Abstract

   To effectively treat osteoarthritis (OA), the existing inflammation
   must be reduced before the cartilage damage can be repaired; this
   cannot be achieved with a single type of extracellular vesicles (EVs).
   Here, a hydrogel complex with logic‐gates function is proposed that can
   spatiotemporally controlled release two types of EVs: interleukin 10
   (IL‐10)^+ EVs to promote M2 polarization of macrophage, and SRY‐box
   transcription factor 9 (SOX9)^+ EVs to increase cartilage matrix
   synthesis. Following dose‐of‐action screening, the dual EVs are loaded
   into a matrix metalloporoteinase 13 (MMP13)‐sensitive self‐assembled
   peptide hydrogel (KM13E) and polyethylene glycol diacrylate/gelatin
   methacryloyl‐hydrogel microspheres (PGE), respectively. These materials
   are mixed to form a “microspheres‐in‐gel” KM13E@PGE system. In vitro,
   KM13E@PGE abruptly released IL‐10^+ EVs after 3 days and slowly
   released SOX9^+ EVs for more than 30 days. In vivo, KM13E@PGE increased
   the CD206^+ M2 macrophage proportion in the synovial tissue and
   decreased the tumor necrosis factor‐α and IL‐1β levels. The aggrecan
   and SOX9 expressions in the cartilage tissues are significantly
   elevated following inflammation subsidence. This performance is not
   achieved using anti‐inflammatory or cartilage repair therapy alone. The
   present study provides an injectable, integrated delivery system with
   spatiotemporal control release of dual EVs, and may inspire logic‐gates
   strategies for OA treatment.

   Keywords: controlled release, extracellular vesicle, logic‐based,
   osteoarthritis, stimulus‐response
     __________________________________________________________________

   Osteoarthritis requires a long and stepwise treatment process. In this
   study, two types of extracellular vesicles (EVs) with anti‐inflammatory
   and cartilage repair functions are injected into the joint cavity. By
   using the “matrix metalloproteinase 13 sensitivity” AND
   “microsphere‐in‐gel” structure of KM13E@PGE delivery materials, the two
   EVs are spatiotemporal controlled release for macrophages and
   chondrocytes, achieving an innovative strategy of “anti‐inflammatory
   first” AND “cartilage repair second” for osteoarthritis treatment.

   graphic file with name ADVS-11-2403227-g006.jpg

1. Introduction

   Osteoarthritis (OA) is a prevalent degenerative joint disorder with a
   range of symptoms, including pain, limited mobility, and physical
   deformity.^[ [56]^1 ^] These symptoms can cause instability and
   physical disability, ultimately reducing the overall quality of life
   for the patient. The pharmacological treatment of OA primarily targets
   the accompanying joint inflammation and cartilage damage via
   nonsteroidal anti‐inflammatory drugs, corticosteroids, sodium
   hyaluronate, and glucosamine.^[ [57]^2 , [58]^3 , [59]^4 ^] The
   research‐ and clinical treatment‐based consensus is that
   anti‐inflammatory and cartilage‐repair therapies should not be pursued
   concurrently.^[ [60]^5 , [61]^6 ^] This recommendation is largely based
   on the presence of high levels of inflammatory factors,^[ [62]^7 ^]
   such as tumor necrosis factor‐alpha (TNF‐α) and interleukin‐6 (IL‐6).
   TNF‐α promotes protease activation and continuous dissolution of
   cartilage collagen in the inflammatory environment, whereas IL‐6
   prompts matrix metalloproteinase (MMP) expression following binding to
   receptors, thereby causing cartilage matrix destruction.^[ [63]^8 ^]
   Within the cells, IL‐1β stimulates the chondrocytes to produce
   prostaglandin E2 (PGE2), an inhibitor of cartilage matrix synthesis
   metabolism.^[ [64]^9 ^] Therefore, improving the OA inflammatory
   microenvironment is of primary significance, as this environment
   directly affects the cartilage destruction behavior and the ultimate
   curative effect of cartilage repair. In accordance with current
   regenerative medicine research.^[ [65]^10 , [66]^11 ^] Applications of
   stem cells, growth factors, and extracellular vesicles (EVs) to OA
   target one aspect only: either anti‐inflammatory action or cartilage
   repair. Thus, there is a lack of integrated strategies that follow the
   AND logic‐gate, as “anti‐inflammatory AND cartilage repair”, and more
   importantly “anti‐inflammatory first, cartilage repair second”.

   EVs offer significant advantages over stem cells for OA therapy. They
   provide a safer alternative by retaining therapeutic properties while
   eliminating the risk of tumorigenesis associated with stem cell
   transplantation.^[ [67]^12 , [68]^13 , [69]^14 ^] Moreover, EVs allow
   for improved targeting, ease of storage, and administration, making
   them promising candidates for the development of effective and scalable
   OA treatments.^[ [70]^15 ^] Mesenchymal stem cell‐derived EVs (MSC‐EVs)
   may stimulate chondrocyte matrix synthesis, or activate stem cells
   during OA repair.^[ [71]^16 ^] In the context of anti‐inflammatory
   therapy, EVs rich in anti‐inflammatory cytokines, such as IL‐10 and
   IL‐6, may serve as macrophage M2 polarization switchers.^[ [72]^17 ^]
   Furthermore, EVs from the macrophage cell line RAW264.7 may enhance the
   targeted therapeutic effects of immune regulation by promoting
   internalization by macrophages via integrin interaction.^[ [73]^17 ,
   [74]^18 ^] In the context of cartilage repair, MSC‐EVs and
   platelet‐rich plasma EVs can encourage chondrocyte proliferation,
   promote chondrocyte matrix synthesis, and reduce joint cartilage
   degeneration.^[ [75]^19 ^] While infrapatellar fat pad‐derived
   mesenchymal stem cells (IPFP‐MSCs) located near the OA joint may also
   contribute to disease development, their EVs, which are rich in
   proteoglycan 4 (PRG4) and microRNA (miR)−100‐5p, can potentially
   support cartilage repair.^[ [76]^20 ^] Although EVs have multiple
   roles, a certain type of EV may possess relatively singular therapeutic
   effects and may require appropriate engineering or specific cultural
   environments for activation of their therapeutic effects.^[ [77]^21 ^]
   Considering the Boolean “AND” logic‐gates therapeutic function required
   for OA, which targets first anti‐inflammatory action and then cartilage
   repair, the use of at least two types of EVs may be necessary. And
   obviously, it is challenging to deliver two types of EVs at different
   times and in different areas.

   Intravenous injection of EVs usually yields a low EV concentration
   within the joint, owing to the relatively closed anatomical structure
   of the joint cavity.^[ [78]^5 ^] However, injection into the joint
   cavity ensures local, high‐concentration EV enrichment.^[ [79]^22 ^]
   Sodium hyaluronate and chitosan are commonly used for joint‐cavity
   injections in clinical settings.^[ [80]^5 , [81]^23 ^] Hydrogels with
   similar properties may serve as excellent EV‐delivery materials by
   providing a physical buffer to the joint cavity.^[ [82]^2 ^]
   High‐water‐content materials prevent damage to EVs and ensure a higher
   EV load.^[ [83]^2 , [84]^24 ^] Self‐assembling peptides (SAPs) are
   short peptides that spontaneously form beta folds based on specific
   amino acid sequences.^[ [85]^25 ^] These peptides form hydrogels at the
   macroscopic scale, enabling the insertion of functional peptides; this
   aspect renders them useful as hydrogels for biological response
   strategies.^[ [86]^26 ^] In cases of OA, the level of matrix
   metalloproteinase 9 (MMP9) and MMP13 increase within the joint
   cavity.^[ [87]^27 ^] The incorporation of enzyme‐substrate sequences
   into SAPs enables their breakdown in response to the MMP9‐ and
   MMP13‐containing inflammatory environment, thereby controlling the
   release of the drugs they contain. Our previous research highlighted
   the use of MMP1‐sensitive SAP in clinical treatment.^[ [88]^28 ^] The
   ability of MMP1‐sensitive SAP to respond to neovascularization and
   regulate exosome release spatially and temporally offers a novel
   strategy for controlled release. In the context of OA pathology,
   MMP13‐sensitized hydrogels can initiate drug release and achieve an
   “anti‐inflammatory first” effect. Challengingly, the “cartilage repair
   second” effect needs to be combined with another system with
   differential release rates.

   In this study, we construct an MMP13‐sensitive hydrogel complex that
   responds to the OA inflammatory environment and releases
   anti‐inflammatory EVs as an initial step, followed by
   cartilage‐repairing EVs. In detail, we genetically engineer IL‐10^+ EVs
   (RIE) for anti‐inflammatory treatment and induce SOX9^+ EVs (CIE) for
   cartilage repair. The two EVs are packed in SAPs and microspheres,
   respectively, which are combined to form a composite
   “microspheres‐in‐gel” system. Due to the different release rates of the
   two EVs in the hydrogel complex, and the effect of the two EVs’
   different targeting properties, spatiotemporal controlled release of
   the drug was achieved. This hydrogel complex can be injected directly
   into the joint cavity with minimal invasiveness. The anti‐inflammatory
   RIE is released in response to MMP13. Slow release of the CIE then
   occurs, which promotes cartilage repair. Hence, step‐by‐step,
   integrated treatment of OA is enabled via a new type of logic‐based
   material that is jointly controlled by biological responses and its
   physical structure.(Scheme [89]1 ).

Scheme 1.

   Scheme 1
   [90]Open in a new tab

   Schematic illustration: the KM13E@PGE system spatiotemporally releases
   of two types of EVs according to its MMP13‐sensitive characteristics
   and “microspheres‐in‐gel” structure, resulting in step‐by‐step efficacy
   of anti‐inflammatory and cartilage repair to improve OA.

2. Results

2.1. IL‐10 Overexpression within EVs Enhances Anti‐Inflammatory Capabilities

   To increase the IL‐10 content within the EVs, we overexpressed IL‐10 in
   the RAW264.7 macrophage cell line by using a lentiviral vector (Figure
   [91]S1, Supporting Information) to increase the IL‐10 mRNA (Figure
   [92]1A). We obtained EVs produced with WT‐RAW (RE) and IL‐10^+ RAW
   (RIE) from a cell culture medium. Both the RE and RIE had cup‐shaped
   morphologies (Figure [93]1B) with particle sizes within the 80–150 nm
   range (Figure [94]1C). No significant difference in protein
   concentration was determined for the RE and RIE secretion levels
   (Figure [95]S2A, Supporting Information). Both the RE and RIE were
   highly enriched with CD9, CD81, and ALIX proteins, which are well‐known
   EV markers (Figure [96]1D). Therefore, we successfully isolated RE and
   RIE. To clarify the impact of the IL‐10 overexpression on the RIE
   protein composition, protein profiling was performed: 1101 proteins
   were detected in the RIE and RE, of which 360 and 487 were down‐ and
   upregulated, respectively, by the log[2] fold change (Figure [97]1E).
   Gene ontology (GO) analysis revealed that the biological processes of
   the differentially expressed proteins (DEP) were primarily inflammatory
   and cytokine regulation and that the DEP cellular components and
   molecular functions were primarily associated with extracellular
   proteins and adenosine triphosphate (ATP) binding (Figure [98]1F).
   Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis
   revealed that the phosphoinositide 3‐kinase–protein kinase B (PI3K‐Akt)
   signaling pathway was the most significant (Figure [99]1G). These
   results demonstrate that RIE significantly influences anti‐inflammation
   activity. Furthermore, we quantified the DEP via an enzyme‐linked
   immunosorbent assay (ELISA); the results showed significantly higher
   IL‐10, C─C motif chemokine ligand 8 (CCL18), and integrin subunit alpha
   4 (ITGα4) levels in the RIE compared to the RE. The nitric oxide
   synthase (iNOS) level was significantly lower, and no significant
   differences were obtained for ITGβ1, ITGβ2, and ITGαM (Figure [100]1H;
   Figure [101]S2B, Supporting Information).

Figure 1.

   Figure 1
   [102]Open in a new tab

   RIE characterization and screening of optimal RIE concentrations. A)
   IL‐10 mRNA expression in RAW264.7 before and after transfection. B) EV
   morphologies observed via transmission electron microscopy (TEM); scale
   bar: 50 nm. C) EV particle size distribution measured via nanoparticle
   tracking analysis (NTA). D) Western blotting of EV markers. E) DEP
   volcano map between RIE and RE; p < 0.05; log[2] fold change. F,G) GO
   enrichment and KEGG pathway analyses, respectively, of DEP between RIE
   and RE. H) Inflammation‐related DEP quantified via ELISA, for RE and
   RIE. For M1‐polarized RAW264.7 samples treated with various RIE
   concentrations for 24 h: I) CD206 fluorescent staining of samples;
   scale bar: 100 µm; J) CD206^+ cell percentages following CD206
   fluorescent staining; and K) M1 polarization‐related protein iNOS and
   M2 polarization‐related protein ARG1 concentrations, quantified by
   ELISA. ^***: p < 0.001, ^**: p < 0.01, ^*: p < 0.05, ns: p > 0.05.

   Following the co‐culturing of the RIE with M1‐polarized RAW264.7 cells
   in vitro, immunofluorescence analysis was performed; the results showed
   that RIE concentrations exceeding 200 µg mL^−1 significantly reversed
   the M1 polarization, yielding a significant amount of CD206^+ M2
   polarization (Figure [103]1I,J). ELISA results also confirmed this
   finding, as the iNOS decreased with increasing RIE concentration and
   was negatively correlated with the M2 polarization. In contrast, the
   Arginase‐1 (ARG1) expression was positively correlated with the M2
   polarization and increased with increasing RIE concentration. Regarding
   the effective RIE dose, there was no significant difference between 200
   and 300 µg/mL (Figure [104]1K). These results indicate the successful
   construction of IL‐10^+ EVs promoting macrophage M2 polarization, that
   is, RIE.

2.2. Chondrogenically Differentiated IPFP‐MSC‐Derived EVs Promote Cartilage
Matrix Synthesis

   Through enzymatic digestion of infrapatellar adipose tissue discarded
   during arthroscopic surgery, we successfully obtained IPFP‐MSCs (Figure
   [105]S3A,B, Supporting Information). These cells showed a strong
   preference for chondrogenic differentiation (Figure [106]S3C,
   Supporting Information). Immunofluorescence analysis showed that the
   chondrogenically differentiated IPFP‐MSCs (C‐IPFP‐MSCs) formed dense
   chondrospheres rich in type‐II collagen (COL2), as shown in Figure
   [107]2A. We extracted IPFP‐MSC‐derived EVs (IE) and C‐IPFP‐MSC‐derived
   EVs (CIE) from a cell culture medium soaked with chondrospheres. Both
   the IE and CIE assumed cup‐shaped morphologies (Figure [108]2B) with
   particle sizes in the range of 90 to 140 nm (Figure [109]2C). There
   were no statistically significant differences in the IE and CIE
   secretion levels (Figure [110]S4, Supporting Information). Both the IE
   and CIE expressed the CD9, CD81, and ALIX EV markers, but did not
   express calnexin (Figure [111]2D). To further clarify the impact of
   chondrogenic differentiation on the CIE protein components, protein
   profiling was performed. As shown in the volcano plot (Figure [112]2E),
   521 proteins were totally quantified in the CIE and IE groups, of which
   98 downregulated and 131 upregulated, DEP in the CIE group, compared
   with the IE group, by the log[2] fold change. GO analysis revealed that
   the DEP were significantly enriched by the regulation of the cellular
   chondrogenic differentiation and the extracellular matrix (ECM)
   secretion, with the cellular components and molecular functions
   primarily involving extracellular proteins and RNA binding
   (Figure [113]2F). Moreover, KEGG pathway enrichment analysis suggested
   that the wingless (Wnt) and cyclic adenosine monophosphate (cAMP)
   signaling pathways were the most DEP‐related (Figure [114]2G). We
   performed ELISA to validate the growth differentiation factor 5 (GDF5),
   SOX9, proteoglycan 4 (PRG4), and POU class 5 homeobox 1 (OCT‐4) protein
   contents, which were related to the stem cell fate; the quantitative
   results were consistent with the GO‐analysis expression trend
   (Figure [115]2H). The above results indicate successful isolation of IE
   and CIE, and that the CIE obtained after chondrogenic differentiation
   was rich in GDF5, forkhead box O1 (FOXO1), and SOX9, which have
   potential cartilage matrix synthesis functionality.

Figure 2.

   Figure 2
   [116]Open in a new tab

   CIE characterization. A) COL2 fluorescence staining of IPFP‐MSC and
   C‐IPFP‐MSC chondrospheres over 14 d; scale bar: 1 mm. B) EV
   morphologies observed via TEM; scale bar: 50 nm. C) EV particle size
   distribution measured via NTA. D) Western blotting of EV markers. E)
   DEP volcano map between IE and CIE; p < 0.05; log[2] fold change. F,G)
   GO enrichment and KEG pathway analyses, respectively, of DEP between IE
   and CIE. H) Chondrogenesis‐related DEP quantified via ELISA. ^**: p <
   0.01, ^*: p < 0.05.

   To further clarify the CIE ability to promote cartilage matrix
   synthesis, we co‐cultured CIE cells with chondrocytes (the C28/I2 cell
   line) to form chondrospheres in vitro. Four groups were considered: a
   proliferation medium, a chondrogenic differentiation medium, a
   proliferation medium containing 300 µg mL^−1 IE, and a proliferation
   medium containing 300 µg/mL CIE. Immunohistochemical staining revealed
   that the CIE‐induced chondrospheres had the highest expression of
   aggrecan (ACAN), COL2, and SOX9 proteins among all groups, surpassing
   the chondrogenically differentiated group (Figure [117]3A,B).
   Similarly, regarding the mRNA expression levels, the highest ACAN,
   COL2A1, and SOX9 expression levels were obtained for the CIE group,
   along with the lowest levels of the hypertrophic cell marker COL10A1
   (Figure [118]3C). This outcome suggests that CIE not only delivers SOX9
   protein to chondrocytes but also promotes SOX9 mRNA self‐expression in
   target cells. We further detected SOX5 and SOX6, which functionally
   interacted with the SOX9 and were highly expressed in the CIE group
   (Figure [119]3D). Following co‐cultivation with CIE, no significant
   change in the SOX9 downstream target GLI family zinc finger 1 (GLI1)
   was obtained. Another downstream target related to osteogenic
   differentiation, RUNX family transcription factor 2 (RUNX2), decreased
   significantly. To clarify the SOX9 cellular self‐expression pathway,
   the following upstream transcription factors were detected: RELA
   proto‐oncogene (RELA), hypoxia‐inducible factor 1 subunit alpha
   (HIF1A), the recombination signal binding protein for the
   immunoglobulin Kappa J region (RBPJ), and FOXO3A. Only the mRNA
   expression of FOXO3A was significantly enhanced (Figure [120]3E). This
   result suggests that FOXO1, carried by the CIE, activated SOX9
   expression in the target cells through the FOXO1–FOXO3A/SOX9 pathway.
   Under the effects of direct SOX9 delivery and stimulation of endogenous
   SOX9 expression through FOXO1 delivery, the CIE promoted chondrocyte
   synthesis in the ECM (Figure [121]3F).

Figure 3.

   Figure 3
   [122]Open in a new tab

   CIE concentration screening and cartilage matrix‐synthesis promotion
   mechanism. Groups: Control: proliferation medium; Positive:
   chondrogenic differentiation medium; IE: proliferation medium
   containing 300 µg mL^−1 IE; CIE: proliferation medium containing
   300 µg mL^−1 CIE. A) Immunohistochemical staining of ACAN, COL2, and
   SOX9 at 21 d; Magnification: 50× (left) and 200× (right); scale bar:
   50 µm. B) Proportions of cells with positive immunohistochemical
   staining. C) ACAN, COL2A1, SOX9, and COL10A1 mRNA expression levels. D)
   SOX9‐related gene expression levels: SOX5, SOX6, GLI1, and RUNX2. E)
   SOX9 upstream transcription‐factor expression levels: RELA, HIF1A,
   RBPJ, and FOXO3A. F) Schematic diagram of CIE‐induced cartilage
   matrix‐synthesis promotion mechanism. For C28/I2 treated with various
   CIE concentrations for 48 h: G) COL2 fluorescence staining samples;
   scale: 100 µm; H) COL2^+ cell percentages following COL2 fluorescent
   staining; and I) ACAN, COL2A1, SOX9, and COL10A1 mRNA expression levels
   via RT‐qPCR. ^***: p < 0.001, ^**: p < 0.01, ^*: p < 0.05, ns: p >
   0.05.

   We investigated the CIE chondrocyte internalization efficiency of
   chondrocytes via fluorescence imaging, obtaining absorption peaks at
   48 h (Figure [123]S5, Supporting Information). Therefore, we
   co‐cultured different CIE concentrations with C28/I2 cells for 48 h and
   screened for the optimal CIE concentration. Comparable induction levels
   were obtained for 200 and 300 µg mL^−1 CIE concentrations,
   significantly elevating the expression of COL2, a cartilage marker
   (Figure [124]3G,H). This finding was confirmed using the quantitative
   polymerase chain reaction with reverse transcription (RT‐qPCR) method
   (Figure [125]3I).

2.3. Characteristics of MMP13‐Sensitive Hydrogel

   An MMP13‐sensitive hydrogel was used as an RIE carrier, based on the
   insertion of the MMP13 substrate sequence into a KLDL‐12 self‐assembled
   peptide. The molecular weight of KLDL‐MMP13 (KM13) is 2.71 kDa, and its
   theoretical isoelectric point is 7.01. Rheological tests demonstrated a
   gelation time of 5 min for 1% (w/v) KM13 (Figure [126]4A). In a
   gel‐formation experiment, KM13 formed a solid hydrogel block after
   20 min (Figure [127]4B). The above characteristics of KM13 indicate
   that it can spontaneously but slowly change from a solution to a
   hydrogel. Scanning electron microscope (SEM) images confirmed that the
   KM13 hydrogel had a 3D porous structure (Figure [128]4C).

Figure 4.

   Figure 4
   [129]Open in a new tab

   MMP13‐sensitive hydrogel (KM13) characterization. A) Gelation profile
   of 1% (w/v) KM13 over 60 min. B) Image of 1% (w/v) KM13 dissolved in
   DPBS for 5 and 20 min at room temperature; scale bar: 2 cm. C) SEM
   micrographs of 1% (w/v) KM13 hydrogel; scale bar: 1 µm. D) TEM and E)
   AFM micrographs of 1% (w/v) KM13 nanofibers before and after digestion
   by MMP13; scale bars: 100 and 400 nm, respectively. F) 1% (w/v)
   KM13‐nanofiber lengths before and after digestion by MMP13. G) Storage
   and loss moduli (G’ and G’’, respectively) of KM13 before and after
   digestion by MMP13 for angular frequencies of 0.1–10 rad/s (1% strain).
   (H) COF of 1% (w/v) KM13 hydrogel before and after digestion by MMP13.
   BCA assays of KM13E containing different RIE concentrations (0, 0.5, 1,
   1.5, 2 mg mL^−1) in (I) SBF solution, (J) MMP13 solution (SBF
   containing 5 ng mL^−1 MMP13), and K) SBF solution (with MMP13 addition
   at 6 d, to a final concentration of 5 ng mL^−1), over 18 d. (L)
   Schematic diagram of KM13E releasing RIE in response to MMP13. ^**: p <
   0.01, ns: p > 0.05.

   We observed the enzymatic response degradation of KM13 via TEM and
   atomic force microscopy (AFM), in a simulated body fluid (SBF)
   solution. For 1% (w/v) KM13, interwoven nanofibers were observed. We
   measured the intra‐articular MMP13 content and activity in the
   joint‐fluid of OA rats, and used a consistent 4.78 ng mL^−1 of MMP13 to
   prepare the enzymatic solution (Figure [130]S6A,B, Supporting
   Information). In the MMP13 solution, the KM13 nanofiber structure was
   broken and loosely arranged (Figure [131]4D,E). Quantitative analysis
   revealed that the KM13 nanofiber length before contact with MMP13 was
   135.71 ± 13.64 nm; after enzymatic digestion, the nanofiber length
   decreased significantly to 74.89 ± 24.38 nm (Figure [132]4F). We also
   observed a significant decrease in the KM13‐hydrogel modulus after
   enzymatic digestion; however, their coefficient of friction (COF)
   values remained relatively unchanged (Figure [133]4G,H). Notably, these
   enzymatically digested KM13 gels exhibited modulus and COF values
   similar to those of commercially available hyaluronic acid, indicating
   that KM13 gels may be a viable option for intra‐articular injection
   (Figure [134]S6C,D, Supporting Information). In addition, the shear
   modulus of the KM13 hydrogels did not differ significantly before and
   after enzymatic digestion, demonstrating that they remained well stable
   at high MMP13 levels (Figure [135]S6E, Supporting Information). To
   confirm that KM13 could be extruded smoothly through a syringe, we also
   tested its shear‐thinning properties (Figure [136]S7, Supporting
   Information). The drug release rates of KM13 samples mixed with various
   RIE concentrations (0.5–2 mg mL^−1, labeled KM13E0.5, KM13E1, KM13E1.5,
   and KM13E2, respectively), and of KM13 without RIE, were evaluated by
   quantifying the protein concentration released into the solution.
   Without MMP13, KM13E released all the RIE within 18 d (Figure [137]4I;
   Figure [138]S8A, Supporting Information). Of the various samples, the
   KM13E1.5 average release concentration was closest to 200 µg mL^−1
   (Figure [139]S8B, Supporting Information). In the presence of MMP13, a
   sudden RIE release was observed, with complete release occurring within
   9 d (Figure [140]S8C, Supporting Information). Note that KM13 also
   released protein components when digested by MMP13 (Figure [141]4J).
   Again, the KM13E1.5 average release concentration was the closest group
   to 200 µg mL^−1 (Figure [142]S8D, Supporting Information). Midway
   addition of MMP13 significantly increased the KM13E protein release
   rate (Figure [143]4K; Figure [144]S8E,F, Supporting Information). In
   DMEM medium, KM13E also released all RIE within 18 d and the release
   profile was similar to that in SBF solution without MMP13 enzyme
   (Figure [145]S9, Supporting Information). We also considered some of
   the MMPs expressed in the OA joint fluid, including MMP1, MMP3, and
   MMP9. The KM13 hydrogel were specifically responsive to MMP13 (Figure
   [146]S10, Supporting Information). These results show that the KM13
   hydrogel can be degraded by MMP13 in particular to rapidly release RIE,
   while maintaining certain structural and mechanical properties
   following enzymatic degradation (Figure [147]4L).

2.4. Droplet‐Based Microfluidics Facilitating Microsphere Formation

   Polyethylene glycol diacrylate/gelatin methacryloyl hydrogel
   (PEGDA/GelMA) microspheres were used as CIE carriers. The synthesized
   GelMA was first examined via proton nuclear magnetic resonance
   analysis; a clear characteristic peak was observed (Figure [148]S11A,
   Supporting Information). The GelMA was then mixed with PEGDA to form
   gels and examined via Fourier‐transform infrared spectroscopy, which
   confirmed the successful formation of PEGDA/GelMA (PG) hydrogels
   (Figure [149]S11B, Supporting Information). The storage and loss moduli
   (G’ and G’’, respectively) of the PG hydrogels exhibited slight
   fluctuations and low strain dependence (Figure [150]S11C, Supporting
   Information). Figure [151]5A is a schematic of the microfluidic
   focusing chip. The oil and water phase solutions intersected to form
   water droplets in the oil. Following ultraviolet curing, the materials
   in the water phase formed a stable 3D structure, thus maintaining the
   microsphere structures. SEM images showed that the PEGDA/GelMA/CIE
   (PGE) microspheres prepared using the microfluidic focusing chips had a
   consistent size and smooth surfaces (Figure [152]5B). The microsphere
   sizes were predominantly within the 70–80‐µm range, with an average
   diameter of 75.172 ± 13.682 µm (Figure [153]5C).

Figure 5.

   Figure 5
   [154]Open in a new tab

   PGE microsphere characterization. A) Schematic diagram of microfluidic
   focusing chip used for PGE microsphere preparation. B) SEM micrographs
   showing uniform spherical PGE microspheres. C) Histogram of
   PGE‐microsphere diameter distribution. BCA assays of PGE microspheres
   in (D) SBF solution, (E) MMP13 solution (SBF containing 5 ng mL^−1
   MMP13), and (F) SBF solution (with MMP13 addition at 6 d, to a final
   concentration of 5 ng mL^−1), over 18 d. G) Immunofluorescence images
   of single PGE2 microsphere (the CIE were labeled with PKH26) in SBF
   solution over 18 d; scale bar: 50 µm. J) Quantified relative
   fluorescence intensity of single PGE2 microsphere.

   To investigate the controlled release effect of the PGE microspheres in
   the MMP13 solution, we prepared microspheres with CIE concentrations of
   0.5–2 mg mL^−1 (labeled PGE0.5, PGE1, PGE1.5, and PGE2, respectively),
   and also considered PG without CIE. In the absence of MMP13, all the
   PGE groups released ≈90% of their CIE within 18 d (Figure [155]5D;
   Figure [156]S12A, Supporting Information). Among the various samples,
   the PGE2 average release concentration was closest to 200 µg mL^−1
   (Figure [157]S12B, Supporting Information). In the presence of MMP13,
   the release rates of all PGE groups increased from 50% to 70%. Here,
   the PG may have been digested by the MMP13, increasing the protein
   release concentration of the PG group (Figure [158]5E; Figure
   [159]S12C, Supporting Information). We speculate that this phenomenon
   is caused by the degradation of gelatin components in GelMA by MMP13.
   However, this sudden release effect was not as significant as that of
   KM13E (40–90% release rate). Again, the PGE2 average release
   concentration was closest to 200 µg mL^−1 (Figure [160]S12D, Supporting
   Information). Midway addition of MMP13 enhanced the PGE protein release
   rate (Figure [161]5F; Figure [162]S12E,F, Supporting Information), but
   not as significantly as for the KM13E response to MMP13. Confocal
   microscopy was used to observe the release of PKH26‐labeled CIE from
   single microsphere over 18 d. The PGE2‐group fluorescence intensity
   gradually decreased over time, indicating CIE release (Figure [163]5G).
   Quantitative analysis of the fluorescence intensity showed that the
   PGE2 retained weak fluorescence until Day 18, when the CIE was
   completely released (Figure [164]5H).

2.5. The “Microspheres‐In‐Gel” Structure Leads to Differential Release Rates
of the Two EVs

   Following the construction of RIE‐containing KM13E hydrogel and
   CIE‐containing PGE microspheres, we expected that the
   “microspheres‐in‐gel” hydrogel complex, that is, KM13E@PGE, could
   achieve different release rates for the two EVs. We performed
   quantitative analysis of IL‐10 and SOX9 to assess the released RIE and
   CIE, respectively. Based on the separately marked fluorescence of the
   two EVs, the KM13E@PGE was co‐cultured with macrophages and
   chondrocytes in vitro and injected into joint cavities in vivo. Hence,
   the differential release rates of the two EVs was examined for
   KM13E@PGE (Figure [165]6A). ELISA results showed that, regardless of
   the presence of MMP13, the IL‐10 and SOX9 concentrations in the pure
   EVs (both RIE and CIE) decreased continuously, indicating the
   possibility of EV degradation or MMP13 enzymatic hydrolysis. In terms
   of IL‐10 release, there was no significant difference between the KM13E
   and KM13E@PG groups, as the KM13E was located in the outer layer of the
   “microspheres‐in‐gel” system. However, with the involvement of MMP13,
   both the KM13E and KM13E@PG groups exhibited significantly increased
   IL‐10 releases. The difference in their performance pertained to their
   SOX9 release. The PGE and KM13@PGE groups exhibited more sustained
   release rates, which was due to the PGE encapsulation performed by the
   KM13 in the “microspheres‐in‐gel” system. In the presence of MMP13, the
   SOX9 releases of the PGE and KM13@PGE groups were slightly upregulated,
   possibly because of the enzymatic hydrolysis of the GelMA component in
   the PGE (Figure [166]6B,C). These results indicate that the
   enzyme‐sensitive hydrogel outer layer of KM13E@PGE can rapidly release
   RIE in the presence of MMP13. However, MMP13 does not significantly
   affect the CIE release rate from the KM13E@PGE system.

Figure 6.

   Figure 6
   [167]Open in a new tab

   Controlled release of KM13E@PGE in vitro and in vivo. A) Schematic
   diagram of KM13E@PGE controlled‐release detection. B,C) ELISA assay of
   KM13E@PGEs in SBF and MMP13 (SBF containing 5 ng mL^−1 MMP13)
   solutions, respectively, over 18 d. D,E) Immunofluorescence images and
   quantitative fluorescence intensity analysis results, respectively, for
   PKH26‐labeled RIE and PKH67‐labeled CIE uptake by RAW264.7 and C28/I2
   within 30 d; scale bar: 20 µm. F) In vivo imaging at 0, 7, and 14 d
   post‐injection into rat joint cavities. G,H) Quantification of
   KM13E@PGE relative fluorescence intensity in vivo, for RIE and CIE,
   respectively.

   We labeled the RIE and CIE with PKH26 and PKH67, respectively, and
   added them to the KM13E@PGE system. Direct observation of the
   fluorescence intensity in the KM13E@PGE material was not feasible
   because the dispersed PKH26‐labeled RIE was easily mis‐identified as
   background fluorescence. However, we observed changing fluorescence
   trends for the two types of EVs (Figure [168]S13A,B, Supporting
   Information). Therefore, we co‐cultured KM13E@PGE with RAW264.7 and
   C28/I2, respectively, to facilitate observation of the fluorescence
   intensities after the EVs were internalized in the cells. The results
   showed that both cell types first absorbed the PKH26‐labeled RIE. Then,
   as the fluorescence carried by the RIE gradually weakened, the PKH67
   fluorescence carried by the CIE gradually increased (Figure [169]6D).
   Quantitative analysis showed that the fluorescence peaks of RAW264.7‐
   and C28/I2‐internalized RIE appeared on Day 3. The fluorescence peak of
   the RAW264.7‐internalized CIE appeared on Day 27, whereas that for the
   C28/I2 appeared on Day 24 (Figure [170]6E). This result indicates that
   the proposed KM13E@PGE system has a programmable release function in
   which RIE is released first, followed by CIE.

   Because of the inability of in vivo imaging to distinguish between the
   two fluorescent markers, we separately labeled a single EV type with
   fluorescence in our in vivo experiments. The results showed that, for
   direct injection of RIE or CIE into the joint cavities of wild‐type
   rats, the EVs could not be maintained for 14 d. Moreover, the
   fluorescence had decreased by more than 50% on Day 7. In OA rats, the
   fluorescence nearly disappeared on Day 7 (Figure [171]6F–H). When KM13E
   or PGE were injected into wild‐type rats, the EV fluorescence persisted
   for up to 14 d. However, in the joints of the OA rats, less than 50% of
   the fluorescence intensity of KM13E and PGE remained at 7 d, and had
   almost disappeared at 14 d (Figure [172]6F–H). For KM13E@PGE, the
   fluorescence in the wild‐type rats was maintained for 14 d; at that
   time, the RIE and CIE fluorescence had decreased by ≈50% and ≈30%,
   respectively. However, in OA rats, the RIE fluorescence had decreased
   by 60% by Day 7 and had disappeared completely by Day 14. The CIE
   fluorescence was maintained for 14 d, with the fluorescence decreasing
   by ≈40%; however, there was no significant difference in the trend
   between the OA rats and wild‐type rats (Figure [173]6F,H). This outcome
   further demonstrates the sensitivity of KM13E to OA biochemical
   indicators, and that the KM13E@PGE system quickly released RIE within 7
   d and slowly released CIE for more than 14 d. In summary, the
   MMP13‐sensitive characteristics and “microspheres‐in‐gel” structure of
   KM13E@PGE contribute to the different release rates of RIE and CIE.
   Additionally, the homologous targeting effect of EVs and the relatively
   active immune cells during the early stages of injection collectively
   result in a spatiotemporal controlled release effect of dual EVs
   (Figure [174]S14, Supporting Information).

2.6. Anti‐Inflammatory Effect of KM13E@PGE In Vivo

   Plantar pain threshold analysis was performed for four groups, which
   received the following treatment: saline injection without modeling
   (Blank group); monosodium iodoacetate (MIA) injection modeling for 2 w
   (OA group); injection of KM13E hydrogel containing 1.5 mg mL^−1 RIE in
   OA rats (KM13E group); injection of PGE microspheres containing
   2 mg mL^−1 CIE in OA rats (PGE group); and injection of KM13E@PGE
   containing 1.5 mg mL^−1 RIE and 2 mg mL^−1 CIE in OA rats (KM13E@PGE
   group). The results reveal a pain‐threshold decrease for all treatment
   groups by 2 w postoperatively, with the KM13E@PGE group showing the
   smallest decrease. By 8 w postoperatively, the plantar pain threshold
   of the KM13E@PGE group had returned to a level similar to that of the
   control group (Blank), which was significantly different from those of
   the other groups. Notably, differences were apparent in the plantar
   pain thresholds obtained for the OA group and the KM13E and KM13E@PGE
   groups at 4 w; however, there was no significant difference between the
   OA‐ and PGE‐group results (Figure [175]7A). Furthermore, we tried to
   treat with direct injection of CIE at different treatment times, and
   all groups were injected with RIE on the first day (Figure [176]S15A,
   Supporting Information); the results showed that all groups had a
   decrease in plantar pain thresholds at 2 w postoperatively, which was
   similar to that of the KM13E@PGE group (Figure [177]S15B, Supporting
   Information). However, at 8 w postoperatively, the plantar pain
   threshold could not recover to a level similar to that of the Blank
   group, which was slightly different from that of the KM13E@PGE group.
   Quantitative analysis of the OA biochemical indicators in the joint
   cavity indicated considerably higher levels of MMP13, TNF‐α, and IL‐1β
   for the OA and PGE groups at 2 w relative to those for the KM13E and
   KM13E@PGE groups. However, by 4 w, the MMP13 levels had decreased
   significantly with no discernible differences between the treatment
   groups; in contrast, the TNF‐α and IL‐1β levels exhibited slight upward
   trends. Furthermore, for the KM13E and KM13E@PGE groups, the IL‐10
   levels were consistently elevated at 2 and 4 w postoperatively, with
   significant differences compared to the other groups (Figure [178]7B).
   In contrast, the levels of MMP13, TNF‐α, and IL‐1β in the groups
   injected directly with RIE and CIE were significantly higher than those
   in the KM13E@PGE group at 2 w and 4 w postoperatively, and the increase
   in IL‐10 levels was not as great as in the KM13E@PGE group (Figure
   [179]S15C, Supporting Information). Direct injection of RIE and CIE did
   not significantly improve inflammatory factors, but delaying the
   injection interval of CIE to 14 or 21 d can improve pain to some
   extent. It can be seen that there is a need for a time interval between
   anti‐inflammatory treatment and cartilage repair treatment, and the
   sustained release effect of delivery materials can be more beneficial
   for the long‐term treatment of OA.

Figure 7.

   Figure 7
   [180]Open in a new tab

   KM13E@PGE‐induced improvement of the joint‐cavity inflammatory
   microenvironment in OA rats. Groups: Blank: saline injection without
   modeling; OA: MIA injection modeling for 2 w; KM13E: KM13E (containing
   1.5 mg mL^−1 RIE) hydrogel injection; PGE: PGE (containing 2 mg mL^−1
   CIE) microsphere injection, KM13E@PGE: KM13E@PGE (containing
   1.5 mg mL^−1 RIE and 2 mg mL^−1 CIE) injection. A) Plantar pain
   threshold analysis results, before modeling, 2 and 4 w after modeling.
   B) ELISA detection of IL‐10, MMP13, TNF‐α, and IL‐1β concentrations in
   joint fluid. C) H&E staining of articular cartilage tissue at 2 and 4
   w; scale bar: 200 µm. The yellow arrows indicate cartilage traps that
   have lost chondrocytes; the yellow dotted lines distinguish cartilage
   layers. D) Chondrocyte viability counted via H&E staining. E) Safranin
   O/fast‐green staining of articular cartilage tissue at 2 and 4 w; scale
   bar: 1 mm. F) ORASI score statistics based on H&E staining and Safranin
   O/fast‐green staining. G) TNF‐α fluorescent staining of articular
   cartilage tissue at 2 and 4 w; scale bar: 200 µm. H) Quantification of
   TNF‐α relative fluorescence intensity. I) CD86/CD206 fluorescent
   staining of synovial tissue at 2 and 4 w; scale bar: 200 µm. J)
   CD206^+‐cell proportions following CD86/CD206 fluorescent staining.
   ^***: p < 0.001, ^**: p < 0.01, ^*: p < 0.05, ns: p > 0.05.

   H&E staining revealed numerous cartilage traps with missing
   chondrocytes on the articular surface at 2 w postoperatively for all
   groups, except the control group. However, for the KM13E@PGE group at 4
   w post‐surgery, a significant number of newly formed chondrocytes
   appeared in the cartilage cavities. The PGE had the second highest
   number of these chondrocytes and, thus, more than the KM13E and OA
   groups (Figure [181]7C). For the chondrocyte survival‐rate calculation,
   multiple views were considered. The chondrocyte survival rate was found
   to be enhanced in the KM13E, PGE, and KM13E@PGE groups at both 2 and 4
   w postoperatively, with the KM13E@PGE group having the highest rate
   (Figure [182]7D). Safranin O–fast‐green staining demonstrated varying
   degrees of cartilage ossification at 2 w for all groups, excluding the
   Blank group, with less ossification observed for the KM13E and
   KM13E@PGE groups. After 4 w, the KM13E, PGE, and KM13E@PGE groups
   retained a significant amount of cartilage tissue on their articular
   surfaces, whereas the OA group exhibited severe cartilage ossification
   (Figure [183]7E). From the OA Research Society International (OARSI)
   score, the KM13E@PGE group exhibited the best recovery, with a reduced
   score at 4 w (Figure [184]7F). Immunofluorescence analysis indicated
   that the OA and PGE groups expressed high levels of TNF‐α protein,
   whereas the KM13E and KM13E@PGE groups expressed the opposite. Note
   that TNF‐α protein is an inflammation‐related protein that can serve as
   a predictor of inflammation severity. Quantitative analysis showed that
   the inflammation degree at 2 w was considerably lower for the KM13E and
   KM13E@PGE groups than for the other groups. At 4 w, the KM13E@PGE group
   exhibited the lowest TNF‐α protein levels among the groups, and the
   most statistically significant difference (Figure [185]7G,H).
   Immunofluorescence imaging of the synovium demonstrated positive
   expression of CD206 and CD86 proteins (M1 and M2 macrophage markers,
   respectively) in all groups during OA progression (Figure [186]7I).
   Quantitative analysis revealed that, at 2 and 4 w, the KM13E and
   KM13E@PGE groups had higher proportions of M2 macrophages than the
   other groups (Figure [187]7J). Intriguingly, at 4 w, the PGE group
   showed an improvement in the M2 macrophage proportion, approaching that
   of the Blank group. However, the difference between these two groups
   remained statistically significant. These results indicate that the
   KM13E and KM13E@PGE groups effectively improved the OA‐induced
   pathological inflammation by promoting M2 macrophage polarization,
   which is attributed to the RIE. Although the CIE‐carrying PGE had a
   certain effect on cartilage regeneration, the articular cartilage
   exhibited a high level of inflammation when PGE alone was employed.

2.7. Cartilage Repair Effect of KM13E@PGE In Vivo

   Alcian blue staining showed that, at 4 w, all groups exhibited various
   degrees of cartilage regeneration, with the KM13E@PGE group showing the
   highest and most significant difference compared with the OA group
   (Figure [188]8A). The KM13E@PGE group displayed the highest
   glycosaminoglycan (GAG) expression levels at both 2 and 4 w, close to
   those of the Blank group (Figure [189]8B). Masson's trichrome staining
   was performed to determine the local‐collagen type and density. At 4 w,
   all treatment groups, except for the KM13E@PGE group, exhibited varying
   degrees of type‐I collagen infiltration. Notably, the KM13E@PGE group
   displayed the highest COL2 density (Figure [190]8C). The KM13E@PGE
   group had the highest COL2 expression levels at both 2 and 4 w, which
   were similar to those of the Blank group (Figure [191]8D).
   Immunohistochemical analysis revealed that all groups displayed
   significant ACAN protein expression, with the KM13E@PGE group
   exhibiting the highest levels and the most noticeable differences
   relative to the other groups (Figure [192]8E,F). Furthermore,
   immunofluorescence imaging indicated positive SOX9 protein expression
   in all groups during the repair phase, suggesting the involvement of
   chondrocyte differentiation (Figure [193]8G). Quantitative analysis
   revealed a significantly higher degree of chondrogenic differentiation
   for the KM13E@PGE group than the other groups (Figure [194]8H).
   Notably, the PGE group displayed the second highest level of
   differentiation, albeit still lower than that of the Blank group; this
   level was not significantly different to that of the Blank group at 4
   w. These results indicate that the KM13E, PGE, and KM13E@PGE groups
   promoted cartilage repair, with the latter being the most effective.
   Therefore, anti‐inflammatory action and tissue regeneration are equally
   important for OA treatment, and the order of these treatments is even
   more important (Figure [195]S16, Supporting Information).

Figure 8.

   Figure 8
   [196]Open in a new tab

   KM13E@PGE promotion of articular cartilage repair in OA rats. A) Alcian
   blue staining of articular cartilage tissue at 2 and 4 w; scale bar:
   200 µm. B) Relative GAG levels are based on Alcian blue staining
   statistics. C) Masson's staining of articular cartilage tissue at 2 and
   4 w; scale bar: 200 µm. D) Collagen volume fractions from Masson's
   staining. E) ACAN staining of articular cartilage tissue at 2 and 4 w;
   scale bar; 200 µm. G) Positive areas of ACAN staining. H) SOX9
   fluorescent staining of articular cartilage tissue at 2 and 4 w; scale
   bar: 200 µm. I) SOX9^+‐cell proportions following SOX9 fluorescent
   staining. ^**: p < 0.01, ^*: p < 0.05, ns: p > 0.05.

3. Discussion

   Compared to stem cells, EVs can be cryopreserved, have lower
   immunogenicity, and do not cause tumorigenesis.^[ [197]^13 , [198]^14
   ^] EVs also have the ability to regulate target cell functions, showing
   great potential as biological macromolecule medicines.^[ [199]^13 ,
   [200]^14 ^] Several studies have reported that MSC‐, probiotic‐, and
   tumor‐derived EVs induce macrophage M2 polarization.^[ [201]^29 ,
   [202]^30 , [203]^31 , [204]^32 , [205]^33 ^] The present study aimed to
   enrich the M2‐polarizing stimulus, IL‐10, in EVs. Previously, Tang
   et al.^[ [206]^17 ^] used a similar genetic engineering approach to
   modify the parental cells, thereby achieving IL‐10^+ EVs that regulated
   the immune responses during the treatment of ischemic acute kidney
   injury. In this study, we obtained IL‐10^+ EVs (RIE) without
   dexamethasone, which reduced the time for engineering treatment of RIE
   and removed the potential effects of dexamethasone, but focused more on
   the effects of IL‐10. We also found that RIE is rich in CCL18, which
   promotes the M2 polarization of macrophages through the PI3K‐Akt
   pathway.^[ [207]^34 ^] This pathway is critical for the ATP/adenosine
   diphosphate (ADP) response and may contribute to the promotion of
   aerobic glucose metabolism in macrophages, thereby facilitating their
   polarization toward the M2 phenotype.^[ [208]^35 ^] Upon detection of
   several integrins, we found that the α4 integrin expression in the RIE
   was enhanced after the IL‐10 overexpression. The α4 integrin may
   interact with the β1 integrin expressed by the macrophages, thus
   promoting targeted action.^[ [209]^36 ^] However, the RIE expression of
   αmβ2, which is related to phagocytosis, did not change significantly.^[
   [210]^37 ^] This indicates that the RIE is more prone to macrophage
   internalization but the possibility of phagocytosis and degradation
   exists. This is important because macrophages are the immune cells that
   respond most rapidly to exogenous EVs in the innate immune system,^[
   [211]^38 ^] and also key for us to ensure that RIE takes effect in the
   joint cavity. Note that we screened the in vitro effective RIE
   concentration and the KM13E release rate to ensure accurate dosing and
   high RIE utilization.

   The IPFP tissue plays a role in promoting inflammation and angiogenesis
   during the development of OA.^[ [212]^39 , [213]^40 ^] However, the EVs
   secreted by IPFP‐MSCs possess potential therapeutic effects.^[ [214]^41
   , [215]^42 ^] These dual effects prompted us to adopt an engineered
   strategy involving chondrogenic induction cultures to induce EVs. Wu
   et al.^[ [216]^20 ^] found that EVs derived from IPFP‐MSCs are rich in
   miR‐100‐5p, which can downregulate the mammalian target of rapamycin
   (mTOR) pathway to protect articular cartilage. These EVs also inhibit
   fibroblast proliferation by suppressing metallothionein‐2A in an
   inflammatory environment. In this study, we found that FOXO1, GDF5, and
   SOX9 were enriched in the CIE through proteomics. In chondrocytes, CIE
   promotes COL2 and ACAN secretion, which improves the chondrosphere
   cohesion. This process is caused by exogenous SOX9 delivery from the
   CIE and, also, because CIE delivers FOXO1 to chondrocytes exogenously,
   specifically binds to FOXO3A, and promotes the transcription of
   endogenous SOX9. This process suppresses fatty acid oxidation and
   induces chondrogenic commitment to a low‐fat environment.^[ [217]^43 ^]
   Considering the tendency of IPFP‐MSCs to undergo adipogenic
   differentiation, this mechanism further confirms the necessity for
   chondrogenic differentiation of IPFP‐MSCs. As for the RIE, the effects
   of different CIE concentrations were screened in vitro in this study.
   As to whether CIE has a targeting effect on chondrocytes, we have not
   been able to study this in‐depth, but based on the homologous targeting
   theory,^[ [218]^44 ^] this possibility exists.

   At the onset of OA, inflammatory factors such as IL‐1β, IL‐7, and TNF‐α
   stimulate proteinase production, including MMPs and a disintegrin and
   metalloproteinase with thrombospondin motifs (ADAMTS).^[ [219]^27 ,
   [220]^45 ^] MMP‐1, −2, −3, −9, and −13 are expressed in the OA
   pathological environment.^[ [221]^27 ^] MMP13 degrades proteoglycans,
   inducing a dual degradation effect on the cartilage matrix.^[ [222]^46
   ^] Materials that stimulus‐responsed to MMPs biomarker have attracted
   attention, such as MMP2‐sensitive drug delivery materials used in renal
   ischemia/reperfusion injury.^[ [223]^47 ^] In our previous study, we
   presented an MMP1 enzyme‐responsive hydrogel that responds to
   angiogenesis during bone repair.^[ [224]^28 ^] Previously, Kisiday
   et al.^[ [225]^48 ^] compared the KLDL‐12 SAP system with the RADA‐16
   SAP system and revealed effective mimicry of the natural ECM by the
   former, which facilitated cartilage repair. Building upon this finding,
   we focused on developing a KM13 hydrogel, which is an MMP13 degradable
   gel based on a similar KLDL‐12 SAP system. We integrated the MMP13
   substrate sequence proposed by Hubbell et al.^[ [226]^49 ^] into this
   system, thereby creating a KM13 short peptide sequence. Our KM13
   hydrogel demonstrated the capacity for spontaneous gelation within a
   rapid 5 min timeframe. Moreover, its storage and loss moduli and COF
   closely resembled those of sodium hyaluronate, a commercially available
   injectable drug widely used in joint cavity treatment. Through in vivo
   and in vitro experiments, we confirmed the ability of this material to
   serve as the outermost release structure for programmed‐release
   materials, enabling EV release in accordance with the degree of
   inflammation in the OA joint cavity.

   The differential release rate realized by the integrated delivery of
   the two EV types has not been achieved in previous studies. To
   sustainably release the CIE after the initial RIE burst release,
   microspheres were incorporated as physical structural systems into the
   stimuli‐responsive system. We observed differential release rates of
   fluorescence and biochemical markers both in vitro and in vivo. This
   “microspheres‐in‐gel” system enables step‐by‐step treatment, which is
   not only recommended in clinical therapy, but has also been explored by
   Mitra et al.^[ [227]^50 ^] and Leung et al.^[ [228]^51 ^] Although the
   direct construction of bilayer‐structured microspheres can achieve
   step‐by‐step release to some extent, in this study, we did not use an
   MMP13‐sensitive sequence‐carrying SAP to prepare microspheres via a
   microfluidic chip. This is because an SAP with an MMP13‐sensitive
   sequence is electroneutral, which causes faster gel formation. Thus, we
   previously investigated the “microspheres‐in‐gel” system.^[ [229]^52 ^]
   Note that the GelMA component of PGE may serve as a substrate for
   collagenase MMP13; therefore, PGE exhibits a less‐intense sudden
   release in an MMP13 solution. However, KM13E@PGE extended the CIE
   release cycle in this study, especially in the in vivo experiments,
   which confirmed the role of the “microspheres‐in‐gel” physical
   structure in prolonging the CIE release. Further, the microspheres in
   the KM13E@PGE composite system are smaller than the diameter of a
   syringe needle, and the outer layer of KM13 has shear‐thinning
   properties, so the whole system can be injected. Based on the fact that
   KLDL‐12 has self‐healing properties,^[ [230]^53 ^] we speculate that
   the KM13 has a slow self‐healing property after injection, so the
   mechanical properties and structural arrangements of the hydrogel
   network will not be particularly affected, and the release
   characteristics of the EVs will not be altered.

   The M2 macrophages showed increased phagocytic activity upon RIE
   regulation, potentially hindering EV retention. Therefore, CIE needs to
   be released later on. Controlled drug release using stimuli‐responsive
   materials is essential for targeted and timed delivery, especially with
   the advent of multidrug protocols.^[ [231]^54 , [232]^55 ^] Zhang
   et al.^[ [233]^56 ^] introduced an innovative and highly programmable
   platform for sequential drug release in tissue engineering, which
   enables the controlled release of more than three proteins in a
   specific sequence. The present study constitutes significant progress,
   this is not only achieved by the MMP13 stimuli‐responsive of KM13E@PGE,
   but also the “microspheres‐in‐gel” structure ensures that the two types
   of EVs have different release rates. In addition, EVs exert different
   therapeutic effects on macrophages and chondrocytes, constituting a
   “temporal AND spatial” logic‐gate function, collectively achieving
   spatiotemporally controlled release. This approach distinguishes the
   proposed method from the current practice of simply mixing multiple EVs
   during treatment, as demonstrated by Wu et al.^[ [234]^57 ^]

   To establish an OA model, we employed the joint cavity injection method
   using MIA, which caused significant mobility impairment, cartilage
   degradation, and the development of an inflammatory environment within
   the joint cavities of rats.^[ [235]^58 ^] Compared with surgical
   excision of the anterior cruciate ligament, our MIA‐induced OA model
   yielded a more pronounced alteration in the inflammatory environment
   within the joint cavity. We used a method of injecting RIE and CIE at
   different intervals to simulate sequential anti‐inflammatory and
   cartilage repair treatments. We found that the overlap of the two
   treatment times was not conducive to the improvement of validation
   indicators and pain. When the injection interval between RIE and CIE
   was more than 2 weeks, the pain improved, but due to the lack of a
   sustained‐release system for EVs, the therapeutic effect was not
   satisfactory. KM13E, which was developed by the authors, exhibited
   significant anti‐inflammatory properties by enhancing the M2
   polarization of the macrophages and reducing the inflammatory‐factor
   levels. However, despite the lack of pro‐chondrogenic differentiation
   factors, the effectiveness of KM13E for cartilage repair is limited,
   which can be attributed to the repair effects of the M2 macrophages and
   KLDL‐12 on tissue. The persistent inflammatory microenvironment further
   deteriorates the cartilage ECM, contributing to OA progression.^[
   [236]^59 ^] Conversely, KM13E@PGE exhibits a comprehensive therapeutic
   effect, as evidenced by its OARSI and synovial inflammation scores,
   which were the lowest, as well as a notable increase in the cartilage
   repair indicators, including ACAN and SOX9. Additionally, although the
   in vitro and in vivo experiments revealed that KM13E@PGE releases RIE
   almost completely within 7 d in response to MMP13, the TNF‐α level at 4
   w was lower than at 2 w, which was not obvious for the KM13E group. We
   believe a feedback regulation mechanism exists; that is, the reduction
   in inflammatory factors synthesized by the CIE‐targeted chondrocytes
   consolidates the low‐inflammatory microenvironment. Several studies
   have investigated the potential of EVs in terms of their
   anti‐inflammatory and cartilage‐repair properties. Wei et al.^[
   [237]^60 ^] examined the use of MSC‐derived exosomes to alleviate
   temporomandibular joint OA by reducing inflammation. Similarly, Lin
   et al.^[ [238]^61 ^] developed MSC‐derived exosomes that promoted
   chondrocyte proliferation via the lncRNA‐KLF3‐AS1/miR‐206/GIT1 axis.
   However, none of these studies comprehensively addressed the dual
   requirements of anti‐inflammatory and reparative strategies in clinical
   OA therapy.

   Based on the excellent OA efficacy of KGM13E@PGE, we have further
   demonstrated that the “anti‐inflammatory first AND cartilage repair
   second” logic‐gate treatment strategy is necessary. Currently,
   achieving both anti‐inflammatory and reparative effects using EVs
   remains a challenge, which necessitates the development of drug
   delivery materials with an affinity for EVs and an intelligent response
   for controlled release.^[ [239]^62 ^] In addition, these materials must
   be injectable and minimally invasive in certain cases. Compared to
   clinical treatment methods for OA, we believe that a single
   intra‐articular injection of KM13E@PGE can be expected to have a
   sustained‐release effect lasting up to one month, potentially replacing
   a course of treatment with four injections of sodium hyaluronate. This
   work has used MMP13 concentration which is consistent with the animal
   model. This concentration may vary from that of patients. We believe
   that by adjusting the concentration of KM13, we can meet the release
   rate required for clinical treatment. Our KM13E@PGE delivery system not
   only effectively facilitates therapeutic EV application, but also
   serves as a potential mediator for integrated step‐by‐step therapies
   involving drug and stem cell delivery. As a flexible and
   spatiotemporally controlled delivery system for multiple EVs, with
   minor modifications, this approach can also serve as an effective
   therapeutic tool for diseases requiring combination drug scenarios,
   such as novel coronavirus pneumonia. In this work, we focused on the
   effects of EVs on macrophages and chondrocytes, however, other types of
   immune cells and tissue cells could not be detected in their entirety,
   and we believe that the wide range of regulatory effects of EVs may
   have a more diverse impact on OA treatment, which will be gradually
   refined in our subsequent work.

4. Conclusion

   In this study, we have developed a logic‐based system involving two
   engineered EVs for intra‐articular injection for OA treatment. Early
   release of IL‐10^+ EVs, which acted as an anti‐inflammatory agent, was
   achieved, followed by later release of SOX9^+ EVs, which had
   cartilage‐repair effects. This integrated in vivo delivery effectively
   enabled step‐by‐step therapy of OA. To achieve differential release
   rates of the two EV types in the joint cavity, an MMP13‐sensitive
   hydrogel and related “microspheres‐in‐gel” structure were used for
   spatiotemporally controlled EVs release. This work may inspire
   multi‐targeted, integrated applications of multiple EVs, providing
   informative approaches to differential rate release and
   enzyme‐sensitive controlled release.

Conflict of Interest

   The authors declare no conflict of interest.

Supporting information

   Supporting Information
   [240]ADVS-11-2403227-s001.pdf^ (1.7MB, pdf)

Acknowledgements