Human Milk and Matched Serum Demonstrate Concentration of Select miRNAs

Background

More than 45,000 U.S. women younger than the age of 50 are diagnosed with breast cancer yearly, and until age 40, parous women are more likely to develop breast cancer than nonparous women, with an average age of onset of breast cancer more than 5 years earlier than nonparous women. ¹ Parity influences breast cancer risk for at least 10 years after childbirth, suggesting that it influences the development of a large fraction of premenopausal breast cancers. Pregnancy-associated breast cancers (PABCs) diagnosed after childbirth are often aggressive, with a poor prognosis. ² One study reported the mean overall 5-year survival for women with PABC was 52% versus 80% with non-PABC. ³ Our understanding of the human biology of PABC is limited, and there are no clinically useful predictors of the disease.

There are many potential influences of breastfeeding on PABC. Breastfeeding increases breast density, making clinical examination and mammography more difficult to assess. ¹ However, PABC has not been consistently shown to be diagnosed at a larger size/later stage than other breast cancers. ¹ Another influence of pregnancy, with or without breastfeeding, is the process by which the breast undergoes involution to its prepregnancy state. This process and the environment that it creates have been characterized as similar to what occurs during wound healing and inflammation, which is pro-oncogenic. ¹

There is increasing evidence that miRNAs, which are RNA fragments of ∼22 nucleotides, influence breast cancer development and progression, ⁴ and miRNAs are present in human breast milk. ⁵ miR10a is able to transcriptionally inhibit Hoxd4 expression in human breast cancer cells. ⁶ miR155 and miR21 have been implicated in breast cancer epithelial to mesenchymal transition (EMT), cell migration, and invasion control.^4,7 Serum levels of miR16, ⁸ miR-21, ⁹ and miR-155¹⁰ are significantly higher in patients with breast cancer than controls. The expression of miR100 has been correlated with improved patient survival. ¹¹ We previously observed higher expression of miR100 in normal tissue than in hormone-sensitive rat mammary tumors. ¹²

miR140 expression decreases with increasing breast cancer grade. ¹³ Downregulation of miR145 was reported to predict postmenopausal breast cancer risk. ¹⁴ miR181 was evaluated in the serum of 88 patients with colorectal carcinoma (CRC) and 11 healthy controls. Expression was significantly higher in patients with CRC. ¹⁵ miR199¹² and miR204¹⁶ expression was found to be lower in mammary tumors than in matched controls. miR205 is frequently detected in exosomes from body fluids; it can act both as a tumor suppressor and as an oncogene. ¹⁷ As a tumor suppressor, miR205 inhibits cell proliferation, migration, and invasion. On the contrary, as an oncogene, miR205 promotes tumor initiation and development. miR205 was measured in the serum of 58 breast cancer patients and 93 healthy controls. Levels in the serum of healthy women were significantly higher (p < 0.01) than in women with breast cancer. ¹⁸ A second article ¹⁹ observed the opposite trend, with higher levels of miR205 in the serum of women with breast cancer compared to controls. miR212 has been shown to act a tumor suppressor, inhibiting the growth of lung cancer ²⁰ and cervical cancer. ²¹

As serum levels of many of the miRNAs that we studied were significantly higher in the serum of patients with breast cancer than controls, there is reason to believe that miRNAs should also be differentially expressed in milk from a breast with cancer compared with a clinically normal breast. As a first step to address this, our objectives were to determine if cancer-related miRNAs are present in breast milk fractions (supernatant, a.k.a. skim milk; fat; and cells) and matched serum from lactating women, as well as their relative expression. The availability of breast milk for assessment makes it a potentially ideal body fluid for analysis.

Materials and Methods

After obtaining Institutional Review Board approval from the University of North Dakota for this research, matched milk and blood samples were collected from six lactating women with their written consent. The women ranged in age from 29 to 36 years. All women were Caucasian. One of the women had a family history of breast cancer. None of the women had a personal history of breast cancer. This was the first full-term pregnancy for four women, the second full-term pregnancy for the other two. Women were asked to pump as necessary for collection of hindmilk. Milk was frozen as soon as possible after collection, shipped frozen, and stored frozen. Collected milk was thawed, centrifuged (1,500 g, 20 minutes, 4°C), and the fat and cellular layers separated. The aqueous phase was then centrifuged at 12,000 g for 15 minutes at 4°C, the second lipid layer removed and stored at −80°C before analysis. The samples were separated by centrifugation into three fractions as follows: fat, skim milk, and cell pellet. The separated fractions were then snap-frozen, labeled, and placed in a −80°C freezer until analysis. Blood was placed in a serum separator tube, spun at 1,200 g × 10 minutes, the serum fraction aliquoted and stored at −80°C until analysis.

One hundred sixty nanograms of total RNA, including miRNAs, was extracted from human breast milk and serum samples using the miRNeasy Micro Kit (Qiagen, Valencia, CA) following the manufacturer's instructions. Eleven miRNAs (10a-5p, 16, 21, 100, 140, 145, 155, 181, 199, 205, 212) were analyzed. Snord95, whose expression exhibited minimal variability between participants, was used as a control. Primers were purchased from Qiagen. cDNA was generated using the miScript Reverse Transcription kit (Qiagen). The miScript SYBR Green PCR kit (Qiagen) was used in real-time PCR for analysis of miRNA expression. Reverse transcription and qPCR steps were followed according to the manufacturer's instructions. Relative miRNA concentration (ΔCt) was calculated as unknown Ct – Snord95 Ct for each sample. Normalized gene expression (NGE) was calculated as 2 exp(−ΔΔCt), where ΔΔCt = ΔCt (milk fraction) − ΔCt (serum). Pairwise correlations between miRNA ΔCt data were calculated using Pearson's correlation coefficient.

Results

miRNAs were detectable in all samples of skim milk, milk cells, and milk fat. For all miRNAs, the relative concentration (compared to Snord95 control) was highest in skim milk, intermediate in milk fat, and lowest in milk cells (Table 1). miRNAs were also detectable in all serum samples. NGE was calculated for milk fractions compared to matched serum. NGE was highest in skim milk for all miRNAs (Table 2). Of the 11 miRNAs, miR205 had the highest concentrations relative to serum for 5/6 skim milk specimens, 6/6 milk fat specimens, and 6/6 milk cell specimens. Forty-six percent and 38% of the pairwise correlations between miRNAs in milk cells and milk fat, respectively, were >0.7, compared with only 19% in skim milk.

Discussion

PABC is an aggressive form of the disease that impacts women in the prime of their lives. Standard screening for the disease both through imaging and physical examination is difficult due to the effects of pregnancy and lactation on the breast. There is no proven clinically useful tool to specifically screen for PABC. miRNAs are being evaluated in a variety of malignancies, both to better understand tumor biology and their potential usefulness in disease detection and prognosis. Easy access to breast milk means that biomarker(s) in the fluid found to be clinically useful would be easy to assess in screening for disease or for monitoring response to treatment.

Most studies of miRNAs have been performed in bovine milk. ²² Relatively few publications have evaluated miRNAs in human milk and those generally focus on a single fraction of human milk: skim milk, ⁵ milk exosomes, ²³ or less commonly, milk fat. ²⁴ Alsaweed et al. determined the concentrations from two participants' samples of total RNA and two miRNAs (miR30a-5p and miR148a-3p) in milk cells, fat, and skim milk, ²⁵ with variable relative miRNA concentrations in the cell, lipid, and skim milk fractions.

Our study evaluated in six nursing women 11 miRNAs known to influence cancer risk and found that all were expressed in each milk fraction and in serum. For all miRNAs, expression was highest in the skim fraction of milk. Moreover, for 7/11 miRNAs, expression was higher in skim milk than in matched serum. Based on the known function (outlined in Background) of the seven miRNAs with higher NGE than serum, some promote and others inhibit cancer. miR205, the only miRNA with higher mean NGE expression in all milk fractions compared to matched serum, is pleiotropic, acting as either a tumor suppressor or oncogene, depending on the context. It is downregulated in cells that have undergone EMT, and forced overexpression leads to mesenchymal to epithelial transition. ²⁶ Transforming growth factor (TGF)β isoforms are known to be upregulated in the lactating breast, especially during the weaning process. ²⁷ miR205 may counterbalance the EMT-related effects of TGFβ in the lactating breast, and indeed it has been shown to block TGFβ-induced EMT in other systems. ²⁶

When conducting additional studies, we suggest analysis of the skim fraction as a first step, as opposed to all fractions in milk, since miRNA concentrations were highest in the skim fraction and less highly correlated than in other milk fractions.

Conclusion

miRNAs are detectable in skim milk, milk cells, milk fat, and in matched serum. The highest concentrations in milk were in the skim fraction. miR205 had the highest concentrations relative to serum for all three milk fractions. Fewer miRNAs tested were highly correlated in skim milk than in milk fat and milk cells.