Photoactivatable Surface-Functionalized Diatom Microalgae for Colorectal Cancer Targeted Delivery and Enhanced Cytotoxicity of Anticancer Complexes

Systemic toxicity and severe side effects are commonly associated with anticancer chemotherapies. New strategies based on enhanced drug selectivity and targeted delivery to cancer cells while leaving healthy tissue undamaged can reduce the global patient burden. Herein, we report the design, synthesis and characterization of a bio-inspired hybrid multifunctional drug delivery system based on diatom microalgae. The microalgae’s surface was chemically functionalized with hybrid vitamin B12-photoactivatable molecules and the materials further loaded with highly active rhenium(I) tricarbonyl anticancer complexes. The constructs showed enhanced adherence to colorectal cancer (CRC) cells and slow release of the chemotherapeutic drugs. The overall toxicity of the hybrid multifunctional drug delivery system was further enhanced by photoactivation of the microalgae surface. Depending on the construct and anticancer drug, a 2-fold increase in the cytotoxic efficacy of the drug was observed upon light irradiation. The use of this targeted drug delivery strategy, together with selective spatial–temporal light activation, may lead to lower effective concentration of anticancer drugs, thereby reducing medication doses, possible side effects and overall burden for the patient.


Introduction
Targeted drug delivery for the confined treatment of colonic diseases is becoming increasingly important to address locally life-threatening disorders like colorectal cancer (CRC). Colon-focused drug delivery (CFDD) permits direct treatment at the disorder site, lowers drug dosing and decreases the chance of systemic side-effects. Indeed, most available chemotherapeutics are poorly tolerated by patients and induce severe side effects that are unbearable in many cases. CFDD is a strategy actively pursued to address inflammatory bowel diseases [1][2][3] and most importantly CRC. Indeed, despite advances in diagnostic and therapeutic modalities, CRC is the third most commonplace cause of cancer-related deaths worldwide. Generally, CRC originates from the inner wall of the colorectal epithelium, develops as a polyp and finally spreads by invading nearby lymph nodes and other organs. As pointed out in recent reviews, issues like the path of the gastro-intestinal (GI) tract, dynamic pH changes and the inability to discern healthy tissues from cancerous ones must be overcome in order to develop effective materials for CFDD treatment [4][5][6].
Pharmaceutics 2020, 12 Different porous materials such as zeolites, nanoparticles, dendrimers/polymers, nano-hydrogels or metal organic frameworks were shown to be prospective tools in the field of drug delivery systems (DDSs) and cancer-targeted treatments [7][8][9][10][11][12][13][14][15][16][17]. In terms of material design for colon-focused drug delivery, several approaches have been described. Salleh et al., e.g., evaluated the performance of gelatin-coated type Y zeolite for the controlled release of the anticancer drug zerumbone [18] encapsulated in its pores, showing sustained and prolonged drug release over 24 h [19]. Li and coworkers addressed CFDD via a nano-in-micro dual drug delivery platform composed of halloysite nanotubes and a pH-responsive polymer. This clay mineral was loaded with atorvastatin and celecoxib, which were released only at pH 7.4 and effectively inhibited colon cancer cell proliferation [20]. Xu et al. recently presented up-conversion nanoparticles (UCNPs) to trigger cancer immunotherapy in CRC by NIR-induced photodynamic therapy (PDT) to directly destroy tumor cells and to stimulate immune responses by triggering the maturation of dendritic cells and secretion of cytokines [21]. Others have used gold- [22] and folate-coated nanoparticles for active CFDD of nanotherapeutics [23]. Finally, different authors have used reactive oxygen species (ROS)-responsive nanoplatforms and polymers as drug delivery systems with excellent results [6].
We have recently begun to develop CFDD materials based on environmentally sustainable, abundant and inexpensive diatom microalgae (DEMs) [24] with the aim of designing an innovative structure able to simultaneously target the tumor site and release loaded chemotherapeutics in its immediate proximity [25]. DEMs, as the fossil frustules of diatoms, recently gained attention for their use in drug delivery due to their biocompatibility and ability to shuttle and slowly release different drugs [26][27][28][29]. Diatoms are enclosed in a three-dimensional, highly ordered silica shell (called frustule) and represent an inexpensive and well-engineered source of microporous silica [30][31][32][33]. Our original proof of concept relied on the chemical functionalization of the diatom microalgae's surface with vitamin B 12 , which allowed specific binding of the material to CRC cells and, therefore, discrimination between healthy and diseased tissues. The next step in our concept is illustrated in Figure 1. This bio-inspired hybrid multifunctional drug delivery system is modified by the introduction of photoactivatable units (orange spheres in Figure 1), which are envisioned to act in concert with the loaded chemotherapeutic drugs (light blue squares) in order to (a) sensitize the tumor and (b) lower the overall drug dose needed for effective treatment.
Pharmaceutics 2020, 12, x FOR PEER REVIEW 2 of 22 Different porous materials such as zeolites, nanoparticles, dendrimers/polymers, nanohydrogels or metal organic frameworks were shown to be prospective tools in the field of drug delivery systems (DDSs) and cancer-targeted treatments [7][8][9][10][11][12][13][14][15][16][17]. In terms of material design for colonfocused drug delivery, several approaches have been described. Salleh et al., e.g., evaluated the performance of gelatin-coated type Y zeolite for the controlled release of the anticancer drug zerumbone [18] encapsulated in its pores, showing sustained and prolonged drug release over 24 h [19]. Li and coworkers addressed CFDD via a nano-in-micro dual drug delivery platform composed of halloysite nanotubes and a pH-responsive polymer. This clay mineral was loaded with atorvastatin and celecoxib, which were released only at pH 7.4 and effectively inhibited colon cancer cell proliferation [20]. Xu et al. recently presented up-conversion nanoparticles (UCNPs) to trigger cancer immunotherapy in CRC by NIR-induced photodynamic therapy (PDT) to directly destroy tumor cells and to stimulate immune responses by triggering the maturation of dendritic cells and secretion of cytokines [21]. Others have used gold- [22] and folate-coated nanoparticles for active CFDD of nanotherapeutics [23]. Finally, different authors have used reactive oxygen species (ROS)-responsive nanoplatforms and polymers as drug delivery systems with excellent results [6].
We have recently begun to develop CFDD materials based on environmentally sustainable, abundant and inexpensive diatom microalgae (DEMs) [24] with the aim of designing an innovative structure able to simultaneously target the tumor site and release loaded chemotherapeutics in its immediate proximity [25]. DEMs, as the fossil frustules of diatoms, recently gained attention for their use in drug delivery due to their biocompatibility and ability to shuttle and slowly release different drugs [26][27][28][29]. Diatoms are enclosed in a three-dimensional, highly ordered silica shell (called frustule) and represent an inexpensive and well-engineered source of microporous silica [30][31][32][33]. Our original proof of concept relied on the chemical functionalization of the diatom microalgae's surface with vitamin B12, which allowed specific binding of the material to CRC cells and, therefore, discrimination between healthy and diseased tissues. The next step in our concept is illustrated in Figure 1. This bio-inspired hybrid multifunctional drug delivery system is modified by the introduction of photoactivatable units (orange spheres in Figure 1), which are envisioned to act in concert with the loaded chemotherapeutic drugs (light blue squares) in order to (a) sensitize the tumor and (b) lower the overall drug dose needed for effective treatment. Concept showing the bio-inspired hybrid multifunctional drug delivery system described in this study. Natural diatom microalgae (DEMs) are functionalized at their surface with photoactivatable elements (orange spheres) linked to vitamin B12 (red spheres) acting as the outer tumor-targeting layer. The construct can be loaded with a chemotherapeutic drug (light blue squares) to be selectively delivered to the tumor site in the intestinal tract. Once DEMs are bound to the tumor mass, the chemotherapeutic drug is slowly released. In addition to that, the micro-capsules can be photoactivated to generate a toxic CO or 1 O2, sensitizing the tumor to the treatment. Right. Photoinduced chemical reactions at the surface of the diatom microalgae (vide infra for more details).
For this purpose, two complex bio-vehicles were synthetized. In both cases, the outer layer of the microalgae was still composed of cobalamin (vitamin B12, red spheres in Figure 1). The inner Concept showing the bio-inspired hybrid multifunctional drug delivery system described in this study. Natural diatom microalgae (DEMs) are functionalized at their surface with photoactivatable elements (orange spheres) linked to vitamin B 12 (red spheres) acting as the outer tumor-targeting layer. The construct can be loaded with a chemotherapeutic drug (light blue squares) to be selectively delivered to the tumor site in the intestinal tract. Once DEMs are bound to the tumor mass, the chemotherapeutic drug is slowly released. In addition to that, the micro-capsules can be photoactivated to generate a toxic CO or 1 O 2 , sensitizing the tumor to the treatment. Right. Photo-induced chemical reactions at the surface of the diatom microalgae (vide infra for more details).
For this purpose, two complex bio-vehicles were synthetized. In both cases, the outer layer of the microalgae was still composed of cobalamin (vitamin B 12 , red spheres in Figure 1). The inner layers of the constructs were built with either a chemical photosensitizer (for photodynamic therapy, PDT) or a photo-triggered CO-releasing molecule (see right inserts in Figure 1). Photodynamic therapy is a minimally invasive treatment incorporating three different components: a light source, a chemical photosensitizer (PS) and tissue oxygen. This process aims to generate endogenously highly toxic singlet oxygen ( 1 O 2 ) and reactive oxygen species (ROS) upon light irradiation [34]. The produced excited-state singlet oxygen species induce cell apoptosis in tumors [35,36]. The advantages of this technique are the spatio-temporal controlled activity allowed by the photoactivation and the potential to be combined with other therapeutic treatments as chemotherapy, surgery, radiotherapy or immunotherapy [37]. As chemical photosensitizer for our hybrid multifunctional drug delivery system, we chose porphyrins, as they absorb at several wavelengths in the visible region (Soret band in the blue and Q-bands in the red) and show long-lived triplet states allowing high 1 O 2 quantum yields [38][39][40][41]. Furthermore, coordination of metal ions such as Zn 2+ , Ga 3+ or Si 4+ allows tuning 1 O 2 generation capacities of porphyrins, giving long excited-state lifetimes and permitting high singlet oxygen yields [42]. As a photo-triggered CO-releasing molecule, we selected a (2,2 -bipyridine)-3,4 -dicarboxylic acid manganese tricarbonyl complex. Such types of complexes (vide infra for detailed chemistry) are well known to decompose if photo-irradiated, liberating in the process carbon monoxide. CO, in turn, is known to exhibit antiproliferative and cytotoxic effects in cancer treatments [43][44][45][46][47]. In this contribution, we describe the preparation of these new bio-inspired hybrid multifunctional drug delivery systems. In addition to that, we detail CRC cellular interaction of the same, their drug-loading and -releasing properties, and their effects on the proliferation of CRC HCT-116 cells in the dark and when photoactivated.

Chemicals and Materials
Diatomaceous earth microalgae (DEMs), in the form of Celatom ® Fw-14, were obtained from Applied minerals Ltd.

Density Functional Theory (DFT) Calculations
All computations were performed with the Gaussian 09 programs on the DALCO Cluster of the University of Fribourg. The procedure of minenkov et al. [50] was used, and different combinations of functional and basis sets were tried. Geometry optimizations were performed employing a solvent continuum dielectric model for water. The hybrid-GGA (GGA = generalized gradient approximation) functional M06 and B3LYP were used in combination with the standard LanL2DZ basis sets for the optimization of B 12 -TCPP and B 12 -Mn, respectively. For the spin state of the molecules (singlet state in all cases), the default spin formalism was followed in the calculations, and default Gaussian 09 values were adopted for the numerical integration grids, self-consistent-field (SCF) and geometry optimization convergence criteria. Geometries were optimized without symmetry restrictions. The nature of the stationary points was checked by computing vibrational frequencies in order to verify true minima. No negative frequencies were observed for the reported values. The calculated molecular geometries were visualized using GaussView. TCPP was synthetized via the monopyrrole tetramerization method originally described by Adler et al. and slightly modified [51,52]. First, 1.5 g of 4-carboxybenzaldehyde (10 mmol) was added to 50 mL acetic acid, stirred and heated to 80 • C for better dissolution of the aldehyde. Afterward, distilled pyrrole (0.7 mL, 10 mmol) was added, and the reaction mixture was brought to reflux and stirred for 2 h. After the reaction mixture was allowed to cool to room temperature, the reaction flask was placed in the fridge to induce precipitation of the porphyrin. Vacuum filtration of the reaction mixture, washing the residue with DCM (5 times 50 mL) and recrystallization permitted the isolation of a dark purple solid (1.1 g, 1.4 mmol, 55% yield), which was dried under vacuum in the oven. 1 Figure S1), which was consistent with previous studies [53]; HR-ESI-MS (ESI+) (m/z): Cyanocobalamin was modified by pegylation on the 5 -hydroxylic function of the ribose moiety by slight modification of published procedures [54,55]. Briefly, 100 mg of reduced B 12 (0.0739 mmol) was mixed with 75 mg of CDT in 3 mL of DMSO and stirred overnight. The product was then precipitated in ethyl acetate (150 mL) and centrifuged for 10 min at 6000 rpm in order to recover a red precipitate. The dried precipitate was resolubilized in 1 mL anhydrous DMF, and 100 µL of 4,7,10-trioxa-1,13-tridecanediamine in 1 mL anhydrous DMF was added and stirred for 24 h. The mixture was precipitated in 3:1 ether:ethyl acetate (150 mL) and centrifuged for 10 min at 6000 rpm before being purified by preparative HPLC. The fraction corresponding to the pegylated B 12 (B 12 -1) was evaporated under vacuum and further lyophilisated to recover a pure dried red powder.

B 12 -ZnTCPP
To a solution of B 12 -TCPP (3 mg, 1.26 × 10 −6 mol) in 3 mL MeOH, 0.5 mg Zn(OAc) 2 ·2H 2 O were added, and the reaction mixture was stirred under reflux for 1 h. Afterward, the reaction mixture was allowed to cool to room temperature. Vacuum filtration and washing the residue several times with water allowed removing excess zinc and permitted the isolation of a brown-red solid that was dried under vacuum in the oven. UV-Vis (in MeOH, λmax, nm): 360, 423.5, 517, 560, 601.

Vitamin B 12 Derivative B 12 -Mn
Cyanocobalamin was modified by pegylation on the hydroxylic function of the ribose part (5 -OH) of the molecule as previously described [25,57]. The pegylated B 12 was then reacted with the manganese complex Mn to give the B 12 -Mn. For this purpose, 1.6 mg (0.0035 mmol, 1.5 equivalent) of Mn were activated in MES buffer pH 5.5 with 46 µL EDC during 30 min in the dark before 17 mg of NHS were added for another 30 min and finally 3.7 mg (0.0023 mmol, 1.0 equivalent) of pegylated B 12 was poured dropwise in 1 mL MES buffer pH 5.5. For the next 5 h, 15 mg of EDC were added each hour. Two hours after the last EDC addition, 50 µL of TEAwas added to the mixture that was allowed to react overnight in the dark. The crude was finally purified by preparative HPLC. Each manipulation was done in diminished light due to the photo-sensitivity of Mn. Yield 3.7 mg (79%). 1 Figure S10).

DEMs Functionalizations
The isolation, purification and surface activation of DEMs were done as previously described by Delasoie et al. [25]. The aminated DEMs surface was generated by APTES condensation.
Afterward, the functionalized DEMs were separated by centrifugation and washed several times with MeOH and water, yielding a slightly reddish powder. The material was characterized by UV solid-state ( Figure 2) and zeta potential.

Mn-DEMs
In order to functionalize the aminated DEMs with Mn, 0.5 mg of Mn were dissolved in 1 mL DMF with 1 mg of EDC under diminished light and stirred for 30 min. Then, 1 mg of NHS was added and let react for the next 30 min. After direct addition of ca. 10 mg of aminated DEMs and 20 µL TEA, the mixture was reacted overnight at RT in the dark. The mixture was then centrifuged Pharmaceutics 2020, 12, 480 8 of 23 and the supernatant discarded, and this was repeated with 10 mL DMF and 10 mL methanol thrice. The sample was finally dried under vacuum. The recovered powder was denoted as Mn-DEMs.

B 12 -Mn-DEMs
To achieve the functionalization of aminated DEMs with B 12 -Mn, typically 0.8 mg of B 12 -Mn was added to 0.15 mg of EDC chloride and 0.10 mg of NHS in 1 mL DMF for 1 h diminished light. Then, 2 mg of DEMs in 0.5 mL DMF was added to the mixture. Twenty microliters of TEA were added directly after. The mixture was reacted at RT in the dark overnight. The mixture was then centrifuged and the supernatant discarded. The recovered powder was washed with 20 mL DMF and with 20 mL methanol thrice and finally dried under vacuum overnight. The recovered powder was denoted as B 12 -Mn-DEMs.

Photodynamic Measurements
where K R and K S are the slopes of the kinetic plot of the difference in absorbance vs. irradiation time of DMA photo-oxidized by the reference and sample, respectively; I R and I S the total light intensities absorbed by the reference and sample and Φ S ∆ the singlet oxygen quantum yield of the sample. Light intensities were calculated according to the following equation: where I is the absorbed light intensity from the sample, I 0 the light intensity from the light source and A (λ = 420 nm) the absorbance of the photosensitizer at the excitation wavelength.

Kinetic and Equivalent CO Released by Photoirradiation
For both Mn and B 12 -Mn complexes, the kinetic of CO release was evaluated by monitoring the spectral changes at a specific wavelength (410 and 365 nm respectively) by UV/Vis spectroscopy. For this purpose, the complexes were firstly dissolved in DMSO (1% v/v, final concentration) before being diluted in 0.1 M PBS. Samples were irradiated with a 420 nm light. Then, the equivalent of CO release was evaluated under the conditions of myoglobin assay for all complexes. The solutions Mn (15 µM) and B 12 -Mn (10 µM), measured in 0.1 M PBS buffer at pH 7.4, 10 mM dithionite, with 3 equivalent of myoglobin (respectively 45 and 30 µM), under argon, were irradiated with a 420 nm light. The stability of each compound in the presence of dithionite was evaluated prior to each experiment, and none of them showed spontaneous CO release. The concentration of Myoglobin-CO complex c(Mb-CO) formed through time due to CO release from the photoCORMs upon irradiation at 420 nm was monitored, and the CO release equivalent was calculated as described by Atkin et al. [59]. Due to its poor solubility, Mn was previously dissolved in DMSO (1% v/v, final concentration).

Resistance of B 12 -X-DEMs in Simulated Digestive Fluids
The test was performed in simulated gastric fluid (SGF). For this purpose, ca. 4 mg of each B 12 -X-DEMs sample was digested in SGF (6 mL) for 2 h before the solid residue and the supernatant were recovered. The supernatant was diluted to 10 mL with HNO 3 2% (v/v). The solid fraction was digested for a second time in 2 mL piranha solution (H 2 SO 4 :H 2 O 2 7:1) overnight before being diluted to 10 mL with deionized water. All the samples were analyzed by ICP-OES (Co monitoring). The result was calculated as a mass ratio of the B 12 equivalent compared to the total amount of sample.

Drug Loading and Release from DEMs
Around 2 mg of the rhenium complexes used as model drugs (1 or 2, vide infra) was weighed and dissolved in 20 µL of DMSO, and then 10 µL of this solution of 100 mg/mL was diluted (100 fold) to 1000 µL with aqueous buffer PBS pH 7.4 (1% v/v, DMSO), giving the first standard solution (S1, 1.0 mg mL −1 ). Dilutions of S1 in PBS pH 7.4 (1% v/v, DMSO) gave the standards used for the calibration curves (5 standards, R 2 > 0.9). The standards and samples were analyzed by UV-Vis spectroscopy. For each measurement, the sample was quickly vortexed before being transferred for UV-Vis measurement. The drug loadings (DL) were performed by weighing around 10 mg of the drug and the same amount of DEMs in an Eppendorf. Afterward, 1 mL of acetone was added to obtain a 10 mg mL −1 drug solution. The samples were shaken on the plate for 48 h. The mixtures were then centrifuged and the supernatant removed. A quick wash was performed by adding 500 µL of acetone, centrifuging 5 s and removing supernatant. The samples were then evaporated under argon to dryness. Drug release experiments were performed by adding 1000 µL of PBS pH 7.4 (1% DMSO) to the drug-loaded DEMs samples. Samples of 100 µL were taken at each time point and replaced with fresh buffer. The concentrations of these samples were evaluated by UV-VIS spectroscopy by monitoring the absorption of the complexes at specific wavelengths. The values for the cumulative drug release in percentage were calculated from the linear regressions previously established. The graphs were plotted using Origin 7.5. The DL were calculated as the mass percentage (wt%) of drug over the sum of drug and DEMs.

Cytotoxicity Assessment
Rhenium anticancer complexes 1 and 2 were freshly dissolved first in DMSO and then in RPMI 1640 medium (Merck, Munich, Germany), while the other analyzed compounds were immediately dissolved in RPMI 1640 medium and used for bioactivity assessments. In vitro cytotoxicity in terms of antiproliferative effects was determined by (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay [60] on colorectal cancer HCT116 cell line, obtained from American Type Culture Collection (ATCC). The cells, cultured in the complete RPMI 1640 medium as a monolayer (1 × 10 4 cells per well), were incubated with test compounds for 48 h in humidified atmosphere of 95% air and 5% CO 2 at 37 • C, and the MTT assay was carried out two times in four replicates. The extent of MTT reduction was measured spectrometrically at 540.0 nm using Tecan Infinite 200 Pro multiplate reader (Tecan Group Ltd., Männedorf, Switzerland), and the cell survival was expressed as a percentage of the control arbitrarily set to 100%. Cytotoxicity was expressed as the concentration of the compound inhibiting cell growth by 50% (IC 50 ) in comparison to untreated control. For the photoactivation experiment, Osram L30W/840 cool white light Lumilux (Munich, Germany) has been used two times for 10 min, immediately upon the addition of compounds to the cells and 1 h after.

Synthesis and Characterization of Surface Coating molecules
Vitamin B 12 (tumor-targeting agent) surface-coated molecules bearing either a (2,2 -bipyridine)-3,4 -dicarboxylic acid manganese tricarbonyl complex (photoCORM, B 12 -Mn) or a zinc of 5, 10, 15, 20-tetrakis(4-carboxyphenyl)porphyrin molecule (PDT agent, B 12 -ZnTCPP) were prepared according to the synthetic procedure illustrated in Scheme 1. Vitamin B 12 (represented by a simplified structure) was first modified with an aminated PEG-chain on the 5 -OH ribose moiety and then reacted directly with the manganese complex (Mn, giving B 12 -Mn, path A in Scheme 1) or with 5,10,15,20-tetrakis(4-carboxyphenyl)porphyrin (TCPP) respectively (path B in Scheme 1). The latter was finally metalated with zinc(II) acetate, giving B 12 -ZnTCPP. Alternatively, B 12 -ZnTCPP could be prepared by PEGylation of TCPP followed by metalation and coupling to vitamin B 12 (path C in Scheme 1). We noted that the length of the selected PEG-chain allows vitamin B 12 derivatives to be still recognized by the transcobalamin receptors (TC(II) and TC(II)-R), leading to increased adherence towards cancer cells [25,61,62]. Density functional theory (DFT) calculations of B 12 -Mn and B 12 -TCPP performed in water (polarizable continuum model, vide infra, Figure 3) indicate a distance of ca. 18 Å between the molecular units. Thus, the PEG chain does not only act as a linker but also provides a certain degree of freedom to the system. Indeed, we expected this distance to minimize steric hindrance in the final composite materials, allowing the different components to play their role with minimal interferences [63]. Once grafted onto the surface of diatom microalgae, this arrangement facilitates the role of B 12 as the tumor-targeting agent [25]. At the same time, photoactivation of the surface allows for CO release from B 12 -Mn or 1 O 2 generation from B 12 -ZnTCPP respectively. Both surface-coating derivatives were characterized by standard techniques. 1 H-NMR, visible spectroscopy and MS data are fully consistent with the proposed molecular structures of B 12 -Mn and B 12 -ZnTCPP. Particularly revealing were the 1 H-NMR spectra of the molecules (ESI Figures S1-S4). As expected, the homotopicity of the aromatic protons of the (2,2 -bipyridine)-3,4 -dicarboxylic acid and TCPP of B 12 -Mn and B 12 -ZnTCPP, respectively, was lost upon coupling to B 12 -1 (i.e., PEGylated vitamin B 12 , see Scheme 1). In the case of B 12 -ZnTCPP, e.g., the symmetric 1:1 doublets of TCPP split into three doubles of relative intensity of 4:3:1; the latter being assigned to the 4-carboxyphenyl protons in ortho-position to the newly formed amide bond (ESI Figure S1). Metalation of B 12 -TCPP was confirmed by UV-Vis spectroscopy. Indeed, the Soret band of the porphyrin undergoes a bathochromic shift from 415 to 423 nm upon reaction with Zn(OAc) 2 ·2·H 2 O (Figure 3). Moreover, upon metalation, the degeneracy in the porphyrin orbital causes the coalescence of the four TCPP Q-bands (512, 546, 588 and 645 nm) into two Q-bands in B 12 -ZnTCPP (560 and 601 nm), as previously described [64,65]. The peaks at 360 nm and 517 nm were respectively attributed to the α-band and β-band of vitamin B 12 ( Figure 3).

1 O 2 Generation and CO Release of Surface Coating molecules
Fluorescence quantum yield (Φ f l ), lifetime (τ f ) and singlet oxygen ( 1 O 2 ) quantum yield (Φ ∆ ) were assessed for PEG-ZnTCPP and B 12 -ZnTCPP in DMSO (Table 1). Fluorescence spectra of PEG-ZnTCPP and B 12 -ZnTCPP in DMSO show emission bands around 612 and 666 nm (ESI Figure S7), and the fluorescence quantum yields and lifetimes are similar for the two species. The results are consistent with values reported in the literature for comparable porphyrin analogs and support the hypothesis that the structural change at the periphery of the porphyrin does not drastically affect Φ f l and τ f since the fluorescence phenomenon arises from the inner π-electron system on the macrocycle [65][66][67]. Nevertheless, it is known that introducing electron-donating or -withdrawing groups onto porphyrin center will slightly influence the fluorescence properties [68]. Longer fluorescence lifetime and improved inter-system crossing positively influence the singlet oxygen quantum yield, but the same is affected by many other parameters such as solubility or the tendency of the fluorophore to form aggregates. However, Φ ∆ does not correlate to the fluorescence lifetime and quantum yield improvement. Porphyrins can easily aggregate in solution, thereby lowering their efficiency in energy transfer, with 3 O 2 resulting in lower 1 O 2 Φ ∆ [69,70]. The solubility and aggregation propensity of the zinc(II) porphyrin are substantially modified by introducing a hydrophilic PEG chain or a large and highly water-soluble structure like vitamin B 12 . The coupling of vitamin B 12 to ZnTCPP slightly increases the singlet oxygen quantum yield by 15%. Indeed, vitamin B 12 covalently linked to the porphyrin enhances the solubility of the structure and lowers its tendency to aggregate, thereby leading to an efficient generation of 1 O 2 singlet oxygen. B 12 -ZnTCPP is stable to light. We observed no decomposition of the molecule under irradiation, suggesting the great potential of B 12 -ZnTCPP in PDT. Pseudo-half-life and equivalents of CO released were measured for B 12 -Mn and its constituent photoCORM complex (Mn) [71] according to the method described by Atkin et al. (Table 1) [59]. To this end, changes in the UV-Vis spectrum at 410 and 365 nm were monitored as a function of irradiation time for Mn and B 12 -Mn, respectively (ESI Figures S18-S21). The stabilities of both complexes were demonstrated by the absence of spontaneous CO releases in the dark in the presence of sodium dithionite. Both molecules, upon light irradiation, released ca. one equivalent of CO (ESI Figures S16 and S17). Table 1. Fluorescence quantum yield (Φ f l ), lifetime (τf) and singlet oxygen quantum yield (Φ ∆ ) of the porphyrin complexes in DMSO. Photo-induced CO-release kinetics of the manganese complexes in PBS buffer pH 7.4.

Surface Functionalization, Characterization and DEMs Cellular Interaction
Surface functionalization of DEMs with B 12 -ZnTCPP and B 12 -Mn was achieved via amide-bond formation with one of the free carboxylic functions available on the molecules and the previously aminated surface of the DEMs (Figure 2). In the case of B 12 -TCPP and B 12 -ZnTCPP, the presence of free carboxylic functions on the porphyrin macrocycle in both cisand trans-positions to the PEG arm offers the possibility of several binding modes to DEMs. Given the fact that B 12 -ZnTCPP-DEMs interact strongly with cancer cells (Figure 2) we posit either that the trans-4-carboxyphenyl is preferentially favored or that the system is flexible enough so that steric hindrance (i.e., in the case of cis-4-carboxyphenyl binding) does not play a significant role in B 12 -ZnTCPP-DEMs cell interaction.
Successful surface functionalization of diatoms microalgae was assessed via solid-state UV-Vis spectroscopy ( Figure 2D,E). For B 12 -TCPP-DEMs, comparison of the spectra of the new material to the non-functionalized DEMs, shows new absorbance peaks at 362 nm (Soret band of vitamin B 12 ), 423 nm (Soret band TCPP), 518, 552, 591, and 648 nm (Q-bands of TCPP). By further reacting B 12 -TCPP-DEMs with zinc acetate in methanol under reflux, metalation of the porphyrin centers occurred, leading to B 12 -ZnTCPP-DEMs. The solid-state UV-Vis spectrum of B 12 -ZnTCPP-DEMs compared to the non-metaled equivalent shows a bathochromic shift of the TCPP Soret band to 432 nm and coalescence of the Q-bands to two peaks at 561 and 601 nm ( Figure 2E). This evidence confirmed the successful metalation of the material. Similarly, solid-state UV-Vis spectroscopy was performed to evaluate the newly synthetized material B 12 -Mn-DEMs. When comparing the DEMs spectra before and after surface functionalization with pure Mn, B 12 and B 12 -Mn, the appearance of new peaks corresponding to each compound could be observed ( Figure 2D). For the Mn-DEMs (synthetized to assess the feasibility of the surface coating), a peak at 431 corresponding to the mLCT absorbance spectrum of the Mn complex is detected. In the case of B 12 -DEMs [25], peaks at 361 nm, 521 nm and 545 nm are consistent with absorbance of pure cyanocobalamin [72]. Finally, the spectrum of B 12 -Mn-DEMs shows peaks corresponding to cobalamin at 363 nm, 515 nm and 545 nm as well as a very small peak at 423 nm attributed to the presence of Mn. The solid-state UV-Vis spectrum of B 12 -Mn-DEMs is consistent with the spectrum of B 12 -Mn in solution ( Figure S13), which shows a signal at 420 nm, very low in intensity when compared to the peaks attributed to the vitamin B 12 at 361, 521 and 545 nm.
Further direct characterization of successful DEMs surface functionalization came from zeta potential (ζ-potential) analysis of B 12 -ZnTCPP-DEMs. Zeta potential of different samples of DEMs were recorded in PBS buffer at pH 7.4. Clean hydroxylated diatoms (DEMs-OH) expressed a negative potential of −23.4 ± 0.5 mV, which shifted to positive +9.5 ± 0.4 mV upon surface amination with APTES. This positive shift is indicative of amine functions on DEMs, which can be easily protonated in the buffer [25,[73][74][75]. Further functionalization with TCPP moves the ζ-potential back towards the negative value of −53.0 ± 0.4 mV, logically deriving from the presence of carboxylic acid groups. The same potential slightly increases to −48.8 ± 0.3 mV for B 12 -TCPP-DEMs and remained virtually the same following Zn metalation for B 12 -ZnTCPP-DEMs (−47.1 ± 0.2 mV).
The surface functionalization of B 12 -Mn-DEMs and B 12 -TCPP-DEMs, as well as the resistance of the surface coating in simulated gastric fluid, were assessed by inductively coupled plasma atomic emission spectroscopy (ICP-OES). Naturally occurring diatoms microalgae may contain small traces of cobalt in their frustules composition but considerable traces of manganese [76]. For example, in Cyclotella meneghiniana and Stephanodiscus hantzschii, common freshwater diatom species, zinc and manganese compete for cell wall incorporation with Mn accumulating preferentially at the girdle band of frustules [77,78]. In contrast, cobalt can only replace zinc when this ion is totally absent from the growth medium [79]. ICP-OES measurements of the cobalt content in clean DEMs indicated values below the limit of quantification (4 ppm). The difficulty of generating Co ions from cobalamin in the plasma has been reported before [80]; nevertheless, the method we used in the present study was already proven efficient to determine about 95% of the B 12 present in a sample [25]. The B 12 -equivalent content was extrapolated from Co concentration of each sample after 2 h digestion in SGF. B 12 -Mn-DEMs showed concentrations of 450 ± 19 ppm and 131 ± 10 ppm of B 12 for supernatant and solid residue, respectively. B 12 -TCPP-DEMs showed concentrations of 224 ± 28 ppm and 92 ± 21 ppm of B 12 for supernatant and solid residue, respectively. Results mean that ca. 29% and ca. 41% of the surface coating remain intact after 2 h in the stomach for B 12 -Mn-DEMs and B 12 -TCPP-DEMs, respectively. We plan to coat with gel to increase performance.
Scanning electron microscopy was used to assess the adherence of B 12 -ZnTCPP and B 12 -Mn on colorectal cancer HCT-116 cells. In order to demonstrate cellular adherence, cells were treated with either unmodified (DEMs) and B 12 -functionalized diatoms under identical conditions (200 µg biomaterial mL −1 , 1 h exposure). After intensive washes, cells exposed to B 12 -functionalized diatoms still presented and retained a large amount of micro-particles ( Figure 2I,L), while in the sample of unmodified DEMs, almost all the micro-particles were removed via the washing steps.

In vitro Drug Loading and Release from Diatom Microalgae
We have recently described the anti-proliferative efficacy of a series of rhenium(I) tricarbonyl N-derivatized N-((2,2 -bipyridin)-6-ylmethyl) complexes of increasing lipophilicity against different cell lines [48]. Among the series of these compounds, the fac-(Re(CO) 3 Br) complexes with 6-(bromomethyl)-2,2 -bipyridine (1 in this study, Figure 4) and N-((2,2 -bipyridin)-6-ylmethyl)-N-isobutyl-2-methylpropan-1-amine (2, Figure 4) showed excellent in vivo toxicity profiles and in vivo anticancer/antimetastatic efficacy and effective inhibition of angiogenesis in the zebrafish-human CRC tumor xenograft model [48]. These two complexes were selected for the present study. The molecules were successfully loaded into DEMs according to the vacuum infiltration method described by Vasani et al. [81] (giving 1@B 12 -ZnTCPP-DEMs and 2@B 12 -Mn-DEMs, respectively) with respective loading degrees of 3.7 (1) and 1.5 (2) wt% ( Table 2). These are in the typical range of drugs loaded into DEMs biosilicas [25,26,[82][83][84]. The drug-material composite selection was based on the most favorable drug-release characteristics for each combination. Following successful encapsulation, the drug release in PBS buffer pH 7.4 (1% v/v, DMSO) was initiated by the addition of solvent and monitored by UV-VIS spectroscopy over a period of 5 days in both cases. As often observed in different studies, an important initial burst release occurs during the first hours ( Figure 4) [27,29]. Ca. 60% and 50% of complexes 1 and 2 were respectively released in solution after the first hour. A second more gradual and sustained release was then observed over the course of a few days. Indeed, 1 and 2 were nearly completely discharged from B 12 -ZnTCPP-DEMs and B 12 -Mn-DEMs after 2 and 5 days, respectively.

PDT and Cytotoxicity Assessment
The cytotoxicity of the 1@B 12 -ZnTCPP-DEMs and 2@B 12 -Mn-DEMs formulations in dark and under light irradiation were assessed using the human colorectal HCT-116 cell line ( Figures 5 and 6). In the case of 1@B 12 -ZnTCPP-DEMs, light activation promotes singlet oxygen generation by the ZnTCPP component. Photoactivation of 1@B 12 -ZnTCPP-DEMs increased the cytotoxicity of the construct up to 2.4-fold for the most concentrated sample (λν 200 µg/mL (+IC 50 ), Figure 5 and Table 2) when compared to the effects of the administration of 1 alone. Notably, 1@B 12 -ZnTCPP-DEMs containing only 25% of the IC 50 concentration value of 1, caused a 45% increase in cell death upon light exposure in comparison to dark conditions, essentially reaching the cell-killing effects of the IC 50 value of 1 at 1 4 the complex concentration. 1@B 12 -ZnTCPP-DEMs with 2.5% the IC 50 of 1 caused an increase of 14% of cell death under tested similar conditions. Overall, as it can be appreciated from the graphs in Figure 5 that the data suggest that lower amounts of cytotoxic compounds are needed in combination with the photoactivatable surface-functionalized diatom microalgae for an equally effective drug formulation or cytotoxic effect. Unmodified natural DEMs, either in the dark or exposed to light, showed no toxicity at concentrations >100 µg/mL, while both DEMs containing adsorbed 1 and B 12 -TCPP-DEMs particles alone showed slightly higher activity upon the light exposure ( Figure 5). The photoactivity of complex 1 was somewhat surprising. We posit that it may be due to activation of the pending α-diimine ligand-CH 2 -Br arm towards nucleophilic substitution with bio-available amines via reactions of the complexes previously [48].
For 2@B 12 -Mn-DEMs, light activation induces CO release from the photoCORM Mn component at the surface of the microalgae. CO released from this type of molecule is known to promote pro-apoptotic effects in different cancer cell lines [85][86][87][88][89][90] via CO-mediated attenuation of glutathione and nuclear metallothionein levels [91] and inhibition of cystathionine β-synthase [92]. When HCT-116 cells were exposed to 2@B 12 -Mn-DEMs, photoactivation of the material similarly increased the overall cytotoxicity effect of complex 2. Light irradiation of 2@B 12 -Mn-DEMs increased the cytotoxicity of the construct by ca. 2.0-fold in the case of the most concentrated sample (λν 200 µg/mL (+IC 50 ), Figure 6 and Table 2) when compared to the same sample in the dark or to the effects of the administration of 2 alone. In particular, 2@B 12 -Mn-DEMs containing 25% of the IC 50 concentration value of 2 caused a 31% increase in cell death upon light exposure in comparison to dark conditions. Photoactivation of 2@B 12 -Mn-DEMs containing 1 2 the IC 50 concentration value of 2 was more effective than administration of 2 alone at its IC 50 concentration or B 12 -Mn-DEMs, inducing ca. 64% cell death. Photoactivation of 2@B 12 -Mn-DEMs containing only 2.5% of the same IC 50 was not effective but still induced a 13% increase of cell death compared to the same concentrations of 2 or B 12 -Mn-DEMs. As concluded for 1@B 12 -ZnTCPP-DEMs, the data for 2@B 12 -Mn-DEMs also suggest a potential of this poly-systemic approach in cancer treatment.
Pharmaceutics 2020, 12, x FOR PEER REVIEW 15 of 22 exposure ( Figure 5). The photoactivity of complex 1 was somewhat surprising. We posit that it may be due to activation of the pending α-diimine ligand-CH2-Br arm towards nucleophilic substitution with bio-available amines via reactions of the complexes previously [48].  For 2@B12-Mn-DEMs, light activation induces CO release from the photoCORM Mn component at the surface of the microalgae. CO released from this type of molecule is known to promote proapoptotic effects in different cancer cell lines [85][86][87][88][89][90] via CO-mediated attenuation of glutathione and nuclear metallothionein levels [91] and inhibition of cystathionine β-synthase [92]. When HCT-116  Figure 5). The photoactivity of complex 1 was somewhat surprising. We posit that it may be due to activation of the pending α-diimine ligand-CH2-Br arm towards nucleophilic substitution with bio-available amines via reactions of the complexes previously [48].  For 2@B12-Mn-DEMs, light activation induces CO release from the photoCORM Mn component at the surface of the microalgae. CO released from this type of molecule is known to promote proapoptotic effects in different cancer cell lines [85][86][87][88][89][90] via CO-mediated attenuation of glutathione and nuclear metallothionein levels [91] and inhibition of cystathionine β-synthase [92]. When HCT-116 DEMs, B12-ZnTCPP-DEMs, B12-Mn-DEMs and their respective combinations with rhenium anticancer complexes 1 and 2 (Figure 7) clearly demonstrate the improved efficiency of this approach (data not shown for raw DEMs alone, which are non-toxic). Data confirm the additive (and possible synergistic) effect of drugs 1 and 2 together with the 1 O2-generating porphyrin and the CO-releasing photoCORM-functionalized microalgae surface, respectively.

Conclusions
In this study, we have reported the synthesis and characterization of two bio-inspired hybrid multifunctional drug delivery systems based on diatom microalgae. Both photoactivatable materials, either bearing a tetracarboxyphenyl porphyrin PS or tricarbonyl bipyridyl Mn(I) photo-CORM, are able to bind to HCT-116 colorectal cancer cells via the outer cobalamin layer and slowly release highly active rhenium(I) tricarbonyl anticancer complexes. The ability of the B12-X-DEMs micro-particles to retain a high level of surface functionalization in simulated gastric fluids was also demonstrated. Furthermore, the potentiation effect of the chemotoxic drug by the different light-activated molecules at the surface of diatom microalgae was demonstrated by MTT assay under dark/light conditions on the HCT-116 cell line. At least a 2-fold increase of cytotoxicity toward HCT-116 was observed upon light irradiaton for both constructs (at 200 μg mL −1 , incubated with drug concentration corresponding to respective IC50 values) with cell survival falling from ca. 40% to ca. 20%. Our results are in line with what was reported earlier in the literature for different constructs [93][94][95][96][97], indicating that synergistic chemo-photodynamic therapy mediated by diatom microalgae is a strategy worth pursuing in colonfocused drug delivery. It is well known that undesirable side effects of most chemotherapeutics like cisplatin are dose-dependent [98]; thus, this approach can reduce the need for chemotherapeutics concentration for equal effectiveness, hence leading to diminished undesirable side effects for the patient. Future efforts for the optimization of this hybrid multifunctional drug delivery system will be directed towards coating formulations (e.g., pH-dependent polymers such as cellulose acetate

Conclusions
In this study, we have reported the synthesis and characterization of two bio-inspired hybrid multifunctional drug delivery systems based on diatom microalgae. Both photoactivatable materials, either bearing a tetracarboxyphenyl porphyrin PS or tricarbonyl bipyridyl Mn(I) photo-CORM, are able to bind to HCT-116 colorectal cancer cells via the outer cobalamin layer and slowly release highly active rhenium(I) tricarbonyl anticancer complexes. The ability of the B 12 -X-DEMs micro-particles to retain a high level of surface functionalization in simulated gastric fluids was also demonstrated. Furthermore, the potentiation effect of the chemotoxic drug by the different light-activated molecules at the surface of diatom microalgae was demonstrated by MTT assay under dark/light conditions on the HCT-116 cell line. At least a 2-fold increase of cytotoxicity toward HCT-116 was observed upon light irradiaton for both constructs (at 200 µg mL −1 , incubated with drug concentration corresponding to respective IC 50 values) with cell survival falling from ca. 40% to ca. 20%. Our results are in line with what was reported earlier in the literature for different constructs [93][94][95][96][97], indicating that synergistic chemo-photodynamic therapy mediated by diatom microalgae is a strategy worth pursuing in colon-focused drug delivery. It is well known that undesirable side effects of most chemotherapeutics like cisplatin are dose-dependent [98]; thus, this approach can reduce the need for chemotherapeutics concentration for equal effectiveness, hence leading to diminished undesirable side effects for the patient. Future efforts for the optimization of this hybrid multifunctional drug delivery system will be directed towards coating formulations (e.g., pH-dependent polymers such as cellulose acetate phthalates, hydroxypropyl methyl-cellulose phthalate or biodegradable liposomes) that can respond to dynamic pH changes of the gastro-intestinal (GI) tract in order to expose the active material in the colon only. This approach should further increase surface coating retention after the passage through the stomach in future formulations.