Abstract

   Dysregulation of the JAK2/STAT3 pathway disrupts immune balance and
   tissue homeostasis, leading to persistent inflammation and subsequent
   bone loss. Yet, commonly used JAK2 inhibitors with hydrophobicity
   primarily suppress inflammation but often fail to provide sustained
   pharmacological effects and address ongoing inflammation-induced bone
   destruction. Here, using computer-aided drug design, we developed
   hydrophilic calcium-doped carbon dots (Ca-CDs) as a multifunctional
   nanoinhibitor targeting JAK2. The Ca-CDs can block the excessive
   activation of the JAK2/STAT3 pathway by binding to JAK2, thereby
   reducing the secretion of pro-inflammatory cytokines and exerting
   anti-inflammatory effects. Furthermore, the Ca-CDs promote bone
   regeneration under inflammatory conditions and serve as crosslinking
   junctions that facilitate the formation of a Ca-CDs-based alginate
   hydrogel, thereby enabling prolonged drug retention and controlled
   release. Their application in the representative inflammatory
   microenvironment of periodontitis successfully validated the efficacy
   of the Ca-CDs-based therapeutic strategy. Overall, targeting JAK2 with
   Ca-CDs nanoinhibitor offers a promising treatment option for
   JAK2-associated inflammatory diseases.

Graphical Abstract

   [50]graphic file with name 12951_2025_3744_Figa_HTML.jpg

Supplementary Information

   The online version contains supplementary material available at
   10.1186/s12951-025-03744-w.

   Keywords: Carbon dots, JAK2 nanoinhibitor, Sustained release,
   Anti-inflammation, Bone repair

Introduction

   Janus protein tyrosine kinases (JAK1, JAK2, JAK3, and TYK2) are crucial
   mediators of the Janus kinase/signal transducer and activator of
   transcription (JAK/STAT) pathway, essential for transmitting
   extracellular cytokine-mediated signals across the cell membrane to the
   nucleus, ultimately regulating gene transcription [[51]1–[52]3]. This
   pathway is pivotal for maintaining hematopoiesis, immune balance, and
   tissue homeostasis. Dysregulation of JAK/STAT signaling is closely
   associated with inflammatory and immune diseases such as rheumatoid
   arthritis, periodontitis, and inflammatory bowel disease [[53]4–[54]6].
   In particular, the overactivation of the cytokine-mediated JAK2/STAT3
   pathway induces the expansion and persistence of inflammation, which in
   turn causes the continuous destruction of bone in these diseases
   [[55]7]. Given its critical role in human health, targeting this
   pathway presents a significant therapeutic opportunity for managing a
   broad spectrum of conditions [[56]8].

   Currently, commonly used JAK2 inhibitors are small molecules
   characterized by NH/O spacers, amide bonds, and N-heterocyclic
   structures [[57]9–[58]11]. These inhibitors are typically administered
   orally due to their poor water solubility, which presents several
   limitations, including side effects and low therapeutic concentrations
   at target sites [[59]12–[60]14]. This issue is particularly problematic
   for inflammatory diseases requiring localized treatment, such as
   periodontitis, where hydrophobic inhibitors fail to remain at the
   lesion site, thus diminishing their effectiveness [[61]15]. More
   importantly, these inhibitors lack multifunctionality and fail to
   address the ongoing bone destruction driven by inflammation. Therefore,
   to overcome these challenges, there is a significant need to design an
   inhibitor that not only more effectively regulates the JAK2/STAT3
   pathway to suppress the secretion of pro-inflammatory cytokines but
   also prevents bone loss and promotes bone regeneration associated with
   inflammation [[62]16].

   Carbon dots (CDs) have garnered extensive attention due to their unique
   advantages [[63]17–[64]19], such as the ability to engineer various
   functional groups by selecting specific chemical structures as
   precursors [[65]20–[66]22], potentially allowing direct involvement in
   cellular signaling regulation. In this study, leveraging the structures
   of JAK2 protein and known inhibitors, pharmacophore models were
   constructed using the HipHop algorithm, and potential fragments that
   fit well within the JAK2 binding site were identified through
   fragment-based drug discovery. Consequently, calcium-doped CDs
   (Ca-CDs), a multifunctional nanoinhibitor targeting JAK2, were
   synthesized through space-confined vacuum heating of acetylsalicylic
   acid, metformin, and calcium chloride. The synthesized Ca-CDs have been
   demonstrated to regulate the JAK2/STAT3 pathway effectively, thereby
   inhibiting the secretion of pro-inflammatory cytokines. The capability
   of Ca-CDs to control the JAK2/STAT3 pathway mainly stems from their
   efficient binding to JAK2 protein, which was elucidated through
   molecular dynamics simulation and surface plasmon resonance (SPR)
   analysis, exhibiting an affinity dissociation constant (K[D]) of 64.2
   µM. Additionally, the calcium incorporated within the Ca-CDs
   facilitates the repair of bone defects caused by inflammation.
   Significantly, the inherent calcium in the Ca-CDs can trigger the
   formation of a Ca-CDs-based alginate (Ca-CDs@ALG) hydrogel, where
   Ca-CDs serve as crosslinking junctions, thereby enabling long-term
   retention at local lesion sites and sustained drug release. To validate
   the efficacy of Ca-CDs, periodontitis, a condition requiring local
   administration and prevalent in moist oral environments, was selected
   as a model. In vivo experiments demonstrated that our nanoinhibitor
   could effectively reduce the inflammatory response and promote the
   regeneration of bone defects caused by inflammation, showcasing its
   substantial potential for clinical applications in JAK2-associated
   inflammatory diseases (Fig. [67]1).

Fig. 1.

   [68]Fig. 1
   [69]Open in a new tab

   Schematic illustration of the design strategy for Ca-CDs targeting JAK2
   and their anti-inflammatory and osteogenic mechanisms for the sustained
   treatment of inflammatory diseases. Ca-CDs, computer-aided designed as
   a multifunctional nanoinhibitor targeting the JAK2/STAT3 pathway,
   reduce pro-inflammatory cytokines secretion and promote osteogenic
   differentiation. The Ca-CDs also serve as crosslinking junctions to
   form the Ca-CDs@ALG hydrogel, enhancing therapeutic efficacy in vivo

Results

Computer-aided drug design, synthesis and characterization of Ca-CDs

   A chemically diverse set of 53 known JAK2 inhibitors was selected as
   the training set to generate a ligand-based common feature
   pharmacophore model using the HipHop algorithm. The selected compounds
   displayed key features such as hydrogen bond acceptors (HBA), hydrogen
   bond donors (HBD), ring aromaticity (RA), and hydrophobic groups
   (HYD) associated with their aromatic structures (Fig. S1).
   Additionally, all training set compounds shared an essential
   nitrogen-containing heterocyclic functionality. After calculations, ten
   pharmacophore models were generated, and the top two were shown based
   on their rank values (Fig. [70]2A and Video S1). Analysis of the
   resulting pharmacophore models revealed that, besides the presence of
   HBA, HBD, RA, and HYD features, HBA and RA were adjacent in
   pharmacophore model 1, while HBD and RA were adjacent in pharmacophore
   model 2, suggesting the presence of aromatic rings containing
   heteroatoms. Furthermore, fragment-based drug design was performed
   using the structures of the 53 inhibitors and the JAK2 protein from the
   UniProt database (ID: [71]Q62120) (Fig. [72]2B and Video S2). The
   results indicated that suitable fragments for interacting with JAK2
   primarily contained nitrogen and/or oxygen aromatic rings
   (Fig. [73]2C). Therefore, the aforementioned pharmacophore model and
   fragment-based drug design can serve as a preliminary screening
   strategy, guiding the selection of precursors capable of forming such
   heterocyclic structures for CDs synthesis, thereby greatly reducing the
   workload of experimental validation.

Fig. 2.

   [74]Fig. 2
   [75]Open in a new tab

   Computer-aided drug design and characterizations of Ca-CDs. (A) Top 2
   ranked pharmacophore models, highlighting key structural features: HBA
   (green), HBD (pink), RA (orange), and HYD aromatic (blue). (B) A
   general view of the three-dimensional structure of JAK2, shown as a
   rainbow-colored cartoon (left panels) and a gray-shaded surface model
   (right panels). A close-up highlights its binding site, represented by
   a purple sphere. (C) Five fragments suitable for interacting with the
   JAK2 binding site (upper panels) and their structures formulas (lower
   panels). JAK2 is depicted as a rainbow-colored cartoon and gray-shaded
   surface. The fragments are shown in ball-and-stick mode, with carbon,
   hydrogen, oxygen, and nitrogen atoms colored gray, white, red, and
   blue, respectively. (D) Schematic diagram of Ca-CDs synthesis (ASA:
   acetylsalicylic acid). Photographs of (E) Ca-CDs sphere and (F) Ca-CDs
   solution taken under sunlight (upper panels) and UV light (lower
   panels), respectively. (G) Excitation-emission map of Ca-CDs. (H) HRTEM
   image of Ca-CDs, with insets showing the corresponding particle size
   distribution histogram and an enlarged view of the Ca-CDs lattice. (I)
   FTIR spectra of acetylsalicylic acid (purple), metformin (blue), and
   Ca-CDs (red). The enlarged image shows the peak fitting of the FTIR
   spectra of Ca-CDs in the 1600–1800 cm^−1 range. High-resolution XPS
   spectra of Ca-CDs: (J) C 1 s and (K) N 1s. (L) EDS pattern of Ca-CDs.
   Inset: the flame test image of Ca-CDs solution

   Based on the above analysis results, multiple compounds containing
   nitrogen and/or oxygen aromatic rings from our drug library were
   selected as candidate precursors for CD synthesis and pharmacological
   evaluation (Fig. [76]S2). Consequently, acetylsalicylic acid and
   metformin were selected as organic precursors for preparing CDs. Using
   our previously developed space-confined vacuum heating method [[77]23],
   a solution containing acetylsalicylic acid, metformin, and calcium
   chloride was heated and evacuated (Fig. [78]2D), resulting in the
   formation of a foamy, porous spherical structure (Fig. [79]S3).
   Continued heating of the reaction system yielded a porous yellow
   spherical product that emitted bright fluorescence under ultraviolet
   (UV) excitation (Fig. [80]2E). After dialysis and purification, a light
   yellow Ca-CDs solution was obtained (Fig. [81]2F). The
   ultraviolet-visible (UV-vis) absorption spectrum and photoluminescence
   (PL) spectrum of this solution differed from those of the raw materials
   (Fig. [82]S4 and S5), confirming the successful synthesis of Ca-CDs.
   Additionally, the excitation-emission map presented a single emission
   center (Fig. [83]2G), demonstrating size uniformity, further confirmed
   by high-resolution transmission electron microscopy (HRTEM)
   (Fig. [84]2H). The HRTEM image revealed that the Ca-CDs are uniformly
   sized, with an average diameter of 3.5 nm. The well-defined lattice
   spacing of 0.24 nm corresponds to the (110) crystallographic planes of
   graphitic carbon, confirming the formation of a graphene-like,
   conjugated aromatic structure. To further clarify the structure of the
   Ca-CDs, Fourier transform infrared (FTIR) spectroscopy and X-ray
   photoelectron spectroscopy (XPS) characterization were performed.

   As depicted in Fig. [85]2I, the FTIR spectrum of acetylsalicylic acid
   reveals the presence of carboxyl group, evidenced by a C = O stretching
   vibration at 1690 cm^−1. The peaks at 1458 and.1605 cm^−1 indicate the
   presence of the carbon framework in benzene ring, which provides the
   structural basis for the formation of conjugated structures in Ca-CDs.
   The FTIR spectrum of metformin displays characteristics of amine group,
   confirmed through an N-H stretching vibration between 3040 and
   3450 cm^−1, and an N-H bending vibration at 1570 cm^−1. Within the FTIR
   spectrum of Ca-CDs, a broad band peak ranging from 2800 to 3500 cm^−1
   is attributed to O-H/N-H stretching vibrations. The peaks at 1654 cm^−1
   and 1558 cm^−1 correspond to C = O stretching vibrations and N-H
   bending vibrations, respectively, which are associated with the amide I
   and amide II bands. These findings indicate the presence of amide bonds
   in Ca-CDs, formed through dehydration condensation between carboxyl and
   amine groups in the raw materials. Additionally, unsaturated
   carbon-nitrogen bond stretching vibrations are observed at 1631 and
   2183 cm^−1, with C-N stretching vibration also detected at 1366 cm^−1,
   confirming the presence of arylamine [[86]24]. Furthermore, peaks at
   1611 and 1466 cm^−1 are associated with the carbon framework (i.e.,
   C = C stretching vibrations) in conjugated structures. The presence of
   numerous functional groups within the Ca-CDs is further substantiated
   by XPS characterizations.

   The C 1 s spectrum of acetylsalicylic acid showed four peaks at 284.7,
   285.1, 286.3, and 288.8 eV, indicating the presence of sp^2 C, sp^3 C,
   C-O, and COO, respectively (Fig. S6A). The C 1 s spectrum of metformin
   displays three peaks at 284.9, 285.9, and 288.3 eV, indicative of sp^3
   C, C-N, and guanidino groups (Fig. S6B). The N 1 s spectrum of
   metformin reveals three peaks at 398.4, 399.4, and 400.2 eV,
   corresponding to N-H, C-N, and C = N, respectively (Fig. S7). In
   comparison, as shown in Fig. [87]2J, K, the C 1 s spectrum of the
   Ca-CDs shows six peaks at 284.6, 285.1, 285.8, 286.4, 287.6, and
   288.9 eV, corresponding to sp^2 C, sp^3 C, C-N, C-O, CON, and
   COO/guanidino groups, respectively (Fig. [88]2J). The N 1 s spectrum of
   Ca-CDs displays four peaks at 398.4, 399.5, 400.3, and 402.3 eV,
   representing N-H, pyrrolic N, pyridinic N, and graphitic N/CON,
   respectively (Fig. [89]2K). These findings underscore the amide
   formation and the presence of a highly conjugated sp^2 domain and
   heteroatom-containing rings in Ca-CDs, setting the stage for specific
   interactions with target protein.

   Energy-dispersive spectroscopy (EDS) characterization confirmed the
   presence of C, N, O, and Ca elements in the Ca-CDs, with the atomic
   ratio of Ca being 9.1% (Fig. [90]2L). To further verify the presence of
   calcium visually, a flame test was conducted, resulting in a brick-red
   flame, which is characteristic of calcium (Fig. [91]2L and S8; Video
   S3). The presence of calcium provides favorable conditions for the
   regeneration of bone damaged by inflammation.

Biocompatibility and anti-inflammatory effects of Ca-CDs

   Before applying to in vivo and in vitro studies, the biocompatibility
   of Ca-CDs should be assessed. A cell counting kit-8 (CCK-8) was
   utilized to evaluate the cytotoxicity of RAW264.7 cells treated with
   Ca-CDs for 24 h. The results showed that Ca-CDs had no significant
   effect on the viability of RAW264.7 cells within the concentration
   range of 0–300 µg/mL. However, the viability decreased to 84.7% at a
   concentration of 400 µg/mL (Fig. [92]3A). Flow cytometry and live/dead
   cell staining further demonstrated that there were no significant
   changes in apoptosis and morphology of RAW264.7 cells at working
   concentrations ranging from 0 to 300 µg/mL (Fig. [93]3B and S9). These
   findings suggest that Ca-CDs have good biocompatibility. To directly
   observe the distribution of Ca-CDs within cells, the unique PL
   properties of Ca-CDs were employed for cell imaging using confocal
   laser scanning microscopy (CLSM). As shown in Fig. [94]3C, the blue
   fluorescence of Ca-CDs was primarily observed in the cytoplasm,
   indicating that the Ca-CDs can be uptaken and internalized by the
   RAW264.7 cells. This establishes a proof of concept for the
   pharmacological functionality and further research both in vivo and in
   vitro of Ca-CDs.

Fig. 3.

   [95]Fig. 3
   [96]Open in a new tab

   Ca-CDs exert anti-inflammatory effects via regulating JAK2/STAT3
   pathway. (A) CCK-8 assay of RAW264.7 cells, which are treated with
   different concentrations of Ca-CDs (0, 50, 100, 200, 300, and
   400 µg/mL). (B) Cell apoptosis assay of RAW264.7 cells, which are
   treated with different concentrations of Ca-CDs (0, 100, 200, and
   300 µg/mL), respectively. (C) Cell imaging of Ca-CDs-treated RAW264.7
   cells. (D) Detection of mRNA levels of Nos2, Il1b, and Tnfa by RT-qPCR.
   (E) Western blot analysis of pro-inflammatory cytokines: iNOS, IL-1β,
   and TNF-α. (F) Corresponding quantitative data from Western blot. (G)
   Volcano diagram (q-value < 0.05 & |log[2]FC|>1) and (H) heat map of
   differentially expressed genes. (I) Bubble diagram of KEGG pathway
   enrichment analysis. (J) GSEA enrichment plot of JAK/STAT pathway. (K)
   Western blot results of p-STAT3, STAT3, IL-1β, and TNF-α treated
   without or with LPS, Ca-CDs (300 µg/mL), and Colivelin (500 nM),
   respectively. (L) Western blot results of p-JAK2, JAK2, p-STAT3, STAT3,
   IL-1β, and TNF-α treated without or with LPS, Ca-CDs (300 µg/mL), and
   Coumermycin A1 (CA1, 10 µM), respectively. (M) Schematic diagram of
   Ca-CDs inhibiting pro-inflammatory cytokines secretion via regulating
   JAK2/STAT3 pathway. Data are presented as mean ± SD from three
   independent experiments. * indicates P < 0.05, ** indicates P < 0.01,
   *** indicates P < 0.001

   Considering that Ca-CDs exhibit structural similarities in their
   functional groups with JAK2 inhibitors, it is anticipated that Ca-CDs
   may possess anti-inflammatory effects. To investigate the effects of
   Ca-CDs on the secretion of pro-inflammatory cytokines in RAW264.7
   cells, lipopolysaccharide (LPS) was used to establish an inflammatory
   model in vitro. The cells were then co-incubated with Ca-CDs and
   another group received acetylsalicylic acid and metformin (ASA + MET)
   for 24 h, respectively. Reverse-transcriptase quantitative polymerase
   chain reaction (RT-qPCR) results indicated that compared to the LPS
   group, Ca-CDs significantly reduced the mRNA expression of Nos2, Il1b,
   and Tnfa, which are representative pro-inflammatory cytokines, as the
   concentration of Ca-CDs increased from 0 to 300 µg/mL (Fig. [97]3D).
   Western blot analysis further confirmed a significant reduction in the
   expression of iNOS, IL-1β, and TNF-α when treated with Ca-CDs
   (Fig. [98]3E, F). Moreover, these reductions were more pronounced than
   those observed in the ASA + MET group. This suggests that Ca-CDs
   possess potent pharmacological activity. Given that Ca-CDs are explored
   as a JAK2 nanoinhibitor, their anti-inflammatory effects were compared
   with those of Fedratinib, a clinically approved JAK2 inhibitor. At
   biosafe concentrations, RT-qPCR analysis revealed that our Ca-CDs
   exhibited anti-inflammatory activity comparable to that of the
   commercially available Fedratinib (Fig. S10), demonstrating the
   potential of Ca-CDs as a JAK2 nanoinhibitor.

Ca-CDs inhibit the JAK2/STAT3 pathway to exert anti-inflammatory effects

   Having established the anti-inflammatory effects of Ca-CDs, RNA
   transcriptome analysis was conducted to explore the underlying
   mechanisms of this biological phenomenon. Compared to the LPS group,
   the addition of Ca-CDs to LPS-induced RAW264.7 cells resulted in the
   identification of 97 differentially expressed genes, including 12
   upregulated and 85 downregulated genes (Fig. [99]3G). As depicted in
   Fig. [100]3H, the heat map shows significant downregulation of
   cytokines and chemokines receptors related to inflammation, such as
   Il6, Il23a, Lif, Il1a, Il1b, Cxcr5, and Ccr7. Further analysis through
   Gene Ontology (GO) enrichment and Kyoto Encyclopedia of Genes and
   Genomes (KEGG) pathway enrichment of these differentially expressed
   genes aimed to identify the main biological functions and pathways
   involved. It was found that the Ca-CDs primarily participate in immune
   and inflammatory responses and are predominantly enriched in
   cytokine-related pathways, particularly the JAK/STAT pathway
   (Fig. [101]3I and S11). Gene Set Enrichment Analysis (GSEA) revealed
   substantial activation of the JAK/STAT pathway following LPS
   stimulation, which was effectively inhibited by Ca-CDs (Fig. [102]3J).
   Given the heterocyclic structures in Ca-CDs that are similar to those
   of JAK2 inhibitors, it is hypothesized that Ca-CDs may exert their
   anti-inflammatory effects by inhibiting the JAK2/STAT3 pathway. To
   validate this hypothesis, Western blot experiments were conducted.

   We first examined whether the expression and activation of the STAT3 in
   the JAK2/STAT3 pathway could be regulated by Ca-CDs. The results
   indicated that, compared to the control group, the phosphorylation
   level of STAT3 was significantly increased in the LPS group, confirming
   STAT3 activation during the inflammatory process (Fig. [103]3K and
   S12). However, after treating LPS-induced cells with Ca-CDs, both the
   phosphorylation level of STAT3 and pro-inflammatory cytokines
   expression were significantly reduced. To further explore whether the
   anti-inflammatory effects of Ca-CDs could be reversed, Colivelin, an
   STAT3 activator, was used [[104]25]. As shown in Fig. [105]3K and S12,
   when Colivelin was added to the LPS group, there were no significant
   variations in the phosphorylation level of STAT3, nor were there
   changes in the expression of IL-1β and TNF-α. However, when Colivelin
   and Ca-CDs were simultaneously administered to LPS-induced cells, there
   was a noticeable increase in the phosphorylation level of STAT3
   compared to the LPS-induced cells treated solely with Ca-CDs, along
   with elevated expression of IL-1β and TNF-α. These results imply that
   the downregulation of pro-inflammatory cytokines by Ca-CDs is achieved
   through the inhibition of phosphorylation of STAT3.

   To investigate whether the inhibition of STAT3 phosphorylation is due
   to the upstream blockade of JAK2 by Ca-CDs, Coumermycin A1, a JAK2
   activator, was used [[106]26]. As shown in Fig. [107]3L and S13,
   compared to the control group, there was a significant increase in the
   phosphorylation levels of JAK2 and STAT3 in the LPS group, indicating
   activation of the JAK2/STAT3 pathway. However, the phosphorylation
   levels of JAK2 and STAT3, as well as the expression of IL-1β and TNF-α,
   were significantly suppressed after treating LPS-induced cells with
   Ca-CDs. Additionally, when Coumermycin A1 and Ca-CDs were
   simultaneously administered to LPS-induced cells, the phosphorylation
   levels of JAK2 and STAT3 were essentially restored, and the expression
   of IL-1β and TNF-α increased accordingly.

   JAK2, an intracellular protein, comprises seven conserved domains
   without transmembrane regions. The carboxy-terminal domain possesses
   tyrosine kinase activity, which activates downstream signaling
   [[108]27]. As shown previously in Fig. [109]3C, Ca-CDs are
   predominantly distributed in the cytoplasm, providing a precondition
   for the interaction of Ca-CDs with the cytoplasmic domains of JAK2.
   Moreover, amino acid residues surrounding the JAK2 binding site
   typically interact with nitrogen-containing five- or six-membered
   heterocyclic compounds, thereby blocking the pathway [[110]28]. Based
   on previous HRTEM, FTIR, and XPS analyses, which confirmed that Ca-CDs
   possess a graphene-like carbonized core and a rich nitrogen-containing
   heterocyclic structure, it is reasonable to infer that Ca-CDs can
   interact with JAK2 to form a stable complex, thus reducing the
   phosphorylation level of JAK2, inhibiting the JAK2/STAT3 pathway and
   finally exerting anti-inflammatory effects (Fig. [111]3M).

Ca-CDs inhibit the activation of JAK2/STAT3 pathway by binding to JAK2

   Molecular dynamics simulations were used to model the binding process
   between Ca-CDs and JAK2 and to assess the structural stability. Before
   the simulation, an initial model was constructed using the JAK2 protein
   structure from the UniProt database, and a graphene-like fragment was
   attached to the ligand molecule as a simplified model of Ca-CDs
   (Fig. [112]4A). Since the JAK2 protein structure was predicted using
   AlphaFold, a Ramachandran plot was applied to verify the predicted
   torsion angles. As shown in Fig. S14, 98.5% of the residues in the
   protein were located in the allowed regions, demonstrating a
   high-quality model. A 120 ns molecular dynamics simulation was then
   performed. An animated movie of the JAK2/Ca-CDs complex, based on 600
   frames extracted from the trajectory, was provided in Video S4. The
   movie and snapshot images of the simulation trajectory at 0, 60, and
   120 ns showed that the Ca-CDs were stably bound to the JAK2 throughout
   the simulation process (Fig. [113]4A-C).

Fig. 4.

   [114]Fig. 4
   [115]Open in a new tab

   Ca-CDs inhibit the activation of JAK2/STAT3 pathway by binding to JAK2.
   (A-C) Structures of Ca-CDs binding to JAK2, extracted from the
   molecular dynamics simulation trajectory at (A) 0 ns, (B) 60 ns, and
   (C) 120 ns, respectively. JAK2 is depicted as a rainbow-colored cartoon
   and gray-shaded surface. Ca-CDs are shown in ball-and-stick mode, with
   carbon, hydrogen, oxygen, and nitrogen atoms colored gray, white, red,
   and blue, respectively. Enlarged views highlight the binding
   conformations of Ca-CDs within the JAK2 binding pocket at different
   simulation times. (D) RMSD of the Cα atoms of JAK2 in JAK2/Ca-CDs
   complex as a function of time. (E) Numbers of residues in diverse
   secondary structures of JAK2. (F) RMSF values of amino acid residues
   750–1130 in JAK2/Ca-CDs complex and JAK2, respectively. (G) A schematic
   representation of SPR analysis illustrating the interaction between
   Ca-CDs and JAK2. JAK2 protein is covalently immobilized on a CM5 sensor
   chip as the stationary phase, with Ca-CDs introduced as the mobile
   phase. The binding of Ca-CDs to JAK2 causes a shift in the reflected
   angle. (H) The sensorgram of Ca-CDs binding to JAK2-immobilized sensor
   chip. (I) The equilibrium responses versus Ca-CDs concentrations

   A least-squares fit of the entire trajectory was conducted to calculate
   the root mean square deviation (RMSD) of the backbone atoms from their
   initial structure. The RMSD values stabilized at approximately 6 Å
   after 90 ns of simulation, indicating that the simulation reached
   stability beyond this point (Fig. [116]4D). Additionally, the number of
   residues in various secondary structures of the JAK2, including
   α-helix, β-sheets, turns, coil, and bend, remained nearly constant
   throughout the simulation (Fig. [117]4E). Further analysis involved
   measuring the residue-wise root mean square fluctuation (RMSF), which
   reflects deviations in the positions of JAK2 amino acid residues.
   Before evaluating the RMSF of the complex, a 120 ns molecular dynamics
   simulation of JAK2 without Ca-CDs was performed (Fig. S15A-C), with the
   RMSF values of JAK2 serving as a reference for comparison. The RMSD
   results indicated that JAK2 reached a stable conformation after 100 ns
   of simulation (Fig. S15D). Then, to calculate the RMSF, the simulation
   trajectories of both the JAK2/Ca-CDs complex and JAK2 alone were
   obtained from the last stable 20 ns of the molecular dynamics
   simulation. As shown in Fig. [118]4F, the RMSF values near the binding
   pocket of the JAK2/Ca-CDs complex were significantly lower than those
   of JAK2 alone, indicating that these amino acid residues were
   restrained, suggesting several interactions between Ca-CDs and JAK2.

   To further investigate the interaction between Ca-CDs and JAK2, we
   employed surface plasmon resonance (SPR) technology, recognized as the
   gold standard for real-time detection of molecular interactions
   [[119]29]. Purified recombinant JAK2 protein was covalently immobilized
   on a CM5 sensor chip to serve as the stationary phase for
   protein-ligand binding in the SPR analysis. Ca-CDs were introduced over
   the chip as the mobile phase. The binding of Ca-CDs to JAK2 led to
   real-time changes in the refractive index at the sensor surface,
   causing shifts in the resonance angle. By monitoring these changes, we
   obtained kinetic parameters and the equilibrium dissociation constant
   (K[D]) of the interaction (Fig. [120]4G). The results showed that the
   K[D] value between Ca-CDs and JAK2 was approximately 64.2 µM,
   exhibiting rapid association and dissociation kinetics (Fig. [121]4H,
   I). These findings confirm a significant interaction between Ca-CDs and
   JAK2, which contribute to the regulation of inflammatory pathways and
   thereby inhibit the expression of pro-inflammatory cytokines.

Promotion of Ca-CDs on the osteogenic differentiation

   Dysregulation of the JAK2/STAT3 pathway disrupts immune homeostasis,
   leading to the expansion and persistence of inflammation, which
   subsequently contributes to continuous bone destruction. Consequently,
   the ultimate objective of treatment is to reshape the inflammatory
   microenvironment and reconstruct the bone tissue [[122]30]. To this
   end, our Ca-CDs show promise for promoting bone regeneration under
   inflammatory conditions by leveraging the incorporated calcium element
   (Fig. [123]5A). Initially, rat bone marrow stromal cells (rBMSCs) were
   isolated and identified (Fig. S16), followed by biosafety testing. The
   results demonstrated that Ca-CDs exhibited good biocompatibility within
   the concentration range of 0–300 µg/mL (Fig. [124]5B and S17-19) and
   were successfully taken up into the cytoplasm (Fig. [125]5C), providing
   a prerequisite condition for the biological functions of Ca-CDs in
   rBMSCs.

Fig. 5.

   [126]Fig. 5
   [127]Open in a new tab

   Ca-CDs promote the osteogenic differentiation under inflammatory
   conditions. (A) Schematic diagram of Ca-CDs protection for rBMSCs under
   LPS stimulation. (B) CCK-8 assay of rBMSCs, which are treated with
   different concentrations of Ca-CDs (0, 50, 100, 200, 300, and
   400 µg/mL). (C) Cell imaging of Ca-CDs-treated rBMSCs. ALP staining on
   (D) day 7 and (E) day 14. (F) ARS staining on day 21. (G) Validation of
   osteogenic genes expression by RT-qPCR on day 7. (H) Western blot
   analysis of osteogenic-related proteins: RUNX2, SP7, and OCN. (I)
   Corresponding quantitative data from Western blot. (J) IF staining of
   OCN using CLSM. Data are presented as mean ± SD from three independent
   experiments. * indicates P < 0.05, ** indicates P < 0.01, *** indicates
   P < 0.001

   Based on the biosafety results, the dosing concentration range for
   rBMSCs was set to match that of RAW264.7 cells. Initially, the effect
   of Ca-CDs on rBMSCs under normal conditions was examined, confirming
   their ability to promote osteogenic differentiation (Fig. S20-23),
   whereas Fedratinib, a JAK2 inhibitor, did not upregulate osteogenic
   gene expression (Fig. S22). These results indicate that, in contrast to
   a conventional JAK2 inhibitor, Ca-CDs possess additional osteogenic
   activity, highlighting their multifunctionality as a nanoinhibitor.
   Subsequently, LPS was used to stimulate rBMSCs to mimic an inflammatory
   microenvironment, followed by alkaline phosphatase (ALP) and alizarin
   red S (ARS) staining. The staining results showed that under LPS
   stimulation, almost no positive cells were observed, whereas the
   addition of Ca-CDs significantly increased both the number of positive
   cells and the staining intensity in a dose-dependent manner
   (Fig. [128]5D-F). RT-qPCR and Western blot results further demonstrated
   that Ca-CDs promoted the expression of osteogenic markers at both the
   genetic and protein levels (Fig. [129]5G-I). Immunofluorescence (IF)
   staining was further used to assess the expression of osteocalcin
   (OCN), a specific biochemical marker of bone formation. The results
   revealed that the intensity of red fluorescence from OCN in the Ca-CDs
   group was significantly higher than in the LPS group, indicating
   increased expression (Fig. [130]5J). These comprehensive experimental
   findings confirm that Ca-CDs effectively promote osteogenic
   differentiation in rBMSCs under both inflammatory and normal
   conditions.

   To further investigate whether the osteogenic effect of Ca-CDs is
   associated with the incorporated calcium, undoped CDs were synthesized
   from ASA and MET. RT-qPCR and ALP staining results showed that undoped
   CDs exhibited negligible osteogenic activity (Fig. S22 and S24),
   directly confirming that the osteogenic function of Ca-CDs depends on
   calcium doping. To clarify the contribution of calcium in Ca-CDs, the
   osteogenic effect of Ca-CDs were compared with those of calcium
   chloride (CaCl[2]) at an equivalent calcium concentration. The results
   demonstrated that Ca-CDs induced stronger osteogenic activity than
   CaCl[2] (Fig. S22 and S24), suggesting that the osteogenic function of
   Ca-CDs is not solely attributable to calcium release. The difference
   between Ca-CDs and CaCl[2] is most likely due to the nanoscale size of
   Ca-CDs, which facilitates enhanced cellular uptake and consequently
   improves Ca^2+ bioavailability. To verify this hypothesis, Rhod-2, a
   calcium-sensitive fluorescent probe, was used to measure intracellular
   calcium levels in rBMSCs (Fig. S25). The results showed that Ca-CDs
   significantly increased intracellular Ca^2+ levels compared with
   CaCl[2], thereby enhancing osteogenic effects. These findings confirm
   that the osteogenic effect of Ca-CDs is achieved through both calcium
   doping and enhanced cellular calcium uptake.

Construction of Ca-CDs-based self-triggered hydrogel delivery system

   It is well known that reshaping the inflammatory microenvironment and
   reconstructing bone tissue is a slow and complex process requiring
   long-term regulation [[131]31]. However, maintaining prolonged drug
   residence at the target site remains challenging during treatment,
   limiting sustained pharmacological effects. Hydrogel-based delivery
   systems offer a promising solution for addressing these challenges in
   vivo [[132]32–[133]34]. Notably, the inherent calcium in the Ca-CDs can
   serve as crosslinking junctions by coordinating with carboxyl group,
   thereby triggering the formation of a Ca-CDs@ALG hydrogel
   (Fig. [134]6A). As demonstrated, the blue fluorescence of the formed
   hydrogel confirms the uniform distribution of Ca-CDs within the matrix
   (Fig. S26). The Ca-CDs@ALG hydrogel exhibits both injectable and
   moldable properties (Video S5). As shown in Fig. [135]6B, C and S27,
   the hydrogel can be injected to form the letters “JLU” and molded into
   a cylindrical shape. Furthermore, rheological testing was conducted to
   assess the mechanical properties of the hydrogel. As the angular
   frequency increased, the storage modulus (G’) remained consistently
   higher than the loss modulus (G’’) (Fig. [136]6D), indicating that the
   hydrogel can withstand rapid deformation and maintain a stable network
   structure, even under high oscillatory strain (Fig. [137]6E).

Fig. 6.

   [138]Fig. 6
   [139]Open in a new tab

   Construction of Ca-CDs@ALG hydrogel and its therapeutic effect on
   periodontitis. (A) Schematic illustration of the construction of
   Ca-CDs@ALG hydrogel. Photographs demonstrating the (B) injectability
   and (C) plasticity of Ca-CDs@ALG hydrogel taken under sunlight and UV
   light, respectively. Variations of storage and loss modulus (G’ and
   G’’, respectively) versus (D) angular frequency and (E) oscillation
   strain. (F) Images of Ca-CDs@ALG microsphere. Inset: enlarged view of
   the microspheres with a scale bar of 5 μm. (G) Cumulative release of
   Ca-CDs from Ca-CDs@ALG microsphere at different time points. (H)
   Degradation curves of Ca-CDs@ALG microsphere over time. (I)
   Experimental design for treating rat ligature-induced periodontitis.
   (J) Representative micro-CT three-dimensional reconstructed images
   (upper panels) and scanning buccopalatal images (lower panels) along
   the longitudinal direction of the maxilla. (K) Quantitative analysis of
   BV/TV, Tb.N, Tb.Th, and Tb.Sp, determined by micro-CT images. Data are
   presented as mean ± SD from three independent experiments. * indicates
   P < 0.05, ** indicates P < 0.01, *** indicates P < 0.001. (L) H&E
   staining showing inflammatory cells infiltration under 4X objective
   lens (left panels) and a locally enlarged image (right panels). Arrows
   indicate inflammatory cells. (M) IF staining of IL-1β under 10X
   objective lens. The white line indicates the boundary of the second
   molar root. (N) Goldner staining showing alveolar bone height and
   morphology, depicted by gray lines, under 4X (left panels) and 10X
   (right panels) objective lens, respectively

   Nevertheless, inflammatory and immune diseases often result in
   irregularly shaped defects, which require a delivery system with
   improved flow properties and small particle sizes. To address this
   need, a Ca-CDs-based alginate hydrogel microsphere (Ca-CDs@ALG
   microsphere) system was developed. CLSM images revealed that the
   microspheres were spherical with an average diameter of approximately
   5.7 μm (Fig. [140]6F and S28). The microsphere system facilitated the
   continuous release of Ca-CDs, with a cumulative release of 75.1% by day
   7 (Fig. [141]6G). Additionally, the microspheres degraded by
   approximately 73.5% after 7 days (Fig. [142]6H). These findings suggest
   that Ca-CDs can initiate self-crosslinking with alginate to form an
   injectable, moldable hydrogel for sustained drug release, enhancing the
   long-term biological effects of Ca-CDs at inflammatory sites and
   demonstrating significant potential for in vivo applications.

Therapeutic efficacy of Ca-CDs and Ca-CDs-based hydrogel in treating
periodontitis

   To comprehensively verify the anti-inflammatory and osteogenic effects
   of Ca-CDs, and to assess the sustained release properties and long-term
   biological effects of the Ca-CDs@ALG delivery system, it is essential
   to select a disease model that closely aligns with these research
   objectives. Periodontitis, one of the most prevalent chronic
   inflammatory diseases worldwide, affects the periodontium and is
   characterized by extensive inflammatory cell infiltration and irregular
   destruction of the alveolar bone [[143]35, [144]36]. Effective
   treatment of periodontitis requires drugs that can be precisely
   injected and fully penetrate the irregular defect areas of periodontal
   pockets, while also withstanding the flushing effect of saliva to
   ensure prolonged action and enhanced efficacy. Therefore, a rat
   periodontitis model was established to fulfill these experimental
   requirements (Fig. [145]6I and S29). Micro-computed tomography
   (micro-CT) scans revealed significant alveolar bone height reduction in
   the periodontitis group, with vertical bone resorption reaching up to
   one-third of the root, along with horizontal bone loss, confirming the
   successful establishment of the periodontitis model. In contrast, both
   the Ca-CDs and Ca-CDs@ALG microsphere groups showed minimal bone
   resorption, comparable to the control group (Fig. [146]6J). To quantify
   the extent of bone loss, the distance from the cemento-enamel junction
   (CEJ) to the alveolar bone crest (ABC) was measured. As shown in Fig.
   S30, the CEJ-ABC distance in the periodontitis group was
   0.77 ± 0.08 mm, while it was significantly reduced by 46.8% in the
   Ca-CDs group (0.41 ± 0.02 mm) and by 68.9% in the Ca-CDs@ALG
   microsphere group (0.24 ± 0.03 mm). Further analysis of alveolar bone
   volume and mineralization levels demonstrated significant increases in
   bone volume fraction (BV/TV), trabecular number (Tb.N), and trabecular
   thickness (Tb.Th), along with a notable reduction in trabecular
   separation (Tb.Sp) in both the treatment groups compared to the
   periodontitis group (Fig. [147]6K). These findings indicate that
   treatment with Ca-CDs and Ca-CDs@ALG microsphere can effectively
   mitigate alveolar bone resorption in inflammatory conditions, with the
   Ca-CDs@ALG microsphere formulation showing superior therapeutic effects
   due to its sustained release properties.

   Histological staining of maxillary specimens was conducted to
   thoroughly assess the periodontal tissues. Hematoxylin and eosin (H&E)
   staining in the periodontitis group revealed characteristic
   pathological changes, including the extension of the junctional
   epithelium toward the root, detachment of the epithelial attachment
   from the root surface, and dense infiltration of inflammatory cells
   (Fig. [148]6L). In contrast, both the Ca-CDs and Ca-CDs@ALG microsphere
   groups exhibited significantly reduced inflammatory cell infiltration
   around the epithelium, indicating a reversal of the pathological
   changes associated with periodontitis. IF staining further demonstrated
   that treatment markedly reduced IL-1β expression, particularly in the
   Ca-CDs@ALG microsphere group, which significantly alleviated local
   inflammation (Fig. [149]6M and S31). Goldner staining showed
   substantial bone crest resorption close to the root apex and
   discontinuous trabeculae in the periodontitis group (Fig. [150]6N). In
   comparison, both the Ca-CDs and Ca-CDs@ALG microsphere groups displayed
   minimal bone resorption, continuous bone morphology, and significantly
   upregulated OCN expression, as confirmed by IF staining (Fig. S32).
   Overall, our therapeutic strategy, which inhibits inflammation and
   promotes tissue regeneration, presents great potential for the
   treatment of chronic inflammatory diseases. Additionally, the H&E
   staining of vital organs in both the Ca-CDs and Ca-CDs@ALG microsphere
   groups showed no significant differences compared to the control group
   (Fig. S33), confirming the safety of Ca-CDs and their microsphere
   formulations for in vivo applications.

Discussion

   The JAK2/STAT3 pathway is recognized as the central communication hub
   in cytokine signaling, playing a critical role in maintaining immune
   balance and tissue homeostasis, and has thus emerged as an essential
   target for drug development [[151]37–[152]40]. Recent clinical trials
   have shown that targeting JAK2 and preventing its phosphorylation can
   effectively suppress abnormal immune and inflammatory responses
   mediated by cytokines, providing solid evidence for the use of JAK2
   inhibitors in treating inflammatory and autoimmune diseases
   [[153]41–[154]43]. Most of the reported research on JAK2 inhibitors has
   focused on the design and synthesis of small molecules with specific
   structural characteristics, such as NH, O or CH[2] spacers,
   heterocyclic moiety, and amide bonds [[155]44–[156]46]. However, these
   inhibitors usually present limitations, including poor water
   solubility, low therapeutic concentrations at target sites, and
   inadequate management of ongoing inflammation-induced bone loss. In
   response, this study integrates nanomaterial properties with inhibitory
   functions [[157]47–[158]49], developing a multifunctional JAK2
   nanoinhibitor, Ca-CDs, which holds promise as an effective therapeutic
   strategy.

   Pharmacophore models and fragment-based drug design can serve as
   preliminary screening tools to effectively identify key functional
   groups closely associated with target binding, thereby theoretically
   excluding inactive structures. This approach not only reduces the
   workload and cost of experimental validation but also holds significant
   potential for the design and development of targeted therapeutics.
   Based on the binding site of JAK2 protein and the structures of
   existing JAK2 inhibitors, we generated a target-specific training
   dataset, thereby constructing pharmacophore models and conducting
   fragment-based drug design. Computational biology confirmed that
   nitrogen and oxygen-containing aromatic rings facilitate stable binding
   to JAK2. By leveraging the unique characteristics of CDs, such as their
   ability to incorporate various functional groups [[159]50–[160]53], and
   by selecting precursor materials with relevant chemical structures
   [[161]54–[162]56], we successfully designed and synthesized Ca-CDs, a
   multifunctional JAK2 nanoinhibitor featuring a highly conjugated sp^2
   domain and heteroatom-containing rings [[163]57]. A series of
   characterization methods, including RNA sequencing, molecular dynamics
   simulations, and SPR, confirmed that Ca-CDs can stably bind to JAK2,
   inhibiting JAK2/STAT3 pathway, interrupting cytokine-mediated positive
   feedback loops, and effectively attenuating excessive inflammatory
   responses [[164]58]. These findings demonstrate that Ca-CDs possess
   potent anti-inflammatory properties similar to those of traditional
   JAK2 inhibitors, but with enhanced biocompatibility and water
   solubility due to their unique carbon-based nanomaterials
   [[165]59–[166]62], suggesting that Ca-CDs have the potential to address
   the challenges faced by conventional JAK2 inhibitors, offering a new
   approach to managing JAK2-related diseases.

   A further advantage of Ca-CDs over existing JAK2 inhibitors is their
   multifunctionality. Traditional JAK2 inhibitors primarily target the
   JAK2/STAT3 pathway to suppress inflammation but often fail to address
   the ongoing inflammation-induced bone destruction. In contrast, Ca-CDs
   demonstrate the capacity to promote bone regeneration under
   inflammatory conditions through two mechanisms. First, Ca-CDs inhibit
   inflammatory signaling, mitigating bone loss associated with
   inflammation. Second, the inherent calcium in Ca-CDs plays a crucial
   role in the bone matrix, facilitating effective bone regeneration
   [[167]63]. Furthermore, this calcium can serve as crosslinking
   junctions, triggering the formation of Ca-CDs@ALG hydrogel. Although
   alginate hydrogels generally exhibit relatively low mechanical
   strength, their lower mechanical rigidity actually facilitates complete
   filling of complex anatomical structure of periodontal pockets,
   ensuring more uniform drug distribution and localized therapeutic
   effects. This multifunctionality distinguishes Ca-CDs from existing
   inhibitors, presenting them as a viable option for inflammatory and
   immune-related diseases by offering a more comprehensive therapeutic
   approach. By demonstrating the efficacy of Ca-CDs in reducing
   inflammation and promoting bone regeneration in a periodontitis model,
   this study highlights their in vivo potential. Future research should
   focus on optimizing the design and synthesis of Ca-CDs to improve
   binding affinity and multifunctionality [[168]64]. Additionally,
   exploring the potential of Ca-CDs for treating other JAK2-related
   inflammatory diseases, such as rheumatoid arthritis and inflammatory
   bowel disease, could expand their therapeutic applications.

Conclusion

   In summary, through the construction of pharmacophore models and
   fragment-based drug design, we successfully designed and synthesized
   Ca-CDs as a multifunctional nanoinhibitor targeting JAK2. Strong
   characterization evidence confirms that Ca-CDs can directly bind to the
   JAK2, effectively modulating the JAK2/STAT3 pathway and reducing the
   secretion of pro-inflammatory cytokines. Importantly, the inherent
   calcium within the Ca-CDs promotes osteogenic differentiation under
   inflammatory conditions, demonstrating dual anti-inflammatory and
   osteogenic properties. Additionally, our Ca-CDs serve as crosslinking
   junctions that trigger the formation of a Ca-CDs-based alginate
   hydrogel, facilitating long-term retention at local lesion sites and
   sustained drug release. In vivo studies using a periodontitis model
   demonstrated that the Ca-CDs and Ca-CDs@ALG delivery system effectively
   reduced inflammatory cell infiltration and facilitated the repair of
   inflammation-associated bone defects. Overall, the computer-aided
   design of Ca-CDs as a JAK2 nanoinhibitor, along with the
   self-triggering hydrogel delivery system, holds significant potential
   for treating periodontitis and other inflammatory diseases associated
   with the JAK2/STAT3 pathway, offering a promising therapeutic strategy
   for clinical application.

Materials and methods

Materials

   Acetylsalicylic acid (99%), metformin (97%) and calcium chloride
   (99.9%) were purchased from Aladdin (Shanghai, China). Dialysis bag
   (MWCO = 100–500 Da) was obtained from Spectrum Laboratories (Rancho
   Dominguez, CA, USA). Dulbecco’s modified Eagle medium (DMEM) with high
   glucose, alpha minimum essential medium (α-MEM) and
   penicillin-streptomycin were purchased from Hyclone (Logan, UT, USA).
   Fetal bovine serum (FBS) was obtained from Clark Bioscience (VA, USA).
   Cell counting kit-8 (CCK-8) was purchased from Invigentech (Irvine, CA,
   USA). Cell cycle kit, Hifair^® III 1 st Strand cDNA Synthesis SuperMix
   for qPCR, Brefeldin A, and Rhod-2 AM were purchased from Yeasen
   Biotechnology (Shanghai, China), respectively. Annexin V-FITC apoptosis
   analysis kit obtained from Simu Biotech (Tianjin, China).
   Lipopolysaccharide (LPS), ascorbic acid, and β-glycerophosphate
   disodium salt hydrate were purchased from Sigma Aldrich (St Louis, MO,
   USA). Calcein/PI cell viability/cytotoxicity assay kit, alkaline
   phosphatase (ALP) assay kit, alizarin red S (ARS) staining solution,
   modified oil red O staining kit, and enhanced bicinchoninic acid (BCA)
   protein assay kit were obtained from Beyotime (Shanghai, China). RNAiso
   Plus reagent was obtained from Takara Bio (Otsu, Japan). Fedratinib,
   Colivelin and Coumermycin A1 were obtained from MedChenExpress
   (Monmouth Junction, NJ, USA).

Generation of ligand-based common feature pharmacophore models

   The two-dimensional (2D) structures of 53 known JAK2 inhibitors were
   obtained from the MedChenExpress database (Fig. [169]S1). Their
   corresponding three-dimensional (3D) structures were constructed using
   Avogadro software, followed by optimization through energy minimization
   within the CHARMM force field. The maximum number of conformations for
   each molecule was set to 255, and conformations were generated under an
   energy threshold of 20 kcal/mol. For a molecule with stereo isomers,
   this threshold applies to conformations for each individual isomer.
   Subsequently, pharmacophore models were generated using the HipHop
   algorithm [[170]65]. Ten pharmacophore models were created, and the top
   two were selected based on their rank values.

Fragment-based drug design

   First, a fragment library was generated from the 53 known JAK2
   inhibitors obtained from the MedChenExpress database using the
   Retrosynthetic Combinatorial Analysis Procedure (RECAP) algorithm
   [[171]66]. During the library construction, bond-type rules, including
   amide, ester, amine, urea, ether, olefin, quaternary nitrogen, aromatic
   nitrogen bonded to aliphatic carbon, lactam nitrogen bonded to
   aliphatic carbon, aromatic carbon bonded to aromatic carbon, and
   sulfonamide, were employed to cleave the molecules. Then, the structure
   of the JAK2 protein, predicted by AlphaFold, was downloaded from the
   UniProt database (ID: [172]Q62120). After defining the protein’s
   binding site, the Ludi algorithm was utilized to search the generated
   fragment library for candidates that bind to the active site.
   Consequently, suitable fragments capable of interacting with the JAK2
   protein were screened, and their positions within the active site of
   the JAK2 protein structures were visualized using ChimeraX.

Synthesis of Ca-CDs

   A mixture of metformin (1 g) and CaCl[2] (2 g) was dissolved in
   deionized water (3 mL). This solution was then transferred to a
   round-bottom flask containing acetylsalicylic acid (0.5 g) and heated
   in an oven at 120 °C for 110 min. Subsequently, the mixture was reacted
   at 250 °C under vacuum for 120 min, yielding a yellow, fluffy, porous
   Ca-CDs sphere. This sphere was dissolved in deionized water (100 mL),
   sonicated for 5 min to accelerate dissolution, and then centrifuged at
   10,000 rpm at room temperature for 10 min to remove large particulate
   impurities. The supernatant was rotary evaporated and then transferred
   to a dialysis bag (100–500 Da) for 12 h of dialysis. The purified light
   yellow Ca-CDs solution was concentrated and freeze-dried to obtain a
   Ca-CDs solid powder. Undoped CDs were prepared using a similar
   procedure, except without the addition of CaCl[2]. Following the same
   approach, CDs synthesized using other organic small molecules as
   precursors were also prepared.

Characterization of Ca-CDs

   Ultraviolet-visible (UV-vis) absorption spectra were obtained using a
   Lambda 800 spectrophotometer. Fluorescence and photoluminescence
   spectra were analyzed with a Shimadzu RF6000 spectrophotometer.
   High-resolution transmission electron microscopy (HRTEM) image was
   captured using a JEM-2100 F microscope at an acceleration voltage of
   200 kV. Fourier transform infrared (FTIR) spectra were recorded using a
   Bruker VERTEX 80 V spectrometer. X-ray photoelectron spectroscopy (XPS)
   was conducted using the NEXSA system from Thermo Scientific.
   Energy-dispersive spectroscopy (EDS) analysis was conducted using an
   Oxford Inca X-Max spectrometer.

Preparation of Ca-CDs-based alginate hydrogels and microspheres

   A Ca-CDs solution was added dropwise to sodium alginate (ALG), followed
   by continuous stirring at 1500 rpm to produce injectable hydrogels. The
   rheological properties of the hydrogels were analyzed using a DHR-2
   rheometer. The variation in storage modulus (G’) and loss modulus (G’’)
   was monitored across an angular frequency range from 0.1 to 100 rad/s
   in oscillation mode. Using a similar method, Ca-CDs solution (1 mL) was
   added dropwise to 20 mL of an aqueous ALG solution at a flow rate of
   0.1 mL/min with a syringe pump, followed by magnetic stirring at
   400 rpm for 30 min to prepare hydrogel microspheres. The microspheres
   were alternately washed with deionized water and anhydrous ethanol
   twice. The size and morphology of the microspheres were characterized
   using confocal laser scanning microscopy (CLSM).

Degradation and release performance of Ca-CDs-based alginate hydrogel
microspheres

   For degradation studies, microspheres were immersed in deionized water
   at 37 °C. The microspheres were centrifuged and washed at predetermined
   intervals, followed by freeze-drying and weighing. The degradation rate
   (D%) of the microspheres was calculated using the formula: D% = (N0 -
   Nt)/N0 × 100%, where N0 (mg) is the initial weight of the microspheres,
   and Nt (mg) is their weight after soaking for t days. For release
   studies, the microspheres were incubated at 37 °C, and 3 mL of
   supernatant was collected at various time points, while maintaining the
   total volume at 10 mL by replenishing with deionized water. Given that
   the concentration of Ca-CDs correlates with their UV-vis absorbance
   values, the concentration of Ca-CDs released at various time points was
   determined by measuring the absorbance of the supernatant. The
   cumulative release rate (%) of Ca-CDs at different time points was
   calculated as: (Accumulated release of Ca-CDs at the respective time
   point/Total amount of Ca-CDs) × 100%.

Culturing and identification of RAW264.7 cells and rat bone marrow stromal
cells (rBMSCs)

   RAW264.7 cells were resuspended in DMEM high glucose medium
   supplemented with 10% FBS and cultured in an incubator at 37 °C with 5%
   CO[2]. Cell density was monitored, and upon reaching 70%−80%
   confluency, cells were subcultured for further experiments. Male Wistar
   rats (age 3–4 weeks) were used for isolating rBMSCs employing the whole
   bone marrow adherence method. The cell suspension was cultured in a 5%
   CO[2] incubator at 37℃. After 4 days, the medium was replaced with
   fresh α-MEM complete medium containing 10% FBS, penicillin (100 U/mL),
   and streptomycin (100 µg/mL). At the third passage, adipogenic and
   osteogenic differentiation capabilities were assessed using oil red O
   and ARS staining, respectively. Flow cytometry was employed for
   phenotypic identification of rBMSCs using CD29 and CD90 as positive
   surface markers and CD45 and CD34 as negative surface markers.

CCK-8 assay for cell toxicity

   RAW264.7 cells were seeded in a 96-well plate at a density of 5 × 10^3
   cells/well. A control group and five experimental groups were
   established, treating cells with 100 µL of complete medium containing
   varying concentrations of Ca-CDs (0, 50, 100, 200, 300, and 400 µg/mL).
   After 24 h, 100 µL of medium containing 10 µL of CCK-8 solution was
   added to each well. The absorbance at 450 nm was measured using a
   microplate reader 30 min later, and cell viability was calculated
   relative to the control group. The same experimental setup and groups
   were used to assess the viability of rBMSCs seeded at densities of
   5 × 10^3 and 2 × 10^3 cells/well, after incubation periods of 24 h and
   7 days, respectively.

Detection of cell apoptosis and cell cycle by flow cytometry

   RAW264.7 cells and rBMSCs were seeded in 6-well plates at a density of
   2 × 10^5 cells/well and cultured overnight for adherence. The medium
   was then replaced with complete medium containing different final
   concentrations of Ca-CDs (0, 100, 200, and 300 µg/mL). After 24 h,
   cells were collected, washed twice with phosphate buffer saline (PBS),
   and resuspended in 100 µL of binding buffer. Each sample was stained
   with 5 µL of Annexin V-FITC and 5 µL of 7-AAD. The samples were
   incubated for 15 min in the dark, followed by flow cytometry analysis
   to detect fluorescence signals. rBMSCs were cultured and collected as
   described above. Cells were stained with PI/RNase staining solution for
   cell cycle analysis.

Live/Dead cell staining

   RAW264.7 cells and rBMSCs were seeded in confocal dishes at a density
   of 10^5 cells/well and cultured overnight for adherence.
   Post-adherence, the medium was replaced with complete medium containing
   different concentrations of Ca-CDs (0, 100, 200, and 300 µg/mL) and
   incubated for 24 h. The medium was then discarded, and the cells were
   stained with 500 µL/well of Calcein/PI staining solution. The cells
   were incubated at 37 °C in the dark for 30 min before capturing images
   using CLSM.

Cellular uptake of Ca-CDs

   RAW264.7 cells and rBMSCs were seeded in confocal dishes at a density
   of 10^5 cells/well and cultured overnight for adherence. After
   adherence, cells were incubated for 12 h in complete medium containing
   Ca-CDs at a final concentration of 300 µg/mL. Following incubation,
   cells were washed three times with PBS to remove any uninternalized
   material. The internalization of Ca-CDs within the cells was observed
   under CLSM by examining the fluorescence of the Ca-CDs.

Establishment of RAW264.7 cell inflammation model

   Cells were seeded in culture dishes and divided into seven groups, each
   with three replicates: (1) control; (2) LPS group; (3) LPS + 100 µg/mL
   Ca-CDs; (4) LPS + 200 µg/mL Ca-CDs; (5) LPS + 300 µg/mL Ca-CDs; (6)
   LPS + 300 µg/mL acetylsalicylic acid and metformin (300 µg/mL
   ASA + MET); (7) LPS + Fedratinib. After cell adherence, the control
   group’s medium was replaced with complete medium, while the LPS model
   and treatment groups received complete medium containing LPS (100
   ng/mL) to establish the inflammation cell model.

Osteogenic induction and model establishment in rBMSCs

   rBMSCs were seeded and cultured until 70–80% confluency. The medium was
   then switched to osteogenic induction medium (containing 10 mM
   β-glycerophosphate, 50 µg/mL ascorbic acid, in α-MEM complete medium)
   for osteogenic differentiation induction. Normal state rBMSCs grouping:
   (1) control; (2) 100 µg/mL Ca-CDs; (3) 200 µg/mL Ca-CDs; (4) 300 µg/mL
   Ca-CDs; (5) 300 µg/mL ASA + MET; (6) Fedratinib; (7) undoped CDs; (8)
   CaCl[2]. The osteogenic induction medium was replaced every three days.
   Inflamed state rBMSCs grouping: (1) LPS group; (2) LPS + 100 µg/mL
   Ca-CDs; (3) LPS + 200 µg/mL Ca-CDs; (4) LPS + 300 µg/mL Ca-CDs; (5)
   LPS + 300 µg/mL ASA + MET, with five groups in total, each having three
   replicates. After cell adherence, the control group was switched to
   osteogenic induction medium, while the LPS model and treatment groups
   received complete medium containing LPS (10 µg/mL) to establish the
   inflammation model. Osteogenic induction medium was replaced every
   three days.

Reverse transcriptase quantitative polymerase chain reaction (RT-qPCR)

   RAW264.7 cells and rBMSCs were seeded in 6-well plates at a density of
   2 × 10^5 cells/well and cultured as per the above methods for 1 day and
   7 days, respectively. RNA was isolated using 1 mL RNAiso Plus reagent
   per well. RNA concentration was determined using the Thermo Scientific
   NanoDrop spectrophotometer. RNA was reverse transcribed into cDNA using
   the Hifair^® III 1 st Strand cDNA Synthesis SuperMix for qPCR. RT-qPCR
   was conducted on a Roche LC96 fluorescent quantitative PCR instrument,
   analyzing gene expression levels with Actb as the reference gene. The
   primers are listed in Tables S1 and S2.

Western blot

   RAW264.7 cells were seeded in 6 cm diameter culture dishes at a density
   of 10^6 cells/dish and grouped according to the inflammation cell
   model. After 20 h of culture with appropriate treatments, Brefeldin A
   was added to a final concentration of 1 µg/mL, followed by a further
   4-hour incubation. Cells were lysed using RIPA lysis buffer containing
   protease and phosphatase inhibitors. Protein concentration was
   determined using the enhanced BCA protein assay kit. Proteins (30 µg)
   were separated using 10% SDS-PAGE and transferred to polyvinylidene
   fluoride (PVDF) membranes. Membranes were blocked with 5% BSA for 1 h
   to prevent nonspecific binding, followed by overnight incubation at
   4 °C with primary antibodies: anti-GAPDH (10494-1-AP, 1:5000,
   Proteintech), anti-IL-1β (ab254360, 1:1000, Abcam), anti-iNOS
   (ab178945, 1:1000, Abcam), anti-TNF-α (17590-1-AP, 1:1000,
   Proteintech), anti-β-actin (66009-1-Ig, 1:5000, Proteintech), anti-JAK2
   (3230, 1:1000, CST), anti-p-JAK2 (4406, 1:1000, CST), anti-STAT3 (4904,
   1:2000, CST), and anti-p-STAT3 (9145, 1:2000, CST). Membranes were
   washed and incubated with secondary antibodies for 2 h, followed by ECL
   detection and quantification using ImageJ.

   rBMSCs were seeded in 9 cm diameter culture dishes at a density of
   2 × 10^6 cells/dish and subjected to the same treatment protocol as
   defined in the established cell model for a duration of 7 days. Total
   proteins were extracted and incubated with primary antibodies:
   anti-GAPDH, anti-RUNX2 (ab236639, 1:1000, Abcam), anti-SP7 (ab209484,
   1:1000, Abcam), and anti-OCN (23418-1-AP, 1:1000, Proteintech),
   following the same procedure as above, with quantification using
   ImageJ.

ALP and ARS staining

   rBMSCs were seeded in 12-well plates at a density of 10^5 cells/well
   and cultured until 70–80% confluency. Cell models under normal and
   inflammatory conditions were established as previously described, and
   cultured for 7, 14, and 21 days, respectively. The cells were then
   stained using a kit according to the manufacturer’s instructions. The
   stained cells were observed and imaged using an inverted microscope.

Immunofluorescence (IF) staining

   rBMSCs were seeded at a density of 10^5 cells/well in confocal dishes.
   The cells were then cultured for 14 days in osteogenic induction medium
   containing LPS either 0–300 µg/mL Ca-CDs. Following this period, the
   cells were fixed with 4% paraformaldehyde for 10 min and permeabilized
   with 0.1% Triton X-100 for 10 min. Nonspecific binding sites were
   blocked with 5% goat serum for 1 h. Subsequently, cells were incubated
   overnight at 4 °C with anti-OCN (23418-1-AP, 1:400, Proteintech). The
   next day, cells were incubated with CoraLite594-conjugated secondary
   antibody (SA00013-4, 1:200, Proteintech) and Phalloidin-FITC (CA1620,
   1:200, Solarbio) in the dark for 1 h. Images were captured using CLSM.

Measurement of intracellular Ca^2+

   rBMSCs were plated in confocal dishes at a density of 10^5 cells/well.
   The cells were treated with Ca-CDs or CaCl[2] for 12 h, followed
   byincubation with Rhod-2 AM (2 µM) for 30 min, and subsequently imaged
   using CLSM.

Transcriptome analysis by RNA sequencing

   The RAW264.7 inflammation cell model was established with two groups:
   LPS and LPS + 300 µg/mL Ca-CDs. RNA was extracted after 24 h of culture
   and subjected to transcriptome sequencing performed by Shanghai OE
   Biotech Co., Ltd. using the Illumina NovaSeq 6000 platform, generating
   150 bp paired-end reads. Approximately 50 M raw reads were obtained for
   each sample. The raw reads in FASTQ format were processed to remove
   low-quality reads, obtaining clean reads for subsequent data analysis.
   The reference genome alignment was performed using HISAT2 software, and
   gene expression levels were calculated. Reads count for each gene was
   obtained using HTSeq-count and normalized across samples using DESeq2
   software. Differential expression was determined with a threshold of
   q-value < 0.05 and fold change > 2. Gene Ontology (GO) enrichment and
   Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment
   analysis of differentially expressed genes (DEGs) were conducted using
   R, based on the hypergeometric distribution.

Molecular dynamics simulations

   The structure of the JAK2 protein, predicted by AlphaFold, was
   downloaded from the UniProt protein database (ID: [173]Q62120) and
   subsequently assessed for quality using Ramachandran plots. Based on
   analyses from HRTEM and XPS, a simplified model of Ca-CDs was
   constructed and optimized using density functional theory methods.
   Molecular dynamics simulations were performed using the GROMACS
   software package version 2018.8 under the AMBER99SB-ILDNP force field.
   The system was solvated in a cubic box with transferable intermolecular
   interaction potential 3 points (TIP3P) water molecules, ensuring a
   minimum distance of 10 Å between the protein and the box edges. Extra
   system charges were neutralized with Cl- ions. Periodic boundary
   conditions were applied in all directions to simulate an infinite
   system. Steepest descent and conjugate gradient minimization were
   conducted (tolerance up to 1000 kJ/mol/nm) to eliminate any unfavorable
   interactions. Prior to production simulation, the system underwent two
   equilibration steps where all atoms, except for water molecules and Cl-
   ions, were restrained to prevent conformational changes. Initially, the
   system was simulated under constant volume (NVT ensemble) conditions,
   reaching 300 K within 1 ns using the V-rescale method. Subsequently,
   the equilibrated structure of the NVT ensemble was balanced under
   constant pressure (NPT ensemble) at isotropic pressure of 1.0 bar using
   the Parrinello-Rahman barostat for 1 ns. The production Molecular
   dynamics simulation was carried out for 120 ns without any restraints.
   Simulation trajectories were saved every 10 ps for analysis of root
   mean square deviation (RMSD) and root mean square fluctuation (RMSF).

Surface plasmon resonance (SPR)

   SPR assays were conducted to evaluate the interaction between purified
   recombinant JAK2 proteins and Ca-CDs using a Biacore 8 K instrument
   (Cytiva). The His-tagged human JAK2 protein was expressed in E. coli
   (HY-P7989, MedChenExpress) and exhibited 96.3% homology with the mouse
   JAK2 protein. This recombinant His-JAK2 was purified and immobilized on
   a CM5 chip under optimized conditions. The running buffer was prepared
   with the following composition: 50 mM Tris-HCl, 150 mM NaCl, 5 mM
   MgCl[2], and 0.05% Tween-20. The Ca-CDs were serially diluted in
   running buffer, covering a range of concentrations, and injected into
   the sensor chip. Each binding cycle consisted of a 60-second binding
   phase followed by a 60-second dissociation phase, with a flow rate of
   30 µL/min. The equilibrium dissociation constant (K[D]) was calculated
   using the Biacore 8 K Evaluation Software, employing appropriate
   fitting models to ensure accuracy.

Establishment of rat periodontitis model

   Twenty-four male Wistar rats, weighing 150–200 g, were randomly divided
   into four groups: control, periodontitis, Ca-CDs, and Ca-CDs@ALG
   microsphere. Periodontitis was induced in the second upper molar using
   the ligature method. Both the control and periodontitis groups received
   20 µL of PBS injected into the gingival sulcus, while the Ca-CDs group
   received 20 µL of Ca-CDs solution (300 µg/mL), and the Ca-CDs@ALG
   microsphere group received 20 µL of microsphere (300 µg/mL). Treatments
   were administered every three days. After 28 days, the rats were
   anesthetized, perfused with saline and 4% paraformaldehyde, and samples
   of the maxilla, heart, liver, spleen, and kidneys were collected. All
   animal experiments were conducted following Jilin University’s
   committee guidelines for animal experiments (approval number:
   20240290).

Micro-computed tomography (Micro-CT) bone tissue analysis

   The collected maxillary specimens were scanned using µCT-50 (Scanco
   Medical AG, Bassersdorf, Switzerland) with scanning parameters of 70
   kVp and 200 µA. ImageJ software was used to measure the distance
   between the cemento-enamel junction (CEJ) and the alveolar bone crest
   (ABC) at four points on the buccal and palatal sides of the second
   upper molar. Bone tissue parameters were also obtained for each group.

Histological staining analysis

   Maxillary and visceral specimens, fixed and decalcified in 4%
   paraformaldehyde, were dehydrated in an ethanol gradient, embedded in
   paraffin, and sectioned. Sections were stained according to the kit
   instructions for Hematoxylin and eosin (H&E), Goldner’s trichrome, and
   IF antibodies included anti-OCN (23418-1-AP, 1:400, Proteintech) and
   anti-IL-1β (ab254360, 1:50, Abcam).

Statistical analysis

   Each experiment was conducted at least three times. Data are presented
   as mean ± standard deviation. All statistical analyses were performed
   using Origin software. For comparisons involving multiple groups,
   statistical analysis was performed using one-way ANOVA, with a P-value
   of < 0.05 considered to indicate statistical significance.

Supplementary Information

   [174]Supplementary Material 1^ (28.2MB, docx)
   [175]Supplementary Material 2^ (8.7MB, mp4)
   [176]Supplementary Material 3^ (8.3MB, mp4)
   [177]Supplementary Material 4^ (12MB, mp4)
   [178]Supplementary Material 5^ (12.5MB, mp4)
   [179]Supplementary Material 6^ (9.5MB, mp4)

Author contributions

   Yang Liu: Writing - review & editing, Writing - original draft,
   Validation, Supervision, Software, Methodology, Investigation, Formal
   analysis, Data curation, Conceptualization. Duan Wang: Visualization,
   Supervision, Methodology, Investigation, Data curation. Wenqian Zheng:
   Visualization, Methodology, Formal analysis. Yuxuan Wang: Software,
   Validation, Investigation. Peiyu Wang: Visualization, Methodology,
   Investigation. Yuping Zhao: Visualization, Methodology. Xiaofan Liu:
   Visualization, Methodology. Yaru Shi: Investigation, Formal analysis.
   Yi Wang: Writing - review & editing, Supervision. Na Zhou: Writing -
   review & editing, Investigation. Fermín E. González: Writing - review &
   editing, Supervision. Hongchen Sun: Writing - review & editing,
   Resources. Ding Zhou: Writing - review & editing, Writing - original
   draft, Supervision, Software, Resources, Project administration,
   Funding acquisition, Conceptualization. Xiaowei Xu: Writing - review &
   editing, Writing - original draft, Validation, Resources, Project
   administration, Funding acquisition. All authors have read and approved
   the final manuscript.

Funding

   This study was supported by grants from the National Key Research and
   Development Program for Young Scientists (2022YFA1105800), the National
   Natural Science Foundation of China (82470957), Jilin Province Science
   and Technology Research (Project No. YDZJ202501ZYTS020), and the
   Bethune Plan Project of Jilin University (2024B14).

Data availability

   No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

   All animal experiments were conducted following Jilin University’s
   committee guidelines for animal experiments (approval number:
   20240290).

Competing interests

   The authors declare no competing interests.

Footnotes

   Publisher’s note

   Springer Nature remains neutral with regard to jurisdictional claims in
   published maps and institutional affiliations.

   Yang Liu and Duan Wang contributed equally to this work.

Contributor Information

   Ding Zhou, Email: zhouding@jlu.edu.cn.

   Xiaowei Xu, Email: xiaoweixu@jlu.edu.cn.

References