Intestinal Apc‐inactivation induces HSP25 dependency

Abstract

The majority of colorectal cancers (CRCs) present with early mutations in tumor suppressor gene APC. APC mutations result in oncogenic activation of the Wnt pathway, which is associated with hyperproliferation, cytoskeletal remodeling, and a global increase in mRNA translation. To compensate for the increased biosynthetic demand, cancer cells critically depend on protein chaperones to maintain proteostasis, although their function in CRC remains largely unexplored. In order to investigate the role of molecular chaperones in driving CRC initiation, we captured the transcriptomic profiles of murine wild type and Apc‐mutant organoids during active transformation. We discovered a strong transcriptional upregulation of Hspb1, which encodes small heat shock protein 25 (HSP25). We reveal an indispensable role for HSP25 in facilitating Apc‐driven transformation, using both in vitro organoid cultures and mouse models, and demonstrate that chemical inhibition of HSP25 using brivudine reduces the development of premalignant adenomas. These findings uncover a hitherto unknown vulnerability in intestinal transformation that could be exploited for the development of chemopreventive strategies in high‐risk individuals.

Upon loss of Apc, the most common early genetic lesion in colorectal cancer, cells upregulate a genetic program involved in buffering proteotoxic stress, including small heat shock protein HSP25. Here, we reveal that inhibition of HSP25 inhibits oncogenic transformation and prevents polyp formation.

Introduction

The development of colorectal cancer (CRC) is characterized by the stepwise accumulation of mutations (Fearon & Vogelstein, 1990). More than 80% of all CRCs present with early mutations in the tumor suppressor gene APC, leading to overactivation of the Wnt pathway and the formation of premalignant polyps (Muzny et al, 2012). Apc‐driven transformation of the intestinal epithelium is associated with hyperproliferation, impaired differentiation, and massive cytoskeletal remodeling (Sansom et al, 2004; Näthke, 2006; Barker et al, 2009). In order to fulfill the biosynthetic demands associated with these processes, Apc‐mutant cells increase their rate of global mRNA translation (Faller et al, 2014; Smit et al, 2020a) through a coordinated transcriptional network downstream of oncogenic c‐MYC and effector functions of mTORC1 signaling on mRNA translation (Van Riggelen et al, 2010). We have previously shown that proteotoxic stress, which is closely linked to increased protein synthesis levels, compromises the intestinal stem cell state and renders Apc‐mutant cells vulnerable to endoplasmic reticulum (ER) stress (van Lidth de Jeude et al, 2018; Meijer et al, 2021). Since the ER is mainly involved in posttranslational folding and processing of extracellular proteins, which are not directly used for the purpose of cell division or cytoskeletal reorganization, additional factors must be at play to ensure protein integrity in these hyperproliferative Apc‐mutant cells.

Heat shock proteins (HSPs) are a diverse class of molecular chaperones that perform multiple intracellular tasks related to the preservation of proteostasis in conditions of stress. In addition, HSPs are essential for cancer development due to their multifaceted role in controlling proliferation, differentiation, apoptosis, angiogenesis, and metastasis (Calderwood et al, 2006). It is becoming increasingly evident that HSPs also play a crucial role in controlling tumor initiation. For example, squamous cell carcinoma cells with oncogenic mutations in KRAS and TP53 that lack HSF1, a master transcriptional regulator that activates the heat shock pathway, are refractory to oncogenic transformation (Dai et al, 2007). In addition, loss of Hsp70 prevents the initiation of mammary tumors (Gong et al, 2015), indicating that oncogenic transformation itself may impose cellular proteotoxic stress and dependency on HSPs. Although the functions of protein chaperones in promoting cancer initiation and development are increasingly recognized, the influence of HSPs involved in Apc‐driven oncogenic transformation of intestinal epithelial cells is currently poorly defined.

In this study, we sought to identify molecular chaperones involved in Apc‐driven transformation by capturing the transcriptomic profiles of murine small intestinal organoids directly after homozygous deletion of Apc. We detected a significant upregulation of Heat shock protein B1 (Hspb1), which encodes HSP25, a chaperone known to be involved in the refolding of proteins in conditions of cellular stress, the inhibition of apoptosis, and remodeling of the actin cytoskeleton (Arrigo, 2017). We confirmed exclusive expression of Hspb1 in murine and human adenomas and revealed that upregulation of HSP25 is Wnt‐driven. Furthermore, we discovered that CRISPR/Cas9‐mediated knockout of Hspb1 or chemical inhibition of HSP25 using the antiviral compound brivudine (Bromovinyldeoxyuridine, BVDU) drastically impaired oncogenic transformation in vitro. Furthermore, oral administration of BVDU in a conditional knockout mouse model for Apc significantly reduced the abundance and expansion of Apc‐mutant crypts. In line with this, short‐term administration of BVDU resulted in a decrease in adenoma formation and increased survival rates. Taken together, our results indicate that Apc‐mutant cells are critically dependent on HSP25, and inhibition of this small heat shock protein poses a new therapeutic target for the prevention of premalignant adenomas.

Results

Apc‐driven intestinal transformation results in increased expression of Hspb1

To identify genes that are upregulated in response to acute loss of Apc, we generated in vitro organoid cultures derived from Villin‐Cre ^ERT2 (wild type, WT) and Villin‐Cre ^ERT2 ;Apc ^fl/fl mice. Apc was conditionally inactivated by administration of 2 μM 4OH‐tamoxifen, and organoids were dissociated and repassaged to ensure full recombination (Fig 1A; Smit et al, 2020a, Data ref: Smit et al, 2020b). After passaging, Apc‐mutant organoids demonstrated cystic‐like growth, which is a typical sign of elevated Wnt‐β‐catenin signaling in intestinal organoids, while WT cultures maintained their crypt‐like phenotype (Fig 1B). Three days after reseeding, organoids were harvested and RNA was extracted for global gene expression analysis. We identified a set of genes belonging to the heat shock protein family that were specifically upregulated after Apc‐driven oncogenic transformation (Fig 1C), and enrichment in pathways indicative of proteotoxic and cellular stress (Fig EV1A–C). Of particular interest was Hspb1, which was the only gene that was also significantly upregulated in murine adenomas when compared with normal intestinal tissue, together with Wnt target genes Axin2 and Lgr5 (Fig 1D; Reed et al, 2015a, Data ref: Reed et al, 2015b). Hspb1 encodes HSP25, which is an ATP‐independent chaperone that keeps its client protein‐folding intermediates in a folding‐competent state for the HSP70 chaperone machinery during conditions of stress (Ehrnsperger et al, 1997). Moreover, HSP25 also possesses extra‐chaperone functions, including cytoskeleton stabilization and antioxidant activity, making it a potentially relevant factor for adaptation to Apc‐driven transformation (Bakthisaran et al, 2015). We validated the specific expression of Hsbp1 and the concomitant upregulation of HSP25 protein in Apc ^fl/fl organoids after in vitro recombination (Fig 1 E and F). Moreover, we detected and elevated Hspb1 expression in intestinal adenomas collected from Lgr5‐Cre ^ERT2 ;Apc ^fl/fl mice (Fig 1G) and spatially confirmed the expression of Hspb1 to adenomatous regions by RNA in situ hybridization (RNA‐ISH; Fig 1H). Importantly, we confirmed upregulation of the human homolog HSPB1 in adenoma tissues obtained from a cohort of 59 patients of which we received both adenomatous and adjacent intestinal tissue of normal macroscopic appearance, which suggests that HSPB1 upregulation in APC‐mutant cells is conserved from mouse to human (Fig 1I).

Figure 1. Identification of Hspb1 as Apc‐specific heat shock protein.

• A
  Schematic illustration of the in vitro recombination experiment. TAM, 4OH‐tamoxifen.
• B
  Representative phase images of the morphology of WT and Apc ^−/− organoids before mRNA isolation. Scalebar, 250 μm.
• C
  Heat map illustrating differentially expressed genes (DEGs) encoding heat shock proteins between WT and Apc ^−/− organoids.
• D
  Volcano plot of DEGs between normal mouse intestinal tissue and adenomas (GSE65461).
• E, F
  Relative gene expression of Hspb1 (***P = 0.0004, n = 6) (E) and protein expression of HSP25 (F) in unrecombined versus recombined murine organoids.
• G
  Gene expression of Hspb1 in normal (N) versus adenoma (A) tissue (***P = 0.0002, n = 4).
• H
  RNA‐ISH for Hspb1 in murine intestinal adenoma. Scalebar, 100 μm. Zoom panel, 25 μm.
• I
  Relative gene expression of human HSPB1 in paired sets of normal versus adenoma tissue, (****P < 0.0001, paired t‐test, n = 59 pairs).

Data information: All data are mean ± s.e.m., biological replicates, analyzed using unpaired two‐sided t‐test unless otherwise specified.

Source data are available online for this figure.

Figure EV1. Loss of Apc induces cellular stress.

• A–C
  Gene Set Enrichment Analysis for pathways involved in HSF1‐mediated heat shock response (A), chaperonin‐mediated protein folding (B), and p38/MAPK stress response (C). NES, Normalized Enrichment Score; P, nominal P‐value of the enrichment score, which is based on a phenotype‐based permutation test procedure as described in more detail in (Subramanian et al, 2005).

Source data are available online for this figure.

Hspb1 is upregulated in response to Wnt pathway activation

To assess whether the upregulation of Hspb1 is a direct result of Wnt/β‐catenin pathway activation, we used a β‐catenin mutant cancer cell line carrying a construct containing an inducible dominant negative TCF4 (dnTCF4), which upon doxycycline administration acts as inhibitor of the Wnt pathway (Fig 2A; Van de Wetering et al, 2002). As previously reported, incubation with doxycycline for 48 h revealed a marked decrease in Wnt target gene AXIN2 and an increase in P21 expression (Fig 2B; Van de Wetering et al, 2002). In addition, doxycycline treatment resulted in decreased HSPB1 mRNA levels (Fig 2B) and downregulated HSP27 protein, the human ortholog of HSP25 (Fig 2C), indicating that HSPB1 expression is TCF4‐dependent. Vice versa, activation of Wnt signaling by GSK3β inhibitor CHIR99021 increased Hspb1 expression in a dose‐dependent manner in WT murine organoids (Fig 2D). In agreement, treatment with CHIR99021 increased HSP25 protein expression of WT organoids to comparable levels as Apc‐mutant organoids, while levels of other heat shock proteins remained unaffected (Fig 2E). Together, these data indicate that loss of Apc and the subsequent activation of Wnt signaling directly drive the upregulation of Hspb1.

Figure 2. Hspb1 expression is Wnt‐regulated and sensitizes Apc‐mutant cells to HSP25 inhibition in vitro .

• A
  Schematic illustration of the doxycycline‐inducible dnTCF construct.
• B, C
  Relative gene expression of AXIN2 (*P = 0.0129), P21 (*P = 0.0103), and HSPB1(*P = 0.0236) (B), and protein expression of HSP27 (C) in Ls174t‐dnTCF4 cells in the presence or absence of doxycycline.
• D, E
  Relative gene expression of Hspb1 in murine WT organoids treated with different concentrations of CHIR‐99021 (**P = 0.0016, one way ANOVA, n = 4) (D), and protein expression of HSP25 in WT and Apc ^−/− organoids (E).
• F
  Schematic illustration of in vitro recombination experiment using Hspb1 KO clones.
• G, H
  Representative images of single cell Hspb1 KO clones passaged after loss of Apc (G), and quantification of their clonogenic potential (*P = 0.0147(#1), **P = 0.0035(#2), **P = 0.0032(#3)) (H). Scale bar, 500 μm.
• I, J
  Representative images of Apc ^−/− organoids cultured in the absence or presence of 60 μM BVDU (I), and quantification of their clonogenic potential (**P = 0.0039) (J). Scale bar, 500 μm.
• K, L
  Relative gene expression of Wnt target genes Axin2 (**P = 0.0080) and Lgr5 (*P = 0.0142) in Apc ^−/− organoids cultured in the absence or presence of 60 μM BVDU.
• M
  RNA‐ISH for Lgr5 in control or BVDU‐treated Apc ^−/− organoids. Scale bar, 50 μm, zoom panel 20 μm.

Data information: All data are mean ± s.e.m. n = 3 biological replicates, analyzed using unpaired two‐sided t‐test unless otherwise specified.

Source data are available online for this figure.

Inhibition and loss of HSP25 impairs oncogenic transformation in vitro

Given the marked upregulation of Hspb1 after Apc‐inactivation, we next assessed how loss of Hspb1 influences oncogenic transformation using organoid cultures. We first generated single‐cell Hspb1 knockout clones in a Villin‐Cre ^ERT2 ;Apc ^fl/fl background using CRISPR/Cas9 (Fig EV2A–C) and administered tamoxifen to recombine Apc in vitro (Fig 2F). We observed that inactivation of Hspb1 did not influence passaging potential of unrecombined (Apc ^fl/fl ) organoids (Fig 2G and H); however, conditional loss of Apc after tamoxifen administration significantly decreased outgrowth of transforming Apc‐mutant (Apc ^−/−) organoids (Fig 2G and H). Moreover, chemical inhibition of HSP25 using BVDU demonstrated a comparable decrease in clonogenicity in established Apc ^−/− organoids (Fig 2I and J), which was accompanied by a decrease in expression of Wnt target genes Axin2 and Lgr5 (Fig 2K–M), while leaving growth of the organoids unaffected (Fig EV3A). Importantly, BVDU treatment did not influence growth, clonogenicity, or Lgr5 expression in WT organoids (Fig EV3B–E), however, administration of both CHIR99021 and BVDU in WT organoids did result in loss of clonogenicity (Fig EV3F and G), confirming that HSP25 expression is Wnt regulated. Although the role of BVDU as HSP27 inhibitor has been described before (Heinrich et al, 2011), we confirmed functional inhibition of HSP25 by assessing recovery after heat shock stress, since protection against elevated temperatures is one of the best known roles of heat shock proteins (Landry et al, 1989). Short‐term heat shock of WT organoids resulted in a rapid increase in both Hspb1 and HSP25 expression (Fig EV3H and I), and incubation with BVDU decreased clonogenic capacity after heat shock (Fig EV3J–L). Importantly, also the unrecombined (Apc ^fl/fl) Hspb1 KO clones were sensitive to heat shock, thereby functionally validating these clones (Fig EV2D and E). Together, these data reveal that Apc‐mutant epithelial cells are specifically sensitive to loss of Hspb1 or inhibition of HSP25 during oncogenic transformation in vitro. We also observe increased HSPB1 expression in human colon organoids derived from patients with familial adenomatous polyposis (FAP), carrying heritable mutations in the APC gene predisposing them to the development of adenomas and CRC (Fig EV4A) and confirmed that BVDU treatment decreased clonogenicity of these human FAP (APC ^−/−) organoids (Fig EV4B and C).

Figure EV2. Validation of Hspb1 KO clones.

• A
  Overview of Hspb1 gene, positions of the sgRNA's and edited sites within the KO clones.
• B, C
  Sanger sequencing results of the edited regions in exon 1 (B) or exon 2 (C).
• D, E
  Illustration of the heat shock (HS) clonogenicity experiment (D), and quantification of clonogenicity in KO clones in untreated or HS‐treated conditions (****P < 0.0001 for KO#1, 2, and 3, n = 4 experiments, unpaired two‐sided t‐test).

Source data are available online for this figure.

Figure EV3. Effect of in vitro HSP25 inhibition using BVDU.

• A
  Growth curves of Apc‐mutant organoids.
• B–D
  Growth curves of WT organoids (B), representative images of WT organoids cultured in the absence or presence of 60 μM BVDU (C), and quantification of their clonogenic potential (D). Scale bar, 500 μm.
• E
  RNA‐ISH for Lgr5 in control or BVDU‐treated WT organoids. Scale bar, 50 μm, zoom panel 20 μm.
• F, G
  Representative images (F) and clonogenicity (G) of WT organoids treated with CHIR99021 in the absence or presence of BVDU. (****P < 0.0001, n = 4). Scale bar, 500 μm.
• H, I
  Relative Hspb1 (H) and HSP25 (I) expression in wild‐type organoids after heat shock (HS) treatment (**P = 0.0016).
• J–L
  Illustration of the heat shock (HS) clonogenicity experiment (J), representative images of control or BVDU‐treated WT organoids after HS treatment (K), and quantification of clonogenicity (*P = 0.0128) (L).

Data information: All data are mean ± s.e.m., n = 3 biological replicates, analyzed using unpaired two‐sided t‐test.

Source data are available online for this figure.

Figure EV4. Effect of BVDU on human FAP organoids.

• A
  Relative HSPB1 expression in normal colon organoids (n = 3) and organoids derived from patients with familial adenomatous polyposis (**P = 0.0013, n = 4, mean ± s.e.m.).
• B, C
  representative images of control or BVDU‐treated human organoids (B) and quantification of clonogenicity (C) (***P = 0.0009 (FAP1), ****P < 0.0001 (FAP2), *P = 0.0197 (FAP3), n = 4 wells per line, data are mean ± s.d.) scale bar, 500 μm.

Data information: All data are analyzed using unpaired two‐sided t‐test.

Source data are available online for this figure.

Inhibition of HSP25 prevents oncogenic clonal expansion and adenoma formation in vivo

To study the effect of BVDU on Apc‐driven oncogenic transformation in vivo, we first used Villin‐Cre ^ERT2 ;Apc ^fl/fl mice to activate epithelial‐wide recombination of Apc alleles in the intestine. As previously reported, upon injection of tamoxifen, these mice develop massive hyperproliferation throughout the epithelium followed by rapid death within a week (Heino et al, 2021). We injected two doses of 2 mg tamoxifen on two consecutive days, either in the presence or absence of BVDU (400 mg/kg), which was administered daily via oral gavage. Two hours before mice were sacrificed, they were injected with EdU to label actively dividing cells (Fig 3A). Visualization of EdU+ cells in intestinal tissues revealed a marked decrease in the length of the proliferative zone (Fig 3B and C), suggesting that BVDU treatment reversed the hyperproliferative phenotype associated with Apc loss. In line with these findings, isolation of crypts from control and BVDU‐treated mice revealed fewer recombined (Apc ^−/−) alleles in BVDU exposed crypt fractions (Fig EV5A and B), as well as reduced outgrowth in selection medium that lacks R‐spondin1 (EN‐medium; Fig EV5C), both indicative of inhibition of oncogenic transformation. In order to assess the effect of BVDU on adenoma formation, we used Lgr5‐Cre ^ERT2 ;Apc ^fl/fl mice, that allow for low‐dose recombination of Apc in the stem cell compartment and therefore is a more controlled model of intestinal transformation. First, we studied the short‐term effects of BVDU treatment on the number of mutant crypts (Fig 3D), which could be visualized by performing RNA in situ hybridization of Notum, a Wnt antagonist that was previously identified to be exclusively expressed by Apc ^−/− cells (Kleeman et al, 2020; Flanagan et al, 2021; van Neerven et al, 2021). Importantly, BVDU treatment did not influence Notum expression levels (Fig EV5D), nor did tamoxifen injection induce Hspb1 expression (Fig EV5E and F). We demonstrated that daily treatment of BVDU for 10 days decreased the abundance of Notum ⁺ crypts (Fig 3E and F). In addition, we observed a reduction in clone sizes and the number of fully populated “fixed” Notum ⁺ crypts within single crypt bottoms 14 days after tamoxifen injection (Fig 3G and H). These data indicate that inhibition of HSP25 by using BVDU reduces expansion of Apc‐mutant clones and thus also potentially inhibits tumor initiation. Therefore, we performed a subsequent study in which we activated oncogenic Wnt signaling in Lgr5‐Cre ^ERT2 ;Apc ^fl/fl mice in the absence or presence of BVDU for only 10 days surrounding the Apc‐inactivation and studied when these mice developed symptoms of adenoma development and needed to be sacrificed (Fig 3I). This revealed that BVDU‐treated mice display a significant survival advantage compared with untreated mice (Fig 3J). Of note, most BVDU mice eventually have to be sacrificed due to the development of a single tumor in the caecum that obstructed passage to the colon. Furthermore, assessment of the intestines of control and BVDU‐treated mice reveals a marked decrease in the number of adenomas, in particular in the distal region of the small intestine (Fig 3K and L). Taken together, our data reveal that Apc‐driven intestinal transformation drives the upregulation of Hspb1 mRNA and HSP25 protein in a Wnt‐dependent fashion, and inhibition of HSP25 using BVDU only during the tumor initiation phase effectively reduces the subsequent development of premalignant adenomas.

Figure 3. Inhibition of HSP25 prevents oncogenic transformation and adenoma development in vivo .

• A
  Experimental set‐up of short‐term assay using Villin‐Cre ^ERT2 ;Apc ^fl/fl mice. p.o., per os, orally.
• B, C
  Visualization of proliferating cells in tissues incubated with EdU for 2‐h (B), and quantification of the hyperproliferative zone (**P = 0.0057, n = 3 mice) (C). Scalebar, 50 μm.
• D
  Experimental set‐up of short‐term assay using Lgr5‐Cre ^ERT2 ;Apc ^fl/fl mice.
• E, F
  RNA‐ISH for Notum reveals Apc‐mutant clones in crypt bottoms of control or BVDU‐treated mice (E), and quantification of the abundance of mutant crypts (n = 2 mice per condition, n = 5 technical replicates per mouse) (F). Scalebar, 50 μm.
• G, H
  Clone size distributions of Notum + crypts in control (n = 265 crypts) or BVDU‐treated (n = 260 crypts) mice, the average clone size and number of fixed clones are included in the figure.
• I
  Experimental set‐up of tumor development assay using Lgr5‐Cre ^ERT2 ;Apc ^fl/fl mice, mice were sacrificed once they develop signs of adenoma development.
• J–L
  Survival curves for control and BVDU‐treated mice (P = 0.0351, Mantel‐Cox test) (J), macroscopic images of adenoma development in the distal small intestine (K), and boxplots for the number of adenomas per intestinal region (**P = 0.0013) (L), n = 8 control, n = 11 treated.

Data information: All data are mean ± s.e.m., boxplots, the box represents the 25^th–75^th percentile and the median (line), whiskers represent min to max values, biological replicates, analyzed using unpaired two‐sided t‐test unless otherwise specified.

Source data are available online for this figure.

Figure EV5. Validation of in vivo BVDU treatment.

• A
  Illustration of experimental setup for crypt isolation.
• B
  Ratio of unrecombined versus recombined Apc alleles in control and BVDU treated crypts (n = 2 mice per group, C, control, B, BVDU treated).
• C
  Outgrowth of crypts isolated from control and BVDU‐treated mice in ENR (mEGF, Noggin, and Rspondin1) or EN medium (mEGF, Noggin) (**P = 0.0010 (mouse 3), ***P < 0.0001 (mouse 4), n = 4 wells per condition, data are mean ± s.d.).
• D
  Relative Notum expression in control or BVDU‐treated Apc organoids (n = 3 experiments).
• E, F
  Relative Hspb1 expression in control or tamoxifen‐treated organoids (n = 3 experiments) (E) or intestinal tissues (F) (n = 2 mice per condition, n = 3 technical replicates, C, control, T, Tamoxifen).

Data information: All data are mean ± s.e.m., unless otherwise specified, analyzed using unpaired two‐sided t‐test.

Source data are available online for this figure.

Discussion

In this study, we identified HSP25 as a critical mediator of oncogenic transformation following loss of tumor suppressor Apc. HSP25, a small heat shock protein with critical functions in regulating various stress responses, is upregulated after loss of Apc in organoid cultures and in murine adenomatous tissues in a Wnt‐dependent fashion. We demonstrate that Apc‐mutant organoids are particularly sensitive to loss of Hspb1 or inhibition of HSP25 using BVDU, while growth kinetics and clonogenicity of WT organoids remain unaffected. Furthermore, BVDU administration significantly prevented oncogenic transformation and clonal expansion in vivo. Crucially, inhibition of HSP25 delayed the development and abundance of premalignant adenomas in vivo. Our study adds to the pleiotropic role of small HSPs such as HSP25 in facilitating CRC development, such as anti‐apoptotic effects, resistance to chemotherapeutics, and oncogenic mesenchymal signaling (Garrido et al, 1997; Henriques et al, 2018; Liu et al, 2020). However, in contrast to these studies, our data suggest that HSP25 is already crucial at an earlier stage by facilitating oncogenic transformation. This poses HSP25 inhibition using compounds such as BVDU as an interesting strategy to prevent tumor initiation, which is of particular relevance to high‐risk individuals, e.g., FAP patients. Importantly, clinical studies on the effects of BVDU for treatment of viral infections have reported minimal side effects in adults or children indicating potentially high tolerance to these agents (Wildiers & De Clercq, 1984; Benoit et al, 1985; Wassilew, 2005; Salvaggio & Gnann, 2017).

Materials and Methods

Animal experiments

This study made use of Lgr5‐EGFP‐Cre ^ERT2 , Villin‐Cre ^ERT2 , and Apc ^fl/fl mice that all have been described previously (Shibata et al, 1997; El Marjou et al, 2004; Barker et al, 2007) and were bred on a C57BL/6 background in the mouse facility of the Amsterdam UMC. All in vivo experiments are approved by the animal experimentation committee of the Amsterdam UMC – location AMC and are performed according to national guidelines under license number AVD1180020172125. Mice were housed under standard conditions, with temperatures between 20 and 24°C in 12‐h light/dark cycles with 40–70% humidity. For all studies, mice were between 6 and 12 weeks old at the start of the experiments. For short‐term experiments, both male and female mice were used, and for long‐term studies, only females were used. To assess the short‐term effect of brivudine (BVDU, MedChemExpress) on hyperproliferation upon conditional Apc loss, Villin‐Cre ^ERT2 ;Apc ^fl/fl mice were injected intraperitoneally (i.p.) with 2 mg tamoxifen (Sigma) dissolved in cornflower oil for two consecutive days. Mice were randomly assigned into the control group or were treated with BVDU (400 mg/kg), which was administered daily via oral gavage (p.o.) during the entire duration of the experiment. In order to assess hyperproliferation, EdU (Thermo Fisher) was injected i.p. (100 μl of 10 mM stock) 2 h before mice were sacrificed. To study the effects of BVDU on stem cell dynamics and adenoma formation, Lgr5‐EGFP‐Cre ^ERT2 ;Apc ^fl/fl mice were injected i.p. with a single dose of tamoxifen (2 mg) and randomly assigned to the control group or treated daily with BVDU for 10 consecutive days. For short‐term effects, mice were sacrificed 14 days after tamoxifen injection, for long‐term adenoma studies, mice were sacrificed when they started to lose weight (more than 20% weight loss compared with the highest measured weight), or when mice appeared moribund due to other characteristics associated with adenoma development.

Tissue processing and clone size quantifications

After the termination of all in vivo experiments, mice were dissected, and intestines were isolated and processed for further analysis. In short, intestines were cut open longitudinally and fixed in 4% paraformaldehyde (PFA) overnight at 4°C and kept in the dark. The next day, tissues were placed in 30% sucrose solution for 24 h at 4°C in order to preserve tissue integrity, after which tissues were frozen and stored at −80°C. Tissues were sliced using a Cryostar NX70 cryostat and carefully placed on glass slides. To determine the clone sizes of Apc‐mutant cells, RNA in situ hybridization (RNA‐ISH) was used to detect the presence of Notum mRNA according to a previously optimized protocol (van Neerven et al, 2021), slides were counterstained with hematoxylin, and mutant (Notum ⁺ ) clone sizes were quantified as proportions of the crypt circumference (1:8–8:8), and completely full (8:8) Notum ⁺ crypts are considered to be fixed by Apc‐mutant cells. Both adenomas and clone sizes were scored blindly.

Human tissue collection

We collected fresh paired biopsies from adenomas and the surrounding macroscopically normal‐appearing intestinal epithelium from 61 patients undergoing routine colectomy. The collection of material was approved by the medical ethical committee of the Amsterdam UMC under approval number (2015–206), and all patients provided written informed consent. Biopsies were dissociated using the FastPrep‐24‐5G (MP Biomedicals) in combination with Lysis Tubes S (Qiagen) for two times 1 min at 6.5/s with 1 min on ice in between, material was collected using the Allprep DNA/RNA Universal kit (Qiagen). The isolation and maintenance of human colon organoids were described before (van Neerven et al, 2021). Collection of material was approved by the Medical Ethical Committee of the Amsterdam AMC under approval numbers 2014–178 and 09–146 with written informed consent of the patients, and the experiments conformed to the principles set out in the WMA Declaration of Helsinki and the Department of Health and Human Services Belmont Report.

Organoid cultures

Mouse proximal small intestinal crypts were isolated from Villin‐Cre ^ERT2 ;Apc ^fl/fl mice as previously described (Sato et al, 2009), grown in Matrigel (Corning), and cultured in DMEM/F12 medium supplemented with N2, B27, Glutamax, Hepes (5 mM), Antibiotic/Antimycotic (all Gibco), and N‐acetyl‐L‐cysteine (1 mM, Sigma). The culture medium was freshly supplemented with mouse EGF (50 ng/ml, TEBU‐BIO), R‐spondin1, and Noggin (both conditioned media), hereafter referred to as “ENR medium.” Apc‐mutant (Apc ^−/−) organoids were cultured in the same medium but in the absence of R‐spondin1, or “EN medium.” To recombine the LoxP flanked Apc alleles, organoids were seeded and incubated with 1 μM of 4OH‐tamoxifen (Sigma) overnight, the next day medium was replaced for fresh ENR medium. Quantification of organoid growth was performed by bright‐field microscopic assessment of individual organoids that were cultured in the presence or absence of BVDU. At each timepoint, from day 1 to day 4 after seeding, images were taken and the mean organoid area of a minimum of 150 organoids per timepoint was quantified using ImageJ software. To generate Hspb1 CRISPR knockout (KO) lines, unrecombined Villin‐Cre ^ERT2 ;Apc ^fl/fl organoids were transduced with lentiviral particles generated from LentiCRISPR‐V2 plasmid (#52961, Addgene), by spinfection in the presence of 5 μM CHIR99021 (Axon Medchem), 10 μM ROCK inhibitor (Sigma), and 8 μg/ml Polybrene (Sigma) as previously described (van Neerven et al, 2022). To select for integration of the plasmid, organoids were incubated for 3 days in the presence of 1 μg/ml puromycin (InvivoGen), expanded, and sorted into 96‐well plates to generate single‐cell clones. Editing of the Hspb1 gene was confirmed by Sanger sequencing. For in vitro inhibition of HSP25, BVDU (60 μM) was added to the medium immediately after plating the organoids. For heat shock experiments, equal numbers of (control or treated) organoids were harvested, incubated at 37°C (control) or 42.5°C (heat shock) for 60‐min in 15 ml tubes containing ENR medium before replated in fresh matrigel covered with fresh ENR medium, and clonogenic outgrowth was determined 4 days after replating. For all organoid experiments, medium was refreshed every other day. All organoids were routinely screened for mycoplasma.

TCF4 assay

The maintenance and applications of the doxycycline‐inducible dominant negative TCF4 (dnTCF4) reporter cell lines have been described in detail elsewhere (Van de Wetering et al, 2002). For this study, we used the Ls174T colorectal cancer cell line (RRID:CVCL 1384), which is routinely cultured in DMEM/F12 (Gibco) supplemented with 10% FCS (Serana), 1% penicillin/streptomycin (Gibco), and 100× L‐glutamine (Gibco). In order to inactivate endogenous TCF4 and inhibit TCF4‐related transcription, cells were cultured in the presence of 1 μg/ml doxycycline (Sigma) for 48 h before RNA or protein was extracted. Ls174T cells were authenticated using STR profiling and routinely screened for mycoplasma.

Generation of CRISPR constructs

To generate CRISPR KO organoids, three distinct sgRNA's were designed using Benchling software, aimed at targeting either exon1 or exon2 of the Hspb1 gene. Sequences are sgRNA1: 5′ CGGTTGCCCGATGAGTGGTC 3′, sgRNA2: 5′ CTTCGCTCCGGAGGAGCTCA 3′, and sgRNA3: 5′ GAAGAAAGGCAGGACGAACA 3′. sgRNAs are cloned into the LentiCRISPR‐V2 plasmid (#52961, Addgene), and transformed into Stabl3 competent bacteria (Invitrogen). Sanger sequencing was performed to check for correct insertion of sgRNAs, and lentiviral particles were subsequently generated as previously described (van Neerven et al, 2022).

RNA extraction and RT‐qPCR analysis

For analysis of gene expression by qPCR, RNA was isolated using the Nucleospin RNA isolation kit (#740955, Bioke). cDNA was synthesized using SuperScript III RT (Sigma), and RT‐qPCRs were performed on the LightCycler 480 system using SYBR Green (both Roche). To analyze gene expression levels, the ΔΔCt method was applied, and all values were normalized to expression of housekeeping genes Hprt and Rpl37 (mouse) or GAPDH and GUSB (human). The following primers were used for RT‐qPCR:

Gene	Species	Sequence (5′–3′)
Hspb1	Mouse	FW: TCACCCGGAAATACACGCTC
		RV: GGCCTCGAAAGTAACCGGAA
Axin2	Mouse	FW: CCATGACGGACAGTAGCGTA
		RV: CTGCGATGCATCTCTCTCTG
Lgr5	Mouse	FW: TTCGTAGGCAACCCTTCTCT
		RV: TCCTGTCAAGTGAGGAAATTCA
Notum	Mouse	FW: CTGCGTGGTACACTCAAGGA
		RV: CCGTCCAATAGCTCCGTATG
Hprt	Mouse	FW: TGTAATGATCAGTCAACGGGGG
		RV: AGAGGTCCTTTTCACCAGCAA
Rpl37	Mouse	FW: CCAAGGCCTACCACCTTCAG
		RV: CAGTCCCGGTAGTGTTTCGT
AXIN2	Human	FW: CTCCTTATCGTGTGGGCAGT
		RV: CTTCATCCTCTCGGATCTGC
P21	Human	FW: AGTCAGTTCCTTGTGGAGCC
		RV: CATGGGTTCTGACGGACAT
HSPB1	Human	FW: GCGGAAATACACGCTGCCC
		RV: GACTCGAAGGTGACTGGGATG
GAPDH	Human	FW: AATCCCATCACCATCTTCCA
		RV: TGGACTCCACGACGTACTCA
GUSB	Human	FW: TGGTTGGAGAGCTCATTTGGA
		RV: GCACTCTCGTCGGTGACTGTT

Microarray experiment and analysis

Microarray data used in this study were generated and published before (Smit et al, 2020a) and are publicly available through the National Center for Biotechnology Information (NCBI) Gene Expression Omnibus (GEO) under accession number GSE143509. In short, to capture the transcriptional profiles of Apc‐mutant organoids on the verge of transformation, Villin‐Cre ^ERT2 (wild type, WT) and Villin‐Cre ^ERT2 ;Apc ^fl/fl organoids were plated, and recombination was activated using 1 μM of 4OH‐tamoxifen. After 4 days, organoids were passaged to ensure full recombination, and RNA was isolated 3 days later using the ISOLATE II RNA Mini Kit (BIO‐52073, Bioline). A total of 400 ng of purified RNA was amplified and labeled using the 3′ IVT Pico Kit (Affymetrix) and RNA Amplification Kit (Nugene) according to manufacturer's protocols. Microarray analysis of mouse organoids was performed using Affymetrix Clariom S mouse array by the Dutch Genomics Service and Support Provider (MAD, Science Park, University of Amsterdam, The Netherlands). Washing and staining were performed by the GeneChip Fluidics Station 450, and the scanning was performed using the GeneChip Scanner 3000 7G (both Thermo Fisher Scientific). Data normalization, statistical testing, and extraction of differentially expressed genes were performed using the R2 platform (R2 R2 database, 2019).

RNA‐ISH

RNA‐ISH was performed on both fixed‐frozen and paraffin embedded mouse organoids and tissues using the RNAscope 2.5 HD‐Brown kit (ACD Bio) according to manufacturer's protocols. RNAscope was used for the detection of Notum (probe #428981), Lgr5 (probe #312171), or Hspb1 (probe #488361) mRNA, or for the positive control Ppib (probe #313911), and slides were counterstained with hematoxylin.

Digital droplet PCR

To quantitatively assess the amount of unrecombined (Apc ^fl/fl) and recombined (Apc ^−/−) Apc alleles in intestinal crypt fractions, we applied digital droplet PCR using the QX200 Droplet Digital PCR System (Bio‐Rad) in combination with EvaGreen supermix (#1864034, Bio‐Rad). Primers used to detect these Apc alleles are “Apc common” FW 5′ GTTCTGTATCATGGAAAGATAGGTGGTC 3′, “Apc ^fl/fl” RV 5′ CACTCAAAACGCTTTTGAGGGTTG 3′, and “Apc ^−/−” RV 5′ GAGTACGGGGTCTCTGTCTCAGTGAAG 3′.

EdU assays

EdU incorporation assays were performed to assess the hyperproliferative phenotype of Apc‐mutant cells in vivo. EdU was administered 2 h before mice were sacrificed and tissues were immediately fixed in 4% PFA. For the visualization of proliferating cells, 10 μM thick cryosections were cut and processed for analysis using the EdU Click‐IT imaging kit (Thermo Fisher, C10234) according to the manufacturer's protocol. Cells were counterstained using Hoechst‐33342 and visualized using an SP8 confocal microscope (Leica).

Immunofluorescence

Stainings were performed on intact, fixed mouse organoids. Organoids were harvested at designated timepoints together with surrounding Matrigel, carefully resuspended in Cell Recovery Solution (Corning), and incubated for 30 min on ice. After centrifugation, supernatant was removed and pellets were washed once in 1%FCS/PBS, before getting fixed in 4% PFA for 10 min at room temperature, in the dark. Organoids were permeabilized using 0.1% Triton X‐100 in PBS, for 15 min at room temperature, whereafter they were incubated overnight with anti‐HSP25 protein (Enzo Life Sciences, ADI‐SPA‐801‐D, RRID:AB_1193481, 1:100) dissolved in antibody diluent (ScyTek) at 4°C. The next day, samples were washed twice and incubated with anti‐Rabbit‐Alexa647 (Thermo, A‐21244, RRID:AB 2535812, 1:500) for 1 h at room temperature, protected from light. Samples were washed twice, and the actin cytoskeleton was stained using ActinGreen (Thermo Fisher, R37110), whereafter samples were washed again and counterstained with Hoechst‐33342 for 5 min. Next, samples were mixed with prolong, placed on microscopic slides, and covered with glass coverslips.

Western blot

Protein lysates were prepared using RIPA Buffer (Thermo Fisher) and Halt protease inhibitor (Thermo Fisher) and incubated on ice for 10 min. Next, samples were centrifuged at 6,000 g for 10 min at 4°C, and the lysate (supernatant) was collected. Protein quantification was performed using the Pierce BCA protein assay kit (Thermo). After protein quantification, 30 μg lysates were prepared in 4× leammli sample buffer (Bio‐Rad), incubated for 5 min at 95°C, and centrifuged at 6,000 g for 10 min at 4°C. Protein samples were loaded into 4–15% pre‐cast gels (Bio‐Rad) and run on Mini‐PROTEAN Tetra system (Bio‐Rad). Gels were blotted on PVDF membranes (Bio‐Rad) using the TransBlot Turbo system (Bio‐Rad), blocked in 5% skim milk powder (Sigma) for 1 h, and washed in 1× TBST. Blots were incubated with primary antibodies in 5% BSA (Fitzgerald Industries) overnight at 4°C. The next day, blots were washed, incubated with secondary antibody for 1 h at room temperature. After incubation, blots were thoroughly washed, and proteins were visualized using Pierce ECL Western Blotting substrate (Thermo) according to manufacturer's protocol.

Primary antibodies are anti‐HSP25 (Enzo Life Sciences, ADI‐SPA‐801‐D, RRID:AB_1193481, 1:1,000), anti‐HSP27 (Santa Cruz, sc‐13132, RRID:AB 627755 1:1,000), anti‐HSPB8 (CST, 3059 S, RRID:AB 2248643, 1:1,000), anti‐DNAJB6 (Abcam, ab198995, 1:1,000), anti‐GAPDH (Millipore, MAB374, RRID:AB 2107445 1:1,000). Secondary antibodies are anti‐mouse‐HRP (Southern Biotech, 1070‐05, RRID:AB 2650509 1:10,000), anti‐rabbit‐HRP (Cell Signaling, 7074, RRID:AB 2099233, 1:5,000).

Statistical analysis

Visualization of the data and statistical analyses were performed using GraphPad Prism. For every experiment, the statistical test used is noted in the figure legends, significant results are represented with asterisks, *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001, ns, not significant. Exact P‐values are displayed in the figure legends. For all, sample sizes were determined based on previous studies with a comparable study design (Vermeulen et al, 2013; Van Der Heijden et al, 2016; van Neerven et al, 2021).

Author contributions

Sanne M van Neerven: Conceptualization; data curation; formal analysis; supervision; validation; investigation; visualization; methodology; writing – original draft; writing – review and editing. Wouter L Smit: Conceptualization; data curation; formal analysis; investigation; methodology; writing – review and editing. Milou S van Driel: Investigation. Vaishali Kakkar: Conceptualization; investigation. Nina E de Groot: Investigation. Lisanne Nijman: Investigation. Clara C Elbers: Funding acquisition. Nicolas Léveillé: Supervision. Jarom Heijmans: Conceptualization; supervision. Louis Vermeulen: Conceptualization; supervision; funding acquisition; validation; writing – original draft; writing – review and editing.

Disclosure and competing interests statement

LV received consultancy fees from Bayer, MSD, Genentech, Servier, and Pierre Fabre, but these had no relation to the content of this publication.

The paper explained.

Problem

Chemoprevention strategies for high‐risk CRC patients such as familial adenomatous polyposis (FAP) patients, who carry a heterozygous germline mutation in tumor suppressor gene APC, are currently lacking. Upon a second genetic hit in APC, the master tumor suppressor gene in most colorectal cancers (CRCs), intestinal epithelial cells rapidly undergo oncogenic transformation. These transformed cells give rise to excessive polyp formation, leading to a 100% risk of CRC development in these patients. Therefore, mechanistic insight and means to prevent oncogenic transformation are needed.

Results

We set out to capture the transcriptomic profiles of Apc‐mutant intestinal epithelial organoids during transformation and observed marked upregulation of a set of genes involved in protein folding and buffering of proteotoxic stress, in particular the small heat shock protein and chaperone HSP25, a mouse ortholog of human HSP27. Both genetic deletion and chemical inhibition in vitro studies revealed that clonal expansion of Apc‐mutant organoids was strongly dependent on HSP25, without affecting wild‐type organoids. To translate these findings to an in vivo setting, we conditionally deleted Apc in two distinct mouse models and demonstrated that oral administration of HSP25 inhibitor brivudine, a commonly used antiviral drug, significantly impaired adenoma formation. Critically, we also observed elevated HSP27 expression in human adenomas and in human FAP organoids and demonstrated that these FAP organoids were also sensitive to HSP27 inhibition.

Impact

This study reveals HSP25/27 as a specific vulnerability to transforming APC‐mutant cells, due to its role in channeling proteotoxic stress associated with oncogenic transformation. We propose HSP27 as a novel therapeutic target for chemoprevention strategies in patients at a high risk of developing APC‐driven CRCs such as FAP patients.

Supporting information

Expanded View Figures PDF

Source Data for Expanded View

Source Data for Figure 1

Source Data for Figure 2

Source Data for Figure 3

PDF+

EMBO Mol Med (2022) 14: e16194

Data availability

This study includes no data deposited in external repositories.