Antisense Oligonucleotide Activation via Enzymatic Antibiotic Resistance Mechanism

The structure and mechanism of the bacterial enzyme β-lactamase have been well-studied due to its clinical role in antibiotic resistance. β-Lactamase is known to hydrolyze the β-lactam ring of the cephalosporin scaffold, allowing a spontaneous self-immolation to occur. Previously, cephalosporin-based sensors have been developed to evaluate β-lactamase expression in both mammalian cells and zebrafish embryos. Here, we present a circular caged morpholino oligonucleotide (cMO) activated by β-lactamase-mediated cleavage of a cephalosporin motif capable of silencing the expression of T-box transcription factor Ta (tbxta), also referred to as no tail a (ntla), eliciting a distinct, observable phenotype. We explore the use of β-lactamase to elicit a biological response in aquatic embryos for the first time and expand the utility of cephalosporin as a cleavable linker beyond targeting antibiotic-resistant bacteria. The addition of β-lactamase to the current suite of enzymatic triggers presents unique opportunities for robust, orthogonal control over endogenous gene expression in a spatially resolved manner.


■ INTRODUCTION
−6 A drawback of oligonucleotides can be their susceptibility to nuclease-mediated degradation, which can be addressed through the incorporation of chemical modifications. 1 Morpholino oligonucleotides (MOs) are one class of modified nucleic acid analogs, containing a six-membered morpholine ring in place of the ribose and a neutral phosphoramidite linkage within the backbone.Despite the structural differences from natural nucleic acids, MOs maintain their high specificity and affinity to hybridize to the corresponding target RNA sequences. 7,8−15 The development of caged MOs (cMOs) that can be activated with external triggers has broadened the utility of these tools for analysis of dynamic cellular processes with precise spatial and temporal control.This is particularly important for investigations of gene function during later developmental stages or tissue-specific cellular processes, both of which are impeded by immediate, global silencing.Upon delivery, cMOs are functionally inert and unable to engage their target sequences until activated with an external trigger, imparting conditional control over MO-induced gene silencing. 16We and others have devised various MO caging strategies by incorporating light-, 17−22 enzyme-, 23 and more recently, small molecule-responsive elements 24 within the MO structure.−24,28−33 While all of these designs have demonstrated varying degrees of success in aquatic embryo systems, end-to-end macrocyclization has proven to be a highly modular approach that can be used with a wide range of immolative linkers 34 and commercially available MOs bearing suitable reactive handles.The curvature induced by the cyclization (and more recently bicyclization) 20 of the MO impedes target binding until a single cleavage event restores the active, linear species, which can then hybridize to its mRNA target and induce gene silencing.
Enzymatic triggers provide unique opportunities for conditional control that complement current light-and small molecule-activated cMOs.As an example, precise activation by light exposure can be complicated by the inherent structural complexity, opacity, and physical movement of individual cells within developing organisms.In the case of small molecule delivery, controlled activation of the MO can be limited by diffusion of the small molecule trigger.These shortcomings can be addressed through enzymatic activation, as the triggering enzyme can be expressed in specific cells and at controlled time points through the use of inducible or tissue-specific promoters. 35,36The Chen lab has demonstrated this previously using an Escherichia coli nitroreductase, Nfsb, to control the activity of a cyclic cMO tethered with a 4-nitrobenzyl linker in live zebrafish embryos. 23Additionally, the Dmochowski group has developed a caspase-3 activatable peptide nucleic acid (PNA)-based oligonucleotide sensor to probe apoptosis in cells. 37Taken together, these reports demonstrate the versatility of enzyme-mediated strategies that can be applied to nucleic acid-based tools.
Herein, we expand upon this previous work and demonstrate for the first time the use of the bacterial enzyme β-lactamase to control oligonucleotide function.The use of βlactamase as a triggering enzyme is particularly appealing since the structure, mechanism, and substrate specificity are wellestablished due to the enzyme's prominent role in antibiotic resistance. 38,39β-Lactamases also possess many other favorable characteristics, including stability, catalytic efficiency, and the fact that their enzymatic function is unique to bacterial systems.−42 Cephalosporin motifs have been previously functionalized for β-lactamase-mediated activation of prodrugs against antibiotic resistant bacterial strains 43,44 and fluorescent sensors probing β-lactamase expression in mammalian cells and zebrafish embryos. 45,46ere, we leverage β-lactamase-mediated antibiotic resistance mechanisms to control MO silencing function in vivo by designing and synthesizing a cyclic cMO tethered by a cephalosporin-based linker that is efficiently cleaved via βlactamase-mediated hydrolysis.

■ RESULTS AND DISCUSSION
Upon β-lactamase catalysis, the core β-lactam ring of the linker is hydrolyzed, generating the corresponding amine, which then undergoes self-immolation and spontaneous elimination via the carbamate linkage.This elimination reaction releases the 5′-terminus of the MO to generate the active, linear MO capable of hybridizing to the mRNA target and subsequently inducing gene silencing (Figure 1).
To synthesize the β-lactamase-activated cMO, the linker 8 was designed using a cephalosporin core scaffold.The C-3 and C-7 positions of the core cephalosporin scaffold were synthetically modified with orthogonal azido and chloroacetamide handles, respectively, necessary for bioconjugation to the terminally modified linear MO. 39 The carbamate linkage was strategically installed at the C-3 position as it has been previously shown that leaving groups attached to the exocyclic methylene can undergo spontaneous elimination following hydrolysis of the β-lactam ring. 47,48e linker 8 was synthesized in four steps from commercially available 7-aminocephalosporanic acid (4) (Figure 2A).The tert-butyl ester 5 was obtained by treating 4 with tert-butyl acetate in the presence of p-toluenesulfonic acid (TsOH) and concentrated sulfuric acid. 49The chloroacetamide at the C-7 position was installed by reacting the free amine 5 with 2-chloroacetyl chloride to form the corresponding amide 6. Selective cleavage of the acetate at  the C-3 position was achieved chemoenzymatically through lipase-mediated hydrolysis of 6 using the serine hydrolase Candida antarctica lipase B (CAL B) to generate the free alcohol 7. 49 To install the azide handle, 3-azidopropan-1-amine (2) was synthesized from the corresponding alkyl chloride 1 as previously reported. 50The amine 2 was converted to the corresponding isocyanate 3 through treatment with diphosgene.The isocyanate 3 was then added to a solution of the alcohol 7 in the presence of dibutyltin dilaurate to generate the corresponding carbamate-containing cephalosporin linker 8.It should be noted that while the canonical cephalosporin scaffold contains a carboxylic acid moiety at the C-4 position, there have been no direct studies of enzyme reactivity with any synthetic C-4-protected analogs.To test that the bulky tertbutyl ester in 8 still presents a suitable enzyme substrate, the linker (1 mM) was incubated with recombinant β-lactamase (0.5 μg), and the reaction progression was monitored by LCMS (Supporting Figure S1).Gratifyingly, nearly full cleavage of the linker 8 was observed after 3 h, and the expected cleavage product was generated.These results suggest that the presence of the tert-butyl ester at the C-4 position does not impede enzymatic cleavage of the lactam ring.With knowledge that the linker can be cleaved by β-lactamase, we moved forward with the macrocyclization of our desired MO with the functionalized linker 8.
A cyclic cephalosporin cMO targeting the T-box transcription factor Ta (tbxta), also known as no tail a (ntla, 5′-GACTTGAGGCAGCATATTTCCGAT-3′, anti-start codon underlined) was prepared in three steps using the synthesized linker 8 (Figure 2B).MO-mediated silencing of ntla during early embryogenesis induces an obvious and distinct morphant phenotype that includes truncation of the embryo tail and misshapen somites. 51,522][23][24][27][28][29]31 The ntla MO, containing 5′-amine and 3′-disulfide handles, was subjected to macrocyclization 28,29 by first acylating the terminal amine with the NHS ester 9 to yield the corresponding alkyne suitable for bioconjugation to the azide-containing linker. The linker 8 was clicked to the alkyne-functionalized ntla MO via a copper(I)-catalyzed alkyne−azide cycloaddition (CuAAC). 53 Latly, the 3′disulfide was reduced in the presence of resin-immobilized tris(2-carboxyethyl)phosphine (TCEP) to generate the free thiol, which then undergoes a spontaneous intramolecular cyclization via thioether formation with the chloroacetamide handle of 8 to generate the cyclic cephalosporin ntla cMO.Reaction progress for each step was analyzed via MALDI-TOF MS and purified by high-performance liquid chromatography (HPLC) (Supporting Figure S2).To ensure the removal of any remaining linear MO species, the final cMO product was subject to additional purification steps using iodoacetyl-and NHS-functionalized resins, as previously described.24 After confirmation of successful macrocyclization, we next sought to evaluate enzyme-mediated linearization and activation of the cephalosporin ntla cMO.Linearization was performed by incubating the cMO (20 μΜ) with recombinant β-lactamase for 3 h. We utlized a reticulocyte lysate-based translation system to analyze activation of the cMO.24 The consensus ntla binding sequence (ntlaBS) was cloned directly upstream of and in frame with the coding sequence of the firefly luciferase reporter gene.Hybridization of the MO to this target sequence blocks the translational machinery from accessing the start site, thereby repressing luciferase expression (Figure 3A).Following treatment with linear ntla MO (1 μM), luciferase expression is reduced by 50%, whereas treatment with the cyclic cephalosporin ntla cMO did not impact luciferase levels.However, MO silencing function is fully restored following linearization of the cyclic cMO with βlactamase, as evidenced by a nearly 65% reduction of reporter gene expression (Figure 3B).Taken together, this data suggests that enzymatic catalysis by β-lactamase can be used as an efficient off to on switch to conditionally control MO gene silencing activity.
While switching of MO activity looked excellent, we were surprised by the overall limited silencing of luciferase expression in the in vitro translation assay and thus conducted a gel-shift analysis of the MO:target interaction.A 25-mer DNA oligonucleotide that is fully complementary to the ntla MO (cDNA, Supporting Table S2) was used as an mRNA mimic.Heteroduplex formation with the MO or cMO was visualized by native PAGE (Figure 3C).No heteroduplex formation was observed with the cMO in the absence of βlactamase, confirming the results of the in vitro translation assay, as the curvature induced by macrocyclization of the cMO prevents target binding.It is only after incubation with the triggering enzyme that nearly full sequestration of the target cDNA is observed, as evidenced by the shift in mobility of the heteroduplex in the gel.This shift indicates successful cleavage of the linker and subsequent target hybridization by the decaged linear MO.To quantitatively determine the activated amount of cMO, the circular antisense agent was incubated with recombinant β-lactamase for up to 1 h.At defined time points, the enzyme was heat-inactivated and the MO was hybridized with an excess of cDNA.The band densities in a gel-shift assay were quantified using a standard curve (Supporting Figure S3) to determine the percentage of activated cMO.After 1 h, the cMO was 88% linearized (Figure 3D and Supporting Figure S4).The nearly complete heteroduplex formation observed by the decaged MO suggested promise for in vivo gene silencing.
Next, we sought to investigate whether β-lactamase could be used to control MO-mediated gene silencing of an endogenous gene target during early zebrafish development.β-lactamase activity can be easily evaluated using nitrocefin (10), a cephalosporin-based chromogenic substrate that changes in color from yellow to red following β-lactam hydrolysis (yielding 11; Figure 4A). 54The utility of this substrate was validated using recombinant β-lactamase (Supporting Figure S5).We generated a β-lactamase expression construct by cloning an N-terminally HA-tagged mammalian codonoptimized β-lactamase into the pCS2+ vector, which can be used for enzyme expression in mammalian cells and as a template for in vitro mRNA production (Supporting Table S2). 55The expression of β-lactamase was confirmed in HEK293T cells and enzymatic function was validated by monitoring cleavage of the nitrocefin substrate (Supporting Figure S6).
In order to evaluate β-lactamase expression and activity in vivo, we next generated synthetic β-lactamase mRNA via in vitro transcription (Supporting Figure S7).The synthetic βlactamase mRNA was injected into the yolk sac of 1-to 4-cellstage zebrafish embryos.At 24 h post-fertilization (hpf), the injected embryos were collected, lysed, 56 and β-lactamase expression was confirmed by Western blot (Figure 4B).To confirm that β-lactamase also maintained hydrolase activity, embryos were injected with increasing amounts of mRNA, lysed, and then incubated with nitrocefin.Hydrolysis of the substrate was monitored over time.An obvious dosedependent correlation in β-lactamase activity was observed (Figure 4C).Further, quantification of the initial rates of nitrocefin hydrolysis demonstrated a clear positive correlation between the amount of mRNA injected into the embryos and the rate of enzymatic activity (Figure 4D).These results suggest that the extent of enzymatic activity can be finely tuned with the mRNA injection amount, with maximal activity plateauing following injection of approximately 500 pg of βlactamase mRNA.
We then utilized our cyclic cephalosporin ntla cMO to determine whether we could induce enzymatic activation of MO function in vivo.−29 Silencing of ntla expression during early embryogenesis results in a distinct phenotype observable at 24 hpf, including truncation of tail length, loss of the notochord, and U-shaped somites (Figure 5A). 51,52The quantification of embryo body length was used to evaluate the strength of the morphant phenotype (Supporting Figure S8). 24Zebrafish embryos were injected in the yolk with 200 pg of either a negative control MO, the linear ntla MO, or our cyclic cephalosporin ntla cMO following previously established methods for zebrafish microinjections. 14,56,57Based on the diameter (0.7 mm) and cytoplasmic bridging of early-stage embryos, we estimate the final concentration of the morpholino to be approximately 120 nM in the injected embryos. 58,59Microinjection into the embryo yolk sac, as opposed to directly into the cell, enables a higher throughput workflow (200-300 embryos from one mating) 60 and reduces injection errors that could result in embryo-to-embryo variation.However, a caveat of this injection technique is that precise assessment of the intracellular MO concentration is difficult to discern. 59Where indicated, 400 pg of β-lactamase mRNA was coinjected with the specified MO or cMO, and embryos were incubated until 24 hpf (Figure 5B).Embryos that were not injected or injected with the control MO developed as expected and displayed minimal signs of toxicity.However, we observed a slight increase in toxicity in embryos coinjected with MO and mRNA.To further investigate this toxicity, control MO (200 pg) was injected with increasing amounts of β-lactamase mRNA (up to 1000 pg).No trend was detected and no increase in phenotypic defects or toxicity was observed (Supporting Figure S9).Importantly, embryos injected with the cyclic cephalosporin ntla cMO alone also exhibited minimal phenotypic defects, confirming the inactivity of the  cyclic cMO in vivo.However, upon coinjection of the cephalosporin ntla cMO with β-lactamase mRNA, full rescue of the ntla morphant phenotype was observed, comparable to the frequency of phenotype observed with the linear ntla MO.Gratifyingly, zebrafish coinjected with the cyclic cephalosporin ntla cMO and synthetic Nfsb mRNA did not show any ntla morphant phenotypes (Supporting Figure S10A).The same observation was made when injected embryos were exposed to 365 or 405 nm light (Supporting Figure S10B).Together, these results demonstrate the orthogonality of the two enzymatic cMO activation approaches and to the optical stimulation used for photocaged cMOs.There was a higher frequency of nonspecific phenotypic defects in Nfsb-mRNA injected embryos, showing lower toxicity of the β-lactamase expression (Supporting Figures 9A and 10A). 52Taken together, these results support the utility of the β-lactamase enzyme for conditionally controlling cephalosporin-modified antisense oligonucleotides to investigate endogenous gene function in vivo.

■ CONCLUSION
In summary, we leveraged enzyme-mediated antibiotic resistance mechanisms to develop a β-lactamase-triggered antisense agent.A novel linker containing a cephalosporin core-scaffold was synthesized and used to cyclize a MO.The curvature of the cyclic cMO successfully abrogated hybridization to and silencing of its mRNA target.β-lactamasemediated cleavage of the cMO linker, MO linearization, and restoration of gene silencing were confirmed in vitro by gelshift assay and silencing of reporter gene expression.Efficient enzyme-triggered silencing of endogenous gene expression was demonstrated in live zebrafish embryos coinjected with the cyclic cephalosporin ntla cMO and mRNA expressing βlactamase.7][18][19][20][22][23][24][27][28][29]31 Despite only one prior example of β-lactamase use in zebrafish embryos, 45 the work presented herein demonstrates that the highly specific interaction between β-lactamase and cephalosporin provides a novel and robust means to control gene expression in a potentially tunable manner. Future studies placing β-lactamase expression under the control of tissuespecific promoters (similar to what has been shown previously with nitroreductase to enable targeted cell ablation experiments) 23 will provide spatial and temporal control of enzyme function. While numerous fish lines expressing nitroreductase have been reported, 61 transgenic fish lines stably expressing βlactamase have not yet been generated, thereby precluding these studies.20,[22][23][24]28,29 Lastly, the activation of MOs demonstrates that β-lactamase expression has potential for being a broadly applicable trigger for molecular function in zebrafish, significantly expanding its utility.

Figure 1 .
Figure 1.Following enzyme-mediated hydrolysis of the β-lactam ring by β-lactamase (PDB ID 4F6H), the resulting amine intermediate undergoes self-immolation.The release of the 5′-end generates the active, linearized MO capable of hybridizing to its respective target mRNA and silencing gene expression.

Figure 3 .
Figure 3. (A) Luciferase reporter assay in which luciferase expression is silenced in the presence of linearized MO. (B) Luminescence signal after treatment of the indicated MO.Data represent average ± SEM from three independent experiments.(C) Gel-shift assay showing activation of the circular morpholino sequestering its complementary target DNA (cDNA) in the presence of β-lactamase.(D) Quantification of a time course analysis of cyclic cephalosporin ntla cMO cleavage following incubation with recombinant β-lactamase.Bars represent averages and error bars represent standard deviations from three independent experiments.

Figure 4 .
Figure 4. (A) Structure of the chromogenic β-lactamase substrate nitrocefin (10), which is hydrolyzed to 11 following β-lactamase catalysis.(B) Western blot analysis of β-lactamase expression in embryos injected with 400 pg of mRNA.(C) Zebrafish lysates analyzed for β-lactamase activity by monitoring production of 11 (absorbance at 486 nm) over time.(D) Initial velocities were determined through a linear regression analysis of the first 10 min of the data presented in (C).The error values indicate the standard deviation of values for the best fit parameters from the linear regression analysis.

Figure 5 .
Figure 5. (A) Representative images of ntla morphant phenotypes in zebrafish embryos at 24 hpf.Scale bar equals 1 mm.(B) Phenotypic scoring of embryos injected with the indicated morpholino and amount of β-lactamase mRNA.