The Network of Antigen-Antibody Reactions in Adult Women with Breast Cancer or Benign Breast Pathology or without Breast Pathology

The Immunoglobulin G (IgG) antibody response to different protein antigens of the mammary ductal carcinoma by adult women affected by Breast Cancer (BC) distinguishes at least 103 proteins that differ in their molecular weights (MW). The IgG producing cell clones (nodes) coexist with each other in each individual organism and share energy resources among themselves, as well as factors that control the level of expression and Specificity of their IgG antibodies. So, it can be proposed that among them there is a Network of interconnections (links) unveiled by the antigens, which specifically react with the IgG antibodies produced by the clones. This Network possibly regulates IgG antibodies' activity and effectiveness. We describe the Network of nodes and links that exists between the different antigens and their respective IgG producing cell clones against the extracted protein antigens from the cells of the T47D Cell-Line, in 50 women with BC, 50 women with Benign Breast Pathology (BBP) and 50 women without breast pathology (H). We have found that women with BBP have the highest number of Links, followed by the H group and, lastly, the women with BC, a finding which suggests that cancer interferes with the Connectivity between the IgG producing cell clones and blocks the expression of 322 links in women with BBP and 32 links in women with H. It is also plausible that the largest number of links in the women with BBP indicates the Network’s state of arousal that provides protection against BC. On the other hand, there were many missing links in the BC group of women; the clone which lost more links in the BC group was the hub 24, which point to some of the antigens of T47D as potentially useful as vaccines, as the immune system of women with BBP is well aware of them.


Introduction
Breast Cancer (BC) is the most common cancer in women worldwide, with nearly 1,380,000 diagnostics per year [1]. In 2008, 460,000 deaths were registered, of which 269,000 (58%) were in determines its own identity and is self-regulated by idiotype/anti-idiotype interactions [25]. According to this theory, lymphocyte clones are regulated by other lymphocyte clones within a network. In that context, even in the absence of interactions with the external environment, the Immune Network could manifest autonomous activities [26]. Following this perspective, the essence of the immune system is the regulation of lymphocytes by the interaction of their antigen-binding molecules (mainly immunoglobulins), thus implying that the immune system contributes to the integration of an internal environment or molecular ecology [27][28][29][30].
In the case of BC, the presence of antibodies against breast tumor antigens may be analyzed as a network in which the connections and the dynamics among the clones can maintain a healthy state, or coexist with a benign pathology or a cancer.
Based on the foregoing, the present article presents an analysis from the perspective of the networks, that is, their composition and geometric structure among the 121 bands (nodes) found in the 1D-II of women with Breast Cancer (BC), with Benign Breast Pathology (BBP) or without breast pathology (H).
The composition and structure of the network in each of the women's groups, as well as their similarities and differences, might help to identify and understand the mechanisms by which the pathology is triggered, whilst the more connected nodes (hubs) of each network will indicate the antigens and antibodies that are more useful in the design of therapeutic and diagnostic tools, and those that disappear in women with BC or BBP as possibly protective antigens and antibodies.

Grouping of Participants and Sample Sizes
Experimental groups were made up of 50 female patients. Breast cancer patients were recruited, all of which were over 30 years old, with histologically confirmed untreated breast cancer (BC) (94% of the infiltrating canalicular variety, 2% Lobular, 2% Mucinous, 2% Tubular), as well as 50 patients suffering from benign breast pathology (BBP) (fibroadenoma or cystic fibrosis or mastitis) during their first consultation at the Hospital General de México. We also recruited 50 women without breast pathology who were considered healthy (H), older than 22 years of age, without any signs or symptoms of breast pathology (Table 1). All participants were informed of the details of the study (scientific and technical basis, exclusive use of their blood sample for immunodiagnostic of BC, anonymity, confidentiality) and signed a letter of informed consent. The protocol was reviewed and approved by the Committee of Ethical Research at the Hospital General de México.

Blood Samples
Ten milliliters of venous blood were collected from each participant using sterile and discardable Vacutainer Kits. Blood samples were left to coagulate at room temperature, from which sera were collected after their centrifugation at 1,500 rpm and they were then aliquoted in five sub-samples of equal volume before storage at -80°C.

Cell Culture and Antigen Preparation
The human breast cancer cell line T47D was obtained from ATCC (Manassas, VA, USA), cultured in phenol-free RPMI 1640 media supplemented with 10% fetal calf serum, 100 U/ml penicillin, 100 mU/ml streptomycin and 250ng/ml amphotericine B. The cells were grown on plastic tissue culture plates (Costar, Cambridge) within atmospheric conditions of 95% humidity and 5% CO 2 at 37°C. When the cells reached the desired confluence, they were harvested by treatment with Versene (Tris, NaCl, EDTA) and pelleted by centrifugation at 1,500 rpm for 5 minutes. Cell pellets were washed three times with PBS and stored at -80°C.
As a control, the human cell-line MCF10, which was originated from a mammary gland fibrocystic disease, was obtained from ATCC (Manassas, VA, USA). MCF10A cells were grown on Petri dishes with culture medium (2/3 DMEM high glucose, 1/3 F12K) supplemented with 10% FBS, Anti-Anti (Gibco), 2mM glutamine, 1μg/mL Hydrocortisone, 100 mU/mL insulin and 10 ng/mL EGF, and incubated at 37°C. For obtaining protein extracts, once the culture dishes reached a 85-90% confluence, cells were washed with PSB and incubated with Versene for 7 minutes at 37°C. Then, cells were detached from the dish plates washing the surface of the dish plates with a 1 mL pipette tip and once all the cells were detached, the cell suspension was centrifuged at 1150 rpm for 5 minutes and the cell pellet was washed three times with PBS.
Cell pellets destined for 1D-II analysis were lysed for 30 minutes at 4°C in a denaturing lysis buffer composed of 7 M urea, 4% (w/v) CHAPS, 65 mM DTT and Halt protease inhibitor cocktail (PIERCE, 10 μL per 875 μL of lysis buffer), using 40 μl per 10 6 cells. The lysate was centrifuged at 16,000 g for 10 minutes at 4°C, the supernatant was recuperated and its protein concentration was measured using the commercial Bradford reagent assay (Bio-Rad, Hercules, CA). The different cell lysate batches were pooled, aliquoted and their protein concentrations were similarly measured. Aliquots of the pooled batches were kept at -80°C until further use.
Uni-dimensional Electrophoresis (1D-Electrophoresis) and Immunoblots of the Protein Extracts from the T47D Cell-Line 100 μg of the pooled protein extract (PE) from the T47D Cell-Line were subjected to electrophoresis using polyacrilamide gels (4-20% TGX Bio-Rad) to separate the proteins by molecular weight, at 80V for about 2 hours. After electrophoresis, the separated proteins were then transferred electrophoretically onto nitrocellulose membranes (HyBond ECL, Amersham Biosciences) using a mini Trans-Blot cell (Bio-Rad; 1h 15 min, 100 V). To confirm the transference of proteins, membranes were reversibly-stained using CPTS (copper phthalocyanine tetrasulphonic acid, ALDRICH) diluted in 12mM HCl, scanned and destained followed by blocking with 5% skimmed Nestlé's Svelty powdered milk in PBS containing 0.3% Tween 20, pH 7.4 (PBS-T) (16h; 4°C). Each membrane was marked with two pencil lines, one at its upper limit and another one at its lower limit, and then cut vertically into seventeen to eighteen 4 mm wide strips. Each strip was then probed individually with the serum of a different participant diluted 1:300 in 2 mL of fresh blocking solution (5h; room temperature). The same identical serum from a healthy woman (H111) was included in each set of 17-18 strips as a control used to match the problem sera's electrophoretic runs between the runs of the 17 different membranes. Bound antibodies were detected by incubation with HRP-conjugated second antibody (goat anti-human IgG (H+L); THERMO; 1h, room temperature) diluted 1:2500 in PBS-T. Detection of second antibody binding was performed by incubation with DAB substrate (SIGMA; 0.1 mg/ml, 0.015% peroxide in PBS-T) for 5 minutes at room temperature and the reaction was stopped by rinsing the strips five times with deionized water. Five washes with 2 ml of PBS-T each were performed between each step.

Image Analysis
The strips of the Immunoblots were left to air-dry overnight, protected from exposure to light and were prepared for scanning. An HP Scanjet G4050 scanner was used to scan all images (reversible-stained membranes and immunoblot strips) at a resolution level of 300 dpi in TIF format. The digitized images of the strips were aligned in Photoshop and then the images were analyzed with Quantity-One software (Bio-Rad) for the detection of bands and calculation of Molecular Weights (MW, kDa). All images were studied with the same settings of brightness, contrast and color to minimize bias. Banding patterns between strips from different gels and membranes were compared using the control strips (H111) in order to identify the total number of different bands and create a binary database with the presence (1) or absence (0) of each band in each strip.
In order to compare immunoblots obtained with T47D and MCF10 extracts, both extracts were loaded onto 10-well polyacrylamide gels. One gel was stained with Coomassie Blue and the other gel was transferred onto nitrocellulose membranes. Each lane was incubated individually with sera. Three sera were used: 7, 10 and 111. Each serum was probed both for T47D and MCF10 extracts, each in a different lane. The banding patterns obtained with the sera probed with T47D extract were compared to those obtained for the same sera previously; the same was done for the sera probed with MCF10 extract. This allowed the horizontal comparison of all the immunoblots obtained with T47D and MCF10 extracts.

Network Analysis
In order to establish the presence of a Network among the 121 bands (nodes) of 1D-II and evaluate whether the Networks are different among the groups of women with BC, BBP and H, Pearson correlations were calculated among the bands (nodes) in each group of women, using the Statistical Package for the Social Sciences software (SPSS), as a measure of the statistical validity of the bands' interconnections by links.
Statistically significance tests were performed by SPSS, which are based on the t distribution, then the significant value (p-value) can be obtained through hypothesis testing about rho, the null hypothesis. In this case, the null hypothesis implies that the true correlation is zero and that, therefore, the alternative hypothesis' correlation is different from zero. Furthermore, the hypothesis testing algorithm suggests that the testing value should be calculated as follows: Where, r is the Pearson´s Correlation Coefficient and n is the number of sample observations. Finally, the decisions rule establishes that if jtj > ta = 2 ; n À 2, where ta = 2 ; n À 2 is the t distribution value with probability a = 2 and n−2 degrees of freedom, thus, the conclusion is the rejection of the null hypothesis, which means that the correlation between two variables is significantly different from zero and it can be considered statistically significant. In terms of probability, the p-values can be obtained by the probability calculation of t (p-value), based on the t distribution and the decisions rule indicates that if p À value < a = 2 then the null hypothesis should be rejected [31].
In order to create the Networks, the correlation matrix (R) from k group was transformed into a Binary matrix (B), where k = 1,2,3. The transformation indicates that if r i,j (the Pearson's Correlation Coefficient between variable i and variable j with i 6 ¼ j) was considered statistically significant, then, b i,j = 1, if the opposite occurs then b i,j = 0. In this way, the number of Links was given by the number of the ones in B, which means the number of significant Pearson Correlations found in the k group.
Subsequently, if the Pearson Correlations obtained were statistically significant, they were plotted by Agna software [32]. Agna is a platform-independent application designed for social network analysis, sociometry and sequential analysis. Agna allows the user to create, edit, analyze, store and visualize Networks.
Network Analysis (or Social Network Analysis) is a set of mathematical methods used in social psychology, sociology, ethology, and anthropology [32]. This methodology assumes that the way the members of a group communicate with each other affects some important properties of the group (such as performance, leadership, work satisfaction, etc.). We propose that equivalent actors (i.e., tumor cells and clones of antibody producing cells) and equivalent immunological events (i.e., protein synthesis, expression and secretion, cellular differentiation, reproduction or death), occur in immunological networks and affect the performance of the tumor cells (i.e., tissue invasion, metastasis) and of the immune system of the patient with BC in successfully or disastrously dealing with the tumor cells.
Sequential analysis deals with chains of behavior by way of recording the behavior of an animal under specific personal and circumstantial conditions and, then, by dividing the complete set of behavior types into basic sequential units, or links, to make with them a single filed sequential chain of nodes [32].
Here we studied the differences between the 121 bands of the 1D-II of the 50 women with Breast Cancer (BC), those with Benign Breast Pathology (BBP) or without breast pathology (H), and we postulate that the clones of IgG-producing cells are the equivalent components of the immune Network, and adopt the equivalent structures and behavior found in the social networks.
Statistical differences in the number of nodes and links between groups were calculated with the Kruskall-Wallis test. The differences in the intensity of the networks' Connectivity among the groups of women were also analyzed. In order to compare the connection Intensity, it is necessary to define two measures. First, the number of significant correlations in k group is denoted by N k . Then, the expected value of all the significant correlations in k group is calculated by the following formula: Where r i,k represents the significant correlation i of the group k. Finally, the connection Intensity (I i ) can be calculated as follows: As can be inferred from the equation above, I k is a measure which indicates the expected value of Links in group k, based on the Pearson´s Correlations between nodes. Also, disconnected Nodes were identified using the follow algorithm: by applying it as a logical formula in the Excel program addressing the corresponding Excel worksheets. Values (1,0) assigned to "a" and "b" correspond to link presence (1) or link absence (0) between the matrix compared (H vs BBP; H vs BC and BBP vs BC). All matrices were studies to identify output = 2 which point a lost connection, and then a table of disconnections by nodes was constructed.
The network analysis for both T47D and MCF10 were carried out in a similar way.

Results
The Networks' Attributes and Comparisons among the Different Groups of Women Table 2 shows the networks' attributes in each group of the participating women, while Fig. 1 exhibits the networks' visual features. The network with the highest number of links is that of BBP (1932), followed by H (1642) and finally by BC (1610). Despite their numerical differences in connectivity, these numbers were not statistically different from each other (P = 0.207). However, the most connected node (hub) varies among networks: for example, the hub in the H group is node 24 (39 links), for the BBP group it is node 115 (45 links), and node 111 for the BC group (32 links), but there are many other differences in this respect among the networks of the different groups of women (Table 2).
It is noteworthy that the analysis of the frequency (%) in the number of links in each node shows some differences among the groups of women: particularly striking is the decrease in the number of nodes having 19 to 24 links, which is also more pronounced in the groups of women having some pathology: from 17.4% in H, to 13.12% in BBP, and to 6.6% in BC (Fig. 2).
On the other hand, if one takes into account not only the number of links in each node (N, the number of statistically significant correlations), but also their magnitude (M, the average value of their correlation coefficients of the significant correlations in each node), as well as their Intensity (I = NxM), and assigns an intensity ranking order of the most connected nodes (hubs), it can be seen that the network of H women shares 5 of the 10 most connected nodes (hubs 24, 112, 115, 106, and 39) with the BBP women's network, while it only shares a single hub (hub 55) with that of the BC group (Table 3). A similar pattern was observed in the MCF10 networks (S1 Table).
Also, unlike the analysis in the total number of links, when comparing the overall I (SI) in each of the networks, there are statistically significant differences (P = 0.001) among the networks of the women's groups (X = 0.5568595 in H, X = 0.78975207 in BBP and X = 0.41578512 in BC) ( Table 3).
As previously noted, the comparisons among networks show that there are significant differences in their connection intensities. Table 4 shows the number of links that are missing in each node and the nodes with which each node lost its links.
The greatest amount of disconnections by node in the three groups is given in Table 3. It is noteworthy that it is node 24, which is the most disconnected, both in the H vs BC groups (27 disconnections) as in BBP vs BC (33 disconnections). This is remarkable since node 24 is a hub in both networks (H and BBP) ( Table 3). A similar pattern was observed in the MCF10 Networks (S2 Table) when the H and BC groups were compared.
It is also important to note that node 11 is the most disconnected (19 lost links) in BBP with respect to the network of H (Table 4). There are also some nodes that do not lose any links among networks while some are conserved in the three groups of women (nodes 117, 119, 120, 121).

Discussion
No disease exists in isolation [33]. That is, given the functional interdependencies among the molecular components in a human cell, a disease is rarely a consequence of an abnormality in a single gene, but reflects the perturbations of the complex intracellular network [34]. Cancer contains many manifestations of networks at various levels of organization, including the genetic [35,36], cellular [37,38], and phenotypic [39]. The emerging tools of network medicine offer a platform to explore systematically not only the molecular complexity of a particular disease, leading to the identification of disease modules and pathways, but also the molecular relationships between apparently distinct pathological phenotypes [34]. Our results are compatible with the hypothesis of there being a complex immunological network interconnecting the IgG antibody producing clones against at least 121 protein antigens (nodes) differing in their molecular weights (MW). Although the networks of the different groups of women differed in their total numbers of links, the numeric differences were not found to be statistically significant (P = 0.207). This lack of differences among networks could be explained by the shared protein motifs or immunological cross-reactivity, in a manner in which the number of nodes tend to equality and differences become not statistically significant. However, the connectivity intensities [I = NxM] of the association) of the 121 nodes in each of the networks (H, BBP and BC) were statistically different (P = 0.001). That is, the differences found in breast pathology (BBP and BC), and particularly in BC, may be due to the connectivity of the network rather than to the number of nodes in each network.
But there were not only differences in the network connectivity among groups, the number of the most connected nodes (hubs) decreases with the disease, since the diminishing of the nodes having 19 to 24 links were more pronounced in the groups of women having some pathology (Fig. 2). As such, the loss of connectivity in the BC group is probably the consequence of the loss of connectivity of the hubs, as has been seen in other networks [39,40]. More importantly, by way of the disconnection analysis performed among networks, it was possible to identify the lost node in the BC group (node 24), which is the most connected hub in the H network. This striking result makes node 24 a potentially therapeutic target. Similarly, when the By this definition, less than the 10% of the nodes in BC should be considered as hubs. Also BC group is the one that present the lower frequency of hubs, followed by H and BBP group.   MCF10 Networks were analyzed, we found that the most connected node of the H network (node 111) was also one of the most disconnected nodes in BC (S1 Table and S2 Table). Nonetheless, since the antigens' characterization is currently being carried out by our group, it is yet not possible to assign any function to the missing nodes' links. However, as for the possible actors and events involved in the immunological control of cancer cells, one may state: Complement (C) fixing antibodies reacting with antigens on the surface of the cancer cells and drilling lethal holes in the cells' membranes, as well as the recruiting into the site of the tumor immunologically active natural killer cells and macrophages, which exert their  noxious action though increasing their activation and promoting their production of hydrogen peroxide, superoxide, the hydroxyl radical and singlet oxygen, altogether key elements which can deliver a deleterious impact upon the membranes of the tumor's cells, thus limiting their reproduction and, in fact, killing many of them [30,[32][33][34][35][36][37][38][39][40][41][42][43].
As for the network's state of arousal of their hubs, nodes and links, which seems to provide protection against BC, as well as the expectation of an effective vaccine, one would anticipate that, at least, be the same which activates against a preexisting BC, and which appears to inflict so much damage to it, or that its strength be increased and/or enriched by incorporating other actors or combining them in a different manner, more effective and durable, as do other networks that are biologically active such as: the networks of neurons and with the striated muscles in the neuromuscular plaques; the networks of mutualist symbiont plants such as beans and corn [44]; the network of flocks of birds migrating southbound [45]; the network of scouting ants weaving trails that lead thousands of sister ants to the leaves of trees and take them back to their specific anthill; the group of foraging gazelles while being guarded against predators by their fellow vigilantes connected to the group by means of a communicative system consisting of a certain pattern of ground waves created by their hooves kicking the ground; that of the cancer genes in cancer patients [46]. Supporting Information S1 Table. Classification of Immunological Bands of the MCF10 cell-line, according to the Number (N) of Variables with which each one of them significantly correlated (P<0.05), the Average Magnitude (M) of the correlations, the Connection Intensity (I = NxM) it contributes to the Network and its Intensity Ranking Order (R). (DOCX) S2 Table. MCF10 Network Disconnections. A comparison between groups are display in each column (H vs BC, H vs BBP and BBP vs BC). The number of missed links in each node are ordered by their magnitude from the highest to the lowest. Also it is possible to visualize the links missed in each node. Bolded and underlined are the most connected Nodes ranked from 1 to 10. (DOCX)