Design of an artificial natural killer cell mimicking system to target tumour cells

Abstract

NK cell mimics are assemblies of a cell membrane and a template that replicate biomimetic features and physicochemical properties, respectively. To develop this targeted drug delivery system, gelatin microspheres (cG) were fabricated using a water-in-oil emulsion and reinforced via DMTMM cross-linking to exhibit tunable Young's modulus, a critical parameter for cell-material interactions. These microspheres were subsequently coated with membranes derived from the human NK cell line KHYG-1 to form biomimetic NK cell mimics (cGCM), combining physicochemical control with bioinspired functionality. These engineered cGCM were non-toxic, non-inflammatory, and capable of reducing macrophage uptake by ~10% when incubated with differentiated THP-1 cells. In vitro studies demonstrated significant interaction/ proximity of the cGCM with cancer cells in 2D cultures of breast cancer cells (MDA-MB-231), 3D spheroids of liver (HepG2), and colon (HT-29) cancer cell models, and a zebrafish breast cancer xenograft (MDA-MB-231) model. The cGCM also evaded macrophage detection in a Kdrl:EGFP Spil:Ds Red zebrafish model. Furthermore, in a pilot assessment, loading and release of the sialyltransferase inhibitor (STI, 3Fax-Peracetyl Neu5Ac) using cGCM significantly reduced α-2,6 sialylation in 2D cultures of MDA-MB-231 cells, demonstrating the STI's intact functionality in inhibiting sialylation. By integrating bioinspired membranes with mechanically tunable gelatin-based carriers, our system demonstrates a multifunctional immune-mimicking platform with relevance to tissue engineering, tumour modelling, immune modulation, and drug delivery. These findings offer a promising foundation for future therapeutic strategies in cancer research and immuno-engineering.

Keywords: 3D spheroids; NK cell membrane; biomaterials; gelatin microspheres; zebrafish xenograft breast tumour model.

Scheme 1.

Overview on key conceptual elements of the study, including biomimetic design, tumour targeting capability, interaction with macrophages, and the localized delivery strategy.

Scheme 2.

Overview of NK cell mimics’ comprehensive study: design, interactions and therapeutic potential in tumour therapy. NK cell mimics were designed by coating KHYG-1 cell membrane onto gelatin microspheres, exhibiting moderate elasticity. NK cell mimics’ interaction with macrophages (2D in vitro differentiated THP-1 model) were investigated to examine their pro-inflammatory response and phagocytosis. The NK cell mimics’ interaction towards tumour cells without prior activation were evaluated in various in vitro models using 2D breast cancer cell cultures (MDA-MB-231), 3D spheroids of liver (HepG2), and colon (HT-29) cancer cells. Further, their interaction with breast cancer cells and macrophages in an in vivo zebrafish model was investigated. Finally, NK cell mimics’ loading and drug release behaviour was assessed using sialyltransferase inhibitor (STI, 3Fax-Peracetyl Neu5Ac) as a relevant model drug.

Figure 1.

Characterization of gelatin microspheres: (a) field emission scanning electron microscopic (FESEM) image cross-linked gelatin microspheres (cG) using 50 mM DMTMM, scale bar = 5 μm, (b) size distribution of cG by laser diffraction, (c) surface charge on gelatin microspheres before crosslinking (G) and after crosslinking (cG) using zetasizer (N = 3), Error bars represent standard deviations, (d) Young’s modulus estimation of gelatin hydrogels cross-linked with 10, 50, and 100 mM DMTMM, termed as cG1, cG2, cG2 respectively, N = 3, Error bars represent standard deviations, (e) design of microfluidics chip, consists of deformability array with gap sizing of 8 μm, and capture array with v-cup sizing of 13 μm, (f) representation of deformable gelatin microspheres (>8 μm) captured in the deformable array, scale bar = 10 μm, and (g) representation of aggregated spheres/smaller spheres/non-deformable spheres captured in V-cups, scale bar = 10 μm. Graphs are plotted in box and whiskers format with max and min value showing all data points.

Figure 2.

Characterization of NK cell mimics: (a) surface charge on cG, CM and cGCM, N = 3, (b) Fourier transform infrared (FT-IR) spectra of cG, CM, and cGCM, (c) SDS-PAGE gel stained with coomassie brilliant blue dye to assess the protein profile on cG, CM and cGCM, (d) Protein loading yield (%) on cGCM, N = 3, (e) Western blot to identify specific protein (NKp30, NKG2D, DNAM-1) on CL, CM and cGCM, and (f) Protein absorption analysis on cG and cGCM after incubation in FBS for 3 h at 37°C (cG vs cG_FBS, p < 0.0001, 95% CI = −3.004 to −2.355), N = 3. Abbreviations: cG: cross-linked gelatin microspheres; cG_FBS: cross-linked gelatin microspheres incubated in foetal bovine serum; cGCM_FBS: NK cell mimics incubated in foetal bovine serum; CL: KHYG-1 cell lysate; CM: isolated KHYG-1 cell membrane; cGCM: NK cell mimics; MW: molecular weight. Graphs are plotted in box and whiskers format with max and min value showing all data points. *p < 0.05. **p < 0.01. ***p < 0.001. ****p < 0.0001.

Figure 3.

Visualization of NK cell membrane onto gelatin microspheres: (a) transmission electron microscopy (TEM) images of uncoated gelatin microspheres (cG) and NK cell membrane-coated microspheres (cGCM). The cGCM images reveal a dark gelatin core with a surrounding lighter layer corresponding to the NK cell membrane coating. Due to the thin and low-contrast nature of the membrane layer, enlarged views of the red-highlighted regions are included to aid in visualizing coating integrity. Scale bar = 500 nm, (b) field emission scanning electron microscopic (FESEM) images of cG and cGCM, Scale bar = 1 μm, and (c) confocal laser scanning microscopic (CLSM) images of dextran cascade blue loaded gelatin spheres coated with WGA labelled KHYG-1 cell membrane, Scale bar = 10 μm. Abbreviations: cG: cross-linked gelatin microspheres; cGCM: NK cell mimics, WGA-FITC: wheat germ agglutinin-fluorescein isothiocyanate.

Figure 4.

In vitro cellular uptake studies of zymosan A (S. cerevisiae) bioparticles™, gelatin microspheres (cG), and NK cell mimics (cGCM) by differentiated (diff.) THP-1 cells as macrophages for 3 h using Image Stream X (cell/particle: 1:1): (a) For each condition (zymosan, cG–Cascade Blue, and cGCM–Cascade Blue), two representative single-cell uptake images are shown. Each pair includes a brightfield (BF) and the corresponding fluorescence image captured from the same field of view, highlighting the internalization and localization of particles within individual cells at 37°C and (b) comparative analysis of the uptake at 4°C and 37°C (cG vs cGCM, p = 0.0004, 95% CI = 4.397 to 16.45; cGCM vs Zymosan (p = 0.0035, 95% CI = −15.26 to −2.556), N = 3–5. Zymosan A was used as a positive control and uptake analysis at 4°C was used as a negative control. Zymosan A (S. cerevisiae) bioparticles™ was tagged with Alexa Flour™ 488, cG and cGCM was loaded with dextran cascade blue, diff. THP-1 cells was tagged with CD45_APC_Cy7 dye. Graphs are plotted in box and whiskers format with max and min value showing all data points. *p < 0.05. **p < 0.01. ***p < 0.001. ****p < 0.0001.

Figure 5.

Pro-inflammatory assessment of gelatin microspheres (cG), NK cell mimics (cGCM), NK cell membrane (CM) with differentiated THP-1 cells. Quantitative measurement of pro-inflammatory cytokines, (a) interleukin 1β (IL-1β), p< 0.0001, 95% CI [Media vs LPS] = −133.3 to −72.21, (b) tumour necrosis factor α (TNF-α) after 24 h incubation using enzyme-linked immunosorbent assay (ELISA), p < 0.0001, 95% CI [Media vs LPS] = −406.8 to −227.1, N = 3. Lipopolysaccharides (LPS) was used as positive control. Graphs are plotted in box and whiskers format with max and min value showing all data points. *p < 0.05. **p < 0.01. ***p < 0.001. ****p < 0.0001.

Figure 6.

In vitro interaction studies of gelatin microspheres (cG) and NK cell mimics (cGCM) with 2D cultures of MDA-MB-231 (human breast cancer cell line) and 3D spheroids of HT-29 (human colon cancer cell line) and HepG2 (human liver cancell cell line) using various cell: sphere ratios (1:1, 1:5, 1:10, 1:20) after 24 h incubation: (a) comparative analysis of number of cG and cGCM interacting or in the close proximity with the MDA-MB-231(One-way ANOVA, cG vs cGCM (1:5), p = 0.0003, 95% CI = −0.8344 to −0.2330; cG vs cGCM (1:10), p = 0.0018, 95% of CI: −0.7513 to −0.1499), N = 3, (b) comparative analysis of number of cG and cGCM in close proximity to spheroids of HT-29 small spheroid (one-way ANOVA, cG vs cGCM (1:5), p = 0.0019, 95% CI = −0.5650 to −0.115; cG vs cGCM (1:10), p = 0.0014, 95% CI: −0.5752 to −0.1216); HT- 29 large spheroid (one-way ANOVA, cG vs cGCM (1:5), p = 0.0011, 95% CI = −0.4875 to −0.1083; cG vs cGCM (1:10), p = 0.0002, 95% CI: −0.5436 to −0.1644), and HepG2 small spheroid (one-way ANOVA, cG vs cGCM (1:5), p = 0.0040, 95% CI = −0.3505 to −0.05568; cG vs cGCM (1:10), p = 0.0010, 95% CI: −0.4110 to −0.1162); HepG2 large spheroid (one-way ANOVA, cG vs cGCM (1:5), p = 0.0325, 95% CI = −0.2586 to −0.008217; cG vs cGCM (1:10), p = 0.0010, 95% CI: −0.3244 to −0.07408), N = 3; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, and (c) representative 2D image of interaction of cGCM with HT-29, 1:5 (cell: sphere ratio), NK cell membrane was tagged with CM-DiI dye, NK cell mimics were loaded with dextran-FITC, and nuclei of HT-29 with Hoechst, Scale bar = 100 μm. Graphs are plotted in box and whiskers format with max and min value showing all data points.

Figure 7.

In vivo interaction studies with MDA-MB-231 (breast cancer cell line) in zebrafish xenograft breast tumour model: (a) schematics of the site and time of injection of MDA-MB-231 cells, dextran-FITC loaded gelatin microspheres and NK cell mimics, (b) microscopic images (bright field, red and green fluorescence) of same embryos at 0 h post injection (hpi) of MDA-MB-231 cells, at 3 and 24 hpi of dextran-FITC loaded gelatin microspheres (cG) and NK cell mimics (cGCM), Scale bar = 500 μm, Zoom view of the selected region is also presented for 24 hpi, Scale bar = 500 μm, (c) quantification of the interaction with respect to tumour cells (One-way ANOVA, cG vs cGCM (24 hpi), p = 0.0032, 95.00% CI = −3.496 to −0.5426), and (d) with respect to spheres (One way ANOVA, cG vs cGCM (3 hpi), p = 0.0131, 95% CI = −10.51 to −0.9147, cG vs cGCM (24 hpi), p = 0.0002, 95% CI = −12.73 to −3.131), N = 20. Note. No cytotoxic payload was incorporated in these experiments, as the primary aim was to evaluate the interaction of NK cell mimics with tumour cells—specifically their attachment and spatial proximity—rather than inducing tumour cell death or inhibiting spheroid growth. Graphs are plotted in box and whiskers format with max and min value showing all data points. *p < 0.05. **p < 0.01. ***p < 0.001. ****p < 0.0001.

Figure 8.

In vivo interaction/cellular uptake studies of gelatin microspheres (cG) and NK cell mimics (cGCM) with macrophages in zebrafish model: (a) microscopic images (bright field, red and green fluorescence) presenting the interaction at 24 h post injection (hpi) of dextran-FITC loaded gelatin microspheres (cG) and NK cell mimics (cGCM) with DsRed macrophages in 3 day post fertilized (dpf) Kdrl:EGFP Spil:DsRed Zebrafish strain model, Scale bar = 200 μm. The zebrafish embryos’ blood vessel were EGFP tagged, and spheres were FITC labelled. Therefore, both blood vessels and spheres were shown in the green channel. The background green fluorescence for the blood vessels was removed in Image J for analysis and marked the position of the spheres with white arrow for better analysis of the images to avoid the overlap of green channels, and (b) quantification of interaction/cellular uptake of cG and cGCM with macrophages at 24 hpi of spheres (unpaired t-test, cGCM vs cG, *p < 0.0001, 95.00% CI = −8.894 to −4.596), N = 10. Graphs are plotted in box and whiskers format with max and min value showing all data points. *p < 0.05. **p < 0.01. ***p < 0.001. ****p < 0.0001.

Figure 9.

Sialyltransferase inhibitor (STI, 3Fax-Peracetyl Neu5Ac) studies: (a) Release of STIanalysis from gelatin microspheres (cG) and NK cell mimics (cGCM) using a high-performance liquid chromatography (HPLC) system, N = 3, Effect of STI, STI loaded gelatin microspheres (cG_STI), and STI loaded NK cell mimics (cGCM_STI) on MDA-MB-231 cells. Mean fluorescence intensity/cell of (b) Sambucus nigra (SNA) lectin (C vs STI, p< 0.0001, 95% CI = 2458–1208; C vs cG_STI, p = 0.0015, 95% CI = 644 to 393.5; C vs cGCM_STI, p = 0.0011, 95% CI = 1686 to 436.2; cG vs cG_STI, p = 0.0363, 95% CI = 1286–35.53; cGCM vs cGCM_STI, p = 0.0228, 95% CI = 1337 to 86.87), (c) fluorescence microscopic images of sambucus nigra (SNA) lectin expression, (d) peanut agglutinin (PNA) lectin, (e) Maackia amurensis (MAA) lectin, and (f) qheat germ agglutinin (WGA) lectin expression after treatment of 200 μM STI, cG, cGCM, cG_STI, cGCM_STI for 72 h at 37°C, N = 3. Abbreviations: C: Control (cells with only media), cG: gelatin microspheres, cGCM: NK cell mimics. Note. No cytotoxic payload was incorporated in these experiments. Note: This study aimed to assess the controlled release of functional glycan modulators, not tumour cell killing or spheroid growth inhibition. Graphs are plotted in box and whiskers format with max and min value showing all data points. *p < 0.05. **p < 0.01. ***p < 0.001. ****p < 0.0001.