Prodigiosin-functionalized Lactobacillus acidophilus ghost: A bioinspired combination against colorectal cancer cells


 Lactobacillus acidophilus ghosts (LAGs) with the unique safety of a probiotic, inherent tropism for colon cells and multiple bioactivities offer great promise as drug carrier for colon targeting. We report herein on a novel bioinspired drug delivery system against colorectal cancer (CRC) cells based on LAGs functionalized with prodigiosin (PG), a proapoptotic bacterial metabolite. LAGs were prepared by a chemical method and highly purified by density gradient centrifugation. Multiple microscopic and staining techniques characterized LAGs by a relatively small size, size uniformity, a relatively large internal volume devoid of cytoplasmic and genetic materials and an intact negatively charged envelop. PG was highly bound to LAGs cell wall, generating a physiologically stable bioactive entity (PG-LAGs) active against HCT116 CRC cells at the cellular and molecular levels. Cell viability data underlined cytotoxicity of PG and LAGs and LAGs-induced enhancement of PG selectivity for HCT116 cells. Combination and Dose reduction indices anticipating dose reduction of PG and LAGs. Molecularly, PG-LAGs significantly modulated the expression of apoptosis-related biomarkers, caspase 3, P53 and BCL2, in favor of cytotoxicity against HCT116 cells relative to PG and 5-fluorouracil. Accordingly, LAGs offer promise as novel drug carrier for targeted colon delivery and PG-LAGs may bring therapeutic benefits in colorectal carcinoma.


Introduction
Bacterial cells have attracted growing attention as a sustainable source of bacterial-derived structures and byproducts of great bene t in biomedical applications [1][2][3] . Amongst bacterial-derived structures, bacterial ghosts (BGs) represent an advancing biotechnological platform in the prevention and treatment of disease. BGs are intact envelopes of bacterial cells, emptied of their cytoplasmic and genetic materials by gentle poring methods. While the genetic method has been commonly utilized for the preparation of Gram negative bacterial ghosts 4 , chemical methods allowed the generation of BGs of both Gram negative and Gram positive bacteria 5,6 .
BGs retain the cellular morphology and surface structural components of native cells, most importantly antigenic proteins to be recognized by the immune system as well as mbriae and adhesins to facilitate targeting and binding to different cells and tissues 7,8 . Such outstanding surface features endow BGs with the ability to stimulate immune responses, provide natural adjuvant properties and serve as vector in the delivery of human and veterinary vaccines 9 . Equally important, BGs are emerging as a multifunctional platform in drug and gene therapy. In this regard, BGs may have therapeutic potentials on their own and may synergize the activity of other drugs via BG-induced immunostimulatory effects 10 .
BGs show several intrinsic merits as drug carriers compared with biomaterial-based delivery systems.
These include primarily stability in the biological milieu, large internal space, greater safety allowing for larger doses in addition to inherent tropism-targeted drug delivery 2,11 . Promising effects have been achieved recently with BG-mediated targeted delivery of anticancer drugs 12,13 and macrophage delivery of anti-infective agents 14 .
However, bacteria used for the preparation of BGs to date are mostly pathogenic and the utilization of non-pathogenic bacteria would greatly enhance the safety pro le of BG-based drug delivery systems. To this end, Gram positive lactic acid probiotic bacteria with signi cant health-promoting effects, are generally recognized as safe (GRAS) and have demonstrated immunomodulatory capability 15 . Moreover, several anticancer effects have been associated with their structural components and metabolites 16,17 .
Among the best known lactic acid bacteria, Lactobacillus acidophilus (LA) naturally colonize the human colon with resilient adhesive properties as a result of binding to mucin 18 and recognition of their antigens by colon cells and the adjacent immune system 19 .
Such distinct properties render LA ghosts (LAGs) highly promising drug carriers for targeting colorectal cancer (CRC) cells. To date, drug delivery by BGs has been generally restricted to chemotherapeutic agents such as 5-uorouracil 13 and doxorubicin 20 . Nevertheless, functionalizing BGs with bacterial metabolites having colorectal cancer inhibitory activity may establish a new microbially-derived biotherapeutic platform for the treatment of CRC, one of the most common malignancies worldwide. In this regard, prodigiosin (PD), a secondary metabolite of Serratia marcescens (S. marcescens), induces signi cant cytotoxic effects in diverse cancer cell lines 21 without induction of marked toxicity in nonmalignant cell lines 22 . PG's selective anti-colorectal actions are believed to occur via apoptosis as a result of alteration of the expression of apoptosis-related genes 23,24 and restoration of p53 tumor suppressor activity in chemoresistant CRC stem cells 25 . PG was also shown to reduce survivin levels and increase caspase-3 and miRNA-16-1 levels in CRC stem-like cells 22,24 . Besides, PG sensitizes CRC cells to cell death induced by anti-cancer drugs 22,26 .
The objective of the present study was to develop PG-functionalized LAGs (PG-LAGs) as a novel microbially inspired anti-cancer drug delivery system integrating the aforementioned potentials of PG and LAGs for targeting CRC cells. The study involves preparation, puri cation, and characterization of blank LAGs as well evaluation of LAGs as drug carrier in terms of size, size uniformity, PG loading capacity, and release properties. Finally, the activity of PG-LAGs against CRC cells in comparison with PG, LAGs, and 5uorouracil as standard treatment was assessed at the cellular level using HCT116 CRC cell line and molecularly by determining the level of three apoptosis-related biomarkers, caspase 3, P53 and BCL 2 in CHT116 cellular proteins.

Results
Preparation and differentiation of Lactobacillus acidophilus ghosts (LAGs). The chemical SLRP used for the preparation of LAGs using NaOH and SDS as described in the methods section generated a bacterial mass consisting of a mixture of bacterial LAGs, dead un-evacuated LA cells and the remaining live LA cells which tolerated chemical lysis. Differentiation of LAGs from LA cells in the bacterial mass was achieved using a new light microscopic method based on toluidine blue, a dye known to bind to DNA. Figure 1 indicated that live LA cells acquired the intense toluidine blue color (Fig. 1a) while the presence of LAGs, presumably devoid of genetic material, led to an obvious reduction in the blue staining of the bacterial mass (Fig. 1b). Moreover, comparison of intracellular content by TEM imaging of live LA cells ( Fig. 1c) and the bacterial mass (Fig. 1d) veri ed the presence of a mixture of evacuated LAGs and unevacuated LA cells in the mass. Attempts to increase the yield of LAGs in the bacterial mass by increasing the concentration of NaOH and SDS or applying the chemical treatment under static conditions resulted in rupture of the ghost cell wall ( Fig. 1e and f).
Puri cation of L. acidophilus ghosts (LAGs) by density gradient centrifugation. Subjecting a suspension of the bacterial mass in saline to a two-step density gradient centrifugation resulted in effective separation of light-weight LAGs as a high purity fraction. The rst centrifugation steps at 168 xg for different time intervals (6, 9, 12, 15, 18 min) were effective in eliminating the heavy dead un-evacuated LA cells as well as the remaining live cells. The nal centrifugation step (20 min at 672 xg) precipitated the lighter bacterial ghosts. As shown in Fig. 2, light microscopy indicated absence of toluidine blue-stained LA cells (Fig. 2a) which was veri ed by TEM (Fig. 2b) and the absence of growth upon inoculating a MRS plate with the puri ed LAGs for 48 h at 37°C (Fig. 2c).
Characterization of puri ed L. acidophilus ghosts (LAGs). Puri ed LAGs were characterized for morphology and intracellular content by SEM and TEM respectively ( Fig. 3Aa-f), expulsion of genetic materials and intracellular cytoplasm and cell wall proteins by gel electrophoresis (Fig. 3Ba and b), in addition to integrity of the ghost capsule (Fig. 3Ca) and preservation of the cell wall surface negative charge (Fig. 3Cb) using staining methods.
LAG morphology and intracellular content. As indicated by SEM at 35000x of live LA cells (Fig. 3Aa) and LAGs (Fig. 3Ab), the surface structures of LAGs appeared intact similar to those of live cells except for the existence of pores from which the intracellular contents were expelled. TEM imaging ( Fig. 3Ac-f) indicated that compared with live LA cells (Fig. 3Ac), LAGs appeared empty with retention of a cohesive cell wall ( Fig. 3Ad-f). The calculated internal volume of LAGs was 0.13 µm 3 approximately and the pores with a mean size of 153.63 ± 12.23 nm were located at the division sites representing the weak point of the bacterial wall (Fig. 3Ae, f).
Elimination of genetic materials and proteins. Elimination of genetic materials was con rmed by agarose gel electrophoresis of DNA following DNA extraction from LAGs and live LA cells. As revealed by Fig. 3Ba, LAGs were devoid of nucleic acids. Expulsion of the total protein content of live LA and proteins of LA envelops were veri ed by SDS-PAGE electrophoresis (Fig. 3Bb). The total protein content of live LA cells (Lane 1) was larger than that of LAGs (Lane 2), con rming evacuation of LAGs from cytoplasmic materials. In addition, Lane 3 for the protein pro les of the supernatant after chemical treatment revealed expulsion of the LAGs proteins into the supernatant.
Integrity and negativity of the L. acidophilus ghosts. The integrity and surface negativity of LAGs envelop were demonstrated by light microscopic imaging of LAGs in comparison with live LA cells using crystal violet/copper sulphate (Antony's stain) and nigrosine staining, respectively (Fig. 3C). Antony's staining revealed existence of an intact capsule in both live bacteria (Fig. 3Ca) and LAGs (Fig. 3Cb). Insets in both gures showed a faint copper sulphate blue halo (capsule) around crystal violet-stained purple cells indicating integrity of the capsule. Nigrosine staining a rmed that the negativity of the cell wall of live LA cells (Fig. 3Cd) was preserved in LAGs ( Fig. 3Cd) after chemical treatment, both appearing as bright halos on the dark background.
Prodigiosin-functionalized L. acidophilus ghosts Preparation. Loading of PG via incubating blank LAGs with PG solution in a methanol: acetic acid solvent system was affected by the solvent composition, PG concentration, incubation time and agitation rate. In a series of single point experiments, the highest PG loading e ciency achieved by incubating LAGs with a 2 mg/mL PG in a methanol: acetic acid (1:1) for 2 h with agitation at 200 rpm was 3.9 %.
Veri cation of PG entrapment in LAGs. PG entrapment in LAGs was veri ed by digital photography, light microscopy, and TEM ( Fig. 4a-h). Compared with blank LAGs, the pellet of PG-LAGs acquired the characteristic red color of PG as indicated by digital photos (Fig. 4a and e). PG-LAGs also appeared as red vesicles under the 100x lens of the light microscope without staining ( Fig. 4b and f). TEM imaging at 20000x revealed that the density of the cell wall of blank LAGs (Fig. 4c)  Release studies. Data for PG release from PG-LAGs at 37°C in media of different pH (acetate buffer pH 5.5 and phosphate buffer saline pH 7.4 with or without the addition of 5% methanol) at 100 rpm for 24 h indicated high PG retention by LAGs which could not be overcome by the inclusion of up to 3% Tween 80. Similar results were obtained for the release of PG in simulated gastric and simulated intestinal uids. The IC50 values computed from cell viability data (Table 1) indicated that the IC50 of PG (2.026 ± 0.17 µg/mL) was slightly higher than that of 5-FU (1.483 ± 0.19 µg/mL) with a non-statistically signi cant difference (p > 0.05). Moreover, a combination of PG and LAGs in PG-LAGs induced ≈ 8.5-fold reduction in the IC50 of LAGs. The selectivity of the test preparations for HCT116 CRC expressed as the selectivity index (SI) (IC50 values in normal human broblasts relative to those in HCT 116 cells) indicated signi cantly greater (p < 0.05) selectivity of PG (17.87), LAGs (28.24) and PG-LAGs (43.60) for HCT116 cells relative to 5-FU (8.63). Furthermore, the SI index of PG and LAGs was signi cantly increased (p < 0.05) by their combination in PG-LAGs. Combination index (CI) and dose reduction index (DRI). Values for the Combination Index (CI) and Dose Reduction Index (DRI) are shown in Table 2. CI is a parameter used to indicate synergistic (CI < 1), additive (CI = 1) or antagonistic (CI > 1) effects of 2 drugs in combination while the Dose Reduction Index (DRI) expresses the synergy of two drugs and indicate the fold-decrease in the dose of each drug independently related to their dose in the combination. Both CI and DRI values were generated by analysis of the combinatorial cytotoxic effect of PG-LAGs on HCT116 cells following 24 h treatment at EC50 (Effective dose for 50% cell viability inhibition achieved by the combination). The CI was 0.997, indicating a synergistic cytotoxic effect of PG and LAGs. The concentrations of PG and LAGs as single components at EC50 were 2.026 µg/mL and 393.440 µg/mL, respectively. These were reduced by combining both agents in PG-LAGs to 1.79 µg/mL and 44.68 µg/mL respectively, producing DRI values > 1 with 1.13 -and 8.81-fold dose reduction for PG and LAGs, respectively. Effect of test preparations on apoptosis-related biomarkers. The effect of PG, LAGs, and PG-LAGs in comparison with 5-FU on the level of three apoptosis-related biomarkers, namely caspase-3, P53 and BCL 2 per mg of HCT116 cellular protein is illustrated in Fig. 6a-c.
Caspase 3 activity. Figure 6a indicated a signi cant increase (p < 0.05) in the activity of the apoptotic caspase 3 by all treatments relative to control. The difference between single treatments was not statistically signi cant. However, PG-LAGs exerted a signi cantly greater (p < 0.05) increase in caspase 3 activity relative to its single PG and LAGs components and 5-FU. p53 protein level. As shown in Fig. 6b, expression of the pro-apoptotic P53 protein was signi cantly increased (p < 0.05) by all treatments, though the greatest upregulating effect was exerted by PG-LAGs. The PG-LAGs effect was signi cantly greater (p < 0.05) than that of its single components but not 5-FU.
BCL 2 protein level. The level of the anti-apoptotic cellular BCL 2 protein (Fig. 6c) was signi cantly (P < 0.05) reduced by all treatments. However, reduction by PG-LAGs was signi cantly greater (P < 0.05) than that exerted by its single components but not 5-FU.

Discussion
LAGs were prepared, puri ed, and characterized for application as novel drug carriers for colon targeting.
The chemical SLRP method 27 with modi cation involving replacement of H 2 O 2 and CaCO 3 with NaOH and SDS at their respective MGC and MIC generated a bacterial mass containing LAGs in addition to unevacuated dead LA cells and live LA cells which resisted chemical lysis. Differentiation of LAGs from other cells could be achieved utilizing a new simple and economic method based on toluidine blue, a basic thiazine molecule known to highly bind to nucleic acids 28 . The differentiative ability of the toluidine staining method was veri ed by light microscopy and TEM (Fig. 1a-d), which indicated incomplete lysis of LA cells. Attempts to increase the proportion of LAGs in the mass by increasing NaOH and SDS concentrations or preparing LAGs under static conditions resulted in cell wall rupturing ( Fig. 1e and f). As such, the product of chemical lysis was a bacterial mass consisting of LAGs in combination with live and dead un-evacuated LC cells.
A puri ed fraction of LAGs could be separated from the bacterial mass by density gradient centrifugation 29 , a method not documented to date for the separation of ghosts from un-evacuated bacterial cells.
Microscopical and microbiological veri cation of the purity of the separated ghost fraction (Fig. 2) ascertained the e ciency of the density gradient centrifugation as a practical method for the separation of ghosts from a bacterial mass. Characterization of the puri ed LAGs by SEM and TEM, gel electrophoresis and light microscopy (Figs. 3A-C) collectively veri ed the morphology of LAGs as empty vesicles devoid of cytoplasmic content with an internal volume of 0.13 µm 3 approximately and surrounded by an intact and cohesive negatively charged cell wall having pores with a mean size of 153.63 ± 12.23 nm located at the division sites representing the weak point of the bacterial cell wall. Agarose gel electrophoresis and SDS-PAGE con rmed elimination of genetic materials and intracellular proteins respectively from LAGs with preservation of their membrane proteins. Retention of an intact capsule around LAGs following chemical treatment is a crucial factor in their adhesion to cellular membranes, a process mediated by the capsular polysaccharides cohesive layer 30 . Moreover, the surface negativity of LAGs denoted retention of lipoteichoic acid, a strongly negatively charged surfaceassociated element that contributes to the integrity of the membrane of LA cells 31 and their adhesion to colon cells 32,33 . This implied maintenance of targeting ability of the live cells by the ghost.
Findings obtained so far suggested bene ts of puri ed LAGs as a drug carrier characterized by an intracellular space of 0.13 µm 3 approximately and a negatively charged intact membrane as potential sites for drug loading in addition to surface characteristics favoring inherent tropism for colon cells. Applicability of LAGs in drug delivery was supported by favorable pharmaceutical attributes including a relatively small size (1.13 ± 0.13 µm) and size distribution (PDI 0.15 ± 0.09). Promotion of LAGs as bioinspired anti-CRC delivery system was achieved by the incorporation of PG, a secondary bacterial metabolite with established apoptotic activity against CRC. However, the development of PGfunctionalized LAGs was challenged by the high hydrophobicity of PG (log P octanol−water 5.16) and its poor solubility in aqueous physiological media, a well-documented PG formulation problem [34][35][36] . A simple incubation method allowed 3.9% loading of LAGs by varying the composition of a methanolacetic acid solvent system, temperature, incubation time and agitation rate. The solvent system had a key role in PG loading as it was shown to promote permeabilization of the bacterial phospholipid bilayer 37,38 and protonation of PG 39,40 . This would enhance PG binding to the negatively charged LAG membrane via strong electrostatic interaction PG loading which was con rmed by digital and microscopic imaging (Fig. 4a-h).
In vitro release data obtained in buffers with different pH and simulated gastrointestinal uids at 100 rpm  13 . This is a valuable merit of LAGs as bioactive carrier for colon targeting knowing that some bacterial ghosts might exert a cancer cell proliferating effect 43 . The cytotoxic activity of LAGs was signi cantly enhanced by PG loading, achieving ≈ 8.5-fold reduction in LAGs IC50 as well as an increase in their selectivity for HCT116 cancer cells, surpassing 5-FU selectivity ( Table 1). The notable safety of LAGs to normal human broblasts cells, their inherent tropism for colon cells and possible cellular uptake may account for the relatively high selectivity of PG-LAGs for CRC cells. Importantly, analysis of the combinatorial anticancer effects of PG and LAGs in PG-LAGs following a relatively short 24 h incubation with HCT116 cells pointed to synergism that was associated with an anticipated 1.13-fold and 8.81-fold reduction in the dose of PG and LAGs respectively in the combination at EC50 (Table 2).
Anti-cancer merits of PG-LAGs demonstrated at the cellular level were substantiated at the molecular level by signi cant modulation of the levels of apoptosis-related biomarkers in HCT116 intracellular protein (Fig. 6a-c). PG and LAGs in comparison with 5-FU showed a signi cant increase in the intracellular apoptotic caspase 3 activity and P53 protein level and signi cant downregulation of the anti-apoptosisrelated B-cell lymphoma 2 (BCL 2) protein level relative to untreated control cells. LAGs-induced molecular effects supported the intrinsic tropism and cytotoxicity of LAGs against CRC cells, an issue warranting further investigation. Upregulation of caspase 3 activity and apoptosis of HCT116 by PG was shown previously to depend on a decrease in the mRNA and protein levels of the proto-oncogene survivin 24 . Induction of caspase-3 activation through the survivin inhibition pathway in both HTC116 and HT-29 CRC cells 23,24 may provide a molecular mechanism for PG-induced apoptosis.
PG and LAGs as well as 5-FU also upregulated P53, a tumor suppressor frequently mutated or inactivated in colorectal cancer. The individual effects of PG and LAGs were signi cantly exceeded by that of PG-LAGs. PG was shown to restore the p53 pathway known to target CRC stem cells representing a considerable proportion of HCT116 cells via activation of p73, a member of the p53 family 25 , leading to cell growth inhibition. BCL 2 which belongs to the BCL 2 protein family plays an essential role in the intrinsic mitochondrial apoptotic signaling pathway. Thus, the anti-apoptotic BCL 2 protein may support cell survival and induce drug resistance in cancer cells 49 . The level of BCL2 protein in the untreated control HCT116 cells was signi cantly reduced by all test preparations, though reduction by PG-LAGs was signi cantly greater than that of PG and LAGs but not 5 Puri cation of L. acidophilus ghosts (LAGs) by density gradient centrifugation. LAGs were puri ed by separation from live and dead un-evacuated LA cells by subjecting the resuspended bacterial mass to a density gradient centrifugation technique 29 for different time intervals (6, 9, 12, 15, 18 min) at 168 x g. This was followed by centrifugation of each separated fraction for 20 min at 672 xg to precipitate LAGs. To assess the e ciency of the separation procedure, the puri ed LAGs obtained at the end of the Release studies. The in vitro release of PG from PG-LAGs was studied at 37°C by a dialysis method 43 using acetate buffer pH 5.5 and phosphate buffer saline (PBS) pH 7.4 with or without the addition of 5% methanol or up to 3% Tween 80 as release media. Brie y, 1 mL of PG-LAG suspension in 0.5% saline was placed in a dialysis bag (VISKING ® dialysis tubing MWCO 12,000-14,000) and shaken in 10 mL of the release medium at 60 rpm for 24 h protected from light. In addition, PG release was examined in simulated gastrointestinal uids by immersing PG-LAGs in simulated gastric uid (SGF, 10 mL pepsin / HCl, 320 mg/100 mL, pH 1.2) for 2 h followed by immersion in simulated intestinal uid (SIF, 10 mL of pancreatin / PBS, 1 g/100 mL, pH 7.2) for 4 h. Samples of the release medium (2 mL) were withdrawn for analysis at different time intervals and replaced with 2 mL of fresh medium at 37°C. The concentration of PG released was determined by UV-Vis spectrophotometry at λ max 535 nm.
Cytotoxicity studies. The cytotoxicity of PG-LAGs in comparison with blank LAGs, PG, and 5-uorouracil  Caspase-3 activity. The level of active caspase-3 in cell lysates was determined using a colorimetric kit (# ab39401, Abcam) as reported 55 . A p-nitroaniline moiety released after hydrolysis of the peptide substrate (Ac-DEVD-pNA) by active caspase-3 in cell lysates was quanti ed using a calibration curve constructed from absorbance at 405 nm measured on a microtitre plate. Data are the mean ± SEM of three replica.
P53 and BCL 2 proteins. The levels of P53 and BCL 2 proteins per gram of total cellular protein in the cell lysate were determined using Human Immunoassay Elisa kits (ab171571-p53 Human SimpleStep ELISA ® Kit) and (ab119506 -Bcl-2 Human ELISA ® Kit) respectively according to the manufacturer's instructions. The p53 and Bcl-2 protein levels were normalized by cell viability for each treatment. Independent experiments with three replicates were performed for each protein and the mean abundances from each of the experiments were pooled for statistical analysis. Finally, the amount of P53 and BCL 2 per g of total cellular protein was determined.

Statistical analysis
Data were analyzed using Graph Pad Prism ® version 6 software (GraphPad Software Inc., CA, USA).