Membrane Bound Peroxiredoxin-1 Serves as a Biomarker for In Vivo Detection of Sessile Serrated Adenomas

Abstract

Aim: Sessile serrated adenomas (SSAs) are premalignant lesions driven by the BRAF^V600E mutation to give rise to colorectal cancers (CRCs). They are often missed during white light colonoscopy because of their subtle appearance. Previously, a fluorescently labeled 7mer peptide KCCFPAQ was shown to detect SSAs in vivo. We aim to identify the target of this peptide.

Results: Peroxiredoxin-1 (Prdx1) was identified as the binding partner of the peptide ligand. In vitro binding assays and immunofluorescence staining of human colon specimens ex vivo supported this result. Prdx1 was overexpressed on the membrane of cells with the BRAF^V600E mutation, and this effect was dependent on oxidative stress. RKO cells harboring the BRAF^V600E mutation and human SSA specimens showed higher oxidative stress as well as elevated levels of Prdx1 on the cell membrane.

Innovation and Conclusion: These results suggest that Prdx1 is overexpressed on the cell surface in the presence of oxidative stress and can serve as an imaging biomarker for in vivo detection of SSAs. Antioxid. Redox Signal. 36, 39–56.

Introduction

Colorectal cancer (CRC) is a leading cause of cancer-related deaths, and the incidence is growing rapidly in developing countries (46). Over the next decade, the global burden of this disease is expected to increase steadily (1). Screening for CRC is often performed endoscopically by visual detection of adenomas (43). Previously, CRC was believed to progress exclusively from the traditional adenoma–carcinoma sequence (13). Recently, sessile serrated adenomas (SSA) have been found to result in CRC via an alternate pathway (12, 29, 42). These premalignant lesions are frequently either flat or subtle in appearance and are difficult to detect with imaging using white light illumination alone (25, 37, 59). SSAs are found primarily in the proximal colon and have a high miss rate that may explain the reduced mortality benefit of colonoscopy in this segment of colon (4, 34, 47). SSAs are characterized by a somatic point mutation in BRAF^V600E that results in enhanced MEK and ERK signaling (10, 14, 35).

The peptide KCCFPAQ, hereafter KCC*, was identified empirically using phage display methods by performing an unbiased screen against human HT29 CRC cells (22). This approach was taken to maximize binding interactions and hence the fluorescence signal for real-time in vivo imaging. This peptide was labeled with fluorescein isothiocyanate (FITC) and administered topically in human subjects to detect SSA in vivo using a fluorescence colonoscope. At that time, the target for this peptide was unknown. In vivo molecular imaging has great potential for early detection and staging of cancer, personalized medicine, targeted therapy, and response to therapy. Therefore, imaging biomarkers are promising for advancing personalized medicine. Identifying new imaging target can also provide insight into disease mechanisms and lead to development of new therapeutic targets.

Innovation

Cancer cells are characterized by overexpression of membrane-associated proteins, including cell surface receptors, membrane transporters, and cell adhesion molecules. However, expression of antioxidant enzymes on the cell surface has not been reported previously. We showed that peroxiredoxin-1 translocates to the surface of cells with the BRAF^V600E mutation in the presence of oxidative stress. This effect provides an important imaging biomarker for sessile serrated adenomas that can be developed for early detection of premalignant lesions that are often missed during white light colonoscopy. This finding highlights the importance of exploring altered cellular processes during neoplastic transformation for identification of new cancer biomarkers.

Peroxiredoxin-1 (Prdx1) is a 22 kDa enzyme that protects tissues from oxidative damage induced by reactive oxygen species (ROS), including superoxide ions, hydrogen peroxide, and hydroxyl radicals (11). Prdx1 regulates ROS-dependent signaling pathways that play an important role in the progression and metastasis of many tumors, including lung (20, 26, 27), breast (52, 53), bladder (21), ovary (30, 63), esophagus (17, 41), biliary tract (64), liver (48), pancreas (5), and glioblastoma (49). Although usually found in the cytoplasm (44, 53), Prdx1 has also been found in the nucleus and mitochondria (6, 53, 56). Membrane-associated Prdx1 can be phosphorylated transiently resulting in inactivation of peroxidase activity (55). Prdx1 has been found on membrane protrusions of pancreatic ductal adenocarcinoma cells and plays an important role in cell motility and invasion (50). In this study, we aim to validate Prdx1 as the binding partner of the KCC* peptide and elucidate mechanisms for membrane translocation of Prdx1 in SSAs.

Results

Pull-down of peptide interaction partners

The workflow for target identification is shown in Supplementary Figure S1A. The KCC* peptide was immobilized on NHS-activated sepharose beads to perform a pull-down from an HT29 cell lysate. Intense bands were observed (red boxes) in elutions 1 and 2 (Fig. 1A). No bands were seen without peptide (none). The bands were excised and evaluated by mass spectrometry. Several proteins were found (Supplementary Data). Proteins with more than 300 peptide scans were detected and >70% coverage were considered as potential binding partners. A total of four proteins met these criteria (Supplementary Table S1). Structural proteins, including α-tubulin, β-tubulin, and β-actin, were excluded. Prdx1 with a molecular weight of 22 kDa was identified with 92% coverage and was found on mass spectrometry analysis in 420 MS-MS spectra with 36 unique peptides and an error of <10 ppm (Supplementary Fig. S1B). Mass spectrometry data were validated by Western blot. Western blot analysis of the eluate showed a band (red oval) that confirms precipitation of Prdx1 by KCC* (Fig. 1B). By comparison, the eluate without peptide showed very weak signal. Previous studies have reported that this protein is mostly localized in the cytoplasm (44). However, immunofluorescence (IF) staining with KCC* peptide shows surface staining (22), which suggests that the binding partner is localized on the membrane. Analysis of the HT29 cell lysate confirms the presence of Prdx1 in the membrane (M) and cytoplasm (C) (Fig. 1C).

FIG. 1.

KCC* interacts with Prdx1. (A) Three bands (red boxes) were identified from elutions 1 and 2 of a pull-down of the KCC* peptide interacting with an HT29 cell lysate. No bands were seen without peptide (none). (B) Western blot of KCC* pull-down shows the presence of Prdx1 (red oval). (C) Prdx1 expression was found in the membrane (M) and cytoplasm (C) of HT29 cells. (D) A pull-down assay showed increased binding of KCC* with purified Prdx1 protein (0–250 nM). (E) An apparent dissociation constant k_d = 10.4 ± 1.1 nM with R² = 0.91 was determined (equation described in the Materials and Methods section). (F) Precipitated Prdx1 protein increased with greater concentration of immobilized peptide. Prdx1, peroxiredoxin-1. Color images are available online.

A pull-down assay showed increased binding of purified Prdx1 protein to fixed quantities of KCC* in a concentration-dependent manner (Fig. 1D). These results were used to estimate an apparent dissociation constant of k_d = 10.4 ± 1.1 nM for binding by KCC* to Prdx1 (Fig. 1E). We also performed a pull-down assay using different concentrations of peptide to a fixed concentration of protein. We incubated a fixed concentration of Prdx1 (250 nM) with an increasing concentration of KCC* peptide. The quantity of precipitated Prdx1 protein increased with greater concentration of immobilized peptide, confirming specificity of the interaction (Fig. 1F).

Immunofluorescence

IF was performed to validate cell surface localization of Prdx1. Without permeabilization, the fluorescence was visualized predominantly on the cell surface (Supplementary Fig. S2A). In permeabilized cells, fluorescence was detected both on the cell membrane and in the cytoplasm (Supplementary Fig. S2B).

Identification of scrambled control peptide

A structural model (4XCS) for the Prdx1 protein was used to identify a scrambled peptide for control (Supplementary Fig. S3A). After evaluating several sequences, FPAQCKC, hereafter FPA*, was selected. Docking analysis revealed a lower binding energy for KCC* versus FPA*. The C-terminus of either peptide was labeled with FITC via a GGGSK linker, hereafter KCC*-FITC and FPA*-FITC (Supplementary Fig. S4A and B). The linker separates the peptide from the fluorophore to prevent steric hindrance. The absorbance and fluorescence emission spectra are shown in Supplementary Figure S4C and D. The peptides were synthesized with >95% purity by high-performance liquid chromatography (HPLC), and an experimental mass-to-charge ratio (m/z) of 1181.55 was measured for unlabeled peptides by mass spectrometry, which agrees with the expected value of 1182.38 (Supplementary Fig. S4E and F). KCC*-FITC shows strong binding to the surface (arrow) of HT29 cells by comparison with FPA*-FITC (Supplementary Fig. S3B). The mean fluorescence intensity was found to be significantly greater for KCC*-FITC versus FPA*-FITC (Supplementary Fig. S3C).

Target validation by two unique antibodies

Two distinct antibodies were used to validate Prdx1 as the target for KCC* on the pull-down assay. On Western blot, bands at 22 kDa were detected from the total (T) HT29 cell lysate and from the cytoplasm (C) and membrane (M) using anti-Prdx1 antibodies (#ab109498; Abcam and #HPA007730; Sigma) (Supplementary Fig. S5A and B). Both antibodies showed a strong pull-down of Prdx1 protein by KCC* compared with that for FPA* (Supplementary Fig. S5C). These antibodies were labeled with either FITC or Cy5.5, respectively. From confocal microscopy, strong binding by both antibodies was seen to the surface (arrows) of HT29 cells confirming Prdx1 localization on the membrane (Supplementary Fig. S5D).

Binding interactions between peptides and purified Prdx1 protein

We further validated the interaction between KCC* and Prdx1 through dot blot. We used His-tagged KCC*, hereafter KCC*-His to probe purified Prdx1. From a dot blot analysis, decreased reactivity was observed when purified Prdx1 protein was probed with increasing dilutions of KCC*-His (Supplementary Fig. S6A). In contrast, greater spot intensities resulted from increasing quantities of Prdx1 being probed with a fixed dilution of KCC*-His (Supplementary Fig. S6B). These results confirmed the interaction between KCC* and purified Prdx1. Specificity of the interaction was further confirmed through a competition assay. When preincubated with unlabeled KCC*, the spot intensity from KCC*-His reduced in a concentration-dependent manner (Supplementary Fig. S6C). Interestingly, KCC* peptide also showed competition with Prdx1 antibody. Preincubation of HT29 lysate with unlabeled KCC* inhibited immunoprecipitation of Prdx1 with anti-Prdx1 antibody in a concentration-dependent manner (Supplementary Fig. S6D). By comparison, competition from the scrambled peptide FPA* (control) showed no change (Supplementary Fig. S6E). Similarly, when HT29 cells were prestained with KCC*-FITC, the fluorescence intensity from anti-Prdx1-Cy5.5 antibody binding to Prdx1 decreased with increasing concentrations of KCC*-FITC (Supplementary Fig. S6F). These results confirmed that KCC* binds with Prdx1.

siRNA knockdown of Prdx1

siRNA-mediated knockdown of Prdx1 was performed to further validate KCC* binding to Prdx1. Knockdown of Prdx1 in HT29 cells was performed using two different siRNA, including Silencer™ select and Dermacon. KCC*-FITC (green) and anti-Prdx1*-Cy5.5 antibody (red) showed strong binding to the surface (arrows) of control HT29 cells transfected with siCL (Silencer select) (Fig. 2A). Reduced fluorescence intensities were observed with HT29 knockdown cells transfected with Silencer Select siRNA (siPrdx1) (Fig. 2B). Quantified intensities showed these differences to be significant (Fig. 2C). Western blot shows a significant reduction in Prdx1 expression in siRNA-treated cells (Fig. 2D). KCC*-FITC and anti-Prdx1-Cy5.5 binding colocalized to the surface of HT29 cells with a Pearson's coefficient of ρ = 0.71, as shown in the magnified image in Figure 2E. Similar results were found with the Dermacon siRNA (Supplementary Fig. S7A–D).

FIG. 2.

siRNA (Silencer™ select) knockdown of Prdx1 reduces fluorescence signal from KCC*-FITC. (A) KCC*-FITC (green) and anti-Prdx1-Cy5.5 (red) show strong binding to the surface (arrows) of control (siCL)-treated HT29 cells. Colocalization of binding (orange) can be appreciated on the merged image. (B) Peptide and antibody showed minimal intensity with knockdown HT29 cells treated with Silencer select siRNA (siPrdx1). (C) The mean intensities for KCC*-FITC and anti-Prdx1-Cy5.5 were significantly higher for control versus knockdown. (D) Western blot shows Prdx1 expression in control (siCL)- and knockdown (siPrdx1)-treated HT29 cells. (E) Expanded view of the merged image shows peptide and antibody colocalize (arrow) with a Pearson's coefficient of ρ = 0.71. Color images are available online.

Target validation with human colon specimens

We stained formalin-fixed paraffin-embedded (FFPE) colon specimens with anti-Prdx1 antibody. Strong binding with anti-Prdx1-Cy5.5 (red) was observed to the cell surface (arrow) of SSA (Fig. 3A). Minimal signal was seen with normal, adenoma, and hyperplasia. Similarly, with KCC*-FITC, strong fluorescence signal on the surface (arrow) of SSA was observed while minimal staining was found for normal, adenoma, and hyperplasia (Fig. 3B). Sections of SSA stained with KCC*-FITC and anti-Prdx1-Cy5.5 showed colocalization of binding in the merged image (Fig. 3C). Representative histology (H&E) for SSA shows the characteristic sawtooth pattern (arrow) (Fig. 3D). Histology for normal, adenoma, and hyperplasia is also shown. The mean fluorescence intensity with anti-Prdx1-Cy5.5 was significantly greater for SSA versus normal (N), adenoma (A), and hyperplasia (Fig. 3E). ROC curves showed high sensitivity and specificity for use of anti-Prdx1-Cy5.5 to distinguish SSA (Fig. 3F).

FIG. 3.

Prdx1 distinguishes SSA from normal, adenoma, and HP with high specificity and sensitivity. On confocal images, (A) anti-Prdx1-Cy5.5 (red) and (B) KCC*-FITC show (green) strong binding to the cell surface (arrows) of SSA while minimal staining was seen for normal, adenoma, and HP. (C) Colocalization of binding (arrows) by peptide (green) and antibody (red) to SSA was seen on the merged image. A correlation of ρ = 0.63 was measured from the expanded view of the region in the dashed box. (D) Representative histology (H&E) is shown for SSA, normal, adenoma, and HP from colon. (E) The mean fluorescence intensity was significantly greater for SSA than for normal (N), adenoma (A), and HP. The mean fluorescence intensity was significantly higher for anti-Prdx1-Cy5.5 to SSA versus normal (2.5-fold), adenoma (1.65-fold), and HP (2.5-fold). An ANOVA was fit with terms for four groups to log-transformed data. (F) ROC curves showed high sensitivity and specificity for KCC*-FITC to distinguish SSA from normal, adenoma, and HP. ANOVA, analysis of variance; SSA, sessile serrated adenoma; HP, hyperplasia. Color images are available online.

In vivo imaging of human colonoids implanted in mouse colon

Prdx1 overexpression by SSA was further validated using patient-derived colonoids. Human SSA, adenoma, and normal colonoids were implanted in the colon of immunocompromised mice. Using white light illumination, colonoids (arrows) appeared subtle in images collected endoscopically in vivo (Fig. 4A and Supplementary Videos S1– S3). After intrarectal administration of KCC*-FITC, bright fluorescence was seen from the SSA colonoids (arrows) but not from adenoma and normal colonoids (Fig. 4B and Supplementary Videos S4 and S5). Adjacent normal colonic mucosa showed minimal background. The mean fluorescence intensity for SSA was significantly greater than that for adenoma and normal (Fig. 4C and Supplementary Videos S6 and S7). After 3 days passed to allow for KCC*-FITC to clear, white light and fluorescence images were collected from the same colonoid using FPA*-FITC (control) (Supplementary Fig. S8A–C). The mean fluorescence intensity from SSA was significantly greater with KCC*-FITC than with FPA*-FITC (Supplementary Fig. S8D). Representative histology (H&E) for patient biopsies used for isolation of SSA, adenoma, and normal colonoids is shown in Supplementary Figure S8E–G.

FIG. 4.

KCC*FITC detects patient-derived SSA colonoids implanted on mouse colon. (A) Human SSA, adenoma, and normal colonoids (arrows) implanted in the colon of NOD-scid mice are indistinguishable on WL images collected endoscopically. (B) FL images collected after topical administration of KCC*-FITC shows strong signal from human SSA colonoid (arrow) but not adenoma and normal. (C) The mean fluorescence intensity was 3.6-fold and 2.8-fold greater for SSA (n = 7) than for adenoma (n = 7) and normal (n = 6), respectively. An ANOVA model with terms for three groups was fit to log-transformed data. WL, white light; FL, fluorescence. Color images are available online.

After imaging was completed, the mice were euthanized, and the colon was resected, divided, and formalin-fixed. IF was performed using anti-Cam5.2-FITC, an anti-cytokeratin antibody specific for human tissues. Strong staining by Cam5.2-FITC (green) and anti-Prdx1-Cy5.5 (red) was observed on the cell surface (arrows) of human SSA colonoids (Supplementary Fig. S9A). Mouse crypts (arrowhead) showed minimal staining. For the human adenoma and normal colonoids, strong signal was seen with Cam5.2-FITC (green), but minimal fluorescence was seen with anti-Prdx1-Cy5.5 (red) (Supplementary Fig. S9B and C).

Prdx1 expression in BRAF-mutated cells

Expression of Prdx1 in the membrane and cytoplasm of a panel of human CRC cells with and without the BRAF mutation was evaluated. HT29 and RKO cells (BRAF^V600E+) show higher membrane expression of Prdx1 versus SW620, DLD-1, and CCD841 cells (BRAF^V600E−) (Supplementary Fig. S10A). The band intensities were quantified, and these differences were found to be significant (Supplementary Fig. S10B). No difference was seen in the expression of Prdx1 in the cytoplasm of HT29, RKO, SW620, and DLD-1 cells (Supplementary Fig. S10C and D). CCD841, a normal epithelial cell, showed low expression of Prdx1 in both the membrane and cytoplasm. IF staining with anti-Prdx1 antibody was performed to validate the Western blot results. Anti-Prdx1-Cy5.5 (red) showed strong staining to the surface (arrows) of HT29 and RKO cells, whereas SW620, DLD-1, and CD841 cells showed weak intensity (Supplementary Fig. S10E). KCC*FITC (green) also showed bright signal with HT29 and RKO but not SW620, DLD-1, and CD841 cells (Supplementary Fig. S10F). These findings suggest that colon cancer cells with the BRAF^V600E mutation show higher expression of Prdx1 on the cell membrane.

Isogenic RKO cells with and without the BRAF mutation were used to validate the above findings. RKO parental cells, hereafter RKO^mut, have the BRAF mutation (BRAF^V600E+), and RKO isogenic mutant cells, hereafter RKO^wt, have wild-type BRAF (BRAF^V600E-). Prdx1 expression is significantly greater in the cell membrane of RKO^mut versus RKO^wt cells (Fig. 5A and B). By comparison, no significant difference was seen in the cytoplasm. Anti-Prdx1-Cy5.5 (red) and KCC*-FITC (green) showed strong staining to the surface (arrow) of RKO^mut by comparison with RKO^wt cells (Fig. 5C and D). The mean fluorescence intensity was significantly greater for RKO^mut versus RKO^wt cells using anti-Prdx1-Cy5.5 and KCC*-FITC (Fig. 5E).

FIG. 5.

RKO cells with BRAF^V600E mutation have higher membrane Prdx1 expression and oxidative stress in comparison with isogenic cells with wild-type BRAF. (A) Western blot analysis shows Prdx1 expression in the membrane and cytoplasm of RKO^mut (BRAF^V600E+) and RKO^wt (BRAF^V600E-) cells. (B) Quantified results showed that RKO^mut cells have significantly greater Prdx1 expression in the membrane but not the cytoplasm versus RKO^wt. An ANOVA model was fit with terms for four groups and six blots. (C) Anti-Prdx1-Cy5.5 (red) and KCC*-FITC (green) shows strong binding to the surface (arrows) of RKO^mut cells, (D) but minimal intensity with RKO^wt cells. (E) The mean fluorescence intensity was 1.8-fold higher for RKO^mut versus RKO^wt with antibody and 2.1-fold higher with peptide using t-tests on log-transformed data in triplicate measured from three slides per condition. (F) CellRox deep red showed strong signal (arrows) on RKO^mut cells but not RKO^wt. (G) The intensity was 6.1-fold higher for RKO^mut versus RKO^wt cells using the same experimental design and analysis as for panel. Color images are available online.

Previous studies have shown that Prdx1 secretion is dependent on cysteine oxidation and redox signaling, and mutant BRAF^V600E generates ROS in nontransformed epithelial cells (32, 45). Therefore, we hypothesized that RKO cells with BRAF^V600E mutation will have higher oxidative stress compared with those with wild-type BRAF. CellRox deep red was used to assess oxidative stress. Strong staining (arrow) in RKO^mut cells was observed compared with RKO^wt (Fig. 5F). The mean fluorescence intensity was significantly greater for RKO^mut versus RKO^wt cells (Fig. 5G). These results confirmed that cells with the BRAF^V600E mutation have higher oxidative stress. Next, we compared the oxidative stress in human sections of SSA, normal, adenoma, and hyperplasia using an anti-8-OHdG antibody. Strong staining with anti-8-OHdG labeled with Cy5.5 was observed in SSA by comparison with normal, adenoma, and hyperplasia (Fig. 6A). The fluorescence intensities were quantified, and the mean value for SSA was significantly higher than that for normal, adenoma, and hyperplasia (Fig. 6B). These results were supported by immunohistochemistry (IHC) (Fig. 6C).

FIG. 6.

Human SSA specimen show higher oxidative stress in comparison with normal, adenoma, and HP. (A) Immunofluorescence shows bright signal (red) for representative SSA specimen stained with anti-8-OHdG-Cy5.5 by comparison with normal, adenoma, and HP. (B) The mean fluorescence intensity for SSA was found to be significantly higher than that for normal (N), adenoma (A), and HP. An ANOVA was fit with terms for four groups. (C) Immunohistochemistry shows strong reactivity for SSA by comparison with normal, adenoma, and HP. Color images are available online.

Effect of oxidative stress on Prdx1 expression

HT29 cells were treated with increasing concentrations of ROS inducer (menadione) and quencher (N-acetylcysteine; NAC), and membrane localization of Prdx1 was evaluated. ROS induction and quenching was confirmed using ROS-Glo™ H₂O₂ Assay (Supplementary Fig. S13B and C). Western blot showed that increasing concentrations of menadione results in greater levels of secreted as well as membrane bound Prdx1, whereas cytoplasmic expression of Prdx1 decreased (Fig. 7A and Supplementary Fig. S11A). Similarly, IF staining increased with greater concentrations of menadione (Supplementary Fig. S12A). The fluorescence intensities were quantified, and this increase was found to be significant (Supplementary Fig. S12C).

FIG. 7.

Membrane localization of Prdx1 is dependent on oxidative stress. (A) HT29 cells treated with the ROS inducer menadione show a concentration-dependent increase in Prdx1 secretion (supernatant), dimerization, and membrane localization. (B) HT29 cells treated with the ROS quencher NAC showed a concentration-dependent decrease in Prdx1 secretion, dimerization, and membrane localization. M, membrane; C, cytoplasm; NAC, N-acetylcystine; ROS, reactive oxygen species.

Oxidation of cysteine has been reported to result in the dimerization of Prdx1 and lead to secretion through an exosome mediated pathway (32). We found that levels of membrane-associated Prdx1 dimers increased with greater concentrations of menadione, whereas cytoplasmic dimer levels decreased (Fig. 7A and Supplementary Fig. S11A). These results suggest that elevated cellular oxidative stress causes dimerization and membrane translocation of Prdx1. By comparison, treatment with the ROS quencher NAC resulted in decreased secretion, membrane translocation, and dimerization (Fig. 7B and Supplementary Fig. S11B). Consistent with a decrease in the membrane, the level of cytoplasmic Prdx1 increased. IF staining confirmed the finding from Western blot (Supplementary Fig. S12B). We observed a decrease in surface fluorescence intensities with an increase in NAC concentration (Supplementary Fig. S12B). Quantified results showed this difference to be statistically significant (Supplementary Fig. S12D).

We further confirmed this finding by analyzing membrane translocation and dimerization after treatment with H₂O₂. We found that with increasing concentration of H₂O₂, membrane translocation and dimerization of Prdx1 increased (Fig. 8A and C). Blocking the thiol group by pretreatment with N-ethylmaleimide (NEM) resulted in reduced membrane translocation and dimerization (Fig. 8B and D). To investigate the mechanism of membrane translocation, we blocked exosome secretion by treatment with GW4869. Inhibition of exosomal secretion was confirmed by acetylcholinesterase assay (Supplementary Fig. S13D). We found that blocking exosome secretion by treatment with GW4869 resulted in decreased translocation of Prdx1 in the membrane. Treatment with menadione alone increased membrane expression; however, treatment with GW4869 even in the presence of menadione decreased membrane expression (Fig. 8E and F).

FIG. 8.

Mechanism of dimerization and membrane translocation. (A) HT29 cells treated with H₂O₂ (without NEM pre-treatment) show increase dimerization and membrane localization of Prdx1. (B) Membrane translocation can be inhibited by blocking the thiol group with pretreatment of NEM. (C, D) Quantified results for Western blots. M, membrane; C, cytoplasm. (E) Exosome inhibition by GW4869 resulted in reduced membrane localization of Prdx1. (F) Quantified results for Western blots. NEM, N-ethylmaleimide.

Discussion

In this study, Prdx1 was precipitated from an HT29 cell lysate and was identified as the binding partner for the 7mer peptide KCCFPAQ (KCC*) using mass spectrometry. This peptide was identified previously using phage display methods and was shown to detect SSA endoscopically in human subjects (22). Multiple assays were performed to support specific binding of KCC* to Prdx1. KCC* was found to bind to crude protein from the cell lysate and from purified protein. The binding was concentration dependent. Dot blot and competition assays confirmed interaction specificity. Surprisingly, KCC* inhibited immunoprecipitation by Prdx1 antibody. The binding and membrane localization were confirmed with two unique antibodies. Both KCC*-FITC and anti-Prdx1 antibody showed strong staining on the membrane of SSA compared with normal, adenoma, and hyperplasia. Additionally, KCC*-FITC and anti-Prdx1 antibody were found to localize on the membrane of SSA. These findings confirm that KCC* binds with the membrane Prdx1. These results were confirmed with images collected in vivo from human SSA colonoids implanted in the colon of immunocompromised mice.

We demonstrate efficient methods for colonoid transplantation that enable physiological growth of human tumors in a preclinical model of CRC. Colonoids can be used to recapitulate the molecular and genetic profile of human disease (19, 62). Human genomic expression of these colonoids has been authenticated, and their genetic variations have been characterized (8). SSA colonoids were imaged endoscopically in vivo and appeared with subtle features to mimic the human condition. Target expression levels and genetic heterogeneity were representative of that seen in the clinic. Prdx1 expression in the colonoids was validated ex vivo, and the use of anti-Cam5.2 confirmed the presence of human specimens implanted in mouse colon. Prdx1 was found to be expressed on the cell membrane of SSA. By comparison, other mouse models of CRC have been genetically engineered to overexpress molecular targets and result in levels much higher than that found clinically (18, 58). Our study shows that patient-derived colonoids are promising as a preclinical model for in vivo imaging to validate target expression.

Although Prdx1 is normally expressed in the cytoplasm (44), we showed that this protein is also expressed in the membrane of transformed cells. We performed protein extraction using a method that has been used previously to analyze membrane proteins with <10% cytosolic contamination (38). We found Prdx1 in the membrane fraction of several different cell types, including colorectal, lung, and pancreas, in addition to colonoids and fresh colon biopsies (data not shown). Cell surface biotinylation and mass spectrometry were used to analyze membrane-bound proteins in CRC cells, and Prdx1 was found in the membrane fraction (3). Similarly, several other studies used cell surface biotinylation, and detected Prdx1 in the cell membrane (16, 40, 54). In this study, we used live cell staining to validate membrane localization of Prdx1. Antibodies were used to detect Prdx1 proteins present on the cell surface, whereas after permeabilization, Prdx1 was found in the cytoplasm as well.

Prdx1 is a small protein without a membrane localization signal. The mechanism for translocation of Prdx1 to the cell surface is unknown. Previously, secretion of Prdx1 has been reported to be dependent on cysteine oxidation and redox signaling (32). Furthermore, mutant BRAF^V600E has been shown to generate ROS in nontransformed epithelial cells (45). We hypothesized that isogenic RKO cells with or without mutated BRAF should have different levels of cellular oxidative stress. RKO cells harboring two copies of mutated BRAF^V600E+ showed increased oxidative stress compared with BRAF^V600E- cells. In line with oxidative stress, RKO^mut cells showed elevated Prdx1 expression in the membrane. Interestingly, SSA tissues known to harbor BRAF mutations also showed higher oxidative stress in comparison with normal, adenoma, and hyperplasia, as revealed by immunostaining with anti-8-OHdG. Furthermore, treatment with the ROS inducer menadione increased membrane localization in HT29 cells. In contrast, treatment with the ROS quencher NAC decreased membrane localization. Thus, our study suggests that SSA has higher oxidative stress due to the BRAF mutation. Higher oxidative stress results in the increased secretion and membrane localization of Prdx1 (Fig. 9). In this study, SSA tissues were shown to have higher Prdx1 expression in comparison with normal, adenoma, and hyperplasia. Since KCC* binds with Prdx1, it can detect SSA with high sensitivity and specificity. Thus, Prdx1 is a promising membrane target for cells with the BRAF^V600E mutation and can be used as a biomarker for imaging.

FIG. 9.

Schematic diagram. Point mutation in BRAF^V600E causes oxidative stress to cell. Elevated levels of ROS induce Prdx1 dimerization, secretion, and membrane localization. KCC peptide binds to Prdx1 after dimerization and translocating to the cell surface. Color images are available online.

In this study, we demonstrated that Prdx1 translocates to the cell membrane in an oxidative stress-dependent manner (Fig. 9). We showed that treatment with the oxidizing agent H₂O₂ increases membrane translocation and dimerization. Prdx1 contains 2 cysteine residues, which contain a thiol group, which is oxidized to form disulfide bond (7, 33). In our study, H₂O₂-induced dimerization was inhibited by treatment with the thiol blocker NEM. This result suggested that dimerization of Prdx1 occurs through the cysteine residue. Furthermore, we also found that pretreatment with NEM also reduced membrane expression, suggesting that dimerization is required for membrane translocation. In our study, we found that the exosome inhibitor GW4869 blocked membrane translocation, suggesting that Prdx1 is transported to membrane via exosomes.

Oxidative stress mediates a variety of important human diseases, such as neurodegenerative disorders, cardiovascular disease, rheumatoid arthritis, and cancer (39). In addition, the mechanism of action for many chemotherapeutic agents is based on increased intracellular ROS levels (60). Biomarkers for oxidative stress are important tools to assess disease severity and therapeutic effectiveness. Our results show that Prdx1 translocates to the cell membrane in the presence of oxidative stress, thus can be a promising imaging biomarker with use of fluorescently labeled peptides. In addition, Prdx1 is secreted and potentially can be used as a serological marker. Further studies are needed to demonstrate Prdx1 as a serological marker for oxidative stress. Prdx1 may also serve as a therapeutic target for tumors harboring the BRAF^V600E mutation, such as melanoma. By identifying the target for the KCC* peptide, we have shown that Prdx1 is promising for imaging of oxidative stress that may be generalized for use in disease diagnosis and therapy.

Conclusions

In summary, our study shows that Prdx1 is expressed on the cell membrane and serves as an imaging target for the KCC* peptide to bind. Prdx1 is overexpressed on the membrane of CRC cells that harbor the BRAF^V600E mutation, and translocation of Prdx1 to the cell membrane is induced by oxidative stress. Membrane-bound Prdx1 can be used as an imaging biomarker for endoscopic detection of sessile serrated adenomas, a precursor of CRC.

Materials and Methods

Cells, media, and chemicals

HT29 cells were cultured in McCoy 5A media. RKO, SW480, and DLD1 cells were cultured in RPMI media. CCD841 cells were cultured in EMEM media. All cells were incubated at 37°C in 5% CO₂. Fetal bovine serum (FBS) 10% and penicillin/streptomycin 1 × were added as supplements. The peptides were synthesized using standard Fmoc-mediated solid-phase synthesis. Details of these methods are described previously (24). FITC- and polyhistidine (His)-labeled peptides were obtained from Biomatik. Peptides were dissolved in phosphate buffered saline (PBS) and stored at −20°C. All experiments were conducted at physiological pH. The stability of peptides in PBS was analyzed by HPLC at room temperature (RT).

Pull-down of peptide interaction partners

Unlabeled peptides were conjugated to pre-activated resin per manufacturer's protocol (NHS-activated Sepharose 4 Fast Flow; GE Healthcare Life Sciences). Briefly, 1 mg of peptide was dissolved in 1 mL of coupling buffer (0.2 M NaHCO₃, 0.5 M NaCl, pH 8.3). The peptide solution (200 μL) was incubated with 100 μL of NHS bead slurry for 2 h at RT. Pull-down experiments were performed as described previously (36). Unbound sites were blocked with quenching buffer (0.2 M ethanolamine), followed by washing with low pH (0.1 M acetate, 0.5 M NaCl, pH 5) and high pH buffer (0.1 M Tris-HCl, pH 8.0). After washing 2 × with lysis buffer (50 mM HEPES, 150 mM NaCl, 0.2% NP-40, protease inhibitor cocktail, pH 7.5), the HT29 cell lysate was incubated with peptide coupled to NHS beads and incubated for 2 h at RT. The beads were collected by centrifugation at 1000 g and washed with buffer (50 mM HEPES, 300 mM NaCl, 0.2% NP-40, pH 7.5). Bound proteins were collected by boiling the beads in 1 × Laemmli sample buffer and separated on 4%–12% sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE), followed by staining with Coomassie brilliant blue. Bands of interest were excised and analyzed by mass spectrometry. Resin without bound peptides was used as control.

Excised bands were digested with trypsin and analyzed using LC-MS/MS (Fusion Lumos; Thermo Scientific). The NCBI protein database was interrogated using Mascot and X-Tandem! search engines to identify potential binding partners. Mascot output files were parsed into the Scaffold program (www.proteomesoftware.com) for filtering to assess for false discovery rates and to identify promising proteins.

Mass spectrometry analysis

Proteins were separated by SDS-PAGE (0.8 cm separation; Criterion XT Bis-Tris 12% gel; Bio-Rad). After staining with Coomassie blue, the protein containing regions were excised and digested in situ with trypsin (sequencing grade; Promega). The samples were analyzed by capillary HPLC-electrospray ionization–tandem mass spectrometry (HPLC-ESI-MS/MS) on a Thermo Scientific LTQ Orbitrap Velos Pro mass spectrometer. Online HPLC separation was accomplished with an Eksigent/AB Sciex NanoLC-Ultra 2-D HPLC system: column, PicoFrit™ (New Objective; 75 μm i.d.) packed to 15 cm with C18 adsorbent (Vydac; 218MS 5 μm, 300 Å). Precursor ions were acquired in the Orbitrap in centroid mode at 60,000 resolution (m/z 400); data-dependent collision-induced dissociation spectra of the six most intense ions in the precursor scan were acquired at the same time in the linear trap (30% normalized collision energy). Mascot (v2.6.2; Matrix Science) was used to search the spectra against the human subset of the UniProt database [Uniprot_Human 20161004 (92,910 sequences; 36,855,958 residues)] concatenated with a database of common protein contaminants [contaminants 20120713 (247 sequences, 128,130 residues)]. Cysteine carbamidomethylation was set as a fixed modification, and methionine oxidation and deamidation of glutamine and asparagine were considered variable modifications. Trypsin was specified as the proteolytic enzyme with one missed cleavage allowed. Subset search of the identified proteins by X! Tandem, cross-correlation with the Mascot results, and determination of protein and peptide identity probabilities were accomplished by Scaffold (v4.8.4; Proteome Software). The thresholds for acceptance of peptide and protein assignments in Scaffold were 95% and 99%, respectively, resulting in 0.0% protein false discovery rate.

The quantitative in vitro binding assay was performed as described previously with some modification (2, 15, 28). For interaction studies, we used 6 × histidine tagged KCC* (Biomatik), hereafter KCC*-His, and purified human Prdx1 protein (#PILFP0002; Invitrogen) and a His Protein Interaction Pull-Down Kit (#21277; Pierce) per manufacturer's protocol. Briefly, KCC*-His was immobilized on cobalt chelate agarose resin and incubated with twofold serial dilutions of Prdx1 (0–250 nM) at 4°C for 2 h with rotation. After incubation, the supernatant was removed, and the beads were washed thoroughly. Bound proteins were eluted with 50 μL of elution buffer containing 290 mM of imidazole. Equal amounts of eluted protein (20 μL) were loaded on SDS-PAGE. Western blot was performed to detect bound Prdx1 using monoclonal anti-Prdx1 antibody (#ab109498; Abcam). Anti-His antibody (#ab18184; Abcam) was used to probe KCC*-His. For control, cobalt chelate agarose resin without peptide was incubated with 250 nM of Prdx1. Bands were quantified with the ImageJ, and the intensity of samples without peptide was used for background subtraction. The band intensity for Prdx1 was normalized against the intensity of KCC*-His.

Measurement of apparent dissociation constant k_d

The apparent dissociation constant of k_d was estimated from the plot of the normalized band intensities versus Prdx1 protein concentration. Binding isotherms were defined using Prism (GraphPad Software) using the one site-specific binding equation: Y = B_max × X/(k_d+X), where Y equals the specific binding signal, X equals the concentration of Prdx1 added to KCC*-His bound to cobalt chelate agarose, and B_max is the maximum specific binding value (28).

Immunofluorescence

HT29 cells were cultured on coverslip to 70%–80% confluency. For live staining, the cells were washed and incubated with anti-Prdx1 antibody (#ab109498; Abcam) in blocking buffer at 1:500 dilution for 2 h at RT. The cells were washed 3 × with ambient PBS and incubated with Cy5.5-labeled secondary antibody for 1 h at RT. After washing 3 × with PBS, the cells were fixed with 4% paraformaldehyde (PFA), washed 3 × with PBS, and mounted with ProLong^® Gold Antifade Reagent with DAPI (#8961; Cell Signaling Technologies). Alternatively, HT29 cells were fixed with 4% PFA for 10 min and permeabilized with 0.1% Triton X-100 for 10 min at RT. After blocking with 5% goat serum, the cells were incubated with anti-Prdx1 antibody (#ab109498; Abcam) at 1:500 dilution for 2 h and with Cy5.5-labeled secondary antibody for 1 h at RT with 3 × washes after each incubation with antibody and mounted with ProLong Gold Antifade Reagent with DAPI (#8961; Cell Signaling Technologies).

Target validation by two unique antibodies

An HT29 cell lysate was incubated with peptide coupled to NHS beads and incubated for 2 h at RT. The pull-down eluate was loaded onto SDS-PAGE, transferred onto a polyvinylidene fluoride (PVDF) membrane (Immobilon-PSQ PVDF Membrane; Millipore Sigma), and probed with two unique rabbit monoclonal anti-Prdx1 antibodies (#ab109498; Abcam, and #HPA007730; Sigma). Protein bands were detected by enhanced chemiluminescence.

For antibody staining, HT29 cells were washed with PBS and fixed with 2.5% PFA for 5 min at RT to avoid permeabilization. After blocking with 5% bovine serum albumin (BSA) for 1 h, the cells were incubated with two unique anti-Prdx1 antibodies, including #ab109498 (Abcam) at 1:250 dilution and #HPA007730 (Sigma) at 1:250 dilution for 2 h at RT. After washing 3 × with PBS, either FITC- or Cy5.5-labeled secondary antibody, respectively, was added. After 1 h of incubation at RT, the cells were washed 3 × with PBS and mounted with ProLong Gold Antifade Reagent with DAPI (#8961; Cell Signaling Technologies).

Identification of scrambled control peptide

The crystal structure for human Prdx1 C83S mutant protein (4XCS) was obtained from the Protein Database (10.2210/pdb4XCS/pdb) and used in the docking study. The binding energies were calculated using simulation software, Hex, ver 8.0. The peptide absorbance spectra were collected using a spectrophotometer (NanoDrop 2000; Thermo Scientific). Fluorescence emission from a 5 μM peptide solution diluted in PBS was collected with a fiber-coupled spectrophotometer (Ocean Optics) using a diode-pumped solid-state laser (Technica Laser, Inc.) with excitation at λ_ex = 488 nm.

Binding interactions between peptides and purified Prdx1 protein

Purified human recombinant Prdx1 protein (# LF-P0002; Invitrogen) was spotted on a charged PVDF membrane. After blocking with 5% BSA, the membrane was incubated with decreasing concentrations of His-tagged KCC*, washed, and probed with HRP-conjugated anti-His antibody (# MA1-21315-HRP; Invitrogen). The blots were washed and developed using ECL Western Blotting Detection Reagent (Amersham). Increasing quantities of Prdx1 were spotted and incubated with KCC*-His. Spots were developed with HRP-conjugated anti-His antibody (#MA1-21315-HRP; Invitrogen). Unlabeled KCC* was preincubated with spotted Prdx1 in increasing quantities for competition. After blocking with 5% BSA, the membrane was incubated with His-tagged KCC* and HRP-conjugated anti-His antibody, as described above. See Supplementary Data for immunoprecipitation experiment.

Binding interaction by immunoprecipitation

Immunoprecipitation of Prdx1 by peptides from an HT29 cell lysate (500 μg) was evaluated using mouse monoclonal anti-Prdx1 antibody (#ab16745; Abcam). Competition was performed by preincubating with increasing concentrations (0–500 μM) of unlabeled peptide. Western blot analysis was performed to evaluate the amount of precipitated protein. For the loading control, anti-Prdx1 antibody was probed with HRP-conjugated mouse secondary antibody (#ab6789; Abcam). For IF, HT29 cells were incubated with increasing concentrations (0–100 μM) of KCC*-FITC, and then, IF was performed with anti-Prdx1 antibody (#ab109498; Abcam) as described above. Images were acquired with a confocal microscope (Nikon A-1) using a 40 × oil-immersion objective, and fluorescence intensities were quantified using the ImageJ software.

siRNA knockdown of Prdx1

Knockdown of Prdx1 expression by HT29 cells was performed with siRNA from two different sources: Silencer^® select validated siRNA, Ambion^® by Life Technologies and Dermacon. The specifications of the siRNAs are as follows: Silencer select PRDX1, ID# s10007, control Silencer select negative control #1, ON-TARGETplus human Prdx1 siRNA-SMARTpool (#L-010338-00-0005; Dermacon), and ON-TARGETplus non-targeting pool (#D-001810-10-05; Dermacon). The siRNA transfection was performed per manufacturer's protocol. Briefly, ∼10⁵ cells were seeded in 12-well culture plates at ∼60% confluence in McCoy 5A media supplemented with 10% FBS without antibiotics. The next day, the cells were transfected with 25 nM of siRNA using DharmaFECT transfection reagent (#T-2001-01; Dermacon). After 72 h, the cells were prepared for Western blot analysis and stained with peptide and monoclonal anti-Prdx1 antibody (#ab109498; Abcam). For colocalization studies, the cells were first stained with peptide, fixed with PFA, as described above, and then stained with antibody. Images were acquired with a confocal microscope (Nikon A-1) using a 60 × oil-immersion objective. Images were processed using the NES element software. Pearson's correlation coefficient was calculated using custom Matlab software (Mathworks, Inc.).

Target validation with human colon specimens

FFPE sections of human colon were obtained from the archived tissue bank of the Department of Pathology at the University of Michigan. Sections were cut with 5 μm thickness and mounted on glass slides (Superfrost Plus; Fischer Scientific) and deparaffinzed. Antigen retrieval was performed using sodium citrate buffer before staining. Sections were blocked with DAKO^® protein block serum-free (#X0909; Agilent) for 1 h at RT and incubated overnight with 1:250 dilution of monoclonal anti-Prdx-1 antibody (#ab109498; Abcam) at 4°C. The sections were washed 3 × with phosphate buffered saline with Tween 20 (PBST) for 3 min and incubated with FITC-labeled secondary antibody for 1 h at RT. After washing 3 × with PBST, the sections were mounted with coverslips (1.5 μm thickness) using ProLong Gold Antifade Reagent with DAPI (#8961; Cell Signaling Technologies). Fluorescence images were acquired with confocal microscopy using a 40 × oil-immersion objective. The images were collected with the same exposure for all specimens. The mean fluorescence intensities from each image were measured by placing three boxes with dimensions of 20 × 20 μm² completely within colonic epithelium using custom Matlab software (Mathworks, Inc.). Regions of saturated intensities were avoided. Adjacent sections were processed for routine histology (H&E) and evaluated by an expert gastrointestinal pathologist (H.D.A.).

Preparation of colon organoids (colonoids) for implantation

Human colonoid cultures were established by the Translational Tissue Modeling Laboratory at the University of Michigan (TTML; IRB REP00000105; unregulated designation) (8, 9). Tissue was collected from large adenomas (greater than 1 cm in dimension) that were surgically removed from patients. As per the clinical protocols, a sample of the tissue was FFPE for examination by a University of Michigan pathologist. Another fraction was collected into cold Dulbecco's phosphate buffered saline (DPBS) to establish in culture. Of this DPBS fraction, a small piece was also FFPE preserved for further histological validation, as well as for the isolation of DNA for genomic analysis. The normal specimen (#87; 21 years, male) was derived from deceased donor tissues. The tubular adenoma (TA) colonoid was derived from a biopsy of a 35 mm lesion in a 58 year-old female (#590). The SSA colonoid was derived from a biopsy of a 20 mm lesion in a 54 year-old female (#245). The genomic profile for the TA showed the following variants: BUB1B Arg550*, FLCN His429fs, MLH1 Lys443fs, MSH3 Lys381fs, PALB2 Met296fs, TCERG1 Arg889fs, CTNNA1 Met826Thr, CTNNB1 Ser45Phe, MAP2K4 Val127Ala, MLH3 Pro564Ser, PIK3R1 Arg188Cys (8). The SSA showed the following variants: BRAF Val600Glu, WBSCR17 Ser432Ser (8).

Short tandem repeat analysis was performed to identify human genomic DNA for 15 tetranucleotide repeat loci (AMPFLSTR Identifier Plus Assay, Applied Biosystems; University of Michigan DNA Sequencing Core) and to authenticate the specimens. An amelogenin gender determination marker run was performed using a 3730XL Genetic Analyzer (Applied Biosystems). Cultures were frequently tested for mycoplasma contamination using the Lonza MycoAlert Kit by the University of Michigan Transgenic Animal Model Core.

Culture conditions

Colonoid cultures were grown in Matrigel diluted to 8 mg/mL with growth media (#354234; Corning) in six-well tissue culture plates (#CC7682–7506; USA Scientific CytoOne). Cultures were passaged by triturating and dissociating the Matrigel in cold DPBS, centrifuging at 300 g, and plated the first day with 2.5 μM CHIR99021 (4423; Tocris) and 10 μM Y27632 (#125410; Tocris). The normal (#87) and sessile serrated (#245) colonoids were cultured in medium containing 50% L-WRN conditioned medium (ATCC CRL-3276; source of Wnt3a, R-spondin-3, and Noggin) (31) advanced DMEM/F-12 (#12634028; Gibco), N-2 media supplement (#17502048; Gibco), B-27 supplement minus vitamin A (#12587010; Gibco), 1 mM N-Acetyl-L-cysteine (#A9165; Sigma–Aldrich), 2 mM GlutaMax (#35050–061; Gibco), 10 mM HEPES (#15630080; Gibco), 50 U/mL penicillin, 0.05 mg/mL streptomycin (#15070063; Gibco), 100 μg/mL Primocin (#ant-pm-1; InvivoGen), 100 ng EGF/mL (#236-EG; R&D Systems, Inc.), 10 μM SB202190 (#S7067; Sigma–Aldrich), 500 nM A83–01 (#2939; R&D Tocris), and 10 μM Y27632 (#125410; Tocris) (8, 51).

The TA colonoid (#590) was cultured in Stemline™ Keratinocyte Medium II (#S0196; Sigma) and supplemented with Stemline Growth Supplement (#S9945; Sigma), 2 mM GlutaMax, 4 mM L-glutamine, and 50 μg/mL Primocin. Before harvest for transplantation, the #590 cultures were treated with 5 μM Y27632 for 18 h.

Preparation for transplant

Cultures were harvested from Matrigel in cold DPBS, triturated 30 × with a 1 mL pipette tip, and centrifuged at 300 g for 3 min at 4°C. The colonoid pellet was resuspended in 10 mL of cold DPBS and mechanically disassociated with the gentle MACS Octo Dissociator (130-096-427; Miltenyi Biotec) using the programs h_Tumor_01.01 followed by m_Lung-01.01. The colonoid fragments were further dissociated by 20 × pipetting with a 1 mL pipette tip. Large fragments were removed over a 100 μm BSA-coated cell strainer (#DL 352360; Corning). Slow centrifugation at 100 g was performed to reduce the single cell content. The cell aggregates were resuspended in cold DPBS supplemented with 5% Matrigel and 10 μM Y27632. ∼2.5–5 × 10⁵ cell aggregates in 200 μL were transplanted per mouse as described previously (57). Briefly, NOD-scid mice (005557 NOD Cg-Prkdc<scid> ll2rg<tm1Wjl>SzJ; Jackson Laboratory) were treated with 2.5% disuccinimidyl suberate (DSS) for 5 days. Mice were deprived of food for 12 h before transplantation. Anesthesia was induced and maintained via a nose cone with inhaled isoflurane mixed with oxygen at concentrations of 2%–4% at a flow rate of 0.5 L/min. ∼2.5–5 × 10⁵ cell aggregates in 200 μL were injected intrarectally. The rectum was closed immediately using tissue adhesive (SC361931; Santa Cruz Biotechnology) to promote retention of transplanted colonoids. All plasticware, including gentle MACS C-tubes (#130-093-237; Miltenyi), were treated with 0.1% BSA in DPBS to reduce colonoid adherence.

After completion of imaging, the mice were euthanized, and the colon specimens containing colonoid implants were fixed with formalin and embedded into paraffin. IF and IHC staining of the specimens was performed as described above. Cytokeratin cam5.2 (#345779; BD Biosciences) was used for human-specific staining, and monoclonal anti-Prdx1 antibody (#ab109498; Abcam) was used for staining of Prdx1.

In vivo imaging of human colonoids implanted in mouse colon

Mouse imaging studies were conducted with the approval of the University of Michigan Committee on the Use and Care of Animals (UCUCA). Imaging was performed ∼4 weeks after the colonoids were implanted. Before imaging, the mice were fasted for 4–6 h. Anesthesia was induced and maintained via a nose cone with inhaled isoflurane mixed with oxygen at concentrations of 2%–4% at a flow rate of 0.5 L/min. A small animal endoscope (Karl Storz Veterinary Endoscopy) was inserted into the rectum (57, 61). White light illumination was used first to identify the presence of viable colonoids. The target peptide (200 μM, 100 μL) was delivered intrarectally through the instrument channel. After 5 min for incubation, the colon was rinsed with warm tap water 3 × to remove stool and debris before image collection. After 3 days for the signal from the target peptide to clear, the same mice were imaged using the control peptide. A ratio of the fluorescence and reflectance images was determined to correct for differences in distance and geometry over the image field-of-view (23). A contour around regions of high-intensity pixels was defined for the target region. The neighboring contour with width of 30 pixels was assigned as background. The target-to-background (T/B) ratio was calculated for each lesion.

Prdx1 expression in BRAF-mutated cells

Human HT29, RKO, SW620, DLD-1, and CCD841 colon cells were cultured on chambered slides up to ∼60%–70% confluence. For peptide staining, the cells were washed with PBS, blocked with 5% BSA for 30 min at RT, and incubated with 5 μM of peptide for 10 min. The cells were washed 3 × with PBS, fixed with 4% PFA for 10 min, and mounted with ProLong Gold Antifade Reagent with DAPI (#8961; Cell Signaling Technologies).

For antibody staining, the cells were washed with PBS and fixed with 2.5% PFA for 5 min at RT to avoid permeabilization. After blocking with 5% BSA for 1 h, the cells were incubated with 1:250 dilution of anti-Prdx1 antibody (#ab109498; Abcam) for 2 h at RT. After washing 3 × with PBS, Cy5.5-labeled secondary antibody was added. After 1 h of incubation at RT, the cells were washed 3 × with PBS and mounted with ProLong Gold Antifade Reagent with DAPI (#8961; Cell Signaling Technologies). For colocalization, cells were first stained with peptide, fixed with PFA, as described above, and then stained with antibody. Images were acquired with a confocal microscope (Nikon A-1) using a 60 × oil-immersion objective. Images were processed using the NES element software. Pearson's correlation coefficient was calculated using custom Matlab software (Mathworks, Inc.).

RKO^mut and RKO^wt cells were cultured on chambered slides for analysis of cellular oxidative stress. The cells were washed with PBS, and fresh media was added. 2.5 μM of CellRox™ Deep Red Reagent (#C10422; Life Technologies) was added and incubated for 30 min at RT. The cells were washed 3 × with PBS and fixed with 4% PFA for 15 min followed by washing 3 × with PBS. Slides were mounted with SlowFade™ Gold Antifade Mountant with DAPI (#S36938; Life Technologies). Images were collected immediately with confocal microscopy using a 40 × oil-immersion objective. Alternatively, ROS levels were measured using ROS-Glo H₂O₂ Assay (#G8820; Promega).

Analysis of oxidative stress on human tissue

The oxidative stress on human specimen was analyzed by IF and IHC. IF staining was performed on FFPE sections of human with anti-8-hydroxy-2′-deoxyguanosine antibody (8-OHdG) (#ab48508; Abcam) and Cy5.5-labeled secondary antibody (#ab6947; Abcam). The images were collected with a confocal microscope (Nikon A-1) using 40 × oil-immersion objective at same exposure time for all specimens. For each sections, three images were collected. The mean fluorescence intensity from each image was determined by placing three boxes with dimensions of 20 × 20 μm² using custom software (Mathworks, Inc.). Regions of saturated intensities were avoided. One-way analysis of variance was performed to determine statistical significance.

Immunohistochemistry

IHC was performed on FFPE specimens of human colon. Tissues sections were deparaffinzed, and antigen unmasking was performed by boiling slides in sodium citrate buffer. After blocking with 10% normal goat serum, tissue sections were incubated with 8-OHdG (#ab48508; Abcam) antibody at a dilution of 1:250 at 4°C followed by washing 3 × with PBST for 3 min. Tissue sections were treated with 3% H₂O₂ for 15 min to remove endogenous peroxidase activity. Each section was incubated with biotinylated secondary antibody at 1:200 for 1 h at RT followed by washing 3 × with PBST for 3 min. Pre-mixed Elite Vectastain ABC reagent (Vector Labs) was added to each section and incubated for 30 min at RT. After washing 3 × with PBST for 3 min, the sections were developed with 3,3′-diaminobenzidine tetrahydrochloride substrate. The sections were monitored for 2 min and then quenched by immersing the slides in dH₂0. For nuclear staining, the sections were dipped in Gill's reagent followed by bluing solution and then dehydrated by immersing in 100% ethanol and 3 × washes of xylene for 2 min. Coverslips were mounted using permount mounting medium (#SP15–100; Fisher) in xylene.

Effect of oxidative stress on Prdx1 expression

HT29 cells were cultured in six-well plates up to ∼60%–70% confluence. The cells were washed with PBS, and fresh media was added. Different concentrations (10–50 μM) of menadione dissolved in DMSO (#M5625–25G; Sigma) were added to induce ROS, and the cells were harvested after 24 h. NAC (#A7250–25G; Sigma) was dissolved in PBS and added at different concentrations (0.1–20 mM) to quench the ROS, and the cells were harvested after 3 h. ROS induction or quenching was confirmed using ROS-Glo H₂O₂ Assay (#G8820; Promega). For Western blot analysis, the membrane and cytosolic proteins were isolated using a Mem-PER^TM Plus membrane protein extraction kit (#89842; Thermo Scientific) per manufacturer's protocol. For study of dimers, cells were treated with 150 μM DSS 30 min before harvesting. Protein samples were quantified using Pierce™ BCA Protein Assay Kit (#23225; Thermo Scientific), and 20 μg of protein was separated on SDS-PAGE (#NP0322BOX; Invitrogen). The cells were cultured in serum-free media at different time points, and the media was removed, boiled with sample buffer, and loaded on SDS-PAGE to evaluate the culture supernatant. Proteins were transferred onto PVDF membranes and probed with the anti-Prdx1 antibody (#ab109498; Abcam). After incubation with HRP-conjugated secondary antibody, bands were developed using Amersham ECL Western Blotting Detection Reagent (#RPN2106; GE Health care Life Sciences). For IF analysis, HT29 cells were cultured in eight-well chambered slides (Nunc™ Lab-Tek™ II CC2™ Chamber Slide System; Thermo Scientific™) up to ∼60%–70% confluence. Cells were washed with PBS and treated with either menadione or NAC at various concentrations (0–1000 nM). After 3 h, IF staining with Prdx1 antibody was performed.

Quantitative ROS measurement

Direct ROS quantitation was performed using ROS-Glo H₂O₂ Assay as per manufacturer's protocol.

Effect of H₂O₂ treatment on membrane translocation and dimerization

HT29 cells were cultured in a 60 mm disk up to ∼70%–80% confluence. The cells were washed with PBS, and fresh media was added. The cells were treated with different concentrations of H₂O₂ (0–100 μM). For membrane localization, the cells were harvested after 30 min. For Western blot analysis, the membrane and cytosolic proteins were isolated using a Mem-PER™ Plus membrane protein extraction kit (#89842; Thermo Scientific) per manufacturer's protocol. For study of dimers, after 10 min of H₂O₂ treatment, the cells were incubated with 150 μM of cross-linking reagent DSS containing 20 mM NEM (#E3876–5G; Sigma) for 30 min before harvesting. To block dimerization, HT29 cells were pretreated with 20 mM NEM followed by treatment with H₂O₂. Membrane translocation and dimerization were studied after NEM pretreatment according to the protocol mentioned above.

Mechanism of membrane translocation

To study the mechanism of membrane translocation, HT29 cells were cultured on a 60 mm disk up to 70%–80% confluence and treated with the exosome inhibitor GW4869 at different concentrations (0–20 μM) with or without 50 μM menadione. After 24 h, the cells were harvested and membrane and cytosolic proteins were isolated using a Mem-PER™ Plus membrane protein extraction kit (#89842; Thermo Scientific). To confirm exosome inhibition, an acetylcholinesterase assay was performed using assay kit (#ab138871; Abcam) per manufacturer's protocol.

Use of electronic laboratory notebook

An electronic laboratory notebook was not used.

Abbreviations Used

ANOVA

analysis of variance

CRC

colorectal cancer

DSS

disuccinimidyl suberate

FFPE

formalin-fixed paraffin-embedded

FITC

fluorescein isothiocyanate

FL

fluorescence

FPA

FPAQCKC

HP

hyperplasia

HPLC

high-performance liquid chromatography

IF

immunofluorescence

IHC

immunohistochemistry

KCC

KCCFPAQ

NAC

N-acetylcysteine

NEM

N-ethylmaleimide

PBS

phosphate buffered saline

PFA

paraformaldehyde

Prdx1

peroxiredoxin-1

PVDF

polyvinylidene fluoride

ROS

reactive oxygen species

SSA

sessile serrated adenoma

SDS-PAGE

sodium dodecyl sulfate–polyacrylamide gel electrophoresis

TA

tubular adenoma

Authors' Contributions

S.J. and T.D.W. designed the study. B.J. and J.C. designed and synthesized the peptides. S.J. and F.W. performed the experiments. M.K.D., J.R.P., G.M.N., and E.L.K. isolated and cultured the patient-derived colonoids. Y.M.S. performed colonoid transplantation. G.L. and M.L. performed image acquisition. R.K. performed statistical analysis. H.D.A. evaluated histology. S.J. and T.D.W. wrote and edited the article.

Author Disclosure Statement

The authors declare no conflict of interest.

Funding Information

Funding was provided in part by the National Institutes of Health, including R01 CA1933377, U01 EB028235 (T.D.W.), P30 CA046592 (R.K.), and R01 CA148828 (Y.M.S.).

Supplementary Material

Supplementary Data

Supplementary Figure S1

Supplementary Figure S2

Supplementary Figure S3

Supplementary Figure S4

Supplementary Figure S5

Supplementary Figure S6

Supplementary Figure S7

Supplementary Figure S8

Supplementary Figure S9

Supplementary Figure S10

Supplementary Figure S11

Supplementary Figure S12

Supplementary Figure S13

Supplementary Table S1

Supplementary Video S1

Supplementary Video S2

Supplementary Video S3

Supplementary Video S4

Supplementary Video S5

Supplementary Video S6

Supplementary Video S7