Impact of ambient temperature exposure on miRNA stability in human plasma

Abstract

MicroRNAs are considered more stable than mRNA, but the impact of progressive thawing of biological samples after freezing as may happen during shipping delays has not been quantified. To address this, we utilized digital PCR to estimate the absolute concentrations of select microRNAs following progressive thawing of human plasma and maintenance at ambient temperature. Specifically, we quantified let-7b-3p, miR-144–5p, miR-150–5p, miR-517a-3p, miR-524–5p, and miR-1283, which have varying abundance in plasma. We observed a trend indicating a decline in microRNA concentration as plasma samples were progressively thawed. Notably, miR-150–5p and miR-517a-3p were the least stable and were degraded by 32% and 52% respectively after 24 hours of ambient temperature storage. We found that the variation in sensitivity to temperature was not due to the GC content of the microRNAs nor their initial abundance, suggesting that other factors, such as protein interactors and vesicles carrying these microRNAs, may impact sensitivity.

Methods summary

Plasma samples from nine pregnant women from the Gen3G cohort which were stored in −80°C for 12 years, were used to measure microRNA stability due to progressive thawing often encountered during shipping and experimental handling. We pooled samples from different participants, in addition to two spike-in controls to reduce technical variability.

Introduction

MicroRNAs (miRNAs) are short non-coding RNA molecules, typically 19–25 nucleotides long [1]. They bind to the 3’ untranslated region of target messenger RNAs (mRNAs), resulting in the downregulation of protein synthesis. This occurs either by reducing the translation of the mRNAs or by decreasing their stability [2]. MiRNAs are expressed in all cell types [3] and can be found in a variety of bodily fluids, such as urine, saliva, cerebrospinal fluids, plasma, and serum [4,5]. Altered miRNA profiles in these biofluids have been linked to several pathologies, including different types of cancer [6,7], cardiovascular diseases [8], and obesity [9]. While miRNAs are emerging as a promising novel class of biomarkers, their pre-analytical and analytical stability must be characterized before they can be utilized in clinical settings.

Only a limited number of studies have examined the stability of miRNAs in blood, plasma, or serum samples, an ideal biofluid for biomarker analysis application. Different conditions have been tested, including extreme pH changes [10], freeze-thaw cycles [4,10–13], and physical disturbances, with miRNAs generally maintaining great stability. Extended storage at 4°C, −20°C, and −80°C has also been evaluated, showing no significant differences in miRNA abundance between samples stored at −20°C and −80°C. However, abundance decreased when samples were stored at 4°C for 24 hours or more [11,12,14,15]. Few studies have investigated the effect of ambient temperature storage of plasma and serum, which can occur between blood collection and subsequent analysis, or during transport between laboratories. Nearly all of these studies reported that the few miRNAs tested remain stable for up to 24 hours in fresh plasma and serum samples, but their abundance started to decline with extended storage at ambient temperature [4,13,15]. In contrast, one study found a significant decrease in miRNA concentration in fresh plasma and serum samples within 24 hours of being stored at ambient temperature [12].

Overall, these current studies have tested only a limited number of miRNAs and reported large variability in their response to storage conditions [11,14,16]. Additionally, they analyzed fresh samples which are less representative of biobanked samples generally used in research. Furthermore, these studies used reverse-transcription quantitative polymerase chain reaction (RT-qPCR) to assess changes in miRNA levels, which provides only relative concentrations making it less precise and difficult to replicate [17].

Recently, plasma samples which were retrieved from our biobank for miRNA analyses, were delayed at customs between the U.S. and Canada. Despite being packed with a large quantity of dry ice, they arrived thawed at destination. We estimate that our samples remained at ambient temperature for about 48 hours, leading us to hypothesize that miRNA abundance had decreased in these plasma samples. In this study, we aimed to test this hypothesis and assess whether miRNA concentration in our samples was affected by the delay at customs. To achieve this, we measured the absolute concentration (miRNA copies/μL of plasma) of 6 candidate miRNAs (i.e., let-7b-3p, miR-144–5p, miR-150–5p, miR-517a-3p, miR-524–5p, and miR-1283) in plasma at four time points using digital PCR (dPCR). Our results suggest that progressive thawing of plasma samples is associated with a trend towards decreased miRNA concentration, independent of initial miRNA abundance, or sequence.

Materials and Methods

Participant selection

For this study, we used plasma samples from the Genetics of Glucose regulation on Gestational and Growth (Gen3G) cohort biobank. Gen3G is a prospective pregnancy and early life cohort composed of mother-child dyads, with pregnant women initially recruited between 2010 and 2013 in the Sherbrooke area of Québec, Canada [18]. Leveraging the Gen3G cohort biobank, we selected 9 women for whom 2.0 mL of plasma was collected at the first trimester of pregnancy (visit 1: between the 5^th and 16^th week of pregnancy) between 2010 and 2013. These samples were stored at −80°C for ~12 years and had never been thawed before this study (plasma was unused due to lack of follow-up). Each woman consented, in a free and informed manner, to participate in the study. This project was approved by the ethics committee of the Centre intégré universitaire de santé et de services sociaux de l’Estrie - Centre hospitalier universitaire de Sherbrooke (CIUSSS de l’Estrie - CHUS), Ethics approval number : 2010–198 - 07–027-A1.

Blood sampling

Blood samples were collected in tubes containing EDTA during the first trimester visit, one hour after a 50 g glucose challenge test. The samples were then centrifuged within 30 minutes at 2500 g for 10 minutes at 4°C to isolate plasma, following a standardized procedure as previously described [18]. Plasma was aliquoted into 500 μL samples and stored at −80°C in a monitored freezer in the Gen3G biobank.

Experimental design

The experimental design is illustrated in Fig 1. Briefly, four aliquots of 500 μL of plasma (totaling 2.0 mL) per participant were used. One aliquot remained at −80°C until RNA extraction as an unthawed control while the other three aliquots were slowly thawed in a cooling microtube holder (Ultident Scientific, catalog #87-W24-MCT) placed in room temperature (~20°C). The 9 participants were randomized and pooled into three groups (3 participants x 3 aliquots per group).. Each pool was then aliquoted in nine replicates of 450 μL of plasma. For each pool, three replicates were left at ambient temperature (20°C) for 24 hours (n=9), three replicates were left at ambient temperature for 48 hours (n=9), and three replicates were left at ambient temperature for 72 hours (n=9), before storing again at −80°C for two weeks for RNA extraction. For the unthawed controls (n=9), plasma samples were thawed, mixed into three pools as described above, and divided into three replicates per pool immediately prior to RNA extraction.

Figure 1: Experimental design.

For each of the nine participants selected from the Gen3G cohort biobank, three aliquots of 500 μL of plasma were slowly thawed on ice, pooled in three pools of three participants (randomly grouped) and aliquoted (nine aliquots of 450 μL of plasma per pool). Three aliquots per pool were left at ambient temperature (20°C) for 24 hours (n=9), 48 hours (n=9) or 72 hours (n=9) and were then stored at −80°C until RNA extraction (exposure; yellow box). For each participant, one aliquot of 500 μL of plasma remained frozen at −80°C until RNA extraction (n=9; unthawed controls; green box), prior to which they were pooled and aliquoted similar to the exposure groups. Total RNA was extracted from the plasma samples (n=36) in a random order, and RNA samples were randomized again before microRNA quantification by digital PCR. Plasma samples are represented by dark yellow drops. Abbreviations: Gen3G: Genetics of Glucose regulation on Gestational and Growth prospective pregnancy and early life cohort; P: Pool; V1: Visit 1, between the 5^th and 16^th week of pregnancy.

RNA extraction

Plasma samples (n=36), without apparent hemolysis, were randomized before RNA extraction. Total RNA was extracted from 450 μL of plasma using the miRVana PARIS kit (ThermoFisher Scientific, catalog #AM1556), following the manufacturer’s standard procedure. As an external control for data normalization, 1.8 μL of diluted (8.0 x 10⁶ copies/μl) miRNeasy Serum/Plasma Caenorhabditis elegans cel-miR-39–3p mimic Spike-In Control (Qiagen, catalog # 219610) was added to each sample at the beginning of the extraction procedure. RNA was eluted in 75 μL of preheated nuclease-free water and stored at −80°C until dPCR analyses.

Candidate microRNAs selection

To evaluate the miRNAs stability in plasma at ambient temperature, we selected six miRNAs for which quantification conditions were recently optimized using dPCR in ongoing studies from our group. The selected miRNAs were let-7b-3p, miR144–5p, miR-150–5p, miR-517a-3p, miR-524–5p, and miR-1283. We selected these miRNAs because of their wide range of abundance in plasma (Table 1). Let-7b-3p, miR-144–5p and miR-150–5p are relatively abundant in circulation, while miR-517a-3p, miR-524–5p and miR-1283, are less abundant in circulation. These miRNAs have biological relevance to various diseases including metabolic and pregnancy conditions. Let-7b-3p regulates the MAPK/ERK pathway in lung cancer [19], while macrophage derived miR-144–5p plays a key role in bone healing especially in type 2 diabetes mellitus [20]. Furthermore, miR-150–5p has been studied as a potential biomarker for the detection of advanced heart failure [21], as well as autoimmune complications [22]. On the other hand, miR-517a-3p, miR-524–5p, and miR-1283 are all members of the chromosome 19 miRNA cluster, whose expression in the placenta is relatively low in early pregnancy [23]. MiR-517a-3p plays a role in glucose uptake in skeletal muscles and has been shown to be correlated with insulin resistance in pregnant women [24]. MiR-524–5p levels were significantly reduced in placental tissue analyzed from patients suffering from preeclampsia compared to controls [25]. MiR-1283 has also been shown to play a role in hypertension in an in vitro cell system, where its levels were positively correlated with hypertension [26]. As an exogenous control, we also quantified cel-miR-39–3p for data normalization. Furthermore, we included the UniSp6 spike-in control to evaluate reverse transcription (RT) reaction efficiency, as recommended by Qiagen.

Table 1: Biological and therapeutic relevance of the selected miRNAs.

The biological and disease relevance of the six miRNAs measured in our study following exposure to ambient temperature for varying times.

microRNAs	Abundance in Circulation	Biological Relevance	Disease	Reference number
Let-7b-3p	Relatively abundant	Regulates MAPK/ERK pathway	Lung cancer	(19)
miR-144–5p	Relatively abundant	Bone healing in type 2 diabetes mellitus	Type 2 diabetes mellitus, bone healing	(20)
miR-150–5p	Relatively abundant	Potential biomarker	Heart failure, autoimmune complications	(21,22)
miR-517a-3p	Less abundant	Involved in glucose uptake in skeletal muscles	Insulin resistance in pregnancy	(23,24)
miR-524–5p	Less abundant	Reduced expression in preeclamptic placenta	Preeclampsia, pregnancy complications	(25)
miR-1283	Less abundant	Role in hypertension	Hypertension	(26)

Digital PCR quantification

After randomization, RNA samples were reverse transcribed into complementary DNA (cDNA) using the miRCURY LNA RT Kit (Qiagen, catalog # 339340), following the manufacturer’s procedure. A 20 μL reaction was made containing 4.0 μl of 5x miRCURY RT SYBR Green Reaction Buffer, 2.2 μL of nuclease-free water, 1.0 μl of diluted (1:100) UniSp6 RNA Spike-in Control, 2.0 μl of 10x miRCURY RT Enzyme, and 10.8 μl of total RNA (equivalent to 64.8 μL of original plasma). To avoid multiple freeze-thaw cycles, cDNA samples were aliquoted and stored at −20°C until miRNA quantification.

Selected miRNAs and spike-in controls were quantified by dPCR using the QIAcuity EG PCR Kit (Qiagen, catalog # 250112) and miRCury LNA miRNA PCR Assays (Qiagen, catalog # 339306; assays ID: YP00203952 (cel-miR-39–3p), YP00205653 (hsa-let-7b-3p), YP00204670 (hsa-miR-144–5p), YP00204660 (hsa-miR-150–5p), YP00206019 (hsa-miR-517a-3p), YP00204135 (hsa-miR-524–5p), YP02104470 (hsa-miR-1283) and YP00203954 (UniSp6)). Immediately before use, cDNA samples were thawed, diluted 1:5 in nuclease-free water, and 3.0 μl of this diluted cDNA was used in a 12 μl dPCR reaction, which also contained 4.0 μl of 3x EvaGreen PCR Master Mix, 1.2 μl of 10x miRCury LNA miRNA PCR Assay, and 3.8 μl of nuclease-free water. For each reaction, 11.5 μl were loaded into a QIAcuity 8.5k partitions 96-well Nanoplate (Qiagen, catalog # 250021) and analyzed using a QIAcuity ONE 5 Plex Digital PCR System (Qiagen, catalog # 911022). Samples were amplified for 50 cycles and imaged twice (after 40 cycles and 50 cycles) using the green channel with an exposure of 200 milliseconds. The specific amplification and imaging conditions for each miRNA are outlined in Supplementary Table 1. Quantifications were done in duplicate and combined as hyperwell (8.5k partitions x 2 wells = 17k partitions available for quantification).

Data normalization and quality control

To estimate and reduce the technical variability between samples, particularly in RNA extraction, RT and dPCR efficiency, we added an exogenous spike-in control, cel-miR-39–3p. This control was added at the beginning of the RNA extraction procedure and measured by dPCR. MiRNA quantification data were then normalized as follows: candidate miRNA concentration in plasma (copies/μL) / cel-miR-39–3p concentration in plasma (copies/μL) x 10,000 (arbitrary factor added to avoid very low values). Normalized data are expressed in relative units (RU). Inter-sample variability in RT reaction efficiency was also evaluated by measuring the UniSp6 RNA spike-in control added to the RT reaction mix.

Statistical analyses

MiRNA concentration in plasma (copies/μL) measured by dPCR was calculated as follows, as recommended by Qiagen. Concentrations are represented by brackets and expressed in copies/μL and volumes are expressed in μL:

$$\left[\mathit{\text{miRNA}}\right]\mathit{\text{plasma}}=\left[\mathit{\text{miRNA}}\right]\mathit{\text{RTmix}}\times \left(\frac{\mathit{\text{Volume RT reaction}}}{\mathit{\text{Volume plasma equivalent}}}\right)$$

Where:

$$\left[\mathit{\text{miRNA}}\right]\mathit{\text{RTmix}}=\left[\mathit{\text{miRNA}}\right]\mathit{\text{dPCRmix}}\times \left(\frac{\mathit{\text{Volume dPCR reaction}}}{\mathit{\text{Volume diluted cDNA in dPCR reaction}}}\right)\times \mathit{\text{cDNA Dilution factor}}$$

and:

$$\mathit{\text{Volume plasma equivalent}}=\mathit{\text{Volume RNA in RT reaction}}\times \left(\frac{\mathit{\text{Volume plasma for extraction}}}{\mathit{\text{RNA elution Volume}}}\right)$$

Coefficients of variation (CV; %) in spike-in concentrations between samples were calculated as standard deviation / mean x 100. The normality of data distributions was assessed using the Shapiro-Wilk test. When normality could not be achieved through logarithmic or square-root transformation of data for a variable, non-parametric tests were applied for this variable. Homogeneity of variances was evaluated using the Levene test. Difference in cel-miR-39–3p concentration in plasma between RNA extraction batches (n=3) was assessed using Analysis of Variance, ANOVA, followed by post-hoc pairwise T-test with Bonferroni adjustment for multiple testing. Spearman correlation was also applied to evaluate the association between the concentration of cel-miR-39–3p in plasma and the sample order of elution during the extraction procedure. Difference in cel-miR-39–3p concentration in plasma between dPCR batches (n= 6 plates) was assessed using a Kruskal-Wallis test. Differences in miRNA concentration following plasma exposure to ambient temperature for 0 h (unthawed control), 24 h, 48 h and 72 h was assessed by a Friedman test. Rates of change (%) in miRNA normalized concentration in plasma exposed to ambient temperature, compared to that in the unthawed control, were calculated using the following: ((median normalized miRNA concentration in plasma left at ambient temperature - median normalized miRNA concentration in unthawed plasma) / median normalized miRNA concentration in unthawed plasma) x 100. Results were considered statistically significant at p ≤ 0.05. Boxplots were generated using the ggplot2 R package version 3.4.4 [27] and ggpubr package version 0.6.0 [28]. Statistical analyses were carried out using R version 4.3.2 [29] in R Studio version 2023.09.1+494 [30].

% GC calculation

% GC was calculated using OligoCalc version 3.27. MiRNA sequences were obtained from miRbase [31].

Results and Discussion

Study participants and sample storage conditions

We used plasma samples collected in nine women initially enrolled in the Gen3G cohort, all of which are European decent (100%), aged 28.9 ± 5.0 [21 – 36] years at enrollment, and the gestational age at blood sampling (Visit 1) was 8.6 ± 1.8 [5.7 – 11.0] weeks. Plasma samples were stored in the Gen3G biobank for ~12 years at −80°C prior to usage. Samples were then pooled, aliquoted, and stored at ambient temperature for 0, 24, 48, and 72 hours. Following ambient temperature storage, plasma was stored in −80°C until RNA extraction. Total RNA was extracted from plasma samples and candidate miRNAs were quantified by dPCR.

Absolute quantification of candidate microRNAs in plasma

Digital PCR is a technology allowing robust absolute quantification (copies/μL) of target molecules, including miRNAs, without the need of a standard curve [32]. Leveraging this technology, we quantified six candidate miRNAs in plasma samples which were either unthawed (controls) or thawed and left at ambient temperature for up to 72 h. Out of 17,000 available partitions (hyperwell mode), we observed an average of 16,498 ± 191 valid partitions per sample, representing 97.0 ± 1.1% of the available partitions used for quantification. Absolute concentration of the candidate miRNAs in plasma at different time points are presented in Table 2. The concentrations ranged from 5.10 ± 3.81 copies/μL of plasma for miR-517a-3p to 751.68 ± 153.75 copies/μL of plasma for miR-150–5p in control samples that had not been thawed. Of note, the Poisson confidence intervals at 95% were relatively low for the most concentrated miRNAs (e.g., 9.3 ± 1.2% on average for miR-150–5p). In contrast, these intervals were high and more variable for the less concentrated miRNAs (e.g., 82.9 ± 57.3% on average for miR-517a-3p), consistent with what is expected for targets that are low in concentration [33]. The mean number of valid partitions as well as the Poisson confidence intervals at 95% for each miRNA and spike-in control are presented in Supplementary Table 2. Overall, the data suggest a trend where progressive thawing of plasma samples at ambient temperature may lead to a decline of the absolute concentrations of miRNA (Table 2).

Table 2: Absolute concentration of candidate miRNA in plasma.

The concentrations of the six miRNAs measured in our study following exposure to ambient temperature for varying times.

microRNAs		microRNA concentration in plasma (copies/μL) following exposure at ambient temperature
		0 h (unthawed controls)	24 h	48 h	72 h
let-7b-3p	mean ± SD	30.3 ± 7.9	32.1 ± 8.3	27.2 ± 9.0	23.5 ± 7.2
	range	16.4 – 44.7	20.3 – 41.8	18.7 – 48.8	13.2 – 36.7
	median (IQR)	31.1 (30.1 – 33.2)	35.5 (23.4 – 37.5)	25.8 (22.0 – 28.7)	23.6 (19.5 – 28.3)
miR-144–5p	mean ± SD	18.7 ± 12.3	17.0 ± 8.0	12.7 ± 6.2	10.9 ± 6.6
	range	2.5 – 38.5	7.4 – 29.4	5.0 – 24.4	2.5 – 20.9
	median (IQR)	15.1 (13.3 – 22.4)	12.5 (12.1 – 23.2)	11.2 (7.3 – 17.4)	9.8 (5.2 – 17.4)
miR-150–5p	mean ± SD	751.7 ± 153.8	607.3 ± 93.5	532.0 ± 44.3	453.4 ± 116.5
	range	554.6 – 995.7	423.4 – 734.6	460.6 – 603.9	311.9 – 698.8
	median (IQR)	750.6 (629.6 – 868.5)	627.2 (565.3 – 667.9)	543.3 (490.7 – 557.3)	457.8 (374.6 – 461.5)
miR-517a-3p	mean ± SD	5.1 ± 3.8	3.0 ± 3.6	1.4 ± 1.6	1.5 ± 1.5
	range	1.2 – 9.9	0.0 – 11.0	0.0 – 3.7	0.0 – 3.7
	median (IQR)	2.4 (2.4 – 8.6)	2.4 (0.0 – 3.7)	1.2 (0.0 – 2.4)	1.2 (0.0 – 2.4)
miR-524–5p	mean ± SD	14.8 ± 11.4	15.0 ± 9.4	13.9 ± 9.0	8.1 ± 4.9
	range	1.2 – 35.2	3.7 – 28.2	1.2 – 30.3	1.2 – 17.5
	median (IQR)	15.5 (3.7 – 22.5)	12.3 (6.3 – 24.1)	11.3 (9.3 – 15.5)	8.5 (3.6 – 9.9)
miR-1283	mean ± SD	7.5 ± 5.9	11.7 ± 8.8	8.3 ± 6.0	5.1 ± 3.9
	range	1.2 – 19.0	0.0 – 21.6	1.2 – 20.2	1.2 – 10.8
	median (IQR)	5.8 (3.5 – 12.0)	9.5 (2.4 – 19.3)	8.4 (3.5 – 9.6)	3.6 (2.4 – 8.4)

Normalization with exogenous control reduces technical variability between samples

To evaluate the inter-sample variability in RNA extraction and RT reaction efficiency, two spike-in controls were added to the plasma samples. Cel-miR-39–3p and UniSp6 RNA control were added prior to RNA extraction and RT reaction, respectively, followed by quantification by dPCR. Supplementary Fig 1 shows inter-sample variability in cel-miR-39–3p concentration in plasma and in UniSp6 concentration in the RT reaction mix. On average, 3,247.96 ± 943.18 copies/μL of plasma were detected for cel-miR-39–3p, with a CV of 29.0% across samples, illustrating a relatively high inter-sample variability (Supplementary Fig 1A). This variability partly came from RNA extraction batch effect, as we observed a significant difference in cel-miR-39–3p mean concentration between the three batches of twelve samples extracted (F= 4.931; p= 0.013; Supplementary Fig 2A). Interestingly, we also observed a negative correlation between the concentration of cel-miR-39–3p and the sample order of elution during the RNA extraction procedure (r_s= −0.522; p= 0.001; Supplementary Fig 2B), also suggesting an intra-batch variability. The cel-miR-39–3p concentration was not significantly different between the dPCR batches (n= 6 plates; Kruskal-Wallis χ2= 2.462; p= 0.782; Supplementary Fig 3). On the other hand, RT efficiency is considered homogeneous between samples, with a mean of 5,200.39 ± 269.82 copies/μL detected for UniSp6 in the RT reaction mix and a CV of 5.2% (Supplementary Fig 1B). To reduce the impact of technical variability mainly caused by heterogeneity in RNA extraction efficiency and to ensure accurate comparison of miRNA concentrations across time points, we normalized miRNA quantification data for cel-miR-39–3p concentration.

Stability of candidate miRNAs in plasma left at ambient temperature

To assess candidate miRNA stability in plasma at ambient temperature, samples were left at ambient temperature for 0 h (control; n=9), 24 h (n=9), 48 h (n=9) or 72 h (n=9), and their miRNA concentrations were compared, using data normalized for cel-miR-39–3p. Although the differences between conditions were not statistically significant (Friedman test: 0.733 ≤ χ² ≤ 7.551; 0.056 ≤ p ≤ 0.865), we observed a trend toward a reduction in miRNA abundance in plasma left at ambient temperature for 24 h to 72 h compared to unthawed plasma (Fig 2). Moreover, this effect appears likely to be miRNA-dependent (Fig 3); with certain miRNAs appearing to be more resistant to degradation (e.g., let-7b-3p, median reduction of 18% of its concentration in plasma after 72 h at ambient temperature) and others being affected in a more pronounced and early manner (e.g., miR-517a-3p, median reduction of 52% of its concentration in plasma after 24 h at ambient temperature).

Figure 2: Variation in microRNA concentrations in plasma exposed to ambient temperature during 24, 48 or 72 hours.

Boxplots showing let-7b-3p (A), miR-144–5p (B), miR-150–5p (C), miR-517a-3p (D), miR-524–5p (E) and miR-1283 (F) concentrations in plasma, normalized for cel-miR-39–3p, for unthawed control (0 h) plasma (n=9) and plasma left for 24 h (n=9), 48 h (n=9) or 72 h (n=9) at ambient temperature. Differences in miRNA concentration measured by digital PCR between time points was assessed by a Friedman test and considered statistical significance at p ≤0.05. Rate of change in miRNA median normalized concentration following plasma exposure to ambient temperature compared to unthawed control plasma is expressed in percentage for each miRNA and time point. Abbreviations: CTRL: Unthawed control plasma.

Figure 3: Candidate microRNAs stability in plasma exposed to ambient temperature during 24, 48 or 72 hours relative to unthawed control.

Graph showing the variation in median miRNA concentrations in plasma, normalized for cel-miR-39–3p, after incubation for 24 h, 48 h or 72 h at ambient temperature. Relative median miRNA concentrations are expressed as percentage, with concentration of unthawed control (0 h; baseline) set to 100%.

Percent GC content of candidate miRNAs

Previous studies have suggested that a higher % GC content positively correlates with miRNA stability [34]. To this end, we calculated the % GC content of each mature miRNA and found no clear relationship between % GC content and stability. The most stable miRNAs, let-7b-3p and miR-1283, had a % GC content of 50 and 41 % respectively. The rest of the miRNAs, which were less resistant to ambient temperature storage, had a % GC content of 32–55%. Notably, miR-144–5p had the lowest % GC content at 32% (Supplementary Table 3). Although GC content was calculated for each miRNA, no clear correlation was observed between the stability of miRNAs in our sample set and GC content. Based on this observation, we therefore conclude that GC content alone is not predictive of miRNA stability tested in our sample set.

In this study, we examined how progressive thawing and ambient temperature storage affects absolute plasma miRNA concentrations measured by dPCR. Recently, we experienced unanticipated delays in plasma sample transportation, raising concerns about the quality of samples that spent about 48 hours at ambient temperature. To determine whether prolonged exposure to ambient temperature affects miRNA concentration in previously frozen plasma, we replicated those conditions using stored plasma samples from the Gen3G biobank (Fig 1). We measured the absolute concentration of selected miRNAs in plasma samples using dPCR and observed a trend suggesting that some specific miRNA concentrations decrease with prolonged exposure to ambient temperature (Fig 2, Fig 3). This is partially in contrast with a study demonstrating remarkable stability of miRNAs in freshly obtained plasma samples maintained at ambient temperature for 24 hours [13], as two of the miRNAs we quantified (miR-150–5p and miR-517a-3p) showed degradation after 24 hours. However, our data is in accordance with other studies showing a reduction in miRNA concentration measured by RT-qPCR in fresh serum and plasma stored at ambient temperature for 24 to 96 hours [12,15]. The discrepancies between those studies and ours may partly be due to the different technologies used to quantify the miRNAs (RT-qPCR vs dPCR), and the condition of the original biological sample (e.g., plasma vs serum, fresh vs frozen, time since collection). Although the stability of miRNA was generally reduced by prolonged exposure to ambient temperature, some miRNAs, such as let-7b-3p, are more stable even after 72 hours, while others, such as miR-517a-3p, are less stable (Fig 3). This indicates that other factors, such as abundance, sequence, post-transcriptional modification, and miRNA packaging may have an effect on miRNA stability.

We investigated whether the absolute miRNA abundance in plasma could explain the observed differences in stability. Comparing the more abundant miRNAs, let-7b-3p, miR-144–5p, and miR-150–5p (relative abundance at 0 h: 90–2500 RU; Fig 2 A–C) to the less abundant miRNAs, miR-517a-3p, miR-524–5p, and miR-1283 (relative abundance at 0 h: 15–35 RU; (Fig 2 D–F), we found no correlation between miRNA abundance and stability following ambient temperature storage for 72 hours. This is in contrast with a report that hypothesized that cellular concentration of miRNA impacts miRNA structure which can affect their stability [35]. This may still be true as our experiments measured miRNA levels in plasma, not cells. Additionally, we investigated the potential impact of miRNA sequence on its stability. We measured % GC content of each miRNA and saw no clear correlation between % GC content and miRNA stability at ambient temperature (Supplementary Table 3). Factors such as post-transcriptional modifications may also affect miRNA stability. Adenylation of the 3’ end of miRNA acts as a protective modification, increasing the stability of miRNA [36]. However, the quantification method we used, dPCR, did not allow us to test this hypothesis.

The packaging of miRNA and their interactions may also influence their stability in plasma. A key question in the field is how miRNAs are transported through the bloodstream without being degraded by RNAses and other RNA-degrading factors. A widely accepted idea is that miRNAs are packaged into extracellular vesicles (EVs), including exosomes, which protect them in their role as intercellular communicators [37]. One study demonstrated that vesicle-associated miRNAs are more stable and resistant to RNase A, supporting the idea that vesicles help shield miRNA from RNase degradation [38]. Furthermore, other studies have reported that lipoproteins, including high-density lipoproteins (HDL), can transport miRNAs in plasma and deliver them to recipient cells, where they have regulatory activities [39,40]. To our knowledge, only one study evaluated the stability of HDL-bound miRNAs and demonstrated that RNAse A treatment, or serum storage at ambient temperature for 24 hours, had no effect on miR-223 and miR-135a concentration. This demonstrates the protective capacity of HDL-bound miRNAs [41]. Furthermore, miRNAs can also associate with proteins in plasma including the protein Argonaute 2, a member of the RNA-induced silencing complex [42]. This interaction offers protection to miRNA against degradation by the RNAses present in plasma [43], consistent with previous reports showing that AGO proteins shield the 3’ and 5’ ends of miRNAs from nucleases [44].

Interestingly, miRNA interaction with packaging proteins and vesicles is not random. Indeed, studies have found that several miRNAs are more likely to bind to some carriers, contributing to a distinct miRNA signature. Notably, this has been reported for HDLs, whose miRNA signature was considered distinct from that of plasma, exosomes, and low-density lipoprotein [39,45]. If EVs are more susceptible to degradation at ambient temperatures [46], we speculate that miRNAs may be exposed to RNases once the vesicle is degraded, leading to reduced miRNA stability. Understanding the impact of suboptimal conditions, such as unexpected exposition at ambient temperature that can occur during biobanked sample transport between laboratories, allows for a better assessment of miRNAs as potential biomarkers. Studying the degradation dynamics of miRNAs in plasma improves our ability to evaluate their therapeutic and diagnostic potential.

Strengths and Limitations

By measuring absolute miRNA levels by dPCR, we were able to achieve accurate and sensitive measurements, especially for those miRNAs that are less abundant in the bloodstream. The inclusion of a spike-in control during both reverse-transcription and dPCR protocol further enhanced the accuracy of our results. Another strength is our usage of a diverse range of miRNAs, varying in abundance, allowing us to assess the impact of ambient storage temperature on miRNA stability across different abundance levels in the bloodstream. However, a limitation of our study is the focus on only six miRNAs, which restricts our ability to generalize our findings. A genome-wide analysis of miRNA stability would provide a more comprehensive understanding but would require significantly more resources, and was beyond the scope of this study. Another limitation of our study is that the ambient temperature was maintained at a stable 20°C with low humidity and fixed duration. However, during transport, temperature and humidity levels fluctuate over time, which we did not account for in our study. These fluctuations could potentially further impact miRNA stability.

Conclusions

In conclusion, our results suggest that prolonged storage of plasma samples at ambient temperature is associated with a trend towards decreased miRNA concentration, which appears specific to individual miRNAs. However, this decrease in concentration did not correlate with initial miRNA abundance, or miRNA sequence, thus we hypothesize that binding partners are likely influencing their rate of degradation. This study aims to provide guidance for researchers who may encounter ambient temperature storage due to shipping conditions, offering insights into the potential impact on miRNA integrity.

Article Highlights:

• A trend in microRNA degradation in plasma samples with progressive thawing

• Degradation of microRNA was not dependent on %GC or initial abundance

• We predict that factors such as protein interactors and vesicles carrying these microRNAs impact degradation rate