Abstract
Cartilage-targeted gene therapy is promising for osteoarthritis (OA)
treatment, though its potency critically depends on the effectiveness
of delivery vectors. Here, we modularly develop a series of
non-pathogenic, virus-inspired lipopeptide-based nanoparticles (VPN)
tailored to deliver nucleic acids to cartilage. The cationic moiety of
lipopeptide with variable arginine and histidine residues is the key
functional component, and screened by in vitro performance. The
optimized VPN-2 with a moiety of –[(R)[5]-(H)[4]][2]- facilitates
sufficient endocytosis and effective lysosomal escape, achieving about
2.5-fold improvement in transfection potency over conventional lipid
nanoparticles. To address the tradeoff between penetration and
retention within articular cartilage, si-VPN-2 is further formulated
into ROS-responsive nano-in-gel system, which turns out to alleviate
cartilage degeneration in surgical ACTL mice, and further synergizes
with methylprednisolone to implement superior joint protection in PTOA
mice. Our study underscores the platform’s potential of VPN as
cartilage-targeted RNA delivery vector for innovative OA therapy.
Subject terms: Drug delivery, Nanoparticles, Nanobiotechnology,
Nucleic-acid therapeutics
__________________________________________________________________
Cartilage-targeted gene therapy is dependent on delivery efficiency.
Here, the authors develop non-pathogenic, virus-inspired
lipopeptide-based nanoplatforms to deliver nucleic acids to cartilage
with increased transfection efficiency over conventional lipid
nanoparticles.
Introduction
Contemporary pharmaceutical managements of OA is mainly palliative,
which can’t decelerate cartilage degeneration but only achieve
temporary relief from inflammation or pain^[52]1,[53]2. Articular
cartilage loss is the hallmark of OA matrix metalloproteinase 13
(MMP-13) is deemed as the most pathogenic proteolytic driver to OA
progression, due to its potent ability to degrade critical structural
components of cartilage, especially type II collagen
(Col2)^[54]3–[55]5. However, MMP-13 small molecule inhibitors are
suspended by low selectivity-related safety concerns in clinical
translation^[56]4,[57]6. Alternatively, siRNAs-based gene therapy may
bring new breakthroughs by selectively silencing MMP-13 via
sequence-specific manner^[58]3,[59]7. Nevertheless, nucleic acids
suffer intrinsic instability to reach target site for function, their
therapeutic potentials are largely dependent on delivery vectors. Viral
vectors (e.g., adeno- and retroviruses) can effectively transport gene
cargos into host cells, whereas they are restricted to inherent
immunogenicity and pathogenicity^[60]8,[61]9. Instead, non-viral
delivery vehicles, represented by ionizable lipids formulated lipid
nanoparticles (LNP), have been developed as potentially safer and
easy-manufacturing alternatives^[62]10. Despite LNP technology enabling
the clinical translation of siRNA therapeutics for liver diseases and
mRNA for vaccines, its application within extrahepatic tissues is
hindered by limited transfection competency.
Given the respective strengths and limitations possessed by viral and
non-viral approaches, a series of virus-resembling, non-pathogenic
carriers have been rationally designed, aiming to perform well in both
therapeutic efficacy and safety^[63]11–[64]13. For example, genetically
engineered viral proteins are incorporated into LNP to improve
transfection efficiency by enhancing membrane fusion activity^[65]14.
Besides, virus-derived peptides are also excellent alternatives to
bolster genetic cargos transport. Several virally-derived
cell-penetrating peptides (CPPs) with arginine-rich residues, such as
TAT (originated from HIV)^[66]15–[67]17 and VP22 (originated from
herpes simplex virus 1), are utilized as vectors to condense genetic
materials for intracellular delivery. Moreover, owing to enhanced
endosomolytic capacity due to the histidine imidazole group (pKa 6.0)
protonation, polyhistidine-fused CPPs (e.g, TAT-10H) could further
boost gene transfection relative to counterparts^[68]18,[69]19.
Inspired by the powerful transfection potency of virally-derived
arginine-rich CPPs and their polyhistidine-fused derivatives, we
developed a series of modular-designed lipopeptides with three
functional moieties: (i) a targeting head, i.e., Col2-targeting peptide
WYRGRL to implement non-covalent cartilage binding^[70]20–[71]22, (ii)
a cationic moiety with variable arginine and histidine residues to
facilitate nucleic acids condensation; and (iii) a hydrophobic moiety
to endow amphiphilic property. These rationally-designed lipopeptides
could self-assemble into well-defined unilamellar vesicle
nanostructures, and their morphologies further transformed into solid
spherical nanoparticles upon nucleic acid (i.e., siRNA and mRNA)
packaging, resembling the structural transition from liposomes into
LNP^[72]23,[73]24. Concretely, the notable aspects of VPN are as
follows: (i) The biological properties of VPN, especially transfection
efficiency, could be flexibly modulated by tunning the functional
moieties of lipopeptide. (ii) VPN possessed pH-sensitive ability to
strengthen penetration into anionic cartilage tissue under acidic
condition of OA^[74]25,[75]26. (iii) Compared to commercial ionizable
lipid DLin-MC3-DMA formulated LNP (MC3-LNP), the optimized VPN-2 could
facilitate superior endocytosis, intensive lysosomal escape, and higher
transfection efficacy of both siRNA and mRNA. (iv) To fit physiological
application within articular cartilage, VPN could be feasibly entrapped
into a ROS-responsive hyaluronic acid hydrogels (VPN@HA) to achieve the
tradeoff between retention and penetration^[76]27,[77]28. The
as-prepared si-VPN@HA could effectively alleviate cartilage
degeneration, reduce inflammation and relieve from pain in surgical
ACTL mice, moreover, it could further in synergy with
methylprednisolone (MP), the clinical standard steroid treatment, to
implement superior joint protection in PTOA mice. To conclude, our
study is an innovative endeavor to engineer virus-inspired,
non-pathogenic nanoplatforms to delivery nucleic acids to cartilage for
OA therapy.
Results
Rational design and fabrication of VPNs and si-VPNs
The lipopeptides were developed by multi-modular approach with three
functional moieties: (i) a Col2-targeting head, (ii) a cationic moiety
with variable arginine and histidine residues, and (iii) a hydrophobic
moiety consisting of docosanoic acid (C[22]) (Fig. [78]1A and
Supplementary Fig. [79]1). The versatility of the platform allowed
independent investigation of each moiety. Lipopeptides 1–3 were
designed to consist –(R)[5]-(H)[4]-, –[(R)[5]-(H)[4]][2]- or
-(R)[10]-(H)[8]- with a targeting head and two C[22] tails. By
contrast, lipopeptides 4–6 didn’t have targeting head, lipopeptides 7–9
had only one C[22] tail, and peptide 10–12 were set without C[22] tail
(Supplementary Table [80]1). The molecular weights of synthetic peptide
sequences and representative lipopeptide 2 were confirmed by
matrix-assisted laser desorption ionization-time of flight (MALDI-TOF)
mass spectrometry (Supplementary Figs. [81]2–[82]3). Lipopeptide 1–9
could spontaneously self-assemble into uniform nanoparticles to yield
VPN. VPN-1 and VPN-2 (derived from lipopeptide 1 and 2) shared
comparable hydrodynamic diameter (~50 nm) while the size of VPN-3
(derived from lipopeptide 3) was slightly larger (~90 nm), and all of
them carried positive surface charges (Supplementary Fig. [83]4A, B).
Fig. 1. Fabrication and characterization of VPNs and si-VPNs.
[84]Fig. 1
[85]Open in a new tab
A Schematic diagram of the synthetic multi-modular lipopeptides. B The
zeta potential of VPN-2 under pH 7.4, pH 6.8 or pH 5.5 at different
time points, data points represent mean ± SD (n = 3 independent
samples). C Representative photographs of VPN-2 under pH 7.4, pH 6.8 or
pH 5.5 at 2 h. D The siRNA loading efficacy of si-VPN-1, si-VPN-2 or
si-VPN-3 at various N:P ratios by agarose gel electrophoresis. E
HAADF-STEM images of si-VPN-2 and the corresponding element mapping
images of C, N, O, and P, scale bar = 50 μm. F Stability of si-VPN-1,
si-VPN-2 or si-VPN-3 against RNase cocktail, employing free siRNA as
comparison. G Size distribution of si-VPN-1, si-VPN-2, or si-VPN-3. H
Representative TEM images of VPN-1, VPN-2, VPN-3, and corresponding
si-VPN-1, si-VPN-2, si-VPN-3, scale bar = 50 μm. I Representative high
resolution (HR)-TEM (upper) and FE-SEM (bottom) images of VPN-2, scale
bar = 50 μm. J DPD simulation snapshots depicting the front (upper) and
cross-section view (bottom) of si-VPN-2 to illustrate the bead
distribution and overall architectures. scale bar = 10 nm. K Snapshots
depicting the assembly dynamic of si-VPN-2 from a homogeneous mixture
in long-term 200 ns dynamic simulation trajectory. Data in (D, E, F, H
and I) are representative of three independent experiments with similar
results. Source data are provided as a [86]Source data file.
The zeta potentials and sizes of VPNs progressively increased under
pH6.8 and pH5.5 relative to pH7.4, suggesting their pH-sensitive
abilities (Fig. [87]1B and Supplementary Fig. [88]4C–H)^[89]24,[90]29.
The external appearances of VPNs solutions became more turbid when pH
went down, yet with no precipitation (Fig. [91]1C and Supplementary
Fig. [92]5). siRNA-loaded VPNs (si-VPNs) was prepared by microfluidic
based on electrostatic adsorption (Supplementary Fig. [93]6). The
condensation efficacy was assessed by agarose gel electrophoresis, the
N:P ratios were set as 6:1 for si-VPN-2 and si-VPN-3, while that was
quite higher for si-VPN-1 (Fig. [94]1D). Due to phosphorus (P) only
being presented in siRNA, the distribution of siRNA was further
visualized by HAADF-STEM. As demonstrated, siRNA was uniformly
dispersed within si-VPN-2 with 0.73% atomic fraction of P, consistent
with the given N:P ratio (Fig. [95]1E). RNase protection assay
indicated that si-VPNs could effectively prevent siRNA being degraded
by RNase cocktail, RNase A and RNase I, suggesting the excellent
ability of VPNs to protect nucleotide cargos (Fig. [96]1F and
Supplementary Fig. [97]7A, B). After condensing siRNA, si-VPNs showed
slight elevations in diameter, while notable decreases in surface
charge (Fig. [98]1G and Supplementary Fig. [99]8A, B). Stability
testing implied that si-VPNs could maintain reliable storage stability
as well as intact encapsulation of siRNA for a period of at least 48 h
(Supplementary Fig. [100]9A–D). Then, the morphologies of VPNs and
si-VPNs were compared. As shown, VPNs showed well-defined unilamellar
vesicle structures, distinguishing with most reported amphiphilic
peptide nanoparticles, which self-assembled into micelles
(Fig. [101]1H–I). The formations of vesicle architectures were
speculated mainly depended on the hydrophobic moiety of
VPN^[102]30,[103]31, since lipopeptides 4–6 (without WYRGRL head) still
formed vesicles, while lipopeptides 7–9 (with one C[22] tail)
self-assembled into micelles rather than vesicles, peptide 10–12
(without C[22] tail) can’t form nanoparticles at all (Supplementary
Fig. [104]10A–D). Upon packaging siRNA, si-VPNs transformed into solid
spherical nanoparticles, resembling the structural change from
liposomes into LNP, which might be resulted from structural
rearrangement due to charge interaction.
Furthermore, dissipative particle dynamics (DPD) simulations were
designed to determine the morphological characteristics and assembly
dynamics of si-VPNs, where siRNA and lipopeptides were coarse-grained
using DPD approach (Supplementary
Figs. [105]11–[106]12)^[107]32–[108]34. As shown, all three si-VPNs
displayed solid spherical architectures with similar spatial
arrangements, with lipopeptides-siRNA complex concentrated in the
assembly core, surrounded by a shell composed of peptide backbones and
hydrophilic groups on outermost surface (Fig. [109]1J and Supplementary
Fig. [110]13A, B). Assembly dynamics further indicated aggregations of
siRNA and lipopeptide 2 progressively formed from a homogeneous
mixture, with significant structural organization occurring within the
first 1000 ps, and integrity was maintained throughout the 200 ns
simulation, thus validating the stability of the molecular interaction
within this system (Fig. [111]1K).
In vitro biological assessment and gene silencing efficacy of si-VPNs
The cytotoxicity of VPN was noteworthy owing to positive surface
charge. The IC[50] of VPNs uniformly decreased in pH-dependent manner
under acid conditions (Supplementary Fig. [112]14). By contrast, the
cytotoxicity of si-VPN was significantly weakened. As shown, the IC[50]
of si-VPNs were about 10-fold or greater higher than the corresponding
VPNs, for example, the IC[50] of VPN-2 was 194 μg/mL under pH7.4 while
that of si-VPN-2 was 4701 μg/mL (Supplementary Fig. [113]15).
Consistently, si-VPNs only resulted in far faint hemolysis compared to
VPNs, their hemolysis fractions were no more than 5% below 300 μg/mL
under pH7.4 (Supplementary Fig. [114]16A, B). Collectively, these data
suggested great cytocompatibility of si-VPNs.
In vitro cellular uptake assay was conducted using insulin-induced
ATDC5 cells^[115]35,[116]36, which underwent a progressive chondrogenic
differentiation process to achieve ~10-fold increase of Col2α1
expression (Supplementary Fig. [117]17A). As shown, all three si-VPNs
showed higher cellular uptake than siRNA loaded MC3-LNP (si-MC3-LNP) in
either short (i.e., 1 h and 2 h) or long (i.e., 4 h and 8 h) time
course, and more importantly, si-VPN-2 always possessed the highest
mean fluorescence intensity (MFI) (Fig. [118]2A and Supplementary
Fig. [119]17B). The function of WYRGRL head was investigated, as
indicated, WYRGRL-modified si-VPNs showed significantly elevated MFI
compared to non-modified counterparts (Fig. [120]2B and Supplementary
Fig. [121]17C). Besides, surface plasmon resonance (SPR) suggested high
binding affinity between WYRGRL-consisted peptide and Col2 with a
dissociation constant (KD) of 2.05 × 10^−8 M^[122]22, which was in
sharp contrast to non-WYRGRL-modified peptide (Fig. [123]2C and
Supplementary Fig. [124]18). Moreover, the MFI of WYRGRL-modified
si-VPNs in insulin-induced ATDC5 cells were greater than in non-induced
cells, and even greater than in Col2α1 knockdown cells (Supplementary
Fig. [125]19A-B). Collectively, these results validated Col2-dependent
cartilage-targeted ability of WYRGRL-modified si-VPNs. Additionally,
the cellular uptake of si-VPNs displayed ~2-fold increase under pH6.8
than pH7.4, presumably attributable to their increased surface charge
resulting from enhanced polyhistidine ionization under acid conditions
(Fig. [126]2D and Supplementary Fig. [127]20). Cellular uptake
mechanism study revealed that the internalization of si-VPN-2 was
significantly suppressed by 4 °C, amiloride, and chlorpromazine,
suggesting its internalization was energy-dependent, principally
mediated by macropinocytosis and clathrin-dependent endocytosis
(Fig. [128]2E).
Fig. 2. In vitro biological properties and RNAi efficacy of si-VPNs.
[129]Fig. 2
[130]Open in a new tab
A Quantitative cellular uptake of different formulations by flow
cytometry at 1 h or 2 h in insulin-induced ATDC5 cells, data points
represent mean ± SD (n = 6 independent samples). B Confocal images of
cellular uptake of WYRGRL-modified si-VPNs versus non-modified
counterparts, green fluorescence indicated Cy5-labeled formulations,
red fluorescence indicated F-actin, blue fluorescence indicated the
nuclei stained with DAPI, scale bar = 40 μm. C Characterization of
affinity between WYRGRL-modified peptide and Col2 by surface plasmon
resonance (SPR). D Representative flow cytometric analysis of
Cy5-labeled si-VPN-1, si-VPN-2, or si-VPN-3 under pH 7.4 or 6.8
condition. E Cellular uptake mechanism study of si-VPN-2 under pH 7.4,
data points represent mean ± SD (n = 6 independent samples), control
group indicated cell uptake of si-VPN-2 without pre-treatment. F
Penetration of Cy5-labeled formulations into multicellular chondrocyte
spheroids under pH7.4 or pH6.8 after 12 h incubation, scale
bar = 200 μm. G Representative co-localization images of Cy5-labeled
siRNA (red fluorescence) with lyso-Green (green fluorescence) at 2 h,
6 h or 12 h in si-MC3-LNP or si-VPN-2 group, scale bar = 10 μm. H
RT-qPCR measurement of MMP-13 mRNA expression in TNF-α-stimulated ATDC5
cells after different transfections, data points represent mean ± SD
(n = 5 independent samples), control group indicated ATDC5 cells
stimulated with TNFα, but without transfection. Statistical
significance was determined using unpaired one-way ANOVA (E) and
unpaired two-way ANOVA (A, H): *p < 0.05, **p < 0.01,
***p < 0.001 and ****p < 0.0001, NS means no significance. Data in
(B, F and G) are representative of three independent experiments with
similar results. Source data are provided as a [131]Source data file.
Considering vehicles must diffuse into cartilage tissue before
internalization by chondrocytes, the matrix penetration ability was
further evaluated using 3D chondrocyte spheroids model. As shown,
si-VPN-3 almost localized on outer layer of spheroids, and only faint
fluorescence signals reached the center after 12 h incubation
(Fig. [132]2F and Supplementary Fig. [133]21A, B). By contrast,
si-VPN-2 could penetrate deep into the center with significantly
increased fluorescence, relative to either si-MC3-LNP or si-VPN-1.
Moreover, si-VPN-2 achieved substantially higher fluorescence
intensities under pH6.8 than pH7.4, suggesting its acid-boosted
penetration ability^[134]37,[135]38. Since lysosomal escape was one of
the all-important processes for siRNA delivery, the dynamics of
lysosomal escape were compared between si-MC3-LNP and si-VPN-2. As
shown, si-MC3-LNP was still entrapped in lysosomal vesicles at late 6 h
or 12 h; by contrast, si-VPN-2 was rapidly released into cytoplasm over
time, and its Pearson’s Rr value was only ~0.2 at 12 h (Fig. [136]2G).
The potent lysosomal escape of si-VPN-2 was probably attributed to
proton sponge effect trigged by polyhistidine protonation under acidic
condition^[137]24,[138]39. Finally, MMP-13 silencing assay was
evaluated in chondrogenic ATDC5 cells pre-treated with 25 ng/ml TNFα
for 24 h to mimic pro-inflammatory conditions. The cells were
transfected with different formulations and then prepared for RT-qPCR
assay. As shown, si-VPN-2 demonstrated the most efficient MMP-13
silencing in a dose-dependent manner, with over 85% MMP-13 knockdown at
dose of 100 nM siRNA, much superior than si-MC3-LNP (Fig. [139]2H).
Immunofluorescent staining assay gave consistent results, the optimal
si-VPN-2 showed minimized MMP-13 expression, even comparable to the
without TNFα group (Supplementary Fig. [140]22A, B).
MMP-13 silencing broadly affects gene expression profiles in vitro
In order to verify universal MMP-13 silencing ability of si-VPN-2,
patients-derived primary chondrocytes were further employed
(Fig. [141]3A). Consistent with ATDC5 cells, si-VPN-2 showed the
highest MFI of cellular uptake, with 3.04-fold increase over si-MC3-LNP
group (Fig. [142]3B and Supplementary Fig. [143]23A, B). Moreover,
si-VPN-3 consisting -(R)[10]-(H)[8]- sequence formed apparent
aggregations, by contrast, despite with identical number but distinct
sequence of arginine and histidine residues, si-VPN-2 possessing a
cationic moiety of –[(R)[5]-(H)[4]][2]- did not. And the aggregations
of si-VPN-3 was presumably due to high-degree non-specific protein
adsorption. To validate this hypothesis, we performed protein
adsorption experiments. As shown, si-VPN-3 displayed significantly
higher turbidity and adsorbed protein than si-VPN-2 after exposure to
10% FBS (Supplementary Fig. [144]24A, B). Consistently, SDS-PAGE
indicated fewer bands with lower grayscale intensity of si-VPN-2
relative to si-VPN-3 (Supplementary Fig. [145]24C, D). Additionally,
the superior lysosomal escape of si-VPN-2 was verified, since the
fluorescent colocalization of lyso-tracker red and FAM-labeled siRNA
was diminished after 6 h incubation. And the Pearson’s Rr value of
si-VPN-2 was far below than si-MC3-LNP, just slightly over Free siRNA
(Fig. [146]3C and Supplementary Fig. [147]25A, B). Moreover, si-VPN-2
also exerted the most potent MMP-13 silencing efficacy, as MMP-13 mRNA
and its protein expression were significantly reduced after si-VPN-2
treatment, which were even statistically equivalent to without TNFα
group (i.e., with no TNFα stimulation) (Fig. [148]3D–E, and
Supplementary Fig. [149]26).
Fig. 3. MMP-13 silencing broadly affects gene expression profiles in vitro.
[150]Fig. 3
[151]Open in a new tab
A Schematic diagram to obtain patients-derived primary chondrocytes. B
Cellular uptake of Cy5-labeled different formulations (green
fluorescence) in patients-derived primary chondrocytes, the red
fluorescence indicated F-actin, blue fluorescence indicated the nuclei
stained with DAPI, scale bar = 50 μm. C Representative co-localization
images of FAM-labeled siRNA (green fluorescence) with lyso-tracker red
(red fluorescence) in si-MC3-LNP or si-VPN-2 group after 6 h
incubation, scale bar = 20 μm. D RT-qPCR measurement of MMP-13 mRNA
expression after different transfections, data points represent
mean ± SD (n = 8 independent samples), control group indicated primary
chondrocytes cells stimulated with TNFα, but without transfection. E
The MMP-13 silencing efficiency study of different formulations by
immunofluorescent staining, the red fluorescence indicated MMP-13
expression, the green fluorescence indicated F-actin, scale
bar = 40 μm. F Volcano plot of the DEGs in si-VPN-2 vs TNFα group (only
stimulated with TNFα) (n = 3 independent samples). G Heat map showing
representative DEGs between si-VPN-2 and TNFα group (n = 3 independent
samples). H Schematic diagram of the broad gene expression effects of
si-VPN-2 (Created with BioRender.com). I The KEGG enrichment
bioinformatic analysis of si-VPN-2 vs TNFα group. Statistical
significance was determined using unpaired one-way ANOVA (D):
*p < 0.05, **p < 0.01, ***p < 0.001 and ****p < 0.0001, NS
means no significance. Data in (B, C and E) are representative of three
independent experiments with similar results. Source data are provided
as a [152]Source data file.
To characterize the global effect of si-VPN-2 on gene expression
profiles, RNA-Seq analysis was performed. There were 329 significantly
downregulated genes and 519 significantly upregulated genes in si-VPN-2
group, compared to TNFα group (Fig. [153]3F). Among them, some
representative differentially expressed genes (DEGs) implicated in OA
pathophysiology were particularly concerned. Treatment-associated
downregulated DEGs included chemokines (i.e., CCL2, CCL7, CXCL6 and
CXCL10), immunity related markers (i.e., TLR2), pro-inflammatory
related markers (i.e., IL34, ICAM-1 and NFκB2), negative regulator of
cartilage matrix ADAMTS12, and pro-apoptosis related genes (i.e., TP53)
(Fig. [154]3G). By contrast, upregulated DEGs were anti-inflammatory
markers (i.e., IL11 and IL33) and genes related chondrocyte
differentiation (i.e., GDF5 and BMP6). The changes of above-mentioned
DEGs were also confirmed by RT-qPCR, which were mainly involved in
following biological processes, including chondrocyte differentiation,
skeletal system development, cytokine-cytokine interaction,
inflammation as well as immune response (Fig. [155]3H and Supplementary
Fig. [156]27). Moreover, bioinformatic analysis was performed via Kyoto
Encyclopedia of Genes and Genomes (KEGG) enrichment. There were nine
pathways highly affected by si-VPN-2 treatment, such as TNF,
cytokine-cytokine receptor interaction, P13K-Akt as well as MAPK, which
were reported to highly correlate with OA pathogenesis^[157]40,[158]41
(Fig. [159]3I). Collectively, these results verified that si-VPN-2 can
exert broad gene expression effects via silencing MMP-13.
Construction and release kinetics of nano-in-gel si-VPN@HA
To address the dilemma between retention and penetration faced by
intra-articular delivery (i.e., small vectors facilitate penetration
but suffer rapid clearance, while large vectors act the opposite
way)^[160]26,[161]28, a stimulus-responsive nano-in-gel system
(si-VPN@HA) was constructed by encapsulating proper amount of si-VPN-2
into ROS-responsive hyaluronic acid hydrogels (HA@gels) (Fig. [162]4A).
It presumed that the highly-elevated ROS would trigger dissolution of
HA@gels to achieve sustained-release si-VPN-2 under OA
condition^[163]42. HA@gels were prepared based on non-covalent
host-guest recognition between β-CD (host molecule) and Fc (guest
molecule), which were decorated onto HA backbone respectively
(Supplementary Fig. [164]28A–C)^[165]43,[166]44. Oxidation-induced
detachment of oxidized Fc from β-CD cavity endowed HA@gels with rapid
H[2]O[2]-sensitive gel-sol-transition ability, as it exhibited lessened
cross-linking density and the 3D network became collapsed under 5 mM
H[2]O[2] stimulus (Fig. [167]4B and Supplementary Fig. [168]28D).
Rheological assessment gave consistent results, storage modulus (G’)
were significantly lower than loss modulus (G”) with addition of 5 mM
H[2]O[2], suggesting HA@gels underwent a sharp reduction in
crosslinking density (Fig. [169]4C). Moreover, thixotropic experiment
demonstrated that HA@gels possessed potential self-healing nature
against pressure-induced deformations, as quasi-liquid sol state could
recover into gel state instantly with recycled strains reverted from
400% to 0.2% (Supplementary Fig. [170]28E).
Fig. 4. Construction and intra-articular retention of sustained-release
si-VPN@HA.
[171]Fig. 4
[172]Open in a new tab
A Schematic construction of ROS-responsive nano-in-gel system
si-VPN@HA. B The representative photographs of gel−sol transition of
HA@gel when stimulated with H[2]O[2] (upper row), and the corresponding
representative SEM images (bottom row), scale bar = 20 μm. C Rheology
properties of HA@gel when stimulated with 0.1 or 5 mM H[2]O[2]. D
Fluorescent scanning spectrums of different formulations in 5 mM
H[2]O[2] solutions. E The in vitro release profiles of different
formulations in 5 mM H[2]O[2] solutions (n = 3 independent samples). F
Representative images of ex vivo trypsin-damaged cartilage explants
penetration assay of different Cy5-labeled formulations (n = 3
independent samples), scale bar = 2 mm. G Representative images of
Cy5-labeled formulations (red fluorescence) penetration inside
cartilage section, and the representative 3D reconstructions (bottom
row) of the white dashed box, blue fluorescence indicated the nuclei
stained with DAPI, scale bar = 1 mm. H The fluorescence intensities of
cartilage section in (G), data points represent mean ± SD (n = 5
independent samples). I Representative IVIS images of OA mice knee
joints after intra-articular injection of different formulations over
28 days. J Time course of fluorescent radiant efficiency within joints,
data points represent mean ± SD (n = 3 mice). K Area under curve (AUC)
of different groups in (J) (n = 3 mice). Statistical significance was
determined using unpaired one-way ANOVA (H, K): *p < 0.05,
**p < 0.01, ***p < 0.001 and ****p < 0.0001; NS means no
significance. Data in B are representative of three independent
experiments with similar results. Source data are provided as a
[173]Source data file.
To verify whether loaded siRNA will be dissociated from si-VPN-2 during
release process, in vitro release profiles were assessed. Fluorescence
scanning implied that only free siRNA and free siRNA loaded HA
hydrogels (i.e., siRNA@HA) displayed fluorescence signals in 5 mM
H[2]O[2] solution, while si-VPN-2 and si-VPN@HA did not, indicating
barely dissociative siRNA existed in latter two groups (Fig. [174]4D).
Time-lapse in vitro release kinetics further indicated the
gradual-release of free siRNA from siRNA@HA, by contrast, almost no
fluorescence signals were detected in either si-VPN-2 or si-VPN@HA
group, suggesting siRNA was still wrapped within si-VPN-2 during
ROS-triggered si-VPN@HA dissolution (Fig. [175]4E). And the
morphologies of released si-VPN-2 still maintained mono-dispersed
spherical under 5 mM H[2]O[2] stimulus in either pH7.4 or pH6.8
solution (Supplementary Fig. [176]29A, B). Then cartilage-penetration
ability was evaluated using an ex vivo trypsin-damaged cartilage
explants model (Supplementary Fig. [177]30). It confirmed the optimal
performance of si-VPN-2 among three designed si-VPNs, as it diffused
evenly and achieved a uniform fluorescence profile within cartilage
section after 24 h incubation (Fig. [178]4F–H and Supplementary
Figs. [179]31–[180]32). Despite si-VPN-3 showed higher fluorescence
intensity than si-VPN-2 on cartilage explants, it mainly adsorbed on
external surface and scarcely penetrated into interior cartilage
section. Additionally, si-VPN@HA displayed quite low fluorescence
without H[2]O[2], while its fluorescence signals achieved ~2-fold
increase on cartilage explants and ~3-fold elevation within cartilage
sections under H[2]O[2] stimulus.
Furthermore, in vivo intra-articular retention behaviors were
characterized over 28 days using living imaging system (IVIS). As
shown, the fluorescence signals eliminated fast in free siRNA group,
whereas si-VPN-2 could maintain fairly strong fluorescence signals to 9
days, with significantly superior retention in OA joints than healthy
ones (Fig. [181]4I and Supplementary Fig. [182]33A–C). By contrast,
si-VPN@HA displayed the best performance of intra-articular retention,
its fluorescence signals were considerably extended to over 21 days and
the area under the intensity profile (AUC) was significant increase
relative to si-VPN-2 or si-MC3-LNP, by factors of 2.31, and 3.41,
respectively (Fig. [183]4J–K). Collectively, si-VPN@HA performed well
in prolonging intra-articular retention, and meanwhile upon ROS
stimulus, the released si-VPN-2 also preserved comparable
cartilage-penetrating abilities, thus reconciling the dilemma of
penetration and retention.
Alleviating cartilage degeneration by si-VPN@HA treatment in surgical ACTL
model
The in vivo therapeutic efficacy of si-VPN@HA was evaluated in anterior
cruciate ligament transection (ACTL) mice, a well-established model of
surgically induced OA^[184]45. The affected joints were treated with
one intra-articular injection of different formulations at dose of
1 mg/kg siRNA, all mice were euthanized after four weeks
(Fig. [185]5A). Then the treated joints were stained with H&E and
Safranin O to evaluate whole-joint histopathology (Fig. [186]5B). As
shown, Saline group displayed distinguishing cartilage damage features,
including substantial decline of proteoglycan staining, loss of
cartilage surface, robust synovial thickening and extensive cartilage
erosion. Although si-VPN-2 showed superior alleviation of cartilage
degeneration relative to si-MC3-LNP, there still were detectable
proteoglycan loss and surface discontinuity in cartilage. By contrast,
si-VPN@HA effectively minimized cartilage damages to achieve
better-preserved joint structures with intact cartilage surface and
perichondrium. OARSI histopathology initiative was further adopted to
quantitatively score joint degeneration. Consistently, OARSI score of
si-VPN@HA, with a mean score of 1.4 ± 0.55, was statistically lower
than that of si-VPN-2 or si-MC3-LNP, and statistically equivalent to
Sham group (Fig. [187]5C). Additionally, the bio-safety of si-VPN@HA
was primarily validated by body weight, biochemical indexes and H&E
staining of major organs (Supplementary Figs. [188]34–[189]35).
Fig. 5. Effective alleviation cartilage damage and inflammation in surgical
ACTL model.
[190]Fig. 5
[191]Open in a new tab
A Schematic construction of surgical ACTL model with corresponding
treatment regimen. B Representative images of H&E and Safranin-O
staining of mouse knee joint sections from different treated groups,
scale bar = 100 μm. C OARSI cartilage histopathology assessment scores
of different treated groups, data points represent mean ± SD (n = 5
mice). D Representative images of MMP-13 and NGF immunofluorescent
staining, scale bar = 200 μm in upper row, scale bar = 100 μm in bottom
row. E Semi-quantitation of MMP-13 or NGF expressions on basis of
immunofluorescent staining of (D), data points represent mean ± SD
(n = 5 mice). F Representative images of MMP-9 and ADAMTS5
immunohistochemical staining, scale bar = 100 μm. G Semi-quantitation
of MMP-9 or ADAMTS5 expressions on basis of immunohistochemical
staining of (F), data points represent mean ± SD (n = 5 mice). H
Representative co-immunostaining of F4/80 (green fluorescence), CD80
(cyan fluorescence), and CD206 (red fluorescence) in synovium, and the
nuclei were stained with DAPI (blue fluorescence), scale bar = 100 μm.
I Semi-quantitation of the relative frequencies of M1 macrophages
(CD80^+F4/80^+) or M2 macrophages (CD206^+F4/80^+) on basis of
immunostaining results of (H), data points represent mean ± SD (n = 5
mice). J RT-qPCR measurement of MMP-13, TNF-α, IL-1β, NF-κB1, CCL-2,
CXCL10, ICAM-1 or COX-2 mRNA levels in combined synovial and cartilage
tissues, data points represent mean ± SD (n = 5 mice). Statistical
significance was determined using unpaired one-way ANOVA (C, E, G, I
and J): *p < 0.05, **p < 0.01, ***p < 0.001 and ****p < 0.0001,
NS means no significance. Source data are provided as a [192]Source
data file.
Then, in vivo MMP-13 silencing was monitored by immunofluorescent (IF)
staining, it revealed over 75% reduction of MMP13 expression within
cartilage and meniscus in si-VPN@HA relative to Saline group, which
represented the minimized level among all treated groups
(Fig. [193]5D–E). Inadequately controlled pain, the defining clinical
presentation of OA, is the major reason driving people to seek final
total joint replacement^[194]46. As one of the leading mediators to
activate nociceptive pathway, the expression of nerve growth factor
(NGF) was monitored^[195]47. As indicated, si-VPN@HA treatment
substantially reduced NGF expression by nearly 80% relative to Saline
group in synovium, which was only slightly higher than Sham group,
suggesting si-VPN@HA could effectively relieve from pain. In addition
to knockdown MMP-13, si-VPN@HA also significantly down-regulated the
expressions of MMP-9 and ADAMTS-5, the representative enzymatic markers
of chondrocyte catabolism, relative to si-VPN-2 or si-MC3-LNP group,
implying that si-VPN@HA might mediate collaborated network to alleviate
cartilage degeneration (Fig. [196]5F-G).
Accumulating evidences suggest the essential role of synovial
macrophages in OA pathogenesis, as pro-inflammatory phenotype (M1,
CD80^+F4/80^+) potentiates while anti-inflammatory phenotype (M2,
CD206^+F4/80^+) attenuates OA development^[197]48,[198]49. Therefore,
the phenotypic characterization of macrophages was further identified
(Fig. [199]5H–I). There was a significant elevation in M1 macrophages
in si-MC3-LNP or si-VPN-2 group relative to Sham group, by factors of
8.37 and 6.07, respectively. While the abundance of M1 macrophages
wasn’t statistically different between si-VPN@HA and Sham group. By
contrast, si-VPN@HA group demonstrated highest abundance of M2
macrophages among all treated groups, with 3.51-fold higher than
si-MC3-LNP group. Collectively, si-VPN@HA treatment could effectively
increase synovial macrophage M2 polarization and meanwhile decrease M1
polarization. Subsequently, RT-qPCR assay was conducted to verify
whether si-VPN@HA could broadly impact the expression of
inflammation-associated genes (Fig. [200]5J). As shown, all these
indicators, including cytokines (i.e., TNF-α, IL-1β), transcription
factor NF-κB1, chemokine (i.e., CCL2, CXCL10) as well as inflammatory
mediators ICAM-1 and COX-2 were significantly decreased in si-VPN@HA
group relative to Saline group. Especially, the levels of TNF-α, CCL2
and ICAM-1 were reduced below 35% of Saline group in si-VPN@HA group.
Implementing superior joint protection in synergy with steroid treatment in
PTOA model
In vivo therapeutic study was also carried out in a post-traumatic
osteoarthritis (PTOA) mouse model subjected to non-invasive repetitive
mechanical loading protocol of 9.8 N, 500 cycles, 3 times per week for
6 weeks^[201]50 (Fig. [202]6A). Methylprednisolone (MP), as one of the
most prevalent intra-articular corticosteroids in clinical for OA
intervention^[203]51,[204]52, was employed as a benchmark comparison to
evaluate the therapeutic efficacy of designed formulations. Due to
loading flexibility of HA@gels, we further constructed an upgraded
preparation (i.e., MP&si-VPN@HA) by co-loading si-VPN-2 and MP into
HA@gels to test whether there was any synergistic effect.
Intra-articular injection of MP (4 mg/kg) was given weekly to match the
clinical standard, the rest formulations were administrated every other
week at dose of 1 mg/kg siRNA. Walking speed, as a direct indicator of
joint motion, was recorded weekly by open field analysis. As shown, it
decreased rapidly and maintained at 18.22% of initial level in MP group
after six weeks (Fig. [205]6B). By contrast, the joint motion was
well-preserved by MP&si-VPN@HA, its walking speed was 606.2 cm/min
after six weeks, which was significantly higher relative to
292.6 cm/min in si-VPN-2 group and 475 cm/min in si-VPN@HA group.
Fig. 6. Implementing superior joint protection in synergy with steroid
treatment in PTOA model.
[206]Fig. 6
[207]Open in a new tab
A Schematic construction of noninvasive repetitive mechanical loading
PTOA model (9.8 N, 500 cycles, 3 times per week) with corresponding
treatment regimen, the arrows indicated administration of MP, and the
circle indicated administration of saline, si-VPN-2, si-VPN@HA,
si-NEG@HA or MP&si-VPN@HA. B The walking speeds of different treated
groups; data points represent mean ± SD (n = 5 mice). C Representative
images of H&E, Safranin O and MMP-13 immunohistochemical staining of
knee joint sections from different treated groups, scale bar = 200 μm.
D Representative micro-CT images of 2D cross-sectional images and 3D
reconstructions of affected joints, with reconstructed images of
trabecular structure. E Measurements of osteophyte size at site of the
largest outgrowth from normal cortical bone structure, data points
represent mean ± SD (n = 3 mice). F Quantitative analysis of
subchondral bone parameters, including bone volume/total volume
(BV/TV), trabecular number (Tb.N), trabecular thickness (Tb.Th) and
trabecular separation (Tb.Sp), data points represent mean ± SD (n = 3
mice). Statistical significance was determined using unpaired one-way
ANOVA (E and F) and unpaired two-way ANOVA (B): *p < 0.05,
**p < 0.01, ***p < 0.001 and ****p < 0.0001, NS means no
significance. Source data are provided as a [208]Source data file.
The affected joints were collected at the endpoint study and scheduled
for H&E and Safranin O staining (Fig. [209]6C). As shown, joint
sections from MP, si-VPN-2, and si-VPN@HA group reproduced several
typical pathological characteristics of clinical PTOA, which displayed
more aggressive soft tissue mineralization and larger osteophyte
formation, relative to MP&si-VPN@HA group. The reduced safranin O
staining and discontinuous cartilage surface indicated large
proteoglycan loss and cartilage erosion in MP or si-VPN-2 group, by
contrast, they were markedly improved in MP&si-VPN@HA group. The OARSI
scores showed 5.67-fold elevation in MP group and 3.33-fold increase in
si-VPN@HA group, relative to MP&si-VPN@HA group, suggesting
well-preserved articular cartilage by MP&si-VPN@HA treatment
(Supplementary Fig. [210]36A). Immunohistochemical staining revealed
that MMP-13 expression was minimized by MP&si-VPN@HA treatment in both
articular cartilage and meniscus, with a decrease of ~75% relative to
Saline group (Supplementary Fig. [211]36B).
Osteophyte and ectopic mineralization associated acute pain brings
immense distress to OA patients^[212]53. Therefore, micro-computed
tomography (micro-CT) was conducted to visualize the presence of
osteophytes and ectopic mineralization (Fig. [213]6D). Neither MP nor
si-VPN-2 could significantly block osteophytes formation and ectopic
mineralization, while MP&si-VPN@HA did. Specifically, the osteophytes
outgrowth was reduced to ~10% of Saline group in MP&si-VPN@HA group,
suggested its minimized burden of osteophytes (Fig. [214]6E).
Subchondral bone density, indicated by ratio of bone volume/total
volume (BV/TV)^[215]54, was well-preserved in MP&si-VPN@HA, which was
statistically equivalent to Sham group (Fig. [216]6F). Moreover,
MP&si-VPN@HA was more potent in increasing trabecular number (Tb.N),
trabecular thickness (Tb.Th) while decreasing trabecular separation
(Tb.Sp), relative to si-VPN-2 or MP group, suggesting the superior
efficacy of MP&si-VPN@HA in protecting subchondral bone
microarchitecture. Lastly, body weight and blood routine examination
implied that MP&si-VPN@HA was well-tolerated during in vivo application
(Supplementary Fig. [217]37A–F).
VPN-2 functionally enabled intra-articular mRNA delivery
Furthermore, eGFP mRNA was elected to test the versatility of VPN as
nucleic acid vectors. mRNA-condensed VPNs (i.e., m-VPNs) were yielded
by mixing VPNs with mRNA using microfluidic mixer (Fig. [218]7A).
Agarose gel electrophoresis assay indicated that m-VPN-2 could achieve
complete retardation of mRNA at N:P ratio of ≥ 2.304 (Fig. [219]7B).
Blank VPN-2 showed well-defined unilamellar vesicle structure, by
contrast, m-VPN-2 transformed into solid spherical structure after
loading mRNA, analogous to the morphological transitions of siRNA
entrapment (Fig. [220]7C). Then ATDC5 cells were transfected with
different formulations at equivalent amount of 500 ng mRNA, the eGFP
expression efficiency was characterized. Scarcely any eGFP positive
cells were detected by administrating free mRNA only. By contrast, eGFP
expression level significantly increased in m-VPN-2 group, with 59.98%
eGFP-positive cell population, which was 3.95-fold higher relative to
mRNA loaded MC3-LNP (i.e., m-MC3-LNP) group (Fig. [221]7D–E). Similar
trends were detected in MFI, m-VPN-2 achieved 3.30- and 2.44-fold
increase of MFI than m-VPN-1 and m-MC3-LNP group (Supplementary
Fig. [222]38). The confocal imaging gave the consistent results,
m-VPN-2 group displayed significant elevation in fluorescence intensity
relative to either m-VPN-1 or m-MC3-LNP group (Fig. [223]7F and
Supplementary Fig. [224]39). Moreover, ACLT mice were employed for in
vivo mRNA transfection by intra-articular injection of m-VPN-2 or
m-MC3-LNP at dose of 10 μg mRNA/mice. The affected joints were
harvested after 24 h, and eGFP fluorescence signals within cartilage
sections were detected. As shown, the fluorescence intensity was
4.30-fold higher in m-VPN-2 group than that of m-MC3-LNP group,
suggesting superior ability of m-VPN-2 in delivering mRNA into
cartilage (Fig. [225]7G and Supplementary Fig. [226]40). Collectively,
these data suggested VPN-2 possessed great potential to be tailored as
mRNA delivery vectors for intra-articular application.
Fig. 7. VPN-2 functionally enable intra-articular mRNA delivery.
[227]Fig. 7
[228]Open in a new tab
A Schematic illustrating the fabrication process of m-VPNs. B mRNA
condensation of m-VPN-2 at various N:P ratio by agarose gel
electrophoresis. C TEM images of VPN-2 and m-VPN-2, scale bar = 200 μm
in left column, scale bar = 50 μm in right column. D Representative
flow cytometric analysis of eGFP expressions after different
transfections. E Quantification of eGFP gene expression by flow
cytometry, data points represent mean ± SD (n = 3 independent samples).
F Confocal images of eGFP expression after different transfections,
scale bar = 100 μm. G Fluorescence signals of eGFP expression within
cartilage sections after intra-articular injection of m-VPN-2, scale
bar = 100 μm. Statistical significance was determined using unpaired
one-way ANOVA (E): *p < 0.05, **p < 0.01, ***p < 0.001 and
****p < 0.0001, NS means no significance. Data in (B, C, F and G) are
representative of three independent experiments with similar results.
Source data are provided as a [229]Source data file.
Discussion
OA is labeled as “a serious disease with an unmet medical need” by FDA
in 2018 due to no available disease-modifying OA drugs (DMOADs).
Progressive cartilage breakdown is identified as a major therapeutic
target for potential DMOADs development^[230]55,[231]56. Accumulated
researches highlighted the fatal role of collagenase MMP-13 in
cartilage breakdown, due to its robust ability to degrade
Col2^[232]5,[233]57,[234]58. For these reasons, inhibiting MMP-13
represents a straightforward approach to alleviate cartilage damage.
However, there are several concerns regarding small molecule MMP
inhibitors^[235]4: (i) broad-spectrum MMP inhibitors could cause
musculoskeletal syndrome adverse events due to non-selective inhibition
of multiple MMPs^[236]6; and (ii) MMP13-selective inhibitors
development was suspended owing to the shared domains of
collagenases^[237]59. Alternatively, RNA interference (RNAi) therapy
offering a sequence-specific manner to silence MMP-13 expression by
rationally designing siRNA with selective complementarity to MMP13
mRNA, thus obviating the low selectivity concerns faced by MMP-13 small
molecule inhibitors.
The real-life therapeutic potential of nucleic acids critically depends
on delivery technologies^[238]60. Inspiring by the powerful
transfection competency of virally-derived arginine-rich CPPs and their
polyhistidine-fused derivatives, we innovatively designed a
lipopeptide-based nanoplatform for cartilage-targeted delivery of
nucleic acids. The cationic moiety of lipopeptide, possessing variable
arginine and histidine residues, was the main functional component. The
side chain imidazole ring within histidine could be protonated under
acidic conditions, thus endowed VPNs with pH-sensitive ability to
mediate efficient lysosomal escape via proton sponge
mechanism^[239]24,[240]29. As the inflamed joints were inclined to form
weakly acidic environment (i.e., about pH6.8 or lower), the
acid-sensitive property of VPNs was an additional leverage to
facilitate boosted endocytosis and fortified chondrocyte spheroids
penetration^[241]25,[242]61. Based on in vitro performance, VPN-2 with
a moiety of –[(R)[5]-(H)[4]][2]- was ultimately identified as optimal
vector for siRNA and mRNA delivery. Inversely, despite with identical
number but distinct sequence of arginine and histidine residues, VPN-3
with a moiety of -(R)[10]-(H)[8]- formed apparent aggregations when
applied to cells, which was presumably due to high-degree non-specific
protein adsorption^[243]62,[244]63.
Cartilage is an avascular tissue composed of dense, highly anionic
extracellular matrix. Delivery vectors intended to target chondrocytes,
the only resident cell within cartilage, face the dilemma between
penetration and retention. Specifically, small nano-carriers possess
great potential to penetrate into cartilage to reach chondrocytes, but
they inevitably suffer rapid clearance due to fast physiological
turnover of synovial fluid^[245]26. By contrast, lager-size carriers
(i.e., microparticles or hydrogels) could achieve better-retained
within the joint, however, they can barely penetrate into cartilage
tissue^[246]28. Herein, to fit the physiological feature of articular
cavity, we innovatively designed a nano-in-gel system, by loading
si-VPN-2 into β-CD and Fc-based ROS-responsive HA
hydrogels^[247]43,[248]44. The as-prepared si-VPN@HA could undergo
quick depolymerization to release si-VPN-2 under ROS stimulus, thus
achieving a perfect tradeoff between penetration and retention.
In summary, we have presented a VPN platform based on virus-inspired
lipopeptide for cartilage-targeted nucleic acid delivery. The optimized
VPN-2 could achieve ~85% knockdown of MMP-13 expression in vitro. After
formulated into nano-in-gel system, si-VPN@HA could considerably
extended intra-articular retention to over 21 days, and meanwhile the
released si-VPN-2 still preserved considerable cartilage-penetrating
ability under ROS stimulus. In vivo study demonstrated that si-VPN@HA,
achieving ~75% in vivo MMP-13 silencing, could effectively alleviate
cartilage degeneration, reduce inflammation, and relieve from pain in
surgical ACTL mice. Moreover, the therapeutic potency of this approach
was further amplified by synergy with steroid treatment in non-invasive
repetitive mechanical loading PTOA mice. Additionally, preliminary
evaluation declared that VPN-2 also functionally enabled
intra-articular mRNA delivery. Our study primarily indicated the
platform’s potential of VPN as cartilage-targeted nucleic acid delivery
vectors to provide therapeutic benefit for OA treatment, highlighting
the necessity and significance for further study.
Methods
Animals
C57BL/6J mice (8 weeks) were obtained from Hunan SJA Laboratory Animal
Co., Ltd. (Hunan, China). Mice were housed in specific pathogen-free
condition with a 12 h light-dark cycle at room temperature 23 ± 1 °C,
relative humidity 40–60%, with ad libitum access to food and water. All
animal studies were complied with the guidelines evaluated and approved
by the Institutional Animal Care and Use Committee (IACUC) of Southwest
University Laboratory Animal Center (Approval No. IACUC-20211030-01).
Cells
ATDC5 cell line was obtained from Sigma-Aldrich (99072806-1VL). ATDC5
was cultured in DMEM medium containing 10% FBS and 100 U/mL of
penicillin-streptomycin under 5% CO[2] at 37 °C. The cells at passage
within 5–10 were employed in experiments.
Patient and samples
Three human knee cartilage samples were taken from OA patients (1 male
and 2 females, age range 56–83 years) with Kellgren-Lawrence grade 4
before receiving total knee arthroplasty at Southwest Hospital, the
First Affiliated Hospital of Army Medical University. Informed consent
was obtained from all patients, and the study was approved by the
Ethics Committee of the First Affiliated Hospital of Army Medical
University (approval No. (A) KY2021040).
To isolate primary human chondrocytes, the cartilage tissue specimens
were firstly washed with PBS, and then minced into small pieces and
digested with 0.2% collagenase II at 37 °C for 6 h. After digested, the
triturate suspension was filtered through 100 μm cell strainer to
remove matrix debris, and the resulting cell suspension was centrifuged
at 500 × g for 3 min to move the supernatant. The obtained primary
human chondrocytes were cultured in DMEM medium (KeyGEN) containing 10%
FBS (Gibco, USA) and 100 U/mL of penicillin-streptomycin under 5% CO[2]
at 37 °C. Chondrocytes at passage 2–5 were employed in experiments.
Synthesis and characterization of lipopeptides
The lipopeptides were produced by solid-phase peptide synthesis (SPPS)
protocol using Liberty Blue peptide synthesizer (CEM, USA).
Specifically, Wang-resin was applied as solid phase, N,
N′-methanediylidenebis (DIC) and ethyl cyanohydroxyiminoacetate (Oxyma)
were served as activator and activator base, respectively. Docosanoic
acid (C[22]) was coupled to the peptides before deprotection and
cleaving. The resulting lipopeptides were subsequently cleaved from the
resin and took off the protective groups by incubation in solution
composed of trifluoroacetic acid/triisopropylsilane/water (95:2.5:2.5)
at room temperature (RT) for 2 h. After that, the resin was filtrated,
the solution was added with diethyl ether for precipitation on ice, and
the yield crude lipopeptide products were purified using reverse-phase
HPLC and further characterized by matrix-assisted laser desorption
ionization-time of flight (MALDI-TOF) mass spectrometry (Bruker, USA).
The sequences of all synthetic lipopeptides were showed in
Supplementary Table [249]1.
Construction and characterization of VPNs and si-VPNs
VPNs were prepared based on the self-assembly behavior of lipopeptides.
Briefly, 3 mg of lipopeptides dissolved in 60 μL DMSO was slowly added
into 2 mL DEPC water, VPN were obtained after stirring (400 g) for
10 min. The morphologies of VPNs were assessed by transmission electron
microscope (TEM, HT7800, Hitachi, Japan) and field emission scanning
electron microscopy (FE-SEM, Sigma 500, Carl Zeiss, UK). To evaluate
the acid-sensitive properties of VPNs, VPNs were incubated in 0.1 M PBS
with different pH (i.e., 7.4, 6.8, or 5.5) at 37 °C, and the changes of
size and zeta potential were monitored using particle size analyzer
(Malvern, UK) over time.
siRNA-loaded VPNs (si-VPNs) were prepared using a microfluidic mixer
(Aitesen, China), where equal volume of VPNs and siRNA solution with
appropriate concentration was evenly mixed. The resulting preparations
were further stirred by vortex oscillator for 10 min, and then let them
sit for 2 h. The optimized N:P ratio was obtained using agarose gel
electrophoresis. The electrophoresis was performed under condition of
100 V, 60 mA. For RNase protection assay, si-VPN formulations were
incubated with RNase cocktail enzyme mix (AM2286, Thermo Fisher), RNase
A (EN0531, Thermo Fisher), and RNase I (EN0601, Thermo Fisher) at 37 °C
for 0, 10, 20, 30, and 60 min, respectively. The residual siRNA was
displaced from si-VPN by incubating with 1% sodium dodecyl sulfate
(SDS), then electrophoresis was subsequently conducted to visualize the
amount of extracted siRNA. Equal amount of free siRNA receiving the
same RNase treatment was employed as control. For stability testing,
proper amount of si-VPN-1, si-VPN-2 and si-VPN-3 were incubated in
0.1 M PBS at 4°C, the sizes and zeta potentials of these samples were
monitored. After 48 h, these solutions were further determined by gel
electrophoresis assays with or without addition of SDS. The
morphologies of si-VPNs were assessed by TEM. The elemental analysis of
si-VPN-2 was detected by high angle annular dark field (HAADF)-scanning
transmission electron microscopy (STEM). siRNA,
carboxy-fluorescein-conjugated siRNA (FAM-siRNA), and cyanine 5
conjugated siRNA (Cy5-siRNA) were all purchased from Sangon Biotech
(Shanghai, China).
For MC3-LNP preparation, a mixture of DLin-MC3-DMA, DSPC, cholesterol
and DMG-PEG2000 at a molar ratio of 50:10:38.5:1.5 were dissolved in
ethanol as organic phase^[250]64,[251]65. The organic phase was mixed
with acetate buffer (25 mM, pH 4.0) containing siRNA or mRNA using a
microfluidic mixer at a flow rate ratio of 3:1 (water: ethanol), the
obtained preparations were further purified by dialysis against
deionized water, to yield si-MC3-LNP or m-MC3-LNP.
Fabrication and characterization of HA@gel and si-VPN@HA
The ROS-responsive HA@gel was constructed based on host-guest
interaction of β-CD and Fc^[252]66. β-cyclodextrin conjugated
hyaluronic acid (HA-CD) and ferrocene conjugated hyaluronic acid
(HA-Fc) were synthesized by incorporating CD or Fc onto HA backbone
through amidation reaction. To prepare HA hydrogels, equal volume of
HA-CD and HA-Fc were mixed at ratio of 1:1 (wt%/wt%). The rheology
properties of HA@gel, including storage modulus (G′) and loss modulus
(G′′), were recorded using a rotary rheometer (MCR302, Anton Paar,
Austria) with 20 mm flat plates, when stimulated with 0.1 mM or 5.0 mM
H[2]O[2]. The reversible crosslinking property of HA@gel were detected
with a cyclic stress (0.2%–400%). And 3D network morphologies of HA@gel
were characterized using SEM (Hitachi, SU3500, Japan).
To fabricate si-VPN@HA, 25 μL si-VPN-2 solution and 25 μL 10 mg/ml
HA-CD were firstly mixed to yield mixture 1, then 25 μL si-VPN-2
solution and 25 μL 10 mg/ml HA-Fc were mixed to yield mixture 2,
finally the mixture 1 and mixture 2 were added together to prepare
si-VPN@HA. The release kinetic of Cy5-labeled siRNA from si-VPN-2,
siRNA@HA (i.e., free siRNA loaded HA@gel) and si-VPN@HA under 5 mM
H[2]O[2] conditions was investigated by detecting fluorescence
intensity at 664 nm using fluorescence spectrometer (FL6500,
PerkinElmer, USA). The morphologies of released si-VPN-2 from si-VPN@HA
under 5 mM H[2]O[2] stimulus was characterized by TEM (HT7800, Hitachi,
Japan). Specifically, 200 μL of si-VPN@HA was placed into 1.5 mL tubes,
then 800 μL PBS (pH7.4 or pH6.8) containing 5.0 mM H[2]O[2] was added
into the tubes. After that, the tubes were put on a shaker with gentle
agitation (100 g) at 37 °C. And supernatant (50 μL) was withdrawn from
each tube to prepare TEM samples after 12 h.
Dissipative particle dynamics simulation
Mesoscale dissipative particle dynamics (DPD) simulation was employed
to investigate the structural characteristics and assembly dynamics of
si-VPNs^[253]32–[254]34. Firstly, siRNA molecule was coarse-grained
using DPD approach, its structure was represented by beads
corresponding to phosphate group (P), sugar (S), and bases (A, T, C, G,
U). Similarly, lipopeptides was coarse-grained by representing
docosanoic acid (C[22]) as 11 D beads, and the peptide backbone was
coarse-grained into PB beads, their side chains were represented as 1–2
beads per group. The blend module in Materials Studio 2016 software and
COMPASS force field were employed to calculate the mixing energies
between beads, from which the DPD interaction parameters were derived.
Coarse-grained molecular units of siRNA, lipopeptides and water were
randomly distributed within the template to simulate mesostructured
si-VPNs. Using the derived DPD parameters, a DPD force field was
constructed, and initial conformations were optimized for mesoscale
dynamics simulations. A long-term dynamics simulation for 200 ns was
carried out to analyze the stability and structural evolution of the
system.
Hemolytic activity
The hemolytic activity of VPNs and si-VPNs was evaluated using rabbit
erythrocytes. Equal volume of 4% cell suspension and tested
nanoparticle samples at concentration of 40, 80, 120, 160, 200, 400,
500, 600, 700, and 800 μg/mL (referred by lipopeptide mass) were mixed
and incubated at 37 °C for 1 h. The mixed solutions were set as pH 7.4,
pH 6.8, or pH 5.5. Afterwards, the solutions were centrifuged
(1000 × g, 5 min) to obtain supernatant containing free hemoglobin for
further photometrical measure at 570 nm. And erythrocytes suspended in
PBS and dH[2]O were adopted as the minimal and the maximal hemolytic
controls. The hemolysis fraction (%) was calculated based on the
following equation:
[MATH:
Haemoly<
mi>sisfraction(%)=A<
mrow>T−APBSAdH2O
−APBS×100 :MATH]
Where A[T], A[PBS], and A[dH2O] were the absorbance of the tested
nanoparticle sample, PBS, and dH[2]O, respectively.
Cytotoxicity study
ATDC5 cells were seeded in 96-well plates at a density of 5000
cells/well, and MTT experiments were performed when reaching 70–80%
confluency. Specifically, appropriate volume of VPNs, and si-VPNs at
various concentrations (referred by lipopeptide mass) was added into
the medium, the medium solutions were set as pH 7.4, pH 6.8, or pH 5.5,
and then the cells were cultured for an additional 24 h. After that,
the medium was discarded, and MTT solution (5 mg/ml) was added into
each well to incubate at 37 °C for 4 h. Then 150 μL DMSO was added into
each well after removing the supernatant. Cell viability was determined
by measuring the absorbance of the DMSO solutions at 570 nm.
Cellular uptake study
To induce chondrogenesis, ATDC5 cells were cultured in basic DMEM
medium supplemented with 10 μg/ml recombinant insulin (P3378, Beyotime)
for 7 days^[255]36. Then the effects of incubation time, pH, WYRGRL
targeting and cellular uptake mechanism were studied. For time study,
insulin-induced ATDC5 cells were incubated with free siRNA, si-VPN-1,
si-VPN-2, si-VPN-3 or si-MC3-LNP at final dose of 25 nM Cy5-siRNA for
1 h, 2 h, 4 h, or 8 h, respectively, employing untreated cells as
control, and then prepared for flow cytometer (FACSverse, BD) analysis.
For pH study, insulin-induced ATDC5 cells were incubated with si-VPN-1,
si-VPN-2, si-VPN-3 at final dose of 25 nM Cy5-siRNA under pH 7.4 or pH
6.8 for 4 h, and then prepared for flow cytometer analysis. For WYRGRL
targeting study, three types of cells were used: (i) non-induced ATDC5
cells (control); (ii) insulin-induced ATDC5 cells; iii) Col2α1 siRNA
knockdown ATDC5 cells, which were transfected with Col2α1 siRNA
(antisense sequence: 5′-UUACCAGUGUGUUUCGUGC-3′) using Lipofectamine
2000 (Cat.11668019, Invitrogen). These three cells were incubated with
si-VPNs at final dose of 25 nM Cy5-siRNA, respectively. After that, the
cells were detected by flow cytometer or photographed using Operetta
CLS high content analysis system (Perkin Elmer, USA) after stained with
DAPI dye (Dingguo Prosperous, China) and rhodamine labeled phalloidine
(Cat.R415, Invitrogen). For cellular uptake mechanism study,
insulin-induced ATDC5 cells were pre-treated with 4 °C, Amiloride
(100 μM), Chlorpromazine (20 μM), Filipin (1.25 μM), Brefeldin A
(100 μM), or β-CD (10 mg/ml) for 1 h. Then si-VPN-2 was added into the
cells at final dose of 25 nM Cy5-siRNA and incubated for 2 h. After
that, the cells collected for flow cytometer analysis.
Primary human chondrocytes were also employed for cellular uptake
study. Cells were incubated with free siRNA, si-VPN-1, si-VPN-2,
si-VPN-3, or si-MC3-LNP at final dose of 50 nM Cy5-siRNA for 2 h, and
then analyzed by flow cytometer or photographed using Operetta CLS high
content analysis system. The obtained flow cytometry data were analyzed
by FlowJo V10.
SPR measurements
Binding kinetics between assigned peptides and Col2 were determined by
SPR (Nicoya Lifescience, Waterloo, Canada)^[256]65,[257]67. Firstly,
the COOH sensor chip was operated with PBS running buffer at a constant
flow rate of 150 μL/min, and preconditioned with 80% isopropanol to
eliminate air bubbles and stabilize baseline drift. Col2 protein
(25 μg/mL) was immobilized on the COOH sensor chip by using a standard
amine coupling procedure after activation with 0.2 M EDC and 0.05 M NHS
for 5 min at 20 μL/min. Analytes were injected at a flow rate of
20 μL/min to measure the interaction between the peptide and the
immobilized Col2 protein. Equilibrium dissociation constants (KD) were
calculated using Tracedrawer software.
In vitro protein adsorption evaluation
To evaluate protein adsorption, proper amount of siVPN-1, siVPN-2, and
siVPN-3 was added into 10% fetal bovine serum (FBS) and incubated at
37 °C for 2 h. Firstly, the turbidities of samples were analyzed by
measuring the optical densities at 600 nm using a microplate reader
(BioTeck Synergy H1)^[258]62. Then the mixtures were centrifuged at
14,000 × g for 40 min at 4 °C, the pellet was washed and resuspended
with cold PBS for twice. And the protein concentration was further
measured using the Micro BCA assay (Cat. 23235, Thermo Scientific)
following the manufacturer’s instructions^[259]63. For SDS-PAGE
analysis, FBS was employed as a control. The samples were resuspended
in 60 μL PBS with additional 15 μL of 5× SDS-PAGE sample buffer. After
being boiled for 10 min to denature proteins, the samples were loaded
and analyzed by SDS-PAGE (Bio-Rad), followed by Coomassie Brilliant
Blue staining^[260]62,[261]63. Semi-quantitative analysis of adsorbed
protein bands (approximately 75 kDa) was performed using Image J based
on grayscale intensity.
Multicellular chondrocyte spheroids penetration test
The multicellular chondrocyte spheroids were obtained by using
ultra-low attachment 96-well plates (Nunclon Sphera, round bottom,
Thermo Scientific). Briefly, 1 × 10^3 of insulin-induced ATDC5 cells
were seeded into each well. After slow and continuous shake for 5 min
to ensure close contact of cells, the plate was put into cell
incubator, and cultured under 5% CO[2] at 37 °C. The formation of
spheroids was monitored under electron microscope, and the spheroids
were ready for penetration test when their sizes reached over 200 μm.
Appropriate amount of Free siRNA, si-VPN-1, si-VPN-2, si-VPN-3, or
si-MC3-LNP was added into each well at a final concentration of 250 nM
Cy5-siRNA, the cells were incubated under pH 7.4 or pH 6.8 for
additional 4 h, 8 h, or 12 h. After that, the cartilage spheroids were
transferred to a confocal microscopy dish, and the fluorescent images
of spheroid were acquired using Z-stack scanning of CLCSM (Zeiss).
Lysosome co-localization
For insulin-induced ATDC5 cells, the cells were seeded at a density of
1 × 10^4 cells per well in 24-well plates, the medium was replaced with
fresh medium containing Free siRNA, si-VPN-2, or si-MC3-LNP at a final
concentration of 50 nM Cy5-siRNA. Then the cells were incubated for
additional 2 h, 6 h, or 12 h. After that, the cells were stained with
Lyso Green (KGE2209, KeyGEN, China) at 37 °C for 30 min. The nuclei
were stained with Hoechst 33342, and fluorescent images were visualized
using CLCSM (Zeiss). For primary human chondrocytes, the cells were
seeded at a density of 1 × 10^4 cells per well in 24-well plates, the
medium was replaced with fresh medium containing Free siRNA, si-VPN-2,
or si-MC3-LNP at a final concentration of 50 nM FAM-siRNA. Then the
cells were incubated for additional 6 h. After that, the cells were
stained with Lyso-Tracker Red (KGE2207, KeyGEN, China) at 37 °C for
30 min. And fluorescent images were visualized using Operetta CLS high
content analysis system (Perkin Elmer, USA). Colocalization analysis
was done with Image Pro Plus 6.0 software.
Cellular immunofluorescent staining
After transfection, cells were then washed with PBS for 3 times, fixed
in 4% formalin for 15 min, permeabilized with 0.1%Triton X-100 for
10 min and blocked with 1% BSA for 2 h, respectively. After that, the
cells were incubated with primary anti-MMP-13 (ab39012, Abcam, 1/200
dilution) for 2 h and Cy5 labeled goat anti-Rabbit IgG (H + L) second
antibody (Cat. A10523, Thermo Scientific, 1/1000 dilution) for 2 h.
Subsequently, the cells were further stained with DAPI dye (Dingguo
Prosperous, China) for nuclear labeling, and rhodamine labeled
phalloidine (Cat.R415, Invitrogen) for cytoskeleton labeling. Finally,
the cells were visualized using Operetta CLS high content analysis
system (Perkin Elmer, USA).
Quantitative RT-PCR
MMP-13 silencing was detected in both insulin-induced ATDC5 cells and
primary human chondrocytes. Firstly, the insulin-induced ATDC5 cells
were seeded at a density of 1 × 10^5 cells per well in 6-well plates.
After reaching 50–60% confluency, the cells were stimulated with
25 ng/ml TNFα for 24 h. Then si-VPN-1, si-VPN-2, si-VPN-3 or si-MC3-LNP
were introduced into medium at final dose of 50 nM or 100 nM siRNA.
Fresh medium was added to the plates after 12 h post transfection, and
cells were cultured for additional 24 h. After that, the cells were
harvested, and total RNA was extracted by using RNA Easy Fast Kit
(Cat.DP451, TianGen, China). The first-strand cDNA was produced through
reverse transcription of the total RNA using Quantscript RT Kit
(Cat.KR103-04, TianGen, China). Quantitative real-time PCR was
performed with the resulting cDNA, corresponding primers, and SuperReal
PreMix Plus kit (Cat.FP205-02, TianGen, China) to monitor the relative
level of MMP-13 mRNA; GAPDH was employed as internal control. The
MMP-13 silencing efficiency of si-VPN-1, si-VPN-2, si-VPN-3, or
si-MC3-LNP in primary human chondrocytes was measured in the same way.
The antisense sequence of mouse and human MMP-13 siRNA was shown in
Supplementary Tables [262]2–[263]3, respectively. And the sequences of
primers were shown in Supplementary Tables [264]4–[265]5, which were
purchased from Sangon Biotech (Shanghai, China).
RNA-Seq analysis
To characterize the global change of gene expression, TNFα activated
primary human chondrocytes were treated with si-VPN-2 for 6 h, and then
cultured for an additional 24 h, taking TNFα activated untreated cells
as control. Total RNA was extracted using the TRIzol reagent (Cat.
15596026CN, Invitrogen) and sequenced using the Novaseq 6000 platform
(Illumina, USA). Libraries were constructed using VAHTS Universal V6
RNA-seq Library Prep Kit for Illumina, and raw data of fastq format
were processed using Trimmomatic. Differentially expressed genes (DEGs)
were identified under the filter conditions when P value was less than
or equal to 0.05, and fold Change > 2 or fold Change < 0.5. Heat map
and KEGG pathway enrichment analysis of DEGs were also performed.
Ex vivo trypsin-damaged cartilage explants penetration
Fresh pig cartilage was obtained from excised knee joint and sliced
into 6-by-1-mm cylindrical cartilage explants. After washing with PBS
containing penicillin-streptomycin, cartilage explants were further
treated with 500 μL 2.5% trypsin in 48-well plate for 30 min at 37 °C
to mimic the lesion of OA. After that, different formulations with
equal amount of Cy5-labeled siRNA (at final concentration of 1 μM): (i)
free siRNA; (ii) si-VPN-1; (iii) si-VPN-2; (iv) si-VPN-3; (v)
si-MC3-LNP; (vi) si-VPN@HA; (vii) si-VPN@HA + H[2]O[2] (added with 5 mM
H[2]O[2]), were added into the medium and incubated with cartilage
explants at 37 °C for 24 h with gentle agitation. After incubation, the
medium was discarded, the cartilage explants were washed with PBS, and
then detected using VISQUE in vivo Smart-LF System (Vieworks, Korea).
After that, the cartilage explants were embedded using tissue
OCT-freeze medium (Sakura, Japan), and sectioned into 15 μm slices
using freezing microtome (Leica, Germany) and mounted on glass slides.
After fixed by 4% paraformaldehyde and stained with DAPI, the
fluorescence images of sections were acquired using Operetta CLS high
content analysis system (Perkin Elmer, USA). Quantitative analysis was
further performed.
Intra-articular retention
Male ACLT mice (4 weeks after surgery) were randomly divided to four
groups (n = 3) and treated with an intra-articular injection of
following formulations: free siRNA, si-MC3-LNP, si-VPN-2, or si-VPN@HA,
with equal amount of Cy5-labeled siRNA (0.5 mg/kg) in each group.
Besides, an intra-articular injection of si-VPN-2 was also assessed in
healthy mice. Fluorescence images of joints were acquired at
predetermined time using a VISQUE in vivo Smart-LF System (Vieworks,
Korea). And the radiation efficiency-time curve and the area under
curve were quantified correspondingly.
Surgical ACTL mice model and in vivo therapeutic study
Male C57BL/6 J mice (8 weeks) were adopted for surgical ACLT model
study. Briefly, mice were firstly anesthetized by 5% chloral hydrate,
and then their right hindlimb was shaved and disinfected with iodophor
sanitizer. The joint capsule was exposed by cutting a shallow medial
parapatellar incision and separating the musculature. Then, the joint
capsule was opened by a second incision, which was further extended
using tweezers to make the patella subluxate laterally. The sham
operation was completed at this point. For surgical ACTL model, the ACL
was exposed by flexing the knee at 90°, and further transected with
micro-scissors, anterior drawer test was conducted to confirm the
successful transection. After that, the capsule and skin were closed
separately, and the wound was disinfected with iodophor sanitizer. Five
days after surgery, mice were randomly divided into six groups: Sham,
Saline (treated with saline), si-MC3-LNP, si-VPN-2, si-VPN@HA, and
si-NEG@HA (loaded with scrambled MMP-13 siRNA), with one
intra-articular injection of different formulations (1 mg/kg siRNA)
given into the affected joint. After treatment, body weight was
assessed every week. Mice were euthanized after 4-week treatment, the
affected joints were collected and fixed, blood was collected for serum
biochemical indicator test, and major organs were collected for H&E
staining.
The fixed joints were firstly decalcified and embedded in paraffin,
followed by sectioned into 10 μm thick slices. Then the joint sections
were stained with H&E and Safranin O to evaluate whole-joint histology.
For immunohistochemistry staining, the joint sections from different
groups were stained with antibodies against anti-MMP-13(ab39012, Abcam,
1/200 dilution), anti-NGF (ab216419, Abcam, 1/100 dilution), anti-MMP-9
(ab283575, Abcam, 1/200 dilution), anti-ADAMTS5 (PA5-27165, Thermo
Scientific, 1/100 dilution), anti-F/80(14-4801-82, eBioscience, 1/100
dilution), anti-CD206 (PA5-46994, Invitrogen, 1/40 dilution) and
anti-CD80 (ab254579, Abcam, 1/200 dilution). For RT-qPCR assessment,
joint tissues were collected at the end of treatment, where cartilage
tissue and synovial tissue of each sample were mixed in equal mass, and
total RNA was extracted by using the RNAprep Pure Tissue Kit
(Cat.DP431, TianGen, China), and prepared for quantitative RT PCR to
monitor the relative level of MMP-13, TNF-α, IL-1β, NF-κB1, CCL2,
CXCL10, ICAM-1 and COX-2 mRNA. The sequences of all primer pairs were
showed in Supplementary Table [266]5, which were purchased from Sangon
Biotech (Shanghai, China).
Mechanical loading PTOA mice model and in vivo therapeutic study
Female C57BL/6 J mice (8 weeks) were adopted for non-invasive
repetitive mechanical loading PTOA model study. Briefly, the mice were
firstly anesthetized by 5% chloral hydrate, the hindlimb was positioned
into a custom 3D-printed fixator fitting for the knee, ankle, and leg,
to make the tibia approximately vertical and placed directly under the
loading point. Non-invasive repetitive mechanical loading was conducted
by 500 cycles of 9.8 N on the knee joint of mice and preformed 3 times
per week for six weeks. Mice were randomly divided into seven groups:
Sham (without mechanical loading), Saline, si-VPN-2, si-VPN@HA,
si-NEG@HA, methylprednisolone (MP), and MP&si-VPN@HA (i.e., MP and
si-VPN-2 were co-loaded into HA@gel), and intra-articular injection of
these formulations were given at equivalent dose of 1 mg/kg siRNA and
4 mg/kg MP (if appliable). Except MP was administrated once a week, the
rest formulations were administrated every two weeks. Body weight and
walking speed (cm/min) were recorded every week. For walking analysis,
it was measured by open field test using a clear chamber of 50 cm in
length × 50 cm in width × 25 cm in height, each mouse was placed at the
center of chamber and allowed to move freely. The mouse movement
trajectory was recorded by a tracking system with a duration of 30 min.
The average walking speed was calculated as total moving distance (cm)
divided by total moving time (min). Mice were euthanized after six-week
treatment, and the treated joints were collected for H&E, safranin O,
MMP-13 immunohistochemical staining (ab39012, Abcam, 1/200 dilution),
and micro-CT, and the blood was collected for complete blood count.
Micro-CT analysis
For Micro-CT analysis, the treated mice knee joints (including the
proximal tibia and femoral head) were collected from different groups
and fixed with 4% paraformaldehyde for 24 h, and then were examined
using a micro-CT imaging system (80 kV, 500 μA, 18 μm resolution,
SkyScan, Beckman Coulter, USA). 3D renderings were further
reconstructed using Scanco software. Bone morphometric parameters
including bone volume/total volume (BV/TV), trabecular number (Tb.N),
trabecular thickness (Tb.Th), and trabecular separation (Tb.Sp) were
quantitatively analyzed.
mRNA transfection
eGFP mRNA was acquired from GenScript BioTechnologies(C9450JJSG0-1).
Equal volume of VPNs and mRNA solution were mixed using a microfluidic
mixer (Aitesen, China) to prepare mRNA-condensed VPNs (i.e., m-VPNs).
Agarose gel electrophoresis was performed to investigate the binding
ratios, where 200 ng mRNA was mixed with various concentrations of
VPN-2. The electrophoresis was performed under 100 V, 60 mA, and the
bands were visualized by using SYBR Gold nucleic acid gel stain
(Invitrogen, Waltham, USA). The morphologies of m-VPN-2 were assessed
by TEM. For in vitro mRNA transfection, insulin-induced ATDC5 cells
were seeded in 24-well plate and transfected with different
formulations at dose of 500 ng mRNA for each well for 6 h. After that,
the solutions were removed and fresh medium was added; the cells were
further incubated for 24 h. The efficiencies of eGFP expression were
detected by both flow cytometry and Operetta CLS high content analysis
system (Perkin Elmer, USA). For in vivo mRNA transfection, surgical
ACTL mice (n = 3) were treated with intra-articular injection of
m-VPN-2 or m-MC3-LNP at dose of 10 μg mRNA per mice, then the mice were
euthanized after 24 h and their joints were harvested. After fixed with
4% paraformaldehyde and decalcified with nitric acid, the joints were
embedded in OCT and cut into 15 μm frozen slices. After stained with
DAPI, the fluorescence signals of eGFP within cartilage sections were
acquired using Operetta CLS high content analysis system (Perkin Elmer,
USA).
Statistical analysis
Statistical analysis was performed by GraphPad Prism software (Version
7). All quantitative data were reported as mean ± standard deviation
(SD). Two-sided unpaired Student’ t-test was used for two-group
comparisons, and unpaired one-way or two-way analysis of variance
(ANOVA) test was used for multiple comparisons. Differences were
considered statistically significant when P < 0.05.
Reporting summary
Further information on research design is available in the [267]Nature
Portfolio Reporting Summary linked to this article.
Supplementary information
[268]Supplementary Information File^ (11.1MB, pdf)
[269]Reporting summary^ (14.8MB, pdf)
[270]Transparent Peer Review file^ (2.8MB, pdf)
Source data
[271]Source Data^ (3MB, xlsx)
Acknowledgements