Evaluation of five column‐based isolation kits and their ability to extract miRNA from human milk

Abstract MicroRNA can be found in various body fluids, including breast milk. MicroRNA may be transferred from mother to infant via breast milk and potentially regulate the development of the infant's immune system on a post‐transcriptional level. This study aimed to determine the microRNA extraction efficiency of five RNA extraction kits from human skim milk samples. Their efficiency was determined by comparing microRNA concentrations, total RNA yield and purity. Furthermore, hsa‐miR‐148a‐3p expression and the recovery of an exogenous control, cel‐miR‐39‐3p, were quantified using qPCR. Each kit extracted different amounts of microRNA and total RNA, with one kit tending to isolate the highest amount of both RNA species. Based on these results, the extraction kit ReliaPrep™ miRNA Cell and Tissue Miniprep System from Promega was found to be the most appropriate kit for microRNA extraction from human skim milk. Moreover, further research is needed to establish a standardized protocol for microRNA extraction from breast milk.


| INTRODUC TI ON
Several factors in breast milk (BM), for example immunoglobulins, have immunomodulatory effects. 1 Interestingly, BM also harbours a vast array of small RNA species that can act as an alternate, and less-explored route, to facilitate immune programming in infants. 2,3 miRNAs are very short RNA molecules (18)(19)(20)(21)(22) nucleotides long) that could potentially influence immune maturation through direct effects, for example by inhibiting expression of key transcription factors for T cell polarization. 2,3 The miRNA that presents in BM is believed to be produced by the mammary epithelial cells, subsequently encapsulated in exosomes and released into the fluid. 4 Milk exosomes, containing the miRNAs, may then be absorbed by the offspring with maintained biological function and be distributed to a range of organs via the circulation. 5 Efficient isolation of miRNAs from biofluids is essential for downstream studies of their functional capacities. However, as standardized protocols for BM miRNA extraction are lacking, it is critical to establish appropriate methodology as has been done for other biofluids like plasma 6,7 and serum. 8,9 A paper from 2015 has compared different RNA extraction kits, including phenol/chloroform, combined phenol column and column filter-based kits in BM; the findings suggest that phenol-free kits should be the primary choice for isolating miRNA. 10 Therefore, we aimed to investigate miRNA extraction efficiency in human BM in a set of chloroform-and phenol-free RNA extraction kits that are currently available on the international market.

| Sample collection and milk fractioning
BM was collected at the first morning feeding by ten healthy women 4 months post-partum and stored at −20°C. After transportation to the university, all samples were kept at −70°C until analysis. The milk was thawed and fractionated by centrifugation, 800 × g, 10 minutes, 4°C; skim milk was further purified by repeating the centrifugation in new tubes, see Figure 1. The study was approved by the Regional Ethics Committee in Linköping, Dnr 2011/45-31.

| Kits and RNA extraction
This study investigates the performance of five phenol/chloroformfree RNA isolation kits that, at the time of the study, were commercially available on the international market.
Although the manufacturers' protocols were generally followed, some modifications were established, as shown in Appendix 1. Two µL (250 pM) of cel-miR-39-3p from Caenorhabditis elegans (Cat# 4464066, Invitrogen) was added before extraction, and the elution step was repeated to extract as much miRNA as possible from the matrix.

| Quantification and purity
RNA quantification was performed using Qubit microRNA Assay kit and Qubit RNA HS Assay kit together with the Qubit 3.0 Fluorometer (Invitrogen), following guidelines; Nanodrop ND-1000 spectrophotometer (Thermo Fisher Scientific) was used for purity assessment.

| miRNA recovery
The exogenous miRNA cel-miR-39-3p and the endogenous miRNA hsa-miR-148a-3p were quantified by qPCR. cDNA template was synthesized using TaqMan Advanced miRNA cDNA Synthesis Kit (Applied Biosystems), following the manufacturers' guidelines. The qPCR product was preformed using TaqMan Fast Advanced Master Mix (Applied Biosystems), TaqMan Advanced miRNA Assay primers cel-miR-39-3p (Applied Biosystems, Assay ID 478293_mir) and hsa-mir-148a-3p (Applied Biosystems, Assay ID 477814_mir). The reactions were processed in a 7500 Fast Real-Time PCR instrument (Applied Biosystems) using the following settings: 95°C for 20 seconds, followed by 40 cycles of 95°C for 3 seconds and 60°C for 30 seconds. All reactions, including the non-template controls, were run in duplicates. Ct values were determined using fixed-threshold and analysed using the Thermo Fisher Connect Software (Thermo Fisher Scientific) available online.

| Statistics
Statistical analysis and data visualization were performed using GraphPad prism version 7.04 (GraphPad software, Inc). The performance of the extraction kits was compared in terms of (a) their miRNA and total RNA recovery, (b) their hsa-miR-148a-3p yield and (c) their ability to recover the exogenous cel-miR-39-3p. Wilcoxon matched-pairs signed rank test was used to evaluate statistical significance between the extraction kits.
Two outliers were removed from the RNA purity data using the ROUT method 11 before further comparisons using Wilcoxon matched-pairs signed rank test. P values of <.05 were considered significant.

| RNA concentration
Refer to Table 1 for median and range.
The total RNA concentration after extraction with Norgen Single Cell (NSC) kit was lower compared with Promega, Zymo and Norgen RNA (NR) (P < .05), and the Promega kit yielded higher concentrations compared with Zymo (P < .05). F I G U R E 1 Overview of the experimental design The NSC kit had lower miRNA yield compared with Promega, Zymo and NR (P < .05), the Promega kit yielded higher concentrations compared with Zymo and Sigma-Aldrich (P < .05), and NR yielded higher concentrations than that of the Sigma-Aldrich kit (P < .05).

| Purity
Only the Promega and Sigma-Aldrich kits yielded reasonable 260/280 ratios, Table 1. The Promega kit had a higher 260/280 ratio than Zymo and both Norgen kits (P < .01), and the Sigma-Aldrich kit had a higher ratio than Zymo (P < .05). All the included kits produced 260/230 ratios <1 except the Sigma-Aldrich kit, which was higher than the other kits (P < .01). Also, a higher 260/230 ratio was observed in the Promega kit compared with the NSC (P < .05).

| Extraction efficiency
The Promega, Zymo and NR kits recovered similar amounts of cel-miR-39-3p (Figure 2), showing better recovery than the NSC and Sigma-Aldrich kits (P < .05). The Zymo kit recovered the most hsa-miR-148a-3p, significantly more than the NR, the NSC and the Sigma-Aldrich kits (P < .05). No difference was found between Zymo and Promega, or the Promega and Sigma-Aldrich kits; NSC had the lowest recovery of all five kits.   Together, these factors contribute to small but important differences, for example in lysis and binding abilities, and thus limit comparability between the kits and between different studies. In the optimization process proceeding the laboratory work for this study, we made some minor changes to the protocols in order to make them work more efficiently with our samples. Only the Norgen kits had a protocol for biofluids and did not need any major modifications; we followed the protocol described for blood. The other kits, however, were primarily designed for RNA extraction from cells. Hence, we modified the lysis step slightly by adjusting the amount of ethanol added and disregarded the steps for purification of sample from cellular debris. We, however, followed the proportion stated for lysis and ethanol for each kit. It is possible that, although we increased performance with these set of changes, the increased exposure of the samples to ethanol may have contributed to the lower than expected RNA quality in our isolates. However, in our case when we have a low miRNA concentration and likely salt contamination from the lysis buffer, spectrophotometry is potentially a less suitable method to estimate RNA quality.
The Nanodrop measurements of RNA purity are evaluated by the ratio of absorbance at the 260/230 nm and the 260/280 nm spectrum. This is because nucleotides, that is RNA and DNA, will absorb at 260 nm and common contaminants will appear at the other wavelengths. 14  In our case, low RNA concentration and guanidinium salt contamination are likely the cause of the low RNA purity in our samples.
One strength with this study is that this is the first study, to the best of our knowledge, that uses Qubit for miRNA quantification and not exclusively the Bioanalyzer or Nanodrop in the BM field.
We chose to include this method because of the previously reported high selectively and reproducibility of Qubit, as compared to Bioanalyzer and Nanodrop. 20 Indeed, we found the Qubit to be sensitive and reliable with a high reproducibility in its measurements, both between samples and between readings of the same samples at Nanodrop, as compared to Qubit, is that it quantifies the RNA based on absorbance at 260 nm, including all nucleotides present in the sample meaning that the Nanodrop will also take into account fragmented RNA. Moreover, even though the Bioanalyzer may be more selective for small RNA as compared to the Nanodrop, Garcia-Elias et al 20 observed low reproducibility, also limiting accurate quantification. Furthermore, Qubit tolerates contaminants, such as salts, better than the Nanodrop, which may be of significance when the samples are likely to contain chemical residues as discussed above.
Lastly, one additional strength with this study is that an exogenous control, cel-miR-39-3p, was added in a standardized amount to all samples prior to the RNA extraction. This control allowed for normalization of technical variability between the samples and was used to assess miRNA recovery. A low coefficient of variance (CV) between the samples within a kit would indicate consistency in the amount of miRNA recovered; in this study, Promega had the lowest CV (4.53%) of the included kits.

| CON CLUS ION
In conclusion, our results suggest that at least two of the extraction kits, the ones from Promega and Zymo, were reasonably efficient in terms of recovery and consistency. Acceptable 260/280 ratios were also observed for the Promega kit, although this kind of spectrophotometry method is not so well suited to evaluate purity in miRNA samples. As the Promega kit was also one of the best performers on the other investigated parameters (ie total RNA extraction efficiency, extraction of hsa-miR-148a-3p and recovery of cel-miR-39-3p), as of today, this would be our primary choice for miRNA extraction from BM. For future validation studies, it is important to keep in mind that the input volume must be standardized, as must the elution volume, to produce comparable results between kits. A spike-in of an exogenous miRNA is also needed to control for technical variability and for evaluating miRNA recovery. The exogenous control should also be used to normalize levels of endogenous miR-NAs and is particularly important in the absence of reliable reference miRNAs comparable to the 'housekeeping genes' utilized in mRNA qPCR. Furthermore, we would recommend the Qubit miRNA assay as a reliable, sensitive and reproducible method to estimate miRNA content in RNA isolates.

ACK N OWLED G EM ENTS
We want to thank Dr Ashley Hutchinson, Örebro University, Sweden, for language checking; Rebecka Thune (BSc), Linköping University, Sweden, for isolating miRNA from human mammary epithelial cells for validation purposes; Dr Chrysanthi Alexandri, University Libre de Bruxelles, Brussels, for supplying data on previous kit performance.
This work was supported by a grant from Lisa and Johan Grönberg Foundation, Sweden.

CO N FLI C T O F I NTE R E S T
The authors confirm no conflicts of interest. Writing-review & editing (equal).

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.

Sigma mirPremier microRNA Isolation Kit
The protocol provided with the kit was modified as follows: 126 µL Lysis buffer was mixed with 64 µL binding solution and 10 µL 100% ethanol. 100 µL sample was lysed with 2 volumes lysis buffer, followed by vortexing for 15 s and incubation for 5 min in room temperature. 1.1 volume of 100% ethanol was added to the lysate, followed by vortexing for 15 s. The lysate was transferred to a Binding column, followed by centrifugation at 16 000 × g for 30 sec.
The flow-through was discarded, and 700 µL of 100% ethanol was added to the column, followed by centrifugation at 16 000 × g for 30 s. The flow-through was discarded and 500 µL binding solution was added, followed by centrifugation at 16 000 × g for 30 s. The column was transferred to a new collection tube and was washed twice with 500 µL Washing solution, followed by centrifugation at 16 000 × g for 30 s. The flow-through was discarded and the column was dry-centrifuged at 16 000 × g for 1 min. The column was transferred to a new tube and 20 µL RNase-free water was added, followed by 1-min incubation in room temperature and later centrifuge at 16 000 × g for 1 min. To maximize the miRNA recovery, a second elution was performed with the eluted RNase-free water.

Zymo Quick-RNA MicroPrep Kit
The protocol provided with the kit was modified as follows: 100 µL sample was lysed with 2 volumes of lysis buffer, followed by vortexing for 15 s. Two volumes of 99.5% ethanol was added to the lysate followed by vortexing for 10 s, before it was transferred to a Zymo-Spin™ IIICG Column and centrifuged at 12 000 × g for 30 s. The flowthrough was discarded and 400 µL of RNA Prep Buffer was added to the column before centrifugation at 12 000 × g for 30 s. The flowthrough was discarded, and 700 µL RNA wash buffer was added to the column before centrifugation at 12 000 × g for 30 s. The flowthrough was discarded, and the washing step was repeated by adding 400 µL RNA wash buffer, before centrifugation at 12 000 × g for 2 min, followed by dry centrifugation of the column at 12 000 × g for 1 min. The column was transferred to a new tube and 20 µL RNasefree water was added to elute the miRNA, followed by centrifugation at 16 000 × g for 30 s. To maximize the miRNA recovery a second elution was performed with the eluted RNase-free water.

Promega ReliaPrep miRNA Cell and Tissue Miniprep System
The protocol provided with the kit was modified as follows: 100 µL sample was lysed using 3 volumes lysis/1-Thioglycerol buffer and 600 µL 95% ethanol was added to the lysate, followed by vortexing for 10 s. The lysate was transferred to a ReliaPrep™ Minicolumn, followed by centrifugation at 12 000 × g for 30 s. The flow-through was discarded and 500 µL wash buffer was added to the column, followed by centrifugation at 12 000 × g for 30 s. The flow-through was discarded and 500 µL wash buffer was added to the column, followed by centrifugation at 12 000 × g for 2 min.
The column was transferred to a new tube and 20 µL RNase-free water was added, followed by centrifugation at 12 000 × g for 1 min. To maximize the miRNA recovery, a second elution was performed with the eluted RNase-free water.

System
The protocol provided with the Norgen kits were modified as followed: 100 µL sample was lysed using 3.5 volumes of Buffer RL, followed by vortexing for 15 s. 200 µL 99.9% ethanol was added to the lysate, followed by vortexing for 10 s. The lysate was transferred to the column and centrifuged at 3500 × g for 1 min. The flow-through was discarded and 400 µL wash solution A was added to the column, followed by centrifugation at 14 000 × g for 1 min. The flow-through was discarded, and the same washing step was preformed twice. After the third washstep, the flow-through was discarded, followed by 2 min drycentrifugation of the column at 14 000 × g. The column was

Disclosure
The two kits from Norgen Biotek Corp were provided to our laboratory as free samples.