Proteomics Analysis of Human Breast Milk by Two-Dimensional Polyacrylamide Gel Electrophoresis (2D-PAGE) Coupled with Mass Spectrometry to Assess Breast Cancer Risk

Abstract

Breast Cancer (BC) is one of the most common cancers and one of the most common causes for cancer-related mortality. Discovery of protein biomarkers associated with cancer, is considered as important for early diagnosis and prediction of the cancer risk. Protein biomarkers could be investigated by large-scale protein investigation or proteomics, using mass spectrometry (MS)-based techniques. Our group applies MS-based proteomics to study the protein pattern in human breast milk from women with BC and controls and investigates the alterations and dysregulations of breast milk proteins in comparison pairs of BC versus control. These dysregulated proteins might be considered as potential future biomarkers of BC. Identification of potential biomarkers in breast milk may benefit young women without BC, but who could collect the milk for future assessment of BC risk. Previously we identified several dysregulated proteins in different sets of human breast milk samples from BC patients and controls using gel-based protein separation coupled with MS. Here, we performed two-dimensional polyacrylamide gel electrophoresis (2D-PAGE) coupled with nano-liquid chromatography-tandem mass spectrometry (nanoLC-MS/MS) in a small-scale study on a set of 6 human breast milk samples (3 BC samples vs. 3 controls) and we identified several dysregulated proteins which have potential roles in cancer progression and might be considered as potential BC biomarkers in the future. Proteomics data is available via ProteomeXchange with identifier PXD040188.

1. Introduction

Early detection of BC is needed, because there is a higher probability for survival if the cancer is caught sooner [1,2]. Therefore, several modern analytical techniques focus on finding a non-invasive way to diagnose and possibly predict and assess BC risk earlier than traditional methods. Protein biomarkers in bodily fluids provide a route to non-invasive detection of cancer with and without tissue specificity. Breast milk contains both fluid and exfoliated cells and provides a non-invasive sampling of the breast, including the ductal epithelia, which is the origin of 85% of breast tumors. Our groups have probed both the cellular and fluid components of milk studying both the intra- and extra cellular proteins along with cellular DNA. Using breast milk for BC biomarkers discovery could help to perform early diagnosis and standardize screening methodologies like mammography, which performs poorly with the lactating breast.

One robust analytical technique for identification of protein biomarkers in human breast milk is MS-based proteomics. MS has been used widely in proteomics related studies [3–9]. In our group, we use MS-based proteomics techniques coupled with gel-based protein separation (1D-SDS-PAGE or 2D-SDS-PAGE) methods to investigate the protein pattern of human breast milk in mothers identified with BC vs. mothers without BC as controls. Recently we were able to identify several dysregulations in protein levels in breast milk with BC vs. controls [10] and our outcomes (dysregulated proteins) were in agreement with current literature, i.e., that the identified proteins in our work have shown to play roles in cancer progression and most of them reported to be dysregulated by other groups, either in BC or in different types of cancers, for instance by working on cancer tissues, cell lines or serum samples from BC patients. This could verify our claim that breast milk is an appropriate candidate for protein biomarkers discovery in BC patients [11].

A 2D-SDS-PAGE was used for this experiment to not only fractionate and identify proteins, but this method also allows for quantification of the proteins identifed by using a software such as Progenesis Same Spots (version 4.5, 2011, TotalLab, UK) which is used to overlay the BC and its matched control to identify any spots that are dysregulated. This also allows for the identification of both upregulated and down regulated spots at the same time. Additionally, the 2D-SDS-PAGE gives greater separation than a 1D-SDS-PAGE, as a 2D gel first separates each protein by its isoelectric point, and then further separated by the proteins molecular weight, the combination of these two techniques separates the proteins very well, thus allowing for the identification of protein spots which are dysregulated [12]. This 2D analysis gives a better protein separation and identification than 1D can give itself, and was chosen as we can identify proteins which we were unable to identify in a 1D-PAGE, but also quantify the relative abundance between the two comparison groups [13].

Here, we performed 2D-SDS-PAGE coupled with nanoLC-MS/MS on 3 comparison pairs of human breast milk with BC vs. control. We identified several statistically significant dysregulated proteins in BC vs. control which had potential cancer-related roles and most of which have shown alteration in cancer previously (some of them have shown alteration in our previous works on different set of human breast milk samples). Our data could suggest that the dysregulated proteins have the potential to be considered as future biomarkers for BC and could be beneficial in early detection and risk prediction of BC in the future.

2. Materials and methods

2.1. Reagents

Chemical reagents were purchased from Sigma-Aldrich (St. Louis, MO), (apart from the ones that are mentioned differently).

2.2. Human breast milk samples

Human breast milk samples were collected by University of Massachusetts Amherst (UMass-Amherst following the method described elsewhere [14,15]. The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by the Institutional Review Board of Clarkson University (protocol code 12–34.1E, approved on 05/16/2012). Comparison pairs were chosen based on both the participant’s age and the age of the baby. This allows for more similar factors between each BC sample and its matched control. Donors and milk samples characteristics are described in Table 1.

Table 1:

Descriptions of donors and breast milk samples

Sample Code	BC Diagnosis Time	Age	Age at First Birth	# of Live Births	Baby’s Age Days	Family History of BC
BC_1 *	~17 months after milk donation	30	30	1	510	Yes***
C ** _1 Control	NA	30	19	3	780	Yes
BC_2	~12 months after milk donation	29	27	1	600	No
C_2 Control	NA	33	29	2	600	Yes
BC_3	39 days after milk donation	37	34	2	210	Yes***
C_3 Control	NA	35	33	2	210	Yes

^*Donor Code

^**Control

^***Reported having a familial BRCA2 mutation

2.2.1. 2D-SDS-PAGE

One Coomassie blue-stained gel for each breast milk sample (6 gels in total) and two replicates of silver-stained gels for each sample (12 gels in total) were run. The amount of proteins per gel was 750 μg for Coomassie blue-stained gels and 200 μg for silver-stained gels. The 2D-SDS-PAGE gels were run by Kendrick Labs, Inc. (Madison, WI) using Carrier ampholine method of isoelectric focusing [16,17]. The entire procedure is described in [11].

2.2.2. Computerized Comparisons

Twelve silver-stained gels (2 replicates for each sample) were scanned with a PDSI laser densitometer (Molecular Dynamics Inc, Sunnyvale, CA). Prior to scanning, a calibrated neutral density filter set (Melles Griot, Irvine, CA) was used to check the linearity of the scanner. The gel images were analyzed by Progenesis Same Spots software (version 4.5, 2011, TotalLab, UK) and Progenesis PG240 software (version 2006, TotalLab, UK). Spot percentages (integrated density of the spot above background) were measured for each of the dysregulated spots. The spot percentages were then calculated by taking the spot density above the background noise over the total density above background noise for all spots being measured. The spot percentages were used to calculate the fold change (fold increase or fold decrease) for each comparison group of BC vs. control and the average of all groups for BC vs. control.

2.2.3. In-Gel Trypsin Digestion of the statistically significant dysregulated spots

After computerized comparisons on silver-stained gel images, total of 546 dysregulated spots were identified and 110 spots with statistically significant dysregulation, were selected and the corresponding spot from the Coomassie blue-stained gels were picked and underwent an in-gel trypsin digestion and sample clean-up Using C18 Ziptip^™(Millipore, Billerica, MA).

2.2.4. Nano LC-MS/MS and data processing

The extracted, zip-tipped peptide mixtures were analyzed by nanoliquid chromatography tandem mass spectrometry (nanoLC-MS/MS) as described previously [18], using a nanoAcquity UPLC coupled with a QTOF Xevo G2 mass spectrometer (Waters, Milford, MA).

The resulting MS raw data files were converted to pkl (peak list) files by ProteinLynx Global Server (PLGS version 2.4, Waters) as described previously [19] The resulted pkl files from PLGS were run in Mascot using in-house Mascot server (www.matrixscience.com, Matrix Science, London, UK, version 2.5.1). The following parameters were used: database of NCBInr 20150706 (69,146,588 sequences; 24,782,014,966 residues) (NCBI: National Center for Biotechnology Information), Homo sapiens (human) (312,165 sequences) as taxonomy, propionamide (cysteine) as fixed modification, acetylation (lysine), oxidation (methionine), phosphorylation (serine, threonine and tyrosine) as variable modifications, ±1.3 Da as peptide mass tolerance (one ¹³C isotope), ±0.8 Da fragment mass tolerance and maximum of 3 missed cleavages. The identified proteins with Mascot score lower than 50 were verified by de novo sequencing on MS/MS raw data files.

3. Results and Discussion

3.1. 2D-PAGE

For comparative analysis of milk samples by 2D-PAGE, for each breast milk sample, one Coomassie blue-stained gel (6 gels in total) and two replicates of silver stained gels (12 gels in total) were run. The gel images of Coomassie blue-stained gels are shown in Figure 1 (Figure 1A: BC_1 vs. C_1, Figure 1B: BC_2 vs. C_2 and Figure 1C: BC_3 vs. C_3) and the gel images of silver-stained gels (1 of 2 replicate gels for each sample) are shown in Figure 2 (Figure 2A: BC_1 vs. C_1, Figure 2B: BC_2 vs. C_2 and Figure 2C: BC_3 vs. C_3).

Figure 1:

Gel images of Coomassie blue-stained gels (Figure 1A: BC_1 vs. C_1, Figure 1B: BC_2 vs. C_2 and Figure 1C: BC_3 vs. C_3).

Figure 2:

Gel images of silver-stained gels (1 of 2 replicate gels for each sample) Figure 2A: BC_1 vs. C_1, Figure 2B: BC_2 vs. C_2 and Figure 2C: BC_3 vs. C_3).

For the comparative analysis, we compared the individual comparison pairs (BC_1 vs. C_1, BC_2 vs. C_2 and BC_3 vs. C_3), and the comparison for the average of BC samples vs. average of controls). After computerized analysis of the gel images, a total of 546 spots were found to be dysregulated in BC vs. control. Among these spots, we picked 110 that had statistically significant dysregulation between BC and control samples, meaning that we picked the upregulated spots in BC vs. control (shown in blue in Figure 3) with a fold increase of ≥ 10.0 or a fold increase of ≥ 3.0 and a p value ≤0.05 and downregulated spots in BC vs. control (shown in red in Figure 3) with a fold decrease of ≤ −10.0 or a fold decrease of ≤ −3.0 and a p value ≤ 0.05. The gel difference images and spot numbering for these comparisons are shown in Figure 3 (Figure 3A: total of the 546 dysregulated spots, Figure 3B, 3C and 3D: Statistically significant dysregulated spots in BC_1 vs. C_1, BC_2 vs. C_2 and BC_3 vs. C_3 respectively, Figure 3E: statistically significant dysregulated spots in averaged BC vs. averaged control). All other spot information are shown in Supplemental Table 1 (for individual comparison pairs) and Supplemental Table 2 for averaged comparisons. The 110 statistically significant dysregulated spots were picked from Coomassie blue-stained gels and an in-gel trypsin digestion was performed followed by nanoLC-MS/MS analysis, data conversion using PLGS software, and data base search using in-house Mascot server. Most of the analyzed spots resulted in protein identification and several dysregulated proteins were identified by Mascot database search, some of which were identified to be dysregulated in other set of human breast milk samples in our previous works as well [10]. For the proteins with Mascot score lower than 50, de novo sequencing was performed on MS raw data files in order to verify the Mascot results. The MS and MS/MS data and comparison with Mascot results for theses spots (Mascot score < 50) are shown in the Supplemental Figures. Dysregulated proteins identified in our three comparison pairs (BC_1 vs. C_1, BC_2 vs. C_2 and BC_3 vs. C_3, as well as the average of BC vs. average of control are shown in Tables 2 to 5, respectively. Apart from these dysregulated proteins several antibodies or members of the immune system were found to be dysregulated in BC versus control (data not shown here). The alterations in the levels of antibodies and members of the immune system could possibly have several different reasons other than BC-related alterations from person to person. Therefore, the dysregulated antibodies and members of the immune system are not discussed here.

Figure 3:

Gel difference images and spot numbering for: Total of the 546 analyzed spots (Figure 3A), Statistically significant dysregulated spots in BC_1 vs. control_1 (Figure 3B), Statistically significant dysregulated spots in BC_2 vs. control_2 (Figure 3C), Statistically significant dysregulated spots in BC_3 vs. control_3 (Figure 3D), Statistically significant dysregulated spots in averaged BC vs. averaged control (Figure 3E). upregulated spots in BC vs. control are shown in blue and downregulated spots in BC vs. control are shown in red.

Table 2.

Dysregulated proteins in BC_1 versus control_1

Spot #	Protein Hits	NCBI gi (GenInfo) Identifier	Protein Score	Fold Change	T-test (p value)
197	perilipin-3 isoform 3	gi|255958306	202	6.5	0.005
240	Zn-alpha2-glycoprotein	gi|38026	53	4.3	0.018
269	carbonic anhydrase isozyme VI	gi|179732	114	6.5	0.021
418	ubiquitously transcribed tetratricopeptide repeat protein Y-linked transcript variant 226	gi|157829387	54	−3.5	0.005
	alpha S1-casein	gi|1359714	37
481	beta-casein isoform 1 precursor	gi|4503087	37	−4.4	0.039
531	heart fatty acid binding protein, partial	gi|458862	152	−6.0	0.019

Table 5.

Dysregulated proteins in BC_Average versus C_Average

Spot #	Protein Hits	NCBI gi (GenInfo) Identifier	Protein Score	Fold Change	T-test (p value)
347	beta-casein	gi|29674	49	4.0	0.048
47	Plakoglobin	gi|762885	39	−3.7	0.038
324	beta-casein isoform 1 precursor	gi|4503087	60	−3.7	0.035
531	heart fatty acid binding protein, partial	gi|458862	152	−3.0	0.000

The dysregulated proteins we found in our individual comparison pairs and average of BC vs. average of control are discussed here.

3.2. Proteins

3.2.1. Proteins from casein family: Downregulated in all four comparison pairs

Caseins have multiple functions based on their type, including providingamino acids required for growth. Several different isoforms of casein were found to be downregulated in all of our comparison groups. The different isoforms could be due to alternative splicing, post translational modifications proteolytic modifications and genetic variations [20]. The only exception was one isoform of beta casein that was upregulated in BC_2 vs. C_2, BC_3 vs. C_3 and BC_Average vs. C_Average. In our previous studies on human breast milk samples, several proteins and isoforms of casein family were downregulated in multiple BC samples versus controls [10]. In one study on human BC tumor tissues, as well as other different types of tumors, downregulation of casein has been reported in cancer [21].

3.2.2. Perilipin-3: Upregulated in BC_1 vs. C_1

Perilipin, a protein that plays a role in metabolisms of lipids, was shown to be involved in cancer development previously [22]. Based on Human Protein Atlas, this protein has shown high levels in BC patients [23].

3.2.3. Zn-alpha2-glycoprotein: Upregulated in BC_1 vs. C_1

This protein is involved in degradation of lipids and when protein levels are high it can cause cachexia and lower levels of body fat. In our previous 2D-PAGE study on a different pair of human breast milk, upregulation of this protein was also observed in BC sample vs. control [24]. High levels of Zn-alpha2-glycoprotein have been reported in different types of cancers [25] and particularly upregulation in BC has been reported [26–28].

3.2.4. Proteins from the carbonic anhydrase family: (Carbonic anhydrase isozyme VI: Upregulated in BC_1 vs. C_1, Carbonic anhydrase 6: Downregulated in BC_2 vs. C_2)

Isozymes (isoenzymes) are a group of enzymes that play the same role, in terms of the reaction they catalyze, but are different in structure. Carbonic anhydrase is an enzyme for formation of hydrogen carbonate from carbon dioxide and water and which regulates the pH of the blood. Different enzymes from carbonic anhydrase family have shown to be involved in tumor development in BC and is upregulated in the hypoxia condition in BC [29,30].

3.2.5. Ubiquitously transcribed tetratricopeptide repeat protein Y-linked transcript: Downregulated in BC_1 vs. C_1

This protein has a possible role in protein-protein interactions. Based on Human Protein Atlas, this protein is either in low levels or even not presented in BC tissues [31].

3.2.6. Proteins from the fatty acid binding protein family: Downregulated in BC_1 vs. C_1, BC_3 vs. C_3 and BC_Average versus C_Average

Proteins from fatty acid binding protein family play a role in fatty acids metabolism. This protein also was downregulated in our previous 2D-PAGE study on a different pair of human breast milk samples [24]. Downregulation of proteins from fatty acid binding protein family has been reported by others in BC cell lines [32], as well as in prostate cancer cell lines and tumors [33].

3.2.7. Plakoglobin: Downregulated in BC_1 vs. C_1 and BC_Average versus C_Average

Plakoglobin (also called gamma-catenin) is from catenin family of proteins which are involved in cell-cell interactions and cellular adhesion and it is also involved in tumor development [34]. Low levels of this protein are reported in invasive BC cell lines [31] and these low levels could cause more invasion in BC cells [35]. Also lower expression of the gene that encodes plakoglobin has been reported in triple-negative BC [36].

3.2.8. Alpha-amylase: Downregulated in BC_2 vs. C_2

Alpha-amylase is an enzyme that is involved in polysaccharides hydrolysis. Although in our previous study on one pair of milk samples alpha-amylase was upregulated, here we found it to be downregulated in one pair of samples. Levels of alpha-amylase have been reported to be dysregulated in different types of cancers such as lung [37] and ovarian [38] tumors. Different levels of alpha amylase in different stages of lung cancer were reported before [39], therefore it might be possible that the inconsistency of amylase in the current study and previous study could be related to the stage of BC.

3.2.9. BAI2 protein: Downregulated in BC_2 vs. C_2

BAI2 protein (brain-specific angiogenesis inhibitor 2) is from the family of adhesion-G protein-coupled receptors which are involved in cell-cell interactions. The proteins from BAI family have an inhibitory effect on tumor development. [40].

3.2.10. Proteins from the ATP-binding cassette, sub-family A: Downregulated in BC_2 vs. C_2

These proteins play a role in transportations through the cell membrane. The levels are altered based on subtype of BC and stage of the disease and are higher in advanced stages of BC [41].

3.2.11. Protein tyrosine phosphatase, non-receptor type 20A: Downregulated in BC_2 vs. C_2

This protein is involved in cellular processes. Loss of the gene that encodes this protein has been observed in BC [42].

3.2.12. Proteins from the heat shock protein family: Downregulated in BC_2 vs. C_2 and BC_3 vs. C_3

Heat shock proteins play different roles including their role in the formation of other proteins structures (such as protein folding) they also have a potential role in metastasis [43]. In our previous study, heat shock protein showed to be downregulated in one of our comparison pairs [10]. We found two heat shock proteins, dnaJ homolog subfamily C member 12 and Endoplasmin precursor also downregulated in BC.

3.2.13. Human Protein Disulfide Isomerase, NMR, 40 Structures: Downregulated in BC_3 vs. C_3

Human Protein Disulfide Isomerase is an enzyme involved in protein folding through formation or breakage of disulfide bridges and is involved in cancer development [44].

3.2.14. Glutamate carboxypeptidase: Downregulated in BC_3 vs. C_3

Glutamate carboxypeptidase is an enzyme for catalysis of c-terminal glutamate removal from specific peptides. The gene that encodes this protein has shown to be dysregulated in breast and prostate cancers [45].

3.2.15. SHROOM3 protein: Downregulated in BC_3 vs. C_3

This protein plays a role in cellular processes, such as maintaining cell shape. Medium to low levels of this protein are reported in BC patients based on Human Protein Atlas [46].

3.2.16. Plasminogen: Downregulated in BC_3 vs. C_3

Plasminogen is involved in cellular processes and angiogenesis (inhibitory effect). This protein has shown to be involved in BC invasion [23,47].

3.2.17. Proteins from the enolase family: Downregulated in BC_3 vs. C_3

Enolase is an enzyme that plays a role in glycolysis. Enolase showed to be dysregulated in BC, but the reported data showed upregulation of this enzyme in BC [48] [49] while in our study it is downregulated in one pair of samples.

3.2.18. Proteins from actin family: Downregulated in BC_3 vs. C_3

Actins have different functions including roles in protein-protein interactions and cellular processes. Although here, actin is downregulated in one pair of our samples, in our previous study, it was upregulated in one of our comparison pairs. It has been reported that actins could be involved in tumor development [50].

3.2.19. AKR1CL2 protein: Downregulated in BC_3 vs. C_3

This protein is an enzyme for catalysis of fructose reduction. Low levels and loss of this protein is reported in BC based on the Human Protein Atlas [51].

3.2.20. Proteins from the lysozyme family: Downregulated in BC_3 vs. C_3

Different proteins from lysozyme family were downregulated in BC_3. Lysozyme acts as an antibacterial and hydrolyzes the glycosidic bonds in the bacterial cell wall. In the current study, different proteins from lysozyme family were downregulated in BC_3. Also in our previous 2D-PAGE study on one pair of breast milk samples, different proteins from lysozyme family were downregulated. Low levels of lysozyme were reported for early stages of BC based on a study done on BC tumor tissues [52].

Table 6 give a summary of the dysregulated proteins.

Table 6:

Summary of dysregulated proteins and their isoforms identified in both current and prior studies

Protein Identified	Dysregulation pattern in this study	Protein comparison group	Dysregulation pattern in previous study	Reference from previous study
Casein family proteins	under expressed	BC_1 vs. C_1, BC_2 vs. C_2, BC_3 vs. C_3, BC_Average vs. C_Average	under expressed	[10]
zinc-alpha2-glycoprotein	overexpressed	BC_1 vs. C_1	overexpressed	[24]
fatty acid binding family proteins	under expressed	BC_1 vs. C_1, BC_3 vs. C_3, BC_Average vs. C_Average	under expressed	[24]
alpha-amylase	under expressed	BC_2 vs. C_2	overexpressed	[11]
heat shock protein family	under expressed	BC_2 vs. C_2, BC_3 vs, C_3	under expressed	[10]
actin family	under expressed	BC_3 vs. C_3	overexpressed	[10]
lysozyme	under expressed	BC_3 vs. C_3	under expressed	[24]

3.3. GSEA Analysis

Gene set enrichment analysis (GSEA) is a method of determining data based on gene sets that share a common biological function, chromosomal location or regulation [53]. It uses predefined gene sets based on prior knowledge. The goal is to determine whether a priori defined set of genes is randomly distributed throughout a list or primarily at the top or bottom [53]. The GSEA software is using multiple statistical methods to determine whether the gene is overexpressed or under expressed based on the position in the list. The two important factors for our experiments are the normalized enrichment score (NES) and the false-discovery rate (FDR). NES is the measurement which reflects the degree to which a gene is overrepresented at the top or bottom of a ranked list of genes. Normalizing this statistic allows for the software to account for the differences in gene set sizes. The FDR corresponds to the probability that a gene set with a given NES has a false positive finding [53]. We consider a pathway to be statistically significant if it follows two parameters; a NES greater than 1.00, but the higher the score, the better and the FDR is less than 25% (0.25). A FDR of <25% indicates that there is less than a 25% chance that the pathway identified could be a false positive. The molecular signature database (MSigDB) is a database that provides known gene sets for the GSEA analysis. “Hallmark” gene sets are defined as being a refined gene set without redundancy across and heterogeneity within gene sets [54]. This analysis uses MSigDB of all the identified hallmark gene sets. GSEA allows genes to be identified as part of one of the predefined ‘hallmark” pathways and which genes are responsible for these pathways. We understand that many of these changed pathways identified did not achieve statistical significance. This is due to the limitation of the software, which uses a NCBI gene number for an intact protein, but also for isoproteins/proteoforms, and unless one of these proteins is not greatly dysregulated, then the statistical significance is not achieved. However, we believe that the outcome of GSEA analysis has value (not as a definitive statement, but as an indication that a pathway is possibly dysregulated), as it may still identify pathways whose roles in cancer development could eventually be considered significant with a larger study.

3.3.1. BC_1 versus Control_1

There were 3 differentially expressed upregulated protein and 4 downregulated proteins in comparison to the control in the first comparison group. PLIN3, AZGP1, and CA6 are all upregulated and UTY, CSN1S1, CSN2 and FABP3 are downregulated. For all of the following sets GSEA analysis was performed for this comparison group using the hallmark (H) gene sets collection in MSigDB. The Hallmark analysis revealed a total of no-significant upregulated pathways and one significant downregulated pathway. Fatty acid metabolism and the androgen response were upregulated. The downregulated pathway which is shown to be significant is myogenesis. Table 7 shows the identified pathways for BC_1 v C_1.

Table 7:

Pathways Identified for BC_1 versus Control_1

	Pathways	NES	FDR q-val
Upregulated	Fatty Acid Metabolism	1.00	1.000
	Androgen Response	0.81	0.780
Downregulated	Myogenesis*	−1.34	0.155

^*denotes statistically significant pathway according to GSEA analysis with a FDR < 25% (0.25)

NES-Normalized Enrichment Score

FDR q-val-false discovery rate q-value, where a score of <25% is considered significant.

3.3.2. BC_2 versus Control_2

There were 10 proteins for which levels changed these include CSN2, JUP, AMY1A, CA6, CSN1S1, ADGRB2, ABCA9, DNAJC12, ABCA10 and PTPN20 (Table 8). All proteins are downregulated, except CSN2, which was found to be upregulated and downregulated in different spots. Analysis showed 4 total pathways identified, apical junction, KRAS Signaling Up, xenobiotic metabolism and fatty acid metabolism.

Table 8:

Pathways Identified for BC_2 versus Control_2

	Pathways	NES	FDR q-val
Downregulated	Apical Junction	−1.02	1.000
	KRAS Signaling Up	−1.01	1.000
	Xenobiotic Metabolism	−1.01	0.850
	Fatty Acid Metabolism	−0.76	0.821

NES-Normalized Enrichment Score

FDR q-val-false discovery rate q-value, where a score of <25% is considered significant.

3.3.3. BC_3 versus Control_3

A total of 18 proteins were identified to be differentially expressed with the majority being downregulated. CSN2 was found to be both upregulated and downregulated, and HSP90B1, HSP90B2, HSP90B3P, PDI, CNDP2, PLIN3, PLG, ENO1, ENO2, ACTG1, ACTA1, POTEE, POTEKP, ACTBL2, AKR1E2, LYZ AND FABP1 are all shown to be downregulated (Table 9). Analysis of these proteins identified six upregulated pathways which are apical junction, cholesterol homeostasis, myogenesis, hedgehog signaling, complement and coagulation pathways. Ten downregulated pathways were identified, unfolded protein response, PI3K/AKT/MTOR signaling, fatty acid metabolism, xenobiotic metabolism, hypoxia, MTORC1 signaling, glycolysis, apoptosis, UV response Up, and the epithelial to mesenchymal transition (EMT). The only significant pathway identified is the upregulated apical junction.

Table 9:

Pathways Identified for BC_3 versus Control_3

	Pathways	NES	FDR q-val
Upregulated	Apical Junction*	1.58	0.046
	Cholesterol Homeostasis	1.25	0.416
	Myogenesis	1.15	0.585
	Hedgehog Signaling	0.78	1.000
	Complement	0.78	0.964
	Coagulation	0.77	0.818
Downregulated	Unfolded Protein Response	−1.00	1.000
	PI3K/AKT/MTOR Signaling	−0.99	1.000
	Fatty Acid Metabolism	−0.92	1.000
	Xenobiotic Metabolism	−0.88	1.000
	Hypoxia	−0.83	1.000
	MTORC1 Signaling	−0.81	1.000
	Glycolysis	−0.80	1.000
	Apoptosis	−0.69	1.000
	UV Response Up	−0.69	1.000
	Epithelial to mesenchymal transition(EMT)	−0.68	0.966

^*denotes statistically significant pathway according to GSEA analysis with a FDR < 25% (0.25)

NES-Normalized Enrichment Score

FDR q-val-false discovery rate q-value, where a score of <25% is considered significant.

3.3.4. BC_Average versus Control_Average

Three proteins were identified to be dysregulated from this comparison group, CSN2, which similarly to the other comparison groups was also found to be both upregulated and downregulated, JUP and FABP3 were both downregulated. The hallmark (H) gene set collection in MSigDB revealed no significant pathways affected, but did identify myogenesis as being upregulated and KRAS signaling up, apical junction and xenobiotic metabolism as being downregulated (Table 10).

Table 10:

Pathways Identified for BC_Average versus Control_Average

	Pathways	NES	FDR q-val
Upregulated	Myogenesis	1.00	1.00
Downregulated	KRAS Signaling Up	−0.66	1.00
	Apical Junction	−0.65	1.00
	Xenobiotic Metabolism	−0.65	1.00

NES-Normalized Enrichment Score

FDR q-val-false discovery rate q-value, where a score of <25% is considered significant.

3.3.5. Significant pathways identified

Dysregulation in the Myogenesis Pathway

Myogenesis is the process of generating muscle. In adults, myogenesis relies on the turnover of terminally differentiated cells to maintain tissue homeostasis. When there is muscle damage, satellite cells will differentiate and repair tissue to maintain homeostasis [55]. When muscles are injured, stem cells are activated, express the transcription factors Pax7 and MyoD which allow them to repair the muscle injury [56]. In cancer, tumors have been shown to impact the number of regenerating muscle fibers. Muscle dysfunction in breast cancer patients generally presents as severe muscle fatigue [57]. We identified that heart fatty acid binding protein (FABP3) as the only protein found to be involved with myogenesis. FABP3 is essential in skeletal muscle to move fatty acids towards the mitochondria for energy expenditure [58]. In breast cancer models, reduced ATP levels and increased activity of AMP-activated protein kinase (AMPK) imply that the energy production is decreased in skeletal muscles [56]. FABP3 has been found in many cancers such as non-small cell lung cancer (NSCLC) [59], gastric cancer [60], leiomyosarcoma and melanoma [61]. In breast cancer and embryonic cancer cells FABP3 acts as a tumor suppressor. Inactivation will promote proliferation and inhibit apoptosis [46,59]. Decreased expression of FABP3 will lead to decreased muscle generation, increased muscle fatigue and promote proliferation of cancer cells. Heart fatty acid binding protein (FABP3) was found downregulated in BC_average v control_average comparison group, alpha-skeletal muscle actin (ACTA1) was found to be downregulated in BC_3 v control_3, but neither were significant according to the GSEA algorithm.

Dysregulations in Apical Junction Complex

The main function of the apical junction (AJ) is to initiate and stabilize cell-to-cell adhesion and to modulate actin dynamics. In normal epithelial cells, the upper portion of the lateral membrane contains the cell adhesion apical junction complex (AJC) [62]. In cancer, the AJC is involved in regulating gene transcription, cell proliferation, differentiation and cell polarity and has a role in preventing an invasive phenotype in cancer [62,63]. The AJC was shown to be upregulated in the BC_3 v control_3 comparison group, with the two proteins shown to be associated are ACTG1 and ACTA1. The AJC was also shown to be dysregulated in other comparison groups (downregulated in BC_2 v control_2 and BC_average v control_average), but was not considered significant according to the GSEA algorithm. Actins are a main component of cell and are essential for many cellular processes such as motility and cell division [64]. ACTA1 has been shown to be dysregulated in many cancers, higher expression is thought to be correlated with increased metastatic potential, and shorter survival times [51,65]. Lower expression is associated with a more aggressive phenotype. Downregulated ACTA1 has been associated with tumorigenesis in head and neck squamous cell carcinomas. In colorectal, prostate and pancreatic cancers, the downregulation is associated with aggressive carcinogenesis [51]. Downregulated ACTA1 also encodes a protein that is involved in cell motility, structure and integrity [66]. ACTG1 is encodes proteins that aid in the internal motility [51]. ACTG1 has been associated with centrosome integrity and regulation of mitotic progression and disrupting the cytoskeletal fiber networks [63,67]. In patients with ovarian cancers, reduced levels of ACTG1 have been shown to be negatively correlated with the clinical stage of cancer [68]. Other proteins that are involved with the insignificant comparison groups are plakoglobin (JUP), which plays a role in cell-to-cell adhesion [69].

Other Pathways Identified

There were many other pathways identified from our protein list, but are considered not significant for our constraints of a FDR <25% and also having a NES of 1.00 or higher. Three proteins were identified as being part of the fatty acid metabolism pathway, carbonic anhydrase (CA6), neurone-specific enolase (ENO2) and fatty acid binding protein (FABP1). The fatty acid metabolism pathway is involved in energy, growth and membrane synthesis [70]. CA6 was identified as being upregulated in one comparison group (BC_1 v control_1) and downregulated in another (BC_2 v control_2), and aids in the conversion of carbon dioxide to carbonic acid and bicarbonate [71]. In metastatic solid tumors, continuous division can lead to an acidic microenvironment that is also low in oxygen levels [71]. ENO2 was downregulated in BC_3 v control_3. ENO2 plays an important role in glycolysis and has been used to identify neuroendocrine differences in breast cancer [72,73]. FABP1 was downregulated in BC_3 v control_3, and it functions as a regulator of lipid metabolism and cellular signaling [74].

The androgen response pathway can play roles in breast cancer as androgen receptors are expressed in about 60–70% of BC. There is speculation on whether the androgen receptor acts as an oncogene or tumor suppressor, as the receptor is involved in multiple cellular pathways [75]. Zinc-alpha2-glycoprotein (AZGP1) is upregulated in BC_1 v control_1 and is involved in many signaling pathways that are involved in both apoptosis and proliferation. AZGP1 also promotes lipid metabolism and glucose utilization. Upregulation in BC tumors could cause proliferation and cancer related cachexia [76,77].

The KRAS pathway is shown to regulate cell growth, division and differentiation. When activated, this pathway can cause downstream signaling cascades such as the RAL, RAF-MEK-ERK and PI3K-AKT pathways [78]. Many of these pathways are involved in the progression of cancer, and when dysregulated can cause cancer cells to metastasize. Plakoglobin (JUP), is involved in cell to cell adhesion [69] and is shown to be downregulated in two comparison groups (BC_2 v control_2 and BC_average v control_average).

Xenobiotic metabolism is a chemical transformation via enzymes of relatively lipophilic compounds into compounds that are easier to secrete [79]. Four proteins were identified as being involved in this pathway by GSEA analysis, but no significance. Plakoglobin (JUP), plasminogen (PLG), fatty-acid binding protein (FABP1) and glutamate carboxypeptidase (CNDP2). JUP was shown to be downregulated in two comparison groups (BC_2 v control_2 and BC-average v control_average). FABP1 was only downregulated in one comparison group of BC_3 v control_3. PLG was downregulated in BC_3 v control_3. CNDP2 is a catalyst of hydrolysis of N-acetyl-L-aspartyl-L-glutamate (NAAG) to N-acetyl-L-aspartate and L-glutamate [80]. This enzyme is found to be downregulated in BC_3 v control_3.

Cholesterol homeostasis is an essential pathway in all cells to maintain proper function and homeostasis [81]. ACTG1 is shown to be downregulated in BC_3 v control_3, and is strongly involved in cell division [82].

A protein involved in 3 different signaling pathways-the hedgehog pathway, the complement pathway and coagulation pathway-is PLG, where it is downregulated in BC_3 v control_3 for all three pathways it is part of. PLG aids in removing blood clots and fibrin deposits [83]. Hedgehog pathway is necessary for tissue maintenance and regeneration [84]. The complement pathway is part of the immune system and aids in clearing damaged cells, and activating the immune system in case of an infection or injury [85]. The main function of the coagulation pathway is to keep hemostasis which allows for healing and prevention of excessive bleeding [86].

Heat shock proteins (HSP) are produced in response to cellular stress and work to keep the cells integrity by keeping critical signaling pathways functioning [87]. HSP90B1 is a protein shown to be dysregulated in two pathways (BC_3 v control_3); the unfolded protein response pathway and the PI3K/AKT/MTOR signaling pathway. The unfolded protein response pathway maintains homeostasis of cells by initiating protein folding molecules [88]. The PI3K/AKT/MTOR signaling pathway can lead to cell growth and tumor proliferation in cancers [89].

Hypoxia is limited oxygen supply, and can lead to disease progression and recurrence. Most hypoxic cancer tumors are generally more aggressive, increased risk of metastasis and drug resistance [90]. Glycolysis is a metabolic process that takes glucose and converts it to energy [91]. Alpha enolase (ENO1) and neurone-specific enolase (ENO2) are downregulated in the third comparison group for both the hypoxic pathway and glycolysis pathway. ENO1 is a glycolytic enzyme, and when knocked down has effects on cell proliferation [92]. ENO2 is also involved in glycolysis and has been used to identify different tumors that are releasing hormones [72,73]. ENO2 is also the main protein downregulated in three other pathways within the same comparison group; apoptosis, UV-response pathway and the epithelial to mesenchymal transition (EMT) pathway. UV signaling will activate the AKT/MTOR signaling pathway which can increase proliferation and cell growth [93]. ENO2 downregulation could affect the proliferative rates of cancer cells. The EMT pathway when downregulated can induce changes within tumor cells and their microenvironment which will promote an invasive and migratory phenotype [94].

The mTORC1 signaling pathway regulates cell growth by taking signals from growth factors, energy supply and nutrients available to promote cell growth or catabolism when needed [95]. mTOR is activated in tumor cells which encourages them to grow and metastasize. The two proteins found to be involved with this pathway are; alpha-enolase (ENO1) and heat shock protein (HSP90B1). HSP90B1 works to keep the cells integrity and respond to stress [87]. Both proteins are found to be downregulated in the BC_3 v control_3 comparison group. Downregulation in HSP90B1 could be due to cellular stress and downregulation in ENO1 might allow proliferation in cancer cells. In cancer cells, abnormal activity of the mTOR pathway is common because tumors are requiring large amount of nutrients for growth and division [95].

4. Concluding Remarks

Here we analyzed 6 human breast milk samples including 3 comparison pairs (BC vs. control) to identify statistically significant protein dysregulations in human breast milk which could be potentially related to BC development. 2D-PAGE coupled with nanoLC-MS/MS technique was applied to each of the breast milk samples for our analysis. We identified several dysregulated proteins with cancer-related roles and functions and potential involvement in cancer progression. Some of these proteins were reported to be dysregulated by other researchers, working on cancer cell lines or tumor tissues. Examples of downregulated proteins in BC vs. control were proteins from casein family, proteins from heat shock protein family, proteins from enolase family, proteins from actin family, proteins from lysozyme family. Examples of upregulated proteins in BC vs. control were perilipin, Zn-alpha2-glycoprotein and carbonic anhydrase isozyme VI. Current cancer related literature for the identified proteins helps to verify our outcomes and validates the fact that breast milk could be used as an appropriate BC microenvironment for investigation of BC biomarkers. Dysregulated proteins have the potential to be considered as candidates for BC biomarkers discovery for future methods of early diagnosis and risk assessment of BC. This method could be beneficial for young women who have the possibility of giving birth by collecting the breast milk at the time of lactation and analysis of the milk for BC risk assessment.

4.1. Limitations

The sample group for this study consists of 6 human breast milk samples and is a small number for comparative analysis, therefore the dysregulated proteins still have to be verified in larger groups of samples. Also, the time between sample donation and cancer diagnosis is different between our comparison pairs, which could cause inconsistency. Although there are limitations, we still found several dysregulated proteins, some of which were found to be dysregulated in our previous studies on a different set of milk samples. Also, the identified proteins have potential roles and involvement in cancer progression.

Supplementary Material

supinfo2

Supplemental Figure: Validation of Mascot results by de novo sequencing for dysregulated proteins with Mascot score lower than 50

Table 3.

Dysregulated proteins in BC_2 versus control_2

Spot #	Protein Hits	NCBI gi (GenInfo) Identifier	Protein Score	Fold Change	T-test (p value)
411	beta-casein	gi|29674	36	3.0	0.003
441	beta-casein	gi|29674	38	3.8	0.027
47	Plakoglobin	gi|762885	39	−7.0	0.024
125	alpha-amylase	gi|178585	48	−3.9	0.027
271	carbonic anhydrase 6 isoform 1 precursor	gi|70167127	183	−5.6	0.001
	carbonic anhydrase 6 isoform 2 precursor	gi|395132469	183
320	beta-casein isoform 1 precursor	gi|4503087	56	−10.0	0.015
324	beta-casein isoform 1 precursor	gi|4503087	60	−5.5	0.004
343	beta-casein isoform 1 precursor	gi|4503087	50	−4.5	0.014
348	alpha S1-casein	gi|1359714	75	−3.3	0.033
364	BAI2 protein	gi|223460380	36	−5.5	0.007
368	alpha S1-casein	gi|1359714	49	−5.1	0.009
412	beta-casein	gi|29674	39	−5.3	0.042
430	beta-casein	gi|29674	36	−51.9	0.018
471	ATP-binding cassette, sub-family A (ABC1), member 9, isoform CRA_c	gi|119609479	51	−17.8	0.011
	DnaJ homolog subfamily C member 12 isoform a	gi|11141871	51
	ATP-binding cassette sub-family A member 10	gi|32350914	51
	beta-casein isoform 1 precursor	gi|4503087	48
478	protein tyrosine phosphatase, non-receptor type 20A	gi|55959983	54	−3.9	0.010

Table 4.

Dysregulated proteins in BC_3 versus control_3

Spot #	Protein Hits	NCBI gi (GenInfo) Identifier	Protein Score	Fold Change	T-test (p value)
340	beta-casein	gi|29674	39	5.9	0.000
355	beta-casein isoform 1 precursor	gi|4503087	57	8.4	0.003
15	heat shock protein gp96 precursor	gi|15010550	210	−7.4	0.002
	endoplasmin precursor	gi|4507677	210
	heat shock protein 94c	gi|61104923	51
97	Chain A, Human Protein Disulfide Isomerase, NMR, 40 Structures	gi|159162689	39	−3.4	0.045
157	glutamate carboxypeptidase	gi|15620780	39	−5.6	0.013
191	SHROOM3 protein, partial	gi|3095186	34	−3.0	0.045
192	Plasminogen	gi|38051823	42	−3.6	0.014
200	PREDICTED: alpha-enolase isoform X1	gi|578798587	105	−3.6	0.036
	neurone-specific enolase	gi|930063	94
222	actin, cytoplasmic 2	gi|4501887	207	−3.2	0.042
	gamma-actin, partial	gi|178045	188
	actin, alpha skeletal muscle	gi|4501881	116
	PREDICTED: POTE ankyrin domain family member E isoform X2	gi|767918225	107
	RecName: Full=Putative beta-actin-like protein 3; AltName: Full=Kappa-actin; AltName: Full=POTE ankyrin domain family member K; Contains: RecName: Full=Putative beta-actin-like protein 3, N-terminally processed	gi|74739412	93
	beta-actin-like protein 2	gi|63055057	83
279	AKR1CL2 protein	gi|12804019	46	−7.7	0.000
362	Chain A, Structural And Functional Analyses Of The Arg-Gly-Asp Sequence Introduced Into Human Lysozyme	gi|157831914	52	−4.3	0.050
	Chain A, Changes In Conformational Stability Of A Series Of Mutant Human Lysozymes At Constant Positions	gi|157835338	52
	Chain A, Contribution Of Hydrophobic Effect To The Conformational Stability Of Human Lysozyme	gi|157835340	52
	beta-casein	gi|29674	41
397	beta-casein isoform 1 precursor	gi|4503087	40	−11.7	0.046
451	beta-casein isoform 1 precursor	gi|4503087	51	−7.9	0.003
473	beta-casein	gi|29674	36	−5.7	0.010
535	lysozyme precursor	gi|307141	63	−3.2	0.033
	Chain A, Changes In Conformational Stability Of A Series Of Mutant Human Lysozymes At Constant Positions	gi|157835338	63
	Chain A, Contribution Of Hydrophobic Effect To The Conformational Stability Of Human Lysozyme	gi|157835340	63
	Chain A, T43v Mutant Human Lysozyme	gi|4930021	63
	Chain A, T52v Mutant Human Lysozyme	gi|5107556	63
541	fatty acid-binding protein	gi|227994	180	−4.1	0.012

Abbreviations:

2D-PAGE

two-dimensional polyacrylamide gel electrophoresis

BC

breast cancer

gi

GenInfo

MS

mass spectrometry

nanoLC-MS/MS

nano liquid chromatography-tandem mass spectrometry

NCBI

National Center for Biotechnology Information

pI

isoelectric point

pkl

peak list

PLGS

ProteinLynx Global Server

Data availability statement

The data which supports the findings are openly available in ProteomeXchange with an identifier of PXD040188.