NIR-photocatalytic regulation of arthritic synovial microenvironment

Abstract

Synovial microenvironment (SME) plays a vital role in the formation of synovial pannus and the induction of cartilage destruction in arthritis. In this work, a concept of the photocatalytic regulation of SME is proposed for arthritis treatment, and monodispersive hydrogen–doped titanium dioxide nanorods with a rutile single-crystal structure are developed by a full-solution method to achieve near infrared–photocatalytic generation of hydrogen molecules and simultaneous depletion of overexpressed lactic acid (LA) for realizing SME regulation in a collagen-induced mouse model of rheumatoid arthritis. Mechanistically, locally generated hydrogen molecules scavenge overexpressed reactive oxygen species to mediate the anti-inflammatory polarization of macrophages, while the simultaneous photocatalytic depletion of overexpressed LA inhibits the inflammatory/invasive phenotypes of synoviocytes and macrophages and ameliorates the abnormal proliferation of synoviocytes, thereby remarkably preventing the synovial pannus formation and cartilage destruction. The proposed catalysis-mediated SME regulation strategy will open a window to realize facile and efficient arthritis treatment.

Near infrared–activatable nanocatalysts enable photocatalytic therapy of arthritis by regulating the microenvironment.

INTRODUCTION

Arthritis is a common chronic inflammatory disease, which is characterized by persistent joint inflammation and subsequent cartilage destruction (1). In the arthritic synovial microenvironment (SME), the excessive proliferation of synoviocytes and the massive infiltration of immune cells into synovia together cause synovial pannus and metabolic disorder with both abnormal vascularization and inappropriate accumulation of metabolic intermediates such as lactic acid [LA; about 7.2 mM in rheumatoid arthritis (RA)] derived from inefficient aerobic glycolysis, which is similar to the Warburg effect in cancer (2, 3). These abnormal pathological events lead to the acidification of SME, enhance the polarization of anti-inflammatory M2 macrophages into proinflammatory M1 macrophages, and subsequently stimulate fibroblast-like synoviocytes (FLS) to progress toward proinflammatory/invasive phenotypes that further activate chondrocytes to invade cartilage (4, 5). Metabolite-activated M1 macrophages interact with proinflammatory FLS to form a positive feedback, enlarging the inflammatorily responsive deterioration of RA. Here, we hypothesize that the scavenging of abnormal metabolites such as LA might intercept both the activation of M1 macrophages and the macrophage-FLS feedback to correct the SME for arthritis treatment (Fig. 1).

Fig. 1. Schematic illustration of NIR-photocatalytic strategy and mechanism for arthritic SME regulation with the NIR-responsive HTON.

Left: Band structure and NIR-photocatalytic feature of HTON. Middle: The pathways of photocatalysis-mediated SME regulation. Right: Treatment method.

Reactive oxygen species (ROS) positively correlate with the inflammation in the SME and the progression of arthritis, involving the induced M1-like macrophages and the activation of chondrocytes (6). The administration of antioxidants for scavenging ROS is recognized to be a promising strategy of arthritis treatment (7–9). Among various antioxidants, hydrogen molecule has broad-spectrum antioxidation and anti-inflammation effects, outstanding tissue penetration ability, and high biosafety against a variety of inflammation-related diseases including arthritis (10–15). Sung et al. and Li et al. (16, 17) locally injected the Mg powder–encapsulated poly(lactic-co-glycolic acid) microparticles into mouse arthritis knee to deliver hydrogen molecules for inflammation mitigation, arresting the progression of arthritis. Despite these advances, on demand and sustained hydrogen release is vitally important for enhanced efficacy of arthritis treatment but still substantially challenging. Here, we propose a photocatalytic strategy to sustainably generate hydrogen molecules on demand and meanwhile to oxidatively deplete LA for RA treatment (Fig. 1) by virtue of its low redox potential [pyruvate (PA)/LA = −0.19 eV] (18).

Among various photocatalysts, titanium dioxide (TiO₂) has high biocompatibility and has been approved by the Food and Drug Administration to use as the additive of food, drug, and cosmetics. By a doping route, the light absorption of TiO₂ can be shifted to the near infrared (NIR) zone, and its energy band can be adjusted to be suitable for NIR-photocatalytic water splitting for hydrogen generation and organic degradation under mild conditions (19, 20), which is favorable for biomedical applications due to higher tissue penetration and lower phototoxicity of NIR light compared with ultraviolet (UV)–visible light (21–23). However, doped TiO₂ was generally synthesized by using commercial P25 powder with mixed crystal phases of anatase and rutile at high temperature, leading to poor aqueous dispersivity and photocatalytic instability of hydrogen-doped anatase, which limit its biomedical application.

In this work, we doped hydrogen into uniform TiO₂ nanorods (TON) with a rutile single-crystal structure by a room temperature solution method to obtain a kind of highly stable and monodispersive hydrogen-doped TiO₂ (H_xTiO_2−x) nanorods (HTON) with a bandgap of 1.45 eV (−0.3 eV conduction band potential, 1.15 eV valance band potential) for the NIR-photocatalytic regulation of SME in arthritis (Fig. 1). We found that HTON can not only locally generate hydrogen molecules but also oxidatively deplete LA, which is overexpressed in the arthritic SME under the focused NIR irradiation of 808-nm laser on the HTON-injected RA knee (Fig. 1). Generated hydrogen molecules scavenged intracellular ROS in proinflammatory M1 macrophages, inhibiting their proinflammatory phenotype to block their damage to cartilage (Fig. 1). Meanwhile, the depletion of LA intercepted the activation of FLS, M1 macrophages, and chondrocytes, blocking their invasion to cartilage to correct the SME for efficient RA treatment (Fig. 1). Such a “two birds with one stone” photocatalytic strategy opens a avenue for the treatment of arthritis.

RESULTS

Synthesis and structural characterization of HTON

Industrial photocatalysis always pursues the highest catalytic efficiency, which can be improved by engineering the catalysts for solar full-spectrum absorption. However, photocatalytic therapy prefers NIR light rather than UV–visible light because NIR light has higher tissue penetration and lower phototoxicity to cells/tissues. Particle dispersion and aqueous stability of catalysts were frequently ignored but currently have to be carefully considered for biomedical applications. Hydrogen-doped TiO₂ nanoparticles were generally synthesized under annealing at high temperature, causing irreversible hard agglomeration (21, 22, 24–26). Therefore, we developed a full-solution method to synthesized TON and HTON in this work to obtain high dispersion in an aqueous solution. Hydrogenated product derived from anatase was quite unstable in the aqueous solution when TiO₂ with a mixture phase such as commercial P25 was commonly used (21, 22, 24–27). Therefore, we first prepared a kind of rutile single-crystal nanorods as TiO₂ source to ensure high stability of synthesized HTON.

To obtain good aqueous dispersion of the rutile TON, we modified the routine calcination method with a hydrothermal strategy (26). From figs. S1A and S2, the synthesized TON exhibited high dispersion and uniform dimension as expected. Moreover, x-ray diffraction (XRD) result indicated that the synthesized TON was a single crystal of rutile along (110) orientation (Fig. 2G; JCPDF card number 73-1232) in accordance with high-resolution transmission electron microscopy (HR-TEM) result (fig. S1B). To maintain good aqueous dispersion during hydrogen incorporation, we used an electronical solution method to incorporate hydrogen into TON. Typically, Li was first inserted into TON in an ethylene diamine solution of Li and then exchanged with hydrogen in the aqueous solution of HCl. From Fig. 2 (A to G) and fig. S3, the as-synthesized HTON well inherited the rod-like morphology, crystal orientation, uniform dimension, and good dispersion of TON. The increase of (110) interplanar distance implied that hydrogen had been incorporated into TON (fig. S4). Moreover, from the HR-TEM, selected area electron diffraction, and XRD patterns in Fig. 2 (B, C, and G), HTON maintained the rutile sing-crystal structure. Furthermore, from x-ray photoelectron spectroscopy (XPS) spectra of two samples (fig. S5), Ti 2p_3/2 and 2p_1/2 peaks centered at 458.5 and 464.3 eV (Fig. 2H) and O 1s peak at 529.7 eV (Fig. 2I) were typically attributed to the Ti─O bonds in TiO₂. By comparison, the emergence of two Ti 2p peaks with lower binding energy at 457.8 and 463.6 eV (Fig. 2H) was caused by Ti─H bonds, suggesting successful hydrogen incorporation. In addition, the remarkable increase of O 1s peak at 531.1 eV corresponded to bridging hydroxyls (Ti─OH─Ti), revealing a large amount of defects in HTON, which was much more than calcined H_xTiO_2−x products previously reported (24, 26). Moreover, typical Raman vibration modes of TiO₂, including E_g, B_1g, and A_1g, were obviously weakened and broadened after hydrogenation, and the most intense E_g mode centered at 142 cm⁻¹ blue-shifted to 149 cm⁻¹ (as indicated by dashed lines in fig. S6). In addition, the characteristic symmetric stretching vibration band of Ti─O in the TiO₄ tetrahedra at 400 to 1000 cm⁻¹ (as indicated by blue framework in fig. S7) was also weakened and broadened after hydrogenation. These Raman and Fourier transform infrared spectroscopy (FTIR) results consistently indicated the successful hydrogen incorporation of HTON, which caused the destruction of the symmetry of TiO₂ lattice (24, 28).

Fig. 2. Morphology, composition, structure, and dimension characterization of HTON.

TEM image (A), HR-TEM image (B), selected area electron diffraction pattern (C), and corresponding elementary mapping of (A) (D to F) of HTON and XRD (G) and XPS patterns (H and I) of TON and HTON. a.u., arbitrary units.

Photoelectric behaviors and NIR-photocatalytic performances

TON without hydrogen incorporation exhibited a typical UV absorption behavior (Fig. 3A and fig. S8A) and a broad bandgap of 3.02 eV (Fig. 3B) and thus had no NIR-photocatalytic activity. After hydrogen incorporation, HTON demonstrated a strong localized surface plasmon resonance absorption in the visible-to-infrared range (Fig. 3A and fig. S8C), owing to large amounts of hydrogen incorporation–induced defects (Fig. 2I). Hydrogen incorporation narrowed bandgap to 1.45 eV (Fig. 3b), consequently enabling the NIR-photoelectric conversion of HTON at ≤855 nm. Moreover, the XPS-derived valance band of HTON (Fig. 3C) was calibrated to be 1.15 eV versus normal hydrogen electrode, and thus, the conduction band of HTON was calculated to be −0.3 eV according to its bandgap, providing enough high oxidation and reduction potentials for LA/PA and H⁺/H₂ evolutions whose redox potentials are 0.19 and 0 eV, respectively (Fig. 3D).

Fig. 3. Photoelectric features and NIR-photocatalytic hydrogen generation and LA consumption performances of HTON.

UV–visible–NIR absorption spectra (A), the plots of (αhν)² versus hν for bandgaps (B), valance band potentials of TON and HTON (C), the schematic illustration of the mechanism for HTON-mediated NIR-photocatalytic hydrogen generation and LA consumption (D), the controllability of HTON for NIR-photocatalytic hydrogen generation in the aqueous solution of LA (1 mM) at various power densities (E), and NIR-photocatalytic LA consumption behavior of HTON under irradiation of an 808-nm laser at various power densities (F). a.u., arbitrary units.

Furthermore, we checked the NIR-photocatalytic capability of HTON for hydrogen generation and LA consumption in an SME-simulated solution using an 808-nm NIR laser. It was worth noting that HTON-free detection solution was separated from NIR-irradiated HTON solution by a piece of filter membrane [molecular weight cutoff (MWCO) = 3000 Da] to avoid the negative influence of laser and HTON on the hydrogen electrode, which was used to monitor the hydrogen generation process of HTON in real time, and NIR irradiation was executed in the dark to avoid the potential effect of natural light. From Fig. 3 (E and F), HTON can indeed generate hydrogen gas and meanwhile consume LA under irradiation of an 808-nm laser. The NIR-photocatalytic hydrogen generation and LA consumption of HTON were dependent on power density of laser and its irradiation time. Moreover, the NIR-photocatalytic reaction of HTON was highly controllable (Fig. 3E) in great favor of on-demand therapy. In addition, HTON exhibited especially high compatibility with cells, achieving 83% of cellular viability at the particle concentration of 1000 μg/ml (fig. S9).

Effects and mechanisms of SME regulation in vitro

The occurrence and evolution of RA are closely related to the inflammatory harm in SME, mainly involving the synoviocytes invasion of cartilage. Some metabolic intermediates, such as ROS and LA (7.2 mM), play an important role in SME induction. In this work, we confirmed that ROS-mediated oxidative stress can induce the M1-phenotype polarization of RAW264.7 macrophages according to fluorescence imaging and real-time polymerase chain reaction (PCR) assay (Fig. 4B and fig. S10), while LA (0.5 and 1 mM) can distinctly promote the abnormal proliferation of FLS in a concentration-dependent way (fig. S11) and induce the transformation of their proinflammatory/invasive phenotypes (fig. S12) and further polarized M1-phenotype macrophages (fig. S13) and activated chondrocytes in accordance with previous reports (29, 30). These activated synoviocytes, macrophages, and chondrocytes would destruct cartilage mainly through the overexpression of matrix metalloproteinases (MMPs) (31, 32). As confirmed above, NIR-photocatalysis based on HTON (50 μg/ml of HTON, 808-nm laser NIR irradiation at 0.5 W/cm² for 6 × 5 min) can generate molecular hydrogen, which has well proven to be able of scavenging ROS in inflammatory cells/tissues (Fig. 4B) (10–13) and, meanwhile, consume LA to save chondrocytes at the therapeutic window concentration of 25 to 1000 μg/ml (fig. S14). Therefore, we here hypothesized that the HTON-mediated NIR-photocatalytic scavenging of LA and ROS could block the M1 polarization of macrophages and the proinflammatory/invasive phenotype of synoviocytes to correct the SME for RA treatment (Fig. 4A).

Fig. 4. In vitro NIR-photocatalytic therapy performances and mechanisms of HTON.

The proposed pathways for NIR-photocatalytic regulation of arthritic SME (A), the effects of H₂ on macrophage polarization and intracellular ROS level (B), the effects of HTON photocatalysis on LA-activated FLS (C), the effects of HTON photocatalysis on LA/FLS-activated macrophages (D), the effects of HTON photocatalysis on LA-/FLS-activated chondrocytes (E), and the levels of corresponding representative indicators (F). P values were calculated by the two-tailed Student’s t test (***P < 0.001). Data were presented as means ± SD. Scale bars, 20 μm (B to E). LPS, lipopolysaccharide.

To verify the hypothesis, we checked the effects of HTON-based NIR-photocatalysis (molecular hydrogen plus LA consumption) and individual molecular hydrogen on LA-stimulated FLS (Fig. 4C and fig. S18A), RAW264.7 macrophages (Fig. 4D and fig. S18B), and chondrocytes (Fig. 4E and fig. S18C), respectively. From Fig. 4 (C to E) and figs. S15 to S17, individual LA activated the proinflammatory/invasive phenotype of FLS with remarkably enhanced expression of Fibroblast activation protein–α (FAPα), interleukin-1β (IL-1β), tumor necrosis factor–α (TNF-α), and MMP-3; further mediated M2 to M1 polarization of RAW264.7 macrophages with remarkably enhanced expression of inducible nitric oxide synthase (iNOS) and MMP-3 and decreased expression of CD206 and CD163; and also mediated the activation of chondrocytes with remarkably enhanced expression of cyclooxygenase 2 (COX2), MMP-3, and intercellular adhesion molecule–1 (ICAM-1) and decreased expression of Aggrecan. Noticeably, all these three kinds of LA-stimulated cells highly expressed MMP-3, which is one of most critical cartilage-destructing cytokines, and the morphology of chondrocytes became a star from fusiformate under LA stimulation (Fig. 4E), implying their cartilage-invading phenotypes in SME. Compared with LA, neither LA + NIR nor LA + HTON treatment caused distinctly influence on these kinds of cells. But LA + HTON + NIR treatment (50 μg/ml of HTON, 808-nm laser NIR irradiation at 0.5 W/cm² for 6 × 5 min) significantly reduced all these negative effects of LA, even correcting them to approximately normal levels (Fig. 4F and figs. S15C, S16C, and S17C). The efficient NIR-photocatalytic consumption of LA can make a major contribution to the correction outcome.

Because two half-reactions in catalysis cannot be separated, two exact individual controls, hydrogen generation and LA consumption, cannot be realized to investigate their individual effects and contributions to SME regulation in the same conditions. Therefore, we used a specifically made hydrogen incubator to primarily investigate the effect of molecular hydrogen. Although we cannot guarantee that the dosage of hydrogen in hydrogen incubator was the same to the case of NIR-photocatalytic hydrogen generation from HTON, this control experiment can provide some clues to recognize the role of hydrogen in SME at least. From fig. S18A, hydrogen treatment can indeed inhibit the LA-activated invasive phenotype of FLS as indicated by the suppressed expression of MMP-3 and IL-1β. When the LA-stimulated FLS cells were cocultured with RAW264.7 macrophages or chondrocytes in a Transwell plate, hydrogen treatment can also both enhance the anti-inflammatory phenotype of macrophages (fig. S18B) and reduce the proinflammatory phenotype of chondrocytes (fig. S18C) as suggested by decreased expression of MMP-3 and COX2 and increased expression of COD206. Similar to the NIR-photocatalytic treatment with HTON, the expression of MMP-3 in all these three kinds of LA-stimulated cells was efficiently depressed by hydrogen treatment, suggesting the role and contribution of hydrogen molecules in blocking cartilage destruction. Furthermore, we performed the transcriptome analysis of chondrocytes to investigate the effect of treatment. From the Venn diagram in fig. S19A, LA induced 567 of differentially expressed genes compared with control, 172 of which were shared with the HTON + NIR treatment group. The 394 up-regulated and 173 down-regulated mRNAs in the LA group (fig. S19B) mainly involved oxidoreductase activity, ferroxidase activity, and programmed cell death, while the HTON + NIR treatment group mainly regulated acute inflammatory response and protein and complement activation (fig. S19C), indicating the anti-inflammation and antiapoptosis role of the NIR-photocatalytic treatment.

In vivo photocatalytic therapy outcomes

Encouraged by in vitro therapeutic results, we further investigated the efficacy of NTON-based NIR-photocatalytic therapy on an RA mouse model. The collagen-induced arthritis (CIA) mouse model was built by immunization with an emulsion of complete Freund’s adjuvant (CFA) and type II collagen according to a classic protocol (33) and used for RA treatment after arthritis induction for 28 days. For treatment experiment, the phosphate-buffered saline (PBS) solution of HTON (2 mg/ml, 20 μl) was injected into articular cavity of RA mice at day 28 followed by NIR irradiation (0.3 W/cm², 6 × 5 min) on the knee joints, and NIR irradiation was executed every week for five times (Fig. 5A). After 0.3 W/cm² NIR irradiation on HTON-injected joints for 30 min, the level of LA in joint fluid was significantly reduced (Fig. 5B), and a high amount of H₂ (15 μM) was generated, exhibiting high photocatalytic activity in vivo in good agreement with in vitro–simulated results (Fig. 3, E and F). From Fig. 5C and fig. S20, the NIR-photocatalytic therapy with HTON + NIR remarkably attenuated the typical RA symptomes, including swelling and redness on all the paws of RA mice after 32 days treatment, but neither individual HTON nor individual NIR did. The distinct decrease of knee diameter of RA mice by NIR-photocatalytic therapy was further quantified in Fig. 5D. It suggested that NIR-photocatalytic hydrogen generation and LA consumption can indeed improve RA effectively as HTON did in vitro. Moreover, we also measured paw volume and conducted the subjective evaluation of arthritis severity according to a classic method in real time (33, 34). From Fig. 5 (E and F), neither individual HTON nor individual NIR obviously influenced the gradual deterioration of RA in the whole 32-day treatment process. By comparison, the NIR-photocatalytic therapy with HTON + NIR sharply exhibited clear therapeutic efficacy in the early days and continuously and effectively suppressed the evolution of RA, suggesting high sustainability of NIR-photocatalytic therapy.

Fig. 5. In vivo NIR-photocatalytic therapy performances of HTON.

The therapeutic protocol with HTON on a CIA mouse model (A), the intra-articular LA consumption and H₂ generation after HTON + NIR treatment (B), the representative digital pictures of left lower paws of RA mice after 39-day treatment (C), knee diameter change of RA mice after 32-day treatment (D), paw volume change (E) and arthritis score (F) of RA mice from CIA model building (first 28 days) to treatment ending (later 32 days). P values were calculated by the two-tailed Student’s t test (*P < 0.05, **P < 0.01, ***P < 0.001). Data were presented as means ± SD.

Furthermore, we conducted the histological examination of NIR-photocatalytic therapy outcomes using computed tomography (CT) and tissue staining methods at the end of treatment. From Fig. 6A, blank control group (PBS) demonstrated some typical characteristics of RA, including the osteoporosis of limb bone, severe bone erosion at the articular surface and on the heels, and the narrowing of joint space. Other two control groups (PBS + NIR and HTON) did not obviously improve or worsen these RA symptoms. But in the NIR-photocatalytic treatment group (HTON + NIR), no osteoporosis was visible, and the bone surface of joints and heels became very smooth, suggesting no bone erosion. Moreover, the profile of joints became clearer, and joint space became normal. These CT results indicated that NIR-photocatalytic therapy effectively blocked the typical bone destruction of RA, well explaining the abovementioned therapy outcomes (Fig. 5).

Fig. 6. Histological examination of NIR-photocatalytic therapy by HTON.

CT images of mice paws with different treatments (A), H&E staining images of synovial tissues of normal mice and RA mice with different treatments (B), H&E staining (C), Safhnin−fast green staining (D), and Alcian blue−periodic acid Schiff staining (E) images of cartilage tissues of normal mice and RA mice with different treatments. Scale bars, 1 mm (A) and 100 μm (B to E).

Moreover, articular tissue was collected for the histological examination of synovium and cartilage (fig. S21). By comparison of synovial tissues in normal (normal group) and RA (PBS group) mice in Fig. 6B, serious synovial hyperplasia and inflammatory infiltration were clearly visible from hematoxylin and eosin (H&E) staining results, which is one of most typical characteristics of RA. But the NIR-photocatalytic treatment (HTON + NIR) effectively inhibited both synovial hyperplasia and inflammatory infiltration (Fig. 6B), owing to NIR-photocatalytic hydrogen generation and LA consumption in support of anti-inflammation and antiproliferation (Fig. 4). In comparison of cartilage tissues in normal (normal group) and RA (PBS group) mice in Fig. 6 (C to E), cartilage in RA was largely eroded because its structure was seriously destructed with defect and attenuation (Fig. 6, D and E), and its key components including chondrocytes (Fig. 6, C and D) and glycosaminoglycan (Fig. 6D) largely lost, which are some of most typical characteristics of RA. Other two control groups (PBS + NIR and HTON) did not affect these pathological changes in RA cartilage. But the NIR-photocatalytic treatment (HTON + NIR) thoroughly eliminated these pathological changes, demonstrating the intact structure and component of cartilage similar to the case of normal mice. These histological examination results accordantly indicated excellent outcomes of NIR-photocatalytic therapy based on HTON in good agreement with abovementioned observation results (Fig. 5). Moreover, major organs including the injected joints, the adjacent lymph, heart, liver, spleen, and lung were extracted at the different time points after intra-articular injection of HTON to investigate the clearance of HTON. From fig. S22, it can be found that HTON can be cleared from the joints gradually by lymph transportation, and about 50% of injected HTON can be cleared after intra-articular injection for 3 weeks. In addition, the NIR-photocatalytic treatment with HTON exhibited high compatibility to normal tissues and blood (figs. S23 to S25).

In vivo SME regulation effect of HTON

Furthermore, the effect of NIR-photocatalytic therapy on arthritic SME was investigated. The inflammatory phenotype of synoviocytes and macrophages and the level of LA and H₂ in synovial tissue were evaluated. In the blank control group (PBS) in Fig. 7A, synoviocytes exhibited an obvious inflammatory phenotype with high levels of THY1 and FAPα, and synovial pannus was formed with plenty of abnormal vascularization (CD34) in synovial tissue. From Fig. 7B and fig. S26, there were large numbers of macrophages in an inflammatory phenotype (CD68, iNOS, and CD80) in the synovial tissue of blank control group. Other two control groups (PBS + NIR and HTON) made no influence on both synovial pannus and phenotypes of synoviocytes and macrophages (Fig. 7 and fig. S26). By comparison, NIR-photocatalytic therapy (HTON + NIR) distinctly improved the synovial pannus and inhibited the inflammatory phenotype of synoviocytes (Fig. 7A) and converted M1 phenotype of macrophages into M2 phenotype with much reduced CD68/iNOS/CD80 expressions and enhanced CD206/MerTK expressions (Fig. 7B). Such an SME regulation effect of HTON could be highly related to NIR-photocatalytic hydrogen generation and LA consumption (Fig. 5b) in good accordance with in vitro results (Fig. 4). NIR-photocatalytically generated molecular hydrogen could effectively scavenge ROS to mediate the anti-inflammatory polarization of macrophages in SME for anti-inflammation, and meanwhile, the NIR-photocatalytic consumption of overexpressed LA can correct metabolic disorder in SME to a certain extent to inhibit the inflammatory/invasive phenotypes of synoviocytes and macrophages and the abnormal proliferation of synoviocytes for the prevention of synovial pannus formation. Abovementioned excellent outcomes of RA therapy were possibly owing to the clear SME regulation effect of HTON-based NIR-photocatalysis.

Fig. 7. The effect of NIR-photocatalytic therapy on the arthritic SME.

The effect of HTON treatment on synovial pannus with abnormal vascularization (CD34) and inflammatory phenotype of synoviocytes (A) and macrophages (B) in synovial tissue and corresponding quantitative analysis (C). P values were calculated by the two-tailed Student’s t test (***P < 0.001). Data were presented as means ± SD. Scale bars 100 μm (A and B). DAPI, 4′,6-diamidino-2-phenylindole.

DISCUSSION

The incorporation of titanium dioxide is generally executed by facile calcination routes, but hard aggregation among calcined titanium dioxide nanoparticles cannot be avoided, limiting their biomedical application. Here, the developed full-solution method maintains the aqueous dispersion of TON after hydrogen incorporation. During the incorporation, the use of the ethylene diamine solution of Li as an electronic solution is the key and can be applied to the incorporation of other materials for biomedical application.

Titanium dioxide has high biocompatibility and has been widely used as an additive to medicine and cosmetics. After hydrogen incorporation with slight hydrogen contents, the synthesized HTON has preserved high biocompatibility, ensuring its potential and promising bioapplications as a specific biomedical photocatalyst. After intra-articular injection, HTON can be excreted from articular cavity through the lymph route, which is possibly mediated by the phagocytosis and transportation of macrophages. Moreover, residual HTON within 1 month exhibits excellent tissue compatibility. These biosafety profiles will favor the clinical translation of HTON.

The used NIR power density of ≤0.5 W/cm² has proven safe to skin tissue without visible damage. Moreover, the level of LA in RME is considerably high, which provides enough large numbers of LA as a sacrificial agent for NIR-photocatalytic hydrogen generation, ensuring sustainable NIR-photocatalytic therapy. Compared with most previously reported therapeutic NIR-photocatalysts and hydrogen donors such as Mg particles, HTON for RME regulation and RA treatment neither needs to carry any sacrificial agent nor is exhausted, realizing longtime sustainable and anytime repeatable therapy. Our groundbreaking discoveries will bring the catalytic nanomedicine closer toward clinical practice.

It is worth noting that the NIR-photocatalytic treatment on the HTON-injected knee joint can not only improve the symptom of RA locally but also brings positive effects on remote distant paws. It means that the vicious circle of RA is possibly corrected to form a positive feedback in support of systemic improvement. Such a therapeutic effect has an important clinical value and can bring inspiration for developing the strategy of RA treatment.

In summary, we developed the HTON nanocatalyst by the full-solution method to realize the NIR-photocatalytic generation of hydrogen molecules and simultaneous LA depletion for arthritic SME regulation. Successful hydrogen incorporation into HTON enabled efficient NIR-photocatalysis and also maintained high aqueous dispersivity in favor of its biomedical applications. The HTON nanocatalyst exhibited controlled NIR-photocatalytic hydrogen generation and LA consumption in vitro and in vivo, effectively intercepted the evolution of RA and blocked bone destruction in the CIA mouse model by inhibiting the inflammatory/invasive phenotypes of synoviocytes and macrophages as well as the abnormal proliferation of synoviocytes. By the photocatalytic SME regulation strategy, the HTON nanocatalyst obtained great benefits in the efficacy of RA treatment. The proposed catalyst-based arthritis therapy is a promising strategy for clinical arthritis treatment.

MATERIALS AND METHODS

Material

All chemicals with an analytical-reagent grade were used without further purification, including titanium tetrachloride (TiCl₄; Shanghai Macklin Biochemical), ethylenediamine (Macklin Biochemical), silane–polyethylene glycol (PEG; MW, 5000Da; Aladdin Biochemical), Li foil (Aladdin Biochemical), LA detection kit (Sigma-Aldrich), cell counting kit-8 (CCK-8; Beyotime Biotechnology), 2’,7’-dichlorodihydrofluorescein diacetate (DCFH-DA; Beyotime Biotechnology), heat-killed Mycobacterium tuberculosis strain H37Ra (BD Biosciences), arlacel and heavy paraffin mineral oil (Thermo Fisher Scientific), and type II collagen (Chondrex). FLS, RAW264.7 macrophages, and chondrocytes were purchased from Procell Life Science and Technology Co. Ltd.

Synthesis and characterization of TONs and HTONs

Rutile TONs were synthesized by a solvothermal method. In a typical procedure, 10 ml of TiCl₄ was dropwise added into 30 ml of ice water in an ice bath under stirring to prepare a transparent solution. After further stirring for 30 min, 30 ml of TiCl₄ aqueous solution was added to 50 ml of Teflon-lined autoclave for static crystallization at 453 K for 24 hours. After natural cooling to room temperature, the precipitate after crystallization was separated by centrifugation, washed once with water and three times with anhydrous ethylenediamine in turn, and then ultrasonically dispersed in anhydrous ethylenediamine for the following reaction.

Next, Li was inset into TONs and then replaced with hydrogen to obtain HTONs. A metallic Li foil (14 mg) was completely dissolved in ethylenediamine (20 ml) to form a solvated electron solution, and then the dried TONs (200 mg) were added under stirring under Ar protection. After reaction for 6 days, 30 ml of aqueous solution of HCl (1 M) was slowly dropped into the above solution to quench excessive electrons. Last, the obtained HTONs were collected by centrifugation and then washed with deionized water and anhydrous ethanol in turn for several times. Then, HTONs were modified with PEG. Silane-PEG (5 mg) was resolved into the ethanol solution of HTONs (5 ml, 4 mg/ml) under stirring, followed by sonication for 1 hour. After stirring for another 24 hours, the precipitate was collected and then washed with ethanol.

The morphology and size of nanorods were characterized by scanning electron microscopy (APREO, FEI) and TEM (JEM-2100F). The hydrodynamic size of nanorods was measured on a Malvern Zetasizer Nano ZS90. The composition of nanorods was determined by FTIR on a Thermo-Nicolet Nexus 670 attenuated total reflectance (ATR)–IR spectrometer and by Raman spectroscopy on a Raman imaging microscope (Dxr3xi, Thermo Fisher Scientific). XRD patterns were recorded on an M21X diffractometer (Cu Kα, λ = 1.54056 Å) operated at 40 kV and 200 mA at room temperature. XPS data were collected on a PerkinElmer PHI 5000C system equipped with a hemispherical electron energy analyzer at 15 kV and 20 mA, where the C 1s line (284.6 eV) was used as the reference to calibrate binding energy.

Measurement of photocatalytic hydrogen generation and LA depletion

A specifically made tube, which was separated by a piece of filter membrane (MWCO = 3000 Da) in the middle was used for hydrogen detection. On one side, hydrogen concentration was quantitatively monitored with a hydrogen microelectrode (Unisense, Denmark). In the simulated solution, the reaction solution of nanorods (100 μg/ml) and LA (1 mM) was put on one side of tube and completely covered with the light spot of an 808-nm laser, while a hydrogen microelectrode was set in the detection reaction solution on the another side of tube. The controllability of NIR-triggered hydrogen generation was measured by repeatedly switching on/off the laser at fixed time points in the dark. In addition, LA concentration during photocatalytic reaction was quantitatively determined using an LA detection kit (Sigma-Aldrich). In the simulated solution, the natural light–sealed reaction solutions of nanorods (100 μg/ml) and LA (1 mM) were irradiated in the dark with an 808-nm laser at different power densities (0, 0.3, and 0.5 W/cm²) for different time periods (5 to 30 min). The nanoparticles in the solution were removed by centrifugation, and the super clear solution was collected for LA detection using the LA kit.

In vitro cytotoxicity assessment

Chondrocytes were seeded in 96-well cell culture plates at the density of 5 × 10⁴ cells per well and incubated with various concentrations of LA or/and nanorods for 12 hours at 37°C in a humidified 5% CO₂ atmosphere. After incubation for another 24 hours, cell viability was measured on a microplate reader (BioTek) using the standard CCK-8 assay. Moreover, cells were also treated with HTON at different particle concentrations together with LA (1 mM) to investigate the therapeutic window concentration of HTON under the NIR-irradiation conditions (808 nm, 0.5 W/cm², 6 × 5 min). In addition, gene expression profile was investigated by RNA sequencing analysis.

To evaluate the effect of LA on the proliferation of FLS, joint synovial cells were seeded in 96-well cell culture plates at the density of 1 × 10⁴ cells per well and incubated with various concentrations (0, 0.5, and 1.0 mM) of LA for 24 hours at 37°C in a humidified 5% CO₂ atmosphere. After incubation for another 24 hours, the standard CCK-8 assay was executed to detect cell viability.

Evaluation of the inflammatory phenotype of FLS, macrophages, and chondrocytes in vitro

First, the effects of LA and ROS on the phenotype of FLS and RAW264.7 macrophages were investigated, respectively. Synovial cells were seeded in a glass-bottomed confocal dish at a density of 1 × 10⁴ cells per well and incubated with various concentrations (0, 0.5, and 1 mM) of LA for 24 hours at 37°C in a humidified 5% CO₂ atmosphere. After incubation for another 24 hours, cells were stained with primary antibody against podoplanin (Abcam) and Alexa Fluor 488–labeled secondary antibody (Abcam) in turn, followed by confocal observation. In addition, RAW264.7 macrophages were treated with lipopolysaccharide or/and hydrogen gas, and then intracellular ROS levels were detected with DCFH-DA (Beyotime Biotechnology), and their phenotypes were determined with antibodies against MMP-3 and CD206 (Abcam) and by the real-time PCR technique.

Next, the effects of HTON-mediated photocatalytic consumption of LA on the inflammatory phenotype of FLS, macrophages, and chondrocytes were investigated in vitro. Synovial cells were seeded in a glass-bottomed confocal dish at a density of 1 × 10⁴ cells per well, then stimulated with 1 mM LA, and lastly treated with HTONs (final concentration, 50 μg/ml) followed by irradiation with the NIR laser (808 nm, 0.5 W/cm², 5 min every time at six times) in the dark. After incubation for 24 hours, synovial cells were stained with a primary antibody against FAPα or IL-1β (Abcam) and an Alexa Fluor 488–labeled corresponding secondary antibody (Abcam). In addition, synovial cells were stimulated with 1 mM LA in a humidified 5% CO₂ atmosphere in the routine incubator or in a special H₂ incubator to investigate the effect of H₂ on the phenotype of FLS, whose phenotypes were determined with the antibodies against MMP-3 and IL-1β (Abcam).

Joint synovial cells and RAW264.7 macrophages or chondrocytes were inoculated on the insert and bottom well of a dual-chamber Transwell plate with a 3-μm-thick microporous membrane (Corning). These cells were cocultured with cell culture medium containing 1 mM LA and then treated with HTONs (final concentration, 50 μg/ml) followed by irradiation with the NIR laser (808 nm, 0.5 W/cm², 6 × 5 min) in the dark. After incubation for 24 hours, RAW264.7 macrophages were stained individually with various primary antibodies against iNOS, CD206, MMP-3, and CD163 (Abcam) and then with Alexa Fluor 488–labeled corresponding secondary antibody (Abcam), while primary antibodies against COX2, MMP-3, ICAM-1, and Aggrecan were used for chondrocyte treatment.

CIA mouse model building protocol

A CIA mouse model was established by following a standard protocol with immunization. CFA was prepared by mixing heat-killed M. tuberculosis strain H37Ra (4 mg/ml; BD Biosciences) with incomplete Freund’s adjuvant (IFA; 15% arlacel + 85% heavy paraffin mineral oil; Thermo Fisher Scientific). Then, CFA and type II collagen were dissolved in diluted acetic acid (2 mg/ml) followed by thorough mixing by a homogenizer in an ice bath. Next, the mixture (100 μl) was injected intradermally into the tail of DBA/1 mice to achieve immunization. On day 21, immunization was reinforced by injecting a mixed emulsion of IFA and collagen II without heat-killed binding mycobacteria for CIA modeling. All the in vivo experiments followed the protocols approved by the Animal Care and Use Committee of the Shenzhen University.

In vivo therapy protocol

To investigate therapeutic efficacy with CIA mice, two experienced researchers randomly divided the CIA mice into four groups (n = 7), and then the PBS solution of HTONs (20 μl, 2 mg/ml) was injected into each knee joint of the hind legs on day 28 after RA induction, and then the joints were irradiated with the NIR laser (808 nm, 0.3 W/cm², 6 × 5 min) in the dark once every week. Sterile PBS as a negative control was injected intra-articularly into the knee joint on the same day. In addition, knee joints were extracted for measurement of the intra-articular LA and H₂ levels by the LA kit and gas chromatography immediately after NIR irradiation, respectively, while the samples from CIA and normal mice without NIR irradiation were used as two controls. Moreover, main organs were extracted at fixed time points after intra-articular injection of HTON to investigate the clearance of HTON by inductively coupled plasma–atomic emission spectrometry (Agilent Technologies, USA).

Arthritis scoring assessment

From day 0 to day 60, the length, width, and thickness of mouse paws were measured every 2 days with a digital caliper. The paw volume was approximately represented with the product of these three parameters by assuming that each parameter was equally important to reflect the degree of inflammation in the paw. The arthritis score (0 to 4) was blind-evaluated by unsuspecting researchers according to the following criteria: 0, normal; 1, mild redness of the ankle or tarsal joint; 2, mild redness and swelling of the ankle to the tarsal bone; 3, moderate swelling from ankle to the metatarsal joints; 4, severe swelling of ankles, paws, and fingers. At the end of the treatment, the diameter of the posterior ankle joint was measured with the digital caliper. One-way analysis of variance was used for data statistics, and the Kruskal-Wallis test was used to analyze the arthritis scoring data.

Micro-CT examination

Hindlimbs were imaged using a Skyscan 1172 micro-CT scanner (Bruker) under 600-kV and 167-μA x-ray beam. Projections were taken every 0.45° under 600-ms exposure, resulting in an image pixel size of 13.59 μm. Image volumes were reconstructed with the beam hardening correction using the Feldkamp algorithm (NRecon software, v1.6.1.5, Buker). The radiodensity of −300 to 3000 Hounsfield unit (HU) was chosen to isolate the bony structure from the imaging medium, and a CT Analyzer software v1.12 (Bruker) was used to extract an isosurface mesh representation of the reconstructed mirco-CT slices, followed by modification with MeshLab v1.3.2 (an open source software developed with the support of the 3D CoForm project).

Histological analysis of mouse knee joints

At the treatment end point, mice were euthanized, and their hind knee joints were collected for H&E, Safranin−fast green, and Alcian blue−periodic acid Schiff staining, followed by microscopic observation. To analyze FLS activation, knee joint sections were stained with rabbit antibodies against CD34, THY1, and FAP-α. Moreover, the rabbit antibodies against CD68, CD206, MerTK, iNOS, and CD80 were used for immunofluorescence costaining to analyze the phenotype of macrophages.

Assessment of liver/kidney functions and hemotoxicity

Health BALB/c mice were randomly divided into two groups (n = 5) for in vivo toxicity evaluation of HTON by intra-articular HTON injection. After 2 weeks, the blood was collected and assessed using a biochemical analyzer (iMagic-M7) and a blood cell analyzer (BC-31s, Mindray).

Supplementary Materials

This PDF file includes:

Figs. S1 to S26

View/request a protocol for this paper from Bio-protocol.