Main

Impaired bone formation occurs in several varieties of dysfunctional bone homeostasis. To date, intermittent injection of recombinant human parathyroid hormone (iPTH) is the only bone anabolic agent clinically approved for stimulating bone formation in severe osteoporosis1,2. However, iPTH treatment is limited to a 2-y period because of increasing bone resorption over bone formation and a potential risk of developing osteosarcoma in patients receiving iPTH treatment3,4,5. This limitation has provided an incentive to search for new, safe bone anabolic drugs that do not activate bone resorption.

RNA interference (RNAi), a natural cellular process that regulates gene expression through a highly precise mechanism of sequence-directed gene silencing, could theoretically be used to target any disease-associated pathogenic gene of interest6. Accordingly, RNAi-based therapies targeting those genes that have been identified to negatively regulate bone formation without modulating bone resorption could facilitate translational therapy for treating diseases marked by impaired bone formation7. However, there is a major concern that the large therapeutic doses of systemically administered siRNA that would be needed to stimulate sufficient bone formation may carry a high risk for adverse effects in nonskeletal tissues8. This concern leaves the field with a great challenge when considering the use of these treatments6. Thus, development of a specific delivery system for RNAi-based therapies that addresses this issue is highly desirable.

The niche surrounding the bone-formation surfaces is predominantly occupied by osteogenic-lineage cells at various stages of differentiation3. All of these cells could be potential targets of pro-osteogenic siRNAs. A practical strategy, then, is to develop a generalized siRNA delivery system that selectively targets bone-formation surfaces to facilitate the delivery of therapeutic siRNAs to the majority of the osteogenic-lineage cells. Such a delivery system would probably allow for a highly targeted dose of therapeutic siRNA to be delivered to the bone while avoiding possible negative side effects to nonskeletal tissues, thus increasing the efficacy and safety of RNAi-based bone anabolic therapy.

To date, two types of stable bone-targeting molecules, bisphosphonates and oligopeptides, have been used to target bone after they have been coupled to nonspecific bone therapeutic agents9. Eight repeating sequences of aspartate (Asp8), one of the representative targeting oligopeptides, has been reported to preferentially bind to bone-resorption surfaces, whereas alendronate, one of the representative bisphosphonates, is distributed to both bone-formation surfaces and bone-resorption surfaces10. However, there remains a lack of bone-targeted molecules that have high selectivity for bone-formation surfaces over bone-resorption surfaces.

The physical chemistry of bone-formation surfaces covered with osteoblasts is characterized by lowly crystallized hydroxyapatite, as well as amorphous calcium phosphonate, whereas the physical chemistry of bone-resorption surfaces covered with osteoclasts is characterized by highly crystallized hydroxyapatite10. The stronger affinity of Asp8 to highly crystallized hydroxyapatite rather than lowly crystallized hydroxyapatite in vitro provides an explanation for the preferential binding of Asp8 to bone-resorption surfaces. Recently, (AspSerSer)6 was found to favorably bind to mantle dentin, which consists of small and randomly oriented crystals, rather than the enamel surface, which consists of elongated and well-oriented hydroxyapatite crystals11. Therefore, we postulated that these different bindings of (AspSerSer)6 might depend on the crystallinity of hydroxyapatite. In addition, (AspSerSer)6 also showed favorable binding to osteoblast-mediated mineralizing nodules and amorphous calcium phosphate in vitro, implying its potential as a selectively targeting moiety for bone-formation surfaces.

Here we confirm that (AspSerSer)6 is a targeting moiety in vivo for bone-formation surfaces. Then, we linked (AspSerSer)6 with a DOTAP-based cationic liposome (approved by the US Food and Drug Administration for clinical trials, NCT00059605) that encapsulates an osteogenic siRNA that targets a recently discovered negative regulator (Plekho1) of osteogenic lineage activity without modulating bone resorption12,13. We examined (AspSerSer)6-liposome with the Plekho1 siRNA in vitro for its physical chemistry and biological characterization. We also performed a series of in vivo studies to examine the biological activities of (AspSerSer)6-liposome plus Plekho1 siRNA for cell-selective delivery, gene knockdown and bone anabolic action in both healthy and osteoporotic rats.

Results

(AspSerSer)6 as a targeting moiety

We compared the differences in the presence of FITC at various bone-formation or bone-resorption surfaces between adult rats injected with FITC-labeled (AspSerSer)6 and those injected with FITC-labeled control peptides (Asp8) after pre-injection of xylenol orange (a red fluorescent calcium-binding dye capable of labeling new bone deposition at bone-formation surfaces)14. We found that bone-formation surfaces (labeled with xylenol orange) were largely co-labeled with (AspSerSer)6 (labeled with FITC) in the rats injected with FITC-labeled (AspSerSer)6, whereas we observed very little co-labeling in the rats injected with the FITC-labeled Asp8 peptides (Fig. 1a). Likewise, there was very little staining of the bone-resorption surfaces that had been labeled by injected FITC-labeled (AspSerSer)6, whereas FITC-labeled Asp8 peptide did show staining at the bone-resorption surfaces (Fig. 1a). Similarly, co-injection of both rhodamine-labeled Asp8 and FITC-labeled (AspSerSer)6 showed little colocalization of Asp8 and (AspSerSer)6 (Fig. 1b). After ruling out nonspecific staining, we did not find that FITC labeled either bone-formation surfaces or bone-resorption surfaces (Fig. 1a).

Figure 1: Differential occupancy characteristics of (AspSerSer)6 compared to Asp8 at bone-formation or bone-resorption surfaces in nondecalcified bone sections using a confocal laser scanning microscope.
figure 1

(a) Fluorescence micrographs from rats injected with (AspSerSer)6-FITC (top), Asp8-FITC (middle) or unlinked FITC (bottom). Left, the white arrows point to the bone-formation surfaces labeled with xylenol orange (XO) (red), and the white arrowheads point to bone-resorption surfaces (eroded surface). Middle, the white arrows point at the locally accumulated (AspSerSer)6-FITC (top) or Asp8-FITC (middle) (green), and the white arrowheads point to bone-resorption surfaces. Right, a merged image of the left and middle images. Co-staining of (AspSerSer)6-FITC and xylenol orange was found (top row). Scale bars, 50 μm. (b) Differential distribution of (AspSerSer)6 from Asp8 in undecalcified bone sections after co-injection. Fluorescence micrographs from rats co-injected with (AspSerSer)6-FITC and Asp8-rhodamine (top). Left, the white arrows point to (AspSerSer)6-FITC (green) binding sites. Middle, the white arrowheads point to Asp8-rhodamine (red). Right, a merged image of the left and middle images. Fluorescence micrographs from rats co-injected with unconjugated FITC and rhodamine (bottom). There was no locally accumulated FITC (green, left) or rhodamine (red, middle) seen. Right, a merged image of the left and middle images. Scale bars, 25 μm.

Full size image

In vitro characterization of the targeted delivery system

We used standard methods in our preparation for linking the (AspSerSer)6 peptide to the DOTAP-based liposomes that encapsulated the Plekho1 siRNA (Supplementary Methods and Supplementary Fig. 1a). We confirmed the osteoblast-activity–promoting effect of the identified Plekho1 siRNA sequence (Supplementary Table 1 and Supplementary Fig. 2).

We characterized the physical chemistry of the targeted delivery system in vitro (Supplementary Results and Supplementary Fig. 1b). The in vitro biological characterization showed that (AspSerSer)6-liposome plus Plekho1 siRNA bound more favorably to lowly crystallized hydroxyapatite than to highly crystallized hydroxyapatite (Supplementary Table 2 and Supplementary Fig. 1c). Further, (AspSerSer)6-liposome prevented the Plekho1 siRNA from serum-mediated degradation (Supplementary Fig. 1d) and facilitated the internalization of the linked Plekho1 siRNA in both human osteoblast-like cells (hFOB 1.19 cells) (Supplementary Fig. 1e) and human osteoclast-like cells (giant-cell tumors) (data not shown). Functionally, (AspSerSer)6-liposome facilitated Plekho1 gene knockdown in both the hFOB 1.19 cells and giant-cell tumors (Supplementary Fig. 1f).

Characterization of the targeted delivery system in vivo

We used biophotonic imaging technology to examine the organ distribution of FAM-labeled Plekho1 siRNA delivered by the liposome with or without the (AspSerSer)6 moiety or delivered as free siRNA without any transfection reagent in 6-month-old female healthy Sprague Dawley rats. We used the siRNA delivered by in vivo jetPEI (a commercialized in vivo transfection reagent for nucleic acid) as a positive control. We found that the intensity of the intraosseous fluorescence signal was strongest in the rats injected with (AspSerSer)6-liposome plus Plekho1 siRNA among all the groups (Fig. 2a). However, the intensity of the hepatic fluorescence signal was lower in the rats treated with (AspSerSer)6-liposome plus Plekho1 siRNA than in rats treated with Plekho1 siRNA delivered by either in vivo jetPEI or by the liposome without (AspSerSer)6. The fluorescence signal was barely detectable in the heart, spleen, lungs and kidneys of the rats from all of the treatment groups, except for a small signal that was present in the kidney of the rats injected with free siRNA (Fig. 2a). Further, the quantification data from the fluorescence microplate readers were also consistent with the findings from the biophotonic imaging (Fig. 2b).

Figure 2: Organ-selective delivery and gene knockdown in vivo.
figure 2

(a) Localization of labeled siRNA in rats by a biophotonic-imaging–based analysis after administration of free Plekho1 siRNA, in vivo jetPEI plus Plekho1 siRNA, liposome plus Plekho1 siRNA and (AspSerSer)6-liposome plus Plekho1 siRNA. The intensity of the fluorescence signal was analyzed in isolated hearts, livers, spleens, lungs, kidneys and femurs from the rats. n = 3 per group. (b) Quantitative analysis by a microplate reader system for the fluorescence of the FAM-labeled siRNA in livers, kidneys and bone tissues after in vivo administration in a separate set of rats using the same delivery systems outlined in a. The bone tissues included two sets of femur and tibia samples, as well as vertebra samples. *P < 0.05 for comparison with the liposome plus Plekho1 siRNA group by one-way ANOVA with post hoc test. n = 6 per group. Data are means ± s.d. (c) Representative western blots in various organs for Plekho1 protein after in vivo delivery of Plekho1 siRNA or nonsense siRNA using the indicated methods. β-actin served as an internal control. (d) Knockdown efficiency of Plekho1 mRNA expression by a real-time PCR analysis of various organs after in vivo Plekho1 siRNA delivery by the (AspSerSer)6-liposome or liposome methods. Plekho1 knockdown efficiency was calculated by comparing the Plekho1 mRNA expression value in the Plekho1 siRNA group to the knockdown in the nonsense siRNA group. The Plekho1 mRNA expression value was normalized to Gapdh. *P < 0.05 for the (AspSerSer)6-liposome plus Plekho1 siRNA group compared to the liposome plus Plekho1 siRNA group. n = 6 per group. Data are means ± s.d.

Full size image

We also examined Plekho1 protein and mRNA expression by western blot and real-time PCR analysis, respectively, in bone and nonskeletal organs after administration of Plekho1 siRNA delivered by the liposome with or without the (AspSerSer)6 moiety. In 6-month-old female healthy Sprague Dawley rats, we found that the efficiency of the Plekho1 gene knockdown in bone was significantly higher after treatment with (AspSerSer)6-liposome plus Plekho1 siRNA as compared to the knockdown achieved by treatment with the liposome (without (AspSerSer)6) plus Plekho1 siRNA (P < 0.05) (Fig. 2c,d). In contrast, the Plekho1 gene knockdown efficiency in nonskeletal organs (for example, the liver and kidney) was significantly lower after treatment with (AspSerSer)6-liposome plus Plekho1 siRNA compared to the knockdown efficiency seen after treatment with liposome plus Plekho1 siRNA (P < 0.05) (Fig. 2c,d).

In those rats treated with (AspSerSer)6-liposome plus Plekho1 siRNA, liposome plus Plekho1 siRNA or Plekho1 siRNA only, we then immunostained cryosections of the rat proximal tibia and distal femur using markers of osteogenic cells at various differentiation stages, including alkaline phosphatase (Alp), runt-related transcription factor 2 (Runx2), osteocalcin and type I collagen α1 (Col1A1)15,16. We found numerous instances of colocalization of the labeled siRNA with Alp-positive (Fig. 3a), Runx2-positive (Supplementary Fig. 3a), osteocalcin-positive (Supplementary Fig. 3b) and Col1A1-positive (Supplementary Fig. 3c) cells when we administered (AspSerSer)6-liposome plus Plekho1 siRNA to the rats, whereas there were few instances of such overlapping staining when we administered liposome plus Plekho1 siRNA. In addition, the immunohistochemistry for osteoclast-associated receptor (Oscar), a marker specifically expressed in the cell surfaces of pre-osteoclasts and mature osteoclasts17,18, showed an absence of (AspSerSer)6-liposome plus Plekho1 siRNA particles in Oscar-positive cells. However, several siRNAs were present in Oscar-positive cells when we administered liposome plus Plekho1 siRNA to rats (Supplementary Fig. 3d).

Figure 3: Cell-selective delivery and knockdown efficiency in vivo.
figure 3

(a) Fluorescence micrographs of cryosections from proximal tibia after injection with free Plekho1 siRNA (top), liposome plus Plekho1 siRNA (middle) or (AspSerSer)6-liposome plus Plekho1 siRNA (bottom) 4 h before the rats were killed. The Plekho1 siRNA was labeled with FAM (green, left). Immunofluorescence staining was performed to detect Alp-positive osteoblasts (red, middle left). Merged images with DAPI staining showed co-staining of Plekho1 siRNA and Alp-positive osteoblasts (arrows, yellow, middle right). H&E staining of the same sections is shown (right), and red arrowheads point to bone-formation surfaces, enriched by those cells with pink, which is a merged color of red (osteoblast marker) and blue (DAPI staining for nuclei) in the immunofluorescence staining. Scale bars, 20 μm. (b) Plekho1 knockdown efficiency in Alp-positive cells sorted by FACS after in vivo delivery of (AspSerSer)6-liposome plus Plekho1 siRNA, liposome plus Plekho1 siRNA or nonsense siRNA linked to liposomes or (AspSerSer)6-liposomes. Plekho1 mRNA expression was detected by real-time PCR and normalized by glyceraldehyde 3-phosphate dehydrogenase (Gapdh). The Plekho1 knockdown efficiency was calculated by comparing the Plekho1 mRNA expression in the Plekho1 siRNA group to that in the appropriate nonsense siRNA group. Data are means ± s.d. n = 6 per group. *P < 0.05 (One-way ANOVA with post hoc test). NS, no significant difference.

Full size image

Further, we examined Plekho1 mRNA expression in rat bone marrow cells sorted by fluorescence activated cell sorting (FACS) using antibodies to either Alp or Stro-1 (a surface marker on osteoprogenitor cells and pre-osteoblasts)16. We found that the Plekho1 knockdown efficiency in the cells positive for the antibody to Alp was significantly higher than that in either the cells negative for the antibody to Alp (P < 0.05) (Fig. 3b) or cells positive for an antibody to Oscar (P < 0.05) (Supplementary Fig. 3e). The Plekho1 knockdown efficiency in the cells positive for the antibody to Stro-1 was significantly higher than the knockdown in the cells negative for that antibody at both 12 and 48 h after administration of (AspSerSer)6-liposome plus Plekho1 siRNA (P < 0.05) (Supplementary Fig. 3f). In contrast, we found no significant difference in Plekho1 mRNA expression knockdown efficiency between the cells positive for antibodies to Alp or Stro-1 and the cells negative for these antibodies after administration of liposome plus Plekho1 siRNA (P > 0.05) (Fig. 3b and Supplementary Fig. 3e,f).

RNAi-mediated bone anabolic action in healthy rats

We used in vivo micro computed tomography (microCT) to examine the bone mineral density (BMD) and three-dimensional architecture parameters in trabecular bone of the proximal tibia after the administration of the Plekho1 siRNA delivered by the liposome with and without the (AspSerSer)6 moiety through tail vein injection in 6-month-old female healthy Sprague-Dawley rats. The statistic analysis by repeated measures analysis of variance (ANOVA) for the in vivo microCT data showed that both the 'time effect' and 'time-by-group interaction effect' were statistically significant for all the variables we examined with a statistical significance level at 0.05 (Supplementary Table 3). Thus, the data indicate that there was a change over time in the values for all the examined variables and that there were also different change patterns over time among the examined groups of rats. The liposome plus Plekho1 siRNA group showed a significant increase in BMD, relative bone volume (bone volume/tissue volume), trabecular thickness, trabecular number and connectivity density (10.01%, 31.91%, 18.82%, 10.04% and 10.56%, respectively) at week 9 after treatment compared to baseline (P < 0.05), as well as a significant decrease in structure model index (SMI) (35.07%) at week 9 after treatment compared to baseline (P < 0.05). Further, the (AspSerSer)6-liposome plus Plekho1 siRNA group showed a significant increase from baseline in their BMD, relative bone volume, trabecular thickness, trabecular number and connectivity density (23.12%, 66.87%, 36.37%, 21.65% and 19.18%, respectively) at week 9 after treatment compared to baseline (P < 0.05), as well as a significantly larger increase from baseline in these parameters than the group treated with liposome plus Plekho1 siRNA (13.23%, 27.82%, 15.85%, 12.17% and 12.23%, respectively) at week 9 after treatment (P < 0.05). The (AspSerSer)6-liposome plus Plekho1 siRNA group also showed a significant decrease in trabecular space and SMI (11.61% and 50.00%, respectively) at week 9 compared to baseline (P < 0.05), as well as a significant decrease in these two parameters (8.97% and 20.44%, respectively) at week 9 after treatment compared to the liposome plus Plekho1 siRNA group (P < 0.05). However, we found no significant time-course changes from baseline within the 9 weeks of study in all the in vivo microCT variables we examined in either the free siRNA group or the age-matched control group (Fig. 4a). Consistent with the microCT quantification data, better organized three-dimensional architecture and a higher bone mass in trabecular bone from the in vivo microCT reconstruction images in rats treated with (AspSerSer)6-liposome plus Plekho1 siRNA compared to rats treated with liposome plus Plekho1 siRNA, Plekho1 siRNA alone or PBS (age-matched control) at weeks 6 and 9 after starting the treatment (Fig. 4b).

Figure 4: In vivo microCT examinations of the three-dimensional trabecular architecture and an ex vivo bone formation evaluation in nondecalcified bone sections in healthy rats.
figure 4

(a) Plots of the structural parameters (BMD, relative bone volume (BV/TV), trabecular thickness (Tb.Th), trabecular number (Tb.N), connectivity density (Conn.D), structure model index (SMI) and trabecular space (Tb.Sp)) from in vivo microCT examination, monitored over time for the four groups of rats examined (rats treated with liposome plus Plekho1 siRNA, (AspSerSer)6-liposome plus Plekho1 siRNA, free siRNA or PBS). *P < 0.05 compared to the liposome Plekho1 siRNA group at week 9 after treatment. (b) Representative three-dimensional trabecular architecture at the proximal tibia from the respective groups (rats treated with liposome plus Plekho1 siRNA, (AspSerSer)6-liposome plus Plekho1 siRNA, free siRNA or PBS) obtained by in vivo microCT examination at baseline and at weeks 3, 6 and 9 after treatment. (c) Bone formation was examined by sequential labels with fluorescent dye in nondecalcified bone sections from healthy rats. Representative fluorescent micrographs of the trabecular bone sections showed the xylenol (red) and calcein (green) labels in the baseline group and the groups treated with liposome plus Plekho1 siRNA, (AspSerSer)6-liposome plus Plekho1 siRNA, free siRNA or PBS. Arrows indicate the space between the xylenol and calcein labeling. Scale bars, 10 μm.

Full size image

In addition, the bone histomorphometry analysis showed that the mineral apposition rate, bone formation rate, mineralizing surface area, osteoblast surface area and osteoblast number in the group treated with liposome plus Plekho1 siRNA were all significantly lower than the same parameters in the group treated with (AspSerSer)6-liposome plus Plekho1 siRNA, but the measures of these parameters in these two groups were remarkably higher than those in the groups treated with free siRNA or PBS or the baseline group; we found no difference in osteoclast surface and osteoclast number among all the groups (Supplementary Table 4). We found extensive xylenol and calcein labeling and a larger width between the two labeling bands in the rats treated with (AspSerSer)6-liposome plus Plekho1 siRNA compared to the rats treated with liposome plus Plekho1 siRNA, free Plekho1 siRNA or PBS and compared to the baseline group (Fig. 4c). The quantification of the distance between the xylenol and calcein labeling was also reflected in the bone-formation–related parameters (that is, the mineral apposition rate, bone formation rate, mineralizing surface and osteoblast surface).

RNAi-mediated bone anabolic action in osteoporotic rats

We also used in vivo microCT to examine the BMD and three-dimensional architecture parameters in trabecular bone of the proximal tibia after the administration of the Plekho1 siRNA delivered by the liposome with and without the (AspSerSer)6 moiety in 6-month-old female Sprague-Dawley rats with established osteoporosis induced by ovariectomy (OVX). We initiated the siRNA treatment by tail vein injection at 4 weeks after OVX. The statistical analysis by repeated measures ANOVA for the in vivo microCT data (Supplementary Results) showed that both the time effect and time-by-group interaction effect were statistically significant with a statistical significance level at 0.05 for all the variables we examined (BMD, relative bone volume, trabecular thickness, trabecular number, trabecular space, connectivity density and SMI); the data also indicated a change over time in the examined variables and different change patterns over time between the examined groups after the administration of treatment (Fig. 5a). Briefly, all the above in vivo microCT parameters in the group treated with OVX and (AspSerSer)6-liposome plus Plekho1 siRNA were almost restored to the pre-surgery values after 9-week treatment (at week 13 after surgery), whereas we did not observe such restoration observed in the group treated with OVX and liposome plus Plekho1 siRNA within the 9-week siRNA treatment period. Consistently, we found better organized microarchitecture and a higher bone mass in trabecular bone in rats treated with OVX and (AspSerSer)6-liposome plus Plekho1 siRNA compared to rats treated with OVX and liposome plus Plekho1 siRNA, free siRNA or PBS (age-matched control) after 6-week treatment and 9-week treatment. (at week 10 and 13 after surgery) (Fig. 5b).

Figure 5: In vivo microCT examination of the three-dimensional trabecular architecture in OVX-treated rats.
figure 5

(a) Plots of the structural parameters (BMD, relative bone volume (BV/TV), trabecular thickness (Tb.Th), trabecular number (Tb.N), connectivity density (Conn.D), structure model index (SMI) and trabecular space (Tb.Sp)) from an in vivo microCT examination, monitored over time for the rats from the sham-treated group and the groups treated with OVX and PBS, OVX and free siRNA, OVX and liposome plus Plekho1 siRNA and OVX and (AspSerSer)6-liposome plus Plekho1 siRNA. The repeated measures ANOVA analysis for the in vivo microCT data showed that both the time effect and the time-by-group interaction effect were statistically significant for all the examined parameters. *P < 0.05 compared to the liposome plus Plekho1 siRNA group at week 13 after treatment. The diamond indicates the initiation of treatment. (b) Representative three-dimensional trabecular architecture at the proximal tibia from the respective groups (sham-treated rats and rats treated with OVX and PBS, OVX and free siRNA, OVX and liposome plus Plekho1 siRNA, and OVX and (AspSerSer)6-liposome plus Plekho1 siRNA) obtained by in vivo microCT examination at weeks 0, 4, 7, 10 and 13 after OVX.

Full size image

Discussion

Currently, there is no available bone-specific targeting delivery system used for siRNA delivery in bone metabolic disorders. Here we designed a siRNA delivery system to specifically target bone-formation surfaces and, thus, facilitate the delivery of therapeutic cargos to the osteogenic-lineage cells. This delivery system could establish the foundation for translating RNAi-based therapies from basic science to clinic applications in the musculoskeletal field.

In our in vitro studies described here, we find that (AspSerSer)6 has a higher binding affinity to lowly crystallized hydroxyapatite (similar to what is found at the bone-formation surface) than to highly crystallized hydroxyapatite (similar to what is found at the bone-resorption surface)19, implying that this moiety has the potential to selectively bind to bone-formation surfaces rather than bone-resorption surfaces. Furthermore, in our in vivo studies that included dynamic bone histomorphometry, we found that (AspSerSer)6 favorably binds to bone-formation surfaces rather than bone-resorption surfaces, the mechanism for which is related to the chemical biology that is involved in the interaction between (AspSerSer)6 and crystallized hydroxyapatite (Supplementary Discussion)4,9,14,20,21,22,23,24.

The data from both the biophotonic imaging and microplate reader systems consistently suggested that (AspSerSer)6-liposome could facilitate the delivery of the linked siRNA to bone (with an approximately tenfold more siRNA delivered to the bone in rats treated with (AspSerSer)6-liposome plus Plekho1 siRNA than that seen in rats treated with liposome plus Plekho1 siRNA) and reduce its delivery to nonskeletal organs. To date, there have been no reports of bone-targeting delivery systems for RNAi-based therapy. Furthermore, real-time PCR and western blot analyses in our study consistently suggested that (AspSerSer)6-liposome could facilitate RNAi-based gene knockdown in a bone-selective manner.

The immunohistochemistry data indicated that (AspSerSer)6-liposome could facilitate the delivery of siRNA to osteogenic cells at various differentiation stages. Thus, we did flow cytometry and found that (AspSerSer)6-liposome plus Plekho1 siRNAs, unlike liposome plus Plekho1 siRNAs, facilitate RNAi-based gene knockdown in osteogenic cells at various stages of differentiation but, notably, not in osteoclastic cells. We observed that this exclusion of the knockdown of Plekho1 that was mediated by (AspSerSer)6-liposome plus Plekho1 siRNA in osteoclasts in vivo was not present in vitro, where we found a similar knockdown efficiency facilitated by the (AspSerSer)6-liposome plus Plekho1 siRNA in both human osteoblast-like cells and human osteoclast-like cells. This inconsistency could be explained by the lack of a skeletal domain in the cell culture we used and the existence of this domain in the in vivo animal studies. The mechanism of the cell-specific delivery for siRNA is postulated to involve the increased interaction between Plekho1 siRNA and the osteogenic cells, as facilitated by (AspSerSer)6 (Supplementary Discussion and Supplementary Fig. 4).

The flexible extravasation of the (AspSerSer)6-liposome plus Plekho1 siRNA into bone fluid and the increased cellular internalization of the linked siRNA induced by the (AspSerSer)6-liposome platform are also reflected in the present study (Supplementary Discussion)25,26,27.

The bone histomorphometry data from the healthy rats indicated that the targeting moiety (AspSerSer)6 was able to facilitate the functional longevity and activity of osteoblasts after it facilitated Plekho1 gene silencing that was mediated by liposome plus Plekho1 siRNA. Further, the in vivo microCT data from the healthy rats indicated that the targeting moiety (AspSerSer)6 could significantly facilitate increased bone mass and an improved trabecular architecture. It is known that iPTH stimulates both bone-forming and bone-resorbing cells, which complicates its clinical use2,28. In contrast, bone resorption was not activated by (AspSerSer)6-liposome plus Plekho1 siRNA. Notably, our in vivo microCT data from the osteoporotic rats indicated that the targeting moiety (AspSerSer)6 could also significantly facilitate a liposome plus Plekho1 siRNA–mediated increase in bone mass and an improvement in the trabecular architecture to the pre-surgery values at week 9 after treatment, whereas treatment with liposome plus Plekho1 siRNA did not restore these values to their pre-surgery levels. Taken together, these data suggest that the targeting moiety (AspSerSer)6 can facilitate improvements in bone anabolic action mediated by liposome plus Plekho1 siRNA in both healthy rats and osteoporotic rats.

In summary, (AspSerSer)6-liposome is a promising targeting system for specifically delivering siRNA drugs to bone-formation surfaces and the osteogenic cells that reside there, thus providing a potential solution to the bottleneck in clinical translation of RNAi-based bone anabolic therapies.

Methods

Study profile.

We prepared the (AspSerSer)6-liposome plus Plekho1 siRNA after examination of the nature of (AspSerSer)6 and Plekho1 siRNA (pre-study work 3). After those examinations, eight specific studies were performed, including 'study 1' for physical chemistry characterization in vitro, 'study 2' for binding affinity to hydroxyapatite and resistance against serum-mediated degradation in vitro, 'study 3' for cellular internalization and knockdown efficiency in vitro, 'study 4' for organ-selective delivery in vivo, 'study 5' for organ-specific gene knockdown in vivo, 'study 6' for tissue- or cell-selective delivery in vivo, 'study 7' for cell-selective gene knockdown in vivo and 'study 8' for bone anabolic action in vivo. Details are provided in the Supplementary Methods. The procedures described in the above studies were approved by Animal Experimentation Ethics Committee of The Chinese University of Hong Kong (ref. no. 09/072/MIS).

Additional methods.

Detailed methodology is described in the Supplementary Methods.