Reversal of Phospholamban Inhibition of the Sarco(endo)plasmic Reticulum Ca2+-ATPase (SERCA) Using Short, Protein-interacting RNAs and Oligonucleotide Analogs*

The sarco(endo)plasmic reticulum Ca2+-ATPase (SERCA) and phospholamban (PLN) complex regulates heart relaxation through its removal of cytosolic Ca2+ during diastole. Dysfunction of this complex has been related to many heart disorders and is therefore a key pharmacological target. There are currently no therapeutics that directly target either SERCA or PLN. It has been previously reported that single-stranded DNA binds PLN with strong affinity and relieves inhibition of SERCA in a length-dependent manner. In the current article, we demonstrate that RNAs and single-stranded oligonucleotide analogs, or xeno nucleic acids (XNAs), also bind PLN strongly (Kd <10 nm) and relieve inhibition of SERCA. Affinity for PLN is sequence-independent. Relief of PLN inhibition is length-dependent, allowing SERCA activity to be restored incrementally. The improved in vivo stability of XNAs offers more realistic pharmacological potential than DNA or RNA. We also found that microRNAs (miRNAs) 1 and 21 bind PLN strongly and relieve PLN inhibition of SERCA to a greater extent than a similar length random sequence RNA mixture. This may suggest that miR-1 and miR-21 have evolved to contain distinct sequence elements that are more effective at relieving PLN inhibition than random sequences.

The sarco(endo)plasmic reticulum Ca 2؉ -ATPase (SERCA) and phospholamban (PLN) complex regulates heart relaxation through its removal of cytosolic Ca 2؉ during diastole. Dysfunction of this complex has been related to many heart disorders and is therefore a key pharmacological target. There are currently no therapeutics that directly target either SERCA or PLN. It has been previously reported that single-stranded DNA binds PLN with strong affinity and relieves inhibition of SERCA in a length-dependent manner. In the current article, we demonstrate that RNAs and single-stranded oligonucleotide analogs, or xeno nucleic acids (XNAs), also bind PLN strongly (K d <10 nM) and relieve inhibition of SERCA. Affinity for PLN is sequence-independent. Relief of PLN inhibition is length-dependent, allowing SERCA activity to be restored incrementally. The improved in vivo stability of XNAs offers more realistic pharmacological potential than DNA or RNA. We also found that microRNAs (miRNAs) 1 and 21 bind PLN strongly and relieve PLN inhibition of SERCA to a greater extent than a similar length random sequence RNA mixture. This may suggest that miR-1 and miR-21 have evolved to contain distinct sequence elements that are more effective at relieving PLN inhibition than random sequences.
Calcium cycling in cardiomyocytes is a tightly regulated process that ensures proper muscle contractility (1,2). Many of the proteins involved in this cycle have been implicated in cardiac failure, the leading cause of death world-wide (3,4). Given the complexity of heart failure phenotypes, strategies for reversing declining cardiac performance are diverse and necessary. Gene therapy efforts for inherited forms of heart disease have been growing, but with the recent failure of the Calcium Upregulation by Percutaneous Administration of Gene Therapy in Car-diac Disease (CUPID) trials, drug development targeting specific proteins remains an essential, complimentary effort (5)(6)(7). A therapeutic that is tunable to specific phenotypes would be ideal for reversing aberrant calcium cycling.
The sarco(endo)plasmic reticulum Ca 2ϩ -ATPase (SERCA) 3 is an essential Ca 2ϩ -handling protein that removes Ca 2ϩ ions from the cytosol, causing the heart muscle to relax. SERCA is a P-type ATPase, which in human cardiomyocytes is responsible for the removal of ϳ70% of Ca 2ϩ from the cytosol (8). This process, coupled with other Ca 2ϩ transport mechanisms, lowers the cytosolic Ca 2ϩ concentration enough to allow for muscle relaxation (diastole) (1). Phospholamban (PLN) is a 52amino acid, single-pass transmembrane protein that inhibits SERCA when not phosphorylated by lowering its apparent calcium affinity and hampering Ca 2ϩ transport into the sarcoplasmic reticulum (9,10). Upon ␤-adrenergic stimulation, protein kinase A (PKA) will phosphorylate PLN at Ser-16, restoring the apparent calcium affinity and Ca 2ϩ transport of SERCA (11).
We previously reported that random sequence, singlestranded DNA and RNA bind to PLN with high affinity (K d Ͻ 10 nM), and more importantly, relieve its inhibitory effects on SERCA, restoring the enzyme's basal activity. We found that PLN remained bound to SERCA upon the addition of ssDNA, and that the effects of ssDNA on the SERCA-PLN complex are length-dependent and tunable (12). We propose to refer to these unique sequences as short, protein-interacting DNAs and RNAs (i.e. SPIDRs).
In the current study, we probe the chemical nature behind this unexpected interaction and the reversal of SERCA inhibition. In addition, although ssDNA and RNA are readily degraded in vivo, the chemically modified oligonucleotide analogs, or XNAs, chosen in this work are nuclease-resistant and are therefore much better scaffold candidates for drug development.
Secondly, we examine two specific microRNAs (miRNAs), determining both how strongly they bind PLN and their functional effects on the SERCA-PLN complex. Non-coding miRNAs play central roles in gene regulation and pathology (13). In the heart, miRNAs are involved both in cardiogenesis (14,15) and in disease including: diabetic cardiomyopathy, hypertrophy, ischemia, and electrical remodeling (16 -22). Recent developments in both therapeutic inhibition and enhancement of miRNA function have demonstrated great promise for counteracting cardiac diseases (22,23). However, the off-target effects of miRNAs and antisense oligonucleotides that target miRNAs (i.e. the pleiotropic action of miRNA, affecting multiple genes in different tissues) may prevent the progression of miRNA regulation to therapeutic applications (24).
In the current study, we report that many XNAs bind PLN with similar strong affinity to what was found previously for SPIDRs (12). These molecules, which are tunable by length, would allow clinicians to match the reversal of SERCA inhibition to the severity of the disease. In addition, we found that although most XNAs display similar functional effects to what was seen for ssDNA, specific miRNAs (both miR-1 and miR-21) have a greater effect on the SERCA-PLN complex activity than similar length random sequence RNA libraries. Furthermore, our results suggest that endogenous, non-coding miRNAs may play a more complex role in cardiac regulation than previously thought, targeting SERCA-PLN function via direct physical interactions.

RNA Sequences
Bind Phospholamban with Low Nanomolar K d Reversing SERCA Inhibition-Previously, we found that ssDNA of varying lengths could relieve the inhibition of SERCA by PLN to different extents (12). We also determined that a 50-mer of RNA bound PLN with similar affinity and relieved inhibition of SERCA to a similar extent as ssDNA (12). In the current study, we determined the full extent of different lengths of RNA binding to PLN and its functional effects on the SERCA-PLN complex. Because non-coding RNAs are abundant in cardiomyocytes, we assayed whether RNA sequences would interact with PLN and reverse SERCA inhibition (Figs. 1 and 2). Using affinity capillary electrophoresis (ACE) and fluorescence polarization (FP), we found that random sequence RNA oligonucleotides (5-50-mers) have high affinity for PLN if longer than a 5-mer (Fig. 1). The 20-mer library, which is representative of the typical length of endogenous miRNAs, shows 9 -19 nM binding affinity (Fig. 1). RNA demonstrated no measurable affinity for SERCA (K d Ͼ 4 M). Similarly, only weak affinity was observed for a group of cardiac proteins including actin, myosin, and troponin, suggesting that the observed RNA affinity is specific for PLN.
We performed a competitive assay to determine the contribution of the FAM (fluorescein) label to binding. As shown in Fig. 1B, unlabeled, random sequence, 20-mer RNA effectively competes with the FAM-labeled RNA library at the PLN binding site. The slope of the curve is 1.32 Ϯ 0.07, suggesting that if anything, the addition of the FAM label diminishes the affinity of RNA for PLN.
We performed coupled enzyme assays at different free calcium concentrations (pCa) to assess the extent with which RNAs of different lengths relieve PLN-mediated SERCA inhibition. Coupled enzyme assays use a reconstituted lipid system containing SERCA and PLN and are commonly used to determine SERCA activity (25,26). These assays determine the ATPase activity under near physiological conditions (26). SERCA affinity for calcium is decreased upon the addition of PLN (Fig. 2). The pK Ca corresponds to the pCa at half-maximal activity (pCa at "normalized SERCA activity" ϭ 0.5). The extent of PLN inhibition is quantified by a shift in the pK Ca to lower values and can be visualized in Fig. 2 by the shift in the calcium curve to the right upon the addition of PLN. Upon the addition of RNA, the apparent calcium affinity of SERCA is augmented and the activity curve shifts back toward the left (higher pK Ca values) (Fig. 2). The extent of relief of inhibition of SERCA is dependent on RNA length (Fig. 2). As RNA length increases, inhibition is gradually relieved, with the 50-mer re-establishing SERCA basal activity. This is similar to the incremental relief we have previously observed for ssDNA (12). The random sequence library that was used in these experiments confirms that the RNA-PLN interaction is sequence-independent. These results imply that cytosolic miRNAs are able to interact with the SERCA-PLN complex, providing a natural mechanism of regulation through direct, physical interaction with PLN, a novel phenomenon that has not been seen before with a nontranscription related protein.
miR-1 and miR-21 Bind to PLN Strongly and Relieve Inhibition of SERCA-To determine whether endogenous miRNA sequences are able to reverse the inhibition of PLN, we chose miR-1 and miR-21, which are 22 and 21 nucleotides in length (Fig. 3A). Both miR-1 and miR-21 are implicated in many disease states (13), and miR-1 is highly abundant in cardiomyocytes (13). Using ACE and FP, we found that both miR-1 and miR-21 bind PLN tightly, with low nM dissociation constants. Although displaying slightly higher dissociation constants (ϳ40 -60 nM) than the random sequence mixture (ϳ10 nM), they are still able to bind with remarkably high affinity (Fig. 3, B-D). The difference in K d may be due to the presence of secondary structure in both miR-1 and miR-21, which may incur a thermodynamic penalty associated with unfolding to bind PLN. The secondary structures, as predicted by mfold (University of New York at Albany), are shown in Fig. 3A. Nonetheless, both miR-1 and miR-21 effectively reverse PLN inhibition and fully restore the basal activity of SERCA more efficiently than the random mixture of 20-nucleotide RNA sequences (Fig. 3E).  1) was used to estimate pK Ca (i.e. pCa 2ϩ at activity half-maximum) from each curve. Each data point is the average of n ϭ 3 replicates. Error bars were omitted for clarity. B, pK Ca estimates from A are plotted for different length random sequence RNAs. Error bars are the standard errors from the regression analyses used to estimate pK Ca (Equation 1, n ϭ 3 replicates measured at 12 Ca 2ϩ concentrations). PLN inhibits SERCA by lowering the apparent Ca 2ϩ affinity of SERCA, resulting in a lower pK Ca value. The addition of 50-mer RNA fully restores SERCA activity (no statistical difference from SERCA alone, p Ͼ 0.05). All lengths other than the 5-mer and 10-mer are statistically different when compared with WT PLN (p Ͻ 0.05). p values were calculated using an unpaired t test. Both miR-1 and miR-21 remove all PLN-mediated inhibition. The resulting pK Ca is not statistically different from SERCA alone (Fig. 3E). This suggests that miRNAs may have sequence elements that relieve inhibition more effectively than random sequence RNA.
Oligonucleotide Analogs (XNAs) Bind Phospholamban with Low Nanomolar Dissociation Constants-We analyzed several modified oligonucleotide analogs to determine the specific chemical moieties of ssDNA and RNA that are responsible for the high affinity for PLN (Fig. 4). These chemically modified oligonucleotides (XNAs) also display increased nuclease resistance that is essential for in vivo applications. The XNAs assayed include nuclease-resistant and chemical modifications of DNA such as L-DNA (the enantiomer of natural DNA), 2Ј-Omethyl RNA, phosphorothioate DNA (PT DNA, replacement of one of the non-bridging oxygens with a sulfur), morpholinos (neutral backbone), and peptide nucleic acids (PNA, neutral peptide backbone with DNA bases). All of the chosen synthetic XNAs could be reliably synthesized as random mixtures at a length of 20 nucleotides. We used ACE and FP to determine the binding affinity of PLN with various XNAs (Fig. 5).
We found that charged XNAs (i.e. PT DNA, L-DNA, and 2Ј-O-methyl RNA) bind PLN with an affinity similar to ssDNA or RNA (K d Ͻ 4 nM). Neutral oligonucleotides (i.e. PNA and morpholinos) did not demonstrate any measurable affinity for PLN (Fig. 5), suggesting that affinity is driven by the negatively charged backbone. The cytoplasmic domain of PLN has four positively charged amino acids (Lys-3, Arg-9, Arg-13, and Arg-14) that we previously implicated in interactions with the negatively charged backbone of ssDNA (12). It should be noted that there are other well known examples of proteins, including Taq polymerase (27)(28)(29) and histones (30,31), that interact with nucleic acids primarily through electrostatic and polar interactions at the phosphodiester backbone in a sequence-independent manner.
XNAs Relieve the Inhibition of SERCA by PLN to Different Extents-We also performed coupled enzyme assays using the same lipid-reconstituted system as in Fig. 2. For these experi-   OCTOBER 7, 2016 • VOLUME 291 • NUMBER 41

JOURNAL OF BIOLOGICAL CHEMISTRY 21513
ments, the corresponding 20-mer XNA, random sequence library was added to the complex before beginning the assay. Previously, we showed that different lengths of ssDNA have a graded effect on the SERCA-PLN complex (12), with the 20-mer showing partial relief of inhibition. Fig. 6A depicts the pK Ca values obtained from normalized SERCA-PLN activity curves upon the addition of the different 20-mer XNAs. Without the addition of oligonucleotide, SERCA alone displays a higher Ca 2ϩ affinity (Fig. 6, leftmost black curve and bar). PLN binding to SERCA results in inhibition or decreased SERCA Ca 2ϩ affinity (i.e. the lowest pK Ca ), as depicted with the red curve and bar (Fig. 6). The addition of an XNA to the SERCA-PLN complex shifts the pK Ca higher, indicating that the apparent affinity of SERCA for Ca 2ϩ ions is partially restored. Although the affinity for all of the charged XNA molecules (DNA, RNA, L-DNA, and 2Ј-O-methyl RNA) is essentially equal, their functional effect is distinct. All activity assays were performed at saturating concentrations of XNA well above observed K d values with PLN. The observed differences in XNA effect on PLN inhibition are therefore concentration-independent and unrelated to relative affinity. The uncharged XNAs (PNA and the morpholino) display no effect on the SERCA-PLN complex. This was expected because we observed no affinity between these molecules and PLN. The pK Ca shift for all of the charged molecules was expected to be very similar if electrostatics is the only contributing factor. L-DNA and 2Ј-Omethyl RNA display less relief of SERCA inhibition than the corresponding ssDNA 20-mer (Fig. 6). 2Ј-O-Methyl RNA may have an unfavorable steric effect arising from the added methoxy group at the 2Ј position. The decreased potency of the L-DNA may suggest that although the chirality of the molecule has no effect on binding affinity, the structural conformation adopted upon binding contributes to the extent to which the inhibition of SERCA by PLN is relieved. In contrast, PT DNA displays a statistically significant, higher potency than the corresponding ssDNA 20-mer (Fig. 6, p Ͻ 0.05). This could be due to the fact that sulfur has a slightly larger atomic radius than oxygen and thus a more disperse, negatively charged electron cloud. Only the additions of the ssDNA and the PT DNA show a statistically different pK Ca when compared with WT PLN (Fig. 6). Due to the limitation in the synthesis of these oligonucleotides, we utilized only 20-mers for these experiments. As longer lengths have a more potent effect for RNA (Fig. 2), we expect that longer XNA sequences would also have a greater impact on pK Ca .

Discussion
Our previous work demonstrated that random sequence ssDNA could be used in a tunable manner to alter SERCA function both in vitro and in vivo. However, ssDNA is not a viable therapeutic due to its short in vivo lifetime. Therefore, we turned to XNAs for two reasons: 1) with their nucleaseresistant properties, they have a much longer in vivo lifetime (32) and are potential therapeutic candidates if a suitable carrier can be developed, and 2) the properties of XNAs enabled us to understand the chemical requirements behind this high affinity interaction between random sequence ssDNA and PLN.
We confirmed that RNA Ͼ5 nucleotides long tightly binds PLN (Fig. 1). Additionally, RNA demonstrates a similar lengthdependent relief of inhibition (Fig. 2) to what was seen for ssDNA on SERCA activity (12). PLN-mediated inhibition of SERCA is removed in a stepwise, tunable manner with increasing lengths of RNA. Furthermore, we demonstrated that electrostatics are necessary for strong affinity for PLN (Figs. 5 and 6); all XNA molecules with a negatively charged backbone display high affinity for PLN. However, electrostatics are not the only important factor when considering relief of SERCA inhibition (Fig. 6). In fact, 2Ј-O-methyl RNA displays a similar affinity for PLN as ssDNA, but does not demonstrate a similar relief of inhibition. The addition of the methoxy group on the 2Ј-carbon may sterically hinder the structural conformation or eliminate a hydrogen-bonding site needed for relief of SERCA function.  1) was used to estimate pK Ca (i.e. pCa 2ϩ at activity half-maximum) from each curve. Each data point is the average of n ϭ 6 replicates. Error bars were omitted for clarity. B, pK Ca estimates from A) are plotted for different oligonucleotide analogs (20-mer, random sequence). Error bars are the standard errors from the regression analyses used to estimate pK Ca (Equation 1, n ϭ 6 replicates measured at 12 Ca 2ϩ concentrations). p values were calculated using an unpaired t test (*, p Ͻ 0.05, ***, p Ͻ 0.001).
The homeostatic balance of Ca 2ϩ cycling in cardiac muscle cells is essential for proper function (1,2,33). SERCA and PLN are key proteins within this tightly regulated process (3) and have been identified as potential drug targets (34,35). Improper regulation of SERCA function, either by PLN mutants or by changes in PLN phosphorylation level, leads to cardiomyopathies (36 -45). To reverse adverse effects of heart failure, Hajjar and co-workers (6) pioneered the use of gene therapy to augment Ca 2ϩ transport. Unfortunately, their recent clinical trials did not show any improvement with respect to patients treated with placebo. In a parallel approach, chemically modified oligonucleotide analogs have been used to target miRNAs for calcium-handling proteins. Recently, this approach was used by Größl et al. (46) to target PLN. They constructed an adenoassociated virus (AAV) vector (scAAV6-amiR155-PLBr) to express an anti-miRNA (amiR155-PLBr) that enhanced Ca 2ϩ transport into the sarcoplasmic reticulum. Here, we propose to use XNAs to directly target the SERCA-PLN complex rather than the miRNA transcripts.
Although random sequence ssDNAs elicit the same effects as XNA in a tunable manner both in vitro and in cellular experiments (12), ssDNA is not a viable therapeutic agent due to its short in vivo lifetime. XNAs, on the other hand, display similar binding properties and activity toward the SERCA-PLN complex, with the advantage of being nuclease-resistant, i.e. possessing a longer in vivo lifetime. Therefore, XNAs, or related small molecules, could be developed into targeted therapeutics that directly interact with the SERCA-PLN complex to relieve inhibition in a tunable manner. This type of design flexibility using molecule length, not concentration, is unique and biologically important because it leads to potential customizable drug development. The length and properties of the molecule could be tuned to provide a graded response, allowing for varying degrees of relief of inhibition based on disease severity. This would be the first customizable therapeutic directly targeting SERCA function. Our studies will pave the way for developing alternative therapeutic approaches to treat cardiomyopathies focused on the SERCA-PLN complex.
An intriguing aspect of miRNA biology is the up-and downregulation of specific miRNAs in response to cardiac stress (13,47,48). The presence of miRNAs in the cytoplasm under pathological conditions suggests that they may be able to interact with cardiac proteins directly, modifying their functions. We demonstrated, for the first time, the potential for miRNAs to physically interact with and change the function of non-transcriptionally related proteins. Specifically, we examined miR-1 and miR-21 as examples of miRNAs with the potential to interact directly with proteins in vivo. We chose these two miRNAs because they have important roles in cardiomyocyte biology. miR-1 is highly conserved throughout species, is the most abundant miRNA in human adult hearts (ϳ24%), and is also lethal in homozygous knockdown mice (13,15,49). miR-21 is one of the most consistently up-regulated miRNAs in cardiac hypertrophy, and inhibition of miR-21 enhances hypertrophic growth (13,47,48). We demonstrated that both miR-1 and miR-21 bind PLN and relieve inhibition of SERCA (Fig. 3). Intriguingly, the specific sequences of miR-1 and miR-21 are more effective at relieving inhibition than a random sequence mixture of 20-mer RNA. Both miR-1 and miR-21 completely relieve the inhibition by PLN, fully restoring SERCA function (Fig. 3E). Considering the high affinity of the miRNA-PLN interactions and the abundance of miRNAs in cardiac cells, it would be surprising if these interactions did not take place in vivo. The increased effectiveness of naturally occurring miRNAs suggests that they may have evolved to contain sequence elements that are particularly effective at relieving PLN inhibition. Together these results support the existence of a previously unreported regulatory role for miRNAs where they are able to modulate function through direct protein interactions.
SERCA Activity Assays-The reconstituted system containing either SERCA alone or SERCA with recombinant PLN was reconstituted in multilamellar lipid vesicles using DOPC and DOPE at a 4:1 DOPC:DOPE molar ratio (26). XNA solutions were prepared using nuclease-free water. XNAs were added directly to the reconstituted vesicles containing SERCA and PLN and incubated for 25 min prior to the ATPase assay. The SERCA-PLN complex was reconstituted to a final molar ratio of 700:10:1 lipid:PLN:SERCA. The coupled enzyme assay measures the calcium dependence of the ATPase activity of SERCA using a previously established method (25). The enzymatic activity at 37°C was monitored by following the decrease of NADH absorbance (A 340 ) with a SpectraMax microplate reader (Molecular Devices) equipped with a 96-well plate. The initial rate of SERCA activity was measured as a function of calcium concentration and then fit to the Hill equation (Equation 1) using Origin 9.1 where V is the initial rate; V max is the maximum rate; pCa is the negative logarithm of calcium concentration; pK Ca is the pCa value where V ϭ V max /2, and n is the Hill coefficient. Data sets were normalized to V max .
Affinity Capillary Electrophoresis-All experiments were performed on a Beckman Coulter commercial capillary electrophoresis system (P/ACE MDQ) with laser-induced fluorescence detection at ex ϭ 488 nm, em ϭ 520 nm. All sequences were labeled with FAM fluorophores for detection. FAM-labeled XNAs at a concentration of 10 nM were mixed with increasing concentrations of PLN and injected into a 40 cm ϫ 50-m fused silica capillary (Polymicro Technologies, Phoenix, AZ) using hydrodynamic injection (1 p.s.i. for 4 s). ACE separations were performed in reverse polarity mode at 30 kV for 9 min. Electropherograms were analyzed using Cutter 7.0 (52), where the peak heights and area of the free XNA peak were used to determine the K d , according to Ref. 53, where f b is the bound fraction; c is the maximum bound fraction; and [P] t and [D] t are the total concentrations of PLN and DNA, respectively. The denominator represents the free concentration of PLN.
Fluorescence Polarization-FP experiments were performed using a Synergy TM 2 microplate reader from BioTek Instruments, Inc. (Winooski, VT). All FP experiments with XNAs were performed using filter settings as follows: ex ϭ 485 Ϯ 20 nm, em ϭ 528 Ϯ 20 nm. Samples were made identically to the ACE samples. From these samples, 15-l aliquots were loaded into a Corning 384-well plate (3540 microplate from Corning Inc., Corning, NY). Parallel and perpendicular intensities were measured with a sensitivity setting of 120, and polarization values were calculated from these values. Calibrations were performed using the G-factor, which was calculated in the Gen 5TM software (BioTek Instruments, Inc., Winooski, VT). Bound fractions of XNA were determined according to where f b is the bound fraction; and P, P m , and P o are the measured polarizations of a sample, complex, and free ssDNA, respectively. Polarization of the complex (P m ) is the plateau value at a saturating concentration of PLN. The K d was then determined from fitting the data to Equation 2 using Origin 9.1. Overall fluorescence intensities of the samples were monitored, and the bound fraction was modified if the overall fluorescence intensity was biased, according to the published method (53). Competition Assay-Solutions of 100 nM FAM-labeled, random sequence, 20-mer RNA ϩ 100 nM PLN were prepared with 0 -500 nM unlabeled, random sequence, 20-mer RNA. Fluorescence polarization of each solution was measured to plot the fraction of labeled RNA dissociated from PLN versus mole fraction of unlabeled RNA (i.e. [RNA] unlabeled /[RNA] total ) Fluorescence polarization of 100 nM FAM-labeled, 20-mer RNA alone was measured as the "free labeled RNA" control.