Selection of reference genes for quantitative analysis of microRNA expression in three different types of cancer.

MicroRNAs (miRNAs) are promising biomarkers in cancer research. Quantitative PCR (qPCR), also known as real-time PCR, is the most frequently used technique for measuring miRNA expression levels. The use of this technique, however, requires that expression data be normalized against reference genes. The problem is that a universal internal control for quantitative analysis of miRNA expression by qPCR has yet to be known. The aim of this work was to find the miRNAs with stable expression in the thyroid gland, brain and bone marrow according to NanoString nCounter miRNA quantification data. As a results, the most stably expressed miRNAs were as follows: miR-361-3p, -151a-3p and -29b-3p in the thyroid gland; miR-15a-5p, -194-5p and -532-5p in the brain; miR-140-5p, -148b-3p and -362-5p in bone marrow; and miR-423-5p, -28-5p and -532-5p, no matter what tissue type. These miRNAs represent promising reference genes for miRNA quantification by qPCR.

Introduction

The discovery of the small non-coding RNA lin-4 in Caenorhabditis elegans in 1993 established a base for a new line of research in molecular biology [1]. 2,654 human miRNA genes are currently deposited in the miRBase database [2]. However, some of them are false-positive entries [3]. This led to the need of alternative databases that could give a true view of all miRNAs in existence. One of such databases is MirGeneDB 2.0: in 2020, it contained 556 validated human miRNAs [4]. MicroRNAs play an important regulatory role in many organisms in many biological processes. There is little doubt that aberrant miRNA expression may entail disease initiation and progression [5].

To date, a large number of works showing how miRNAs are distributed in healthy and diseased human tissues have been published [6, 7]. These studies normally seek to identify human disease-specific miRNAs that have promise for personalized medicine, hence the need to have accurate miRNA profiling data on the study samples. The most commonly used approach to addressing this problem is through analysis of miRNA expression levels by qPCR [8]. This technique is highly sensitive, reproducible and relatively cheap. The main problem with qPCR-based protocols is one of the choice of an optimum reference gene. Normalization compensates for variations in the amounts of miRNA following RNA extraction, reverse transcription or that are due to variations in amplification efficiency. A reference gene should have stable expression in all samples, be expressed together with its target in the study cells and have a similar stability with target miRNAs, no matter what sample storage conditions. However, we have yet to know a universal reference gene meeting all these criteria [9]. It has been argued that normalization is the most accurate, if the reference genes used are in the same class of RNA as the study targets [10]. Nevertheless, miRNA expression studies most commonly use non-coding RNAs (U6, U48, U44 etc.). However, non-coding RNAs are not in the same class of RNA as miRNAs and, therefore, the properties of the former are not the same as those of the latter, which may results in different efficiencies of extraction, reverse transcription and PCR amplification. Various authors have demonstrated that, along with miRNAs, small nuclear RNAs are highly variable in expression, too [11-13]. Multiple studies have demonstrated that the best choice of reference gene in the analysis of the relative miRNA expression level is either a miRNA or the geometric mean of several stably expressed miRNAs. Additionally, it is recommended that a reference gene be searched for in each experimental system [9]. The most popular algorithms used for expression stability assessment are geNorm, NormFinder and Best Keeper [14-16].

Although qPCR is the most popular technique for measuring miRNA expression levels, it is not the only one: other molecular genetic methods that can assess the absolute number of miRNA molecules from one to hundreds at once are in use, too, for example, digital PCR, NanoString nCounter and sequencing [17-19]. One of the most popular multiplexed systems used for measuring gene expression is NanoString nCounter. This technique relies on digital barcoding, requires neither mRNA-to-cDNA conversion by reverse transcription nor cDNA amplification by PCR [20]. One experimental session included analysis of about 800 miRNAs in 12 study samples, showing that some miRNAs have variable expression, while the others, stable. Apparently, the miRNA identified by this technique as being stably expressed may further be used as reference genes for qPCR.

The aim of this work was to find the miRNAs that could be used as potential reference genes from among those miRNAs that are stably represented in the thyroid gland, brain and bone marrow, according to nCounter miRNA Assay data.

Materials and methods

Clinical samples

1. Thyroid tissue

All biological material was obtained in compliance with the legislation of the Russian Federation, and written informed consent was provided by all the patients, all the data were depersonalized. This study was approved by the ethics committee of the Institute of Molecular Biology and Biophysics of the Siberian Branch of the Russian Academy of Medical Sciences. We used samples of thyroid tumor tissue surgically removed from 32 patients and representing different neoplasm histotypes and normal tissue: normal tissue, 16; hyperplastic nodule, 5; papillary thyroid carcinoma, 2; follicular variant of papillary thyroid carcinoma, 2; follicular thyroid adenoma, 6; and follicular thyroid carcinoma, 1. Sample collection and histology analysis were controlled by a qualified oncologist (Novosibirsk Municipal Clinical Hospital #1, Oncology Department VI). Detailed clinical information of the patients is given in S1 Table.

2. Brain tissue

Surgery material was two 1-mm3 bioptic samples from each patient’s brain: tumoral tissue and non-tumoral tissue taken farther than 2 cm away from the tumor margin, in a functionally insignificant location. Tumor samples were examined by histology and assigned to groups according to the 2007 WHO grading system of gliomas: Grade I (n = 1), Grade II (n = 2), Grade II (n = 2) and Grade IV (n = 1). Detailed clinical information of the patients is given in S2 Table.

3. Bone marrow tissue

We used 12 cytological specimens obtained by bone marrow aspiration in the Municipal Hematological Center of the Ministry of Health of the Novosibirsk Region. The study groups were patients with myelodysplastic syndrome (MDS) (n = 3), non-Hodgkin’s lymphoma without myelodysplasia ((NHL(−MD)) (n = 3) and NHL with myelodysplasia ((NHL(+MD)) (n = 3); the control group consisted of patients with non-cancerous blood diseases (NCBD) (n = 3). Work with healthy bone marrow donors for allogeneic transplantation is beyond the competence of our clinic. Taken together, we decided to go with a control group composed of people who had no hematologic cancer, but had indications for bone marrow examination to exclude one. They were people with secondary anemic and cytopenic conditions, in whom leukemias were not confirmed by myelography. Detailed clinical information of the patients is given in S3 Table.

Isolation of total RNA

Total RNA was extracted using the RNeasy Mini kit (QIAGEN, Valencia, CA, USA) according to the manufacture’s recommendations.

NanoString nCounter miRNA Expression Assay for miRNA profiling

The expression of 800 miRNAs in tumoral and normal thyroid and brain tissues was evaluated using the nCounter Human v2 miRNA Expression Assay Kit (NanoString Technologies, Seattle, WA, USA); and that in bone-marrow tissue, using the nCounter Human v3 miRNA Expression Assay Kit (NanoString Technologies, Inc., Seattle, WA, USA). These procedures were carried out in accordance with the manufacturer’s protocol. For the NanoString assay, 100 ng of total RNA was isolated from bone-marrow aspiration material. The RNA concentration was measured by a NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific, Inc., Waltham, MA, USA). According to NanoString recommendations, the 260/280 ratio should be not less than 1.9 and the 260/230 ratio, not less than 1.8, and so were our figures. The data were analyzed using the nSolver v4 package (NanoString Technologies, Inc., Seattle, WA, USA). The nCounter assay for each sample consisted of six positive controls, eight negative controls and five control mRNAs (ACTB, B2M, GAPDH, RPL19, RPLP0). Probes that recognize synthetic mRNA targets were included in the CodeSet at specified concentrations.

Statistical analysis

The following groups of samples were considered for analysis of miRNA expression: ‘thyroid gland’, ‘gliomas’, ‘bone marrow’, and ‘pooled’. In each group, the miRNAs that revealed less than 50 molecules in more than 50% of the samples were excluded from analysis. After normalization, the within-group variation of the binary logarithm of the normalized miRNA concentration was analyzed.

For each group, the following normalization strategies were considered: normalization using housekeeping genes; normalization using all miRNAs found; normalization using 75 most represented miRNAs; and normalization using the positive control. Normalized data were calculated as the difference between the binary logarithm of the registered number of miRNA molecules and the binary logarithm of the total number of normalizing molecules.

We used the following measures of variability for the normalized observation data: standard deviation, range (the difference between the maximum and the minimum), interquartile range (the difference between the third quartile and the first quartile) and the mean absolute deviation from the median. For each normalization strategy and each measure of variability, 10 least variable miRNAs are presented in S4–S7 Tables.

Additionally, we determined the most stable miRNAs according to all these measures of variability and two normalization strategies: one to housekeeping genes and one to the total miRNA content. Because we were not aware whether some of the measures of variability or some of the normalization strategies might be more important than some others, we considered them all equally important. To this end, we determined the rank of miRNAs for each measure of variability and each of these two normalization strategies: the least variable miRNAs were ranked 1, the next least variable miRNAs were ranked 2 and so on.

Additionally, for the ‘pooled’ group, we compared the mean of the logarithms of five least variable miRNAs based on the sum of the ranks and the logarithms of five next least variable miRNAs. (Recall that the mean of the logarithms is equal to the logarithm of the geometric mean.) The comparison was performed using the paired Grambsch test which tests difference between variances, and the paired Bonett-Seier test which tests difference between mean absolute deviations from median.

Next, we compared the results of the different normalization strategies within the ‘pooled’ group. We compared the ranking orders of five best miRNAs for each pair of normalization strategies. Five best miRNAs were selected as the miRNAs with the least variances. The comparison was performed using the following permutation test. For each pair of normalization strategies, five best miRNAs were put on a list. The list could have less than ten entries, if some of the five best miRNAs were indicated by both normalization strategies. Next, each miRNA was ranked (that is, was assigned its serial number in the sequence arranged in ascending order of variance) for each normalization strategy. The ranks of the miRNA in this list were compared between two normalization strategies using the permutation test. The p value was the probability of obtaining a permutation “as or more extreme” than the permutation corresponding to the difference between these two normalization strategies, assuming all the permutations are of equal probability. “As or more extreme” meant that the sum of the absolute differences between the ranks is not less than the observed sum of the absolute differences between the ranks for this pair of normalization strategies. (For each miRNA, the absolute difference of its ranks under two normalization strategies is found and then all the absolute differences so found are summed for all miRNAs in the considered sequence.)

Results

Table 1 presents ten most stable miRNAs in the samples: miR-361-3p, miR-151a-3p and miR-29b-3p in thyroid tissue; miR-140-5p, miR-148b-3p and miR-362-5p in bone marrow tissue; and miR-15a-5p, miR-194-5p and miR-532-5p in brain gliomas and surrounding normal tissue. Additionally, miR-423-5p, miR-28-5p and miR-532-5p are the most stable miRNAs in any of these tissues.

Table 1

MicroRNAs with the most stable expression in different tissues according to NanoString data.

All tumors		Thyroid gland		Bone marrow		Gliomas
Best variables by combined rank	Rank sum	Most stable miRNAs by combined rank	Rank sum	Most stable miRNAs by combined rank	Rank sum	Most stable miRNAs by combined rank	Rank sum
miR-423-5p	64	miR-361-3p	27	miR-140-5p	77	miR-15a-5p	125
miR-28-5p	65	miR-151a-3p	122	miR-148b-3p	112	miR-194-5p	153
miR-532-5p	66	miR-29b-3p	126	miR-362-5p	143	miR-532-5p	155
miR-362-5p	70	miR-425-5p	129	miR-191-5p	153	miR-500a-5p + miR-501-5p	173
miR-425-5p	72	miR-361-5p	135	miR-378i	154	miR-660-5p	193
miR-361-3p	110	let-7b-5p	137	miR-98-5p	174	miR-185-5p	201
miR-191-5p	110	miR-423-5p	145	miR-374b-5p	192	miR-99b-5p	234
miR-140-5p	114	miR-330-3p	154	miR-423-5p	194	let-7d-5p	235
let-7a-5p	125	miR-28-5p	154	miR-29b-3p	215	miR-19a-3p	249
miR-106b-5p	131	miR-301a-3p	169	miR-107	227	miR-140-3p	251

For all patients, we compared the mean of the logarithms of the concentrations of five least variable miRNAs based on the sum of the ranks and the mean of the logarithms of five next least variable miRNAs. The results of the comparison are presented in S4–S7 Tables. When we normalized miRNAs to housekeeping genes, p was less than 0.001 for both the paired Grambsch test and the paired Bonett-Seier test, which confirmed the assumption of different variances (the paired Grambsch test) and the assumption of different mean absolute deviations from the median (the paired Bonett-Seier test). However, when we normalized the same miRNAs to the total miRNA content, p was equal to 0.13 for the paired Grambsch test and 0.70 for the paired Bonett-Seier test.

We performed six pair-wise comparisons of four normalization strategies with the permutation test applied to five (for each group) microRNAs with least variances (for results, see Table 2). Each p-value is reported as adjusted p-value and unadjusted p-value. The Benjamini-Hochberg adjustment was used.

Table 2

A comparison of various normalization strategies.

Normalization to	Adjusted p-value (unadjusted p-value)
	Housekeeping genes	Total miRNA content	75 most highly expressed miRNAs	Positive control
Housekeeping genes	1	0.195 (0.13)	0.028 (0.014)	0.77 (0.77)
Total miRNA content	0.195 (0.13)	1	0.77 (0.77)	0.028 (0.014)
75 most highly expressed miRNAs	0.028 (0.014)	0.77 (0.77)	1	0.028 (0.014)
Positive control	0.77 (0.77)	0.028 (0.014)	0.028 (0.014)	1

Statistically significant differences are in bold (adjusted p < 0.05).

As we can see, the different normalization strategies lead to different ‘least variable’ miRNAs. The difference between the normalization to the total miRNA content and the normalization to 75 most highly expressed miRNAs failed to reach significance, as was expected.

Discussion

MicroRNAs are biomarkers, which show promise for diagnosis of tumors of various origins, assessment of antitumor therapy efficacy, survival prognosis and appear to be potential targets for new antitumor drugs [6]. The most frequently used technique for analysis of miRNA expression levels is real-time RT-PCR [21]. The main problem with qPCR-based protocols is one of the choice of an optimum reference gene.

To quantify miRNAs, researchers often use the NanoString system, which directly counts RNA or DNA molecules without amplification, thus preventing PCR-related errors [20]. In this work, we used NanoString to quantify miRNAs in three tissue types: tumoral and normal thyroid tissue; brain gliomas and adjacent morphologically normal tissue; and bone marrow tissue with tumoral and non-malignant bone marrow pathologies.

NanoString detect the expression of up to 800 miRNAs in a sample. Normally, only about 200 miRNAs are expressed at sufficiently high levels, with some of them varying between the samples and the others being invariant. The authors of all publications about NanoString-based miRNA quantification data are largely interested in finding the most significant differences in miRNA expression levels between the groups [22, 23]. By contrast, we wanted to find those miRNAs that have stable expression in all groups. These miRNAs could further be used as reference genes in an analysis of miRNA expression levels by real-time RT-PCR.

We found that ten most stable miRNAs in the thyroid gland are (in descending order) miR-361-3p, -151a-3p, -29b-3p, -425-5p, -361-5p, let-7b-5p, -423-5p -330-3p, -28-5p and -301a-3p. Therefore, any alone or their combinations may be used as a reference gene for analysis of miRNA expression levels in the thyroid gland by real-time RT-PCR.

In practice, analyses of thyroid tissue by qPCR most frequently involve U6, U44 or U48 as reference genes [24-26]. However, some authors prefer miRNAs. Mohamad Yusof et al. used miR-10b-5p and miR-191-5p [27]. Santos et al. proposed a combination of let-7a, miR-103, miR-125a-5p, let-7b, miR-145 and RNU48 for work with cytological specimens of thyroid tissue [28]. Titov et al. used the geometric mean of miR-197, -99a, -151a and -214 for work with cytological specimens of FNA-biopsied thyroid material and, to make their choice of reference genes, relied on NanoString-based miRNA counts [29]. As can be seen, only miRNA-151a appears in that and the present study. This could be due to the difference in the number of study specimens: Titov et al. used 12 and we used 36. Apparently, an increase in sample size may have led to a shift in statistically significant results. However, in a later work, Titov et al. used the geometric mean of miR-197-3p, -23a-3p, and -29b-3p for work with the same material [30]. In the present work, we found one them, miRNA-29b, to be the most adequate reference gene for analysis of miRNA expression levels in thyroid tissue.

We found that ten most stable miRNAs in brain tissue are (in descending order) miR-15a-5p, -194-5p, -532-5p, -500a-5p + -501-5p, -660-5p, -185-5p, -99b-5p, let-7d-5p, -19a-3p and -140-3p. Analysis of literature data showed that U6, U48 or a combination of several small nuclear RNAs are most frequently used as reference genes [31-34]. Thus, the results of this work allow a complex normalizer to be formed, consisting of the geometric mean of miRNAs stably expressed in brain tissue. The use of a complex normalizing factor as a reference gene will increase the accuracy of analysis of miRNA expression variation.

We found that ten most stable miRNAs in the bone marrow were (in descending order) miR-140-5p, -148b-3p, -362-5p, -191-5p, -378i, -98-5p, -374b-5p, -423-5p, -29b-3p and -107. In practice, the researchers tend to give more preference to the use of small nuclear RNAs as reference genes for qPCR in bone marrow tissues [35-37]. However, some authors are confident with miRNAs. Morenos et al. used a combination of miR-16 and miR-26b for work with archived bone marrow specimens [38]. Drobna et al. used a combination of miR-16-5p, -25-3p and let-7a-5p for work with bone marrow samples from patients with acute leukemia [39]. Costé et al. used miR-191-5p for analysis of miRNA expression levels in human multipotent stromal cells [40]. Kovynev et al. used a combination of miR-103a, -191 and -378 for work with bone marrow samples from AML and ALL patients [41]. In our previous work, when we were deciding on the reference gene for analysis of miRNA expression levels in bone marrow samples, we used NanoString nCounter and chose the geometric mean of the expression levels of miR-378 and miR-191. Analysis of miRNA quantification data obtained with NanoString nCounter showed that the abundance of these miRNAs in the samples was well above background values, but these miRNAs had the least variable expression across the groups [22].

In addition to the list of tissue-specific reference genes, we composed a list of miRNAs that can be used as reference genes, no matter which tissue type they were expressed in (10 most stable, in descending order): miR-423-5p, -28-5p, -532-5p, -362-5p, -425-5p, -361-3p, -191-5p, -140-5p, let-7a-5p and -106b-5p.

In this work, miRNA-423 has the least variable expression across the study tissue types, and so it is the best candidate for a universal reference gene. To date, several works with miRNA-423 used as a reference gene have been published. Costa-Pinheiro et al. opine that miR-423-5p is an optimal reference gene for analysis of prostate cancer tissue [42]. Yanokura et al. used miRNA-423-5p as a reference gene for analysis of endometrial cancer tissue [43]. Babion et al. used miRNA-423-5p for analysis of cervical cancer tissue [44].

Analysis of literature data shows small nuclear RNAs are most frequently used as reference genes; however, their expression stability varies. Masè et al. demonstrated that, in atrial tissue samples, the most stable reference gene was SNORD48 and the least stable reference gene was U6 [45]. However, miRNAs are now being used as reference genes increasingly frequently. Remarkably, some of the miRNAs that we have chosen had already been used as reference genes for analysis of miRNA expression levels in tumors of various origins. In particular, Peltier et al., who used geNorm and NormFinder for choosing an optimal reference gene, showed that miRNA-191 and miRNA-103 were the most stable RNAs no matter which tissue type or which storage conditions (frozen specimens/FFPE) [46]. In studies about miRNA expression in renal cell adenocarcinoma, miR-28, -103, -106a and RNU48 were found to be the most stably expressed genes. If it is possible to use a single gene as a reference gene, miR-28 is recommended; otherwise a combination of miR-28 and -103 or a combination of miR-28, -103 and -106a is preferable [47]. Shen et al. and Leitão Mda et al. used a combination of miR-23a and miR-191 as a reference gene for analysis of cervical cancer tissue [48, 49]. When working with lung cancer tissue, a combination of miR-26a, -140-5p, -195 and -30b can be an option [50]. Zhang et al. showed that it is possible to use combinations of miRNA-191 and -103 as a reference gene for analysis of lung cancer FFPE tissue samples [51]. Fochi et al.’ choice of reference gene in melanoma cells was miR-191-5p [52]. Zhu et al. proposed a combination of miR-103 and miR-191 for analysis of hepatocellular carcinoma FFPE samples [53]. Rohan et al. chose miR-191 and RNU6b as reference genes for analysis of cervical cancer FFPE samples [54]. Bignotti et al. used miRNA -191-5p for analysis of ovarian cancer samples [55]. Anauate et al. demonstrated that a combination of miR-101-3p and miR-140-3p was the best reference gene for analysis of stomach cancer samples [56]. As a reference gene, Jacobsen et al. use a combination of miR-24-3p, miR-151a-5p and miR-425-5p for analysis of hepatocytes [12].

Noteworthy, the miRNA most frequently used as a reference gene in work with tumors of various origins is miRNA-191. It is possible that this could is a promising universal reference gene for analysis of miRNA gene expression. However, about two hundred published works indicate that miRNA-191 is associated with various diseases [57]. This is a reminder that a particular reference gene -either as a single miRNA or, to keep the effect of the variable expression of the reference gene on the results to a minimum, as a combination of miRNAs—should be chosen for each experimental system.

To date, a large number of works attempting to find both tissue-specific and universal reference genes have been published, but the problem still persists. Research to uncover the role of miRNA in various diseases has been under way for 20 years. To be sure, much progress has been made in the development of qPCR data normalization strategies in an analysis of miRNA expression levels–from choosing suitable small nuclear RNAs to the development of mathematical approaches minimizing the interference of a reference gene with the end result. In the current work, we have identified the most stable miRNAs by a NanoString-enabled analysis of miRNA expression levels: miR-361-3p, -151a-3p and -29b-3p for thyroid tissue; miR-15a-5p, -194-5p and -532-5p for brain tissue; miR-140-5p, -148b-3p and -362-5p for bone marrow tissue; and miR-423-5p, -28-5p and -532-5p for any of the tissues. The study included very few tissue samples, but even so results of several comparisons were statistically significant. It is deemed logical to further validate these results on larger sample sizes using qPCR followed by analysis of the most stable miRNAs by GeNorm, NormFinder or Best Keeper.

Patient characteristics at the time of diagnosis (thyroid tissue).

(DOCX)

Click here for additional data file.

Patient characteristics at the time of diagnosis (brain tissue).

(DOCX)

Click here for additional data file.

Patient characteristics at the time of diagnosis (bone marrow tissue).

(DOCX)

Click here for additional data file.

MicroRNAs with the least variable expression in all groups of patients.

The names of the microRNA’s together with the observed values of deviations are reported.

(DOCX)

Click here for additional data file.

MicroRNAs with the least variable expression in thyroid tissue.

The names of the miRNAs together with the observed values of deviations are reported. If the names of two miRNAs are linked with the plus sign “+”, then only the total content of these two miRNAs was measured.

(DOCX)

Click here for additional data file.

MicroRNAs with the least variable expression in bone marrow tissue.

The names of the miRNAs together with the observed values of deviations are reported. If the names of two miRNAs are linked with the plus sign “+”, then only the total content of these two miRNAs was measured.

(DOCX)

Click here for additional data file.

MicroRNAs with the least variable expression in brain tissue.

The names of the miRNAs together with the observed values of deviations are reported. If the names of two miRNAs are linked with the plus sign “+”, then only the total content of these two miRNAs was measured.

(DOCX)

Click here for additional data file.

15 Oct 2021

Submitted filename: Review.docx

Click here for additional data file.

15 Dec 2021

We thank the Reviewers for their insightful comments. Below are our answers to the Reviewers’ comments shown in bold.

Reviewer #1

1. On the page 5, lines 116-117, the authors wrote: “the control group consisted of patients with non-cancerous blood diseases”. What was the reason to use these patients as the control group? Because different diseases, even non-cancerous, can significantly change miRNAs expression profile in patients’ tissues in compare with healthy ones.

We have added the following explanation on p.5

“Work with healthy bone marrow donors for allogeneic transplantation is beyond the competence of our clinic. Taken together, we decided to go with a control group composed of people who had no hematologic cancer, but had indications for bone marrow examination to exclude one. They were people with secondary anemic and cytopenic conditions, in whom leukemias were not confirmed by myelography.”

2. Would the authors kindly provide a list of housekeeping genes and positive controls, used for normalization?

To normalize the data, nCounter technology used:

1) Positive Controls (n=6). Probes that recognize synthetic mRNA targets included in the CodeSet at specified concentrations (targets do not require ligation). Positive controls used by the QC metrics in nSolver to confirm linear response to input amounts, and confirm that low input signal is above background.

2) Housekeeping genes: ACTB, B2M, GAPDH, RPL19, RPLP0.

We have added the following explanation on p.6

3. Let-7a, -7b and -7d were listed in Table 1 among a least variable miRNAs in thyroid gland, brain and as universal controls correspondently. But, miRNAs from let-7 family are well known tumor suppressors. Expression of these miRs is downregulated in tumors. How can it be possible that these miRNAs are the most stable?

Analysis of the literature revealed that expression levels of members of the Let-7 family were found significantly different between tumor and normal tissue samples in several publications. The value of the difference varies both over different tissues and within individual tissues. The following ratios of concentrations were reported: Thyroid tumors (cancer sample versus normal): 0.61 (hsa-let-7d-5p) (Swierniak et al,2013); 1.21 (has-let-7f-1) (Pallante et al,2006); approximately 1.5 (has-let-7d) and approximately 1 (has-let-7g) (Braun et al,2010); approximately 3 (has-let-7a) (Zhou et al, 2017). Glioma (cancer sample versus normal): 1.69 (has-let-7b) (Zhang et al,2019), approximately 2 (has-let-7d and has-let-7a) (Yang et al). Bone marrow (cancer sample versus normal): (1.35-3.78) (Veryaskina et al,2020). We see that, in most of the cases, the difference between the expression levels was about 2 or less. In most of the studies, no correction for multiple comparisons was employed, which could lead to chance findings. Besides, generally, authors tend to mention more frequently miRNAs with significant differences than those for which comparisons yielded negative results. Therefore, it seems impossible to evaluate the number of studies where the differences for the Let-7 family miRNAs were not found significant.

On the other hand, only a small number of samples were used in our work. Therefore, the results are to be validated in succeeding studies.

It is also worth noting that let-7a, let-7b, and let-7d were not the most stable normalizers in our study.

4. There is no confidence interval for p-values in the Table 2. Significant p-values in this table may be shown in bold or color.

We might misunderstand the comment. As far as we know, confidence intervals for p-values were suggested only in some theoretical constructs and are not used in practice. Probably, the Reviewer suggested considering confidence intervals for the differences corresponding to the p-values. For Table 2, permutation tests were used, the differences being the sums of absolute differences of the corresponding ranks, which is described in the Statistical analysis section. As far as we know, there is no straightforward way to introduce confidence intervals for differences in permutations tests. Therefore, we did not present confidence intervals in Table 2.

The significant p-values are now shown in bold in Table 2.

5. MiR-28-5p and miR-532-5p were chosen like the universal controls for all three tissues. But, among 10 least variable miRNAs, miR-28-5p is only ninth in thyroid gland and even was not listed in Table 1 for brain and bone marrow. The same story is for miR-532-5p. This miRNA is stable in brain tissues only. What is the possible reason for these discrepancies?

For a miRNA, the variability in all tumors depends on within-group variability in each group and between-the-groups variability of the mean level of concentration of the miRNA. A miRNA may have low within-group variability in the groups, but high between-the-groups variability of the mean level of concentration; or vice versa. Therefore, the order of the miRNAs for all tumors may not correlate with the orders for individual tumors.

Reviewer #2

The only thing that I would suggest is to correct the title into: Selection of reference genes for quantitative analysis of microRNA expression in three different types of cancers or even more specific in three types of cancers with thyroid, brain and bone marrow origin.

According to the Reviewer’s suggestion, the title is corrected into: Selection of reference genes for quantitative analysis of microRNA expression in three different types of cancer.

Submitted filename: Response to Reviewers.doc

Click here for additional data file.

31 Jan 2022

Selection of reference genes for quantitative analysis of microRNA expression in three different types of cancer.

PONE-D-21-19542R1

Dear Dr. Veryaskina:

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Gayle E. Woloschak, PhD

Section Editor

PLOS ONE

Additional Editor Comments (optional):

Thank you for addressing the concerns raised by the reviewers.

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: (No Response)

Reviewer #2: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: I Don't Know

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: (No Response)

Reviewer #2: (No Response)

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

8 Feb 2022

PONE-D-21-19542R1

Selection of reference genes for quantitative analysis of microRNA expression in three different types of cancer.

Dear Dr. Veryaskina:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Gayle E. Woloschak

Section Editor

PLOS ONE