Binding and biomimetic cleavage of the RNA poly(U) by synthetic polyimidazoles

Abstract

Four polyimidazoles were used in the binding and cleavage studies with poly(U). The two polydisperse polyvinylimidazoles were previously described by others, while the other two new polymers of polyethyleneimines were prepared by cationic polymerization of oxazolines. The latter had imidazole units attached to each nitrogen of the polymers. They were characterized by gel permeation chromatography and had very low polydispersities. When they were partially protonated they bound to the poly(U) and catalyzed its cleavage by a process analogous to that used by the enzyme ribonuclease A. The kinetics of the cleavage were followed by an assay we had previously described using phosphodiesterase I from Crotalus venom after the cleavage processes. Cleavage of poly(U) with Zn²⁺ was also examined, with and without the polymers. A scheme is described in which these cleavages could be made sequence selective with various RNAs, particularly with important targets, such as viral RNAs.

The polymeric character of natural enzymes has stimulated us to expand the field of artificial enzymes to those based on synthetic polymers, as we have reported earlier (1). Particularly important would be artificial enzymes that could cleave RNA, preferably with sequence specificity (2, 3). They could potentially be used to destroy dangerous viral RNA molecules if we attached a short DNA to the polymers with the necessary sequence for selective binding. This would be an advance in molecular medicine—using an active catalyst rather than simply a modifier of biological systems, as is typical with other medicines.

We have described an analytical method to follow the rates of RNA cleavage (4) by various catalysts or reactants that imitate the mechanism used by ribonuclease A. In this enzyme, the C-2′ hydroxyl group of a ribose attacks the neighboring phosphate that links the C-3′ hydroxyl with a C-5′ hydroxyl of another ribose. The result is the formation of a C-2′/C-3′ cyclic phosphate and a liberated C-5′ hydroxyl group. Then we treat the product with the phosphodiesterase I from Crotalus venom, which cleaves RNA in a different manner, hydrolyzing the remaining C-3′/C-5′ phosphate link to remove the phosphate from C-3′ and leave it behind on C-5′. With RNA, all the products of the second step are nucleotide 5′-phosphates except those where the ribonuclease-like cleavage has left the 5′-position unphosphorylated. With poly(U) as the substrate, every initial ribonuclease-like cleavage site leaves an unphosphorylated uridine after the phosphodiesterase I treatment, that we assay with quantitative HPLC. We have shown that this assay for RNAse-like cleavage of RNA is quantitatively reliable (4).

In ribonuclease A, the cleavage is catalyzed by imidazole rings of two histidines, one less basic as the free imidazole base (His-12) and the other more basic as an imidazolium cation (His-119) (5). In the most likely mechanism (6), the imidazole of His-12 deprotonates the C-2′ hydroxyl group as the oxygen adds to the neighboring phosphate, forming a phosphorane intermediate with help as the positive charged imidazolium group on His-119 adds a proton to an oxyanion of the phosphate. In subsequent steps, the phosphorane intermediate decomposes, expelling the C-5′ hydroxyl group of the next unit while leaving a C-2′/C-3′ cyclic phosphate. This is then hydrolyzed in subsequent steps to the final product with a phosphate group on C-3′ and a hydroxyl group back on C-2′. We have described biomimetic systems that mimic this process, using either concentrated imidazole buffer with RNA itself (6–10) or cyclodextrins carrying attached imidazole rings to cleave RNA analogs (11–13). We now describe the cleavages of poly(U) (> 300 units) by two classes of synthetic polyimidazoles as ribonuclease A mimics, again forming a C-5′ hydroxyl group on the leaving RNA unit as with ribonuclease A. We also have evidence for binding of some of the polymers to the RNA prior to cleavage. The cleavage was enhanced by the binding characteristics of the polymer itself (compared with those without binding sites), rather than the addition of exogenous metal cation cofactors (although addition of Zn²⁺ did improve the hydrolysis in our cases), which makes it more attractive for further cellular application.

Results and Discussion

Our first study used poly-4-vinylimidazole (P4VIm) and poly-1-vinylimidazole (P1VIm), synthesized by radical polymerization of the corresponding 4- and 1-vinylimidazole (14) (Scheme 1). Overberger et al. had extensively examined them as catalysts for the hydrolysis of phenyl esters (15, 16). We had previously seen that they are good catalysts with our reversibly bound pyridoxamine coenzyme mimic for the transamination of phenylpyruvate to form phenylaniline (14). Those results were gratifying, but polyvinylimidazoles from simple radical polymerization have a wide dispersity of lengths (20- to 40-mer, as we see from mass spectra) and are not homogenous species. Thus we turned to the polymers from cationic polymerization of oxazolines (17).

Scheme 1.

Synthesis of linear polyethylenimines with pendant imidazole groups.

We have reported earlier the use of the well-known class of polymers from cationic polymerization of oxazolines (18, 19). The process leads to defined polymers, whose size is determined by the ratio of alkylating agent to monomer. In general, with 1% of an alkylating agent one can obtain polymers with 100 units, within a narrow size range characterized as the polydispersity index (PDI). With oxazolines derived from the reduction of natural amino acids, the chiral center is undisturbed (18). Thus we considered how we might make an oxazoline polymer derived from histidine. However, the imidazole ring would compete and interfere with the oxazoline alkylation/propagation steps in the polymerization, and it was not clear how this could be easily blocked. Thus we instead used an indirect approach to make oxazoline-derived polyimidazoles. We made the simple polymers (1, Scheme 1) from 2-ethyloxazoline with methyl triflate and terminated the polymerization with sodium carbonate solution (20). We then removed the propionyl groups from the polymers with concentrated hydrochloride acid to obtain the linear polyethylenimines 2 (21). Klotz had made a related catalyst by partially alkylating commercial polydisperse polyethleneimine with chloromethylimidazole, and examined it as a catalyst for ester hydrolysis (22). We treated the polymers 2 with an excess of 4(5)-(hydroxymethyl)imidazole and thus attached one 4(5)-methyleneimidazole group to each secondary amino group of polyethylenimines forming polymers 3 (23). This mild procedure does not lead to further alkylation of the amino groups.

We characterized the polymers of type 1 and 3 with gel permeation chromatography (GPC) to determine their sizes and polydispersities. With an initial 4.5% of methyl triflate we obtained a 23-mer (predicted 22) of 1 with a polydispersity index (PDI) of 1.101 and a 23-mer of 3 with a PDI of 1.161. With an initial 2.5% of methyl triflate, we obtained a 41-mer (predicted 40) of 1 with a PDI of 1.289 and a 41-mer of 3 with a PDI of 1.118. This was a great improvement in dispersity over the large range in the radical polymerizations. Thus we had polyimidazoles with well-defined lengths and compositions. The buffer capacities of polymers 2 and 3 (n ≈ 23) were evaluated by measuring the pH change of the polymer solutions upon the titration (Fig. 1) over pH 2.0–11.0. The midpoint of the titration of polymer 2 was ca. 7.0, consistent with previous measurements on such polyamines (24). The broad titration range reflects the effects of protonated amino groups on the ease of protonation of additional amino groups. The midpoint of the titration of polymer 3 was ca. 6.0, reflecting the lower basicity of imidazoles relative to secondary amines.

Fig. 1.

Titration of aqueous polymer solutions with NaOH. Solutions were adjusted to pH = 2 with 0.1 M HCl, then 200 μL aliquots of 0.05 M NaOH were sequentially added, and the pH measured (black: blank; red: polymer 2; blue: polymer 3).

In these new polymers, we imitate the different pKs of the two imidazoles in the enzyme—reflecting different environments—by using two different classes of nitrogen bases: the more basic nitrogens in the polymer backbones of polyethylenimines and the less-basic nitrogens of the appended imidazoles. Thus, under neutral conditions, the backbone nitrogens will be heavily protonated while the imidazoles will be less so. The heavily protonated backbone nitrogens can contribute to binding of the RNA polyanion and as well to catalysis of the cleavage by hydrogen bonding to the phosphate anions and transferring protons in the catalytic step. Thus the simple polymer 2 shows some catalysis of RNA cleavage (Table 1). However the added imidazoles of polymer 3 contribute additionally, presumably by deprotonating the 3′ hydroxyl groups of RNA during cleavage.

Table 1.

Hydrolysis of poly(U) with catalysis by polyimidazoles 3 and polyvinylimidazoles^*

Entry	pH†	Polymer	kobs × 103 hr-1	kacc
1	7	no polymer	8.763 ± 0.182	1.00
2	7	3, n ≈ 23, 3 mM	10.362 ± 0.915	1.18
3	7	3, n ≈ 23, 15 mM	14.524 ± 0.488	1.66
4	7	3, n ≈ 23, 30 mM	20.708 ± 0.477	2.36
5‡	7	3, n ≈ 23, 40 mM	26.283 ± 1.487	3.00
6	7	3, n ≈ 41, 30 mM	17.204 ± 0.566	1.96
7	7	2, n ≈ 23, 15 mM	11.913 ± 0.510	1.36
8§	7	P4VIm, 15 mM	12.009 ± 0.473	1.22
9§	7	P4VIm, 45 mM	12.825 ± 0.730	1.31
10‡§	7	P4VIm, 150 mM	12.827 ± 0.430	1.31
11§	7	P1VIm, 15 mM	9.525 ± 0.403	0.97
12§	7	P1VIm, 45 mM	15.018 ± 0.533	1.53
13§	7	P1VIm, 150 mM	11.342 ± 1.293	1.15
14§	7	P1VIm, 450 mM	16.757 ± 1.273	1.71
15	6	3, n ≈ 23, 30 mM	5.613 ± 0.153	4.29¶
16	8	3, n ≈ 23, 30 mM	100.825 ± 0.335	2.00¶

^*Hydrolysis in Trizma-HCl buffer 0.1 M, I = 0.1 M (adjusted by NaCl), [poly(U)] = 15 to approximately 20 mM (in U) at 80 °C.

^†pH refers to 80 °C.

^‡The hydrolysis mixture was turbid at 80 °C.

^§Hydrolysis in the Trizma-HCl buffer with 30% ethanol. The background hydrolysis constant k_obs in the Trizma-HCl buffer with 30% ethanol is (9.826 ± 0.192) × 10^-3 hr^-1.

^¶The background hydrolysis constants for poly(U) at pH 6 and 8 are (1.307 ± 0.078) × 10^-3 hr^-1 and (50.499 ± 1.424) × 10^-3 hr^-1, respectively.

We then studied the binding of polymer 3 (n ≈ 23) with poly(U) in phosphate buffer at pH ca. 7.5 (Fig. 2) where the polymer was less than 50% protonated, with extensive free imidazole groups. The polymers 3 did not show any significant absorption in the range of 240–300 nm (Fig. 2A), so we ascribed the changes of UV (approximately 260 nM) and CD spectra (272–276 nm) to the electrostatic attractions of the cationic amino groups (N) in the polyethylenimine backbones and some protonated imidazolium groups binding to the anionic phosphate acid groups (P) in poly(U). There could also be conformational changes of poly(U) due to potential stacking interactions involving imidazole rings and uracil (25). With as much as a 3∶1 excess of ethylenimine units to uridine units, where the cationic and anionic groups were essentially equal in number, we saw complete precipitation of the poly(U) bound with the polymer (26, 27). However, with lower ratios of polycation to polyanion, we saw spectra of solutions in aqueous buffer that presumably involved complexes of the polymer and poly(U) with remaining net negative charges from excess phosphate groups.

Fig. 2.

Binding activity monitoring. (A) Observed changes of the UV spectra, recorded at room temperature, relative to poly(U) (180 nmol in U) and polymer 3 dissolved in 5 mL of 10 mM phosphate buffer (Na₂HPO₄-NaH₂PO₄) at pH 7.55. (black: poly(U); orange: polymer 3; red: poly(U)/3 2/1; blue: poly(U)/3 1/1; purple: poly(U)/3 1/2). (B) Observed changes of the CD spectra, recorded at room temperature, relative to poly(U) (375 nmol in U) and polymer 3 dissolved in 5 mL of 10 mM phosphate buffer (Na₂HPO₄-NaH₂PO₄) at pH 7.5. (black: poly(U); red: poly(U)/3 2/1; blue: poly(U)/3 1/1; green: poly(U)/3 1/1.5).

We did not examine the broad mixtures of polyvinylimidazoles (P4VIm and P1VIm) in binding studies. However, we did examine all the polymers as catalysts for poly(U) (15 mM) cleavage at 80 °C, analyzed using our previously described kinetic procedure (4). The kinetic study was performed of the initial rates with about 5% cleavage, and good pseudo-first-order rate behavior was observed in all cases. For a general review of kinetics and mechanism in RNA cleavage, see ref. 28.

As can be seen from Table 1, all the polymers afforded an increase in the cleavage rate of poly(U), but the accelerations were at best modest, the highest being a three-fold acceleration when polymer 3 was used at pH 7 (entry 5), miniscule compared with natural ribonuclease A [k_obs = (8.64 ± 5.40) × 10⁴ hr^-1, 0.1 MES buffer, pH 6.0, (NaCl) 0.1 M, 25 °C] (29). However, we did find that our pseudo-first-order cleavage rate was larger with increasing amounts of the polymers, until increasing their concentrations led to precipitation and the assay became unreliable. The degree of polymerization did not have a significant effect on the catalytic activity (entry 6). Nevertheless, polyvinylimidazoles were found to be inferior in the cleavage compared with polyethylenimine derived polymers (entries 8–14), which illustrated that the binding of cationic polyethylenimine backbones in polymers 3 with the anionic phosphates in poly(U)—and the likely catalytic role of the protonated backbone amino groups in proton transfer—accelerated the cleavage. That was further confirmed by a control experiment using unsubstituted polyethylenimine 2 as the catalyst (entry 7), from which a small hydrolytic acceleration was observed, indicating that the partially protonated backbone not only binds with the negative charged phosphodiester bonds through proton transfer, but also contributes slightly to the cleavage (30).

The cleavage rate at pH 8 showed some increase of the cleavage at higher pH, but the background uncatalyzed cleavage rate was more greatly increased (entry 16), proportional to hydroxide ion concentration (entries 15 and 16). Thus, at higher pHs, the uncatalyzed reaction dominated that with polymer catalysis. We did not see simple buffer catalysis (varying the buffer concentration) without the polymers [k_obs in Trizma-HCl buffer at concentration of 20 mM was (8.908 ± 0.173) × 10^-3 hr^-1].

We also examined the possible rate of cleavage when Zn²⁺ was included with the polymers (Table 2). We had previously shown that the combination of Zn²⁺/imidazole was much more effective than imidazole or Zn²⁺ alone in the cleavage of phosphate esters and UpU (9). Here we found that Zn²⁺ catalysis was quite effective in all cases, but diminished by the additional polymers. The binding of Zn²⁺ with strong ligands, such as the nitrogen on the polyethylenimine backbone or imidazoles, apparently decreased its Lewis acidity and, therefore, its catalytic activity (9, 31).

Table 2.

Hydrolysis of poly(U) with catalysis by Zn²⁺ with polyimidazoles 3 and polyvinylimidazoles^*

Entry	pH†	Polymer	kobs × 103 hr-1	kacc
1	7	no polymer	8.763 ± 0.182	1.00
2	7	no polymer, [Zn2+] 1.5 mM	64.749 ± 1.765	7.39
3	7	3, n ≈ 23, 3 mM; [Zn2+] 1.5 mM	14.707 ± 1.386	1.68
4‡	7	3, n ≈ 23, 15 mM; [Zn2+] 1.5 mM	11.369 ± 0.567	1.30
5‡§	7	P4VIm, 15 mM; [Zn2+] 1.5 mM	15.781 ± 0.911	1.61
6‡§	7	P4VIm, 45 mM; [Zn2+] 1.5 mM	16.817 ± 1.133	1.71
7§	7	P1VIm, 15 mM; [Zn2+] 1.5 mM	12.095 ± 0.311	1.23
8‡§	7	P1VIm, 45 mM; [Zn2+] 1.5 mM	10.046 ± 0.539	1.02

^*Hydrolysis in Trizma-HCl buffer 0.1 M, I = 0.1 M (adjusted by NaCl), [poly(U)] = 15 to approximately 20 mM (in U) at 80 °C.

^†pH refers to 80 °C.

^‡The hydrolysis mixture was turbid at 80 °C.

^§Hydrolysis in the Trizma-HCl buffer with 30% ethanol. The background hydrolysis constant k_obs in the Trizma-HCl buffer with 30% ethanol is (9.826 ± 0.192) × 10^-3 hr^-1.

Conclusion

In summary, we have prepared polyethylenimines grafted with pendant imidazole groups as ribonuclease A mimics that effectively bind with anionic RNA and catalyze its cleavage. To the best of our knowledge, this is the first example of using well-defined cationic polymers for the cleavage of RNA, and it does not require the addition of any other materials, such as metal ions. For such artificial catalytic cleavage to be useful, we need to add a sequence-selective binding group to such polymer catalysts. One plan is to add a short piece of DNA to the polymer with a sequence congruent to the special sequence of the target RNA, forming a short heterodimer section. We have a number of ways to attach such a selective DNA piece to our catalysts, and in various places, so the potential for cleavage of undesirable RNAs under physiological conditions seems attractive. However, this will be most effective with higher catalytic rates, which we are still pursuing.

Materials and Methods

General Procedure for the Synthesis of Polymers 3.

To a solution of 2-ethyl-2-oxazoline (3.97 g, 40 mmol) in chlorobenzene (10 mL) at room temperature was quickly but carefully added methyl triflate (MeOTf, 197 μL, 1.8 mmol). The mixture was then heated at 80 °C for 1 h and then rapidly cooled by immersing the flask in dry ice. Then 4 mL of a 5% sodium carbonate solution was added to the polymer solution, and the mixture was stirred for 30 min. The aqueous layer was separated, and the organic layer was extracted with 5% sodium carbonate solution. The aqueous layers were separated, combined, and stirred overnight at room temperature to hydrolyze the terminal oxazolinium cation. The cloudy mixture was acidified with diluted HCl to give a clear solution of pH < 6 and was then extracted with methylene chloride. The combined organic layers were dried with anhydrous magnesium sulfate and then concentrated. A white solid was precipitated by dropwise addition of diethyl ether to the concentrated solution, and the solid was collected by filtration and dried overnight in a vacuum oven to afford the desired polymer 1 (n ≈ 23) (3.1 g, yield 78%). ¹H NMR (400 MHz, CD₃OD) δ 3.51–3.70 (m, 4H, N-CH₂-CH₂-), 2.96–3.10 (m, 0.127 H, H₃C-N), 2.36–2.47 (m, 2H, C(O)-CH₂-CH₃), 1.10–1.11 (m, 3H, -CH₂-CH₃). To a suspension of linear polyethylenimine derived from oxazoline polymerization (1, 1 g, n ≈ 23) in 10 mL of water was added 15 mL of concentrated (35%) HCl. The reaction mixture was heated at 100 °C for 48 h to remove the propionyl groups. Excess of hydrogen chloride was removed under reduced pressure and the remaining white solid was dissolved in 10 mL of water. The resulting mixture was made alkaline by adding aqueous sodium hydroxide, and the formed white precipitate was filtrated and washed with water to afford the desired product linear polyethylenimine 2 (230 mg, yield 53%). ¹H NMR (400 MHz, CD₃OD) δ 2.77 (s, 4H, NH-CH₂-CH₂-). To a suspension of polymer 2 (200 mg, 4.65 mmol, n ≈ 23) in 10 mL of deionized water was added 4(5)-(hydroxymethyl)imidazole (1.04 g, 10.6 mmol) and potassium carbonate (4.4 g, 31.9 mmol). The reaction mixture was heated at 100 °C for 24 h and then cooled to room temperature. The residue was put into a dialysis tube and dialyzed against the following solvents (each 12 to approximately 24 h): 50% methanol in water, 25% methanol in water, and water. The final residue was evaporated and dried to afford the desired product 3 as a hydroscopic brown solid (281 mg, yield 49%) (grafting ratio: > 95% of ethylenimine units, according to ¹H integration). ¹H NMR (400 MHz, CD₃OD) δ 7.60–7.61 (m, 1H, imidazol), 6.69–6.96 (m, 1H, imidazol), 3.91 (m, 1H, N-CH₂-C), 3.60 (m, 1H, N-CH₂-C), 2.60–2.67 (m, 4H, NH-CH₂-CH₂-). Detailed description of materials and methods, the synthetic procedures, and characterization spectra are given in SI Text.

Acid-Base Titration.

The buffering capacities of polymers 2 and 3 were determined by acid-base titration. Then 0.25 mmol of 2 or 3 (concentration of ethyleneimine units) were added to 6 mL of deionized water, and then 5 mL of 0.1 N HCl was added to the suspension in order to make a clear solution and adjust the pH to ca. 2. Then 200 μL portions of 0.05 M NaOH were sequentially added to the solution, and the pH was measured after each addition with an UltraBASIC Benchtop pH Meter (UB-5) with an UltraBasic glass body pH electrode.

Determination of Binding Acticity.

UV spectra were recorded at room temperature, relative to poly(U) (180 nmol in U) and polymer 3 (n ≈ 23) dissolved in 5 mL of 10 mM phosphate buffer (Na₂HPO₄-NaH₂PO₄) at pH 7.55. CD spectra were recorded at room temperature, relative to poly(U) (375 nmol in U) and polymer 3 (n ≈ 23) directly dissolved in 5 mL of 10 mM phosphate buffer (Na₂HPO₄-NaH₂PO₄) at pH 7.5.

Enzymatic Kinetic Assay.

Buffer A refers to pH 7.00 (25 °C), 0.125 M Trizma® [tris-(hydroxymethyl)-aminomethane] base/succinic acid, 0.125 M NaCl, 18.8 mM MgCl₂. Buffers for the cleavage reaction were made by Trizma® base/Trizma® hydrochloride. The ionic strength was adjusted to 0.1 M with sodium chloride. The pH of the reaction solution was measured before the reaction at 25 °C and extrapolated to 80 °C, the reaction temperature, with the aid of the known temperature dependence. All buffers were kept frozen before use. Reaction solutions were prepared by combining appropriate amount of poly(U), p-nitrobenzenesulfonic acid sodium salt as internal standard, polymer catalysts, and buffer solution to give a reaction mixture (for polyvinylimidazoles, 30% ethanol in Trizma-HCl solution was used as the buffer). At appropriate intervals, samples were removed from the reaction mixture, cooled to room temperature, and immediately treated with venom phosphodiesterase I solution (approximately 25 units·mL^-1 in buffer A). The tube was capped, vigorously mixed, and submerged in a thermally equilibrated block heater at 40 °C. After incubating for 6 h, the samples were removed and placed in dry ice until ready for analysis by HPLC. Uridine concentrations in the reaction samples were calculated from the equation previously obtained using the initial concentration of the standard. The pseudo-first-order rate constant k_obs for the cleavage were obtained from dividing the slope of a plot of [Uridine] vs. time by the initial concentration of poly(U). A representative procedure, processing of cleavage sample HPLC spectra, and the results are described in SI Text.