The serum protein profile of early parity which induces protection against breast cancer

Early parity reduces the risk of breast cancer in women while nulliparity and late parity increase the risk of breast cancer. In order to translate this protection to women where early pregnancy is not feasible, much work has focused on understanding how parity confers protection against breast cancer, the molecular mechanisms by which this occurs is still not well understood. Healthy parous and nulliparous women were recruited for this study. We assessed serum protein profiles of early parous, late parous, and nulliparous women using the Phospho Explorer antibody array. Significantly altered proteins identified were validated by Western blot analysis. In silico analysis was performed with the data obtained. Our findings indicate increased phosphorylation levels of CDK1, AKT1 and Epo-R increased cell cycle and cell proliferation in late/nulliparous women. Increased levels of LIMK1, paxillin, caveolin-1, and tyrosine hydroxylase in late/nulliparous women demonstrate enhanced cell stress while decreased activity of p-p53 and pRAD51 in late/nulliparous women indicates decreased apoptosis and increased genomic instability. Further, increased levels of pFAK, pCD3zeta, pSTAT5B, MAP3K8 in early parous women favor enhanced innate/adaptive immunity. Overall, we have identified a unique protein signature that is responsible for the decreased risk of breast cancer and these proteins can also serve as biomarkers to predict the risk of breast cancer.


INTRODUCTION
Breast cancer remains the most commonly diagnosed cancer in women worldwide [1,2] and current therapies are plagued by the development of resistance and high toxicity [3]. Therefore, the best treatment approach would be to elucidate and employ strategies that effectively reduce the risk of breast cancer in women. This requires discovery of suitable biomarkers that can accurately predict a woman's risk of developing breast cancer and a woman's risk of disease progression if breast cancer is diagnosed. Pregnancy before the age of 25 reduces a woman's risk of developing breast cancer by as much as 50% [4]. On the contrary, a first pregnancy after the age of 35 is known to increase a woman's risk of developing breast cancer [5]. Despite concerted efforts to understand this phenomenon, the exact etiology behind parity-induced protection against breast cancer is not fully understood.
In previous epidemiologic studies and in studies using rodent animal models, we and others have shown that there are notable differences in the hormonal milieu of parous vs. nulliparous individuals and these differences account, in part, for parity-induced protection against breast cancer [6][7][8][9][10][11][12][13]. These prior studies identified Research Paper alterations in pituitary-gonadal hormones as the primary contributors to the systemic effects of parity, which include differences in prolactin and growth hormone levels [6][7][8][9][10][11]. We and others have also shown that alterations in growth hormone signaling yields the most pronounced effect on breast cancer susceptibility in rodent models [9,[14][15][16]. In fact, growth hormone-deficient rats are completely resistant to mammary carcinogenesis [16]. Further, Abrams and colleagues demonstrated that when susceptible rat mammary epithelial cells were treated with carcinogen and transplanted into virgin or parous hosts, nearly all virgin host transplants developed tumors compared to a negligible number of parous host transplants [17]. Again, this demonstrates that parityinduced protection is largely accomplished by systemwide changes which impacts breast tissue in such a way as to render mammary epithelial cells refractory to carcinogenesis. Further, changes at the cellular level may also contribute to parity-induced protection [18,19].
Given the strong evidence that protection against breast cancer in parous individuals is through systemic changes, it is surprising that virtually no recent study in human subjects has looked at parity-induced systemic changes. Most recent studies in human subjects and rodent animal models have focused on parity-induced changes at the tissue and cellular level [19][20][21][22][23][24]. Current studies in human subjects have compared 1) gene expression profiles of mammary tissue from parous and nulliparous women [12,22]; and 2) gene expression profiles of mammary tissue from premenopausal women undergoing reduction mammoplasty and breast biopsy of benign breast lesions [21]. To our knowledge, the results presented here represent the largest study of completely normal healthy women that includes subjects stratified according to nulliparous, late parous and early parous status. The current study is also unique because we have assessed parity-induced systemic changes in the serum of normal healthy women not undergoing breast reduction surgeries, augmentations surgeries, or lesion biopsies. Here, we analyzed global changes in protein expression and identified serum protein profiles that are unique to 1) low-risk early parous women, and 2) high-risk nulliparous and late parous women. This study is critical for identification of breast cancer biomarkers that could effectively predict the risk and progression of breast cancer, as well as, offer important clues towards developing strategies to prevent breast cancer in highly susceptible women. We expect the unique protein signatures identified here will provide important insights for both the prevention and treatment of breast cancer in women.

Proteomic analysis
A total of 132 healthy volunteers-44 nulliparous subjects, 44 late parous subjects, and 44 early parous subjects-were recruited for this study [12]. Protein expression profiling was carried out using the Phospho-Explorer protein array. Proteomic analysis of the data obtained was performed using GeneSpring bioinformatics software and Metacore knowledge-based bioinformatics software. Raw protein expression values were first normalized and then subjected to a fold change cutoff of ± 1.3. Fold differences in protein expression with a p-value of ≤ 0.05 were considered significant. As expected, there were notable differences in the protein expression of each group ( Figure 1A). Moreover, a large number of proteins were differentially regulated in late and nulliparous samples compared to early parous samples ( Figure 1B) and a subset of these proteins were commonly expressed between late and nulliparous groups ( Figure 1C). In order to distinguish proteins that may be involved in facilitating the protective effects of parity, we specifically focused on proteins with similar expression patterns among the two high risk groups (late and nulliparous) that were distinct from the expression pattern observed in early parous subjects. Of note, the majority of differentially expressed proteins appear to be downregulated in late/nulliparous subjects compared to early parous subjects. Many of these are highlighted by arrows in Figure 1A.

Apoptosis and DNA damage responses
Several interesting patterns of protein expression immediately emerge from late/nulliparous vs. early parous serum protein profiles. First, expression levels of proteins critical for regulating apoptosis and repairing DNA damage appear to be decreased in the late/nulliparous groups in comparison to the early parous group. These include RAD51 and p53 (Table 1, Figure 2A). p53 is particularly important for cell cycle regulation and induction of apoptosis in response to DNA damage. When compared to early parous women, two phosphorylated forms of p53 critical for effective protein activity are downregulated in late and nulliparous women (Table 1, Figure 2A). Phosphorylation of p53 at serine 20 has been shown to alleviate MDM2-mediated inhibition of p53 [25] and this form of the protein is significantly downregulated in late/nulliparous women (Table 1). Similarly, p53 phosphorylated at threonine 81 stabilizes p53 in response to cell stress [26] and this form of the protein is also downregulated in late/nulliparous women (Table 1). This indicates that p53 is less stable and more prone to inhibition in late/nulliparous women. RAD51 phosphorylated at tyrosine 315 is also significantly downregulated in late/nulliparous women compared to early parous individuals (Table 1, Figure 2A), and this form of RAD51 has enhanced binding affinity for damaged DNA, which facilitates DNA repair [27]. This suggests that there may be an impaired DNA damage response in late/nulliparous women compared to early parous women.

P19235
Dysregulation may affect the growth of certain tumors. Erythropoietin growth factor signaling.
Activates JAK/STAT signaling. In the case of p53, there appears to be a deficiency in p53 activation in late/nulliparous women based on the ratio of phosphorylated protein to total protein ( Figure 2A). For RAD51, on the other hand, the ratio of total to phosphorylated protein remains the same for all three groups (Figure 2A), suggesting a deficiency in protein translation. Early parous individuals have much higher levels of active CHK1 (Figure 2A) while there is no significant difference in active CHK2 levels ( Figure 2A).

Cell cycle and cell proliferation
Another intriguing pattern of protein expression in late/nulliparous subjects is indicative of poorly controlled cell cycle progression which may lead to enhanced cell proliferation. This is reflected in the downregulation of CDK1 (p-Thr14) along with the upregulation of the RET and Src proto-oncogenes (Table 1, Figure 2B). Downregulation of HRS (hepatocyte growth factorregulated tyrosine kinase substrate) may also play an important part in "poorly controlled" cell proliferation within these populations ( Table 1). Phosphorylation of CDK1 at threonine 14 inhibits CDK1 activity thus inhibiting cell cycle progression [28]. Therefore, downregulation of this form of CDK1 suggests that cell cycle progression may actually be enhanced in late/nulliparous women. Further, despite significant variation within replicates which precludes statistical significance, there is a trend for increased phospho-RET and phospho-Src activation in late/nulliparous subjects ( Figure 2B). In many cases, RET gain-of-function is associated with breast cancer formation through enhanced cell proliferation, migration and cell scattering, as well as, through enhanced cytokine production [29][30][31][32]. Likewise, Src hyperactivation fuels progression of various cancers [33] and plays a key role in breast cancer resistance to Herceptin therapy [34]. Therefore, overexpression of these proteins may render cells more susceptible to hyperactive signaling and eventually oncogenic transformation.
It is interesting that total protein levels of several proto-oncogenes were upregulated in early parous women, including c-Jun and FosB (Table 1). However, only in the case of FosB there is a corresponding trend for increased protein activity compared to late/ nulliparous subjects ( Figure 2B). Early parous serum protein profiles also indicate alteration of several growth factor signaling pathways such as the erythropoietin pathway (Epo-R), the human epidermal growth factor pathway (HER2, JAK1), and the hepatocyte growth factor pathway (Met) ( Table 1). We actually see higher expression of active Epo-R and Met in late/nulliparous women and no difference in active Her2 between early and late parous individuals (Table 1, Figure 2C). This suggests that these pathways do not enhance cell growth above a normal basal level in early parous women compared to late/nulliparous women.
Further, there is strong evidence of better control over cell cycle progression and cell growth in early parous women based on the following findings: 1) upregulation of phosphorylated CDK1 (Thr14), 2) downregulation of phosphorylated AKT1 (Ser124), 3) downregulation of phosphorylated Epo-R (Tyr368), and 3) upregulation of HRS (Table 1, Figure 2B and 2C). Upregulation of the inhibitory form of CDK1 [28] coupled with downregulation of the form of AKT1 required for optimal downstream activation [35] likely reflects much tighter control over cell cycle progression in early parous individuals. Suppression of the active form of Epo-R also indicates more fine-tuned control of this signaling pathway in early parous women. Also, the HRS protein is more abundant in early parous subjects and is required by many growth factor receptors for appropriate receptor recycling, inactivation, and degradation [36]. This indicates that growth factor signaling is more well-balanced in early parous individuals.

Cytoskeletal remodeling, cell stress and immune responses
A third serum protein expression profile that emerges from this data is the differential regulation of many proteins involved in cytoskeletal remodeling, cell adhesion and migration (Table 2). While a number of proteins in this category were downregulated, several key proteins were significantly upregulated in late/nulliparous subjects, including LIMK1, paxillin, and caveolin-1 (Table 2, Figure 2D). LIMK1, is important for stabilizing the actin cytoskeleton and thus in cell motility and cell cycle progression [37]. Likewise, paxillin is important in the regulation of focal adhesions/cell motility and serves as a docking protein for intracellular signaling [38]. LIMK1 overexpression has been noted in breast cancer progression [39], while paxillin overexpression has been shown to enhance breast cancer metastasis [40]. Caveolin-1 is a scaffolding protein found in certain types of lipid rafts that facilitate cellular signal transduction [41]. Caveolin-1 has been shown to be positively influence many biological processes including cell migration and cell stress responses [41,42]. In fact, caveolin-1 is phosphorylated on tyrosine residues in response to cellular stress [42], and pTyr14-caveolin-1 is significantly upregulated in late/nulliparous women perhaps suggesting increased levels of cellular stress (Table 2, Figure 2D). Upregulation of tyrosine hydroxylase in late/nulliparous women further suggests enhanced cellular stress (Table 2, Figure 2D). Chronic social stress has been tied to increased tyrosine hydroxylase levels which catalyze catecholamine synthesis-a process frequently deregulated by chronic stress in both humans and animals [43].
Compared to late/nulliparous women, early parous subjects demonstrate upregulation of FAK/PAK and MAPK signaling (Table 2). Further, various proteins important for T-cell and NK cell mediated immunity are expressed at higher levels in early parous serum samples,

P07101
Catecholamine biosynthesis, stress responses. Phosphorylation at this site results in the most potent induction of TH activity.

Validation of protein array results
The final validation of our protein array results was accomplished through Western blot analysis of a subset of proteins chosen based on reproducibility in initial analyses, statistical significance, and biological relevance to parity-induced protection. These proteins included tyrosine hydroxylase, PKC theta, LimK1, pRad51 (Tyr315), p-p53 (Thr81), and pCaveolin-1 (Tyr 14) ( Figure 3A). As demonstrated in the representative Western blots, the pattern of protein expression in serum samples assayed via Western blot closely mirrors the expression pattern observed in our protein array results. Although there was biological variation between individual replicates, the overall expression pattern remained similar. This confirms the integrity of the protein array data and indicates the accuracy and reproducibility of our proteomic analysis.
Innate and adaptive immunity represent many of the top scoring gene ontology and biological pathways more significantly associated with proteins expressed at higher levels in early parous individuals. Further, early parous protein expression patterns demonstrate enrichment for gene ontology/biological processes that regulate cell cycle progression, apoptosis, protein phosphorylation, and transcription ( Figure 3B-3E). Late/nulliparous protein expression patterns, on the other hand, were primarily associated with growth factor signaling and hormone and metabolic signaling ( Figure 4A and 4B). All together, these results indicate that early parous populations may benefit from a greater level of "biological control" over basic cellular pathways governing everything from cell proliferation to immunity.

DISCUSSION
The objective of this study was to survey differences in protein expression in the serum of normal healthy women to determine key differences in protein expression patterns in early parous vs. late and nulliparous women. This is critical, because examining differential protein expression according to this stratification allows us to distinguish parity-induced changes that increase or decrease breast cancer risk from changes that may be parity-associated but do not impact breast cancer risk. Changes in protein expression patterns that are unique to early parous women compared to those commonly shared by late parous and nulliparous women are likely related to reduce the risk of breast cancer.
Indeed, here we identified specific patterns of systemic protein expression unique to high-risk late and nulliparous women compared to low-risk early parous women. We expect that this distinct pattern reduces the risk of breast cancer, and also allows us to identify biomarkers useful in predicting breast cancer risk. The molecular signatures identified here will also help in the design of safe and non-invasive breast cancer prevention strategies for those individuals where early parity is not feasible.
The patterns of protein expression unique to late/ nulliparous individuals revealed a protein signature associated with enhanced cell proliferation, enhanced cellular stress, and impaired DNA repair mechanisms. Metacore knowledge-based bioinformatics analysis further revealed that this protein expression signature is strongly associated with the development of various cancers including many forms of breast cancer. In fact, the top scoring network for this protein signature reveals a potential contribution from androgen receptor signaling and various G-protein coupled receptor signaling pathways. This is interesting because androgens are necessary precursors for estrogen synthesis and androgen receptor signaling itself has been demonstrated to increase cell proliferation and breast cancer progression [47]. This suggests an overall higher level of steroid hormone signaling in late/nulliparous women compared to early parous women. Earlier Asztalos et al. [21] reported that estrogen and progesterone signaling was downregulated in the parous breast compared to the nulliparous breast. Therefore, parity not only results in local suppression  of steroid hormone signaling in breast tissue [21], but according to our results, also suppresses steroid hormone signaling systemically throughout the body. Another very interesting observation from this study is the apparent enhancement of immune defense regulation in early parous women compared to late/nulliparous women ( Figure 3B and 3C). Also intriguing is the enhanced level of transcriptional and posttranscriptional regulation observed in early parous individuals-an observation which coincides with the findings of Peri et al. in breast tissue mRNA profiles [22]. This demonstrates that a tighter level of transcriptional regulation is observed at both the local tissue level [22] and systemically in early parous women. Therefore, this likely represents another important protective effect of early parity. But, what may also be of paramount importance to parity-related protection against breast cancer is the increased level of immune system regulation observed in early parous women. Our study reveals a significant contribution from both innate and adaptive immunity in the early parous proteome. We believe this unique signature may actually inhibit breast cancer in early parous women through increased immune surveillance, which would more effectively eliminate any breast epithelial cells that may undergo malignant transformation. Indeed, various epidemiologic studies demonstrate that heightened immunity is associated with protection from various cancers [48,49]. It is therefore entirely possible that parity may have a similar effect on immunity and this heightened immunity may be a key aspect of parity-mediated protection against breast cancer.
In conclusion, this study demonstrates that late/ nulliparous individuals appear to have higher baseline levels of cell proliferation and possibly cellular stress. In addition, there is also decreased cell cycle/apoptosis regulation and impaired DNA damage responses. These characteristics ultimately results in "non-specific" and "uncontrolled" cell proliferation with a much greater potential for malignant transformation in late and nulliparous women. In contrast, early parous individuals benefit from enhanced apoptosis/cell cycle regulation, DNA repair mechanisms, and immune defenses. We hypothesize that this translates into much more "wellcontrolled" cell proliferation where any malignantly transformed cells are more efficiently eliminated through several mechanisms including appropriate DNA repair and recovery or through programmed cell death and immune detection/elimination. This study, together with other recent reports [21,22], provides a strong foundation for precisely defining the protein expression signature associated with decreased breast cancer in early parous women. We expect that this will shortly result in the identification of biomarkers suitable for predicting breast cancer risk and progression and will also lead to the development of effective breast cancer prevention and treatment strategies.

Ethics statement
All experiments performed were reviewed and approved by the Texas Tech University Health Sciences Center Institutional Review Board and Institutional Biosafety Committee.

Volunteer criteria and samples
Healthy parous and nulliparous women between the ages of 40 and 60 yr. were recruited for this study at the Dr. Susan Love Research foundation in Santa Monica, CA [12]. Protocols for obtaining patient consent to donate venous blood and for the preservation of patient confidentiality, no patient identities were obtained. This protocol was reviewed and approved by the Institutional Review Board of Texas Tech University Health Science Center in El Paso, TX and by the Institutional Review Board of the Dr. Susan Love Foundation. Blood samples were drawn, processed to extract serum, the resulting serum samples were aliquoted and preserved at −80ºC until further processing. Recruited women were divided into the following groups: early parous (first full-term pregnancy ≤ 25 yr.), late parous (first full-term pregnancy ≥ 35 yr.) and nulliparous (never undergone a full-term pregnancy). For parous women, samples were collected at least 5 years after the last full-term pregnancy. Only women who had a history of regular menses and who had not regularly taken medications or oral contraceptives in the six months prior to sample collection were admitted to the study. Subjects with liver disease, renal disease, or endocrine problems were excluded from the study.

Phospho Explorer antibody array
Phospho Explorer antibody arrays were used to assess differences in serum protein profiles between the three groups. Protein array assays were performed according to the manufacturer's instructions. Briefly, whole serum was depleted of albumin (see below), diluted in labeling buffer (1:15), and mixed with Biotin/ DMF labeling solution. Biotinylated proteins from each sample were then incubated on individual antibodycoated slides, followed by fluorescence labeling with Cy3-streptavidin (GE Healthcare). The slides were scanned on an Axon GenePix Array Scanner (Molecular Devices) to detect bound proteins based on fluorescence intensity. Fluorescence intensity of each protein array was normalized and analyzed using GeneSpring software v. 12.5 (Agilent, Santa Clara, CA, USA).

Serum albumin depletion
Upon collection, blood samples were centrifuged in serum separator tubes and serum samples were stored at −80ºC until further analysis. In order to minimize interference from the most abundant serum proteins, serum albumin was removed using the PureProteome albumin magnetic beads kit. Briefly, beads were washed twice with 1X PBS; human serum was diluted (1:15) in 1X PBS and added to the magnetic beads. The mixture was then incubated for an hour at room temperature on a rotator and beads were separated from the albumin-depleted serum according to the manufacturer's instructions.

Western blot analysis
For Western blot, 24 microliters of albumindepleted serum was separated on 4-20% sodium dodecyl sulfate polyacrylamide gels and transferred onto PVDF membranes. The membranes were then probed with select primary antibodies (as per manufacturer's instructions) overnight at 4ºC. Secondary monoclonal anti-rabbit antibodies conjugated to horseradish peroxidase (Sigma Aldrich Inc, St Louis, MO, USA) were used to detect the proteins of interest with SuperSignal West Femto Maximum Sensitivity Substrates (Thermo Fisher Scientific Co, Hanover Park, IL, USA). Chemiluminescent signals were visualized digitally on the ImageQuant LAS 4000 digital imaging system (GE Healthcare BioSciences Corp, Piscataway, NJ, USA).

Statistical analysis
Differences in the expression of individual proteins were compared between groups using an un-paired Student's t-test. Volcano plots and Venn diagrams were generated using a fold change cutoff of ≥ 1.3. Asymptotic p-value computation and multiple correction testing were done using the Benjamini-Hochberg procedure. Only the proteins that had a 30% of higher difference in expression and also had a value of p < 0.05 was considered as statistically significant.