Automated Flow Synthesis of Peptide-PNA Conjugates.

Antisense peptide nucleic acids (PNAs) have yet to translate to the clinic because of poor cellular uptake, limited solubility, and rapid elimination. Cell-penetrating peptides (CPPs) covalently attached to PNAs may facilitate clinical development by improving uptake into cells. We report an efficient technology that utilizes a fully automated fast-flow instrument to manufacture CPP-conjugated PNAs (PPNAs) in a single shot. The machine is rapid, with each amide bond being formed in 10 s. Anti-IVS2-654 PPNA synthesized with this instrument presented threefold activity compared to transfected PNA in a splice-correction assay. We demonstrated the utility of this approach by chemically synthesizing eight anti-SARS-CoV-2 PPNAs in 1 day. A PPNA targeting the 5' untranslated region of SARS-CoV-2 genomic RNA reduced the viral titer by over 95% in a live virus infection assay (IC50 = 0.8 μM). Our technology can deliver PPNA candidates to further investigate their potential as antiviral agents.

Introduction

Antisense oligonucleotide (ASO)-based therapeutic development is gathering momentum following the recent FDA approval of the drugs Eteplirsen,[1] Golodirsen,[2] Casimersen,[3] Viltepso[4] (based on phosphorodiamidate morpholino oligomer, PMO), and Spinraza[5] (based on 2′-O-methoxyethyl-phosphorothioate). Next-generation technology for enhanced ASO tissue penetration using cell-penetrating peptides conjugated to PMO (PPMO) is under development at Sarepta Therapeutics, with promising clinical outcomes.[6] As a charge-neutral ASO class, peptide nucleic acids (PNAs) are also under development.[7,8]

PNAs are artificial single-stranded DNA-like molecules in which the sugar-phosphodiester moiety is replaced with an uncharged N-(2-aminoethyl) glycine unit, and the nucleobases are attached via a methyl carbonyl linker.[7,8] Due to the attenuated electrostatic repulsion in PNA-DNA/RNA duplexes, PNAs can hybridize complementary DNA and RNA with higher affinity and specificity than DNA–DNA/RNA duplexes.[9] Additionally, the amide-based PNA backbone offers unique physicochemical properties including superior chemical, thermal, and enzymatic stability.[10]

PNAs have found a wide range of chemical and biological applications.[11−13] PNAs are currently being developed as antisense agents for diseases including cancer,[14] monogenic blood disorders,[15] novel antibiotics,[13,16] and antivirals,[17,18] but to date have yet to become a drug. PNAs can be limited by low solubility, poor intracellular delivery, and renal clearance.[19] Despite these limitations, some PNAs display promising biological activity.[11−13] Specifically, longer PNA sequences (>15-mer) can be endowed with higher affinity and lower off-target toxicities.[20]

The covalent attachment of a cell-penetrating peptide (CPP) to a PNA can help to overcome some of the development challenges.[19] CPP conjugation can improve the PNA solubility, but it also reduces its physiological clearance rate by promoting cellular internalization.[19,21] Despite the utility of CPP–PNA conjugates, they can be toxic to cells and animals[22] and often require systematic structure–function studies to optimize delivery while minimizing toxicity.[23]

Existing standard batch protocols afford efficient access to PNA sequences with fewer than 15 bases while longer sequences remain challenging to synthesize[20] (Figure a). PNA synthesis can be limited by on-resin aggregation and side-reactions including deletion, rearrangement, isomerization, and nucleobase addition, resulting in low yield of the desired product.[20] Efforts to improve the synthetic efficiency include capping and double-couplings, yet these approaches still fail to produce long PNA sequences robustly.

Figure 1

Automated single-shot technology can rapidly produce on-demand customized PPNA sequences that inhibit SARS-CoV-2. (a) Typical coupling, deprotection, and capping procedures during PNA assembly in the solid phase. (b) Antisense PPNA binding to the 5′ UTR of the SARS-CoV-2 genomic transcript can prevent the expression of viral genes and subsequently inhibit viral growth. (c) Workflow for the manual stepwise synthesis of PPNAs with click chemistry. (d) Automated fast-flow synthesizer allows for rapid manufacture of PPNAs in a single-shot. Fmoc: 9-fluorenylmethoxycarbonyl. HATU: O-(7-azabenzotriazol-1-yl)-1,1,3,3-tetramethyluronium hexafluorophosphate. DIEA: N,N-diisopropylethylamine. Ac2O: acetic anhydride. Bhoc: benzhydryloxycarbonyl.

As the causative agent of coronavirus disease 2019 (COVID-19), the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2)[24] is a pathogen of immense importance to global public health. Developing innovative direct-acting antiviral agents is needed to eliminate this virus. PNAs have been investigated and demonstrated efficacy against SARS-CoV in vitro.[17] Given the high sequence similarity between SARS-CoV-2 and SARS-CoV,[25] antisense PNAs provide a promising avenue to achieve SARS-CoV-2 inhibition through a mechanism of reducing the targeted RNA expression via sequence-specific steric blocking (Figure b).

In order to find the most effective sequence for a given indication, investigation of multiple PNA candidates in a high-throughput manner is required. To remove the obstacles for the rapid production of these target-specific libraries of PNA strands, we developed a highly efficient technology in which chemistry is matched with an automated fast-flow instrument. This fully automated platform increases the PNA synthesis rate by nearly an order of magnitude relative to commercial state-of-the-art instruments with improved crude purity. The new flow synthesis protocol described here only requires 10 s for each amide bond formation between PNA monomers, a significant improvement over commercial microwave peptide synthesizers[26] (10 min/amide bond at 45 °C) or the DNA synthesizer Expedite 8909[27] (32 min/coupling cycle with capping at room temperature). Further, with flow-based synthesis, no capping or double couplings are needed. A variable temperature design increases coupling efficiency, while reducing on-resin aggregation and other side-reactions.

Instead of using a stepwise synthesis approach[28] via click chemistry[29] (Figure c), the high coupling efficiency achieved with our flow technology allows direct manufacture of long (>15-mer) PNA conjugates with cell-penetrating peptides in a single-shot (Figure d). This production strategy is convenient for simultaneous investigation of the bioactivity, toxicity, and cell uptake of multiple PNAs in view of establishing structure–function relationships.

To demonstrate the reliability and applicability of our method, we synthesized an 18-mer PPNA which hybridizes to the β-thalassemia gene sequence, IVS2–654,[30] for splice-correction that was covalently attached to a 12-mer CPP. The enhanced green fluorescence protein (EGFP) assay[30] showed that the single-shot synthesized PPNA presented threefold activity compared to transfected PNA. The utility of this automated technology was further demonstrated by the synthesis of eight PPNAs targeting SARS-CoV-2 genomic RNA in a single day. One of the designed PPNAs targeting the 5′ untranslated region (5′UTR) reduced the viral titers by over 95% in a live infection assay.

Results and Discussion

Automated Microscale Flow Synthesizer Design

The automated flow PNA synthesizer consists of seven modules including a central control computer, solution storage system, three HPLC pumps, three multiposition valves, heating elements, reaction zone, and a UV–vis detector. A modular script in the Mechwolf programming environment[31] controls the instrument (Figure a). During a coupling reaction, three HPLC pumps draw reagents stored under nitrogen atmosphere from the storage module, and the desired PNA monomer, activator, and base solutions are merged using a valve. The mixture flows through a module electrically heated at 70 °C, forming an activated ester. The activated PNA monomer flows next through the reaction zone, a packed bed of resin maintained at 70 °C, where amide bond formation is completed within 10 s. During the deprotection step, the piperidine solution flows through a room temperature loop controlled by a multiposition valve (Figure a) and meets with the 70 °C reactor to generate a ∼40 °C environment, which enables rapid and efficient deprotection with minimized nucleobase adducts. An in-line UV–vis detector was used to monitor the composition of the spent reagent solution (Figure b). The deprotection efficiency and the mass transfer rate through the resin can be inferred through the Fmoc-removal absorbance chromatogram.[32] This modular and in-line detection design allows for chemistry optimization not only for PNAs but also for other biopolymer syntheses.

Figure 2

Automated flow PNA synthesis enables 10 s amide bond formation and complete solid-phase synthesis cycles in 3 min. (a) Photographs of the automated flow solid-phase synthesizer: full image, variant temperature valve, and reusable reactors. (b) Process flow diagram. Amino acids, activating agents, and DIEA are mixed by three HPLC pumps. Two multiposition valves control the selection of the amino acids and activating agents. The third valve controls variant temperature flow paths. Amino acids are activated in the 70 °C loop, then flowed over the resin bed housed in a reusable reactor. The effluent is passed through a UV–vis spectrometer to waste. (c) Diagram of the synthesis cycle showing the duration of each step, the solution composition during each step after mixing, and the time used at each step. HBTU: O-(benzotriazol-1-yl)-N,N,N′,N′-tetramethyluronium hexafluorophosphate; DMF: N,N-dimethylformamide.

The reusable reactor body (Figure a) is designed for a 7.5 μmol scale synthesis and is typically loaded with 15 mg of 0.5 mmol/g Rink Amide resin. While reducing expensive monomer consumption and overall cost, this design can deliver milligrams of pure product, which is often sufficient for subsequent biological characterization. A detailed workflow and timeline for a synthesis process is listed in Figure c, which presents the amide bond formation step and the overall solid-phase PNA synthesis cycle within 3 min. Additional details can be found in the Supporting Information, sections 3 to 7.

Optimization of Automated PNA Synthesis

In-line UV–vis monitoring combined with liquid chromatography–mass spectrometry (LC-MS) and high-performance liquid chromatography (HPLC) product characterization allows for rapid optimization of PNA synthesis conditions. We began the optimization with a manually synthesized 4-mer PNA in ∼4 h[20,33] with 57% crude purity (Table , entry 1). The off-target products contained ∼15% isomers, ∼4% deletions, and ∼2% nucleobase adducts (see SI section 6). Next, the same PNA sequence was synthesized on the automated flow synthesizer in 15 min at 70 °C (Table , entry 2). High temperature accelerated both on-target (70%) and off-target reactions, especially nucleobase adducts (7%) and deletions (10%). Further base screening, including piperazine and morpholine, revealed that piperidine was optimal for the Fmoc-deprotection at 70 °C (Table , entries 3–4).

Table 1

Evaluation of Reaction Conditions for the Automated Flow PNA Synthesisa

PNA sequence: ACTG-Gly-CONH2. Conditions: manual synthesis: 100 mg Rink Amide resin (0.18 mmol/g), 6 equiv PNA monomer, 5.8 equiv PyAOP, 6 equiv DIEA, and 6 equiv 2,5-lutidine in DMF. Automated flow synthesis: 15 mg Rink Amide resin (0.5 mmol/g), 10 equiv PNA monomer, 9.6 equiv activator, and 30 equiv DIEA in DMF, flow rate: 2.5 mL/min. See SI for details. The crude purity was determined by HPLC UV absorbance at 280 nm. Total deletions of A, C, T, G, and Gly were summed. PyAOP: (7-Azabenzotriazol-1-yloxy)trispyrrolidinophosphonium hexafluorophosphate.

A major focus of the high-temperature optimization was the prevention of piperidine adduct formation during the deprotection steps. Addition of formic acid to the deprotection solutions to prevent base-mediated aspartimide formation is a common practice,[34] but we found that the lower basicity is also useful in preventing nucleobase adduct formation. Formic acid (1% solution in 20% piperidine-DMF, v/v) was therefore used as an additive for deprotection and successfully decreased the base–adduct ratio to 3% (Table , entry 5). Higher formic acid concentration can further inhibit the formation of base adducts but results in additional impurities (Table , entry 6). The most effective strategy, however, was to decrease the deprotection temperature to 40 °C, which was controlled by a multiposition valve and a “T” connector (Figure a). Under this condition, only negligible nucleobase adducts (<1%) were generated (Table , entry 7).

Higher coupling temperatures (80 °C and 90 °C) for synthesis were tested, and increased amounts of impurities (16% and 28%, respectively) were found, indicating the synthesis temperature of 70 °C on this flow instrument is optimal (Table , entries 8–9). A significant increase in purity (81%) was achieved by using HATU as an activator, but 5% isomerization was still observed (Table , entry 10). Finally, after replacing HATU with HBTU, we found that the condition of entry 11 in Table yielded 90% pure PNA with <1% isomers and <1% nucleobase adducts. The entry 11 conditions therefore were used in all subsequent PNA syntheses.

Automated Single-Shot Synthesis of PPNA

The optimized flow recipe provides PNA sequences with significantly improved purity to manual protocols in a fraction of the time. As depicted in Figure , using batch protocols, production of a 4-mer PNA was completed in 4 h. In comparison, our flow protocol provided the same sequence in 15 min with superior purity (90% vs 57% for batch synthesis), as demonstrated by both HPLC and LC-MS chromatograms (Figure a–d). To further test the synthetic efficiency of this methodology, an 18-mer PNA with a three-lysine (K3) linker,[35] which hybridizes to the β-thalassemia gene sequence (IVS2–654), was prepared in about 1 h with 60% crude purity. We purified 4 mg of pure PNA with >95% purity for downstream biological characterizations (Figure e,f).

Figure 3

Automated fast-flow synthesis of PNAs and PPNAs with enhanced purity over manual protocols. (a–d) Crude HPLC/LC-MS traces of manually and flow-synthesized 4-mer PNA. (e,f) Crude/pure HPLC, pure LC-MS traces of flow synthesized anti-IVS2-654 PNA. (g,h) Crude/pure HPLC, pure LC-MS traces of a single-shot synthesized anti-IVS2-654 PPNA.

This efficient protocol allows for single-shot synthesis of PPNAs. Conjugating cell-penetrating peptides (CPPs) with PNAs through click chemistry is a strategy for PNA intracellular delivery.[36,37] However, this method can be time-consuming (Figure c) and might require multiple purifications, thereby reducing yields. To streamline preparation, we established a single-shot strategy to produce PPNAs using our automated platform. A 12-mer CPP, RXRRBRRXRRBR-CONH2 (Bpep, X = 6-aminohexanoic acid; B = β-alanine)[38] was presynthesized onto the resin on a peptide synthesizer developed previously in our laboratory,[39] and the Bpep-loaded resin was transferred directly on the microscale instrument for PPNA synthesis. In this way, an anti-IVS2-654 PPNA was synthesized in 1.7 h with 52% crude purity and 1.2 mg of pure material (>95% purity) were obtained after purification (Figure g,h). This automated protocol not only achieved rapid production but also enabled an efficient synthesis of PPNA sequences that might be difficult to synthesize with batch.

Synthetic anti-IVS2-654 PPNA Shows Enhanced Activity in an EGFP Assay

The bioactivity of synthesized PPNAs was demonstrated by an enhanced green fluorescent protein (EGFP) assay in HeLa-654 cells.[30] In this assay, the HeLa cells are stably engineered with an EGFP-coding sequence interrupted by an intron from the human β-globin gene (IVS2-654). The intron contains a cryptic splice site that leads to retention of a β-globin fragment in the EGFP mRNA sequence, resulting in the translation of a nonfluorescent protein. The anti-IVS2-654 PNA hybridizes to the aberrant β-globin 5′ splice site, forcing the splicing machinery to use the normal splice sites and producing a functional, fluorescent EGFP. The PNA activity in the nucleus is therefore correlated with EGFP fluorescence and can be analyzed using flow cytometry, reported here as mean fluorescence intensity (MFI) relative to cells treated with phosphate-buffered saline (PBS) as vehicle. We used the synthesis platform for rapid production of an 18-mer anti-IVS2-654 PNA, a scramble 18-mer PNA (negative control), and an anti-IVS2-654 PPNA at milligram scale with >95% purity.

A commercial anti-IVS2-654 PNA was obtained from PNA Bio as a positive control (Figure a). As depicted in Figure b, the synthesized PNA and commercial PNA, with delivery mediated by lipofectamine, presented similar activity in the EGFP assay, validating the function of the flow-produced material. The fluorescence, however, did not increase when cells were treated with the scramble PNA, indicating sequence-dependent activity (Figure b). Importantly, the single-shot synthesized PPNA improved the activity compared to both unmodified PNA (16-fold) and transfected PNA (3-fold) at 10 μM (Figure c). This result highlights the importance of cell-penetrating peptides and demonstrates the potential to quickly build an active PPNA library for further applications.

Figure 4

Automated flow-synthesized PNAs are active in cell assays. (a) Names, sequences, quantities, and purities of EGFP PNAs and PPNA. (b) Relative fluorescence (to PBS vehicle-treated cells) of flow-synthesized anti-IVS2-654 PNA and relevant scramble PNA are compared to commercial anti-IVS2-654 PNA, as determined by an EGFP assay in HeLa-654 cells. PBS: Phosphate-buffered saline; LPF: Lipofectamine. (c) Dose–response curves corresponding to activity in the EGFP assay for synthesized unmodified PNA, PPNA, and PNA with lipofectamine. Activity is shown as fluorescence intensity relative to PBS vehicle-treated HeLa-654 cells.

Synthetic anti-SARS-CoV-2 PPNA Shows Over 95% Viral Inhibition in a Live Infection Assay

The 5′ UTR of the coronavirus genome is responsible for important biological functions, such as viral replication, transcription,[40] and packaging.[41] In previous studies, synthetic antisense agents such as PNAs and phosphorodiamidate morpholino oligomers (PMOs) targeting various sites in the 5′UTR of mouse hepatitis virus[42,43] and SARS-CoV[17,44] demonstrated viral inhibition. Recently, PPMOs were shown to effectively inhibit SARS-CoV-2 replication.[45] To the best of our knowledge, inhibition of SARS-CoV-2 with PPNAs has not been reported, however.

To further demonstrate the utility of our automated single-shot technology, within 1 day we synthesized eight long PPNAs (>15-mer) targeting the 5′UTR, the transcription regulatory site (TRS), and the polyprotein 1a/b translation start site (AUG) of SARS-CoV-2. After purification, each PPNA was obtained in milligram amounts at >90% purity (Figure a). These PPNAs were evaluated for inhibition of SARS-CoV-2 replication in Vero-E6 cells. Anti-IVS2-654 PNA, which has no intracellular target present in this assay, was selected as a negative control, while a known inhibitor of SARS-CoV-2 viral growth, EK1,[46,47] was used as a positive control. A multiplicity of infection (M.O.I.) of 0.1 was used for virus inoculation for 2 h. PPNA treatments were given at the same time as the viral infection and then continued after virus removal. Subsequently, the viral RNA levels were measured 48 h after initial infection. We observed dose-dependent reduction of the viral RNA replication with increasing concentrations of PPNAs (Figure b). Sequences designed to target the 5′UTR and AUG region were efficacious. The 5′UTR-3 sequence reduced titers by 75% in the live SARS-CoV-2 assay, which was at the same level as EK1. Nearly complete inhibition was achieved with the sequence 5′UTR-1 at 10 μM (IC50 = 0.8 μM, Figure c).

Figure 5

Dose-dependent inhibition of live SARS-CoV-2 replication by a synthetic PPNA library. (a) Names, sequences, target locations, quantities, and purities of synthetic PPNAs. Locations on the SARS genomic RNA are based on GenBank NC_045512. Bases targeting the SARS-CoV-2 leader-TRS (nt 70–75) are in green; Bases targeting the AUG translation start site are in blue. (b) Increasing concentrations of PPNAs result in inhibition of viral replication as measured by total viral RNA present. EK1 was selected as a positive control, and anti-IVS2-654 as a negative control. (c) Viral inhibition with PPNAs at 10 μM.

Conclusions

We developed an automated flow technology that provides a route to produce long (>15-mer) PNA–peptide conjugates with high purine content, whereas traditional methods may require several rounds of optimization. Furthermore, this automated synthesis protocol enables access to PNA sequences on an accelerated time scale, more rapid than traditional automated methods that use commercial microwave peptide synthesizers or the DNA synthesizer Expedite 8909. No additional capping or double couplings are needed even for the assembly of long PNA chains with our approach.

We took advantage of this streamlined flow synthesis to produce eight anti-SARS-CoV-2 peptide-conjugated PNAs in a single day. These novel synthetic PPNAs were tested in a SARS-CoV-2 cell entry assay. A PPNA sequence targeting the 5′UTR of SARS-CoV-2 was found to reduce the viral titer by >95%. Given the favorable toxicological profile of PNA compounds,[12] further investigations on the therapeutic window of the PPNAs reported here are warranted.

Although PNAs face challenges in translation to the clinic, the introduction of the efficient and robust synthesis technology described here represents a step forward by enabling on-demand rapid production of candidate antisense oligonucleotides, not only for COVID-19 but also for other diseases and emerging pathogens.