Noninvasive and Quantitative Monitoring of the Distributions and Kinetics of MicroRNA-Targeting Molecules in Vivo by Positron Emission Tomography

MicroRNAs (miRNAs) are endogenous, small, noncoding ribonucleic acids (RNAs) that bind to the 3′ untranslated regions of messenger RNAs (mRNAs) and induce translational repression or mRNA degradation. Although numerous studies have reported that miRNAs are of potential use for disease diagnostics and gene therapy, little is known about their fates in vivo. This study elucidated the whole-body distributions and kinetics of intravenously administered miRNA-targeting molecules in vivo by positron emission tomography (PET) imaging. A 22-mer sequence targeting miR-15b was conjugated with three different chelators and labeled with gallium-68 (68Ga). These tracers were compared with a scrambled 22-mer sequence; 22-mer with two single base substitutions; anti-miR-34 22-mer; hexathymidylate (T6), a 6-mer sequence; and an unconjugated chelator. miR-15b was chosen as a target because it is important for bone remodeling. All three 68Ga-labeled anti-miR-15b molecules had similar biodistributions and kinetics, and they all accumulated in the bones, kidneys, and liver. The bone accumulation of these tracers was the highest in the epiphyses of long tubular bones, maxilla, and mandible. By contrast, the scrambled 22-mer sequence, the 6-mer, and the unconjugated chelator did not accumulate in bones. PET imaging successfully elucidated the distributions and kinetics of 68Ga-labeled chelated miRNA-targeting molecules in vivo. This approach is potentially useful to evaluate new miRNA-based drugs.


■ INTRODUCTION
Ribonucleic acid interference is a post-transcriptional gene silencing mechanism discovered by Fire and coworkers in 1998. 1 In this mechanism, an exogenous double-stranded RNA silences the corresponding messenger RNA (mRNA). These exogenous RNA molecules are called short interfering RNAs. microRNAs (miRNAs) function in the same way but are endogenously expressed. miRNAs are transcribed from exonic or intronic deoxyribonucleic acid and then exported from the nucleus. The enzyme Dicer digests pre-miRNAs into fragments containing 16−23 nucleotides in the cytoplasm. These short mature miRNA fragments are incorporated into the RNAinduced silencing complex and recognize complementary mRNAs. A component of this complex, termed argonaut 2, cleaves mRNAs and thus prevents subsequent translation. 2 miRNA regulation is an important mechanism in various cells and tissues. For example, blockade of the miRNA maturation pathway via tissue-specific knockout of Dicer elicits a multitude of effects in animal models, ranging from impaired organ function to embryonic lethality. 3−6 Complete knockout of an miRNA is not necessary to elicit effects and may have unwanted consequences. A specific miRNA can be overexpressed by transfection or inhibited by various antisense miRNAs (anti-miRs). 7−10 The main problems associated with the use of short interfering RNAs and miRNAs as therapeutic agents are their limited stabilities in vivo, difficulties in targeting their delivery to a desired tissue, their modest cellular uptake, and possible off-target effects. Several strategies have been developed to overcome these difficulties, including the use of viral vectors, liposomes, chemical modifications, and nanoparticles, as reviewed by Gao and coworkers. 11 The concentrations of injected nonmodified oligonucleotides are the highest in the liver and kidneys, followed by the bone marrow, adipocytes, and lymph nodes. Cellular uptake of these oligonucleotides is mediated by endocytosis. 12 This study examined three chemically modified antisense oligonucleotides targeting miR-15b. miR-15b belongs to the miR-15 family and plays important roles in various tissues, including the bones. 13−16 The novel anti-miR-15b oligonucleotides were conjugated with three different chelators to evaluate the chelator effect on biodistributions and biokinetics of the target oligonucleotide (anti-miR-15b). These chelators were chosen because they bind the positron emitter gallium-68 ( 68 Ga) and allow straightforward synthesis of the oligonucleotide conjugates entirely on a solid phase. In previous positron emission tomography (PET) studies, oligonucleotides were usually conjugated with the macrocyclic chelator 4-(4hydroxybut-1-oxy)phenyl 1,4,7,10-tetraazacyclododecane-N′,N″,N‴,N⁗-tetraacetic acid (DOTA). 17−19 In addition to DOTA, we used two triazacyclononane-based chelators, i.e., 4-(4-hydroxybut-1-oxy)phenyl 1,4,7-triazacyclononane-N,N′,N″triacetic acid (NOTA) and tris[4-(4-hydroxybut-1-oxy)phenyl] 1,4,7-triazacyclononane-N,N′,N″-triacetic acid (NOTA(mod)) ( Figure 1). NOTA has been previously used to label oligonucleotides. 19−24 As mentioned in a review by Gao and coworkers, 15   [ 68 Ga]Ga-NOTA-anti-miR-15b(scr) contained the same nucleotides as [ 68 Ga]Ga-NOTA-anti-miR-15b but in different order (5′-AUG ACA GGA CUG UUA GAU CUU C-3′) and lacked a target in the rat genome according to a BLAST search. The RNA sequence in [ 68 Ga]Ga-NOTA-anti-miR-15b(2SBS) was 5′-UGU AAA CCA UGA UGU GCU ACG A-3′. One nucleotide was changed from guanine to adenine and another was changed from uracil to guanine at the 3′-end of this sequence. The RNA sequence in [ 68 Ga]Ga-NOTA-anti-miR-34 was 5′-ACA ACC AGC UAA GAC ACU GCC A-3′. The effect of the 3′-68 Ga-NOTA/DOTA chelate to the RNA duplex is generally slightly stabilizing as we have previously determined by T m (UV) measurements (cold Ga 3+ used as a ligand, the stabilization is of unspecific origin). 23 It may be emphasized that the oligonucleotides in this study were 2′-O-methyl oligoribonucleotides without cell/tissue-specific targeting agents. While modest cellular uptake may be expected, the imaging data may more reliably demonstrate the importance of the sequence (i.e., anti-miR-15b vs anti-miR-15b(2SBS), anti-miR-15b(scr), and anti-miR-34) and plausible chelator effect (NOTA vs NOTA(mod) and DOTA) on the biodistribution and kinetics. The structures of the studied molecules are presented in Figure 1.
The oligonucleotides were synthesized by an automatic DNA/RNA synthesizer using solid-supported DOTA, NOTA, and NOTA(mod) chelators as we previously reported. 19 Thereafter, the oligonucleotides were released from the supports by an optimized two-step cleavage protocol (treat-

Molecular Pharmaceutics
Article ment with 0.1 mol/L aqueous NaOH for 3 h at 55°C and then incubation with concentrated ammonia overnight at 55°C), purified by reversed-phase high-performance liquid chromatography (RP-HPLC), and lyophilized to dryness. The authenticity of the RNAs was verified by electrospray ionization time-of-flight mass spectroscopy [MS (ESI-TOF)] ( Figure S1 and Table S1). 68 Ga was obtained in the form of [ 68 Ga]GaCl 3 using a 68 Ge/ 68 Ga generator (Eckert & Ziegler, Valencia, CA) by elution with 0.1 M HCl composed of 30% ultrapure HCl and ultrapure water. Sodium acetate (18 mg) was added to 500 μL of [ 68 Ga]GaCl 3 (or 3 mg was added to 100 μL) to yield a final concentration of 0.4 M sodium acetate. The pH was adjusted to approximately 3.5 with 2 M HCl. Thereafter, a NOTA/ DOTA-conjugated oligonucleotide (5−12 nmol, dissolved in deionized water to yield a stock solution of 0.4, 0.5, or 1 mM) was added and the reaction mixture was incubated at 95°C for 10−15 min. Radiochemical purity was determined by reversedphase radio-HPLC using a Jupiter C18 column (300 Å, 5 μm Dynamic Whole-Body PET and CT Imaging. The biodistributions of the tracers were evaluated by performing dynamic PET imaging with a high-resolution research tomograph (HRRT) camera (Siemens Medical Solutions, Knoxville, TN) cross-calibrated with a dose calibrator (VDC-202; Veenstra Instruments, Joure, Netherlands) and a 1480 Wizard 3″ γ counter (PerkinElmer, Turku, Finland). Two rats were imaged simultaneously. For PET imaging, rats were anesthetized with isoflurane (Baxter, 3−4% with an airflow of 600−700 mL/min for induction and 1.5−2.5% with an airflow of 400−500 mL/min for maintenance) and placed on a warm pallet. A catheter was inserted into the tail vein (24-gauge cannula; BD Neoflon, Becton Dickinson Infusion Therapy AB, Helsingborg, Sweden). After a 6 min transmission scan using a 137 Cs point source for attenuation correction, 68 Ga tracer (19.1 ± 4.8 MBq) was intravenously injected as a bolus. Dynamic imaging was started at the time of injection and continued for 60 min.
The PET images presented in this study represent the mean of all time frames. To create an average image of the 30 middle planes, the sum plane and arithmetic functions of Vinci 2.54 software (Max-Planck-Institut fur neurologische Forschung, Cologne, Germany) were used.
PET/CT images were processed with Inveon Research Workplace 4.0 software (Siemens), fused, transformed into an maximum a posterior (MAP) projection, scaled, and rotated into the same position.
Pharmacokinetic parameters were calculated using plasma TACs extracted from SUV-normalized PET images. Plasma TACs were extracted by drawing a spheroid volume of interest on the most radioactive area of the heart in the first frames where blood radioactivity was the highest. At this point, the TAC represented the whole-blood radioactivity. To generate plasma TACs, the whole-blood curve was multiplied by 1.80 for correction. This estimation was based on the plasma/blood radioactivity ratio measured ex vivo (Table S2). The area under the curve (AUC 0→∞ ) from 0 to infinity, total clearance (Cl T ), elimination rate constant (k el ), and plasma half-life (t 1/2 ) were calculated using Paucinf 1.4.2 (Turku PET Centre, Finland, http://www.turkupetcentre.net/software/list.php). The Paucinf program calculates the AUC between 0 and 55.8 min using the linear trapezoidal rule. Remaining AUCs (between 55.8 min and infinity) were calculated with the Paucinf program using log-linear regression followed by the linear trapezoidal rule. The maximum concentration (C max )

Molecular Pharmaceutics
Article and time of maximum concentration (T max ) were defined from plasma TACs.
Ex Vivo Analysis. After PET imaging, the rats were euthanized and blood (by cardiac puncture), urine (by bladder puncture), and tissue samples were excised and weighed. Total radioactivity was measured using a γ counter (1480 Wizard 3″, PerkinElmer, Turku, Finland). Measurements were converted from counts per minute to kilobecquerel using the calibration constant of 68 Ga. All radioactivity values were decay-corrected to the time of injection.
Statistical Analysis. Differences between tracers in ex vivo biodistribution and TAC results were analyzed by the unpaired Student t-test using the GraphPad online calculator (https:// www.graphpad.com/quickcalcs/). When determining differences between TACs, only the last measurement points were used in the comparison. P values of less than 0.05 were considered statistically significant.
In Vivo Imaging.  Table S3), which demonstrated that the chelator had no significant effect on the distribution. Radioactivity accumulated most in the kidneys, urinary bladder, and liver. Some radioactivity was also observed in the salivary glands.  Table S3).
Ex Vivo Measurements. Ex vivo measurements were consistent with the PET images. Radioactivity in the kidneys, bones, and bone marrow was much higher in rats injected with

■ DISCUSSION
The present study evaluated the biodistributions of intravenously administered conjugated anti-miR-15b in rats. miR-

Molecular Pharmaceutics
Article 15b functions in various tissues including bones and is expressed in mesenchymal stem-cell-derived osteogenic cells. 15,16 Although 68 Ga-labeled anti-miR-15b has been previously investigated, this study evaluated the fate of anti-miR-15b tracers in more detail, with a particular focus on blood pharmacokinetics and the specificity of bone uptake. 19,22 In vivo PET imaging and ex vivo analysis of excised tissue samples demonstrated that bone and kidney uptake of [ 68 Ga]Ga-NOTA-anti-miR-15b markedly differed from that of [ 68 Ga]Ga-NOTA-T6 and 68 Ga-NOTA. However, it is unclear whether the RNA sequence itself or the length of the RNA oligonucleotide is responsible for these differences. Short RNA molecules accumulate in the kidneys and liver. 25 Nephrons filter short [ 68 Ga]Ga-NOTA-T6 molecules (2.3 kDa), whereas larger 22-mer sequences (7.9−8.2 kDa) are probably filtered and reabsorbed later in the proximal tubuli. The renal filtration cutoff molecular weight for globular proteins is about 70 kDa. 26 Wu and coworkers studied the biodistributions of phosphorothioate oligonucleotides comple-mentary to rat chromogranin A with various sequence lengths (30-, 20-, 12-, and 6-mer). The 20-mer in their study accumulates in the kidneys and liver, similar to the 22-mers used in the present study; however, the levels of 6-mers differed between the previous study and the present study. In contrast with the findings of Wu et al., we demonstrated that [ 68 Ga]Ga-NOTA-T6 was not present at high levels in the kidneys but accumulated in the intestines (Figures 2G, 3G, Table S3).
The question remains whether the bone accumulation observed is due to the anti-miR-15b sequence or the length of the RNA oligonucleotide and the properties of the chelator and 68 Ga. The control measurement using three different chelators for anti-miR-15b confirmed that the chelator played a minor role in the distribution, which is in the line with our previous studies. 19 However, 68 Ga accumulates in bone and consequently the radioactivity detected in bones may correspond to free 68 Ga detached from the chelator, rather than to the oligonucleotide itself. To exclude this possibility,

Molecular Pharmaceutics
Article quality control for radiochemical purity was performed for all tracers using radio-HPLC. Moreover, bone radioactivity was not observed in rats injected with [ 68 Ga]Ga-NOTA-T6 and [ 68 Ga]Ga-NOTA, which harbored the same chelator and labeled [ 68 Ga]Ga-NOTA-anti-miR-15b. A scrambled sequence was synthesized to determine if the bone accumulation is due to the anti-miR-15b sequence. This scrambled sequence contained the same nucleotides as the anti-miR-15b sequence; however, the order of the nucleotides was randomized and this sequence lacked a specific target in the rat genome according to a BLAST search. PET/CT imaging revealed that the level of [ 68 Ga]Ga-NOTA-anti-miR-15b(scr) in the epiphyses was much lower than that of [ 68 Ga]Ga-NOTA-anti-miR-15b. However, ex vivo analyses of larger rats using an HRRT camera showed that the level of [ 68 Ga]Ga-NOTA-anti-miR-15b(scr) was only slightly lower than that of [ 68 Ga]Ga-NOTA-anti-miR-15b in bones. This discrepancy is because the excised bone samples and bone ROIs in PET/CT analyses did not correspond to the same region. In vivo PET/CT imaging indicated that radioactivity was mainly present in the epiphyses of long bones. The ex vivo samples consisted of the entire tibia, not just the epiphyses. TACs can be used to objectively compare the level of radioactivity between samples. The SUVs, which take into account the injected radioactivity dose and animal weight, of [ 68 Ga]Ga-NOTA-anti-miR-15b(scr) and [ 68 Ga]Ga-NOTAanti-miR-15b significantly differed (P < 0.0001, Figure 4A).
The quantities of synthesized RNA conjugates were limited; therefore, we were unable to perform in vivo stability analyses using rat plasma samples. This would have required larger quantities of the injected tracers because the sensitivity of radio-HPLC is limited. Instead, we modeled pharmacokinetics using plasma radioactivity curves derived from dynamic PET data, which represented the total radioactivity concentration as a function of time after injection. The sample size in each group was limited to 4−6 animals. Both female and male rats were analyzed, and the results were pooled because no differences were observed between genders.
In conclusion, PET imaging successfully elucidated the distributions and kinetics of 68 Ga-labeled chelated miRNAtargeting molecules in vivo. Our results revealed that intravenously injected 68 Ga-labeled anti-miR-15b is specifically accumulated in the bone. This study confirms that PET imaging is a useful approach to evaluate new miRNA-based drugs.