Targeted and responsive biomaterials in osteoarthritis

Osteoarthritis (OA) is a degenerative disease characterized by loss of articular cartilage and chronic inflammation, involving multiple cellular dysfunctions and tissue lesions. The non-vascular environment and dense cartilage matrix in the joints tend to block drug penetration, resulting in low drug bioavailability. There is a desire to develop safer and more effective OA therapies to meet the challenges of an aging world population in the future. Biomaterials have achieved satisfactory results in improving drug targeting, prolonging the duration of action, and achieving precision therapy. This article reviews the current basic understanding of the pathological mechanisms and clinical treatment dilemmas of OA, summarizes and discusses the advances for different kinds of targeted and responsive biomaterials in OA, seeking to provide new perspectives for the treatment of OA. Subsequently, limitations and challenges in clinical translation and biosafety are analyzed to guide the development of future therapeutic strategies for OA. As the need for precision medicine rises over time, emerging multifunctional biomaterials based on tissue targeting and controlled release will become an irreplaceable part of OA management.


Introduction
Osteoarthritis (OA) is one of the most common chronic degenerative joint diseases worldwide, often occurring in middle-aged and older adults over the age of 45 [1]. About 10% of men and 18% of women over 60 years old worldwide suffer from OA, and there is a gradual trend toward younger patients with OA due to the effects of obesity and strenuous exercise [2]. The financial cost of an OA patient is estimated to be between $700 and $15,600 a year, placing a significant burden on both society and the individual [3]. The pathological changes of OA mainly include wear and tear degeneration of articular cartilage, formation of bone fragments, synovial inflammation and subchondral bone remodeling [4]. As the specific pathogenesis of OA is still unknown, the current clinical treatment of OA is mainly focused on relieving joint pain and delaying the development of OA [5][6][7]. Currently, drug therapy, physical therapy and surgery are the primary treatments [8,9]. In the early stages of OA, pain relief and reduction of inflammation are achieved through appropriate exercise and weight loss and are supplemented with oral non-steroidal anti-inflammatory drugs (NSAIDs) [10]. For patients with advanced OA, accompanied by severe articular cartilage wear, short-term symptomatic relief can be provided by intra-articular injections of hyaluronic acid (H) or glucocorticoids [11,12]. Due to the unique physiology of the joint cavity and the presence of synovial fluid, many drugs are characterized by poor water solubility, low cellular uptake, premature release or degradation. Targeted delivery and nonspecific release of drugs remain significant limitation to drug efficacy [13]. A well-designed drug delivery system should have the appropriate size, biological barrier permeability, and proper drug release capacity [14][15][16]. Therefore, many Ivyspring International Publisher researchers hope to achieve OA treatment by developing novel and efficient drug carriers and delivery systems.
In recent times, many biomaterial carriers with cartilage or synovial targeting and responsiveness are emerging. Biomaterial carriers not only prolong the release of drug, but also exhibit higher tissue and cell specificity [17,18]. Among the various types of biomaterials, targeted and responsive biomaterials are more suitable for OA applications due to their small size and localizable delivery. They are often designed and assembled into functional structures to ensure accurate drug delivery [19][20][21]. Currently, researchers developed various dual-functional biomaterials for OA treatment, such as cartilage-targeted combined enzyme/pH/ROS/NIR response [22][23][24][25], macrophage-targeted combined enzyme/ultrasound/pH response [26][27][28], and endothelial cell-targeted combined ultrasound response [29]. Compared to passive cellular endocytosis, the introduction of chondrocyte-targeting peptides allows more efficient entry of responsive biomaterials into chondrocytes [22]. Furthermore, Deng et al. designed a bifunctional biomaterial to target subchondral bone osteoclasts and synovial macrophages via RGD-αvβ3 integrin interactions [26]. Subsequently, these biomaterials induced apoptosis by releasing celastrol in response to local MMP-9 [26]. Pro-inflammatory cell infiltration was found to be reduced and bone erosion improved in rats with advanced arthritis [26]. Researchers have assembled engineered modified polymers into nanocarriers or hydrogels containing hydrophilic or hydrophobic regions loaded with drugs for targeted drug delivery or responsive drug release [13,30]. Targeted drug delivery is mainly divided into two types: passive targeting and active targeting. The former increases the local drug concentration by physical methods such as local injection, particle size reduction, and surface charge change, while the latter enhances targeted drug delivery by chemical methods such as modification of specific antibodies or affinity peptides on the carrier surface [31][32][33]. In contrast, responsive biomaterials are usually stimulated by enzymes, reactive oxygen species (ROS), pH or external stimuli at the lesion site to release drugs, thus dramatically preventing the liberation and uptake of drugs in non-lesioned tissues [34].
In this review, we briefly overview the underlying physiopathology of OA and the currently used clinical treatments and their dilemmas. Then, we describe and summarize the characteristics of different types of targeted and responsive biomaterials and their research progress in OA. Finally, we discuss the potential limitations and challenges of biomaterials in the translation of clinical treatments for OA. We hope to provide further references and new perspectives for the future development of targeted and responsive biomaterials in OA therapy.

Physiological properties of articular cartilage
The anatomy of articular cartilage is divided into hyaline cartilage, tidemark, and calcified cartilage [35]. Articular cartilage transmits mechanical stress and aids in joint motion by providing an elastic and lubricated surface overlying the articular bone [36]. It is a layer of avascular, nerve-free connective tissue composed of chondrocytes and extracellular matrix (ECM) [35,37]. Chondrocytes regulate the composition of the ECM through their own anabolism and catabolism, which constitute the most basic cellular units of cartilage homeostasis [38]. ECM is composed of fibrin and proteoglycan, which has a certain mechanical strength and elasticity [36]. Among them, supramolecular arrays with diameters between 25-400 nm are assembled from collagen types I, II, III, V and XI, which constitute the classical fibrous structure [39]. Type II collagen, which accounts for 80% of the total collagen dry weight, is composed of three α1(II)-chains and has more glucosyl and galactosyl residues than other collagens so that they can interact with proteoglycans [39]. The pore size between the cross-longitudinal collagen network is approximately 50-60 nm, which would allow small molecules of nutrients or drugs to disperse in the ECM and penetrate and act on the chondrocytes or subchondral bone [40]. Proteoglycans consist of a number of core proteins attached to glycosaminoglycan (GAG) chains bonded non-covalently to long chains of hyaluronic acid, which give ECM its integral electronegativity, viscosity and resistance to compression [41]. The dynamic stabilization of the relative ratios of collagen fibers and proteoglycans provides the extracellular environment that supports normal metabolism, proliferation, and differentiation of chondrocytes [42]. Stimulated by mechanical stress or pro-inflammatory factors, anomalous higher expression of matrix metalloproteinases (MMPs) disrupts the dynamic balance in the ECM, which ultimately leads to chondrocyte apoptosis and articular cartilage degradation, thus further contributing to the progression of OA [43]. 45]. The latest data show that about 240 million people worldwide are afflicted by OA [2]. Among the aging population over 60 years old, women (18%) are more likely to suffer from OA than men (10%) [2]. In England, 53.23% of respondents over the age of 50 reported that they suffer from OA-related pain in one or more joints [2]. The pathological changes of OA involve the entire joint including the synovial membrane and subchondral bone ( Figure 1). First, the articular cartilage becomes softened and inelastic, and the cartilage layer velvety erupts and exposes the subchondral bone plate [4,46]. Secondly, abnormal ossification of the cartilage at the edge of the joint forms an osteoid with neurovascular invasion and causes joint pain [4]. Subsequently, vascularization at the edge of the cartilage or at the tendon attachment forms a bony mass via cartilage ossification, and this causes joint pain, stiffness, deformity and dysfunction [4]. Finally, collateral synovial inflammation and subchondral bone osteophytes will lead to macrophage infiltration and neurovascular invasion, further exacerbating the OA process [4].
At the molecular level, Interleukin-1β (IL-1β), produced by chondrocytes, osteoblasts, synoviocytes and macrophages, is considered to be the most critical inflammatory factor in the pathogenesis of OA, which can act independently or synergistically with other cytokines to cause articular cartilage degradation and joint inflammatory responses [47,48]. IL-1β increases nitric oxide (NO) and prostaglandin E2 (PGE2) expression by activating its downstream transcription factors, triggering synovial vasodilatation hyperplasia and joint cavity pain [49]. At the same time, the autocrine production of tumor necrosis factor-α (TNF-α) and Interleukin-6 (IL-6) by the above cells is stimulated, synergistically enhancing the pro-inflammatory effect of IL-1β and producing proteases such as the MMPs family and a disintegrin and metalloproteinase with thrombospondin motifs (ADAMTS) family to cleave and disassemble type II collagen and aggrecan in the ECM, thereby destroying cartilage structure [47,48,50]. Furthermore, because ADAMTS plays a critical role in overall tissue renewal, elevated ADAMTS expression is not only associated with joint degeneration and deterioration, but may also reflect the persistence of OA inflammation and injury [51]. Some scholars have also suggested that alterations in the subchondral bone microenvironment precede the occurrence of cartilage destruction, where abnormal mechanical loading and activation of TGF-β in bone-chondral crosstalk are essential factors leading to anomalous bone remodeling, angiogenesis, and sensory innervation [4]. Although numerous studies are available to explain the pathogenesis of OA at the cellular and molecular levels, a substantial body of research combined with clinical data is still needed to fully understand the underlying causative factors of OA.

Clinical treatment of osteoarthritis and its dilemma
The clinical treatment of OA is divided into two main categories: medication and surgery, while medication is split into two groups: symptom relief and condition improvement [7]. NSAIDs are the most widely used drugs that diminish the production of inflammatory factors and exert analgesic effects [52]. However, NSAIDs have gastrointestinal, neurological, hematologic and allergic side effects, as well as symptomatic treatment and relief of patient suffering, but not slowing down the progression of the disease [53,54]. Another group of drugs can significantly relieve the clinical symptoms of OA is glucocorticoids, which have a potent antiinflammatory effect and can suppress the proliferation of aberrant tissues and reduce joint effusion, thus achieving a significant effect of detumescence and analgesia [55]. However, long-term use of glucocorticoids leads to bone loss and accelerates articular cartilage and subchondral bone lesions, which limits their clinical application and eliminates them as a routine treatment option [56]. HA is an integral component of the joint fluid and cartilage matrix, commonly used in intra-articular injection for the treatment of OA [57]. HA could restore the viscoelasticity of joint fluid, enhance lubrication and shock absorption cushioning, reduce joint friction, relieve pain, prevent cartilage degradation, and lower the production of MMPs, which has significant efficacy in clinical practice [12,58]. Nevertheless, joint cavity injection of HA has high demands on operation technique and operation environment, and may cause short-term complications such as joint edema and infection, so there are still plenty of questions and controversies concerning the use of HA [59]. Glucosamine and chondroitin sulfate (CS) suppress the activity of proteases and degradative enzymes, which also target articular cartilage, making them ideal drugs for improving OA [60].
In advanced OA, severe joint deformities will occur, rendering pharmacological treatment futile, and surgical treatment is the only option, most commonly with arthroscopic debridement and total joint replacement [61]. Arthroscopic debridement is indicated for mild to moderate OA and often provides a more satisfactory outcome [62]. Total joint replacement, on the other hand, is suitable for patients with severe OA and can not only significantly reduce pain, but also correct joint deformities and restore joint function, making it the last choice for all OA patients [63]. Although surgery can achieve effective treatment, it is expensive and may lead to serious postoperative complications and medical risks [64]. For example, patients undergoing total knee arthroplasty have twice or more the incidence and mortality of venous thromboembolism due to postoperative complications such as coagulation disorders, electrolyte disturbances, and pulmonary complications [64,65]. In addition, the recruitment of bone marrow cells by microfracture to form a smooth and strong repair tissue can also replace the role of cartilage [66]. The recovery from microfracture surgery is directly related to the patient's own ability to repair [66]. Significant efficacy in young, low-weight patients, but poor efficacy in older, obese patients. After surgery, patients need to recover for 12 months or more with the assistance of a stent [67,68]. With the maturation of in vitro culture techniques, autologous chondrocyte implantation (ACI) is better able to regenerate normal articular cartilage structures [68]. ACI allows for the implantation of mature and active hyaline cartilage in a single operation, while avoiding immune rejection [68]. But ACI technology is still in its infancy, and lab-cultured chondrocytes often mean higher treatment costs [69,70]. Collectively, owing to the drawbacks and limitations of existing clinical treatment strategies, there is an imminent demand for the development of effective, versatile, targeted, and responsive OA therapies.

Responsive biomaterials in osteoarthritis
Currently available anti-inflammatory and analgesic drugs, either orally or by intra-articular injection, suffer from poor water solubility, low cellular uptake, poor drug concentration distribution or premature degradation [71]. By focusing on the responsiveness of biomaterials to different substrates or external stimuli, researchers are developing responsive biomaterials that can release encapsulated drugs efficiently and accurately [34,72]. These responsive biomaterials can minimize early or abrupt release behavior to address the problems of low bioavailability and low efficacy of anti-OA drugs. In this section, the advances in research on enzymes, ROS, pH responsive and external stimulus responsive biomaterials in the treatment of OA are reviewed (Table 1).

Enzyme-responsive biomaterials
Since enzymes possess a high degree of selectivity and efficiency, drug release in a variety of ways such as surface ligand activation and chemical bond breaking can be achieved by introducing functional groups sensitive to enzymes aberrantly expressed in the OA microenvironment into biomaterials [99]. With a high secretion of multiple matrix-degrading enzymes and degradationactivating enzymes in OA, enzyme-responsive biomaterials are designed to control drug release by exploiting the differences in enzyme activity in tissues [100]. Namely, when administered early in OA or prophylactically, the drug is released in low or no volume to avoid side effects, while a large volume of drug can be unleashed swiftly in advanced disease. Enzyme-responsive wise biomaterials often have enzyme-sensitive backbones or biodegradable ester bonds as linkages, resulting in efficient controlled release behavior [71].  Although enzymes have received extensive attention in studies on their application as triggers in malignancies and cardiovascular diseases, their application in OA therapy is still in its infancy so far [101]. MMPs, one of the key enzymes in the deterioration of OA, are commonly manufactured by chondrocytes under inflammatory conditions and lead to cartilage erosion by breaking down type II collagen and aggrecan in the joint matrix [43]. Out of this inspiration, Lan et al. created an MMP-13 responsive therapeutic integrated nanoplatform by modifying specific peptide substrates of MMP-13 [22]. By respective coupling type II collagen-binding peptide and Cy5.5-labeled MMP-13-responsive peptide substrate to poly(2-ethyl-2-oxazoline)poly(ε-caprolactone) (MRC-PPL), self-assembling and loading the Chinese herbal active ingredient psoralidin (PSO), the multifunctional MRC-PPL@PSO achieved the release of therapeutic drugs while being able to report OA via fluorescent signals (Figure 2A) [22]. The fluorescent signal of MRC-PPL micelles in healthy joints or MMP-13 inhibitor-treated OA joints was significantly lower than in the OA group, representing that MRC-PPL micelles were mediated by MMP-13 activity in response to the onset of OA joints ( Figure 2C) [22]. Similarly, Chen et al. achieved responsiveness to altered MMP-13 expression in OA by loading hydroxychloroquine with MMP-13responsive ferritin nanocages (CMFn@HCQ) in an almost identical manner to the former ( Figure 2B) [73]. This shows the promise of multifunctional responsive biomaterials for image-guided OA precision therapy. In addition to MMP-13, synovial inflammation and subchondral bone lesions in OA are often accompanied by MMP-2/MMP-9-mediated matrix remodeling and angiogenesis [102]. Enzyme-responsive hydrogel loaded with triamcinolone acetonide (TA) have favorable biocompatibility. When it was incubated in PBS at 37°C or in synovial fluid without MMPs for 30 days, only 20% of the drug was released [74]. The addition of MMP-2/MMP-9 or inflammatory synovial fluid resulted in a sustained release of TA over 30 days, which was reversed by the addition of MMPs inhibitors [74]. Furthermore, responsive biomaterials using osteoblasts or vascular endothelial cells highly expressing MMP-2/MMP-9 as enzymatic response targets might be a new strategy to treat OA to restore subchondral bone structure and mechanical stress disorders [103,104]. More interestingly, gelatin microspheres delivering anti-inflammatory cytokines in response to collagenase produced by inflammatory cells were able to effectively restrict IL-1β or LPS-induced NO production, which may further validate the promising application of responsive biomaterials in OA ( Figure 2D) [75].

ROS-responsive biomaterials
ROS broadly refers to oxygen-derived radicals and non-radicals, and contains superoxide anions (O 2-), hydrogen peroxide (H 2 O 2 ), hydroxyl radicals (OH·), nitric oxide radicals (NO·) and singlet oxygen ( 1 O 2 ), which have high chemical reactivity because they contain unpaired electrons [38]. ROS under physiological conditions is primarily generated by the electron transport chain of mitochondria, which in physiological concentrations facilitate immune defense, apoptosis regulation, and intracellular signal transduction [105]. After the injury occurrence, the intracellular or microenvironmental antioxidants such as glutathione (GSH) and superoxide dismutase (SOD) are insufficient to eliminate the excessive production of ROS, which will trigger intracellular metabolic disorders and biomolecular damages, eventually leading to apoptosis [105,106]. Recent studies have shown that ROS levels are significantly increased in the joints of OA patients, moreover, it has been shown that ROS clearance can be effective in relieving OA [105,107,108]. Oxidative stress generated by excessive ROS in the joint will lead to chondrocyte membrane disruption and synthesis of proteoglycan matrix [109]. ROS-induced activation of the NF-κB pathway stalled chondrocyte growth and secreted MMPs, exacerbating cartilage inflammation [105]. With the progression of aging, chondrocytes will become increasingly sensitive to ROS damage, inducing more cellular senescence and apoptosis [110]. Such evidence reveals that abnormally elevated ROS is inextricably linked to OA pathological progression, suggesting that utilizing high levels of ROS in the OA tissue microenvironment is a promising trigger for responsive drug release [110].
The mechanisms of ROS-responsive biomaterials are divided into two types: chemical bond breaking (e.g., polythioketal, polyproline, phenylboronic acid, and ester-containing polymers) and hydrophilichydrophobic conversion (e.g., polypropylene sulfide, thioether-containing polymers, monotellurium polymers, and monoselenium polymers) [111]. Zhang et al. used polythioketal (PTK) as a soft linkage, 1,4-butanediol as a chain extender, and Hexamethylene Diisocyanate (HDI) as a hard linkage to synthesize a ROS-responsive polyurethane (PTKU) with superior mechanical properties ( Figure 3A) [76]. Compared to the control, PTKU NPs underwent a dramatic morphological transformation after incubation with OH· (1000 mM), and all thioketal bonds were broken, leading to nanoparticle depolymerization and structural relaxation [76]. In vivo experiments, PTKU NPs not only demonstrated favorable articular targeting, but also boosted the shift of synovial macrophages to M2 type and downregulate the expression of MMP-2 and various inflammatory factors in articular cartilage [76]. Similarly, TKCP@Dex NPs could also be obtained by encapsulating dexamethasone (Dex) with thioketal (TK) and cartilage-targeting peptide (CAPDWRVIIP PRPSA) functionalized PEG ( Figure 3B) [24]. Under the in vitro catalysis of KO2, TKCP@Dex NPs progressively broke down the thioketone bond to release Cy5.5 and Dex to visualize the treatment, scavenging IL-6 and MMP-13 without causing cytotoxicity [24]. In the monosodium iodoacetate (MIA)-induced OA model, high concentrations of ROS in the joint facilitated aggregation of the nanoprobe and rapidly peaked within 24 h [24]. H&E and safranin O-fast green staining revealed a better recovery of the cartilage matrix than Dex injection, demonstrating the effectiveness of the vector in the organism [24]. Zhao et al. achieved high loading and controlled release of Dex by assembling L-dopamine modified with phenylboronic acid into nanoparticles (Dex-pPADN) in aqueous solution [77]. Through a cascade of redox reactions triggered by ROS at the site of inflammatory damage, Dex-pPADN formed melanin-like structures and exhibited extremely strong anti-inflammatory and antioxidant effects ( Figure 3C) [77]. Due to the structural transformation after oxidation by ROS, Dex-pPADN was available as a contrast agent to detect the progression of OA noninvasively, with luminescence intensity proportional to the severity of OA [77].
Compared with the previous study, the waterin-oil-in-water biphasic hollow microspheres (HM) composed of ethanol, ferrous chloride, Dex, and sodium bicarbonate achieved ultra-sensitive ROS responsiveness as shown in Figure 3D [78]. The low concentration of H2O2 passes through the PLAG shell and then oxidizes ethanol to acetic acid by the Fenton reaction with Fe 2+ catalysis, subsequently sodium bicarbonate decomposes in an acidic environment to produce abundant carbon dioxide (CO 2 ), which eventually leads to the collapse of HM [78]. As shown in the SEM images, HM is uniformly dispersed hollow spheres and form large pores at 50 μM H 2 O 2 /pH 6.8, making this unique structure ideal for the release of cargo [78]. Nevertheless, researchers should be mindful of the potential toxicity of these responsive biomaterials with excessive local Dex concentrations. In addition, tannic acid and tetrahydroxydiboron have been reported to form borates with ROS response and NO scavenging ability [79]. In conclusion, ROS-responsive biomaterials hold exciting promise for the treatment of inflammatory diseases not restricted to OA.

pH-responsive biomaterials
In OA lesioned joints, synovial fluid pH decreases from 7.4 to 6.0 owing to local tissue hypoxia metabolism and acidosis [112,113]. Even in different areas of the lesion, the pH of the joint tissue reflects some variability [114]. Articular cartilage (6.3) had a slightly lower pH than the meniscus (6.5) [114]. This significant pH shift in the pathological microenvironment is considered as an ideal response condition to modulate drug delivery [115,116]. Most pH-responsive nanomaterials are based on chemical bond breakage of cleavable groups or protonated dissociation of ionizable groups, leading to carrier acidolysis, representative of which are esters, Schiff bases, amides, vinyl ethers, and imines [34]. To overcome the high lipophilicity limitation of the anti-inflammatory drug andrographolide (AG), He et al. formulated a ph-responsive biomaterial (AG@MSNs-PAA) based on polyacrylic acid (PAA) [80]. Since PAA degrades in an acidic environment, the progressive release of AG over 72 h reversed IL-1β-induced chondrocyte apoptosis, showing recovery of Col2α1 and Aggrecan [80]. Beyond polymers, metal-organic frameworks (MOF) and inorganic materials also hold promising promise for the development of pH-responsive biomaterials [81,82]. Xiong et al. synthesized an acid-dissolvable biomaterial MOF@HA@PCA by HA modification and loading with protocatechuic acid (PCA) [81]. The release of cargoes in MOF@HA@PCA showed "fast followed by slow" pattern, with a 1.6-fold increase in PCA release rate in PBS at pH=5.6 relative to pH=7.4 [81]. MOF@HA@PCA demonstrated a favorable safety and potent anti-inflammatory effect, substantially downregulating the expression of inflammatory markers including MMP-3, MMP-13, COX2, and iNos [81]. More interestingly, the bisphosphonate-conjugated nanoapatite (NP-BP) exhibited distinct release behaviors at different pH [82]. NP-BP targets overactive osteoclasts in subchondral bone to reverse cartilage fibrosis and angiogenesis, reflecting a novel therapeutic strategy for OA [82].
Additionally, some strategies are based on pH-dependent hydrophobic alterations or internal gas generation, allowing for overall morphological structural changes to achieve on-demand drug release [83][84][85]. Jin et al. constructed chitosan-coated hollow mesoporous silica nanoparticles (CSL@HMSNs-CS) loaded with poorly hydrosoluble Celastrol (CSL, a pentacyclic triterpene compound) [85]. The hollow structure (HMSN) was first prepared by hydrothermal etching, followed by anchoring the chitosan to the HMSN surface by cross-linking agent to enclose the CSL in the cavity inside the sphere ( Figure 4A and 4B) [85]. When H + is present at high levels, the protonation of free NH 2 in Cs leads to increased water solubility and rupture of the protective layer, triggering pH-responsive cargo release [117]. CSL@HMSNs-CS significantly reduced IL-1β and TNF-α in chondrocyte culture supernatants by inhibiting the NF-κB signaling pathway, as well as alleviating pain behavior in OA [85]. In the MIA-induced OA model, poly(β-amino ester) (PAE) amphiphilic polymers released curcumin in a pH-sensitive manner, demonstrating superior cartilage matrix reconstitution than the equivalent curcumin ( Figure 4C) [84]. ACP micelles will aggregate more significantly in OA joints ( Figure 4D). Hu et al. synthesized a responsive biomaterial with carbonates for the treatment of OA [83]. Protons in the acidic microenvironment could penetrate the PLGA shell to produce CO2 with NH 4 HCO 3 to lead to lactate disintegration, then Rh-PLGA-NPs@NH 4 released three times as much rhein (Rh) as the control in the OA synovial fluid mimic [83]. In future applications, we must consider the possible safety hazards associated with these degradation products and improve the sensitivity of biomaterials to environmental pH changes to adapt to the needs of different environments in OA therapy.

Stimulus-responsive biomaterials
In contrast to endogenous factors like enzymes, ROS, and pH, simple and manipulable exogenous stimuli have also been used to initiate the release of loaded drugs [118,119]. Operators can modulate the release behavior of responsive biomaterials by applying light, thermal, magnetic or ultrasound to the lesion site [120]. Moreover, as a high-precision drug delivery system, stimulus-responsive biomaterials can be used to add or remove external stimuli at will, or to achieve simultaneous release of multiple sites, as required, ultimately enabling individualized OA treatment modalities (continuous or intermittent) [121]. Thus, all these strengths offer great possibilities for the application of stimulus-responsive biomaterials in OA therapeutics.
After the onset of inflammation, clinical practice reveals swelling of the knee joint accompanied by an increase in temperature of 2-3°C [34]. Kang et al. reported a thermo-responsive nanosphere that can independently release kartogenin (KGN) and diclofenac (DCF) at different rates in response to variations in temperature [86]. The nanospheres consist of a cross-linked network of amphiphilic tri-block polyoxyethylene, pluronic F127 and chitosan loaded with KGN on the outside, wrapped with polyoxypropylene and a pluronic F127 core loaded with DCF [86]. Nanospheres not only promote chondrogenesis of hBMSC in vitro, but also rescue cartilage matrix deficiency and abnormal ossification (e.g. osteophyte, bone spurs) in vivo [86]. The inverse opal hydrogel scaffold was modified with poly(N-isopropylacrylamide) to acquire temperature responsiveness, releasing diclofenac sodium only in response to inflammation or exercise fever, which greatly enhances drug utilization [87]. But endogenous heat is not sufficient for high-precision and spatiotemporally controllable drug release, limiting the clinical translation of thermo-responsive biomaterials. Considering that near-infrared (NIR, λ=780~1100 nm) light has superior tissue penetration (5~10 mm), researchers have developed photothermal therapy (PTT) with specific wavelengths of laser as the arouser [122][123][124]. Xue et al. constructed a MOF-based mesoporous polydopamine (PDA) multifunctional targeting nanocarrier (RB@MPMW) loaded with rapamycin (Rap) and bilirubin (Br) for the first time ( Figure 5A) [25]. Whether in articular cavity or aqueous solution, RB@MPMW exhibited outstanding photothermal properties, reaching over 40°C within 3 min ( Figure 5B) [25]. Meanwhile, under NIR irradiation, Rap and Br released by RB@MPMW intensely suppressed P65 phosphorylation and mitochondrial dysfunction, suggesting the restoration of chondrogenic anabolism [25]. In an anterior cruciate ligament transection (ACLT) rat model, the RB@MPMW group had the lowest OARSI score and LC3B autophagy-positive cell count, confirming the efficacy of NIR-responsive biomaterials in OA [25]. Similarly, Dex-loaded molybdenum disulfide nanosheets diminished macrophage-derived TNF-α and IL-1β by NIR exposure that rescued joint surface erosion and bradykinesia in OA mice ( Figure 5C) [88]. Immunofluorescence showed that cartilage lesion caused by OA was reversed ( Figure 5D) [88].
Magnetic manipulation has attracted interest due to its biocompatibility and non-invasive operation [125]. It is a magnetically driven manipulation achieved by introducing magnetic nanoparticles or magnetic fields. In the presence of an alternating magnetic field, the magnetic nanoparticle transit rate in bovine articular cartilage was increased by nearly 50-fold [89]. Magnetic nanoparticles are often made of trivalent iron complexes. After chitosan modification, meloxicam achieves an encapsulation rate of 82% [90]. In addition, nanoparticles with a gold-iron-gold shell structure coupled PPT and in vivo MRI, enabling the integration of diagnosis and treatment [91]. The methotrexate loaded in the core is precisely released in the inflamed joint by an external magnetic field [91,92]. Superparamagnetic iron oxide particles integrated diagnosis and therapy under an applied magnetic field. It carries anti-inflammatory drugs to combat local inflammation while enhancing magnetic resonance imaging sensitivity [93,125]. Magnetic nanoparticles with a superparamagnetic iron oxide nanoparticle core were used for responsive delivery of active siRNA to rat joints [94]. Magnetically responsive nanoparticles improved siRNA stabilization and macrophage uptake [94]. The subsequently released siRNA inhibited IL-2/IL-15Rβ expression to alleviate joint inflammation [94]. In addition, the addition of dynamic magnetic field is more favorable for the chondrogenic differentiation of BMSC on the surface of biomaterials [95]. All of these results demonstrate the feasibility and efficiency of magnetically responsive biomaterials for OA therapy.
Ultrasound-responsive biomaterials for arthritis treatment are being developed in recent years. Liao et al. successfully delivered diclofenac sodium to rat joints by ultrasound responsiveness [96]. This transdermal delivery reduced angiogenesis and joint swelling in the inflamed area [96]. Under safe low frequency ultrasound stimulation, ultrasoundresponsive liposomes are enriched in the joint and release methotrexate to inhibit arthritis progression [29]. Ultrasound turns drug-loaded liposomes into "nanobombs" by locally triggering the rupture of the liposomes [27,97]. Under folic acid-mediated targeted delivery, the uptake of nanobombs by activated macrophages is increased [27,97]. Subsequently, localized and precise release of Dex alleviated inflammatory cell infiltration and cartilage destruction in the joint cavity under ultrasound stimulation [27,97]. Alternatively, ultrasound responsiveness of biomaterials can be conferred by introducing sonosensitizer sparfloxacin into the structure [98]. A recent study demonstrated that ultrasound-activated H2O2 nanoenzymes produce O 2 to alleviate hypoxia in the joint cavity, thereby downregulating hypoxia-inducible factors to prevent angiogenesis [98]. Unfortunately, the biggest disadvantage of ultrasound is that it is strongly attenuated by the dense bone. The lack of sufficient penetration may also be an important reason limiting the use of ultrasound-responsive biomaterials in bone-related fields. Overall, the application of stimulus-responsive biomaterials for OA therapy is still in the infancy of simple assembly of material precursors and drugs. More systematic and comprehensive preclinical studies are needed in the future to confirm their reliability and stability.

Targeted biomaterials in osteoarthritis
In the treatment of OA, whether by intraarticular injection or systemic injection, the disorganized diffusion of the drug agent in the organism is fundamental to the efficacy of the drug [126]. The diminished efficacy and low utilization of the drug brought about by this disorganized proliferation forced physicians to raise the dose, creating a vicious cycle that ultimately leads to side effects [126]. Therefore, the targeted delivery of drugs by designing reliable OA targeting strategies has become a hot topic of recent research. Targeted biomaterials not only expand the therapeutic window of the drug, but also enhance penetration and retention time [127,128]. Next, we introduce various passive and active targeting strategies in the hope of achieving the healing of OA through targeted drug delivery to cartilage or synovium ( Table 2).  [25]. Copyright 2021, Elsevier. (C) Synthesis process of the drug-loaded nanosystem and its drug release in response to NIR light in vitro [88]. Copyright 2019, American Chemical Society. (D) Immunohistochemical fluorescence staining of articular cartilage matrix markers (Aggrecan) and cartilage inflammation markers (MMP13, ADAMTS5) in mice after 28 days of treatment with Dex-loaded molybdenum disulfide nanosheets [88]. Copyright 2019, American Chemical Society. Reduce expression of IL-1β and IL-6 in articular cartilage and restore bone remodeling in subchondral bone [151] Passive targeting strategy As an avascular dense connective tissue, the cross-linked network formed by type II collagen with a pore size of about 50-60 nm and aggrecan with a chain spacing of about 20 nm hinders the uptake of most drugs [152][153][154]. Many promising experimental drugs have failed in clinical practice due to the fact that they remain on the surface of the cartilage and do not penetrate into the interior and subchondral bone [155][156][157]. The pore size and porosity of the cartilage matrix also vary with the progression of OA, and either too large or too small a particle size will result in poor penetration [158,159]. Thus, suitable and homogeneous particle size will be a shortcut to increasing the effective dose and penetration of the drug. Several studies have shown that morphologically intact healthy cartilage only allows particles below 10 nm to enter the deeper layers, while 15 nm particles remain in the cartilage surface after 24 h [160,161]. However, during the progression of OA, ECM will allow the entry of nanoparticles up to 300 nm, which may be due to the disruption of the cartilage matrix [135,147]. Endocytosis of 15 nm MnO2 nanoparticles by chondrocytes was observed at 30 μm after 24 h incubation with bovine cartilage explants [130]. Furthermore, the penetration of nanoparticles in the synovium is similarly size-dependent. Only 5-nm gold nanoparticles successfully penetrated and would reduce synovial inflammatory biomarker concentrations, corroborating the importance of particle size in OA treatment [129]. Overall, smaller particle size tends to exhibit higher cartilage penetration and cellular uptake, which is one of the important bases for achieving OA targeting.
In addition, the constituent monomers of Aggrecan contain a large number of negatively charged carboxyl residues and sulfate residues, allowing cationic biomaterials to be attracted to the joints by electrostatic forces [162]. Utilizing this unique property, nanoparticles with a positive charge tend to target the ECM and increase the uptake of chondrocytes through electrostatic attraction [162]. A well-designed green fluorescent protein (GFP) was used to probe the optimal range of positive charge [163]. As the GFP surface charge increased from +9 to +36, the permeability of GFP in the cartilage layer gradually diminished and was replaced by a stronger cellular uptake rate [163]. The +9 GFP can be efficiently taken up by chondrocytes and human knee joints, while the electrically neutral GFP is completely inaccessible to cartilage explants [163]. Similarly, when using cationic peptides as carriers, the net charge of +14 exhibited the best cartilage diffusion ability, suggesting that cations need to be applied with attention to the optimal charge range [131]. Bajpayee et al. assembled a multi-armed anti-biotin protein, succinic acid (SA), glutaric acid (GA) and phthalic anhydride (PA) as a +25 mV cationic nanocarrier (mAv-Dex) to deliver Dex ( Figure 6A) [132]. mAv-Dex effectively inhibited IL-1α-induced ECM loss and nitrite release within two weeks [132]. Polyethyleneglycol-functionalized polyamidoamines (PAMAM) with hundreds of primary amine residues are ideal cationic carriers for the treatment of OA ( Figure 6B) [133]. PAMAM loaded with insulin-like growth factor 1 (IGF-1) penetrated into cartilage to an unprecedented depth of 1 mm and extended the retention time of IGF-1 to 30 days ( Figure 6C) [133].
Relative to the poor efficacy of free IGF-1, PAMAM-IGF-1 significantly alleviated osteophyte formation and aggrecan degradation in rats, significantly delaying the pathological changes of OA [133]. Notably, since the interaction of synovial fluid with biomaterials in the articular cavity may lead to "off-targeting", more efficient targeting strategies must be sought, rather than just staying with physicochemical properties [164].

Proactive targeting strategy
Targeting cartilage and extracellular matrix The identification of highly specific targeting ligands is often the paving stone for the study of active targeting strategies. In 2011 cartilage-targeting peptides (CAP, DWRVIIPPRPSA) screened by phage display technology have been shown to specifically induce the binding of nanoparticles to chondrocytes [165]. They are not species-specific for in vivo studies in different model animals [165]. Due to the unique three-dimensional structure and binding sites of short peptides, peptide-functionalized biomaterials have lower immunogenicity and higher operability [166]. Ouyang et al. reported a cartilage-targeted gadolinium carbonate-PDA nanoparticle loaded with hesperidin (HGdPDW) that can be used for magnetic resonance imaging and OA therapy [134]. CAP-modified HGdPDW has a homogeneous morphological structure and elemental distribution, accumulating abundantly in the cartilage of the lower femur and upper tibia in vivo [134]. HGdPDW decreased Toll-like receptor 2 (TLR-2) expression on the plasma membrane surface, thereby downregulating NF-κB/Akt activation and reversing IL-1βinduced chondrocyte apoptosis [134]. In the ACLT model, HGdPDW alleviated chondrogenic malformation and apoptosis marker caspase-3 protein production, revealing a potential mechanism for chondrocyte protection by HGdPDW [134]. In addition to CAP, Rothenfluh et al. identified a type II collagen-targeting peptide (WYRGRL) that is now widely used in biomaterial modification [152]. WYRGRL directs fluorescent probes into the cartilage matrix, so researchers can track articular cartilage lesion progression through in vitro live imaging [167]. As shown in Figure 7A, Ho et al. attached WYRGRL to the surface of ferritin nanocages (CT-Fn) to overcome the drawbacks of conventional oral administration of metformin (Met) [23]. The modified CT-Fn targeted type II collagen in articular cartilage, reducing cartilage macroscopic scores by 47% and 68% at weeks 2 and 6 [23]. Type II collagen-targeting peptidemodified liposomes also exhibited smooth penetration and prolonged retention in the mouse cartilage matrix ( Figure 7B) [136]. Based on the previous work, Lu et al. designed a functionalized nanoparticle (WPV-CuO) with MMP-2 cleavage peptide as a linker for dual-targeting peptides of type II collagen and MSC [137]. WPV-CuO actively homed to cartilage and recruited MSC to cartilage differentiation in vivo [137]. Transcriptome analysis revealed that phosphorylation activation of AKT was significantly downregulated in the WPV-CuO group, implying that the potential therapeutic mechanism of WPV-CuO is through inhibition of the PI3K/AKT/mTOR signaling pathway [137].  To further improve the active targeting of vectors, specific antibodies against articular cartilage have been incorporated into biomaterial fields of application. Compared with other methods, the antibody only binds and tightly attaches to the antigenic epitope on the surface of the corresponding antigen with high affinity and specificity, greatly improving the targeting of the vector [168]. Type II collagen monoclonal antibody (MabCⅡ)-modified liposomes can be used for in vivo imaging by accumulating at the injured joint after intravenous injection [138]. The amount of MabCⅡ binding correlated positively with the severity of OA, while no fluorescent signal was detected in healthy joints [139]. Compared to the negative control, 124nm polymers coupled to MabCII (mAbCII-siNPs) achieved efficient siRNA chondrogenic delivery [140]. mAbCII-siNPs not only silenced the expression of MMP13 and downstream related genes in chondrocytes, but also showed in vivo protection against ectopic calcification of synovial membrane, which was superior to steroid treatment [140]. This suggests that specific antibodybased cartilage targeting strategies are worthy of further long-term study and promising for clinical trials.
For biocompatibility and adaptability considerations, some bionic biomaterials have also been applied for active targeting of OA [169,170]. The use of HA nanoparticles (HA-NP), rather than direct HA injection, allows targeted binding of highly expressed CD44 in damaged cartilage to competitively block activation of the NF-κB signaling pathway ( Figure 7C) [141]. HA-NP not only exhibited hyaluronidase resistance and cartilage penetration, but also dramatically alleviated destabilization of the medial meniscus (DMM)-induced cartilage destruction and subchondral bone plate thickening [141]. The efficacy remained significant at intervals of up to 4 weeks of administration, suggesting that joint cavity injection of HA-NP may be a potential targeted therapeutic strategy for OA [141]. Nanoghosts (NGs) made from mesenchymal stem cell membranes were reported to be a cell-free cartilage targeting platform, and the intracellular transport of NGs was analyzed by fluorescence co-localization techniques [142]. Due to their inherent immunomodulatory and inflammatory homing ability, NGs were first endocytosed by chondrocytes and then entered the lysosome, subsequently downregulating the expression of COX2 and PEG2 at the mRNA and protein levels compared to the untreated group [142]. Moreover, when using M2 macrophage membranes as a shell ( Figure 7D), artificial macrophages (AM2M) block interleukininduced acute inflammatory injury while avoiding chondroitin sulfate (ChS)-induced immune stimulation [143]. AM2M has the greatest articular retention time and stable ChS release rate, successfully restoring the expression of MMP13 to normal levels [143]. It is demonstrated that it is indeed feasible to mimic the secretion of anti-inflammatory cytokines by M2 macrophages in OA. For bionic biomaterials used in OA, there is a lack of sufficient systematic in vivo and in vitro studies to clarify the action mechanisms and long-lasting therapeutic effects, restricting their clinical translation. Overall, active targeting strategies based on cartilage and extracellular matrix are still in their infancy, further development of more desirable OA-targeting biomaterials can be achieved in the future by drawing on experience from other diseases such as cancer or immunomodulation.

Targeting synovium
Synovium is composed of macrophages, fibroblasts and vascular endothelial cells, which perform the functions of material exchange and secretion of synovial fluid [171]. Apart from cartilage damage, synovial inflammation is also a major pathological characteristic of OA lesions [172,173].
Topical or systemic administration of corticosteroid is often used for synovial inflammation in OA, but their potential toxicity and excessive metabolism both make long-term use impossible and increase the risk of osteoporosis [174]. For this reason, biomaterial therapeutic strategies targeting synovium have attracted interest. Initially, the researchers hoped to achieve synovial-targeted delivery by using RGD peptide binding to integrin αvβ3 on synovial vascular endothelial cells [144]. However, the expression of integrin αvβ3 is not synovial specific and may lead to a non-specific accumulation of biomaterials extrasynovial [175,176]. Colombo et al. achieved targeted delivery of degradable nanoparticles (tBNPs-MTX) by coupling cyclic synovial homing peptides [145]. This cyclic peptide showed strong binding capacity to von Willebrand Factor + (VWF + ) vascular endothelial cells and CD34 + hematopoietic stem cells in synovial tissue [145]. tBNPs-MTX significantly reduced neovascularization density and leukocyte counts in rat synovium, further demonstrating the great potential of synovium as a therapeutic target for OA [145]. In addition, a block polymer-based superoxide dismutase nanoparticle (SOD-NP) was reported to have a unique targeting ability to synovial fibroblasts, overcoming the excessive degradation of SOD [146].
In the inflammatory microenvironment, interferon-γ (IFN-γ) and TNF-α induce synovial macrophages to polarize toward the pro-inflammatory M1 type, leading to the release of downstream inflammatory factors IL-1β and IL-6 [177]. Modulation of the M1/M2 ratio of macrophages to suppress the progression of inflammation may be a promising intervention for OA [178]. Zhou et al. grafted an anti-CD16/32 antibody onto a modified zeolitic imidazolate framework-8 (ZIF-8) to restrain hypoxiainducible factor-1α (HIF-1α) and recover macrophage mitochondrial dysfunction [148]. After macrophages were reprogrammed to M2 type, angiogenesis and inflammatory cell infiltration in the synovium were dramatically reduced, and ACLT-induced cartilage damage was reversed [148]. More interestingly, Chen et al. designed and synthesized novel magnesiumbased ligand containers (Rap-FA@MgDHIA) for the stepwise loading and release of Rap utilizing folic acid (FA) as a surface ligand [149]. Folate receptor β (FRβ) is highly expressed on the membrane surface of M1 macrophages [149]. Rap-FA@MgDHIA achieves targeting M1 type macrophages by binding to FRβ [149]. Rap-FA@MgDHIA subsequently releases Rap to convert macrophages from a pro-inflammatory phenotype (M1) to an anti-inflammatory phenotype (M2) [149]. After 14 days of Rap-FA@MgDHIA local injection, iNOS-positive M1-type macrophages in synovial tissue were remarkably minimized [149]. Rap-FA@MgDHIA alleviated synovial inflammatory cell infiltration while articular cartilage and subchondral bone were successfully conserved [149]. In another study targeting synovial macrophages, FA-modified nanoparticles (CPHs) acted as CO donors to inhibit MIA-induced articular damage in rats ( Figure 8A) [150]. CPHs induced apoptosis by depleting H2O2 in activated macrophages while decreasing mitochondrial membrane potential and ATP synthesis, yet showed a satisfactory safety profile for normal macrophages [150]. Lower ROS levels and higher CO levels reduced the macrophage-mediated inflammatory response ( Figure 8B) [150]. Considering the potential uptake by macrophages and lack of targeting in the circulation by conventional drug delivery, Xue et al. fused erythrocyte membranes and neutrophil membranes on a hollow copper sulfide surface to provide a promising platform (D-CuS@NR NPs) for inflammation-associated diseases ( Figure 8C) [151]. The modification of the composite membrane not only non-specifically adsorbs a variety of serum inflammatory factors, but also rapidly accumulates in the articular cavity at 60 min to inhibit cartilage wear [151]. D-CuS@NR NPs can release Dex while also responding to NIR to produce photothermal effects for adjunctive therapy, demonstrating a new mode of OA treatment [151]. Additionally, self-assembled micellar nanoparticles (DS-TA NPs) of dextran sulfate-triamcinolone acetonide conjugates specifically target the scavenger receptor class A (SR-A) on the surface of activated macrophages to reduce the expression of pro-inflammatory cytokines for promising targeted therapy of OA [179]. In summary, the engineering of targeted biomaterials for synovial cells optimizes the biodistribution and potency of drugs in vivo as a promising therapeutic strategy for OA.

Targeting subchondral bone
Although cartilage degeneration and ECM loss have long been considered the primary causative factors of OA, there is growing evidence that subchondral bone lesions precede the onset of cartilage loss [4]. Bone trabeculae and platelets in the subchondral bone are arranged in an appropriate proportional distribution to disperse mechanical stress, and osteoclast-mediated bone resorption and osteoblast-mediated osteogenesis are counterbalanced [4]. After the onset of the lesion, the over-activated osteoclasts of the subchondral bone induce the invasion of blood vessels and nerves into the cartilage layer, triggering inflammation and pain [180]. Several preclinical studies have been conducted to show the effectiveness of OA treatment options targeting abnormal subchondral bone remodeling and angiogenesis [181][182][183]. With only the simplest drug injections to discuss efficacy on subchondral bone, rather than specific delivery, there remains a potential risk of nonspecific distribution of the drug in the articular cavity or systemically. Bisphosphonates (BPs) are used as highly effective bone targeting agents to inhibit osteoclast activity, and biomaterials coupled to BPs are also used to carry other active molecules for basic bone-related therapies [82]. BPsmodified nano-apatite (NP-BP) achieves targeting of subchondral bone by responding to lower pH in the microenvironment [82]. Since the bone resorption of subchondral bone in the early stage of OA was inhibited by NP-BP, the pain behavior of rats was relieved and the three-dimensional structure of the cartilage layer was preserved intact within 90 days, demonstrating that targeting subchondral bone has favorable feasibility and prospect for the management of OA [82]. In addition, targeting crystalline mineral loss and reduction in collagen mineralization in subchondral bone may also be a potential target for OA [184]. Yet, treatment options that target subchondral bone present enormous difficulties due to the natural barrier of articular cartilage and the low blood circulation to the subchondral bone. Although there are limited studies on targeting subchondral bone, the future development of subchondral bone-targeting biomaterials based on surface modification and innovative structures will be a new direction for OA treatment.  [150]. Copyright 2020, Elsevier. (B) Changes in ROS/CO levels in activated macrophages and inhibition of inflammatory factor secretion after treatment with CPHs [150]. Copyright 2020, Elsevier. (C) Schematic diagram of the preparation and therapy of the D-CuS@NR NPs [151]. Copyright 2022, Elsevier.

Limitations and challenges of clinical translation
Although biomaterials hold exciting promise for improving drug solubility and precision release, the potential risks and unknown chronic toxicity of these materials in living organisms limit their clinical translation. The size of most biomaterials reaches the nanometer dimension, which allows them to easily cross multiple biological barriers and be non-specifically endocytosed by immune cells [185]. Although no visceral toxicity or other complications have been reported in any of the biomaterials discussed above, the risk of temporary organ accumulation, sudden release or accidental off-target remains is still a concern. In addition to focusing on recovery from OA lesions, researchers should have a sufficient long-term understanding of the normal physiological functions of joint tissue (e.g., mechanical strength, synovial fluid composition, and nutrient supply) to ensure the safety of biomaterials. The polymers synthesized in the experiments usually do not reach 100% polymerization, although the polymers themselves are not toxic, the residual polymer monomers may promote apoptosis or genotoxicity [186]. Around 100 nm chitosan, the most common natural biopolymer in the biomedical field, would cause significant neurotoxicity, liver damage and reduced hatchability in zebrafish [187,188]. In addition, joints are one of the most important motor tissues in the body, implant particles from sportswear will activate the NF-κB signaling pathway in macrophages and induce bone mineral loss [189]. Chronic inflammation due to prosthetic wear particles may present an unpleasant challenge for patients with advanced OA undergoing total joint arthroplasty [189]. Also, in vivo evaluation of simulated biomaterials using in vitro stem cell models or bone like organs has recently received much attention, which would be more convincing than toxicity experiments with single cell lines [190,191].
The biodistribution and toxicity of the delivery platforms within the joint will also be influenced by the strength of the surface charge. Positively charged cationic materials tend to have easier uptake by cells and are therefore often used for therapeutic RNA delivery [192]. However, when the charge intensity or concentration is too high, either cation or anion, it will trigger mitochondria-mediated cellular autophagy and plasma membrane damage [193]. In previous studies, cationic nanoparticles were considered as ideal carriers for targeting cartilage using electrostatic attraction. However, as the surface charge continues to rise, excessive electrostatic interactions will prevent the penetration of nanoparticles larger than +14 mV into deeper layers, while other non-covalent bonds (e.g. hydrogen bonds, hydrophobic bonds) formed between GAG and cationic peptides will also affect their permeability [131]. On the other hand, for carriers applied for systemic injection, liposomes with high surface charge have increased hepatic accumulation and complement protein-mediated immune clearance relative to electroneutral liposomes [194]. Researchers should be more cautious to examine whether the loss of GAG in the late stages of OA may lead to off-target effects of cation-targeting strategies and subsequently increase the cumulative toxicity in non-targeted tissues.
Due to the presence of synovial fluid, adsorption of synovial proteins on the surface of biomaterials will probably interfere responsive group breakage and exposure of targeting ligands [195]. Although PEG modification can suppress protein adsorption and protein corona formation, it does not completely avoid albumin or complement protein binding [196]. Brown et al. indicated that incubation in the synovial fluid would lead to an increase in cationic nanoparticle size and charge reversal [195]. It is speculated that the protein corona formed by adsorbed synovial proteins may mask the functional ligands on the nanoparticle surface and subsequently affect their tissue transportation and responsive release within the joint. Hence, how to overcome the interference of synovial fluid with biomaterials to achieve efficient OA treatment will be a great challenge for clinical translation. They should be further tested in a diversity of joint tissues or in large animals to understand and confirm the efficacy of OA.
Finally, researchers should be cautious in their choice of model animals and means of inducing OA. Due to differences in species, age, and sex, none of the animal models can fully replicate all pathological processes in human OA [197]. Researchers need to consider a combination of study objectives and concerns to determine the modeling approach. For example, spontaneous OA allows researchers to analyze the disease process from early to late stages in different time sequences. However, such OA models are time-consuming and expensive, while the variation between animals can be significant. Invasive surgery allows experimenters to achieve a less variable OA development and is currently the most commonly used and stable animal model of OA [198]. Nevertheless, it is not suitable for studying the changes in OA during long-term wear and aging [198]. Induction by chemical adjuvants such as collagenase can produce significant cartilage wear and pain perception in a short period of time [198]. Yet, this is far from the pathogenesis of human OA. These different mechanisms of pathogenesis will perhaps influence the efficiency of drug targeting. Some models that produce large amounts of ROS or pH reduction rapidly may exhibit more pronounced responsive drug release relative to naturally occurring OA [199]. Thus, a single model of OA treatment strategy cannot be fully equated with clinical efficacy. Future studies should include more convincing evidence to compare the efficiency of targeted release of drug-loaded biomaterials in different OA models.

Conclusion
OA is a degenerative disease characterized by loss of articular cartilage and local inflammation that affects tens of millions of people worldwide. The unknown pathogenesis of OA leads to clinical treatment strategies that do not completely eradicate cartilage degradation. Conventional intra-articular injections or oral administration of drugs are often rapidly cleared by phagocytes or metabolic organs resulting in low drug utilization and a narrow therapeutic window. The application of targeted and responsive biomaterials in the biomedical field is bringing new light to the treatment of OA. Efficient drug delivery systems based on the OA microenvironment (enzymes, ROS, pH, stimulation) and lesion sites (cartilage, synovium, subchondral bone) have shown promising applications. Not only do they maximize the therapeutic potential of drug molecules, but more importantly they improve the biodistribution and delivery of drugs. The "biomissile" based on biomaterials penetrates nanoscale pores and has long-lasting controlled release, avoiding the pain and economic burden of multiple high-dose administrations to patients. By modifying and optimizing the structure and physicochemical properties, biomaterials overcome the therapeutic bottleneck of drug monomers and enable the integration of treatment. The controlled release of drugs in combination with microenvironmental changes assists in better mastering the pathological stages of OA, while the coupling of targeted ligands opens new horizons for precision therapy. Moreover, these targeted and responsive biomaterials may be used for other OA-affected joint tissues in the future, including fat pads and joint ligaments.
Nevertheless, biomaterial-based clinical therapies for OA are still in their infancy, and there is a lack of sufficient clinical studies to demonstrate the efficacy and safety of these strategies. Previous studies have also shown that the selection of natural cartilage components based on hyaluronic acid and collagen as carriers may yield better therapeutic benefits and safety. Also, the cartilage layer is thicker in primates, which would greatly prolong the diffusion of biomaterials. These findings should be further validated in a more closely related human model (e.g., rhesus monkey). In addition, further validation is needed to determine if the effectiveness of these targeted ECM strategies is compromised in the late stages of OA when the collagen network of chondrocytes is depleted. On the other hand, new multifunctional biomaterials involve more complex and sophisticated technologies, often leading to expensive economic costs in clinical applications, which should be taken into account. Since the criteria for toxicity evaluation of biomaterials at the development stage are not yet standardized, the absence of a quality evaluation system will lengthen the cycle of the approval process, which is not conducive to the clinical translation of biomaterials. To break the existing barriers and accelerate the clinical translation of targeted and responsive biomaterials, we need to deepen our understanding of OA not only at the molecular and cellular levels, but also to speed up the research process of novel active drug molecules and the search for more specific drug carriers to achieve stronger delivery effects. To meet these challenges, broad and close collaboration and cross-disciplinary research among materials scientists, biologists and clinicians are required. Looking ahead, with the continuous improvement of regulatory mechanisms and experimental techniques, the targeted and responsive biomaterials are expected to become a panacea for OA patients.