased in presence of MG132, without affecting the levels of 393 Hakai (Fig. 3E). However, this effect was not observed in presence of lysosome inhibitor Chloroquine nor 394 in the presence of the autophagy inhibitor 3-MA ( Fig. S1 ). Notably, MG132 treatment in Hakai-395 overexpressing cells resulted in the restoration of LRP4 levels ( Fig. 3F ), further supporting that the 396 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 20, 2025. ; https://doi.org/10.1101/2025.11.20.689197doi: bioRxiv preprint 13 induction of LRP4 degradation by Hakai occurs, at least partially, in a proteasome-dependent manner. 397 Moreover, treatment of cells with the previously reported Hakai inhibitor, Hakin-1 that specifically blocks 398 the HYB domain responsible for the E3 ubiquitin-ligase activity of Hakai, was also able to rescue 399 downregulation of LRP4 induced by Hakai in HCT116 cells ( Fig. 3G). Thus, our results support that Hakai 400 mediates LRP4 degradation through its ubiquitin-ligase activity, further supporting LRP4 as a novel 401 substrate for Hakai. 402 403 Hakai counteracts the antagonistic effect of LRP4 on LRP6-mediated Wnt pathway activation. 404 LRP4 is a transmembrane protein of the LDL receptor family and functions as a negative regulator of the 405 canonical Wnt/β-catenin signalling pathway by antagonizing LRP6 receptor [8]. Considering the crucial 406 role of LRP4 in development and its implication in the Wnt/β-catenin pathway, we aimed to investigate 407 the potential mechanism by which Hakai may influence Wnt/β-catenin signalling. Moreover, the effect of 408 Hakai on key Wnt-related proteins was extended to LRP6. First, we analysed the effect on β-catenin-TCF 409 transcriptional activity by using the TOP-Flash reporter gene, in which the luciferase gene is placed under 410 the control of a promoter harbouring ten copies of the TCF/LEF-1 consensus responsive sequence with a 411 c-fos minimal promoter. As negative control, a mutated form of this promoter (FOP-Flash) was used. Cells 412 were transfected with either TOP-Flash or FOP-Flash plasmids and the ratio of luciferase activity from 413 TOP-Flash to FOP-Flash was calculated to obtain a measurement of Wnt-specific transcriptional activity. 414 TK-Renilla was transfected as a transfection control. Cells were co-transfected with the Hakai and LRP4 or 415 LRP6 plasmids ( Fig. 4A-B ). We added either control L-cell conditioned medium (LCM) or WNT3A-416 conditioned media (WCM) to study the effects of Hakai on both basal as well as Wnt-stimulated 417 conditions. As shown, overexpression of LRP4 significantly inhibited both basal WNT signalling as well as 418 WNT3A-induced signalling, in line with LRP4 negatively regulating Wnt/β-catenin signalling. Whereas in 419 basal conditions overexpression of Hakai showed no significant effect, Hakai significantly potentiated 420 WNT3A-induced signalling. However, Hakai was unable to fully rescue the inhibitory effect of LRP4 when 421 both were co-transfected, although a small effect on basal LRP4-mediated suppression was observed (Fig. 422 4A). On the other hand, we confirmed that LRP6 was able to significantly activate Wnt-pathway in both 423 basal and WNT3A-stimulated conditions [35]. Importantly, there was a synergistic effect observed when 424 Hakai and LRP6 were co-overexpressed. LRP4 co-expression was able suppress LRP6-induced WNT 425 pathway activation, suggesting that LRP4 acts as an antagonist of LRP6 ( Fig. 4B ). Interestingly, co-426 expression of LRP4, LRP6, and Hakai showed that Hakai was able to partially rescue/counteract the 427 negative effect of LRP4 on LRP6, potentially by inducing the degradation of LRP4 ( Fig. 4B). 428 429 Hakin-1 disrupts tumourspheres formation and stemness 430 Preclinical studies from our group have demonstrated the therapeutic potential of targeting Hakai 431 through its HYB domain with Hakin-1 [22]. We therefore next investigated the potential effect of 432 pharmacological inhibition of the E3 ubiquitin ligase Hakai using Hakin-1 on tumoursphere formation. 433 HT29-derived tumourspheres were treated with Hakin-1 for six days, concurrently with the induction of 434 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 20, 2025. ; https://doi.org/10.1101/2025.11.20.689197doi: bioRxiv preprint 14 stemness culture conditions ( Fig. 5A ). Phase-contrast microscopy revealed that this treatment 435 phenotypically affected the tumourspheres ( Fig. 5B). Treated tumourspheres exhibited a less compact 436 morphology and poorly defined 3D structures compared to the well-organized and dense tumourspheres 437 observed under control conditions, suggesting a potential loss of stemness and increased cellular 438 differentiation. In addition, Hakin-1 modestly, but significantly, reduced the number of tumourspheres 439 formed (Fig. 5C). Although the area of individual tumourspheres varied, Hakin-1 treatment significantly 440 decreased the average size of tumourspheres compared to the control condition ( Fig. 5D). Importantly, 441 similar effects were observed when Hakin-1 treatment was applied either before or after the induction of 442 stemness. To further investigate the effect of Hakin-1 on stemness and differentiation, we analysed the 443 expression levels of various markers for the Wnt/β-catenin pathway, stem cells, and differentiation in 444 tumourspheres. Hakin-1 significantly reduced the expression levels of LEF-1 and TCF-1, both of which play 445 pivotal roles in the Wnt/β-catenin signalling pathway through their interaction with β-catenin. Moreover, 446 Hakin-1 markedly reduced the expression of the stem cell markers LGR5 and NANOG (Fig. 5E), while there 447 was no significant reduction in the expression of downstream Wnt/β-catenin target genes commonly 448 upregulated in colorectal cancer ( CCND1, C-MYC, and MMP7) (Fig. 5F). Taken together, Hakin-1 impairs 449 Wnt pathway activation, particularly affecting genes related to stemness. At the protein level, we 450 observed a slight upregulation of E-cadherin, accompanied by a downregulation of Wnt-related stem cell 451 markers LGR5 and NANOG ( Fig. 5G). As expected, Hakin-1 treatment did not alter Hakai protein levels. 452 These findings support the notion that Hakai inhibition alters the tumourspheres phenotype by reducing 453 stemness-associated markers. 454 455 Given the observed effects on stemness, we next analysed the potential impact of Hakin-1 on the 456 subcellular localization of LGR5 and MUC2 in tumourspheres. Moreover, we examined the effect of Hakai 457 inhibition on E-cadherin and LRP4 localization in tumourspheres using immunofluorescence, given E-458 cadherin’s role in EMT and LRP4 as a novel Hakai-interacting protein linked to Wnt/β-catenin signalling. 459 Hakin-1 treatment modestly increased E-cadherin levels at cell-cell contacts (¡Error! No se encuentra el 460 origen de la referencia. A). In contrast, Hakin-1 treatment resulted in a reduction in LGR5 levels (¡Error! 461 No se encuentra el origen de la referencia. B), suggesting a potential decrease in the stemness of cancer 462 cells. On the other hand, Hakin-1 treatment increased the expression of the differentiation marker MUC2 463 (¡Error! No se encuentra el origen de la referencia. D). As expected, LRP4, as a target for Hakai, showed 464 increased expression levels upon Hakin-1 treatment (¡Error! No se encuentra el origen de la referencia.C). 465 Taken together our results suggest that Hakin-1 impairs tumoursphere formation by reducing stemness-466 associated features, disrupting 3D structural integrity, and promoting cellular differentiation. 467 468 In conclusion, our findings identify Hakai as a critical regulator of Wnt/β-catenin signalling in colorectal 469 cancer as illustrated by the model in Fig. 7 . Hakai ubiquitinates LRP4, targeting it for proteasomal 470 degradation. The degradation of membrane-bound LRP4 may relieve its inhibitory effect on Wnt 471 signalling, thereby facilitating Hakai-mediated cooperation with LRP6 to potentiate pathway activation. 472 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 20, 2025. ; https://doi.org/10.1101/2025.11.20.689197doi: bioRxiv preprint 15 Consequently, promoting the expression of Wnt/β-catenin target genes that induce stem-like properties 473 in colon cancer. Pharmacological inhibition of Hakai may reduce stemness-associated features and 474 promote cellular differentiation in colorectal cancer cells. 475 476 477 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 20, 2025. ; https://doi.org/10.1101/2025.11.20.689197doi: bioRxiv preprint 16

478 In this study, we provide evidence that the E3 ubiquitin-ligase Hakai plays a pivotal role in regulating CSC 479 properties in colorectal cancer. Using colon cancer tumoursphere as a model of CSC, we show that Hakai 480 modulates self-renewal, differentiation and Wnt/β-catenin signalling. Silencing Hakai significantly impairs 481 self-renewal capacity, as shown by a marked reduction in tumoursphere number and size (Fig. 1B–D), and 482 leads to downregulation of key CSC markers, including the transcriptional Wnt target gene LGR5 (Fig. 1F) 483 [7,36]. These results are consistent with our previous findings showing that tumourspheres derived from 484 CRC cells exhibit increased CBLL1 (Hakai gene) and LGR5 mRNA levels compared to monolayer cultures 485 [29]. Our mechanistic studies suggest that Hakai could promote the acquisition of CSC properties via the 486 hyperactivation of the Wnt/β-catenin signalling pathway in colon cancer. We further identified the 487 negative regulator of Wnt/β-catenin signalling, LRP4, as a novel substrate of Hakai-mediated 488 ubiquitination. Given our biochemical studies and the important role of Wnt signalling in the development 489 of CSCs, we hypothesize that Hakai’s effect on stemness is at least partly through the newly identified 490 substrate LRP4. Notably, this is the first study to implicate LRP4 in Wnt signalling within colorectal CSCs, 491 expanding its role beyond its previously recognized function as a negative regulator during developmental 492 processes. Hakai overexpression decreased LRP4 protein levels and its surface localization ( Fig. 3A-B ), 493 while Hakai depletion led to its accumulation. Co-immunoprecipitation assays confirmed the interaction 494 between Hakai and LRP4, and ubiquitination studies demonstrated that Hakai targets LRP4 for 495 proteasomal degradation ( Fig. 3C-E ). These findings identify LRP4 as a novel substrate of Hakai and 496 suggest that Hakai contributes to Wnt/β-catenin pathway hyperactivation in CRC by removing LRP4-497 mediated inhibition of LRP6. According to this, a recent study demonstrates that although WNT-induced 498 FZD-LRP6 interaction occurs, this is not sufficient to initiate downstream Wnt/β-catenin signalling, 499 indicating that additional regulatory mechanisms are required for full pathway activation [37]. Moreover, 500 the co-expression of CBLL1 and LGR5 was proposed for the identification of a subpopulation of CMS2 501 patients with more aggressive biological characteristics [29]. Notably, this may include cooperation with 502 LRP6 to overcome LRP4-mediated inhibition, a mechanism consistent with previous reports of LRP4 503 antagonizing LRP6 signalling, especially in developmental contexts [9,10]. We further evaluated the 504 impact of pharmacological inhibition of Hakai using Hakin-1, a compound targeting the HYB domain 505 responsible for its ubiquitin-ligase activity [22]. Treatment with Hakin-1 significantly reduced 506 tumoursphere formation and stem cell marker expression ( Fig. 5C-G ), while increasing differentiation 507 markers such as MUC2, which is indicative of goblet cell lineage commitment [38]. Importantly, Hakin-1 508 also exerted effects on pre-established tumourspheres, demonstrating its potential in both preventive 509 and therapeutic contexts ( Fig. 5, Fig. S1-S3 ). These results were further corroborated by increased E-510 cadherin and LRP4 expression ( Fig. 6), suggesting that Hakin-1 inhibits EMT and the hyperactivation of 511 Wnt/β-catenin signalling. 512 513 The interplay between Hakai and Wnt/β-catenin signalling refines our understanding of CSC regulation. 514 While mutations in APC, CTNNB1, or AXIN frequently lead to Wnt/β-catenin pathway activation in CRC 515 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 20, 2025. ; https://doi.org/10.1101/2025.11.20.689197doi: bioRxiv preprint 17 [3,39] our findings suggest that such genetic alterations are not sufficient to induce a full CSC phenotype. 516 Instead, CSCs may require additional signals or co-factors to enhance Wnt/β-catenin signalling activity. 517 Only cells with the highest nuclear β-catenin activity exhibit CSC traits [5], highlighting the importance of 518 post-translational regulators like Hakai in enhancing Wnt/β-catenin pathway. Indeed, multiple E3 519 ubiquitin-ligases modulate Wnt signalling through degradation of pathway components. For example, 520 Mindbomb 1 enhances Wnt/β-catenin activity by targeting the non-canonical receptor RYK [40], while 521 Itch ubiquitinates LRP6 to promote endocytosis and downstream signalling [41]. Conversely, RNF43 and 522 ZNRF3 negatively regulate Wnt/β-catenin signalling by promoting the ubiquitination and internalization 523 of Frizzled receptors, thus attenuating pathway activation [42,43]. Compared to these ligases, Hakai 524 appears to occupy a unique niche by simultaneously targeting both E-cadherin and LRP4, thereby 525 integrating EMT and Wnt/β-catenin signalling into a common regulatory axis. In addition to modulating 526 both signalling pathways, Hakai may also influence metabolic adaptations essential for CSC maintenance. 527 Proteomic profiling revealed that Hakai silencing leads to a marked reduction in PRPS2 ( Fig. S4), a purine 528 biosynthesis enzyme regulated by MYC signalling and previously implicated in supporting CSC survival and 529 proliferation [44,45]. Given that C-MYC was the first gene recognized as a target of the Wnt/β-catenin 530 pathway in CRC [46], this finding suggests that Hakai might promote metabolic activity indirectly through 531 its impact on Wnt-driven C-MYC expression. These results are consistent with our previous results, which 532 demonstrated an association between CBLL1 expression and the CMS2 subtype in CRC patients, also 533 characterized by Myc pathway activation [29]. This opens the possibility that Hakai may exert a metabolic 534 influence that complements its known effects on signalling and differentiation, suggesting a new direction 535 for future investigation. 536 Mechanistically, our findings suggest that Hakai not only facilitates Wnt/β-catenin signalling by promoting 537 the degradation of LRP4, but may also contribute to LRP6-mediated pathway activation. Reporter assays 538 under WNT3A stimulation showed that co-expression of Hakai and LRP6 restored Wnt/β-catenin activity 539 suppressed by LRP4 more effectively than either factor alone ( Fig. 4B ). These observations imply that 540 Hakai helps counteract LRP4’s inhibitory role, possibly by promoting the clearance of suppressive inputs 541 and thereby facilitating the assembly or function of activating complexes. This interpretation is consistent 542 with previous reports describing the antagonistic role of LRP4 in modulating LRP6-mediated Wnt/β-543 catenin activation, particularly in developmental systems such as limb formation, bone development, and 544 the neuromuscular junction [9,10,13,47,48], and aligns with models of Wnt/β-catenin signalling that 545 emphasize the importance of the changing interactions and arrangements of receptors on the cell surface 546 [49]. 547 Hakin-1’s therapeutic profile compares favourably with other agents targeting the Wnt/β-catenin 548 pathway. For instance, anti-LRP6 nanobodies and R-spondin 3 (RSPO3) inhibitors reduce tumour growth 549 and induce differentiation by exhausting the CSC pool [50,51] Similarly, APC restoration in inducible 550 mouse models similarly drives differentiation and suppresses tumour relapse [52] Like these agents, 551 Hakin-1 promotes MUC2 expression and downregulates LGR5, suggesting a robust differentiation 552 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 20, 2025. ; https://doi.org/10.1101/2025.11.20.689197doi: bioRxiv preprint 18 response. Notably, unlike broad Wnt inhibitors, Hakin-1 has not been associated with systemic toxicity, 553 based on previous histopathological analyses in treated animals [22] However, the in vivo impact of 554 Hakai’s pharmacological inhibition on the Wnt/β-catenin signalling cascade remains to be elucidated. 555 Targeting E3 ligases for cancer therapy is gaining traction due to their substrate specificity and regulatory 556 capacity. While bortezomib and carfilzomib target the proteasome broadly, leading to side effects such 557 as neuropathy and myelosuppression [53,54], E3 ligase inhibitors like MLN4924, which targets the NEDD8-558 activating enzyme to inhibit Cullin-RING ligases, and Hakin-1, demonstrate greater substrate specificity 559 and improved lower toxicity[55,56]. Our data show that Hakin-1 does not impair apoptosis but instead 560 shifts tumour cells toward differentiation, a therapeutic endpoint now increasingly recognized as essential 561 for long-term tumour control. Notably, Hakai-silencing shows a more pronounced effect on cell 562 proliferation compared to the more limited impact of Hakin-1 targeting Hakai’s E3 ligase activity. This is 563 consistent with previous reports showing that full knockdown of Hakai markedly inhibits proliferation and 564 decreases Cyclin D1 protein levels [22,34]. In contrast, Hakin-1 treatment shows only a modest effect in 565 proliferation in vitro and in vivo [22] and does not significantly alter proliferation-associated targets such 566 as C-MYC or CCND1 (Fig. 5F). These findings support the hypothesis that other functional domains of Hakai 567 may be responsible for its effects on proliferation, as the inhibition of the HYB domain, responsible for its 568 E3 ubiquitin-ligase activity, shows only a limited impact when targeted alone. From a translational 569 perspective, the ability of Hakin-1 to reduce stemness, enhance differentiation, and interfere with both 570 Wnt/β-catenin signalling and EMT underscores its value as a therapeutic agent capable of targeting key 571 mechanisms underlying cancer resistance and metastasis. Its impact on both the initiation and 572 maintenance of tumourspheres, together with its favourable safety profile, positions it as a strong 573 candidate for combinatorial regimens targeting CSCs alongside standard chemotherapy. 574

575 In conclusion, our findings define a new functional role for Hakai in colorectal cancer, highlighting its 576 involvement in CSC regulation and Wnt/β-catenin signalling through LRP4-mediated modulation. The 577 development of Hakin-1 provides a promising therapeutic strategy to target CSCs by inhibiting Hakai-578 mediated ubiquitination, enhancing differentiation, and attenuating Wnt/β-catenin pathway activity. 579 These insights support further investigation of Hakai inhibitors as components of CSC-directed therapies 580 in colorectal cancer and offer a rational approach for improving long-term treatment outcomes. 581 582 583 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 20, 2025. ; https://doi.org/10.1101/2025.11.20.689197doi: bioRxiv preprint 19 Glossary 584 3-MA 3-Methyladenine 585 AKT Protein kinase B 586 APC Adenomatous polyposis coli 587 APS Ammonium persulfate 588 ATCC American Type Culture Collection 589 b-hFGF Basic human fibroblast growth factor 590 BCA Bicinchoninic acid assay 591 BMP Bone morphogenetic protein 592 BSA Bovine serum albumin 593 C19ORF43 Telomerase RNA component interacting RNase 594 CBLL1 Casitas B-lineage lymphoma like-1, Hakai gene 595 CCND1 Cyclin D1 gene 596 CMS Consensus Molecular Subtype 597 CRC Colorectal cancer 598 CSCs Cancer Stem Cells 599 CT Threshold cycle 600 DAPI 4’,6-diamidino-2-phenylindole 601 DDA Data-dependent acquisition 602 DKK Dickkopf 603 DMEM Dulbecco's Modified Eagle's Medium 604 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 20, 2025. ; https://doi.org/10.1101/2025.11.20.689197doi: bioRxiv preprint 20 DMSO Dimethyl sulfoxide 605 Dox Doxycycline 606 DVL Dishevelled 607 EGF Epidermal growth factor 608 EMT Epithelial-mesenchymal transition 609 ESI Electrospray ionization 610 FA Formic acid 611 FBS Fetal bovine serum 612 FZD Frizzled transmembrane receptors 613 GAPDH Glyceraldehyde 3-phosphate dehydrogenase 614 GSK3β Glycogen synthase kinase-3 β 615 Hakin-1 Hakai inhibitor 1 616 HRP Horseradish peroxidase 617 HYB Hakai-pY-binding 618 IAA Iodoacetamide 619 LC-MS/MS Liquid chromatography with tandem mass spectrometry 620 LFQ Label-free quantification 621 LGR5 Leucine-rich repeat-containing G-protein coupled Receptor 5 622 LRP4 Low density lipoprotein receptor-related protein 4 623 LRP5 Low density lipoprotein receptor-related protein 5 624 LRP6 Low density lipoprotein receptor-related protein 6 625 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 20, 2025. ; https://doi.org/10.1101/2025.11.20.689197doi: bioRxiv preprint 21 MESD Mesoderm development LRP chaperone 626 MNAR Missing not at random 627 MUC2 Mucin2 628 MYC MYC proto-oncogene 629 NIT1 Nitrilase homolog 1 630 PASEF Parallel accumulation–serial fragmentation 631 PBL Passive lysis buffer 632 PBS Phosphate-buffered saline 633 PBST Phosphate-buffered saline, 0.1% Triton X-100 634 PCA Principal component analysis 635 PCs Principal components 636 PCR Polymerase chain reaction 637 PFA Paraformaldehyde 638 PI3K Phosphatidylinositol 3-kinase 639 PMSF Phenylmethylsulfonyl fluoride 640 PRPS2 Phosphoribosyl pyrophosphate synthetase 2 641 pTyr Tyrosine residue phosphorylation, or phosphotyrosine 642 PVDF Polyvinylidene difluoride 643 RABGGTA Rab geranylgeranyltransferase alpha subunit 644 RNF43 Ring finger protein 43 645 RSPO R-spondin 646 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 20, 2025. ; https://doi.org/10.1101/2025.11.20.689197doi: bioRxiv preprint 22 RT Room temperature 647 SLC35F2 Solute carrier family 35 member F2 648 SOST Sclerostin 649 TBS-T Tris buffered saline-Tween 20 650 TCF/LEF T-cell factor/lymphoid enhancer-binding factor 651 TimsTOF Trapped ion mobility spectrometry quadrupole time-of-flight 652 UBXN7 UBX domain-containing protein 7 653 UFSP2 Ufm1-specific protease 2 654 WDR26 WD repeat-containing protein 26 655 WISE Wnt inhibitory secreted protein 656 ZNRF3 Zinc and ring finger protein 3 657 Declarations 658 • Availability of data and materials 659 All data generated or analysed during this study are included in this published article and its 660 supplementary information files. The datasets used and/or analysed during the current study are available 661 from the corresponding author on reasonable request. 662 663 • Competing interests 664 A.F, A.R.A and L.J. are inventors on patents related to inhibitors of the E3 ubiquitin-ligase Hakai. 665 666 • Funding sources 667 This study has been funded by the Instituto de Salud Carlos III (ISCIII) through the project numbers 668 PI21/00238 and FORT23/00010, and co-funded by the European Union. The project that gave rise to these 669

has received funding from “la Caixa” Foundation and the European Institute of Innovation and 670 Technology, EIT (body of the European Union that receives support from the European Union’s Horizon 671 2020 research and innovation program), under the grant agreement LCF/TR/CC21/52490003 and is also 672 supported by Consolidation of Competitive Research (IN607B 2023/12) from Agencia Gallega de 673 Innovación (GAIN) from Xunta de Galicia. This work was supported by a grant from the Scientific 674 Foundation of the Spanish Association against Cancer (INNOV235141FIGU). A.R.A. and G.A. are supported 675 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 20, 2025. ; https://doi.org/10.1101/2025.11.20.689197doi: bioRxiv preprint 23 by a predoctoral contract (PRDLC21591RODR and PRDLC234251ALFO, respectively) from Fundación 676 Científica Asociación Española Contra el Cáncer (AECC/AECC) and L.J. is supported by FPU contract 677 (FPU021/05350) from Ministerio de Universidades (Spain). 678 679 • Authors' contributions 680 Conceptualization: A.F.; investigation and methodology: A.R.-A., L.J., I.J.; data curation and formal 681 analysis: A.R.-A.; bioinformatic analysis: G.A.; writing—original draft: A.F. and A.R.-A; writing—review & 682 editing: A.F., A.R.-A., I.J., M.M.; with contributions from all authors; supervision, project administration, 683 and funding acquisition: A.F. All authors have read and agreed to the published version of the manuscript 684 685 • Acknowledgements 686 We thank Dr. Yasuyuki Fujita (Hokkaido University, Japan) for generously providing the pcDNA-FLAG-Hakai 687 and pBSSR-HA-Ubiquitin constructs, and Dr. Tatsuo Suzuki (Shinshu University School of Medicine, Japan) 688 for kindly providing the pcDNA-LRP4 plasmid. We would like to thank the Proteomics Unit 30 of 689 NANBIOSIS (CIBER-BBN) at the INIBIC for the support with the sample preparation and the Proteomics 690 laboratory at the Interdisciplinary Center for Chemistry and Biology (CICA), UDC for their expertise with 691 the LC-MS/MS and data analysis 692 693

694 [1] Fares J, Fares MY, Khachfe HH, Salhab HA, Fares Y. Molecular principles of metastasis: a hallmark 695 of cancer revisited. Signal Transduction and Targeted Therapy 2020 5:1 2020;5:1–17. 696 https://doi.org/10.1038/s41392-020-0134-x. 697 [2] Siegel Mph RL, Giaquinto AN, Ahmedin |, Dvm J, Siegel RL. Cancer statistics, 2024. CA Cancer J Clin 698 2024;74:12–49. https://doi.org/10.3322/CAAC.21820. 699 [3] Fearon ER, Vogelstein B. A genetic model for colorectal tumorigenesis. Cell 1990;61:759–67. 700 https://doi.org/10.1016/0092-8674(90)90186-i. 701 [4] Clevers H, Nusse R. Wnt/β-catenin signaling and disease. Cell 2012;149:1192–205. 702 https://doi.org/10.1016/j.cell.2012.05.012. 703 [5] Vermeulen L, De Sousa E Melo F, Van Der Heijden M, Cameron K, De Jong JH, Borovski T, Tuynman 704 JB, Todaro M, Merz C, Rodermond H, Sprick MR, Kemper K, Richel DJ, Stassi G, Medema JP. Wnt 705 activity defines colon cancer stem cells and is regulated by the microenvironment. Nature Cell 706 Biology 2010 12:5 2010;12:468–76. https://doi.org/10.1038/ncb2048. 707 [6] Baehs S, Herbst A, Thieme SE, Perschl C, Behrens A, Scheel S, Jung A, Brabletz T, Göke B, Blum H, 708 Kolligs FT. Dickkopf-4 is frequently down-regulated and inhibits growth of colorectal cancer cells. 709 Cancer Lett 2009;276:152–9. https://doi.org/10.1016/j.canlet.2008.11.003. 710 [7] Zhao H, Ming T, Tang S, Ren S, Yang H, Liu M, Tao Q, Xu H. Wnt signaling in colorectal cancer: 711 pathogenic role and therapeutic target. Molecular Cancer 2022 21:1 2022;21:1–34. 712 https://doi.org/10.1186/S12943-022-01616-7. 713 [8] Ahn Y, Sims C, Murray MJ, Kuhlmann PK, Fuentes-Antrás J, Weatherbee SD, Krumlauf R. Multiple 714 modes of Lrp4 function in modulation of Wnt/β-catenin signaling during tooth development. 715 Development 2017;144:2824–36. https://doi.org/10.1242/dev.150680. 716 [9] Johnson EB, Hammer RE, Herz J. Abnormal development of the apical ectodermal ridge and 717 polysyndactyly in Megf7-deficient mice. Hum Mol Genet 2005;14:3523–38. 718 https://doi.org/10.1093/hmg/ddi381. 719 [10] Li Y, Pawlik B, Elcioglu N, Aglan M, Kayserili H, Yigit G, Percin F, Goodman F, Nürnberg G, Cenani 720 A, Urquhart J, Chung BD, Ismail S, Amr K, Aslanger AD, Becker C, Netzer C, Scambler P, Eyaid W, 721 Hamamy H, Clayton-Smith J, Hennekam R, Nürnberg P, Herz J, Temtamy SA, Wollnik B. LRP4 722 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 20, 2025. ; https://doi.org/10.1101/2025.11.20.689197doi: bioRxiv preprint 24 Mutations Alter Wnt/β-Catenin Signaling and Cause Limb and Kidney Malformations in Cenani-723 Lenz Syndrome. Am J Hum Genet 2010;86:696–706. https://doi.org/10.1016/J.AJHG.2010.03.004. 724 [11] Shen C, Xiong WC, Mei L. LRP4 in neuromuscular junction and bone development and diseases. 725 Bone 2015;80:101–8. https://doi.org/10.1016/j.bone.2015.05.012. 726 [12] Asai N, Ohkawara B, Ito M, Masuda A, Ishiguro N, Ohno K. LRP4 induces extracellular matrix 727 productions and facilitates chondrocyte differentiation. Biochem Biophys Res Commun 728 2014;451:302–7. https://doi.org/10.1016/j.bbrc.2014.07.125. 729 [13] Choi HY, Dieckmann M, Herz J, Niemeier A. Lrp4, a Novel Receptor for Dickkopf 1 and Sclerostin, 730 Is Expressed by Osteoblasts and Regulates Bone Growth and Turnover In Vivo. PLoS One 731 2009;4:7930. https://doi.org/10.1371/JOURNAL.PONE.0007930. 732 [14] Zhou X, Xia E, Bhandari A, Zheng C, Xiang J, Guan Y, Zhang X. LRP4 promotes proliferation, 733 migration, and invasion in papillary thyroid cancer. Biochem Biophys Res Commun 2018;503:257–734 63. https://doi.org/10.1016/j.bbrc.2018.06.012. 735 [15] Weissman AM, Shabek N, Ciechanover A. The predator becomes the prey: regulating the ubiquitin 736 system by ubiquitylation and degradation. Nature Reviews Molecular Cell Biology 2011 12:9 737 2011;12:605–20. https://doi.org/10.1038/nrm3173. 738 [16] Hershko A, Heller H, Elias S, Ciechanover A. Components of ubiquitin-protein ligase system. 739 Resolution, affinity purification, and role in protein breakdown. Journal of Biological Chemistry 740 1983;258:8206–14. https://doi.org/10.1016/S0021-9258(20)82050-X. 741 [17] Fujita Y, Krause G, Scheffner M, Zechner D, Leddy HEM, Behrens J, Sommer T, Birchmeier W. Hakai, 742 a c-Cbl-like protein, ubiquitinates and induces endocytosis of the E-cadherin complex. Nat Cell Biol 743 2002;4:222–31. https://doi.org/10.1038/ncb758. 744 [18] Aparicio LA, Blanco M, Castosa R, Concha Á, Valladares M, Calvo L, Figueroa A. Clinical implications 745 of epithelial cell plasticity in cancer progression. Cancer Lett 2015;366:1–10. 746 https://doi.org/10.1016/j.canlet.2015.06.007. 747 [19] Zhou WJ, Geng ZH, Chi S, Zhang W, Niu XF, Lan SJ, Ma L, Yang X, Wang LJ, Ding YQ, Geng JG. Slit-748 Robo signaling induces malignant transformation through Hakai-mediated E-cadherin 749 degradation during colorectal epithelial cell carcinogenesis. Cell Res 2011;21:609–26. 750 https://doi.org/10.1038/CR.2011.17. 751 [20] Liu M, Jiang K, Lin G, Liu P, Yan Y, Ye T, Yao G, Barr MP, Liang D, Wang Y, Gong P, Meng S, Piao H. 752 Ajuba inhibits hepatocellular carcinoma cell growth via targeting of β-catenin and YAP signaling 753 and is regulated by E3 ligase Hakai through neddylation. Journal of Experimental and Clinical 754 Cancer Research 2018;37:165. https://doi.org/10.1186/s13046-018-0806-3. 755 [21] Hui L, Zhang S, Wudu M, Ren H, Xu Y, Zhang Q, Qiu X. CBLL1 is highly expressed in non-small cell 756 lung cancer and promotes cell proliferation and invasion. Thorac Cancer 2019;10:1479–1488. 757 https://doi.org/10.1111/1759-7714.13097. 758 [22] Martinez-Iglesias O, Casas-Pais A, Castosa R, Díaz-Díaz A, Roca-Lema D, Concha Á, Cortés Á, Gago 759 F, Figueroa A. Hakin-1, a New Specific Small-Molecule Inhibitor for the E3 Ubiquitin-Ligase Hakai, 760 Inhibits Carcinoma Growth and Progression. Cancers (Basel) 2020;12:1340. 761 https://doi.org/10.3390/cancers12051340. 762 [23] Taki M, Kamata N, Yokoyama K, Fujimoto R, Tsutsumi S, Nagayama M. Downregulation of Wnt4 763 and upregulation of Wnt5a expression by epithelialmesenchymal transition in human squamous 764 carcinoma cells. Cancer Sci 2003;94:593–7. https://doi.org/10.1111/J.1349-765 7006.2003.TB01488.X. 766 [24] Clevers H. Wnt/β-Catenin Signaling in Development and Disease. Cell 2006;127:469–80. 767 https://doi.org/10.1016/j.cell.2006.10.018. 768 [25] Spit M, Koo BK, Maurice MM. Tales from the crypt: Intestinal niche signals in tissue renewal, 769 plasticity and cancer. Open Biol 2018;8:180120. https://doi.org/10.1098/rsob.180120. 770 [26] Roca-Lema D, Quiroga M, Khare V, Díaz-Díaz A, Barreiro-Alonso A, Rodríguez-Alonso A, Concha Á, 771 Romay G, Cerdán ME, Gasche C, Figueroa A. Role of the E3 ubiquitin-ligase Hakai in intestinal 772 inflammation and cancer bowel disease. Sci Rep 2022;12. https://doi.org/10.1038/S41598-022-773 22295-W. 774 [27] Tauriello DVF, Haegebarth A, Kuper I, Edelmann MJ, Henraat M, Canninga-van Dijk MR, Kessler 775 BM, Clevers H, Maurice MM. Loss of the Tumor Suppressor CYLD Enhances Wnt/β-Catenin 776 Signaling through K63-Linked Ubiquitination of Dvl. Mol Cell 2010;37:607–19. 777 https://doi.org/10.1016/j.molcel.2010.01.035. 778 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 20, 2025. ; https://doi.org/10.1101/2025.11.20.689197doi: bioRxiv preprint 25 [28] Ricci-Vitiani L, Lombardi DG, Pilozzi E, Biffoni M, Todaro M, Peschle C, De Maria R. Identification 779 and expansion of human colon-cancer-initiating cells. Nature 2007;445:111–5. 780 https://doi.org/10.1038/NATURE05384. 781 [29] Alfonsin G, Berral-González A, Rodríguez-Alonso A, Quiroga M, Rivas JD Las, Figueroa A. 782 Stratification of Colorectal Patients Based on Survival Analysis Shows the Value of Consensus 783 Molecular Subtypes and Reveals the CBLL1 Gene as a Biomarker of CMS2 Tumours. Int J Mol Sci 784 2024;25:1919. https://doi.org/10.3390/ijms25031919. 785 [30] Li Y, Xie P, Lu L, Wang J, Diao L, Liu Z, Guo F, He Y, Liu Y, Huang Q, Liang H, Li D, He F. An integrated 786 bioinformatics platform for investigating the human E3 ubiquitin ligase-substrate interaction 787 network. Nat Commun 2017;8:1–9. https://doi.org/10.1038/s41467-017-00299-9. 788 [31] Wang X, Li Y, He M, Kong X, Jiang P, Liu X, Diao L, Zhang X, Li H, Ling X, Xia S, Liu Z, Liu Y, Cui C, 789 Wang Y, Tang L, Zhang L, He F, Li D. UbiBrowser 2.0: a comprehensive resource for proteome-wide 790 ubiquitin ligase/deubiquitinase-substrate interactions in eukaryotic species. Nucleic Acids 791 Research. 2022. http://ubibrowser.bio-it.cn/ubibrowser_v3/ (accessed June 6, 2025). 792 [32] Wang X, Li Y, He M, Kong X, Jiang P, Liu X, Diao L, Zhang X, Li H, Ling X, Xia S, Liu Z, Liu Y, Cui C, 793 Wang Y, Tang L, Zhang L, He F, Li D. E3 ligase CBLL1: Network view for retrieved substrates 794 (UbiBrowser 2.0) 2022. http://ubibrowser.bio-795 it.cn/ubibrowser_v3/Home/Result/index/name/Q75N03/module/Strict/proteinType/E3 796 (accessed June 6, 2025). 797 [33] Mukherjee M, Chow SY, Yusoff P, Seetharaman J, Ng C, Sinniah S, Koh XW, Asgar NFM, Li D, Yim 798 D, Jackson RA, Yew J, Qian J, Iyu A, Lim YP, Zhou X, Sze SK, Guy GR, Sivaraman J. Structure of a 799 novel phosphotyrosine-binding domain in Hakai that targets E-cadherin. EMBO J 2012;31:1308–800 19. https://doi.org/10.1038/emboj.2011.496. 801 [34] Figueroa A, Kotani H, Toda Y, Mazan-Mamczarz K, Mueller EC, Otto A, Disch L, Norman M, Ramdasi 802 RM, Keshtgar M, Gorospe M, Fujita Y. Novel roles of Hakai in cell proliferation and oncogenesis. 803 Mol Biol Cell 2009;20:3533–42. https://doi.org/10.1091/mbc.E08-08-0845. 804 [35] Chen J, Yan H, Ren D ni, Yin Y, Li Z, He Q, Wo D, Ho MS chun, Chen Y, Liu Z, Yang J, Liu S, Zhu W. 805 LRP6 dimerization through its LDLR domain is required for robust canonical Wnt pathway 806 activation. Cell Signal 2014;26:1068–74. https://doi.org/10.1016/j.cellsig.2013.12.020. 807 [36] Wang W, Lokman NA, Barry SC, Oehler MK, Ricciardelli C. LGR5: An emerging therapeutic target 808 for cancer metastasis and chemotherapy resistance. Cancer and Metastasis Reviews 2025 44:1 809 2025;44:1–23. https://doi.org/10.1007/S10555-024-10239-X. 810 [37] Voss JH, Koszegi Z, Yan Y, Shorter E, Grätz L, Lanner JT, Calebiro D, Schulte G. WNT-induced 811 association of Frizzled and LRP6 is not sufficient for the initiation of WNT/β-catenin signaling. 812 Nature Communications 2025;16. https://doi.org/10.1038/s41467-025-60096-7. 813 [38] Qin T, Yang J, Huang D, Zhang Z, Huang Y, Chen H, Xu G. DOCK4 stimulates MUC2 production 814 through its effect on goblet cell differentiation. J Cell Physiol 2021;236:6507–19. 815 https://doi.org/10.1002/jcp.30325. 816 [39] Fearon ER. Molecular genetics of colorectal cancer. Annual Review of Pathology: Mechanisms of 817 Disease 2011;6:479–507. https://doi.org/10.1146/annurev-pathol-011110-130235. 818 [40] Berndt JD, Aoyagi A, Yang P, Anastas JN, Tang L, Moon RT. Mindbomb 1, an E3 ubiquitin ligase, 819 forms a complex with RYK to activate Wnt/β-catenin signaling. J Cell Biol 2011;194:737. 820 https://doi.org/10.1083/JCB.201107021. 821 [41] Vijayakumar S, Liu G, Wen HC, Abu Y, Chong R, Nastri H, Bornstein GG, Pan ZQ, Aaronson SA. 822 Extracellular LDLR repeats modulate Wnt signaling activity by promoting LRP6 receptor 823 endocytosis mediated by the Itch E3 ubiquitin ligase. Genes Cancer 2017;8:613–27. 824 https://doi.org/10.18632/genesandcancer.146. 825 [42] Koo BK, Spit M, Jordens I, Low TY, Stange DE, Van De Wetering M, Van Es JH, Mohammed S, Heck 826 AJR, Maurice MM, Clevers H. Tumour suppressor RNF43 is a stem-cell E3 ligase that induces 827 endocytosis of Wnt receptors. Nature 2012;488:665–9. https://doi.org/10.1038/nature11308. 828 [43] Hao HX, Xie Y, Zhang Y, Zhang O, Oster E, Avello M, Lei H, Mickanin C, Liu D, Ruffner H, Mao X, Ma 829 Q, Zamponi R, Bouwmeester T, Finan PM, Kirschner MW, Porter JA, Serluca FC, Cong F. ZNRF3 830 promotes Wnt receptor turnover in an R-spondin-sensitive manner. Nature 2012;485:195–200. 831 https://doi.org/10.1038/nature11019. 832 [44] Lv Y, Wang X, Li X, Xu G, Bai Y, Wu J, Piao Y, Shi Y, Xiang R, Wang L. Nucleotide de novo synthesis 833 increases breast cancer stemness and metastasis via cGMP-PKG-MAPK signaling pathway. PLoS 834 Biol 2020;18:1–24. https://doi.org/10.1371/journal.pbio.3000872. 835 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 20, 2025. ; https://doi.org/10.1101/2025.11.20.689197doi: bioRxiv preprint 26 [45] Cunningham JT, Moreno M V., Lodi A, Ronen SM, Ruggero D. Protein and Nucleotide Biosynthesis 836 Are Coupled by a Single Rate-Limiting Enzyme, PRPS2, to Drive Cancer. Cell 2014;157:1088–103. 837 https://doi.org/10.1016/j.cell.2014.03.052. 838 [46] Van Der Flier LG, Clevers H. Stem Cells, Self-Renewal, and Differentiation in the Intestinal 839 Epithelium. Annu Rev Physiol 2009;71:241–60. 840 https://doi.org/10.1146/annurev.physiol.010908.163145. 841 [47] Weatherbee SD, Anderson K V., Niswander LA. LDL-receptor-related protein 4 is crucial for 842 formation of the neuromuscular junction. Development 2006;133:4993–5000. 843 https://doi.org/10.1242/DEV.02696,. 844 [48] Kim N, Stiegler AL, Cameron TO, Hallock PT, Gomez AM, Huang JH, Hubbard SR, Dustin ML, Burden 845 SJ. Lrp4 Is a Receptor for Agrin and Forms a Complex with MuSK. Cell 2008;135:334–42. 846 https://doi.org/10.1016/j.cell.2008.10.002. 847 [49] De Lau W, Peng WC, Gros P, Clevers H. The R-spondin/Lgr5/Rnf43 module: Regulator of Wnt signal 848 strength. Genes Dev 2014;28:305–16. https://doi.org/10.1101/gad.235473.113. 849 [50] Fenderico N, Scherpenzeel RC Van, Gold M, Proverbio D, Jordens I, Kralj T, Stryeck S, Bass TZ, 850 Hermans G, Ullman C, Aastrup T, Gros P, Maurice MM. Anti-LRP5/6 VHHs promote differentiation 851 of Wnt-hypersensitive intestinal stem cells. Nat Commun 2019;10:365. 852 https://doi.org/10.1038/s41467-018-08172-z. 853 [51] Storm EE, Durinck S, De Sousa E. Melo F, Tremayne J, Kljavin N, Tan C, Ye X, Chiu C, Pham T, Hongo 854 JA, Bainbridge T, Firestein R, Blackwood E, Metcalfe C, Stawiski EW, Yauch RL, Wu Y, De Sauvage 855 FJ. Targeting PTPRK-RSPO3 colon tumours promotes differentiation and loss of stem-cell function. 856 Nature 2016;529:97–100. https://doi.org/10.1038/nature16466. 857 [52] Dow LE, O’Rourke KP, Simon J, Tschaharganeh DF, Van Es JH, Clevers H, Lowe SW. Apc Restoration 858 Promotes Cellular Differentiation and Reestablishes Crypt Homeostasis in Colorectal Cancer. Cell 859 2015;161:1539–52. https://doi.org/10.1016/J.CELL.2015.05.033. 860 [53] Pancheri E, Guglielmi V, Wilczynski GM, Malatesta M, Tonin P, Tomelleri G, Nowis D, Vattemi G. 861 Non-Hematologic Toxicity of Bortezomib in Multiple Myeloma: The Neuromuscular and 862 Cardiovascular Adverse Effects. Cancers (Basel) 2020;12:2540. 863 https://doi.org/10.3390/CANCERS12092540. 864 [54] Nowis D, Ma̧czewski M, Mackiewicz U, Kujawa M, Ratajska A, Wieckowski MR, Wilczyński GM, 865 Malinowska M, Bil J, Salwa P, Bugajski M, Wójcik C, Siński M, Abramczyk P, Winiarska M, 866 Da̧browska-Iwanicka A, Duszyński J, Jakóbisiak M, Golab J. Cardiotoxicity of the anticancer 867 therapeutic agent bortezomib. Am J Pathol 2010;176:2658–68. 868 https://doi.org/10.2353/AJPATH.2010.090690. 869 [55] Foster JH, Reid JM, Minard C, Woodfield S, Denic KZ, Isikwei E, Voss SD, Nelson M, Liu X, Berg SL, 870 Fox E, Weigel BJ. Phase 1 study of NEDD8 activating enzyme inhibitor pevonedistat in combination 871 with chemotherapy in pediatric patients with recurrent or refractory solid tumors (ADVL1615). 872 Eur J Cancer 2024;209:114241. https://doi.org/10.1016/J.EJCA.2024.114241. 873 [56] Shah JJ, Jakubowiak AJ, O’Connor OA, Orlowski RZ, Harvey RD, Smith MR, Lebovic D, Diefenbach 874 C, Kelly K, Hua Z, Berger AJ, Mulligan G, Faessel HM, Tirrell S, Dezube BJ, Lonial S. Phase I Study of 875 the Novel Investigational NEDD8-Activating Enzyme Inhibitor Pevonedistat (MLN4924) in Patients 876 with Relapsed/Refractory Multiple Myeloma or Lymphoma. Clinical Cancer Research 2016;22:34–877 43. https://doi.org/10.1158/1078-0432.CCR-15-1237. 878 879 APPENDIX A. Supplemental information 880 Gene name Specie Forward sequence (5’-3’) Reverse sequence (5’-3’) CBLL1 Human CTGGATCCTTGGGTGGTCTT AGTTCTTTGAGTTCGCGGTG NANOG Human CAGTCTGGACACTGGCTGAA CTCGCTGATTAGGCTCCAAC LGR5 Human AGCAAACCTACGTCTGGAC ACAGAGGAAAGATGGCAGT C-MYC Human GTTATCTCGCAAACCCCAGA ACAGAATGGGTCCAGATTGC AXIN2 Human GGTCCACGGAAACTGTTGAC TCCATCTACACTGCTGTCCG B-ACTIN Human GGCATCCTCACCCTGAAGTA AGGTGTGGTGCCAGATTTTC LEF-1 Human ACAGATCACCCCACCTCTTG ATAGCTGGATGAGGGATGCC TCF-1 Human GCCAAGAAGCCAACCATCAA CTGCAATGACCTTGGCTCTC preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 20, 2025. ; https://doi.org/10.1101/2025.11.20.689197doi: bioRxiv preprint 27 CCDN1 Human TCTACACCGACAACTCCAT GGTTCCACTTGAGCTTGTT MMP-7 Human GTGTTTCCTGGCCCATCAAA GGCTTTAAACATGTGGGGCA 881 Table S1. Primers sequences used for RT-qPCR. Sequences of the forward and reverse primers used for 882 quantitative reverse transcription PCR (RT-qPCR) to amplify target genes. Primer sequences were 883 designed using BLAST and Primer3. Gene names correspond to those used in the main text. 884 885 SwissProt ID (E3) Gene Symbol (E3) SwissProt ID (Substrate) Gene Symbol (Substrate) Confidence Score Species 1 Q75N03 CBLL1 O00499 BIN1 0.851 H.sapiens 2 Q75N03 CBLL1 O15230 LAMA5 0.851 H.sapiens 3 Q75N03 CBLL1 O15353 FOXN1 0.851 H.sapiens 4 Q75N03 CBLL1 O75096 LRP4 0.851 H.sapiens 5 Q75N03 CBLL1 O75147 OBSL1 0.851 H.sapiens 6 Q75N03 CBLL1 O75473 LGR5 0.851 H.sapiens 7 Q75N03 CBLL1 P01133 EGF 0.851 H.sapiens 8 Q75N03 CBLL1 P02751 FN1 0.851 H.sapiens 9 Q75N03 CBLL1 P07996 THBS1 0.851 H.sapiens 10 Q75N03 CBLL1 P08631 HCK 0.851 H.sapiens 11 Q75N03 CBLL1 P08922 ROS1 0.851 H.sapiens 12 Q75N03 CBLL1 P08F94 PKHD1 0.851 H.sapiens 13 Q75N03 CBLL1 P11362 FGFR1 0.851 H.sapiens 14 Q75N03 CBLL1 P13688 CEACAM1 0.851 H.sapiens 15 Q75N03 CBLL1 P15391 CD19 0.851 H.sapiens 16 Q75N03 CBLL1 P17948 FLT1 0.851 H.sapiens 17 Q75N03 CBLL1 P23458 JAK1 0.851 H.sapiens 18 Q75N03 CBLL1 P23634 ATP2B4 0.851 H.sapiens 19 Q75N03 CBLL1 P25391 LAMA1 0.851 H.sapiens 20 Q75N03 CBLL1 P28827 PTPRM 0.851 H.sapiens 21 Q75N03 CBLL1 P31276 HOXC13 0.851 H.sapiens 22 Q75N03 CBLL1 P36888 FLT3 0.851 H.sapiens 23 Q75N03 CBLL1 P42229 STAT5A 0.851 H.sapiens 24 Q75N03 CBLL1 P43405 SYK 0.851 H.sapiens 25 Q75N03 CBLL1 P49286 MTNR1B 0.851 H.sapiens 26 Q75N03 CBLL1 P51608 MECP2 0.851 H.sapiens 27 Q75N03 CBLL1 P51692 STAT5B 0.851 H.sapiens 28 Q75N03 CBLL1 P52179 MYOM1 0.851 H.sapiens 29 Q75N03 CBLL1 P52333 JAK3 0.851 H.sapiens 30 Q75N03 CBLL1 P52735 VAV2 0.851 H.sapiens 31 Q75N03 CBLL1 P54296 MYOM2 0.851 H.sapiens 32 Q75N03 CBLL1 P54762 EPHB1 0.851 H.sapiens 33 Q75N03 CBLL1 P56539 CAV3 0.851 H.sapiens 34 Q75N03 CBLL1 Q07889 SOS1 0.851 H.sapiens 35 Q75N03 CBLL1 Q13023 AKAP6 0.851 H.sapiens 36 Q75N03 CBLL1 Q16288 NTRK3 0.851 H.sapiens 37 Q75N03 CBLL1 Q16620 NTRK2 0.851 H.sapiens preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 20, 2025. ; https://doi.org/10.1101/2025.11.20.689197doi: bioRxiv preprint 28 38 Q75N03 CBLL1 Q5VTT5 MYOM3 0.851 H.sapiens 39 Q75N03 CBLL1 Q6Q0C0 TRAF7 0.851 H.sapiens 40 Q75N03 CBLL1 Q8WU20 FRS2 0.851 H.sapiens 41 Q75N03 CBLL1 Q8WZ42 TTN 0.851 H.sapiens 42 Q75N03 CBLL1 Q96BA8 CREB3L1 0.851 H.sapiens 43 Q75N03 CBLL1 Q99996 AKAP9 0.851 H.sapiens 44 Q75N03 CBLL1 Q9BXB1 LGR4 0.851 H.sapiens 45 Q75N03 CBLL1 Q9HAZ2 PRDM16 0.851 H.sapiens 46 Q75N03 CBLL1 Q9HC29 NOD2 0.851 H.sapiens 47 Q75N03 CBLL1 Q9NRA0 SPHK2 0.851 H.sapiens 48 Q75N03 CBLL1 Q9NYQ6 CELSR1 0.851 H.sapiens 49 Q75N03 CBLL1 Q9UIU6 SIX4 0.851 H.sapiens 50 Q75N03 CBLL1 Q9UKW4 VAV3 0.851 H.sapiens 51 Q75N03 CBLL1 P00533 EGFR 0.850 H.sapiens 52 Q75N03 CBLL1 P01127 PDGFB 0.850 H.sapiens 53 Q75N03 CBLL1 P02671 FGA 0.850 H.sapiens 54 Q75N03 CBLL1 P02675 FGB 0.850 H.sapiens 55 Q75N03 CBLL1 P05019 IGF1 0.850 H.sapiens 56 Q75N03 CBLL1 P06241 FYN 0.850 H.sapiens 57 Q75N03 CBLL1 P07333 CSF1R 0.850 H.sapiens 58 Q75N03 CBLL1 P07359 GP1BA 0.850 H.sapiens 59 Q75N03 CBLL1 P07947 YES1 0.850 H.sapiens 60 Q75N03 CBLL1 P08514 ITGA2B 0.850 H.sapiens 61 Q75N03 CBLL1 P09327 VIL1 0.850 H.sapiens 62 Q75N03 CBLL1 P10721 KIT 0.850 H.sapiens 63 Q75N03 CBLL1 P16066 NPR1 0.850 H.sapiens 64 Q75N03 CBLL1 P18206 VCL 0.850 H.sapiens 65 Q75N03 CBLL1 P21333 FLNA 0.850 H.sapiens 66 Q75N03 CBLL1 P21802 FGFR2 0.850 H.sapiens 67 Q75N03 CBLL1 P22607 FGFR3 0.850 H.sapiens 68 Q75N03 CBLL1 P29322 EPHA8 0.850 H.sapiens 69 Q75N03 CBLL1 P29350 PTPN6 0.850 H.sapiens 70 Q75N03 CBLL1 P33151 CDH5 0.850 H.sapiens 71 Q75N03 CBLL1 P35367 HRH1 0.850 H.sapiens 72 Q75N03 CBLL1 P40189 IL6ST 0.850 H.sapiens 73 Q75N03 CBLL1 P42338 PIK3CB 0.850 H.sapiens 74 Q75N03 CBLL1 P49747 COMP 0.850 H.sapiens 75 Q75N03 CBLL1 P49767 VEGFC 0.850 H.sapiens 76 Q75N03 CBLL1 Q02763 TEK 0.850 H.sapiens 77 Q75N03 CBLL1 Q06418 TYRO3 0.850 H.sapiens 78 Q75N03 CBLL1 Q12979 ABR 0.850 H.sapiens 79 Q75N03 CBLL1 Q5VY43 PEAR1 0.850 H.sapiens 80 Q75N03 CBLL1 Q68CJ9 CREB3L3 0.850 H.sapiens 81 Q75N03 CBLL1 Q8N9M5 TMEM102 0.850 H.sapiens 82 Q75N03 CBLL1 Q96QS1 TSPAN32 0.850 H.sapiens 83 Q75N03 CBLL1 Q99650 OSMR 0.850 H.sapiens 84 Q75N03 CBLL1 Q9BRC7 PLCD4 0.850 H.sapiens 85 Q75N03 CBLL1 Q9UL54 TAOK2 0.850 H.sapiens preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 20, 2025. ; https://doi.org/10.1101/2025.11.20.689197doi: bioRxiv preprint 29 86 Q75N03 CBLL1 P17813 ENG 0.822 H.sapiens 87 Q75N03 CBLL1 P35580 MYH10 0.822 H.sapiens 88 Q75N03 CBLL1 P36402 TCF7 0.822 H.sapiens 89 Q75N03 CBLL1 Q03395 ROM1 0.822 H.sapiens 90 Q75N03 CBLL1 Q12809 KCNH2 0.822 H.sapiens 91 Q75N03 CBLL1 Q13009 TIAM1 0.822 H.sapiens 92 Q75N03 CBLL1 Q6KC79 NIPBL 0.822 H.sapiens 93 Q75N03 CBLL1 Q6UWI4 SHISA2 0.822 H.sapiens 94 Q75N03 CBLL1 Q9UJU2 LEF1 0.822 H.sapiens 95 Q75N03 CBLL1 Q9UNE0 EDAR 0.822 H.sapiens 96 Q75N03 CBLL1 Q9Y264 ANGPT4 0.822 H.sapiens 97 Q75N03 CBLL1 O75038 PLCH2 0.821 H.sapiens 98 Q75N03 CBLL1 P20827 EFNA1 0.821 H.sapiens 99 Q75N03 CBLL1 P35579 MYH9 0.821 H.sapiens 100 Q75N03 CBLL1 Q01826 SATB1 0.821 H.sapiens 101 Q75N03 CBLL1 Q15173 PPP2R5B 0.821 H.sapiens 102 Q75N03 CBLL1 Q6ZWH5 NEK10 0.821 H.sapiens 103 Q75N03 CBLL1 P07948 LYN 0.810 H.sapiens 104 Q75N03 CBLL1 P09619 PDGFRB 0.810 H.sapiens 105 Q75N03 CBLL1 P16234 PDGFRA 0.810 H.sapiens 106 Q75N03 CBLL1 Q7Z6A9 BTLA 0.810 H.sapiens 107 Q75N03 CBLL1 O75593 FOXH1 0.809 H.sapiens 108 Q75N03 CBLL1 O96020 CCNE2 0.809 H.sapiens 109 Q75N03 CBLL1 P24864 CCNE1 0.809 H.sapiens 110 Q75N03 CBLL1 P25445 FAS 0.809 H.sapiens 111 Q75N03 CBLL1 Q13485 SMAD4 0.809 H.sapiens 112 Q75N03 CBLL1 Q15431 SYCP1 0.809 H.sapiens 113 Q75N03 CBLL1 Q16181 SEPTIN7 0.809 H.sapiens 114 Q75N03 CBLL1 Q86VD1 MORC1 0.809 H.sapiens 115 Q75N03 CBLL1 Q9UH03 SEPTIN3 0.809 H.sapiens 116 Q75N03 CBLL1 Q9Y615 ACTL7A 0.809 H.sapiens 886 Table S2. Predicted Hakai substrate interactions by bioinformatics. UbiBrowser platform was used to 887 determine the predicted Hakai-substrate interaction network. A total of 116 proteins were found as 888 potential predicted substrates for Hakai-mediated ubiquitination [32]. 889 890 891 892 893 894 895 896 897 898 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 20, 2025. ; https://doi.org/10.1101/2025.11.20.689197doi: bioRxiv preprint 30 Figure captions 899 Fig 1. Hakai-silencing results in a decrease in the number and size of tumourspheres while decreasing 900 stem cell markers. (A) Schematic workflow of the self-renewal model. Hakai-silencing was induced in 901 HT29 monolayer cultures and after 72 hours, cells were cultured in cancer stem cell promoting conditions 902 for 5 days to form tumourspheres. (B) Tumourspheres were phenotypical characterized by phase contrast 903 images. Representative images of Hakai-silenced HT29 tumourspheres and control conditions were taken 904 after 5 days of the induction of stemness. Images were obtained using a 10x objective. Scale bar, 200 µm. 905 (C) Quantification of the number of HT29 tumourspheres in Hakai-silenced compared to control 906 conditions. (D) Size of tumourspheres of HT29 tumourspheres in Hakai-silenced compared to control 907 conditions. Results are expressed as mean ± SEM. Quantification was carried out using ImageJ software, 908 and statistical significance was determined using GraphPad Prism software. (E) mRNA expression levels 909 of Hakai (CBLL1 gene) and stem cell markers in HT29 tumourspheres in Hakai-silenced compared to 910 control analysed by RT-PCR. (F) Expression of Hakai and stem cell protein markers was determined by 911 Western blot in HT29 tumourspheres with Hakai silencing compared to control. Protein bands were 912 quantified using ImageJ and normalized to loading control. Data are presented as mean ± SEM from three 913 independent experiments. Statistical significance was determined using an unpaired t-test (*p < 0.05, **p 914 < 0.01, ***p < 0.001, ****p < 0.0001). 915 916 Fig 2. Proteomic study and UbiBrowser database to find novel Hakai-regulated proteins in 917 tumoursphere formation. (A) Schematic workflow of the proteomic study in Hakai-silenced 918 tumoursphere compared to control. Hakai-silencing in HT29 cells was induced and cells were seeded in 919 stem cell promoting media using ultra-low attachment plates to induce tumoursphere formation. Protein 920 extracts were digested and protein profile was analysed by silver staining in SDS-page gels. Proteins were 921 digested with trypsin and the obtained peptides were fractionated and separated by nano-LC-MS/MS 922 analysis coupled to a timsTOF Pro mass spectrometer. (B) Hakai-silencing in tumourspheres was 923 confirmed in three biological replicates by Western blot. GAPDH was used as loading control. (C) Venn 924 diagram of identified proteins (≥ 2 peptides) in tumourspheres obtained from Hakai-silenced HT29 colon 925 cancer tumourspheres versus control. (D) Principle Component Analysis (PCA) plot showing two principal 926 components corresponding to three replicates of Hakai-silenced tumourspheres and three replicates of 927 control tumourspheres. (E) Heatmap showing the regulated proteins. Up and down regulated proteins in 928 Hakai-silenced tumoursphere versus control are represented. (F) Volcano plot of differentially expressed 929 proteins in Hakai-silenced tumourspheres versus control tumourspheres. Black dots represent proteins 930 showing significant fold changes. The protein significance was set to adjusted p-value < 0.05, protein fold 931 change to ≥ 1.5, used peptides to ≥ 2. (G) Box plot showing the protein profiling comparison, based on the 932 label-free quantitation (LFQ) for which levels changed significantly between samples. The LFQ values are 933 plotted on a Log2 scale along the vertical axis. (H) Top 20 Hakai (CBLL1 gene) substrates predicted in Homo 934 sapiens (blue) and the 4 known substrates already described (red) by using UbiBrowser 2.0 database. 935 936 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 20, 2025. ; https://doi.org/10.1101/2025.11.20.689197doi: bioRxiv preprint 31 Fig 3. Identification of LRP4 as a new potential Hakai-interacting protein. (A) Hakai overexpression 937 reduces LRP4 protein levels in HEK293 and HCT116 cells shown by Western blot. (B) Impact of Hakai on 938 cell-surface expression of LRP4 in HEK293T transfected with LRP4 and empty vector or Hakai-V5. Surface 939 levels of LRP4 were detected in unpermeabilized cells after which total levels of Hakai were visualized via 940 permeabilization using confocal microscopy. Pictures were taken using a 63x objective and scale bar 941 represents 30 µm. (C) Coimmunoprecipitation of Hakai and LRP4 in HEK293 cells overexpressing FLAG-942 Hakai, LRP4 and HA-ubiquitin. (D) Hakai-dependent ubiquitination of LRP4. FLAG-Hakai, HA-ubiquitin and 943 LRP4 were transiently transfected into HEK293 cells. Immunoprecipitation was performed with anti-LRP4 944 and analyzed by Western blot. (E) LRP4 and FLAG-Hakai levels in HCT116 cells treated with the 945 proteasome inhibitor MG132 analysed by Western blot. β-catenin were used as a positive control for 946 MG132 treatment. (F) HCT116 cells were transiently transfected with FLAG-Hakai and the next day treated 947 with MG132 (30 µM) for 6 h. Endogenous LRP4 levels were analysed by Western blot. (G) HCT116 cells 948 were transiently transfected with FLAG-Hakai and treated with 50 µM of Hakin-1. LRP4 endogenous levels 949 were analysed by Western Blot. 950 951 Fig 4. Hakai modulates β-catenin–TCF/LEF-1 signalling. Wnt signalling activity upon overexpression of the 952 indicated plasmid in colon cancer HCT116 treated with control L-cell conditioned medium (LCM) and 953 WNT3A conditioned medium (WCM). Cells were transfected with TOP-flash or FOP-flash reporter 954 plasmids together with the indicated plasmids (A) Hakai-V5 and/or LRP4; (B) Hakai-V5 and/or LGR5 and 955 (C) Hakai-V5, LRP4 and/or LRP6. Relative luciferase activity is presented as fold of control in mean ± SEM 956 of three independent experiments.Two-ways ANOVA with Dunnet’s post hoc analysis was performed for 957 statistical analysis (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001). 958 959 Fig 5. Impact of Hakin-1 on tumoursphere formation. (A) Schematic representation of the experimental 960 workflow used to assess self-renewal. Hakin-1 (50 µM) was added at the initiation of cancer stem cell 961 induction. (B) Representative images of HT29 tumourspheres treated with Hakin-1 or DMSO (vehicle 962 control) five days after induction under stemness-promoting conditions. Images were acquired using a 963 using a 4x (upper images, scale bar: 250 µm) or 10x (lower images, scale bar: 200 µm) objective. (C) 964 Quantification of the number of tumourspheres formed from HT29 cells treated with Hakin-1 or DMSO. 965 (D) Quantification of tumoursphere size in the same conditions. Data in (C) and (D) are presented as 966 mean ± SEM. Quantification was performed using ImageJ software, and statistical analysis was conducted 967 with GraphPad Prism. (E-F) Effect of Hakin-1 on Wnt/β-catenin signalling and stem cell markers at mRNA 968 levels. RT-qPCR analysis of the expression of LEF-1 and TCF-1 transcription factors, stem cell markers LGR5 969 and NANOG (E), and CCND1, MMP7 and C-MYC (F). (G) Protein expression levels of E-cadherin 970 (differentiation marker), LGR5, Hakai and NANOG (stem cell markers) upon treatment with Hakin-1 (50 971 µM) were analysed by Western blot, using the indicated antibodies. Vinculin was used as loading control. 972 Protein bands were quantified using ImageJ and normalized to loading control. Results are expressed as 973 mean ± SEM. T-test was performed for statistical analysis (*p < 0.05; **p < 0.01; ***p < 0.001). 974 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 20, 2025. ; https://doi.org/10.1101/2025.11.20.689197doi: bioRxiv preprint 32 Fig 6. Localization of stem cell and differentiation markers in HT29 tumourspheres upon Hakin-1 975 treatment. Immunofluorescence analysis of HT29 tumourspheres treated with DMSO (vehicle control) or 976 Hakin-1 (50 µM). (A–D) Left panels: representative immunofluorescence images of E-cadherin (A), LGR5 977 (B), LRP4 (C), and MUC2 (D). Images were captured using a 20x objective (scale bar = 50 µm); magnified 978 images (circle images) were generated by digital zoom. Right panels: quantitative analysis of marker 979 expression. For E-cadherin, LRP4, and LGR5, fluorescence intensity was normalized to area and quantified 980 across at least 5 tumourspheres per condition, shown as scatter plots. For MUC2, the number of positive 981 cells was quantified by measuring the stained area and normalizing to DAPI-stained nuclear area. 982 Data are presented as mean ± SEM from three independent experiments. Statistical significance was 983 calculated using unpaired t-tests (*p < 0.05, **p < 0.01, ***p < 0.001) in GraphPad Prism. 984 985 Fig 7. Model proposed for Hakai action in Wnt/β-Catenin signalling pathway. Hakai absence: LRP4 acts 986 as an antagonist at the cell membrane, inhibiting Wnt/β-catenin signalling by interfering with LRP6 987 activity. As a result, cytoplasmic β-catenin is targeted for proteasomal degradation, and Wnt target genes 988 remain transcriptionally repressed. In the presence of Hakai, ubiquitination of LRP4 is induced, thereby 989 promoting its degradation. This action of Hakai can overcome the inhibitory effect of LRP4, resulting in 990 increased LRP6-mediated Wnt signalling. The increased Wnt/β-catenin signalling may promote stem cell 991 properties in tumourspheres. Image created with BioRender.com. 992 993 994 Supplementary figures 995 Fig. S1. LRP4 protein level is not recovered in presence of lysosome inhibitor Chloroquine nor in 996 presence of autophagy inhibitor 3-MA. Endogenous LRP4 and Hakai levels in HCT116 cells treated with 997 (A) lysosome inhibitor Chloroquine and (B) autophagy inhibitor 3-MA analyzed by Western blot. LC3 I/II 998 was used as a positive control for chloroquine and 3-MA treatment and GAPDH as a loading control. 999 1000 Fig. S2. Impact of the effect of Hakin-1 prior to stemness induction. (A) Schematic workflow. Hakin-1 1001 treatment was administered 48 hours prior to the induction of tumoursphere formation. (B) 1002 Representative images of tumourspheres in HT29 treated with DMSO (control) or 50 μM Hakin-1 1003 treatment on day 6 after stemness condition. Images were captured using a 4x (upper images, scale bar: 1004 250 μm) or 10x (lower images, scale bar: 200 μm) objective. (C) Tumoursphere size measurement. Images 1005 of tumourspheres derived from HT29 cells treated with DMSO (control) or Hakin-1 were captured, and 1006 surface area quantification was performed using ImageJ software. Tumoursphere area of at least 15 1007 tumourspheres per experiment were measured. (D) Tumourspheres formation quantification. The 1008 quantification of tumourspheres formed by HT29 cells treated with DMSO (control) or Hakin-1 was 1009 performed by self-renewal assay. Statistical analysis was performed using a t-test, with significance levels 1010 as *p < 0.05 and ****p < 0.0001. 1011 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 20, 2025. ; https://doi.org/10.1101/2025.11.20.689197doi: bioRxiv preprint 33 Fig. S3. Impact of Hakin-1 treatment on tumourspheres formation. (A) Schematic workflow. 1012 Tumourspheres already formed were treated with Hakin-1. (B) Representative images of tumourspheres 1013 in HT29 treated with DMSO (control) or Hakain-1 in day 8 after stemness condition a 4x (upper images, 1014 scale bar: 250 μm) or 10x (lower images, scale bar: 200 μm) objective. (C) Size of tumourspheres derived 1015 from HT29 cells treated with DMSO (control) or 50 μM Hakin-1. Surface area quantification was performed 1016 using ImageJ software. Tumoursphere area of at least 15 tumourspheres per experiment were measured. 1017 (D) Quantification of the number of tumourspheres after DMSO (control) or Hakin-1 treatment by self-1018 renewal assay. Results are expressed as mean ± SEM and statistical analysis was performed using a t-test 1019 of GraphPad Prism software (***p < 0.001 and ****p < 0.0001). 1020 1021 Fig. S4. Validation of identified PRPS2 protein as Hakai-regulated protein in tumoursphere by Western 1022 Blot. Levels of Hakai and PRPS2 proteins in Hakai-silenced HT29 colon cancer tumourspheres versus 1023 control tumoursphere were assessed by Western blot. GAPDH was used as loading control. 1024 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 20, 2025. ; https://doi.org/10.1101/2025.11.20.689197doi: bioRxiv preprint Figure 1 (C) Number of tumourspheres /500 cells **** **** (D) (A) (B) Monolayer culture Stemness induction Tumourspheres 72 h 5 days Hakai silencing Hakai-silencing (Dox) - + 200 µm (F) (E) Hakai-silencing (Dox) Hakai LGR5 Tubulin NANOG - + Protein expression levels **** * ** preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 20, 2025. ; https://doi.org/10.1101/2025.11.20.689197doi: bioRxiv preprint Hakai GAPDH Hakai-silencing (Dox) - + - + - + (C) 3381 103118 Hakai-silencing (Dox) - + (A) (B) (D) (H) Figure 2 CBLL1 ANXA2 ART4 CDH1 CTTN MTNR1B MTNR1B FOXN1 FLT1 PRDM16 JAK1 FLT3 SYKHOXC13STAT5A VAV3 LAMA5 MYOM3 LRP4 NTRK2 MYOM2 CAV3 SPHK2 BIN1 AKAP6 4 known substrates Top 20 predicted substrates E3 ligase (F) Hakai-silencing (Dox) - + - + - + - + (G) CSCs conditions LC-MS/MS timsTOF PRO Hakai silencing in HT29 cells Tumourspheres formation Tumourspheres phenotype characterization Hakai silencing confirmation Proteomic workflow Extraction (urea) Peptides Protein analysis Dox. - Dox. + 200 µm Hakai-silencing (Dox) - + Hakai-silencing (Dox) - + (E) Hakai-silenced vs Control preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 20, 2025. ; https://doi.org/10.1101/2025.11.20.689197doi: bioRxiv preprint Figure 3 (C) (D) LRP4 HA (Ub) LRP4 FLAG-Hakai IP:LRP4 IgG - - + + - + + + + + + +HA-Ubiquitin Input - - + + - + + + + + + + GAPDH Hakai LRP4 GAPDH LRP4 β-Catenin Hakai MG132 - 10µM 30µM (F) MG132 (30 µM) - - + GAPDH Hakai LRP4 FLAG-Hakai - + + (E) (G) Hakin-1 (50 µM) - - + GAPDH Hakai LRP4 FLAG-Hakai - + + LRP4 Hakai LRP4 FLAG-Hakai IP:LRP4 IgG - - + + - + + + + + + +HA-Ubiquitin Input - - + + - + + + + + + + GAPDH Hakai LRP4 (B) LRP4 V5-Hakai - + - + - - + + LRP4 V5 (Hakai) HEK293 HCT116 FLAG-Hakai - + - + GAPDH Hakai LRP4 (A) preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 20, 2025. ; https://doi.org/10.1101/2025.11.20.689197doi: bioRxiv preprint Figure 4 (A) (B) Relative luciferase activity * * **** * ** preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 20, 2025. ; https://doi.org/10.1101/2025.11.20.689197doi: bioRxiv preprint Figure 5 (A) Stemness induction Final tumourspheres 6 days Hakin-1 (C) (D) - + 0 100 200 300 400 Hakin-1 **** * Hakin-1 - + (B) (E) Hakin-1 Hakai LGR5 Vinculin NANOG Hakin-1 - + E-cadherin (G) (F) Hakin-1 * * * preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 20, 2025. ; https://doi.org/10.1101/2025.11.20.689197doi: bioRxiv preprint Figure 6 DAPI LGR5 Hakin-1 _ + Hakin-1 _ + DAPI E-cadherin(A) (B) * *** DAPI LRP4 Hakin-1 _ + Hakin-1 _ + DAPI MUC2 (C) ** ** (D) preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 20, 2025. ; https://doi.org/10.1101/2025.11.20.689197doi: bioRxiv preprint Figure 7 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 20, 2025. ; https://doi.org/10.1101/2025.11.20.689197doi: bioRxiv preprint Fig. S1. - 50µM 100µM Chloroquine Hakai GAPDH LRP4 LC3 I/II (A) 3-MA - 5mM 10mM Hakai GAPDH LRP4 LC3 I/II (B) preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 20, 2025. ; https://doi.org/10.1101/2025.11.20.689197doi: bioRxiv preprint Fig. S2.. (D)(C) (B)(A) Monolayer culture Stemness induction Final tumourspheres 48 h 6 days Hakin-1 **** ** Hakin-1 - + preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 20, 2025. ; https://doi.org/10.1101/2025.11.20.689197doi: bioRxiv preprint (B)(A) (D)(C) Stemness induction Final tumourspheres 6 days Hakin-1 48 h - + 0 100 200 300 400 500 Hakin-1 **** *** Fig. S3. Hakin-1 - + preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 20, 2025. ; https://doi.org/10.1101/2025.11.20.689197doi: bioRxiv preprint Fig. S4. GAPDH Hakai PRPS2 Hakai-silencing (Dox) - + preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted November 20, 2025. ; https://doi.org/10.1101/2025.11.20.689197doi: bioRxiv preprint