Targeted gene silencing by small interfering RNAs (siRNAs) offers great potential for the treatment of different diseases ^[1], such as: cancer,viral infection,and genetic disorders. However, similar to most antisense or nucleic acid-based strategies,poor cellular uptake associated with low permeability of the highly regulated lipid cell membrane is the major limitation of siRNAs application. Safe and effective intracellular delivery of siRNAs remains the most challenging barrier to the clinical application of siRNAs. A variety of siRNA delivery systems have been developed and improved,including cationic lipids ^[2],polymers ^[3],peptides, nanoparticles ^[4],and antibodies ^[5]. Cell-penetrating peptides (CPPs),a group of short cationic vectors with remarkable capacity for plasma membrane translocation,have become one of the most efficient techniques utilized for the delivery of a wide range of cargoes,including small molecules,peptides,proteins,liposomes and oligonucleotides ^[6]. Several peptide based strategies have been developed to improve the delivery of siRNA and other oligonucleotides both in vitro and in vivo using either non-covalent complex approaches via electrostatic interaction or covalent approaches ^[7]. Compared to the more complex non-covalent strategy,the covalent strategy uses fewer CPPs linked to oligonucleotide fragment to achieve the same biological function with less non-specific effects and cytotoxicity.

To construct peptide-siRNA conjugates,both fragment conjugation and solid phase stepwise synthesis have been employed. Most of the papers reported synthesizing the CPP-siRNA conjugates using the fragment conjugation method,but few of them mentioned the purity and structure identification of the final product ^[8]. In this way,the silencing effect induced by siRNA could not be confirmed. In general,the fragment conjugation method is easy to carry out,but requires multiple purification steps,and usually the yields are low. With the experience of our previous work on antisense oligonucleotides,we now focus on the solid phase stepwise synthesis strategy for peptide-siRNA synthesis ^[9].

No matter which method will be used,chemical linkage is needed to link the peptide and oligonucleotide fragments,such as amide,thioether and disulfide bonds. Disulfide bonds present in many bioactive proteins play an important role in the construction of their secondary and tertiary structures ^[10]. Since disulfide bonds are reducible in intracellular fluids,thereby releasing the oligonucleotide in the cytoplasm and nucleus ^[11],reversible disulfide linkages have been widely used for drug delivery such as small molecule drugs,proteins,antibodies and oligonucleotides.

In this article,we described the synthesis of a disulfidemodified nucleoside which can be conveniently used for solid phase synthesis of peptide-siRNAs conjugates. After the reduction of the disulfide bond in intercellular fluids,there were no excess groups on oligonucleotide. The disulfide unit here could minimize the side effect of CPPs on the siRNA interference behavior ^[12] compared to conventional methods. The CPP sequences were elongated on the aminated CPG resin by Boc chemistry,and the modified nucleotide 1 was attached to the N terminal of the peptides. The following RNA was synthesized on this modified resin by standard phosphoramidite chemistry. After cleavage and purification,the final product was confirmed by MALDI-TOF-MS spectra.

The CPP sequence LALLAK (H-Leu-Ala-Leu-Leu-Ala-Lys-OH) is a signal peptide mimic developed by Li and Chen ^[9],which was linked to 3'-end of antisense oligonucleotides by amide bond and improved the permeability of the linked oligonucleotides through the cell membrane. In general,the CPG resin is not suitable for solid phase peptide synthesis due to its rigidity and poor expansion. To overcome this problem,a spacer (Fig. 1) was employed to increase the flexibility of the loading peptide and decrease the reaction room blockage. The loading efficiency could be detected from the DMTr protecting group of the spacer. On this modified CPG resin, the LALLAK peptide was prepared in the presence of microwave irradiation by the Boc strategy. The N-terminal amino group and peptide backbone are polar and they constantly try to align with the alternating electric field of the microwave,which helps in breaking up the chain aggregation. Thus the microwave irradiation improves both coupling/deprotection speed and product purity/ yield during the solid phase peptide synthesis ^[13]. After the linkage of peptide,succinyl oxide was introduced to the N-terminal amino groups of the peptide to change the terminal functional group from amino to carboxyl (Scheme 1).

Fig. 1

	Download: JPG larger image
Fig. 1. Structures of spacer and L1.

Scheme 1

	Download: JPG larger image
Scheme 1. Solid phase peptide synthesis on AP-CPG.

In order to link siRNA to CPP,a modified deoxyuridine 1 was designed and synthesized. The disulfide bond was introduced to the 3' terminal of the sugar ring which could keep the modified nucleosides as natural after the cleavage of CPP. Compound 1 was synthesized via a multiple-step reaction using 2'-deoxyuridine as the starting material. Firstly,2'-deoxyuridine was converted to anhydro dU derivative 2 in three steps by reported procedure ^[14]. Then,sodium thiobenzoate was employed as the nucleophile reagent to attack 3'-C and open the intermolecular ring to form compound 3 ^[15]. The thioester 3 was converted to the corresponding thiol derivative by hydrolysis and then the thiol derivative reacted with L1 (Fig. 1) under basic condition to give 20- deoxynucleoside disulfide 1 (Scheme 2).

Scheme 2

	Download: JPG larger image
Scheme 2. Synthesis of modified deoxynucleoside.

This modified nucleoside 1 was coupled to CPG-P-COOH resin under condition HBTU/DMAP in acetonitrile to give CPG-P-DMT (Scheme 3). Initially,the coupling reaction was carried out according to the situ-neutralization protocol HBTU/HOBt/DIEA ^[16],but the loading rate was too low. The possible reason is that the carboxyl group on the resin might not be able to form the active ester for further coupling. Finally,the condition of HBTU (2 equiv.) and DMAP (2 equiv.) in acetonitrile (4 mL) gave the good yield. The coupling was confirmed by DMTr analysis. The calculated loading of nucleoside on the CPG-P-DMT was 54 μmol/g.

Scheme 3

	Download: JPG larger image
Scheme 3. Loading the modified nucleoside on CPG.

The oligonucleotide sequence synthesis was carried out on the CPG-Pep-DMT resin with Applied Biosystem 392 DNA/RNA Synthesizer. A peptide-DNA conjugate I was firstly synthesized to test the feasibility of the method and for further optimizing. After HPLC purification,the molecular weight of the final product was confirmed by MS spectra which were concordant with the calculated value of the designed sequence. The other two peptide-RNA conjugate II and IV were synthesized by the same method. After cleavage from the resin by concentrated aqueous ammonia,the 2'-OTBDMS protect group was removed by TBAF. The product was purified with an anion exchange column and characterized by MALDI-TOF-MS (Table 1) spectra after lyophilization. Finally,the single chain conjugates were annealed with their naturally complementary RNA strands to generate peptide- siRNA conjugates.

Table 1

Table 1 Characterization of peptide-oligonucleotide conjugates.

Table 1 Characterization of peptide-oligonucleotide conjugates.

Conjugates VI and VII were obtained by annealing sequences III, V and II,V. The duplex siRNAs were used to inhibit the expression of cdc-2 mRNA. The inhibition efficiency was detected by siQuant^TM approach ^[17] in HEK-293 cells. By means of the dual-luciferase test,post-treatment activity of mRNA was measured to indicate the silencing effect,wherein a smaller value means better silence of target RNA. Under the same conditions, when targeting the corresponding mRNA of antisense strand,the conjugate VII showed almost the same silencing activity (92.2%) as positive control VI (93.0%),illustrating that peptide modification on the sense strand had no discernible impact on the activity of the antisense strand. With the corresponding mRNA of sense strand as the target,the conjugate VII showed about 30% loss of activity compared to native siRNA (Fig. 2,red bar).

Fig. 2

	Download: JPG larger image
Fig. 2. Gene silencing efficiency of peptide-siRNA conjugates.

Both strands of siRNA can be loaded into the RNA induced silencing complex (RISC),leading to knockdown effect of the corresponding target,which may be due to the competitive loading of the two single strands into the RISC ^[18]. Interestingly, compared with unmodified siRNA VI,the sense strand 3' terminal conjugate VII displayed decreased silence efficiency when targeting its relative mRNA (Fig. 2,red bar). This suggested that 3' terminal peptide modification could mitigate the off-target effect through suppressing of the sense strand loading into RISC ^[19], which is consistent with the results of isonucleosides incorporated siRNAs ^[20]. The activity assay of antisense 3' terminal conjugates and double strands conjugates is ongoing.

In summary,we have developed a concise route to synthesize peptide-siRNA conjugates by a solid-phase chemistry strategy. With this strategy,peptides and siRNAs can be conjugated easily and efficiently,regardless of sequence diversity. Biological assay results showed that these kinds of peptide-siRNA conjugates retain their gene silencing efficiency while the off-target effect is reduced due to the difficulty of modified strand to load into the RISC. Although much more work is necessary to improve the membrane permeability of the conjugates,and to further optimize the sequence of peptide,a new strategy has been explored which provides more choice of modification for the application of siRNAs in the future.

We thank Prof. Zi-Cai Liang,Quan Du and Fan Yi in Peking University for providing siRNA sequence and assistance with biological assay experiments. This work was supported by the National Natural Science Foundation of China (No. 20932001) and the Ministry of Science and Technology of China (Nos. 2012CB720604,2012AA022501).