Establishment Of Method For High-throughput Profiling MicroRNAs Activities In Live Cells And Its Application

microRNAs (miRNAs) are single-stranded, small and non-coding RNAs. In vivo miRNA isformed into miRISC (miRNA associated RNA induced silencing complex) along with cellularproteins such as AGO. miRISC recognizes and combines with target sequences located in the3′-untranslated region (3′UTR) of mRNA to negatively regulate gene expression throughdecrease of mRNA stability and repression of mRNA translation. miRNA plays critical roles inphysiological and pathological processes. It is important to learn more about miRNA functions.Methods for detection of miRNA expression and activity are the basis for miRNA research.Although several methods for profiling of miRNA expression such as qRT-PCR, miRNAmicroarray, and RNA sequencing, have been developed, approaches for profiling of miRNAsactivities are still lacked. miRNA activity, which directly reflects miRNA function, is not onlyrelated with its expression but also affected by other factors such as its localization, quantity oftargets, and cellular-associated proteins. Therefore it is important to establish a method todirectly detect miRNAs activities.There are three key points for establishing method of high-throughput profiling of miRNAsactivities, design of a proper miRNA activity sensor, large-scale preparation of sensors, andtechnique for high-throughput transduction of genes. In this study, we tried to establish a simpleand convenient method, nominated as miRNA Asensor array (Adeno-associated virus (AAV)vector based miRNA sensor array), for high-throughput functional miRNAs profiling in live cells.Then we utilized this method to analyze characteristics of miRNA activities profiles in9celllines, investigate effects of viral oncogenes on cellular miRNA activity, monitor changes ofmiRNA activity in the process of cell differentiation, and search out specifically high activity ofmiRNA in cells. Lastly, we developed a method to detect in vivo miRNA activity in mouse liver.We firstly constructed115plasmids to act as miRNA activity sensors, nominated as DNAbased sensors, Dsensors. miRNA Dsensor carried two independent expression cassettes, one forsecreted luciferase gene Gluc and the other firefly luciferase gene Fluc. The former contained asingle perfect complementary target for miRNA in the3′UTR, and was used to sense miRNAactivity. The latter contained no miRNA target sequence in the3′UTR, and was used to calibratethe differences of transduction efficiency among miRNA Dsensors.55miRNA Dsensors wererandomly chosen from the115miRNA Dsensors to test their functions in HEK293cells. Results showed that activities of55miRNA in HEK293cells were efficiently detected, suggesting thatmiRNA Dsensors were functional.However, high-throughput plasmid transfection is laborious and difficult to fulfill qualitycontrol. Therefore, we chose Adeno-associated virus vector to perform transduction of cellsconveniently and efficiently. So each miRNA Dsensor was packaged into recombinant AAV2vectors, nominated as miRNA Asensor (AAV vector based on miRNA sensor). miRNA Asensorswere orderly loaded onto96-well plate and miRNA Asensor array was obtained. Upon utilization,the same quantity of cells was loaded into miRNA Asensor array. Cells were reversely infectedby AAV vectors coated on the array. Sensors were tranduced into cells and reporter genes wereexpressed. We further investigated the effects of cell numbers loaded into each well as well assampling times post infection on miRNA Asensor array assay. Results demonstrated that themiRNA Asensor array worked well for miRNA activity profiling when the amount of cell loadedranged from3125to25000per well, and the sampling times ranged from36to60h postinfection.Different transduction efficiencies among Asensors will influence miRNA activity resultsdetected by miRNA Asensor array assay. To exclude this impact, we chose BHK21cell linewhich was sensitive to AAV2vector to correct transduction efficiencies by detecting transductioncoefficient (TC) of each miRNA Asensor. Firstly, the relationship between Gluc and Fluc activityin condition of no miRNA inhibition effect was obtained by infecting BHK21cells using ControlAsensor. Then the transduction efficiency of Control Asensor was defined as one unit oftransduction coefficient (TC). And TC value of miRNA Asensor was the ratio of its Gluc activitywithout miRNA inhibition to that of Control Asensor. According to the relationship between Flucand Gluc, TC value of miRNA Asensor was computed as the power ratio of1.32Fluc activity ofmiRNA Asensor to that of Control Asensor. miRNA activity was represented by relativeinhibiting fold (RIF), which was not only based on the principle of miRNA repression of geneexpression but also considered the different transduction efficiencies of miRNA Asensors. RIFvalue of one miRNA was obtained by the ratio of Gluc activity of Control Asensor to that of thismiRNA Asensor multiplied by its TC value. We further compared the relationship betweenactivity and expression levels of several miRNAs in HEK293cells. We found that RIF valuesdetected by miRNA Asensor array assay can efficiently reflect miRNA activities in cells with theexception of members from miRNA families.After establishment of miRNA Asensor array method, we tried to high-throughput profilemiRNA activities in cells using this method. We prepared a set of miRNA Asensor arraycontaining115miRNA sensors and detected miRNA activities profiles for9cell lines. miRNAactivity profile for each cell line can be obtained within48hours and the9cell lines were detected simultaneously. Activities of115miRNAs for one cell line or activities of one miRNAfor different cell lines can be quickly obtained, suggesting that our miRNA Asensor array washigh efficient and convenient. After analysis of the data, some phenomena were found. Firstlytotal miRNA activities of cell lines derived from tumors (e.g. K562, U937, HepG2, andHuh7/CD81) were clearly lower than that of cells derived from normal tissues (e.g. BJ andC2C12). Activity of tissue-specific miRNAs (e.g. miR-122, miR-194, miR-142-3p, andmiR-142-5p) showed high in the cell lines from their original tissues. Activity level of miRNA astumor suppressor (e.g. let-7a, miR-31, miR-199a-3p, and miR-143) in cell lines derived fromnormal tissues (e.g. BJ and C2C12) was higher than that in the cell lines from tumors(e.g.HepG2, Huh7/CD81, U937, and K562). Activity of each oncomir selected (e.g.miR-17-5p,miR-221/222, and miR-21) had its own performance among cell lines from normal or tumortissues. Activity of miR-17-5p in cell lines from tumors (e.g.HepG2, Huh7/CD81, HeLaS3,U937, and K562) was quite higher than that in cell lines from normal tissues (e.g.BJ, C2C12, andBEAS-2B). Activities of miR-221/222were both higher in BJ, BEAS-2B, and C2C12fromnormal tissues and HepG2, Huh7/CD81, and HeLaS3from solid tumors, while lower in U937and K562from blood cancers. Furthermore, miR-21activity was significantly higher than othermiRNAs among all of nine cell lines. As an oncogene or tumor suppressor reported b y otherresearchers, miR-26a activity was higher in BJ and BEAS-2B from normal tissues as well asHepG2, Huh7/CD81, and HeLaS3from solid tumors, while lower in U937and K562from bloodcancers. These results indicated that our miRNA Asensor array assay co uld rapidly discoverdifferences of miRNA activities among cell lines, which provide helpful clues for miRNAfunctions and mechanisms. Our results also suggested that the division of miRNAs intooncogene or tumor suppression gene need more evidences and co nsiderations.After comprehensive assessment of115miRNA activities profiles for9cell lines, weprimarily chose out13specific miRNAs for further analysis. We found that the13miRNAsactivities profiles were quite different among the9cell lines, suggesting that a panel of miRNAsactivities can be used as biomarkers to identify cell lines.We tried to use miRNA Asensor array to analyze effects of viral oncogene on cellularmiRNA activity. We set effects of SV40large T antigen (SV40LT) on miRNA activity inHEK293cell as an example. We firstly obtained115miRNA activities profiles of HEK293andHEK293T cell lines using miRNA Asensor array assay. HEK293T was known stably expressingSV40LT. We found that total miRNA activities of HEK293T were lower tha n that of HEK293.We further compared some miRNAs activities between the two cell lines. Results showed thatactivities of miR-26a and members of let-7family with exception of let-7i in HKE293T cellswere significantly lower than that in HEK293cells (P<0.01). However, activities of miR-18a and miR-20a from miR-17-92cluster in HEK293T were higher than that in HEK293(P<0.05),suggesting that HEK293T was more “tumor-like” compared with HEK293. There was nostatistical difference for miR-221/222activities between HEK293T and HEK293cells.Interestedly, both miR-21activity and expression levels in HEK293T cells were significantlylower than those in HEK293cells (P<0.01), respectively. Subsequently, we constructedexpression vector of SV40LT pSV40LT and confirmed it by detecting SV40LT through WBassay. Then we compared miRNA activity between HEK293cells transfected with pSV40LT ornot. We found total activity of115miRNAs was also shown lower in HEK293cell transfectedwith pSV40LT than naive HEK293cells. Furthermore, activities of miR-26a and members fromlet-7family excepting let-7i were significantly lower (P<0.05) for pSV40LT transfectedHEK293cells than non-tranfected HEK293cells. However, miR-21, miR-17-92, andmiR-221/222activities showed no statistical differences. These results indicated that SV40LThad effects on miRNAs activities for HEK293cells and provided clues for further studies of themechanisms for SV40LT oncogenesis and the roles miRNAs played in the process.We monitored changes of miRNA activity in the process of K562differentiation induced byTPA using miRNA Asensor array. At first, we showed that obvious changes of morphology forK562cells appeared24h post induction by TPA. And48h post induction, total activities of115miRNA were significantly increased in K562cells. We further found that activities for10of115miRNAs (miR-221, miR-222, miR-34a, miR-21, let-7a, let-7b, let-7c, let-7e, let-7f, andmiR-146a) were significantly increased while activities of3miRNAs (miR-106, miR-144, andmiR-32) were markedly decreased. These results suggested that miRNA Asensor array assay is auseful tool to monitor changes of miRNA activity in the cellular physiological process.We found that activity of miR-206was specifically high for BHK21cells in application ofmiRNA Asensor array.58miRNAs, whose mature sequences were highly conserved amongHomo sapiens, Mus musculus, Rattus norregicus, and Danio rerio, were chose. miRNA Asensorarray was prepared. Using the miRNA Asensor array method, we obtained58miRNAs activitiesprofiles for three kidney tissue derived cell lines, BHK21, HEK293, and Vero. Results showedthat activity level of miR-206was specifically high in BHK21. We confirmed this finding bydetecting activity and expression levels of miR-206in BHK21cells simultaneously withHEK293as negative control and myoblast cell C2C12as positive control. We further inducedBHK21cells by culturing it with medium containing horse serum (HS). Changes of morphologyfor BHK21cells appeared and skeletal myosin heavy chain (MHC) was detectable by WB,indicating that BHK21cells was induced to differentiate towards skeletal muscle cells. Afterinduction, both activity and expression levels of miR-206were markedly increased for BHK21cell while the expression level of Connextin43(Cx43), which is reported to be the target gene regulated by miR-206, was decreased, suggesting that miR-206may work throughdown-regulation of Cx43in the process of BHK21differentiation. Our results suggested thatmiR-206can be used as a biomarker to identify BHK21cells.At last, we developed a method for detection of miRNA activity in vivo. miR-21was takenas an example. We constructed three plasmid sensors (miRNA Psensor) for in vivo miRNAactivity detection (miR-21, miR-122, and miR-206Psensor) with empty Psensor (ControlPsensor) as control. miR-122and miR-206Psensors were used as positive and negative controls,respectively. Control Psensor and the three miRNA Psensors were separately injected into mo useby hydrodynamic method through tail-vein. Detected reporter expression, computed the RIFvalues and obtained miRNA activity. Results showed that both miR-21and miR-122activitieswere high in mouse liver and that miR-21was much higher than miR-122while miR-206activity was little. Furthermore, expression level of miR-21was clearly lower than that ofmiR-122. Our results firstly demonstrated that activity of miR-21was markedly higher than thatof miR-122while its expression level was lower than that of miR-122, suggesting that miR-21was critical in the physiological process of mouse liver.In conclusion, we successfully established a method for high-throughput miRNA activitiesprofiling, miRNA Asensor array assay. We used this approach to investigate miRNA activitiesprofiles of nine cell lines, indicating that miRNAs were potential to be as biomarkers to identifydifferent cell lines. As a useful research tool, miRNA Asensor array assay could be applied inmany areas including analysis of effects of environmental factors on cellular miRNA activity,monitor of miRNA activity in physiological process and discovery of specific high activity levelof miRNA in cells. We also established a method to detect miRNA activity in mouse liver. Usingthis method, we firstly found that activity of miR-21was high in normal mouse liver, suggestingthe multi-face of miR-21function in vivo.