Bacterial Cellulose as Drug Delivery System for Optimizing Release of Immune Checkpoint Blocking Antibodies

Abstract

Immune checkpoint blocking therapy is a promising cancer treatment modality, though it has limitations such as systemic toxicity, which can often be traced to uncontrolled antibody spread. Controlling antibody release with delivery systems is, therefore, an attractive approach to reduce systemic antibody spread and potentially mitigate the side effects of checkpoint immunotherapy. Here, bacterial cellulose (BC) was produced and investigated as a delivery system for optimizing checkpoint-blocking antibody delivery. BC was produced in 24-well plates, and afterward, the edges were removed to obtain square-shaped BC samples with a surface of ~49 mm². This customization was necessary to allow smooth in vivo implantation. Scanning electron microscopy revealed the dense cellulose network within BC. Human IgG antibody was included as the model antibody for loading and release studies. IgG antibody solution was injected into the center of BC samples. In vitro, all IgG was released within 24 to 48 h. Cell culture experiments demonstrated that BC neither exerted cytotoxic effects nor induced dendritic cell activation. Antibody binding assays demonstrated that BC does not hamper antibody function. Finally, antibody-loaded BC was implanted in mice, and serum measurements revealed that BC significantly reduced IgG and anti-CTLA-4 spread in mice. BC implantation did not induce side effects in mice. Altogether, BC is a promising and safe delivery system for optimizing the delivery and release of checkpoint-blocking antibodies.

Keywords: bacterial cellulose; cancer immunotherapy; checkpoint blocking therapy; drug delivery system; side effects; sustained release.

Figure 1

Characteristics and morphology of bacterial-derived cellulose (BC). (a,b) The photographs show the appearance of a freshly produced (native) BC fleece (referred to as round BC), which has the shape and size of the wells of a 24-wells culture dish. Image (c) shows a fleece of which the edges were cut (cut BC). Excess liquid droplets were removed by holding the BC fleece with forceps (d), and afterward, the fleeces were weighted (e). Figures (f,g) represent representative SEM images for native and IgG-loaded BC, respectively. Data in (e) are shown as mean ± SD for triplicate measurements.

Figure 2

Human IgG loading and in vitro release were performed with the ‘injection method’. (a) A total of 50 µg IgG in a volume of 25 µL PBS was injected into the center of a BC fleece. To visualize the injection depot, the IgG solution was supplied with Trypan blue. The loaded fleeces were then put in blocking buffer (1% BSA/0.05% Tween-20 in PBS) and IgG release was followed. At several time-points, the fleeces were taken out, and the Trypan blue injection spots were photographed. Graph (b) shows the cumulative IgG release in %, with the inset of the first 24 h shown in (c). Release data are shown as cumulative release, which are displayed as mean ± SD for a triplicate measurement.

Figure 3

Cytotoxicity evaluation of BC extracts in MC38 tumor cells with 7-AAD staining and MTS. BC extracts were prepared by dissolving 1 g of BC in 50 mL culture medium, which yielded an extraction ratio of 20 mg/mL. Empty extracts or extracts supplied with human IgG or anti-CTLA-4 (both starting at a concentration of 50 µg/mL) were tested on MC38 cells. DMSO was used as positive cell killing control. The primary measure was cell viability, which was assessed by 7-AAD staining (live-dead cells exclusion marker; figures a–c). Higher 7-AAD MFI signal is a hallmark of dying cells with leaky cell membranes. Cytotoxicity was also assessed using a cell metabolism assay (MTS), in which higher cell metabolism is indicative of higher cell viability (d–f). Cell metabolism was measured 48 h after incubation. All data are shown as mean ± SD for triplicates. Only the DMSO treatment significantly decreased cell viability, which was assessed with a Student’s t test, with **** denoting p < 0.0001.

Figure 4

BC extracts do not induce DC maturation. BC extracts (max concentration 20 mg/mL) were co-cultured with D1DCs. After 48 h, D1DCs were harvested to assess the expression of CD40, CD86 and MHC-II activation markers with FACS. Debris, doublet cells and dead cells (7-AAD positive) were gated out (a). Next, the fold increases in MFI (the MFI of treated cells relative to untreated controls) were calculated for poly (I:C) (b) and BC extracts (c). Data are shown as mean ± SD for triplicate measurements.

Figure 5

BC does not affect anti-PD-L1 binding capacity to B16F10 cells. In total, 50 µg of anti-PD-L1 in 25 µL PBS was injected in a BC fleece. After 7 d, the released antibodies from the BC were harvested to assess the binding capacity. B16F10 cells were incubated 24 h with 10 IU/mL IFN-γ to upregulate PD-L1 expression. Thereafter, the cells were harvested, washed and incubated with the anti-PD-L1 that was released from the BC samples. In the subsequent step, cells were stained with fluorescent labeled anti-rat IgG (Alexa 488 donkey-anti-rat IgG) (a). Alexa 488 fluorescence was a measure for PD-L1 binding. This was compared with cells to which stock (fresh) of anti-PD-L1 antibody was added (‘soluble control’). In (b), the gating strategy and representative histograms with the Alexa 488 emission peaks are displayed, with the quantification of the mean fluorescent index (MFI) depicted in (c). In (c), data are shown as mean ± SD for a triplicate measurement, with differences being assessed with Student’s t test and N.S. denoting ‘not significant’.

Figure 6

BC reduces serum antibody levels in vivo. Photographs in (a) depict in chronological order the procedures of the BC implantation, starting with loading, placing the BC sample underneath the skin, wound closure, visual inspection of the skin and removing the BC implant at D21 (after the mice were killed). Body weight measurements are depicted in (b), for the BC-treated mice, the body weight differences at various time-points were compared with the initial body weight at D0. Body weight differences were assessed with a paired Student’s t-test, with NS denoting no statistical differences compared to the body weights at D0. In (c) and (d), the serum IgG and anti-CTLA-4 levels are shown for several time-points after treatment, respectively. Statistical differences between the BC and PBS group were assessed with an unpaired Student’s t-test and are denoted as * p < 0.05, ** p < 0.01, *** p < 0.001 and **** p < 0.0001.