Construction and characterization of PIKfyve inhibitor-based aMDFs

Given that PIKfyve inhibition has recently emerged as a promising strategy to potentiate cancer immunotherapy through macrophage reprogramming [15]. We first sought to identify a potent PIKfyve inhibitor suitable for constructing bioactive aMDFs. Fifteen previously reported PIKfyve inhibitors were synthesized and structurally confirmed by the 1H nuclear magnetic resonance (1H NMR, Fig. S1). Their autophagy inhibition activities were subsequently evaluated in both M1-and M2-polarized bone marrow-derived macrophages (BMDMs) by assessing cytoplasmic vacuolization, a hallmark of effective PIKfyve inhibition [16]. Qualitative microscopic observation (Fig. S2) together with quantitative flow cytometric analysis of lysosomal biogenesis revealed that HW-05-131-01 (hereafter denoted as HW) induced the strongest vacuolization response in M2 BMDMs while maintaining substantially lower activity in M1 BMDMs (Fig. 1a and b). Specifically, HW increased vacuolization in M2 BMDMs to 446 ± 6%, whereas only a modest enhancement (65 ± 1%) was observed in M1 BMDMs, resulting in the highest M2/M1 vacuolization ratio (6.8) among all tested compounds. These findings indicate that HW possesses both potent autophagy inhibition capability and preferential activity toward immunosuppressive M2 macrophages, making it an ideal candidate for constructing functional aMDFs.

Fig. 1: Preparation of HF@SLN NPs. Bioactivities of 15 isomers in (A) M2 BMDMs and (B) M1 BMDMs. (C) Theoretical simulation calculation between HW and Zn2+, red dots (oxygen), blue dots (nitrogen), light blue dots (fluorine), grey dots (carbon), light grey dots (hydrogen), purple dots (zinc). (D) 1H NMR titration results of different molar ratio between HW and Zn2+. (E) Counting rate of HF NPs prepared with different molar ratio between HW and Zn2+. (F) The TEM images of HF. Evaluating SLN, HF, and HF@SLN in (G) diameter, (H) zeta-potential, and (I) PDI. (J) TEM images of HF@SLN. TGA in (K) HF and (L) HF@SLN. Evaluating the stability of HF@SLN in (M) diameter and (N) PDI. Data are presented as means ± standard deviation (SD). The error bars are based on standard errors of the individual samples (n = 3), ***P < 0.001, ANOVA with Tukey's post-test. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

Despite its potent bioactivity, the direct encapsulation efficiency of free HW remained relatively low (26.4 ± 3.8%), which may limit its translational potential. Previous studies have demonstrated that pre-organizing hydrophobic therapeutics into particulate intermediates can markedly improve drug processability and encapsulation efficiency [[17], [18], [19]]. Inspired by these observations, we explored the coordination-driven assembly of HW with metal ions to generate aMDFs. Initially, Cu2+ was employed as the coordinating metal ion. However, the resulting nanoparticles exhibited poor colloidal uniformity with a polydispersity index (PDI) exceeding 0.3, indicating heterogeneous particle formation. In contrast, coordination between HW and Zn2+ generated nanoparticles with narrower size distributions, suggesting more controllable amorphous assembly behavior. Therefore, Zn2+ was selected for subsequent studies.

By the theoretical simulation calculation, the Zn2+ can interact with pyridine (N), imine (NH), NH-C=O (N), and C=O (O) with a distance of 3.69 Å, 1.95 Å, 3.10 Å, and 1.85 Å, which can theoretically form the chelates (Fig. 1c). According to the results of 1H NMR titration, HW showed chemical shift on 9.26, 8.57, 8.43, 8.27-8.21 ppm, 7.73-7.64 ppm, and 6.48 ppm, belonging to pyridine, imidazole ring, and benzene, respectively. After interacting with Zn2+, the peaks of 9.26 ppm (pyridine) and 8.57 ppm (imidazole ring) shifted to the high field. In addition, -NH coordinated with Zn2+ causing a shift of 6.48 ppm (benzene) to the high field. With the increasing amount of Zn2+, the peak at 8.65 ppm (pyridine and benzene) shifted to the low field; in addition, the peak shape changed from three peaks to two peaks (Fig. 1d). These 1H NMR titrations indicated that Zn2+ coordinated with N from pyridine, imine and benzene, and NH and C=O from Ar-NH-C=O, which is consistent with the theoretical simulation calculation.

We further investigated the preparation of NPs by using the different molar ratios between HW and Zn2+. The synthesis of NPs was confirmed by the counting rate using dynamic light scattering (DLS) measurements. The NPs (consisting of the coordination between HW and Zn2+ were denoted as HF in the following content) synthesized with the molar ratio at 4:1 (HW: Zn2+) showed the highest counting rate (14319 ± 124 kcps, Fig. 1e), indicating the formation of HF NPs. Furthermore, transmission electron microscopy (TEM) images also proved the successful synthesis of HF (Fig. 1f). TEM images of pure HW are shown in Fig. S3a.

Although metal-coordinated amorphous nanoparticles offer high drug loading capacity, their colloidal instability under physiological conditions remains a major limitation due to competitive interactions with ions and serum proteins [20]. To address this challenge, HF nanoparticles were further coated with a solid lipid nanoshell (SLN) to generate core–shell HF@SLN nanoparticles. This template-directed encapsulation strategy substantially improved nanoparticle stability while maintaining a high encapsulation efficiency of 78.6 ± 6.5% and loading efficiency of 6.7 ± 2.1%. Interestingly, HF@SLN exhibited a reduced hydrodynamic diameter (150 ± 40 nm) compared to bare HF nanoparticles (314 ± 52 nm) (Fig. 1g and h), likely because the SLN shell prevented further uncontrolled coordination-driven aggregation during particle maturation [21,22]. Concurrently, the surface zeta potential shifted from +10.0 ± 0.6 mV for HF to −26.1 ± 0.4 mV for HF@SLN, confirming successful lipid shell coating. Furthermore, HF@SLN displayed a more reduced PDI (0.24 ± 0.01) relative to bare HF nanoparticles (0.43 ± 0.09), indicating enhanced colloidal homogeneity (Fig. 1i), with the yield around 73.2 ± 8.7%.

TEM imaging further revealed a distinct core-shell morphology consisting of an electron-dense HF core surrounded by a lighter lipid shell (Fig. 1j). Thermogravimetric analysis (TGA) additionally confirmed the presence of the SLN coating. Unlike HF nanoparticles, which underwent rapid single-stage degradation between 300 and 400 °C, HF@SLN exhibited characteristic multistage weight loss profiles attributable to sequential lipid decomposition and HF degradation (Fig. 1k and l) [23]. X-ray diffraction (XRD) analysis further demonstrated the amorphous nature of HF@SLN and the successful masking of HF crystalline diffraction peaks by the SLN shell (Fig. S3b–e). Importantly, HF@SLN remained highly stable in both phosphate-buffered saline (PBS) and serum containing medium, maintaining particle diameters below 200 nm and PDIs below 0.3 over time (Fig. 1m and n), highlighting the effectiveness of lipid shell stabilization.

To investigate the generalizability of this coordination-driven strategy, additional therapeutics containing pyridine or imidazole motifs, including sorafenib, pexidartinib, and erlotinib, were coordinated with Zn2+. All resulting formulations formed nanoparticles with hydrodynamic sizes of approximately 200 nm, negative surface charges, and narrow PDIs (Fig. S4). These preliminary results suggest that Zn-mediated coordination may be extended to other structurally related small-molecule drugs possessing suitable coordination motifs. However, comprehensive physicochemical characterization and biological evaluation of these formulations were beyond the scope of the present study.