Distribution of microRNA biomarker candidates in solid tissues and body fluids

ABSTRACT

Small non-coding RNAs, especially microRNAs, are discussed as promising biomarkers for a substantial number of human pathologies. A broad understanding in which solid tissues, cell types or body fluids a microRNA is expressed helps also to understand and to improve the suitability of miRNAs as non- or minimally-invasive disease markers. We recently reported the Human miRNA Tissue Atlas (http://www.ccb.uni-saarland.de/tissueatlas) containing 105 miRNA profiles of 31 organs from 2 corpses. We subsequently added miRNA profiles measured by others and us using the same array technology as for the first version of the Human miRNA Tissue Atlas. The latter profiles stem from 163 solid organs including lung, prostate and gastric tissue, from 253 whole blood samples and 66 fractioned blood cell isolates, from body fluids including 72 serum samples, 278 plasma samples, 29 urine samples, and 16 saliva samples and from different collection and storage conditions. While most miRNAs are ubiquitous abundant in solid tissues and whole blood, we also identified miRNAs that are rather specific for tissues. Our web-based repository now hosting 982 full miRNomes all of which are measured by the same microarray technology. The knowledge of these variant abundances of miRNAs in solid tissues, in whole blood and in other body fluids is essential to judge the value of miRNAs as biomarker.

Introduction

Small non-coding RNAs, predominantly miRNAs, play a central role in life science research and potentially will become valuable biomarkers in clinical care¹ or even novel therapeutics.² An improved understanding of the distribution of miRNAs across different biological specimens helps researchers to understand whether candidate marker or drug miRNAs are selectively expressed in one or few tissues or are rather ubiquitous abundant. The first study to present miRNA patterns across different tissues has already been carried out in 2007 by Landgraf and co-workers.³ Since that time, the number of miRNAs stored in the reference database miRBase has increased by an order of magnitude, calling for an update on miRNA tissue patterns. For such an endeavor several factors are important. Since miRNAs show inter-individual variations we profiled the organs and tissues from the same individuals. Additionally, it is important to keep the measurement platform the same to avoid bias due to the used technique, as it is the case for microarrays compared to NGS.⁴ As investigated by Mestdagh and colleagues, different platforms have different advantages and disadvantages, while “a best” platform outperforming all others does not exist. The data of miRNA profiles in different conditions in this study rely on Agilent's microarray technology, which has shown a high reproducibility and acceptable sensitivity in previous studies.⁵ Using microarrays we recently published a Human miRNA Tissue Atlas, which is freely accessible (www.ccb.uni-saarland.de/tissueatlas).⁶ In this study we first investigated the influence of degradation on tissue miRNA patterns and then profiled altogether 31 different organs from 2 corpses. Tissue patterns were generated for miRNAs from the most recent versions 20 and 21 of the miRBase. The results highlighted that the majority of miRNAs (82.9%) was neither specific for a single tissue nor ubiquitous expressed in all samples but in a subset of all tested organs. In addition to these miRNA profiles generated from different organs of the same bodies, others and we used the Agilent array technique to profile miRNAs in body fluids such as blood,⁷ serum,^8,9 plasma, urine, saliva, in different blood cell types,¹⁰ and under different collection¹¹ and storage conditions. In order to understand how promising biomarker candidates are distributed across tissues and body fluids we included these data to the tissue atlas, now containing the 105 miRNA profiles of 31 organs, 163 profiles measured from selected organs such as normal lung, prostate and gastric tissue, from 253 whole blood samples and 66 fractioned blood cell isolates,^7,10 from body fluids including 72 serum samples,^8,9 278 plasma samples , 29 urine samples , and 16 saliva samples and from different collection¹¹ and storage conditions such that the database now hosts 982 full miRNomes measured by the same microarray technology. Using this resource we analyzed which miRNAs generally overlap in whole blood and different solid tissues and which tissue specific miRNAs are found in blood. As an example we show the distribution for 3 biomarker candidates in detail.

miRNAs in solid tissues and whole blood

We first ask if the overall pattern of miRNAs in whole blood is determined predominantly by a specific solid organ of a group of organs. To this end, we compared how many miRNAs are found both in solid organ tissues and in whole blood. Considering the 31 solid organs included in our original Tissue Atlas we found that approximately one third of the miRNAs (average of 29.2%) found in the respective organ were also present in whole blood as highlighted in Fig. 1. The largest fraction was observed for the vein with 31.7% of the miRNAs found in this organ and likewise in blood and the lowest fraction was observed for spleen with 26.4% of the miRNAs found in this organ and also in blood. Considering all miRNAs we found however no evidence that a specific solid organ predominantly determines the overall miRNAs pattern of whole blood.

Figure 1.

Bar diagram showing for each of the 31 solid organs the number of expressed miRNAs to the right and the number and percentage of miRNAs that are found in the respective organ and blood. Organs are sorted from top to down in decreasing fraction of miRNAs found in the organ and blood. Highest fraction is observed for vein and epididymis, lowest for spleen.

Since this analysis has been carried out for all miRNAs that are expressed in the solid tissues without taking into account in how many tissues the respective miRNA is found, we refined the analysis focusing on miRNAs with an increased “specificity” for specific solid tissues. We considered a miRNA as “specific” for n tissues, if the average quantile normalized expression of this miRNA across the n tissues that show the highest abundance for this miRNA exceeds the highest expression in the remaining tissues at least 5-fold. This means that a “specific” miRNA is expressed at least 5-fold higher in a given single tissue or in a group of tissues than in any of the remaining tissues considered. To largely exclude the influence of miRNAs that showed only expression values close to the detection background, we limited our analysis to miRNAs that are stably expressed (in this context, stably expressed is defined by exceeding 10 intensity counts following quantile normalization across the solid tissues). Using this definition we found 342 miRNAs (26.4%) that were expressed in all solid tissues. Furthermore, the majority of miRNAs were found in more than 10 different solid tissues (65.3%) while only 7% of all miRNAs were predominantly expressed in a single solid organ.

Next we analyzed the presence of these solid tissue miRNAs in whole blood. While it is difficult to draw conclusions for solid tissues that carry only one or 2 specific miRNAs, we clearly observed a tendency of the miRNAs that are more ubiquitous abundant in solid tissues to be also present in blood. Out of the group of miRNAs being present in 11–30 different solid tissues, 31.2% were also found in whole blood. This ratio largely corresponds to the ratio found for the analysis of all miRNAs combined as addressed above. As examples for tissue specific miRNAs that were also expressed in blood we found both mature forms of the mir-140 precursor, being specific for bone, miR-132-3p being specific for brain, miR-22-5p and both mature forms of mir-378a being specific for muscle, and miR-223-3p specific for vein. However, we also found examples of miRNAs that we largely specific for a given solid tissue but not present in whole blood. Out of 22 miRNAs specific for muscle only 3 were found in blood, out of 16 miRNAs that were specific for spleen none was found in blood and likewise none out of 11 miRNAs that were specific for epididymis were found in blood. Fig. 2 summarizes the results for miRNAs specifically found in single tissues or groups of tissues (2-3, 4–5, 6–10, 11–30) and also in whole blood. While miRNAs that are more ubiquitous abundant in solid tissues are likewise present in whole blood, few miRNAs that are rather specific for solid tissues are not necessarily found in whole blood at least not under physiological conditions. The variant abundances of miRNAs in solid tissue and whole blood need to be taken into account when evaluating the use of miRNAs as biomarker.

Figure 2.

Comparable to Fig. 1, this bar diagram highlights for each solid organ the number of specific miRNAs to the right, while to the left the solid organ specific miRNAs that are also observed in blood are shown. The lower bars present groups of miRNAs that are not specific for single organs but are found in 2–3, 4–5. Six-10, 11–30 and across all 31 solid organs.

Detailed distribution of miRNAs across tissues and body fluids

We selected 4 miRNAs as example to demonstrate the high variability of expression of biomarker candidates in solid tissue and whole blood. miR-208a-3p is a known cardiac specific miRNA.¹² This is consistent with our microarray analysis that showed expression of miR-208a-3p in heart only but not in other solid tissues, nor in whole blood or in other body fluids (Fig. 3A). miR-1-3p is an example for a miRNAs that is predominantly found in 2 different tissues namely heart and skeletal muscle. Again, this miRNA is not expressed in blood and other body fluids (Fig. 3B). The third example miR-223-3p is described as marker for multiple diseases, e.g. Crohn's Disease,¹³ synovial sarcoma,¹⁴ and Rheumatoid Arthritis¹⁵ among others. We found this miRNA strongly expressed in the solid tissues vein and artery and weakly abundant in plasma and in whole blood especially in CD14 and CD15 cells (Fig. 3C). miR-320a has been correlated to different cancer types e.g., pancreatic,¹⁶ gastric¹⁷ and lung cancer,¹⁸ but also to heart failure¹⁹ or type 2 Diabetes.²⁰ We found this miRNA abundant in almost all solid tissues, in blood and urine and to a lesser extent in serum, plasma and selected blood cells (Fig. 3D).

Figure 3.

(A-D): For 4 different biomarker candidates the expression pattern across solid tissues and different body fluids is presented.

Such miRNAs may have a limited value as biomarker due to their lack of specificity. While this can be compensated by discovery of more specific miRNA patterns, in the development of drugs non-specific miRNAs may show side effects.

Limitations and future directions

One strength of the first version of the tissue atlas is the autologous measurement of different organs all derived from the same individual. As of now these analyses rely on 2 corpses only. In future versions of the tissue atlas different genders, ages, and races are to be included to broaden the view on miRNA distribution in human tissues. Toward an even more comprehensive picture of miRNA expression in human tissues, it will also be helpful to record different miRNA pattern within tissues especially form rather heterogeneous tissues as reported by Kakimoto and coworkers for a chamber specific expression of the afore-mentioned miR-208a-3p.²¹ With respect to body fluids, our data collection contains a variety of diseases and control samples. As for these body fluids there is a fundamental problem in obtaining several of these samples from the same individual and to compare these data with the data obtained form tissues of the corpses. Furthermore, the atlas is in its early stages and miRNAs that are not present in any of the analyzed samples e.g. miRNAs that are not found in the 278 plasma samples not necessarily present under other physiological or pathological conditions. In addition to these challenges different definitions of “specificity” for an organ and different thresholds can substantially influence the results. In the light of these limitations we consider our miRNA atlas to be a useful resource, which has however to be extended in the aforementioned directions.

Conclusion

In sum it is important to understand in which body fluids, cell types, organs and under which conditions miRNAs are expressed and can be measured. It has substantial advantages to compare only data that have been generated by the same high-throughput approach. The collection of 982 experiments performed using microarrays provides researchers with interesting insights on the distribution and specificity of their biomarker or pharmaceutical candidates.

Disclosure of potential conflicts of interest

A.K. is consulting for Hummingbird Diagnostics GmbH.