Specific Silencing of L392V PSEN1 Mutant Allele by RNA Interference

RNA interference (RNAi) technology provides a powerful molecular tool to reduce an expression of selected genes in eukaryotic cells. Short interfering RNAs (siRNAs) are the effector molecules that trigger RNAi. Here, we describe siRNAs that discriminate between the wild type and mutant (1174 C→G) alleles of human Presenilin1 gene (PSEN1). This mutation, resulting in L392V PSEN1 variant, contributes to early onset familial Alzheimer's disease. Using the dual fluorescence assay, flow cytometry and fluorescent microscopy we identified positions 8th–11th, within the central part of the antisense strand, as the most sensitive to mismatches. 2-Thiouridine chemical modification introduced at the 3′-end of the antisense strand improved the allele discrimination, but wobble base pairing adjacent to the mutation site abolished the siRNA activity. Our data indicate that siRNAs can be designed to discriminate between the wild type and mutant alleles of genes that differ by just a single nucleotide.


Introduction
Alzheimer's disease (AD) is the most common form of a progressive dementia in humans. The most frequently occurring variant is a "late-onset" (LOAD), sporadic form of the disease developing usually after age 65. The LOAD appears as a result of a complex interaction among environmental factors and individual predisposing genetic traits. Almost sixty years ago, Sjögren et al. demonstrated that some patients with AD had an autosomal dominant Mendelian pattern of disease inheritance [1]. Early onset familial cases of Alzheimer's disease (FAD) are rather rare and account for only a few percent of the total population of patients but in this form disease symptoms appear at an unusually early age, between  year of life. Mutations in the three genes coding Amyloid Precursor Protein (APP), Presenilin 1 (PSEN1), and Presenilin 2 (PSEN2) were identified as responsible for development of the dementia in almost 50% of patients with FAD. Identification of the amyloid-beta (Aβ) peptides isolated from brains of patients with AD and trisomy 21 (Down syndrome) suggested localization of the parent APP gene on chromosome 21 [2][3][4][5]. Linkage to disease mapped in FAD families led to the discovery of the first autosomal dominant missense mutations in APP segregating with disease risk [6]. This was followed by identifications of autosomal dominant FAD mutations in the PSEN1 [7] and PSEN2 [8,9] genes, on chromosomes 14 and 1, respectively. The summary of FAD mutations is maintained at the Alzheimer's Disease and Frontotemporal Dementia Mutation Database [10]. Accordingly, there are currently described totally 227 sequence defects in the human FAD genes: 32 mutations in APP gene, 181 mutations in PSEN1 gene, and 14 mutations in PSEN2 gene, identified, respectively, in 89, 399, and 23 families.
Human Presenilin 1 (PSEN1, called alternatively as PS1, AD3, FAD, S182) is a 467-residues, ∼50 kDa protein 2 International Journal of Alzheimer's Disease with 9-TMD (transmembrane domain) topology [11,12]. PSEN1 together with nicastrin (Nct), anterior pharynxdefective-1 (APH-1) and presenilin enhancer-2 (PEN-2) constitutes a multisubunit membrane-bound protease complex of γ-secretase and is essential for its activity, stability and interaction between all γ-secretase components [13][14][15]. In the γ-secretase complex, Presenilin 1 is cleaved, in intramolecular autocatalytic event, into N-and Cterminal fragments (∼30 kDa NTF and 20 kDa CTF), which form an active heterodimer. Several mutations in PSEN1 gene, detected in FAD patients, can inhibit processing of polypeptide chain. The most well-studied PSEN1 function in γ-secretase complex is catalysis of the intramembranous cleavage of a number of proteins, for example, Notch, APP, N-, and E-cadherins. Due to the fact that only a fraction of cellular protein actually forms the γ-secretase complex [16], it was postulated that Presenilin 1 may have functions beyond those of the γ-secretase complex. For example it is involved in neuronal Ca 2+ signaling and homeostasis [17], and it also interacts with and stabilizes β-catenin [18,19]. The main role of the γ-secretase is participation in the APP processing. The CTF 83 and CTF 99 fragments of APP, formed after its proteolysis by αor β-secretases, respectively, are cleaved within their transmembrane regions to release the C-terminal domain (AICD) and generate Aβ peptides. Nearly 90% of the secreted amyloid peptides are Aβ40 or shorter forms, while the remaining 10% are Aβ42 or longer peptides. AD-linked mutations in PSENs genes influence the γ-secretase cleavage-site specificity, favouring cleavage at the position 42 relative to the position 40, thus increasing the Aβ42/Aβ40 ratio. Research from the 1990s indicates that mutations in the PSEN1 gene may be responsible for 30-60% of early onset Alzheimer's cases. The hydrophilic loop (amino acids 263-407) of PSEN1, in which many pathogenic mutations are localized, appears to be crucial for the protein function, since it includes the binding domains to different PSEN1 partners [20].
Dominantly inherited disorders constitute particularly attractive targets for allele-specific gene silencing by short interfering RNAs due to the fact that patients with these disorders carry both the wild type gene and mutant allele causing the disease. Specific silencing of a mutant allele that expresses the toxic form of the protein, without reducing the level of the wild type allele, might constitute a promising approach for the cure or prevention of such disorders. Most of the disease-associated genes differ by a single point mutation, making them targets of choice for allele-specific silencing. The list of trials on silencing mutant alleles associated with neurodegenerative disorders includes Huntington disease (HD) [21][22][23][24][25][26][27], Parkinson disease (PD) [28], amyotrophic lateral sclerosis (ALS) [29][30][31][32][33], spinocerebellar ataxia (SCA) Type 1 (SCA1) [34], and Type 3 (SCA3) causing Machado-Joseph disease [35,36], frontotemporal dementia with Parkinsonism linked to chromosome 17 (FTDP-17) [35], slow channel congenital myasthenic syndrome (SCCMS) [37], and prion proteininduced disease [38]. Although several studies have used RNAi to investigate the effect of silencing of PSEN1 and its homolog PSEN2 on γ-secretase processing and cell metabolism [39,40], to our best knowledge no studies have been reported on the allele-specific silencing of any PSEN1 mutant allele. Mutation L392V in PSEN1 carries the single nucleotide substitution (C → G) at position 1174 of coding sequence in exon 11, followed by amino acid substitution (Leu → Val) at the position 392 of polipeptide chain in TMD 7. Mutant form of PSEN1 is involved in the production of longer forms of the Aβ peptide resulting in its accumulation in the brain, and, in consequence, generation of earlyonset Alzheimer's disease. Mean age onset of the disease in patients with L392V PSEN1 mutation was determined as 42.5 years and mean age of death was determined as 52.6 years [10]. According to data from Alzheimer's Disease and Frontotemporal Dementia Mutation Database, this mutation has been already found in 50 patients with FAD in European (Italy, France) and Asian (Japan) population [10,[41][42][43][44][45][46]. We selected the L392V mutant and wild type alleles of Presenilin 1 as a model system for the evaluation of allelespecificity of siRNA duplexes. A series of siRNA molecules fully complementary to the mutant gene were screened first by a dual fluorescence assay, then by FACS flow cytometry and their activity was confirmed by fluorescence microscopy and observation of reduced level of Aβ42 determined by ELISA. Our data suggest that siRNAs can be designed to discriminate between the wild type and the mutant alleles that differ by just a single nucleotide.

Synthesis and Purification of RNA Oligonucleotides.
The oligoribonucleotides were synthesized according to the routine phosphoramidite approach [47], using LCA CPG glass support and commercially available nucleoside phosphoramidites (ChemGenes). Synthesis was performed on the Gene World DNA synthesizer under the conditions recommended by the manufacturer. Oligonucleotides were cleaved from the solid support as 5 -DMT-derivatives, then deprotected and purified according to the described procedure [48]. Support-bound oligonucleotides were treated with 33% ethanolic methylamine (Sigma-Aldrich) and DMSO 1 : 1 (v : v) mixture at 65 • C for 15 min and then with triethylamine-trihydrofluoride (Sigma-Aldrich) at 65 • C for 15 min. The reaction mixture was frozen at −20 • C for 30 min, quenched with cold 1.5 M NH 4 HCO 3 and poured into a conditioned SepPak cartridge (Waters). Shorter oligomers were eluted with 14% CH 3 CN in 50 mM CH 3 COONa and 50 mM NaCl. The remaining oligomer was treated with 2% aqueous TFA (trifluoroacetic acid) for 15 min at room temperature and washed with water, 1 M NaCl, and water. The product was eluted from the cartridge with 30% CH 3 CN. The structure and purity of oligomers were confirmed by MALDI-TOF mass spectroscopy (Table 1) and by 20% polyacrylamide/7 M urea gel electrophoresis (data not shown).

2.2.
Assembly of siRNA. siRNA duplexes were assembled in phosphate saline buffer (PBS, without Ca 2+ and Mg 2+ ) by mixing equimolar amounts of complementary oligonucleotides, heating at 95 • C for 2 min, and slow cooling down Two cell lines of human fibroblasts (i) expressing wild type PSEN1 (CELMA) and (ii) L382V PSEN1 mutant (NOV4), were cultured in Dulbecco's MEM (Sigma-Aldrich Co., Saint Louis, MO), supplemented with 10% FBS and antibiotics (100 units/mL penicillin and 100 mg/mL streptomycin) in 24-well plate at 37 • C and 5% CO 2 . Before transfection, the culture medium was replaced with fresh medium, free of antibiotics. Transfection was performed using Lipofectamine 2000 transfection reagent (Invitrogen) and appropriate siRNAs at the ratio 2 : 1. To test the effect of siRNA on Aβ level, the cells were transfected with siRNAs (150 or 300 nM). After transfection, cells were incubated for 5-6 hours and then medium with transfection mixture was replaced with the fresh, culturing medium with antibiotics and 5% FBS. After 72-hour incubation at 37 • C and 5% CO 2 , cells were washed three times with PBS buffer (without Ca 2+ and Mg 2+ ), the fresh medium with 0.2% FBS was added and cells were incubated for the next 24 hours. Then the culturing medium was collected with 1 mM AEBSF (4-(2-Aminoethyl) benzenesulfonyl fluoride hydrochloride, serine protease inhibitor, Sigma) for determination of Aβ42 level by ELISA (Enzyme-linked immunosorbent assay).

Dual Fluorescence Assay.
Fluorescence values of EGFP and RFP (red fluorescent protein) were determined using a Synergy HT reader (BIO-TEK). Quantification of data was done with KC4 software. Excitation and emission wavelengths for each protein were as follows-EGFP: λ Ex = 485/20 nm and λ Em = 528/20 nm and RFP: λ Ex = 530/25 nm and λ Em = 590/30 nm. The activity of siRNA duplexes was calculated as the ratio of EGFP to RFP fluorescence values according to the following equation. Activity of siRNA (%) = 100%− (sample EGFP-X/RFP : control EGFP-X/RFP) ×100% were (EGFP-X means the fluorescence value of EGFP-X fusion protein, X = Wt-PSEN1 or Mut-PSEN1).
Every time an average value of fluorescence was mean of eight repeats calculated after eliminating extreme values. Each siRNA activity value given on the plots is the average of mean values from three independent experiments. The level of fluorescence (EGFP/RFP) in control cells (transfected with pDsRed-N1 and pEGFP-Wt PSEN1(400) or pEGFP-Mut-PSEN1(400) plasmids and control nonsilencing siRNA duplex) was taken as a reference (100%).

Determination of Aβ42
Level. Chromogenic sandwich ELISA kit (Signet Laboratory, Denham, MA) was used according to the manufacturer's procedure. To analyze the level of Aβ42 in conditioned medium (after transfection of siRNAs) collected as described in p.2.4., the 100 μL samples of medium were placed in a 96-well ELISA plate, precoated with a capture antibody (specific to the N-terminus of human Aβ42 peptide) and incubated at 4 • C for 16 hours. After washing, a reporting antibody, that binds to the Cterminus of Aβ42, was added. After incubation for the next 2 hours and extensive washing, a horseradish peroxidase conjugated secondary antibody was added. Color developed with substrate o-phenylenediamine was measured at λ 492 using a Synergy HT reader, and received data were quantified using KC4 software. The siRNA activity was calculated as the ratio of Aβ42 in experimental and control sample (cells treated with transfection reagent only). Results showed at the graph ( Figure 6) are the average of two independent experiments.

Statistical Analysis.
The values of EGFP-PSEN1 fluorescence were normalized to the values of RFP fluorescence for each studied sample. All results are given as a percent of the fluorescence of sample toward fluorescence of control cells (transfected with Lipofectamine 2000 only or with control, nonsilencing siRNA) with standard deviation of mean. The Shapiro-Wilks' W test was used to analyze the normality of distribution of the resulted data. Statistical analysis of differences between two groups of data (Wt-PSEN1 versus Mut-PSEN1) were calculated by the use the Student's t-test (for data with normal distribution) or by the nonparametric Mann-Whitney's U-test. The differences with P < .05 were considered as statistically significant.

Construction of the Expression Plasmids Carrying the Wild
Type and Mutant Alleles of PSEN1 Gene. The wild type and 1174 C → G, L392V mutant of PSEN1 gene were selected as a model system to assess the allele-specificity of the siRNAs and to select the duplexes, that discriminate between two target RNAs differing by the single nucleotide. Plasmids coding the wild type and mutant type genes were prepared as described in Section 2. At first, the full length (1404 bp) wild type and mutant PSEN1 genes were cloned in ORF (open reading frame) at the 3 -UTR of EGFP gene coded in the pEGFP-C1 plasmid. The expression products were designed to contain the EGFP protein fused by five amino acids linker (-Ser-Gly-Leu-Arg-Ser-) with intact Wt-PSEN1 or Mut-PSEN1 protein.
The resulting plasmids pEGFP-Wt-PSEN1(1400) or pEGFP-Mut-PSEN1(1400) were cotransfected with the reporter plasmid pDsRed-N1 coding RFP, into HeLa cells. After 48 hour incubation, only very low levels of the fluorescence of EGFP fusion proteins were observed. Probably, attaching the 473-aa long polypeptide chain at the C-termini of EGFP resulted in abnormal folding of the fusion protein and, in consequence, in the low level of the fluorescence. Therefore these plasmids were excluded from further experiments and new, shorten fusion genes were prepared, containing a 344 bp 3 -terminal fragment of PSEN1 gene, in the wildtype and mutated versions, attached to the 3 -UTR of EGFP gene in the pEGFP-C1 expression plasmid. Additionally, the translation stop codons TAG and TAA were introduced into EGFP upstream of the coding sequence of fragment of the PSEN1 gene. The length of the obtained plasmids was confirmed by 0.5% agarose gel electrophoresis (Figure 1). We expected the high value of the fluorescence of EGFP in the case of intact fusion transcript EGFP-PSEN1 and low or no fluorescence if the fusion transcript is degraded by RNAi. Results of the dual fluorescence assay for all reporter plasmids (pEGFP-Wt-PSEN1(1400); pEGFP-Mut-PSEN1(1400); pEGFP-Wt-PSEN1(400) and pEGFP-Mut-PSEN1(400)) are shown in Figure 2. Plasmids pEGFP-Wt-PSEN1(400) and pEGFPMut-PSEN1(400) were used in further studies of the siRNA activity toward the Wt-and Mut-PSEN1 genes both expressed in HeLa cells.

Design, Synthesis, and Evaluation of the Activity of siRNAs
Targeting Mutant Allele of PSEN1 3.2.1. siRNA Molecules. We synthesized a set of 19 siRNAs, designed across the L392V point mutation in the human PSEN1 gene. The L392V mutant contains a guanosine at the position 1174 of mRNA, whereas the wild-type mRNA bears a cytidine at that position. Each siRNA antisense strand was fully complementary to the mutant PSEN1 mRNA, but in relation to the wild type mRNA contained a C:C mismatch ( Figure 3). The structure of all tested oligonucleotides was confirmed by mass spectrometry (MALDI-TOF MS, Table 1), and their purity was assessed on 20% denaturing PAGE (data not shown). All siRNA duplexes used in these experiments had the typical structure of 19 bp fully complementary duplex with 2-nt overhangs at each 3 -end with typically used two thymidine units (TT). This approach was intended for the proper interaction of RNA duplex with proteins of RNAi machinery [49]. Ago2 protein, the main catalytic core of the RISC (RNA-induced silencing complex) binds the guide strand, which leads the complex to the complementary mRNA strand. Ago2-mediated RNA cleavage requires perfect Watson-Crick base pairing between the guide strand of siRNA and the target mRNA, spanning both, the seed region (positions from 2nd to 8th nucleotide) and the cleavage site (phosphate group opposite to the internucleotide linkage located between 10th and 11th nucleotide, counted from the 5 -end of the guide strand [50].

Silencing Activity Measured in the Dual Fluorescence
Assay. In order to evaluate the silencing activity of siRNAs, we used the already described dual fluorescence reporter system [51][52][53][54][55]. This assay is based on the measurement of the relative fluorescence intensity of EGFP and RFP proteins, expressed from exogenously delivered plasmids cotransfected together with siRNA. In our case, the HeLa cells were transfected with pEGFP-Wt-PSEN1(400) or pEGFP-Mut-PSEN1(400) and pDsRed-N1 as well as with P1-P19 siRNAs listed in Table 1 and Figure 3. The results of the fluorescence measurements are shown in Figure 4. The P8-P11 siRNAs, with the perfect complementarity to the mutant allele, but in the same time with one mismatch (cytidine:cytidine, C:C) to the wild type gene placed at the central positions of the duplex, efficiently silenced the mutant gene (up to 90%, P < .0001), while only slightly reducing the wild type gene expression (silencing ca. 10-30%, P < .0001). Interestingly, duplexes P1 and P2 showed a very strong silencing property towards both-the wild type and the mutated genes. This high silencing activity of the latter duplexes may be related to the high thermodynamic asymmetry of the duplex ends (discussed in Section 3.4) in comparison to other siRNA duplexes or to the better access of siRNA to the target region of the mRNA. The remaining P3, P5, P6, P7, and P12-P19 duplexes were almost inactive in relation to both the wild type as well as the mutant gene.

Silencing Activity Measured by Flow Cytometry.
HeLa cells cotransfected with pEGFP-Wt-PSEN1(400) or pEGFP-Mut-PSEN1(400), pDsRed-N1 plasmids, and with selected duplexes P2, P7, P8, P9, P10, P11, P12, and P16 siRNAs were analyzed by FACS flow cytometry. The amount of cells containing the EGFP and RFP fluorescence in the tested samples (cells transfected with siRNAs) was determined and compared with the amount of cells in control samples (cells treated with transfection reagent only or treated with nonsilencing, control siRNA). The duplexes perfectly complementary to the L392V PSEN1 mRNA but with one mismatch (C:C) towards the wild type gene, lockated at the central position of the antisense strand (as P8 and P9), exhibited the highest allele-specificity and silencing efficiency exclusively towards the mutant gene ( Figure 5). Thus, the data obtained in this analysis confirmed the above described results.

Evaluation of the Level of Aβ42 by ELISA.
The activity of siRNAs towards endogenous target genes was evaluated by the measurement of the level of extracellular Aβ42 released from the transfected cells. For this purpose human fibroblasts derived from the patients with FAD, carrying L392V mutantion in PSEN1 (NOV4), were used [56,57]. Human fibroblasts, carrying the wild type of PSEN1 (CELMA), derived from healthy individuals were used as a control [56,57]. The amount of the Aβ42 peptide was determined by the sandwich ELISA test. The normalized results obtained for this assay for both types of cells transfected with control, nonsilencing, and P10 siRNA are shown in Figure 6.  3  3   3  3  3  3  3  3  3  3  3  3  3  3  3  3  3  3  3 3 3 towards the mutated PSEN1 gene did not influence the level of Aβ42 released from control fibroblasts (CELMA Aβ42 ), which was much lower than in NOV4 cells (both NOV4 Aβ42 and CELMA Aβ42 levels are showed at the graph as 100%). The results obtained may constitute a proof-of-concept that silencing of the mutant allele of PSEN1, exerted by fully complementary siRNA, led to the reduced expression of the mutant Presenilin1, which in consequence gave less of the toxic Aβ42 product.

Modulation of Allele-Specific Activity of siRNAs by
Chemical Modifications 3.3.1. Modification of siRNA with 2-Thiouridine. The allelespecific activity was evaluated for siRNA modified with naturally occurring 2-thiouridine (s 2 U) nucleoside. As it has been already demonstrated, s 2 U, due to preferred C3endo conformation of the ribose ring, improves the thermodynamic stability of s 2 U-containing double stranded RNA helices [52,58,59]. Additionally, s 2 U unit, when present in the anticodon region of transfer RNA (tRNA), enhances its specificity of Watson-Crick base pairing with A and restricts wobble base pairing with G [58,59]. We choose P10 duplex, which previously has shown satisfactory discrimination between wild type and mutant alleles, and introduced the s 2 U unit at the position 18 of the antisense strand (P10thio) (Figure 7(b)). The s 2 U modification at this position of the RNA duplex (the 3 -end of duplex) improves its thermodynamic asymmetry and facilitates incorporation of the antisense strand into the RISC complex [52]. Moreover, s 2 U introduction can impact positively on the uptake of siRNA into cells, due to the presence of sulfur moiety, which increases the hydrophobic properties of the molecule. Results of the dual fluorescence assay performed for HeLa cells transfected with P10-thio siRNA are presented in Figure 7(c). It is clearly shown that introduction of the s 2 U unit at the 3 -end of the antisense strand of P10 siRNA further improved allele discrimination (∼10%) due to the increased thermodynamic asymmetry of the duplex.

Silencing Activity of Sulfur-Containing siRNA Evaluated by Fluorescence
Microscopy. The activity of the selected siRNAs was also observed by the fluorescence microscopy. Figure 8 shows fluorescence images of the fixed samples obtained from HeLa cells transfected with P10, P10-thio, and P11 siRNA duplexes. Here again, the cells carrying the wild type gene (EGFP-Wt-PSEN1(400)) expressed green fluorescent protein at the high level and independently of the siRNA sequence. In contrast, slides containing cells expressing the mutant gene (EGFP-Mut-PSEN1(400)) have shown less fluorescently labeled cells, and the most pronounced result was obtained for cells transfected with the P10thio siRNA duplex, confirming high discrimination ratio of siRNA against the wild type and mutant alleles of the PSEN1 gene.

Modification of siRNA with Wobble Base Pair.
The region near the central positions at the antisense strand of siRNA (10th-11th) is interesting, primarily from the perspective of a nucleolytic reaction catalyzed by Ago2 [49,50]. To investigate the influence of correct base pairing in this region on allele-specificity and the overall activity of RNA duplex, the wobble type modification was introduced at the two positions of the antisense strand, namely position 9 or 11 counting from the 5 end, adjacent to the mutation site (Figure 9(a)). According to our earlier assumption, introduc-   tion of such structural disorder into RNA duplex reduces the thermodynamic stability of the duplex and weakens the interaction between guide strand and the target mRNA. We supposed that the wobble modification could increase the allele-specificity of siRNAs targeted towards the mutant gene. Instead, we observed significant deterioration of the silencing activity of both modified siRNAs (P10W9 and P10W11) (Figure 9(b)). Most probably, introduced disorder in the base pairing like wobble G:U considerably disturbs the recognition of the double stranded RNA helix (containing mRNA and the antisense strand of siRNA) by Ago2, what results in inhibition of the silencing process.

Thermodynamic Properties of siRNA Duplexes (P1-P19)
versus Their Silencing Activity. Thermodynamic stability of RNA duplexes and secondary structure of RNA are often predicted by using free-energy parameters from a near-neighbor model [60]. To assess the relative thermodynamic stability Table 2: Differences between the free Gibbs energy of 5 -end and 3 -end of the duplex (resp. to the polarity of the antisense strand), calculated independently for three and for four consecutive base pairs shown for each duplex (ΔΔG 37 • C = ΔG 37 • C (5 -end) − ΔG 37 • C (3 -end)). The values of ΔG 37 • C were calculated according to [60].   [60]. The standard-state free energy of helix formation was counted as the sum of the following terms: (i) a freeenergy change for helix initiation associated with forming the first base pair in the duplex, (ii) the sum of propagation free energies for forming each subsequent base pair, and (iii) freeenergy increments for unpaired terminal nucleotides. Then, differences between the free Gibbs energy of the 5 -end and the 3 -end for each duplex were calculated independently for three and four consecutive base pairs according to the following equation: The results of this analysis are shown in Table 2 Wt-PSEN1 Mut-PSEN1     Positions of the C-C mismatch between the guide strand of siRNA and the wild-type PSEN1 mRNA are indicated by the black dots. Red dot represents the site of the s 2 U modification at the antisense strand. The level of the relative EGFP/RFP fluorescence in cells transfected with control nonsilencing siRNA was used as 100%. Statistical analysis of differences between two groups of data (Wt-PSEN1 versus Mut-PSEN1) were calculated by the use the Student's t-test (for data with normal distribution) or by the nonparametric Mann-Whitney's Utest. The differences with P < .05 were considered as statistically significant. and in Figure 10. The positive ΔΔG 37 • C values indicate the higher thermodynamic stability of the 3 -end of the duplex, thus the antisense strand should be selected as the guide strand and the siRNA duplex should be more active. The negative values of ΔΔG 37 • C mean the higher thermodynamic stability of the 5 -end of the duplex, thus the antisense strand is not preferred to be selected as a guide strand and, in consequence, the siRNA duplex would be expected to be less active towards the target gene. Comparison of the two graphs representing the silencing activity (Figure 10(a)) and the thermodynamic features of the duplex ends ( Figure 10(b)) can explain observed differences in the activity of the used siRNAs, for example, duplexes P3-P6 and P14-P17 are characterized by the negative values of Gibbs free energy and are neglectably active. On the contrary, duplexes P1, P2, and P7-P10 showing positive or close to zero values of Gibbs free energy are more active as RNAi triggers. In some cases, silencing efficacy of siRNA duplexes depends on other factors as the secondary structure of target mRNA or availability of target sequence for RNAi proteins [61]. Also the other, not identified factors may govern the silencing process of given mRNA targets with short interfering RNAs.

Conclusions
Dominantly inherited disorders, caused by the presence of the toxic mutant allele besides the wild type gene, constitute attractive targets for allele-specific gene silencing. Gene therapy aimed at the elimination of the mutant allele causing the disease can be achieved by an RNA interference technology. Clinical success of such therapies depends mostly on the ability of RNAi to discriminate between the wild type and mutant alleles. In this paper, we demonstrate the allele-specific silencing activity of the short interfering RNAs directed towards mutant allele of PSEN1 gene. L392V PSEN1 mutant gene contains 1174 C → G mutation, which results in   The results are mean values from three independent experiments. Statistical analysis of differences between two groups of data (Wt-PSEN1 versus Mut-PSEN1) were calculated by the use by the nonparametric Mann-Whitney's U-test. The differences with P < .05 were considered as statistically significant.
amino acid substitution, Leu to Val, at the position 392 of polypeptide chain of the protein. This mutation was found in patients with early onset FAD. At first, we synthesized a set of siRNAs, designed across the point mutation and determined their activity towards the mutant and wild type genes by a dual fluorescence assay, flow cytometry assay, and fluorescence microscopy images. The most active and allele-specific were the siRNA duplexes fully complementary to the target gene, which paired with the wild type allele with the C:C mismatch located in the central part of the mRNA-siRNA duplex (P8-P11). These duplexes silenced the mutant gene up to 90% and only slightly reduced the wild type gene, up to 10-30%. Introduction of the 2-thiouridine nucleoside into the 3 -end of the siRNA duplex (in respect to the polarity of the antisense strand) increased silencing activity of the duplex and discrimination ratio between the  Figure 10: Comparison of the activity of siRNAs targeted Mut-PSEN1 gene (a) with their thermodynamic data (b). siRNAs were fully complementary to the Mut-PSEN1 transcript. Differences between the free Gibbs energy of 5 -end and 3 -end of the duplex (resp. to the polarity of the antisense strand) were calculated for three and for four consecutive base pairs: (ΔΔG 37 • C = ΔG 37 • C (5 -end) − ΔG 37 • C (3 -end)) (compare with Table 2). wild type and mutant allele (P10-thio). The wobble base pair modification introduced at the 9 or 11 positions of the antisense strand (P10W9 and P10W11), adjacent to the mutation site, reduced the thermodynamic stability of the duplex. We observed significant deterioration of the silencing activity of both modified siRNAs. Most probably, introduced disorder in the base pairing (G:U) considerably disturbed the recognition of the double stranded RNA helix (containing mRNA and the antisense strand of siRNA) by Ago2, what resulted in inhibition of the silencing process. The activity of the screened siRNAs was also evaluated by measurement of the level of extracellular Aβ42 released from the primary fibroblasts, carrying the (L392V) PSEN1 mutant (NOV4) in comparison to control fibroblasts (CELMA), carrying the wild type PSEN1, determined by ELISA. The normalized results obtained for NOV4 cells transfected with P10 and P10-thio siRNA showed reduced amount of Aβ42 by ca. 40% compared to the control fibroblasts treated only with transfection reagent. This result is a proof-of-concept that silencing of the mutant allele of PSEN1, exerted by fully complementary siRNA, led to reduced expression of the mutant Presenilin 1, which in consequence gave less of the toxic Aβ42 product.