Using coded lipid nanoparticles to track how they move in liver zones and which cells they target most

To create barcoded Cre recombinase mRNAs, we inserted unique 10-nucleotide barcodes and a 20-nucleotide PCR adapter sequence between the stop codon and 3’ untranslated region of a Cre recombinase in vitro transcription (IVT) template DNA (Fig. 1b). The adapter sequence enabled unbiased PCR amplification of the barcoded region, while the barcodes contained a unique identifying sequence that can be readout using qPCR and/or next-generation sequencing. To ensure efficient translation in vivo, mRNAs were synthesized with a template encoded polyA tail, 100% N¹-methyl pseudo-uridine substitution, and included a m7G CAP1 structure. After IVT, the size and integrity of barcoded mRNAs were examined using an Agilent Tapestation, which yielded single bands at the expected molecular weight (Supplementary Fig. 1a, b).

We examined whether the barcodes and adapter sequences might influence the functional activity of the Cre mRNA. To study this in vivo, we delivered barcoded or non-barcoded Cre mRNAs to Ai14 mice using liver-targeting LNPs¹¹ (Supplementary Table 1). We chose to utilize vortex mixing because this method is compatible with high throughput, automatable workflows, and produces high-quality LNPs with comparable in vivo properties to microfluidic mixing⁸. Dynamic light scattering was used to measure the size and dispersity of the barcoded LNPs (Supplementary Fig. 1c). Three days after LNP administration, the Cre-mediated DNA editing efficiency and tdTom expression was quantified using ex vivo imaging and flow cytometry. We found nominal differences in efficiency between the barcoded and non-barcoded Cre mRNAs (Supplementary Fig. 1d, e).

Base modifications such as the N¹-methyl pseudo uridine \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$({\Psi }^{1})$$\end{document}(Ψ1) used here are required for efficient translation of LNP-delivered mRNAs in vivo^25–27, but their effects on standard tissue mRNA quantification techniques are not well established. To ensure that PCR-based detection of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\Psi }^{1}$$\end{document}Ψ1 IVT mRNA is a reliable and linear quantification method, we generated a standard curve by spiking a known mass of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\Psi }^{1}$$\end{document}Ψ1 IVT mRNA into total RNA isolated from mouse liver prior to performing reverse transcription, bypassing variables such as LNP delivery and RNA purification. The mass of spiked IVT mRNA spanned five orders of magnitude, from 10⁻⁵ to 10⁰ w/w of total RNA. We then performed qPCR using primers that anneal to the PCR adapter site and ~100 bp upstream of the stop codon. Linear regression analysis found near-perfect correlation between the mass of IVT mRNA in the RT reaction and the resulting fold changes (Supplementary Fig. 2a). To confirm that mRNA LNP delivery could be quantified by extracting RNA directly from target tissues, we intravenously administered liver-targeting LNPs at doses ranging from 0.0001 to 1 mg/kg and collected major organs after 24 h. The qPCR signal in the liver shared a strong correlation with administered doses in vivo, R² = 0.986 (Supplementary Fig. 2b). Additionally, the tdTom response in the liver was also well correlated with the administered dose, R² = 0.974 (Supplementary Fig. 2c).

LNP delivery of mRNA involves numerous steps, including tissue distribution, cellular uptake, endosomal escape, and mRNA translation. Prior studies that used DNA- and RNA-based barcodes selected downstream timepoints associated with blood clearance and protein expression^18,28,29, events that could occur minutes to hours after important upstream events take place. Since both DNA and mRNA-based barcode readouts rely on the assumption that biodistribution patterns correlate with functional tropism, we sought to systematically optimize key experimental parameters, including dose and tissue collection time. This approach enabled us to interrogate the relationship between biodistribution and functional activity across a range of conditions. We performed detailed assessments spanning multiple doses and previously uncharacterized early and late timepoints. We utilized SORT LNPs for this goal because the distinct organ delivery profiles of these formulations provided a controlled framework to investigate factors such as detection thresholds, saturation dynamics, and the extent to which mRNA biodistribution correlates with downstream protein expression.

Barcoded Cre mRNAs packaged in Lung, Liver, and Spleen SORT LNPs were individually administered IV to Ai14 mice at mRNA doses between 0.001–1.0 mg/kg. Cre mRNA biodistribution was subsequently examined in the lung, liver, and spleen at 1- and 24-h using qPCR. Because successful delivery of Cre mRNA leads to activation of tdTom protein, fluorescence microscopy combined with qPCR quantification can correlate mRNA biodistribution that is nonfunctional (no tdTom expression) and functional (tdTom expression). The tdTom+ area fraction of fluorescent micrographs in regions defined by DAPI signal was used to quantify the tdTom response, a measurement that assumes approximately linear scaling with the number of tdTom+ cells.

Consistent with previous reports, Lung SORT LNPs mediated efficient tdTom expression in the lung across the dose range but also showed some tdTom signal in the liver and spleen at higher doses. (Fig. 2f and Supplementary Fig. 3b). Cre mRNA levels measured by qPCR showed strong dose dependence at both 1 h and 24 h, but were ~100-fold higher at 1 h (Fig. 2g, h). Lung tdTom signal correlated with lung Cre mRNA levels at both timepoints; however, correlation was modestly stronger at 1 h (Fig. 2i, j). While the overall distribution of Cre mRNA resembled the tdTom distribution at both timepoints, more liver and spleen signal were measured by qPCR at 24 h than at 1 h (Supplementary Fig. 3k). These results demonstrated that the observed functional activity of Lung SORT LNPs were reflected in their biodistribution, and suggest that 1 h measurements enable increased signal intensity and modestly improved correlation to protein production versus later measurements.

Spleen SORT LNPs mediated efficient and specific tdTom expression in the spleen, with very few tdTom-positive cells in the liver, and no detectable signal in the lung. The tdTom signal was concentrated in the marginal zones of the spleen, a distinctive distribution compared to Liver SORT or Lung SORT (Fig. 2k and Supplementary Fig. 3c). Cre mRNA levels measured by qPCR showed strong dose dependence and were ~10-fold higher at 1 h than at 24 h (Fig. 2l, m). Spleen tdTom signal correlated with spleen Cre mRNA levels at both timepoints; however, correlation was modestly stronger at 1 h (Fig. 2n, o). The overall distribution of Cre mRNA was similar at both timepoints, showing primarily spleen localization (Supplementary Fig. 3l). Taken together with Liver SORT and Lung SORT, these experiments demonstrated on-target enrichment of Cre mRNA in target organs that improved when measured at 1-h post-injection versus 24 h.

To explore whether these findings could impact a multiplexed barcoding experiment, we designed a test case in which Liver, Lung, and Spleen SORT LNPs were formulated with distinct barcodes and administered as a pool to a single mouse at 0.1 mg/kg per formulation. We harvested the liver, lungs, and spleen at both 1 and 24 h after injection and sequenced barcodes from bulk tissues. Samples harvested at 1 h showed substantial on-target enrichment, and sequencing results correctly ranked the observed functional activity of all three formulations in each organ. In contrast, the on-target enrichment at 24 h was weaker (Supplementary Fig.). This study demonstrated that multiplexed barcoding can prioritize formulations with functional activity in target organs and showed that samples harvested at 1-h post-injection provided a more optimized readout. 3m, n

Having identified LNP kinetics as a variable that influenced the relationship between mRNA biodistribution and protein production, we next sought to more precisely describe (1) tdTom activation and (2) mRNA kinetics utilizing the mRNA barcode platform. To identify relevant tissue collection timepoints for tdTom expression, we established a timeline of reporter expression following LNP injections. We delivered 0.1 mg/kg barcoded Cre mRNA to Ai14 tdTom mice using either Liver, Lung, or Spleen SORT LNPs. We collected the lung, liver, and spleen from 10 min to 72 h post injection. Although scattered cells in target tissues were tdTom positive 12 h after dosing, robust signal became detectible after 24 h that continued to expand between 24 and 72 h post-injection (Supplementary Fig.). We quantified Cre mRNA levels in total RNA extracts of the lung, liver, and spleen at each timepoint using qPCR. Peak mRNA levels were observed at 10 min across tissues, which was the earliest collection timepoint, and diminished substantially at each subsequent timepoint. Although detectable out to 72 h, the majority of transcript abundance was observed within the initial 3 h (Supplementary Fig.). 4a–c 4d–f

We observed a reduction of circulating Lung SORT barcodes in the blood just 30 s after injection versus liver SORT and spleen SORT (Fig. 3b). Lung SORT barcodes encoding the 30 s timepoint were the most abundantly detected in the lung, indicating rapid lung association (Fig. 3c). Lung SORT blood clearance occurred more slowly thereafter, which could have been due to reduced liver clearance (Fig. 3d). Lung SORT was enriched versus other LNPs in the lung and heart from 30 s to 3 h and in the kidney from 5 min to 3 h (Fig. 3c, f, g). Spleen SORT blood clearance temporally coincided with accumulation in the liver and spleen, with Tmax at 5 and 10 min, respectively, and splenic enrichment from 5 to 30 min (Fig. 3b, e). Similarly, Liver SORT LNP blood clearance temporally coincided with rapid liver accumulation, reaching Tmax by 10 min, and Liver SORT LNPs remained enriched in the liver from 10 min to 3 h (Fig. 3b, d). The fast mRNA kinetics we observed show that LNPs can rapidly accumulate in tissues where they have functional activity.

qPCR analysis using barcode-specific primers (Supplementary Fig. 5a) revealed distinct spatiotemporal distributions for each SORT LNP. Liver SORT barcodes were initially enriched in the liver but became more abundant in the spleen by 3 h (Fig. 3h). Given that Liver SORT LNPs did not efficiently induce tdTom expression in the spleen, splenic accumulation could indicate an immune clearance mechanism³⁰. Lung SORT barcodes remained predominantly in the lung, with minor splenic presence (Fig. 3i). Spleen SORT barcodes accumulated early in both liver and spleen, with more sustained signal in the spleen, leading to increased splenic proportion over time (Fig. 3j). These findings highlight the dynamic and formulation-dependent biodistribution of SORT LNPs. Area under the curve (AUC) calculations showed that, by integrating over time, the measured biodistributions resemble earlier tdTom results (Fig. 3k).

To verify that multiplexed kinetics results are comparable to single-injection experiments, we performed a 72-h time course using the barcoded kinetics approach to quantify Cre mRNA levels in the liver over time. We compared the results to single injections given at the same-timepoints and found that the resulting curve shapes were similar (Supplementary Fig.). We also asked whether multiple injections containing pooled LNPs appreciably altered LNP biodistributions. We compared biodistributions obtained in the context of multiple pooled injections side-by-side with the same-dose, same-timepoint single injections (Supplementary Fig.). We found that the barcode distributions from multiple pooled injections showed similar trends as single-injection experiments. We note that the dose and injection volumes were minimized in these experiments to mitigate potential effects from multiple pooled injections. 5b 5c

To further understand the relationship between biodistribution at early timepoints and functional mRNA delivery, we conducted a follow-up multiplexed pooled kinetics experiment in which a single mouse received multiple injections containing time-coded barcodes delivered by liver SORT, lung SORT, and spleen SORT LNPs. In addition to isolating barcodes in total RNA from bulk tissue, we also profiled ribosome-associated barcodes in the liver, lung, and spleen by adapting and employing a ribosome immunoprecipitation method³¹. The barcode kinetics in total RNA from these tissues were similar to the preceding experiments, with a rapid accumulation phase followed by subsequent elimination. However, the onset of ribosome- associated barcodes was delayed relative to total RNA, most dramatically in the lung, where despite immediate signal peak, ribosome-associated barcodes peaked at 30 min post injection (Supplementary Fig. 5d–f). These results show that the rapid accumulation and enrichment of SORT LNPs in target organs is associated with functional delivery events, and that this functional activity can peak quickly.

To validate qPCR-based quantification of LNP-mRNA biodistribution, we performed fluorescence microscopy using Cy5-labeled mRNA loaded in Liver and Lung SORT LNPs. At both 30 s and 5 min post-injection, we observed distinct intracellular puncta consistent with endosomal localization, indicating rapid cellular uptake (Supplementary Fig.). Confocal microscopy revealed Cy5 signal that co-localized with elF4E, a cytosolic translation initiation factor that interacts with the 5-prime cap of mRNA, in liver samples fixed 20 min after Liver SORT LNP administration (Supplementary Fig.). These findings confirm that mRNA delivery via LNPs initiates rapid tissue-level biodistribution in vivo, supporting the interpretation of early qPCR signals as biologically meaningful. 5g, h 5i

The use of barcodes for pharmacological profiling enabled us to compare and capture the rapid accumulation kinetics of multiple LNP chemistries head-to-head and highlighted uptake kinetics as a key parameter to optimizing LNP delivery. Furthermore, this approach solved technical challenges such as the requirement for an internal standard to normalize between timepoints, as the barcodes encode the time-axis. This is particularly important for kinetic studies in tissues like blood, or fractionated samples like ribosomes, which lack an obvious internal standard for normalization. Further, we were able to conduct this study using far fewer animals than would normally be required.

The ability to distinguish the biological fates between different LNP chemistries is critical to an optimized discovery process. We next asked whether our mRNA barcoding workflow could identify differences between similar LNP formulations. Previously, our lab observed that by titrating the molar fraction of the permanently cationic lipid DOTAP into LNPs, the resulting functional LNP distribution could be gradually shifted from the liver to the lung as a function of DOTAP inclusion percentage¹¹. Mechanistically, this phenomenon is at least partially explained by an endogenous targeting mechanism, whereby DOTAP LNPs adsorb Vitronectin from plasma to subsequently internalize in αVβ3 integrin (Vitronectin receptor, Vtn-R) expressing cells that are enriched in the lung^15,16,32. We decided to use this systematic set of characterized LNPs to test whether our barcoding platform could recapitulate these findings and potentially reveal new insights. To accomplish this, we formulated Liver SORT and DOTAP-containing LNPs with unique barcodes, then administered the pooled series to reporter mice. We collected major organs 1 h post injection and processed samples according to our established barcoding workflow. Separately, we administered each formulation to mice individually with unbarcoded Cre mRNA, then harvested tissues after 24 h to compare with the barcoded data.

In our multiplexed analysis, mRNA barcodes also showed a general trend of redistribution from the liver to the lung as a function of DOTAP percentage. Additionally, 30% DOTAP LNP-mRNA barcodes were less abundant in the liver than 20 or 40% DOTAP LNPs. Further, DOTAP LNP barcodes were also enriched versus Liver SORT in the kidney (Fig. 4b and Supplementary Table 2). We plotted the normalized distribution of tdTom signal against the normalized barcode counts and found a strong linear correlation (Fig. 4c). We next asked whether isolating cell populations based on their functional status as tdTom positive or negative would enable further differentiation between formulation activity at the barcode level. To activate the tdTom reporter, we administered the pooled LNP series containing Liver SORT and DOTAP LNPs to Ai14 mice and allowed 24 hours for reporter expression. Prior to harvesting the samples, a second dose of the same LNP pool was given to ensure barcodes were collected during the optimized kinetic window. We then used flow cytometry to collect tdTom-positive and negative cells in the lung, spleen, and kidney (Fig. 4d). We found that the set of barcodes enriched in the tdTom+ populations were associated with enriched functional activity, while barcodes with reduced tdTom activity were enriched in the tdTom-negative populations (Fig. 4e). These multiplex results, performed in a single mouse, recapitulate tdTom reporter results that required 6 mice. Moreover, the detection of kidney tropism associated with DOTAP LNPs demonstrates that our platform can efficiently detect uncharacterized activity within small populations of cells beyond the liver and lung.

To show that these findings are applicable to commercially available LNP formulations currently in clinical use, we examined whether Cre mRNA barcodes could distinguish the functional activity of LNPs based on ALC-0315, SM-102, or LP01 ionizable lipids. To accomplish this, we formulated these LNPs using the clinically described molar ratios of ionizable and helper lipids to carry firefly luciferase mRNA or barcoded Cre mRNA. LNPs were administered IV at 0.1 mg/kg luciferase or 0.01 mg/kg of barcode. We quantified total radiance in ex vivo livers and found that ALC-0315 LNPs resulted in the most luciferase protein production, followed by LP01 and SM-102-based LNPs. We pooled the formulations carrying barcoded Cre mRNAs and administered the pool to Ai14 tdTom reporter mice. The barcode readouts from livers collected 1-h after multiplexed LNP administration in all three pooled mice agreed with the luminescence results obtained using individual mice (Supplementary Fig.). 6a–c

Interestingly, in a new insight, we observed that the activity of Liver SORT LNPs was biased toward periportal and midlobular zones, and excluded from pericentral zones, as tdTom reporter signal showed limited overlap with the pericentral hepatocyte marker gene (GS). In contrast, LNPs containing 10% DOTAP retained periportal activity, but also had expanded distribution that overlapped with GS+ pericentral hepatocytes. Increasing DOTAP beyond 10 percent de-targeted the liver altogether (Fig. 5b, c). Line scans performed across the central to portal vein axis in a lower dose condition showed that Liver SORT LNPs preferentially drive tdTom expression in zones 1 and 2, with reduced zone 3 activity, while the distribution of 10% DOTAP LNPs appears not zonally biased (Fig. 5d).

To recapitulate this finding at the barcode level, we administered a pool of barcoded Liver SORT and DOTAP LNPs, then used flow cytometry to isolate zonated populations based on surface E-cadherin expression³⁴. We determined that liver SORT barcodes preferentially distributed to Ecad-high, periportal cells, while 10% DOTAP LNPs were enriched in Ecad-low pericentral cells (Fig. 5e). These liver metabolic zone-specific observations highlight an important consideration for therapeutic applications of LNPs in the liver and could also be leveraged as a tool to study liver phenomena such as metabolic zonation. Moreover, this experiment demonstrates the utility of the barcode platform to distinguish cell-type-specific trends in nanoparticle-mediated mRNA delivery in a multiplexed experiment.

Barcoded Cre mRNAs were packaged in 5-component SORT lipid nanoparticles using previously described lipids and formulation conditions^8,11. Briefly, the 5A2-SC8 ionizable amino lipid, DOPE (Avanti Polar Lipids Cat. No. 850725), Cholesterol (Sigma-Aldrich Cat. No. C3045), DMG-PEG-2000 (Avanti Polar Lipids Cat. No. 880151), and a SORT lipid were dissolved in ethanol at previously optimized lipid molar ratios of 15:15:30:3:X respectively, where X is either 15.75 (20%) DODAP (Avanti Polar Lipids, 890850 O), 63 (50%) DOTAP (Avanti, 890890 P), or 15.75 (20%) 18PA (Avanti, 840875 C). The lipid-containing organic phase was rapidly mixed with an aqueous phase containing RNA diluted in 10 mM pH 4 citrate buffer at a 3:1 (aq.:EtOH) ratio using vortex mixing⁸ or a NanoAssemblr Benchtop system (Precision Nanosystems) at a 40:1 lipid:nucleic acid mass ratio. LNPs were dialyzed in PBS for at least 2 h prior to use.

Barcodes and a PCR adapter were cloned into an in vitro transcription (IVT) vector encoding Cre recombinase with a DNA-encoded polyA tail using Q5 SDM kit according to the manufacturer's instructions (New England Biolabs, E0554S). Purified barcoded Cre IVT template plasmids were linearized for 4 h via restriction digestion, and mRNAs were transcribed using Megascript SP6 IVT kit (Thermo Fisher, AM1330) with 100% N¹-methyl pseudo uridine substituted for uridine. An m7G (CAP1) structure was added to purified mRNA using ScriptCap Cap1 Capping System (CellScript, C-SCCS1710) according to the manufacturer's instructions.

Organs were minced using a sterile blade and homogenized in 250 µL of 1× digestion medium (45 units/µL collagenase I, 25 units/µL DNase I, and 30 units/µL hyaluronidase). The solution was transferred into a 15 mL tube that contained 5–10 mL of 1× digestion medium. Next, the solution containing cells was filtered using a 70 µm filter and washed once with 1× PBS. A cell pellet was obtained by centrifuging for 5 min at a speed of 300×g at 4 °C. The supernatant was removed, and the cell pellet was resuspended in 2 mL of 1× red blood cell lysis buffer (BioLegend, 420301) and incubated on ice for 5 min. After incubation, 4 mL of cell staining buffer (BioLegend) was added to stop red blood cell lysis. The solution was centrifuged again at 300×g for 5 min to obtain a cell pellet. The cells were washed twice with 1 mL 1X PBS, then resuspended in 500 µL 1X PBS for flow cytometry analysis. Ghost Dye Red 780 (Tonbo Biosciences, 13-0865-T500) was used to discriminate live cells. Where applicable, the sample was subjected to antibody staining in preparation for flow cytometry. The cells were analyzed using a LSRForessa SORP machine (BD Biosciences).

About 30 mg of whole tissue was added directly to 1 mL of ice-cold TRIzol reagent (Thermo Fisher, 15596026) and homogenized in a bead beater homogenizer for 45 s. The homogenate was allowed to lyse on ice for 15 min, then 0.2 mL of chloroform was added and vortexed for 15 s. Phase separation proceeded on ice for 5–10 min, then samples were centrifuged at 12,000×g for 15 min. The upper aqueous phase containing RNA was carefully pipetted into a separate tube, and the RNA was precipitated using one volume of isopropanol, then centrifuged for 15 min at 12,000×g. RNA pellets were resuspended in nuclease-free water, and the RNA yield and purity was calculated using UV-Vis spectrophotometry.

Ribosome purification proceeded as previously described³¹ with modifications to adjust for larger tissue volumes. Organs were digested as described for flow cytometry in the presence of 0.1 mg/mL cycloheximide. Cells were filtered through a 70-uM strainer then lysed in 100 µL lysis buffer (20 mM Tris–HCl, pH = 7.4, 250 mM NaCl, 15 mM MgCl₂), 0.1 mg/mL cycloheximide, 0.5% Triton X-100, 1 mM DTT, 0.5 U/µL RNAse inhibitor (Thermo Fisher, AM2696), 0.024 U/µL TURBO DNase (Life Technologies, AM2222), 1 × Protease Inhibitor (Sigma, P8340)). About 2 µL of biotinylated y10b antibody (Thermo Fisher, MA516060↗) was added to each sample and incubated for 4 h on an inversion rotor at 4 °C. After the incubation, 100 uL of prewashed streptavidin-coated magnetic beads (NEB S1420S) were added to each sample. Beads were washed 3x in 1 mL of wash buffer (20 mM Tris–HCl (pH 7.4), 250 mM NaCl, 15 mM MgCl₂, 1 mM DTT, 0.1 mg/mL cycloheximide, and 0.05% v/v Triton X-100). Samples were incubated with beads for 30 min at 4 °C with inversion. Samples were applied to a magnet and washed 3x in 1 mL of wash buffer with resuspension between washes. After the final wash, beads were resuspended in 50 uL of RNAse-free water and heated in a thermocycler at 65 °C for 10 min. The magnetic beads were removed using a magnet, and the supernatant containing ribosome-associated RNA proceeded directly to reverse transcription. Barcodes were quantified using qPCR.

RNA extracts were reverse transcribed into cDNA using the iSCRIPT cDNA synthesis kit (Bio-Rad, 1708891) according to the manufacturer's instructions. cDNA libraries were used as template DNA for quantitative PCR using iTaq supermix (Bio-Rad, 1725124) as follows: for the detection of all barcoded Cre mRNA irrespective of barcode, the forward primer sequence used was:

5’-ACCGGAGATCATGCAAGCTGG-3’

the reverse primer sequence was:

5’-TTGCTAGGACCGGCCTTAAAGC-3’.

The reverse primer sequence annealed to the PCR adapter site and the forward primer sequence annealed ~150 nt upstream. Samples were normalized to the following primers targeting Actin:

5’-TGTTACCAACTGGGACGACA-3’ and

5’-ACCAGAGGCATACAGGGACA-3’.

Reactions were run on a CFX 384 thermocycler using the following cycling conditions: 1x [3 min @95 °C], 39x [10 s @95 °C, 30 s @65 °C, plate read]. CT values were calculated using Bio-Rad CFX Manager v3.1 software. 2^- dCT (fold change) values were calculated as follows: 2^-[CT(Cre)-CT(Actin)].

RNA extraction and cDNA synthesis were performed as described above. Full-length Cre amplicons were obtained via PCR with Q5 DNA polymerase (NEB, M0491L) using primers that anneal the 5’ UTR and 3’ UTR, followed by agarose gel electrophoresis and purification of the resulting 1.3 kb DNA band. NGS libraries were prepared using a two-step PCR procedure with the following primer sets to amplify the barcode region and add Illumina flow cell adapters, sequencing adapters, and indexes:

PCR1:

5’-GTCATGAACTATATCCGTAACC-3’

5’− 5’ GTGACTGGAGTTCAGACGTGTGCTCTTCCGATCTTTGCTAGGACCGGCCTTAAAGC-3’

PCR2:

5’-AATGATACGGCGACCACCGAGATCTACACTCTTTCCCTACACGACGCTCTTCCGATCTGTCATGAACTATATCCGTAACC-3’

5’-CAAGCAGAAGACGGCATACGAGAT NNNNNNNN GTGACTGGAGTTCAGACGTGTGC-3’

Where NNNNNNNN is the sample index.

Indexed libraries were pooled and submitted to the McDermott NGS core at UT Southwestern.

Reported barcode values are presented as the proportion of reads within a given sample index normalized to an LNP pooled input sample. Alternatively, some experiments used barcode-specific PCR primers to read the barcodes using RT-qPCR according to the methods described above. For qPCR quantification, the reported barcode values are presented as the log2 fold change versus actin, normalized to an input pool.

Tissues were fixed with 10% formalin at 4 °C overnight. The fixed specimens were cryoprotected with a graded series of sucrose and then embedded in tissue-freezing medium. Frozen tissue blocks were cut into 12-μm cryosections. For immunostaining, 10% goat serum in 0.1% Triton X-100 1× PBS solution was used for blocking, and 3% goat serum in 0.1% Triton X-100 1× PBS solution was used for primary/secondary antibody incubations. Sections were incubated in primary antibody at 4 °C overnight, then washed 3 × 10 min in PBS to remove unbound antibody. Samples were incubated in secondary antibody for 1 h at RT, then washed 3 × 10 min in PBS. ProLong Gold Antifade Reagent with DAPI (Thermo Fisher) was used to mount the samples and visualize nuclei. Imaging of the samples was carried out on a Zeiss Z1 slide scanning microscope or a Leica DMi8 inverted microscope. Original images were processed using Zeiss ZenBlue software, Leica LasX software, or exported in tif format and processed using ImageJ.

Animal experiments were approved by the Institutional Animal Care and Use Committee of The University of Texas Southwestern Medical Center under APN 2015-101118-G, APN 2015-100933, APN 2016-101430, and APN 2017-102283 and were consistent with local, state, and federal regulations as applicable. B6.Cg-Gt(ROSA)26Sortm14(CAG-tdTomato)Hze/J (also known as Ai14) mice were obtained from The Jackson Laboratory (007914) and bred to maintain homozygous expression of the Cre reporter allele that has a loxP-flanked STOP cassette preventing transcription of a CAG promoter-driven red fluorescent tdTomato protein. Following Cre-mediated recombination, Ai14 mice express tdTomato fluorescence. All animals were maintained on a 12/12-h light/dark schedule at a mean temperature of 22 °C and fed with ad libitum normal chow. All experiments were performed in at least three biological replicates. Male and female mice aged 6–12 weeks were used for experiments. Age and sex were distributed equally across experimental groups.

Lateral tail vein injections were performed under physical restraint. Sequential tail vein injections for multiplexed kinetics study were performed as follows: (1) Each injection contained 0.01 mg/kg/formulation, 0.03 mg/kg total mRNA dissolved in 50 uL PBS. (2) Injections at 3 h, 1 h, and 30 m timepoints were administered under physical restraint sequentially in one lateral tail vein (distal to proximal), while 10 m, 5 m, and 30 s timepoints were delivered under isoflurane anesthesia in the opposite vein, also distal to proximal. (3) Following the last injection, mice were immediately euthanized by cervical dislocation. (4) Blood collection was immediately performed via cardiac puncture using a 25-G needle inserted into the left ventricle until ~100 uL was drawn. Whole blood was used for downstream analysis by directly adding 1 mL of TRIzol and following the RNA extraction procedure described above. (5) Major organs were immediately collected in the following order: heart, lung, liver, spleen, and kidneys. Utensils were briefly wiped between organs to avoid cross-contamination. No perfusion was used prior to collecting tissues; tissue samples were added directly to TRIzol and processed as described for RNA extraction. (6) Replicate injection times were staggered to facilitate handling only one mouse at a time.

Further information on research design is available in thelinked to this article. Nature Portfolio Reporting Summary