Discovery of Breast Cancer Diagnostic Markers Using Lipase‐Based Plasma Proteomics of Lipid‐Bound Proteins

ABSTRACT

Membrane‐associated proteins, including those embedded in exosomal membranes or bound to plasma lipids, are promising disease biomarkers. However, their hydrophobic nature limits detection through conventional proteomic workflows, which primarily capture soluble proteins. In this study, we developed and optimized a lipase‐based pretreatment strategy to improve the detection of insoluble membrane proteins in human plasma. Lipase treatment using porcine pancreas‐derived enzymes was optimized to 37°C, 3 U, and 1 h, based on overall protein yield and membrane protein enrichment. Sequential application after depletion of high‐abundance proteins increased the proportion of membrane proteins from 12%–20%. Using this optimized workflow, we performed data‐independent acquisition‐based LC–MS/MS analysis of plasma samples from patients with early‐stage breast cancer and benign disease (n = 6 each). Notably, 38% of the uniquely detected or enriched proteins in cancer plasma were membrane‐associated. Among them, protocadherin 12 (PCDH12) was significantly elevated in the plasma of patients with breast cancer, indicating its potential as a novel diagnostic biomarker. This study highlights the use of lipase‐enhanced proteomics for revealing plasma membrane proteins and advancing noninvasive biomarker discovery for cancer.

1. Introduction

Insoluble proteins present in the exosome membrane and lipid‐bound proteins in the blood have the potential to be used as biomarkers for disease diagnosis [1, 2]. While plasma proteomics research has primarily focused on soluble proteins, analysis of insoluble proteins has the potential to further contribute to our understanding of the complexity of tumor biology [3, 4]. Membrane proteins are integral components of cellular and extracellular vesicle membranes and play essential roles in cell signaling, molecular transport, and intercellular communication, all of which are frequently dysregulated in cancer [5, 6]. Due to their structural characteristics, membrane proteins are often not adequately accounted for in standard proteomics analysis workflows [7].

Lipases are enzymes that break down phospholipids into glycerol and fatty acids, which can alter the cell membrane structure [8, 9]. By hydrolyzing triglyceride‐rich lipid components in plasma, lipase treatment may alter the lipid microenvironment and facilitate the release of insoluble membrane‐associated proteins, including those derived from blood cell membranes and circulating exosomes [10, 11]. Through this action, insoluble proteins fixed to the cell membrane may be released. This could provide important clues for the discovery of new biomarkers in exosomal membranes and lipid‐bound proteins.

Breast cancer (BCa) is the most common cancer among women worldwide, and it ranks first in cancer incidence in women in South Korea [12, 13]. The major risk factors include sex, age, and genetic factors. With the increasing focus on the early diagnosis of BCa, diagnostic methods using imaging and biopsy are commonly performed. However, biopsy may have lower diagnostic accuracy in certain situations. Therefore, additional information may be required owing to tumor heterogeneity. Additionally, the biopsy procedure is invasive, which may lead to patient hesitation. Owing to these limitations, noninvasive diagnostic methods based on bodily fluids, such as blood, have recently gained more attention.

Plasma contains a wealth of biological information, making it an accessible and promising sample for discovering biomarkers with high sensitivity and specificity for the early diagnosis and monitoring of treatment responses in patients. Furthermore, conventional hematological markers, such as CA15‐3 and CEA, are used as screening tests for breast metastatic cancer; however, their diagnostic sensitivity and specificity are low, limiting their clinical utility [14, 15, 16, 17, 18]. To overcome this, we tried a new method by treating patients with early BCa with lipase to extract membrane‐ or lipid‐bound proteins from the plasma and analyze the insoluble proteins.

2. Experimental

2.1. Plasma Sample Preparation

We investigated a consecutive series of patients who underwent breast surgery performed by a single surgeon between February 2024 and October 2024. The patients were diagnosed with benign disease or BCa before surgery. Informed consent was obtained from patients the day before surgery. This study was approved by the Institutional Review Board of Kangbuk Samsung Hospital, Sungkyunkwan University of Korea, on February 13, 2024 (Approval Number: KBSMC 2024‐01‐003‐001). Blood samples were obtained from all patients within 2 days before elective breast surgery. Peripheral blood sampling (21 mL) was performed regardless of fasting. After centrifugation, plasma samples were aliquoted and stored at −80°C at Biobank, Medical Research Institute, Kangbuk Samsung Hospital.

The lipase‐treated subgroups were incubated at 37°C with 3 units/μg of lipase for 1 h and lipase‐untreated subgroups were processed without enzymatic treatment. The samples were reduced with 200 mM dithiothreitol at 45°C for 30 min and subsequently alkylated with 400 mM iodoacetamide at 25°C for 30 min in the dark. Protein digestion was performed using the SP3 method. Peptide concentrations were quantified using a quantitative colorimetric peptide assay kit (Thermo Fisher Scientific, Waltham, MA, USA) and the samples were dried under a vacuum. The dried peptides were desalted using a Ziptip (Product No. ZTC18S960) and then stored at −80°C.

2.2. Label‐Free LC–MS/MS Quantitative Profiling

The fractionated peptides were reconstituted in a solution containing 98% water and 2% acetonitrile (ACN) and analyzed using a Q‐Exactive HF‐X Quadrupole‐Orbitrap MS platform (Thermo Fisher Scientific) connected to an Ultimate 3000 nano‐LC system (Thermo Fisher Scientific) equipped with a nano‐electrospray ionization source.

Mobile phase A consisted of water with 0.1% formic acid and mobile phase B consisted of ACN with 0.1% formic acid. Liquid chromatography (LC) gradient elution was performed for 120 min at a flow rate of 300 nL/min under the following conditions: 5% mobile phase B for 3 min, 28% mobile phase B for 107 min, and 95% mobile phase B for 10 min. The peptides were separated using a column (50 cm, 75 μm, 2 μm, PepMap RSLC C18). The ionization source voltage was set to 2 kV. Full MS scans were performed in the mass range of 375–1200 m/z at a resolution of 120 000 (m/z 200). The automatic gain control (AGC) target was set to 5 × 10³ and maximum injection time (Max IT) was 25 ms. Data were acquired in profile mode.

For data‐independent acquisition (DIA), precursor m/z scans were performed within the range of 375–975 m/z, using an isolation window of 10 m/z. DIA scans were acquired at a resolution of 30 000 (at m/z 200), an AGC target set to custom (100%), and a Max IT of 22 ms. Data were processed in centroid mode. Peak alignment and quantification were conducted using DIA‐NN (v. 1.8.1) with the default parameters.

2.3. DIA Peak Alignment and Bioinformatic Analysis

Raw DIA files were converted into the mzML format using MSConvert. Peptides were identified using DIA‐NN (v 1.8.1) with the UniProt reference proteome database (April 202404; 20 436 sequences).

The UniProt human reference proteome database was added by selecting the “Add FASTA file” option. The settings for “FASTA digest for library‐free search” and “Deep learning‐based spectra, RTs, and Ims prediction” were enabled. Trypsin/P was chosen as the protease, with “1 or 2” missed cleavages allowed, and a maximum of “1, 2, 3, 4, or 5” variable modifications permitted to generate different predicted libraries. Modifications such as “N‐term M excision,” “C carbamidomethylation,” “Ox(M),” and “Ac(N‐term)” were also enabled. The remaining parameters were set as follows: peptide length range from “7 to 30,” precursor charge range from “2 to 5,” precursor m/z range from “375 to 984,” and fragment ion m/z range from “150 to 1800.” The false discovery rate was set to 0.1%, and the minimum score threshold for peptide identification was adjusted accordingly. Based on these parameter sets, a predicted spectral library was generated using a FASTA file.

Gene Ontology (GO) annotation and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analyses were performed using the DAVID bioinformatics resource (https://david.ncifcrf.gov/). Relevance was analyzed using protein–protein interaction (PPI) data from STRING (https://string‐db.org). Membrane proteins were defined as those classified as “membrane” in the subcellular localization category of the DAVID‐based Functional Annotation Clustering.

2.4. Statistical Analysis

Following the quantification of the fractionated samples for each condition, the average intensity was calculated for each peptide. Differentially expressed proteins (DEPs) were identified based on a threshold of |log₂ fold change| > 1.5. Statistically significant differentially expressed proteins (sDEPs) were determined using a p value cutoff of 0.05, as calculated using Perseus software (v 2.0.7.0).

All quantitative data are presented as mean ± standard deviation derived from a minimum of three biologically independent replicates. Statistical significance was assessed using either unpaired Student's t‐test or one‐way analysis of variance, depending on the experimental design. A p value of less than 0.05 was considered indicative of statistical significance. All statistical analyses were conducted using the GraphPad Prism software (v 5.0; GraphPad Software, CA, USA).

3. Results and Discussion

To optimize the reaction conditions for lipases to degrade insoluble proteins in plasma, we investigated several variables. Porcine pancreatic lipase (Sigma‐Aldrich, St. Louis, MO, USA) was used to optimize the conditions. Experiment 1 aimed to optimize the conditions for lipase activity, including reaction temperature, enzyme concentration, and incubation time (Figure S1A). Reaction temperatures were tested at 20°C, 37°C, and 56°C; enzyme concentrations were set at 1, 3, and 5 units/μg; and incubation times were extended to 1, 2, and 3 h (Figure 1A). These conditions were designed to assess their effects on the release of lipid‐bound insoluble proteins.

FIGURE 1.

Optimization of lipase treatment conditions for plasma proteomics. (A) Effect of temperature, enzyme concentration, and incubation time on protein identification. (B) Bar graph showing the number of membrane proteins identified under various lipase treatment conditions (temperature, concentration, and incubation time). (C) Number of identified proteins according to lipase treatment order evaluation. Group a: only top 14 proteins depletion, Group b: lipase treatment before depletion, Group c: depletion before lipase treatment. (D) Subcellular localization of identified proteins under each condition. Data are presented as mean ± SD (n = 3), and significance was assessed using one‐way ANOVA and Students t‐test (*p < 0.05, ***p < 0.001).

A total of 457, 465, and 434 proteins were identified at 20°C, 37°C, and 56°C, respectively, with the highest yield observed at 37°C. When the enzyme concentrations were set to 1, 3, and 5 U, 426, 471, and 440 proteins were detected, respectively, with 3 U showing the highest number of proteins. Similarly, incubation for 1, 2, and 3 h yielded 471, 459, and 461 proteins, respectively, with minimal variation. The overlap of proteins detected under each condition is shown in a Venn diagram in Figure S2. We confirmed that the set of proteins detected in common remained largely unchanged despite changes in conditions (Figures S2A, S2B, and S2C). For membrane proteins, 70, 71, and 64 were identified at 20°C, 37°C, and 56°C, respectively. With respect to the enzyme concentration, 56, 71, and 60 membrane proteins were detected at 1, 3, and 5 U, respectively, with the highest yield at 3 U. According to incubation time, 71, 73, and 73 membrane proteins were identified after 1, 2, and 3 h, respectively, with no significant time‐dependent differences (Figure 1B). Based on these results, 37°C, 3 U, and 1 h were selected as the optimal lipase treatment conditions. Although the number of membrane proteins identified at 20°C and 37°C was similar, 37°C was chosen as it better reflects physiological plasma conditions in the human body. As a result, the lipase in the plasma sample was 3 U, and the reaction conditions were established at 1 h and 37°C.

Next, we evaluated the order of lipase and high‐abundance protein depletion using plasma proteomic analyses (Figure S1B). Experiment 2 investigated the effect of the sequence of lipase treatment and high‐abundance protein depletion on protein yield and diversity. Samples were divided into the following three groups: (Group a) only high‐abundance top 14 proteins depletion (no lipase treatment), (Group b) lipase treatment followed by protein depletion, and (Group c) protein depletion followed by lipase treatment. This approach aimed to determine the optimal treatment order for lipases to maximize the yield and specificity of membrane protein extraction.

After establishing the optimal lipase treatment conditions, experiments were conducted to assess the effect of treatment sequence on protein identification (Figure 1C). Comparative analysis demonstrated that the total number of identified proteins was markedly higher in both lipase‐treated groups (Groups b and c) than in the untreated control group (Group a), regardless of treatment order. Consistently, the Venn diagram comparing the three conditions revealed a greater number of proteins uniquely identified in Groups b and c than in Group a (Figure S2D). These findings indicate that lipase treatment improves the detection of low‐abundance proteins, while the treatment sequence may differentially influence membrane protein enrichment. In particular, the order of lipase treatment affected the detection of membrane‐associated proteins. The proportion of membrane‐associated proteins was 12% in the untreated control group, 14% when lipase treatment was performed before high‐abundance protein depletion, and 20% when lipase treatment was performed after depletion. These results suggest that applying lipase treatment after the removal of high‐abundance proteins enhances the detection efficiency of membrane‐associated proteins (Figure 1D). Here, we optimized lipase treatment conditions to improve the detection of insoluble membrane‐associated proteins in plasma proteomic analysis. Integration of this lipase‐based pretreatment strategy with next‐generation high‐resolution mass spectrometry platforms, such as Orbitrap Astral, may further expand deep plasma proteome coverage and facilitate the discovery of clinically relevant biomarkers across diverse disease contexts [19, 20].

An insoluble membrane protein assay for plasma proteomics studies was optimized to limit lipase (3 U/μg) after depletion of high‐abundance top 14 proteins to a 1 h reaction at 37°C. We applied this method in a biomarker discovery study to diagnose patients with early BCa (Figure 2A). To evaluate the effects of lipase treatment on protein detectability, plasma samples from benign disease donors (n = 6) and patients with BCa (n = 6) were subjected to top 14 protein depletion. Both samples treated with lipase were subjected to trypsin digestion following sample preparation using SP3, followed by DIA‐based LC–MS/MS proteomic analysis and label‐free quantification. This study was approved by the Institutional Review Board of Kangbuk Samsung Hospital, Sungkyunkwan University of Korea on February 13, 2024 (Approval Numbers: KBSMC 2024‐01‐003‐001 and SKKU 2024‐04‐047).

FIGURE 2.

Comparison of plasma proteomic profiles using lipase treatment. (A) Plasma samples from benign disease donors (n = 6) and patients with breast cancer (BCa) (n = 6) were analyzed using lipase treatment. (B) Venn diagram showing common and unique proteins identified under each condition. (C) Volcano plot of sDEPs between lipase‐treated BCa and control plasma.

A total of 1070 proteins were identified in the benign disease donor group and 1077 in the BCa group following lipase treatment. Among these, 997 proteins were commonly detected in both groups, whereas 73 and 80 proteins were uniquely detected in the benign disease donor and patient groups, respectively (Figure 2B). Through a comparative proteomic analysis between plasma of benign disease donors and patients with BCa treated with lipase, we identified four proteins that were significantly upregulated and 190 proteins that were significantly downregulated in the BCa group relative to those of the benign disease donor group (Figure 2C).

GO annotation analysis of proteins that were significantly decreased in the BCa group or uniquely detected in benign disease donor samples revealed their enrichment in structural and epithelial‐related terms. These proteins were primarily associated with GO biological processes (GOBPs), such as intermediate filament organization, keratinization, and epidermal development. GO cellular component (GOCC) analysis indicated a strong association with extracellular exosomes, cytosol, and keratin filaments, whereas GO molecular function (GOMF) terms included structural constituents of the skin epidermis and cytoskeleton, as well as calcium and cadherin binding. Furthermore, KEGG pathway analysis revealed the involvement of cytoskeleton organization and estrogen signaling (Figure 3A). Collectively, these results suggest that proteins that are reduced or lost in cancer plasma are involved in maintaining epithelial cell integrity and intercellular structure, potentially reflecting the disruption of normal tissue architecture and extracellular vesicle composition during cancer progression.

FIGURE 3.

Characterization of plasma proteome after lipase treatment conditions. (A) GO and KEGG analysis of 73 proteins detected in the control group and 190 downregulated sDEPs. (B) GO and KEGG analysis of 80 proteins detected only in the patient group and four upregulated sDEPs. sDEPs, significant differentially expressed proteins; GO, Gene Ontology; GOBP, GO biological process; GOCC, GO cellular components; GOMF, GO molecular function; KEGG, Kyoto Encyclopedia of Genes and Genomes.

GO annotation was performed for proteins that were either significantly enriched or uniquely identified in lipase‐treated plasma from patients with BCa (Figure 3B) The enriched GOBP processes were predominantly associated with cellular differentiation, tissue morphogenesis, and structural development, indicating a loss of key regulatory proteins involved in organogenesis, regeneration, and homeostatic maintenance in the cancer group. GOMF analysis revealed significant associations among cadherin‐mediated cell adhesion, calcium and actin filament binding, and structural roles in the cytoskeleton. These findings highlight the potential reduction in proteins essential for maintaining cellular integrity and signaling networks in the plasma of patients with cancer. Regarding GOCC classification, the downregulated proteins were distributed across both intracellular compartments, including the cytosol, cytoplasm, and actin cytoskeleton, and extracellular structures, such as exosomes, the plasma membrane, and the extracellular space. This suggests that cancer‐associated proteomic alterations following lipase treatment may reflect changes in membrane dynamics, secretion mechanisms, and cytoskeletal organization.

To investigate the effect of lipase treatment on BCa plasma, protein–protein interaction (PPI) analysis was conducted using proteins that were significantly increased or uniquely identified in the plasma of lipase‐treated patients with BCa. Subcellular localization analysis revealed that many of these proteins were secreted from mitochondria or associated with cellular membranes (Figure S3A). Notably, 38% of the identified proteins were classified as membrane‐associated (Figure S3B). The GOCC terms primarily indicated associations with extracellular regions and membrane‐related structures that are integral to cell signaling and immune interactions (Figure S3C), indicating that lipase treatment facilitated the detection of membrane‐related proteins.

Among the proteins categorized under the membrane‐associated GOCC terms, 13 representative proteins were identified (Table 1). Specifically, membrane‐associated proteins, such as FCMR, FCAMR, LILRA1, and LEPR were localized to the mitochondrial intermembrane space; FSCN1, CLIC4, PCDH12, and NECTIN3 to the plasma membrane; GGT1 to the external side of the plasma membrane; and COL5A2, CHI3L1, TSKU, and ALPL to the apical plasma membrane. Quantitative analysis of these 13 proteins showed a statistically significant increase in the plasma expression of PCDH12 in the lipase‐treated group (Figure 4). Prior research has documented an association between PCDH12 and BCa. Specifically, PCDH12 expression has been reported to differ significantly between high‐risk and low‐risk patient groups [21]. Moreover, in triple‐negative BCa (TNBC), elevated PCDH12 expression has been associated with poorer prognosis among patients receiving radiation therapy [22].

TABLE 1.

List of membrane‐associated proteins in lipase‐treated breast cancer plasma.

Cellular component	Protein accession	Protein description	Gene names
Mitochondrial intermembrane space	O60667	Immunoglobulin mu Fc receptor	FCMR
	Q8WWV6	High affinity immunoglobulin alpha and immunoglobulin mu Fc receptor	FCAMR
	O75019	Leukocyte immunoglobulin‐like receptor subfamily A member 1	LILRA1
	P48357	Leptin receptor	LEPR
Plasma membrane	Q16658	Fascin	FSCN1
	Q9Y696	Chloride intracellular channel protein 4	CLIC4
	Q9NPG4	Protocadherin‐12	PCDH12
	Q9NQS3	Nectin‐3	NECTIN3
External side of plasma membrane	P19440	Glutathione hydrolase 1 proenzyme	GGT1
Apical plasma membrane	P05997	Collagen alpha‐2(V) chain	COL5A2
	P36222	Chitinase‐3‐like protein 1	CHI3L1
	Q8WUA8	Tsukushi	TSKU
	P05186	Alkaline phosphatase, tissue‐nonspecific isozyme	ALPL

FIGURE 4.

Plasma expression levels of 13 membrane‐associated proteins. The box plot shows the relative expression levels of the 13 membrane‐associated proteins identified using proteomic analysis. A statistically significant difference was observed for PCDH12. Data are shown as mean ± SD (n = 29), with significance assessed using one‐way ANOVA and Students t‐test (*p < 0.05).

In summary, we optimized a lipase‐based pretreatment method to improve the extraction and detection of membrane‐associated insoluble proteins from plasma, which are typically underrepresented because of solubility challenges. This strategy led to the identification of PCDH12, a protocadherin family protein, as a novel membrane biomarker candidate that is specifically enriched in the plasma of patients with BCa. This demonstrates the use of lipase‐enhanced proteomics for expanding the detectable plasma proteome and enabling the discovery of new classes of biomarkers.

Author Contributions

Conceptualization: Lee, K.H. and Lee, S. Methodology and analysis: Kang, E.J., Choe, Y., Lee, K.H. and Lee, S. Resources: Lee, K.H. Data curation: Kang, E.J. and Lee, S. Writing: Kang, E.J., Lee, K.H. and Lee, S. Funding acquisition: Lee, K.H. and Lee, S.

Funding

This work was supported by KBSMC‐SKKU Future Clinical Convergence Research Program Grant.

Supporting information

Figure S1: Schematic overview of the experimental setup. (A) Schematic overview of temperature, concentration and time. (B) Schematic overview of three conditions were tested: only top 14 proteins depletion, lipase treatment before depletion, and depletion before lipase treatment. DIA, data‐independent acquisition.

Figure S2: Venn diagrams of proteins identified during optimization of lipase treatment conditions: (A) temperature, (B) lipase concentration, (C) incubation time, and (D) the combined evaluation of the three conditions; (Group a) only high‐abundance top 14 proteins depletion (no lipase treatment), (Group b) lipase treatment followed by protein depletion, and (Group c) protein depletion followed by lipase treatment.

Figure S3: Functional analysis of proteins identified in lipase‐treated breast cancer plasma (A) PPI network of 58 significantly altered or uniquely identified proteins. (B) Subcellular localization analysis indicated that the proteins were predominantly secreted (41%), membrane‐associated (38%), or cytoplasmic (12%), with 9% assigned to other locations. (C) GO and KEGG pathway analyses of the same protein set. GO, Gene Ontology; KEG, Kyoto Encyclopedia of Genes and Genomes; PPI, protein–protein interaction.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request. Proteome data are available via ProteomeXchange with identifier PXD064099.