Age‐related cholesterol and colorectal cancer progression: Validating squalene epoxidase for high‐risk cases

Abstract As people age, the risk and progression of colorectal cancer (CRC), along with cholesterol levels, tend to increase. Nevertheless, epidemiological studies on serum lipids and CRC have produced conflicting results. We previously demonstrated that the reduction of squalene epoxidase (SQLE) due to accumulated cholesterol within cells accelerates CRC progression through the activation of the β‐catenin pathway. This study aimed to investigate the mechanism by which age‐related cholesterol accumulation within tissue accelerates CRC progression and to assess the clinical significance of SQLE in older individuals with elevated CRC risk. Using machine learning‐based digital image analysis with fluorescence‐immunohistochemistry, we assessed SQLE, GSK3βpS9 (GSK3β activity inhibition through serine 9 phosphorylation at GSK3β), p53 wild‐type (p53WT), and p53 mutant (p53MT) levels in CRC tissues. Our analysis revealed a significant reduction in SQLE, p53WT, and p53MT and increase in GSK3βpS9 levels, all associated with the substantial accumulation of intra‐tissue cholesterol in aged CRCs. Cox analysis underscored the significant influence of SQLE on overall survival and progression‐free survival in grade 2–3 CRC patients aged over 50. SQLE and GSK3βpS9 consistently exhibited outstanding prognostic and diagnostic performance, particularly in older individuals. Furthermore, combining SQLE with p53WT, p53MT, and GSK3βpS9 demonstrated a robust diagnostic ability in the older population. In conclusion, we have identified that individuals aged over 50 face an increased risk of CRC progression due to aging‐linked cholesterol accumulation within tissue and the subsequent reduction in SQLE levels. This study also provides valuable biomarkers, including SQLE and GSK3βpS9, for older patients at elevated risk of CRC.


| INTRODUC TI ON
Colorectal cancer (CRC) ranks as the second leading cause of cancer-related deaths worldwide (WHO, 2020).Moreover, most CRC-related deaths result from metastases (Li et al., 2020;Wang et al., 2020); therefore, it is imperative to identify biomarkers associated with high-risk populations vulnerable to malignant CRC, facilitating tailored surveillance and treatment.
CRC incidence and mortality increase with age, not only showing significant growth beyond age 50 (American Cancer Society, 2020); the most prevalent group at risk of CRC is diagnosed between ages 65 and 74 years.Several studies have reported a relationship between a cholesterol/low-density lipoprotein-related lifestyle and increased CRC fatality (Di et al., 2023;Farvid et al., 2021;Murphy et al., 2019).In addition, elevated serum cholesterol levels contribute to malignant colonic transformation by inducing tissue hypoxia (Herbey et al., 2005;Paraf et al., 1999).Cholesterol level has also been shown to increase with age (Downer et al., 2014;Hosseini et al., 2019).However, the age-dependent connection between increased cholesterol and accelerated CRC progression has yet to be established.
Squalene epoxidase (SQLE), a second rate-limiting enzyme in cholesterol biosynthesis (Chua et al., 2020;Gill et al., 2011), has been reported as a bona fide oncogene in several cancers, including breast (Brown et al., 2016) and liver (Liu et al., 2018).In addition, Gill et al. (2011) and Chua et al. (2020) reported that cholesterol exquisitely regulates the SQLE gene and protein by suppressing SQLE expression and inducing proteasomal degradation of the SQLE protein via MARCHF6-ubiquitination at the N-terminal 100 amino acid.
Notably, we demonstrated this disconnection under certain pathologic conditions on the expression of the SQLE gene and protein (Jun et al., 2021); SQLE degradation caused by accumulated cholesterol over a certain threshold accelerates CRC progression and metastasis via activating the β-catenin oncogenic pathway by inhibiting the p53 anti-tumor suppressor path and GSK3β activity.Our findings of the disruption of the p53/p21 pathway through SQLE reduction are substantiated by the following observations; Overexpression of p53 R273H and p53 WT hindered the survival of CRC cells (HCT116 and HT29) treated with small interfering RNAs targeting SQLE (siSQLE) under conditions of anoikis resistance.Conversely, the mutant p53 variant pLNCX-Flag-p53-R273H-mTAD, rendering its transactivation region nonfunctional, did not increase the survival of CRC cells treated with cholesterol or siSQLE under anoikis resistance (Jun et al., 2021).
In this study, we assessed the age-dependent relationship between cholesterol increase and CRC progression using human CRC samples grouped according to patient age and cancer grade with a machine learning analysis of digitized whole images and retrospective evaluation.Furthermore, we investigated the effectiveness of SQLE and the critical players in the β-catenin pathway (p53 wildtype: p53 WT , p53 mutant: p53 MT , and the GSK3β activity inhibition, as measured by the phosphorylation at serine 9 of GSK3β: GSK3β pS9 ) as prognostic and diagnostic biomarkers in high-risk CRC patients.

| Characteristics of the discovery cohort
Table 1 summarizes the clinicopathological characteristics of the discovery cohort.Out of the 1638 specimens, 1311 (80%) were from CRC patients, with a median age of 58 years (range: 24-86 years), consisting of 828 men (63.1%) and 483 women (36.9%).At the time of diagnosis, 646 (49.3%) were in Grade 2, and 162 (12.4%) were in Grade 3.This cohort also included 327 normal colon tissues (20%) to assess the diagnostic potential of the candidates.Additionally, we randomly divided the discovery cohort into training (70%) and testing sets (30%) using a random forest-based machine-learning approach (Figure S1).

| The association between cholesterol increase with age and heightened CRC progression
Recently, we demonstrated that lowering SQLE due to cholesterol expedites CRC progression (Jun et al., 2021).This acceleration is facilitated by activating the β-catenin oncogene and suppressing the p53-p21 pathways.This study assessed the expression levels of candidates (squalene epoxidase: SQLE, p53 wild type: p53 WT , p53 mutant: p53 MT , and GSK3β activity inhibition measured by phosphorylation at serine 9 of GSK3β:GSK3β pS9 ) according to patients' age and CRC aggressiveness.
Multi-color immunofluorescence staining revealed significant decreases in SQLE, p53 WT and p53 MT , along with an increase in GSK3β pS9 , in patients aged 50-70 compared to those aged 20-30 (Figure 1a-d).These findings were further validated in a separate BIO-BANK cohort (Figure 1e-g).Additionally, we confirmed our previous report (Jun et al., 2021), demonstrating the activation of the β-catenin pathway through SQLE reduction according to CRC progression (Figure S2).Notably, thyroid cancer tissue stained with antibodies also used for CRCs extended our previous findings (Jun et al., 2021), underscoring that SQLE reduction, GSK3β inhibition, and p53 degradation due to aging and increased CRC malignancy are specific events in gastrointestinal cancers (Figure S3).Our investigation, using tissue lysates categorized by patient age and cancer grade, corroborated a significant reduction in SQLE, p53 WT , and p53 MT and increase in GSK3β pS9 levels in malignant CRC tissues (Figure 2a,b).Interestingly, we identified elevated total cholesterol and cholesteryl ester accumulation in aged CRC tissue (Figure 2c,d).Next, to better understand the relationship between cholesterol increase and the activation of the β-catenin pathway via SQLE reduction, we evaluated the candidate levels in relation to total cholesterol and cholesteryl ester contents normalized by CRC progression and patient age (Figure 2e,f).We did not observe a buildup of total cholesterol and cholesteryl ester in advanced G3 CRC (Figure 2c,d) caused by the degradation of SQLE, which is involved in endogenous cholesterol biosynthesis (Figure 2a,b).
Nevertheless, in aged and malignant CRCs, we found significantly reduced SQLE, p53 WT and increased GSK3β PS9 levels (Figure 2e,f).
Notably, p53 MT did not exhibit age-dependent reduction despite its CRC grade-dependent reduction due to the range of the used antibody, which detecting not only the R273H p53 variant but also others unrelated to the transactivation region of p53 (Figure 2e,f).
Overall, this study establishes a direct association between agerelated tissue cholesterol accumulation and accelerated CRC progression in humans, mediated by the subsequent reduction of SQLE and activation of the β-catenin oncogenic pathway.
TA B L E 1 Clinicopathological characteristics of the discovery cohort.).Additionally, the anti-GSK3β pS9 antibody was employed to assess the level of the inactive form of GSK3β.*p values were determined using an unpaired two-sided t test.CRC, colorectal cancer; SQLE, squalene epoxidase.

| Clinical value of the candidates for a high-risk CRC population
Next, we evaluated the candidate's impact on the survival of a highrisk CRC population, adjusted for grades 2 and 3. Notably, SQLE showed a significant prognostic effect on overall survival (OS) We then evaluated the candidates' ability to distinguish between patients who would survive or die within a given time (t).For this purpose, we utilized a time-dependent ROC (timeROC) analysis, employing a weighted Cox regression for survival data to determine the area under the ROC curve (AUC) (Kamarudin et al., 2017): the times that maximized the AUC ratio were found to be 123 months for patients before the age of 50 and 137 months for those aged 50 and older (Figure 4a-d diagnostic efficiency in distinguishing CRCs from controls, in contrast to p53 WT and p53 MT (Figure S4a,b).These exceptional efficiencies of SQLE (AUC, 0.83 [95% CI, 0.8-0.9])and GSK3β PS9 (AUC, 0.70 [95% CI, 0.7-0.7])were further confirmed by using ROC curves utilizing the discriminant score (Figure S4c).These findings were validated in the testing set (SQLE, AUC, 0.81 [95% CI, 0.8-0.9];GSK3β pS9 , AUC, 0.70 [95% CI, 0.7-0.8]; Figure S4d).

| DISCUSS ION
This study reveals the correlation between cholesterol accumulation in tissue and CRC progression.It also underscores the significant clinical relevance of SQLE, particularly within high-risk CRC populations, notably among individuals over 50, where there's a distinct increase in both CRC incidence and mortality rates.
Furthermore, this study provides valuable biomarkers-SQLE and GSK3β pS9 , either independently or in conjunction with p53 WT and p53 MT -for risk stratification in populations facing heightened mortality risks due to elevated cholesterol levels.Our previous proof-of-concept demonstration (Jun et al., 2021) showed a cause-and-effect relationship between cholesterol increase and CRC progression.However, this relationship was primarily demonstrated in cell lines and animals.Thus, our demonstration needed to be further corroborated in humans.
The decline in CRC incidence, which had been decreasing annually by 3%-4% during the 2000s, slowed to 1% annually between 2011 and 2019 (Siegel et al., 2023).Consistently, this study identified increased CRC incidence and mortality using the Surveillance, Epidemiology, and End Results (SEER) Program's Public database between 2017 and 2019 (Figure S7a,b).This deceleration is primarily attributed to a rise in cases among individuals under 50 years old and in lifestyles characterized by high-cholesterol diets (Siegel et al., 2023;Xi & Xu, 2021).Our observation, in alignment with others', underlines age-dependent trends in both CRC incidence and mortality and cholesterol increase (Figures S7c,d and S8).
Nevertheless, establishing a cause-and-effect relationship between increased cholesterol and CRC incidence and mortality has not been previously determined.This study convincingly demonstrates the SQLE-mediated link between cholesterol accumulation with age and CRC progression.
Numerous studies have reported the association between obesity and CRC (Mandic et al., 2023;Yang et al., 2023;Ye et al., 2020).SQLE, which is the first oxygen-requiring enzyme involved in cholesterol biosynthesis, is upregulated during hypoxic conditions in premalignant formation (Coates et al., 2023).Nevertheless, cells tend to favor lipid uptake over internal biosynthesis systems in oxygen and nutrient deprivation during cancer progression.Villa et al. (2016) further supported this notion by demonstrating that glioblastoma cells exhibit increased reliance on external cholesterol by elevating the expression of the low-density lipoprotein receptor (LDLR) while inhibiting internal cholesterol production.We consistently revealed the elevated expression of LDLR alongside the downregulation of adenosine triphosphate (ATP)-binding cassette transporter 1 (ABCA1) and SQLE in the progression of gastrointestinal cancers (Jun et al., 2021).This study not only corroborates our previous report but also demonstrated that age-related accumulation of cholesterol within tissues-potentially through LDLR upregulation and Unlike other genes involved in cholesterol biosynthesis, such as FDFT1, which is upregulated by fasting (Weng et al., 2020), SQLE is reduced when cholesterol accumulates beyond a certain threshold (Gill et al., 2011).Both FDFT1 and SQLE are downregulated during the malignant transformation of CRC.However, while FDFT1 acts as a tumor suppressor by negatively regulating AKT/mTOR/HIF1α signaling in CRC cells (Weng et al., 2020), SQLE activates the β-catenin oncogenic pathway.Furthermore, the current study demonstrated that SQLE has a clinical impact on high-risk CRC patients, particularly in the context of cholesterol accumulation associated with aging.
The Wnt/β-catenin signaling pathway plays a crucial role in CRC development, progression, metastasis, and recurrence.Its abnormal activation is frequently observed in most CRC patients (Chen et al., 2023).Despite this, drugs targeting this pathway have primar- The mlDIA-fmIHC analysis elucidated the underlying mechanisms linking intra-tissue cholesterol accumulation due to aging with the progression of CRC.Additionally, 'R' and 'SPSS' were employed to evaluate the superior diagnostic and prognostic abilities of candidates including SQLE and GSK3β pS9 for CRC patients.Our approach integrated machine-learning-based quantitative analysis with multiplex immunohistochemistry, overcoming traditional IHC limitations, which are primarily derived from inter-and intra-observer dependency.Furthermore, this method identified potential candidates for risk stratification and prognosis among CRC patients, thereby ensuring research objectivity and offering valuable avenues for patient care.
Overall, this study effectively resolved the existing controversy surrounding cholesterol and CRC (Fang et al., 2021), confirming that CRC progression accelerates due to cholesterol accumulation within tissues with aging.Furthermore, although large-scale, long-term retrospective studies are still required, our research offers valuable biomarkers, including SQLE, for stratifying populations at higher risk of CRC.

| Sources of patients' samples
This study utilized a total of 1638 formalin-fixed paraffin-embedded (FFPE) CRC tissues obtained from Tissu eArray.com (formerly known as US Biomax, Derwood, MD, USA), constituting the discovery cohort.The characteristics of this cohort, presented in Table 1, were sourced from Tissu eArray.com.Additionally, we acquired 60 extra FFPE CRC samples and 10 frozen CRC samples from BIO-BANK (Chungnam National University Hospital, Daejeon, South Korea).
Our inclusion criteria encompassed primary colorectal cancer and normal colon tissues, along with pertinent clinical information about the patients of origin, which were retrieved from the source.(Cuschieri, 2019).
In summary, we processed deparaffinized and antigen-retrieved slides using a series of xylene, ethanol, and citrate-based buffers.
These slides were then incubated overnight with antibodies against SQLE, p53 WT , GSK3β (Santa Cruz Biotechnology, Dallas, TX, USA; SC-99144, SC-126, SC-377213, respectively), p53 MT (Abcam, Boston, MA, USA; ab32049), and GSK3β pS9 (Cell Signaling Technology, Danvers, MA, USA; 9323s).Next, each slide was treated with Alexa Fluor-633 goat anti-mouse and anti-rabbit, -546 goat anti-mouse, -568 goat anti-rabbit, or 488 goat anti-mouse antibody (Thermo For the prognostic analysis, we performed a time-dependent receiver operating characteristic (timeROC) analysis using a weighted Cox regression for survival data to determine the area under the ROC curve (Figure 4; Kamarudin et al., 2017).We also used ROC curves with or without a discriminant score for the diagnostic analysis.We performed timeROC analysis and ROC curves without a discriminant score using R, whereas ROC curves with a discriminant score were using SPSS.Survival analyses were conducted using the multivariate Cox proportional hazards methods (Figure 3).Survival analyses were performed using the multivariate Cox proportional hazards approach in R. Statistical significance for all measurements was assessed through an unpaired two-sided t test utilizing R software.
).Both SQLE (Figure4b, AUC, 0.89; Figure4d, AUC, 1.00; training and testing set, respectively) and GSK3β pS9 levels (Figure4b,d, AUC, 0.81; training and testing set) demonstrated outstanding prognostic performance in the population aged 50 years and older.However, this prognostic power was not validated in individuals under 50 (Figure4a,c).Additionally, we found that SQLE (AUC, 0.81 [95% CI, 0.8-0.9])andGSK3β pS9 (0.70 [95% CI, 0.7-0.8])demonstrated sufficient F I G U R E 2 The acceleration of CRC progression via the reduction of SQLE and subsequent activation of the β-catenin pathway facilitated by aging-dependent cholesterol accumulation.(a, b) Western blot analysis was performed using ten independent human CRC specimens.GAPDH served as the loading control (left).The expression levels were quantified (right) using ImageJ (NIH, USA).Total cholesterol (c) and cholesteryl ester (d) levels in CRC tissue lysates were measured following the manufacturer's instructions.Each dot in the plot represents a single CRC sample.(e, f) Each candidate's level according to CRC progression was recalculated in relation to the total cholesterol (e) and cholesteryl ester (f) concentration in each age group.p-values were determined by comparing the levels of G2 (a), G1 (b), and G2 20-30s CRC patients (c-f) using an unpaired t test.p53 WT primarily detects wild-type p53 (DO-1), p53 MT primarily detects mutant p53 (Y5), and GSK3β pS9 represents the inactive form of GSK3β.CRC, colorectal cancer; SQLE, squalene epoxidase.

F
Effects of the candidates with overall survival (OS) and progression-free survival (PFS) in aged high-risk colorectal cancer populations.Forest plots for hazard ratios (HRs) based on Cox proportional hazards regression for two groups: before (left) and after (right) 50 years of age, following grades 2 and 3 adjustment.The relationship between OS (a, b), PFS (c, d), and the continuous candidates was analyzed.p values are indicated on the dashed line of HR weights.(a, c) Training and (b, d) testing.

F
Prognostic and diagnostic performance of the continuous candidates in the high-risk CRC population.(a-d) For prognostic analysis, time-dependent ROC analysis was employed, with time intervals determined by maximizing the ROC curve through weighted Cox regression.The time intervals for patients before (left) and after (right) the age of 50 were 123 and 137, respectively.(e-h) For diagnostic analysis, linear discriminant distribution for the merged candidates (left) and ROC curves using the discriminant score for the merged candidate to diagnose CRC (right) are presented.The ratios of AUC, along with their corresponding 95% CIs, are provided.Training sets are depicted in a, b and e, f, while testing sets are shown in c, d and g, h.AUC, area under the ROC curve; CRC, colorectal cancer.ABCA1 downregulation-accelerates CRC progression via SQLE reduction.
ily been utilized in the preclinical stage.This study expands upon our previous findings by uncovering the underlying mechanisms linking age-dependent cholesterol accumulation to CRC progression through SQLE reduction and subsequent activation of the β-catenin/TCF-LEF pathway.As a result, this study suggests that quantifying SQLE in CRC tissues using machine learning-based digital image analysis with fluorescence-multiplex immunohistochemistry (mlDIA-fmIHC) can aid doctors in predicting whether patients are at an increased risk of CRC or in need of therapy targeting the β-catenin-TCF/LEF complex.Immunohistochemistry (IHC) has significantly broadened its role in disease diagnosis and outcome prediction, revealing crucial molecules relevant to pathogenesis and potential diagnostic or therapeutic targets, particularly in cancer research.However, conventional IHC, limited to single biomarker detection, and the subjective nature of semiquantitative scoring, susceptible to intra-and interobserver variability, pose limitations in clinical and research settings.To address these limitations, we employed multiplex immunohistochemistry capable of detecting multiple molecules in conjunction with CellProfiler (https:// cellp rofil er.org/ ), an intuitive image analysis tool utilizing machine learning classifiers: a machine learning-based digital image with fluorescence-multiplex immunohistochemistry (mlDIA-fmIHC).This method has reported remarkable quantitative high-resolution capabilities in assessing protein expression in tissue (Stirling et al., 2021), surpassing traditional methods like 'Dukes.' The studies using specimens from Tissu eArray.com and BIO-BANK received exemptions from the Public Institution Bioethics Committee (designated by the Ministry of Health and Welfare, as well as the Research Ethics Committee of the Korean Research Institute of Bioscience and Biotechnology: exemption codes P01-202104-31-006 and 2021-0838-001, respectively).The analyses were conducted between August 2012 and December 2022.This study adhered to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement

Discovery phase Case-control cohort
a represents the standard deviation.FI G U R E 1 Decreases in SQLE, p53 WT