Nanoharvesting of bioactive materials from living plant cultures using engineered silica nanoparticles

Abstract

Plant secondary metabolites are valuable therapeutics not readily synthesized by traditional chemistry techniques. Although their enrichment in plant cell cultures is possible following advances in biotechnology, conventional methods of recovery are destructive to the tissues. Nanoharvesting, in which nanoparticles are designed to bind and carry biomolecules out of living cells, offers continuous production of metabolites from plant cultures. Here, nanoharvesting of polyphenolic flavonoids, model plant-derived therapeutics, enriched in Solidago nemoralis hairy root cultures, is performed using engineered mesoporous silica nanoparticles (MSNPs, 165nm diameter and 950 m²/g surface area) functionalized with both titanium dioxide (TiO₂, 425mg/g particles) for coordination binding sites, and amines (NH₂, 145 mg/g particles) to promote cellular internalization. Intracellular uptake and localization of the nanoparticles (in Murashige and Skoog media) in hairy roots were confirmed by tagging the particles with rhodamine B isothiocyanate, incubating the particles with hairy roots, and quenching bulk fluorescence using trypan blue. Nanoharvesting of biologically active flavonoids was demonstrated by observing increased antiradical activity (using 2,2-diphenyl-1-picrylhydrazyl radical scavenging assay) by nanoparticles after exposure to hairy roots (indicating general antioxidant activity), and by the displacement of the radio-ligand [³H]-methyllycaconitine from rat hippocampal nicotinic receptors by solutes recovered from nanoharvested particles (indicating pharmacological activity specific to S. nemoralis flavonoids). Post-nanoharvesting growth suggests that the roots are viable after nanoharvesting, and capable of continued flavonoid synthesis. These observations demonstrate the potential for using engineered nanostructured particles to facilitate continuous isolation of a broad range of biomolecules from living and functioning plant cultures.

1. Introduction

Plant cells are capable of synthesizing valuable secondary metabolites, in the form of small organic molecules that are potential and proven therapeutics that cannot readily be made by traditional synthetic chemistry [1–3]. Recent progress has been made in increasing the yields of target metabolites in plant cell cultures with genetic or environmental modification. Genetic sequencing and manipulation of biosynthetic pathways provide a vast resource of potential therapeutic agents [4–8]. It is also possible to select mutant plant cells for survival on the basis of their overproduction of bioactive metabolites to generate a population of mutants with a specific pharmacological phenotype [9]. Flavonoids, a class of secondary metabolites known to be active antioxidants with medicinal, therapeutic and pharmacological applications, have recently been enriched in hairy root cultures by genetic transformation [10–15]. However, these potential therapeutics are usually present in low concentrations, making recovery and purification expensive and complicated. The shortage of commercially viable separation technologies is a main bottleneck in the discovery and application of plant metabolites [16–20]. The conventional method of recovery of natural products from plant cells is to harvest the whole tissues and chemically recover the targeted materials from macerated tissues, usually by solvent extraction. Besides the destruction of the expensive genetically modified plant cell cultures, the activity of labile biomolecules can be reduced depending on the solvent used for the extraction process [21]. Thus, new techniques are required to continuously isolate metabolites from living plant cultures without whole tissue harvesting.

Nanoharvesting, in which nanoparticles are used to carry biomolecules out of living plant cells, can provide continuous harvesting from living and functioning source plants. This is the reverse process of the delivery of bioactive materials (drugs, nucleic acids, etc.) to cells [22–24], but is guided by the same principles in that a nanoparticle must gain entry to the cell with minimal toxicity. Kurepa et al. [25] reported the use of nonporous 2.8 ± 1.4 nm TiO₂ nanoparticles to harvest quercetin-derived flavonoids directly from Arabidopsis thaliana plants based on a high degree of chelation of these molecules by TiO₂ [26,27], but this is the only report of nanoharvesting to date.

While they are effective in nanoharvesting, nonporous TiO₂ nanoparticles lack the high surface area desirable for adsorption, and mesoporous TiO₂ nanoparticles are not readily synthesized as a robust, stable platform [28,29]. Additionally, requirements for stabilization of small nanoparticles limit the functional groups that can be utilized, and the recovery of 3 nm TiO₂ particles from solution is difficult due to their small size [30]. To overcome these limitations, TiO₂ can be dispersed on mesoporous silica nanoparticles (MSNPs) using post-synthesis TiO₂ functionalization [31–34]. MSNPs, which include both nanosized and nanostructured silica particles, have high surface area (~1000 m²/g) and tunable pore size, and the silica surface can be easily modified with other functional groups as necessary. Moreover, MSNPs can be synthesized with a magnetic metal oxide core, which provides the opportunity for facile recovery of the particles using an external magnetic field [35].

Although MSNPs are broadly applied in catalysis, chromatography, and biomolecule loading and cellular delivery [36–41], hydroxyl-terminated bare silica has limited, nonspecific affinity for most plant metabolites. Thus, to exploit the surface properties of MSNPs for nanoharvesting, functionalization of silica with a metabolite binding group such as TiO₂ is necessary. Previous efforts to incorporate TiO₂ in MSNPs have primarily focused on the adsorption and/or photo-degradation of organic compounds using large-size silica nanoparticles and non-spherical silica mesotructures [32,33,42–45], which are not suitable for cellular internalization. In contrast, we recently demonstrated facile TiO₂ functionalization of MSNPs (average particle diameter ~170 nm), where TiO₂ was deposited inside the mesopores by hydrolysis of a TiO₂ precursor [46]. TiO₂ functionalized MSNPs exhibited a high capacity for the model flavonoid quercetin, over 100 times greater than functionalized nonporous silica. A solvent-based ligand displacement method was developed for flavonoid recovery after binding, and quercetin was found to retain most of its antiradical activity throughout the adsorption and desorption steps. Hence, these particles are hypothesized to be capable of isolating flavonoids in high quantity if internalized in flavonoid-rich cells during nanoharvesting, and to provide for the recovery of active flavonoids.

In addition to binding and releasing active flavonoids, an ideal nanoharvesting process should also provide for efficient uptake of nanoparticles into plants with minimal toxicity. Internalization of MSNPs by plant cells were primarily investigated for gene/protein delivery using MSNP carriers [39,47–49], where valuable guidance can be found for the design of nanoparticles for internalization. The interaction of the particles with plant cells and the ability to extract metabolites depends on the particle shape, size, surface area, adsorption capacity, degree of aggregation, and bulk pH and ionic strength [50]. Particle size and surface chemistry are two critical properties for nanoparticle cellular uptake. Nanoparticle uptake is also dependent on the plant species [51,52], and the uptake mechanisms are still under investigation. Endocytotic uptake in root cells is believed to occur through a variety of mechanisms sometimes associated with nutrient uptake [53,54]. Nanoparticles smaller than the pores in plant cell walls (5–20 nm) can directly pass into the cell membrane. Larger particles can be internalized through endocytosis or direct penetration [39,48]. Further, surface functionalization of MSNPs with amine (-NH₂) groups is often used to promote cell membrane penetration through charge interactions, to enhance their colloidal stability, and thus - facilitate their internalization by living cells [55,56]. Amine groups not only provide positive charge for lipid cell membrane penetration during uptake [54,57], but also act as binding sites for fluorescent molecules [58], which are used for visualization of cellular uptake via fluorescence imaging.

Nanoharvesting also requires that the internalized nanoparticles be recovered from the living plant cells. Internalized particles can escape from cellular systems using vesicle fusion with cell membranes or through direct membrane penetration (probably due to electrostatic interactions) [59,60]. Moreover, plant cells must remain viable and capable of re-synthesizing the target biomolecule (i.e., flavonoids) after nanoparticle uptake and release to maintain a continuous nanoharvesting system. Prolonged interaction between particles and plants has been reported to create toxicity in some cases [61]. The effect of silica nanoparticle concentration on the cell viability of A. thaliana plant root cells was studied by Slomberg and Schoenfisch, who observed that particles between 14 and 200 nm in diameter are taken up and not phytotoxic at concentrations up to 1 mg/mL [62]. A. thaliana roots were also found to be viable after 7 days of uptake of anatase TiO₂ nanoparticles [25,63]. The toxicity of MSNPs functionalized with TiO₂ and/or amines to hairy root cultures is unknown.

In this study, the application of high surface area engineered MSNPs are reported for the nanoharvesting of polyphenolic flavonoids from living Solidago nemoralis hairy roots. Dual surface functionalization (with TiO₂ and amines) is used to facilitate metabolite binding and cellular uptake. The accessibility and order of mesopores of the MSNPs (designed with 2.8 nm diameter pores for loading and 170 nm diameter to permit uptake) after surface functionalization are studied by scanning electron microscopy (SEM), X-ray diffraction (XRD) and nitrogen adsorption. Quercetin is used as a model flavonoid to evaluate the adsorption capacity and the subsequent recovery of active biomolecules from particle surfaces using ligand displacement (in 20% w/v ethanolic citric acid solution). Cellular uptake and localization of engineered nanoparticles fluorescently labeled with rhodamine B isothiocyanate (RITC) from Murashige and Skoog (MS) media by S. nemoralis hairy root cultures are visualized using fluorescent microscopy, using trypan blue (TPB) to quench extracellular fluorescence [64,65]. Nanoparticles from the hairy root cultures are isolated using centrifugation. Antiradical activity and radio-ligand displacement activity ([³H]-methyllycaconitine) - from rat hippocampal nicotinic receptor membranes (characteristics of S. nemoralis flavonoids [66]) are measured to demonstrate the presence of active surface-bound flavonoids and flavonoids recovered in solution. Hairy roots are re-cultured following nanoharvesting to examine continued cell viability and the ability of the cultures to re-synthesize flavonoids, as measured by antiradical activity of root extracts. All of these studies provide proof of concept that engineered MSNPs represent a promising platform for in situ recovery of small molecule drug candidates from living plant cell tissues, and thus are a viable route to advance the concept of nanoharvesting.

2. Materials and methods

2.1. Chemicals and reagents

Tetraethyl orthosilicate (TEOS, 99%), sucrose (grade I, ≥99.5%), 2-(N-morpholino) ethanesulfonic acid hydrate (MES hydrate, ≥99.5%) and H₂O₂ (35 wt% in water) were purchased from Acros Organics; Pluronic F127 (tri-block copolymer, bio-grade), quercetin (> 95%), trypan blue (TPB, 0.4% in 0.81% NaCl and 0.06% K₂HPO₄), titanium (IV) ethoxide (TEO, technical grade), (3-Aminopropyl)triethoxysilane (APTES, 99%), ethylenediaminetetraacetic acid (EDTA, ≥99%), citric acid (CA, ≥99.5%), nicotine (≥99.5%), sodium azide (NaN₃, 99%), phenylmethanesulfonyl fluoride (PMSF, ≥98.5%), phosphate buffer saline tablets (PBS, pH 7.4) and rhodamine B isothiocyanate (RITC, mixed isomers) from Sigma-Aldrich; cetyltrimethylammonium bromide (CTAB, 99.8%) from MP Biomedicals; [³H]-methyllycaconitine (³H-MLA) from American Radiolabeled Chemicals; 2,2-diphenyl-1-pi-crylhydrazyl (DPPH, 95%) and fluorescamine from Alfa-Aesar; NaOH pellets (≥97%) from EMD Millipore; Acetone (≥99.5%) from BDH analytical; Ti reference solution (1000 ppm in 10% HCl), ultrapure deionized ultra-filtrated (DIUF) water, methanol (HPLC grade, 99.9%), ethanol (200-proof), 36 N H2SO4 (95–98% in water), tris-HCl buffer (molecular biology grade, ≥99%), NaH₂PO₄ (certified ACS grade), 12 N HCl (ACS grade) and 29.3 wt% NH₄OH solution from Fisher Scientific; and Murashige and Skoog (MS) media supplemented by vitamin B5 and antibiotic cefotaxime sodium (~95%) from PlantMedia (BioWorld, Dublin, OH, USA).

2.1.1. Solidago nemoralis hairy roots

Hairy roots were generated from stem explants of seedlings using Agrobacterium rhizogenes mediated genetic transformation reported earlier in detail [15]. After Agrobacterium treatment, explants were transferred onto half-strength MS media in agar plates supplemented with 400 mg/L cefotaxime and 3% sucrose, where hairy roots were generated within 2 to 3 weeks. Roots were than excised and maintained in continuous culture on MS media supplemented with 250 mg/L cefotaxime and 3% sucrose.

2.1.2. Animals

Adult, male Sprague-Dawley rats (body weight approximately 200–225 g) were purchased from Harlan Laboratories (Indianapolis, IN, USA). Handling, care and use of animals were performed according to the National Institute of Health Guide for Care and Use of Laboratory Animals after the approval of all protocols by the Institutional Animal Care and Use Committee (IACUC) at the University of Kentucky.

2.1.3. Engineered silica nanoparticles

Functional mesoporous silica nanoparticles were synthesized using a surfactant templated sol-gel process and post-synthesis grafting with TiO₂ and amine, successively, using techniques previously established in literature [46,67–69]. Some of the particles were tagged with fluorescent RITC for visualization after internalization by the hairy roots [70,71]. The detailed procedures for particle synthesis, functionalization, and fluorescent group attachment are provided in the Supplementary materials (Section A.1 of Appendix A).

2.2. Material characterization

A Hitachi S-4300 Scanning Electron Microscope (SEM) was used to examine the particle morphology. Particles were mounted onto a 15 mm aluminum stub using double sided carbon tape, excess materials were blown off with dry N₂, and the samples were stored in a desiccator for 24 h. Prior to SEM analysis, the particles were coated with conductive Au–Pd alloy using an Emscope SC400 sputtering system. Average and standard deviation of particle diameters were calculated using 20 random particles with ImageJ Software. Surface characterization was performed from nitrogen adsorption conducted at −196 °C using a Micromeritics TriStar 3000. Samples were degassed at 135 °C for 4 h under flowing N₂ gas before analysis. The specific surface area, average pore diameter and pore size distribution were estimated using the Brunauer, Emmett and Teller (BET) isotherm and by the method of Barrett, Joyner and Halenda (BJH), respectively. Micropore volume and external surface area were estimated using the comparative adsorption method described by Jaroniec et al. [72,73]. Fourier transform infrared (FTIR) spectroscopy was conducted by a Thermo Nicolet Nexus 470 spectrometer with a deuterated triglycine sulfate (DTGS) detector. 0.5 g of anhydrous KBr and particles (0.5–1.0wt%) were crushed with a mortar-pestle, and some of this powder was pressed into a pellet for transmission analysis with N₂ purging. XRD was performed using a Bruker-AXS D8 Discover diffractometer with a Cu K_α source (λ = 1.54 Å) at 0.5°/min for 2θ from 1.5 to 6°. XRD samples were prepared in a powder sample holder and tapped flat with a spatula.

2.2.1. TiO₂ quantification

The amount of TiO₂ on the particle surface was determined using an H₂SO₄/H₂O₂ assay [31,74]. 25 mg of TiO₂-functionalized particles were mixed vigorously with 25 mL of 2 M H₂SO₄ for 20 min at 90 °C and then the solution was filtered through a PTFE (0.02 μm) syringe filter. 1 μL of H₂O₂ was added to 1 mL of the resulting filtrate, and after 10 min, the absorbance was measured using a BioTek plate reader (Winooski, VT) at 407 nm in a 96 well plate. The absorbance was calibrated using samples prepared using a Ti reference solution.

2.2.2. Amine quantification

The amount of amine groups on the functionalized particle surface was determined by a previously reported fluorescamine assay after particle dissolution [75,76]. 30 mg of particles were dissolved over an 8 h period in 30 mL of 0.02 M NaOH at room temperature under vigorous stirring. 100 μL of this solution and 1.0 mL of 1.0 mM fluorescamine in acetone were mixed with 2.0 mL of PBS solution at pH 7.4. The emitted fluorescence intensity of this solution was measured at 480 nm after excitation at 366 nm using a Varian Cary Eclipse fluorescent spectrophotometer. The calibration curve was prepared using known amounts of APTES.

2.3. Adsorption and recovery of the flavonoid quercetin

Quercetin adsorption onto particles was measured using the method of Schlipf et al. [74]. For each measurement, 25 mg of particles was pre-wetted under vortex mixing with 1 mL of ethanol for 24 h in a 2 mL tube, centrifuged for 5 min at 17,000×g (AccuSpin Micro 17, Fisher Scientific), and the supernatant was discarded. Then, 1 mL of quercetin solution in ethanol (0.05–10 mg/mL) was added to the pre-wetted particles for vortex mixing (24 h) in the dark. The particles were centrifuged again and 200 μL of supernatant was analyzed using the plate reader. A calibration curve was plotted using known concentrations of quercetin, and was used to find the amount of adsorbed flavonoid on to the particle surface by the solution depletion method. For quercetin recovery, quercetin-loaded particles were re-suspended in 1 mL of recovery solvent (20% w/v ethanolic CA). After 24 h of vortex mixing, the particles were centrifuged and the supernatant was analyzed in a plate reader.

2.4. Nanoparticle uptake in hairy roots

Hairy roots cultures which overproduce flavonoids, obtained from stem explants of seedlings using A. rhizogenes induced genetic transformation, were studied in continuous culture for a period of 4–6 weeks [15]. Roots were periodically checked for flavonoid content indicated by radical scavenging activity, and only those found to be overproducing flavonoids were used for nanoharvesting experiments. Sterile nanoparticles of desired functionalization were sonicated in MS media for 1 h to make a uniform 10 mg/mL stock solution, which was diluted to different concentrations (1 μg/mL, 10 μg/mL, 100 μg/mL, 1 mg/mL and 2.5 mg/mL) for root uptake experiments. Growing portions of Solidago nemoralis hairy roots were cut from agar plate cultures and rinsed carefully with sterile water. Roughly 500 mg of hairy roots were placed in sterile Nalgene™ centrifuge tubes and 10 mL of a desired nanoparticle solution and the antibiotic cefotaxime (200 mg/L) were added, followed by vortex shaking in the dark for 48 h. Uptake experiments for each concentration of nanoparticles were performed in triplicate.

2.4.1. Fluorescence imaging of nanoparticle uptake

Fluorescence microscopy of RITC tagged amine-functionalized nanoparticles was performed using a Nikon Ti–U inverted microscope. Roots exposed to fluorescent nanoparticle solutions were submerged in trypan blue (TPB) solution (0.04%) for 10 min and then rinsed with water. Roots with and without TPB treatment were sliced in petri dishes using razor blades, and smashed gently onto glass slides with cover slips before bright field and fluorescence imaging. For some of the roots, a drop of TPB solution was added on top of the sliced roots and the images were taken when the roots were in TPB solution.

2.5. Nanoparticle separation and flavonoid recovery

After 48 h exposure, nanoparticles were recovered from hairy roots by centrifugation (7000 rpm) for 15 min. Whole roots were then separated and kept in sterile conditions for viability studies. The remaining solution was centrifuged again at high speed (17,000 rpm) and the pellets re-suspended in 1 mL of ethanol for particle activity measurements. Flavonoids were recovered from the particle surface by suspending nanoharvested particles in ethanolic CA solution (20% w/v) followed by 24 h of vortex shaking (see Section 2.3).

To verify that the centrifugation step was not responsible for antioxidant activity on recovered particles, one experiment was performed in which hairy roots and particles were centrifuged for 15 min (without prior incubation in the vortex mixer).

2.6. Hairy root extract preparation

Solidago nemoralis hairy roots extract was prepared using the method discussed elsewhere [4]. Briefly, roots were removed from the cultures, rinsed with growth medium, flash frozen in liquid nitrogen, and lyophilized for 24 h and stored at −80 °C. For extraction, 100 mg/mL lyophilized tissue was shaken overnight in extraction solvent (0.1 M HCl in methanol) and the extracts were filtered to remove remaining plant material and dried in a Labconco rotary evaporator (Kansas City, MO, USA) under reduced pressure and stored at −80 °C.

2.7. Activity determination

2.7.1. Radical scavenging activity (RSA)

For RSA determination, 100 μL of the flavonoid solution in ethanol (0–0.5 mg/mL) was mixed with 1 mL of DPPH solution (0.1 mg/mL in ethanol) in a 2 mL vial. The vials were covered immediately and after 30 min the absorbance was measured at 517 nm. The initial DPPH absorbance was measured by adding 100 μL of ethanol with 1 mL of DPPH solution, and RSA was calculated from the percent decrease in DPPH absorbance. RSA of flavonoids recovered by ethanolic CA was determined using the same procedure. For RSA of particle-bound flavonoids, particles with a known amount of bound flavonoids (Section 2.3) were dispersed in 1 mL of ethanol by sonication and 100 μL of the solution was added to DPPH.

For RSA measurement of plant extracts, 50 μL of 2 mg/mL extract solution in methanol was added to 450 μL of tris-HCl buffer and 50 μL of the resultant solution were added to a 96 well plate in triplicate. 100 μL of 0.1 mM DPPH solution in methanol was added to each well and the plate was then kept in the dark for 30 min. Absorbance was measured at 517 nm and RSA of root extracts calculated from the percent decrease in DPPH absorbance.

2.7.2. Specific radio-ligand binding displacement activity

Radio-ligand binding displacement activity of nanoharvested particles and their eluents was performed by displacing ³H-MLA from rat hippocampal membranes [66,77]. Membranes were prepared from freshly harvested hippocampal tissues from adult, male Sprague-Dawley rats by homogenizing the tissues in sucrose buffer (0.32 M sucrose with 50 mM NaH₂PO₄, 0.1 mM EDTA, 0.1 mM PMSF and 0.01% w/v NaN₃) at pH 7.4 using a glass homogenization tube and Teflon pestle. The homogenate was washed two times at 1000×g for 10 min and the supernatants re-centrifuged at 50,000×g for 20 min. The pellet was re-suspended in ice-cold buffer and the protein content was determined by the Pierce Method in Bicinchoninic Acid Kit (Sigma-Aldrich). Final protein content was adjusted to 3 mg/mL using extra buffer solution and samples were frozen at −80 °C before use. Individual samples were screened for displacement activity in quadruplicate against 2 nM solution of ³H-MLA. 100 μL of samples containing nanoparticles or nanoparticle eluents were added in quadruplicate to a 96 well plate containing membranes (1 mg/mL protein content). After 15 min of incubation, ³H-MLA was added and the plate incubated for further 2 h, before harvesting onto a 96 well GF/B filter array and rapidly washed 3 times with 350 μL of 50 mM tris-HCl buffer (pH 7.4) and dried overnight. Finally, 35 μL of scintillation fluid (Microscint 20, Packard Inc.) was added to each filter and the plate was counted using a scintillation counter (Packard TopCount NXT). Nonspecific binding in the presence of 300 μM nicotine was subtracted from total binding to find specific binding.

2.8. Viability and flavonoid synthesis of hairy roots after nanoparticle exposure

Viability of Solidago nemoralis hairy roots after exposure to particles was studied by re-culturing them (at least 10 roots for each nanoparticle type and concentration) on agar plates containing MS media supplemented by 3% sucrose and 250 mg/L cefotaxime. Re-cultured root growth and viability were observed after every week up to four weeks. To measure the flavonoid re-synthesis capability of nanoharvested roots, extracts were prepared in triplicate after regrowth and RSA of the extracts measured and compared to a control (roots not treated with nanoparticles).

3. Results and discussion

3.1. Nanoparticle characterization

Spherical MSNPs (approximately 170 nm diameter) were synthesized by a modified Stober method using CTAB, TEOS and Pluronic F127 as structure directing agent, silica precursor, and dispersant, respectively. Complete removal of the organic surfactants was confirmed by FTIR analysis by the disappearance of the peak corresponding to CH₂ stretching (2800–3000 cm⁻¹) after acidic ethanol wash (Fig. A.1 in Supplementary materials) [46], thus preventing any toxic effects of surfactants from impacting cell viability. Surface characteristics (BET area, average pore diameter and BJH pore volume) of the dried particles were determined using nitrogen adsorption and are presented in Table 1. Average pore diameter (2.76 ± 0.23 nm) falls in the expected range of values obtained from the literature when CTAB is used as a templating agent [78]. The BET specific surface area and BJH pore volume were 953 ± 51 m²/g and 1.21 ± 0.09cm³/g for non-functionalized MSNPs, confirming their potential as high surface area platforms for biomolecule adsorption. In a previous study, MSNPs were functionalized with a range of TiO₂ amounts via the hydrolytic condensation of TEO, and optimum capacity for the model flavonoid quercetin (in mg quercetin/g particle) was observed at a TiO₂ incorporation of 425 ± 9.2mg/g particles [46]. This sample is labeled MSNPT in Table 1. TEM and STEM imaging of these particles shows that there is a tradeoff between the creation of binding sites with by TiO₂ functionalization in the mesopores of ordered cubic MSNPs and the blocking of pores with increased TiO₂ functionalization, where the sample with the greatest quercetin loading (MSNPT) has titanium both inside the pores and at the particle surface [46]. The average nanocrystal size for MSNPT is larger than the average pore diameter from nitrogen adsorption, which suggests that nanocrystallites form only on the particle outer surface, and that TiO₂ in the pores is amorphous (a suitable form for adsorption). Reduction in pore diameter occurred not only due to TiO₂ deposition but also due to restructuring and contraction of silica matrices (indicated by reduced d-spacing) resulting from the solvothermal functionalization process. The presence of uniformly distributed amorphous TiO₂ (confirmed by wide angle XRD [46]) was also evident from peak broadening and reduced contrast in low angle XRD patterns after TiO₂ loading, consistent with other literature for TiO₂ incorporation in mesoporous silica [31,33,79].

Table 1.

BET specific surface area, BJH pore volume and average pore diameter of TiO₂ functionalized, amine functionalized, and TiO₂-amine functionalized mesoporous silica nanoparticles (MSNPs) with optimum TiO₂ content compared to non-functionalized MSNPs.

Particle types	Average particle diameter	Amount of TiO2	Amount of amine	BET surface area	BJH pore volume	Average pore diameter
	nm	mg/g particle	mmol/g particle	m2/g	cm3/g	nm
MSNP	165 ± 19a	–	–	953	1.21	2.76
MSNPT	185 ± 29	425 ± 9.2b	–	629	0.59	2.21
MSNPA	172 ± 26	–	2.00 ± 0.16b	400	0.57	2.50
MSNPTA	174 ± 32	407 ± 13	0.65 ± 0.07	325	0.33	1.97

^aStandard deviation values resulted from measurement of 20 particles selected randomly (using ImageJ Software).

^bStandard deviation values resulted from quadruplicate measurement for chemical analysis.

Both MSNPs and MSNPTs were functionalized by amine (-NH₂) groups using an aminosilane precursor, APTES, resulting in materials referred to as MSNPA and MSNPTA, respectively. For MSNPA and MSNPTA, the amount of amine on the particle surface (presented in Table 1) corresponds to 0.0038 mmol/m² and 0.0012 mmol/m² on bare silica and TiO₂-functionalized silica, respectively. Based on the reported area of an amine group in monolayer coverage (50 Ų surface area/amine functional group) [75], the particle coverage of amines corresponds to 114% and 37% of monolayer coverage for MSNPA and MSNPTA, respectively. The reduced monolayer coverage on TiO₂ (MSNPTAs) is probably due to smaller surface area and lower pore size of MSNPTs compared to bare MSNPs. After amine functionalization, XRD peaks were broadened and less resolved (Fig. A.2), but this effect was not as drastic as after TiO₂ functionalization [46]. MSNPs retain their mesostructure after amine functionalization, which is in agreement with literature observation for other types of mesoporous silica [80,81].

Similarly, MSNPs retained their spherical shape after TiO₂ and/or amine functionalization (Fig. 1) and functionalization did not measurably affect particle diameter (Table 1). The effect of functionalization on the surface properties of MSNP was studied using nitrogen sorption isotherms, BJH pore size distribution, and α_s-plots, as shown in Fig. A.3. The nitrogen sorption isotherm for MSNPs is a type IV isotherm with clear capillary condensation, which indicates a well-ordered mesostructure [67]. For all functionalized particles, the capillary condensation step was not evident, consistent with redued mesopore volume in the presence of TiO₂ and amine groups. The pore size distribution of MSNPs showed a clear peak at around 2.2 nm, but for all types of functionalized particles, the peak sharpness (Fig. A.3b) and the average pore diameter decreased, as reported in Table 1. Prior studies of TiO₂ [33,42,79] or amine [82,83] loading onto mesoporous silica are consistent with our observations, where surface area, pore size and pore volume decreased after functionalization. MSNPTAs are less common, and in this case showed an additional surface area reduction (approximately 50%) relative to the starting MSNPTs. The standard reduced adsorption (α_s) plots (Fig. A.3c) [72,73] show no micropore formation (pores < 2 nm). The external surface areas were also measured from α_s-plots tobe137.2m²/g for MSNP, and were reduced to 120.8 m²/g, 77.9 m²/g and 65.3 m²/g for MSNPT, MSNPA and MSNPTA, respectively. The gradual reduction in surface area and pore diameter and the absence of microporosity after both functionalization steps are consistent with monolayer-like coverage in the pores, providing surfaces appropriate to provide positive charge (to promote cell uptake) and adsorption sites for biomolecule binding.

Fig. 1.

SEM image of (a) MSNP, (b) MSNPT, (c) MSNPA and (d) MSNPTA (scale bar 500 nm for all images).

3.2. Nanoparticle uptake in hairy roots and hairy root viability

Plant cells are inherently different from animal cells in that they have rigid cell walls. It has been proposed that the cell wall may place a size limit on nanoparticle uptake based on the size of naturally occurring pores [48,53,61]. Silica nanoparticles with much larger size (up to 200 nm [62]) than the pores were shown to be internalized by plant cells, but the penetration mechanism through plant cell walls is still unknown. Enlargement of pores or new pores/channel creation after the disruption of cell wall proteins and polysaccharides is speculated to be the pathways for nanoparticle entry [48]. The exact uptake mechanism of nanoparticles through the cell membrane is also still unknown or disputed; endocytosis, direct penetration and pore creation by nanoparticles have all been proposed as mechanisms for nanoparticle uptake in plant cells [39,51,53]. Direct penetration occurs for strongly positively charged nanoparticles and after that the nanoparticle can localize in lysosomes, endosomes, cytoplasm, mitochondria, endoplasmic reticulum and cell nuclei [54]. Although nanoparticle transport through a cell membrane reaches equilibrium given sufficient time [54], smaller particles are transported faster than larger ones [84]. Bare and/or amine-functionalized silica nanoparticles (< 200 nm) or TiO₂ nanoparticles (< 140 nm) were reported to be internalized by plant cells [39,62,85,86], but the effect of combined amine and TiO₂ functionalization on internalization of nanoparticles is unknown. Thus, MSNPTAs were tagged with RITC for visualization of uptake in hairy roots by fluorescence microscopy. To confirm internalization, fluorescence of RITC on particles outside of cells [65] was quenched by trypan blue (TPB) added after uptake and fluorescence microscopy was performed, by rinsing off TPB from the roots [87] and also without rinsing [65].

At concentrations of 100 μg/mL MSNPTA, fluorescent images (Fig. 2) showed evidence for nanoparticle internalization. The bright field images (Fig. 2a and c) show clear outlines of cells with green chloroplasts inside. The corresponding fluorescence image (Fig. 2b) of the root cells after 24 h of nanoparticle exposure showed bright fluorescence, which remained after TPB treatment (Fig. 2d), indicating nanoparticle association and partial internalization by the cells. In contrast, the images of root cells not exposed to nanoparticles and TPB (Fig. A.4b) showed no fluorescence. The faint red fluorescence observed when TPB solution was added to the roots without any nanoparticle treatment (Fig. A.4d) is attributed to the auto-fluorescence of TPB in contact with outer cell membranes [87,88].

Fig. 2.

Bright field (left) and corresponding fluorescence microscopic image (right) of Solidago nemoralis hairy roots after RITC-tagged MSNPTA uptake: (a) & (b) without and (c) & (d) with trypan blue addition for 100 μg/mL nanoparticle solution (exposure time 150 ms). (For interpretation of the references to color in this figure, the reader is referred to the web version of this article.)

The fluorescent image presented in Fig. 2b shows some nanoparticles in aggregates, also visible in less quantity after TPB treatment (Fig. 2d). These particles were observed more closely to verify internalization and localization in magnified fluorescence images with or without TPB (Fig. 3). Nanoparticles were found in clusters as well as scattered throughout the cells (Fig. 3a). After TPB addition and rinsing, images of the roots (Fig. 3b) still show intracellular localization, but the reduction in the number of fluorescent spots indicates some particles (or clusters) were quenched. Particle entry into the plant cells and cluster formation inside the cells has been reported based on particle surface chemistry (hydrophobicity) [89]. For MSNPs, Torney et al. found endocytosis clusters inside tobacco plant protoplasts [49], whereas Chang et al. reported direct penetration (energy independent) for amine functionalized MSNPs into Arabidopsis plant roots [39]. It is also possible that TPB entered and quenched RITC in some of the viable cells, which can explain reduced intracellular fluorescence after TPB treatment. When particles enter plant cells, the cell membrane structures can be disrupted, increasing the cell wall permeability [53], and allowing TPB entry.

Fig. 3.

Magnified fluorescence image showing intracellular localization of MSNPTAs for 100 μg/mL nanoparticle concentration (a) in absence of trypan blue and (b) in the presence of trypan blue.

At a higher particle concentration (1 mg/mL), the fluorescent images were too bright for individual clusters to be seen (Fig. A.5 in Supplementary materials). After TPB addition, almost all of the cells were still very brightly fluorescent, suggesting a higher amount of particle internalization, which indicates a concentration-dependent equilibrium in particle uptake. Finally, because imaging in the presence of a TPB-containing solution has been used to confirm particle internalization [90], fluorescent images were collected after adding a drop of TPB solution to the roots to ensure quenching of extracellular particles (Fig. A.6). The bright field images in TPB solution are bluish in color suggesting cells are completely covered with TPB solution. The corresponding fluorescent images appear similar to those for which TPB was rinsed before imaging. Since extracellular particle fluorescence quenching by TPB is very fast (< 2 min [87]), images taken after rinsing the TPB solution provided additional evidence for particle internalization and localization.

Another requirement for the nanoharvesting process is to not only isolate flavonoids from hairy root cultures but also to maintain viable roots capable of regrowth. Uptake and translocation of engineered nanomaterials in vascular plants, and associated phytotoxicity and cytotoxicity were discussed by Miralles et al. [53], where several factors were identified as contributors to toxicity, such as nanoparticle surface characteristics and possible metal ion leaching. They advise that viability should be studied case-by-case. TiO₂ nanoparticles have been reported to inhibit plant leaf growth and transpiration by reducing root water transport, to delay photosynthesis by acting as photocatalysts, and to disrupt cell function and microtubule networks by generating reactive oxygen species (ROS) [52,91,92]. Cell membrane disruption during the uptake process can also lead to cellular dysfunction [54].

The viability of hairy roots exposed to 2.5 mg/mL nanoparticle concentration, the same experimental conditions used for nanoharvesting experiments, was demonstrated over a period of two weeks after particle exposure and reculturing in petri dishes containing agar (Fig. 4). The results are consistent with the absence of an effect on root growth after exposure to nanoparticles. This is shown in Fig. A.7 (Supplementary materials), where data are shown for hairy roots treated with 0 (control), 10 μg/mL, and 1 mg/mL nanoparticle solutions after 0, 2 and 4 weeks of re-culturing. Root viability is similar to the control over the concentration range tested.

Fig. 4.

Viability of Solidago nemoralis hairy roots culture after exposure to 2.5 mg/mL functionalized MSNPs for 48 h and re-culturing: (a) control after 0 week, (b) only TiO₂ functionalized MSNPs after 0 weeks, (c) TiO₂-amine functionalized MSNPs after 0 weeks, (d) control after 2 week, (e) only TiO₂ functionalized MSNPs after 2 weeks and (f) TiO₂-amine functionalized MSNPs after 2 weeks.

3.3. Nanoparticle recovery and flavonoid activity measurements

Genetically modified S. nemoralis hairy roots synthesize polyphenolic flavonoids that are detectable from their radical scavenging activity RSA (antiradical activity using DPPH assay), a broad measure of their ability to reduce oxidative stress [93]. Also, flavonoids from S. nemoralis have demonstrated specific bioactivity in their ability to displace ³H-MLA from nicotinic receptors [66]. After the nanoharvesting process, nanoparticles were recovered from the solution in contact with the hairy root cultures by centrifugation. Particles were analyzed for RSA directly. Particles were also washed with ethanolic citric acid (CA) and the activity of flavonoids recovered in the solution was tested. We have previously demonstrated ethanolic CA for the effective recovery of flavonoids from TiO₂-functionalized mesoporous silica [46], presumably by ligand displacement of the chelated flavonoid. CA is ubiquitous in nature (found blood plasma at ~1 mM) and has tremendous advantages over other extraction agents for its excellent biocompatibility. For these reasons, CA is widely used in living systems for mineral dissolution and detoxification [94]. Moreover, the citrate ion provides very good surface binding, enhances water solubility and reduces aggregation [95], and as a result can increase the bio-availability of hydrophobic flavonoids.

RSA was measured directly from solutions containing suspended MSNPT or MSNPTA particles recovered from the nanoharvesting process (Fig. 5). Both MSNPT and MSNPTA particles were used in nanoharvesting, where the positive charge of the amine groups (in MSNPTAs) is expected to be necessary for localization in the cell. Additional control experiments included RSA measurements for particles exposed only to the media, and particles exposed to media in which hairy roots have been kept (but not incubated with hairy roots directly). All particles were vortexed in their respective solution (media only, media after 24 h of incubation with hairy root cultures (leftover media), and media in which particles and hairy roots were incubated together for 24 h). For MSNPTAs, nanoparticles exposed to roots showed a statistically significant increase in activity compared to particles exposed to leftover solution after shaking with roots and control (no roots), consistent with the color change of the particles after exposure to roots (Fig. A.8 in Supplementary materials). This color change suggests that active metabolites bound to the particles. For MSNPTs, no significant activity difference was found among particles exposed to roots, particles exposed to leftover solution, and particles exposed only to media (Fig. 5). Thus, internalization of the nanoparticle, which was mediated by the amine functionalization on the particle surface, is necessary for nanoharvesting of these flavonoids to occur. Flavonoids in plants are localized throughout plant cells, with quercetin derivatives accumulating in the plasma membrane, but also found in the nucleus [96]. The mechanism of nanoparticle uptake and release are currently under investigation, but positively charged particles have previously been shown to be taken up by direct penetration of cell membranes [54,57]. After particles are internalized, flavonoids bind to the TiO₂ groups of the particles [46], presumably through complexation of their cis-diol group with Ti, and then are carried out of the cells during particle expulsion.

Fig. 5.

Radical scavenging activity (RSA) of MSNPs functionalized with only TiO₂ and TiO₂-amine group when nanoparticles (25 mg, 2.5 mg/mL in MS media) were exposed to 500 mg hairy roots, leftover solution and control solution (error bars are standard deviation from quadruplicate measurements and statistically significant difference in mean (p ≤ 0.05) from unpaired t-test shown by ‘asterisk’).

Several additional controls were investigated to demonstrate that uptake of particles in the hairy root cultures was necessary to observe RSA. Centrifugation of particles in the presence of hairy roots (without incubation of the roots and particles) did not result in any increase in particle activity (data not shown), suggesting that prolonged interaction during vortex shaking is necessary to allow particles to enter and leave cells. Roots were also subject to vortex shaking and centrifugation in the absence of nanoparticles. No measurable RSA activity was found in the supernatant of any of the experiments conducted, which proves that metabolites were not secreted by the roots and subsequently bound to the particles in solution. Metabolites chelating to MSNPTA particles inside of cells is consistent with the elemental analysis of internalized NaYF₄ nanoparticle clusters, which provided evidence for phosphate attachment to the particle surface [97]. Similarly, silica nanoparticles were shown to adsorb nutrients from the cellular environment [62], thus making the nutrients unavailable to plant cells. Regarding potential mechanisms of particle recovery from the hairy root cultures, there is a strong tendency in cells with internalized nanoparticles towards cell membrane regeneration and repair, leading to the spontaneous direction of some of the endocytosed materials towards the cell membrane [98]. Also, endosomal and recycling pathways often gradually reduce the intracellular concentration of internalized materials [98,99]. Particles inside endocytotic vesicles can be removed by fusion with the cell membrane [59] or can become free in the cytoplasm after endosomal escape, which is reported to happen within 12 h for endocytosed mesoporous silica nanoparticles [100]. These free particles, along with those that directly penetrate the cell membrane, can be steadily exocytosed from the cellular interior by secretary vesicles formation and fusion with the cell membrane [101]. Exocytosis, like endocytosis, reaches thermodynamic equilibrium, but is faster for smaller nanoparticles [54,102]. More than 50% of internalized nanoparticle were observed to be transported to the cell periphery and subsequently expelled out of mammalian cells within 1 to 2 h [102,103].

To show that biologically selective compounds could be nanoharvested using MSNPTAs (since both amine and TiO₂ are required for nanoharvesting), we demonstrated that eluted flavonoids specifically bind α7 nicotinic receptors in rat hippocampal cell membranes. Root extracts from the genetically modified S. nemoralis hairy roots used in this study were previously shown to contain specific flavonoids that selectively bind α7 nicotinic receptors [66]. Binding to α7 receptors is presented as the percent displacement of ³H-MLA from hippocampal membrane and compared to the binding of only ³H-MLA without adding anything (as control). Non-specific binding was measured in the presence of excess nicotine and subtracted from total binding to find the specific binding. Specific binding of ³H-MLA in the presence of extract or nanoharvested compounds (nanoparticle and recovered in solution) is presented as percent of specifically bound ³H-MLA in the control. The percentage of specifically bound ³H-MLA (Fig. 6a) decreased in the presence of increasing plant extract concentration, showing that flavonoids present in the extract have nicotinic receptor activity as predicted for specific flavonoids from S. nemoralis [66].

Fig. 6.

[³H]-methyllycaconitine (³H-MLA) binding displacement activity of (a) Solidago nemoralis hairy root extracts and (b) TiO₂₂ and amine functionalized MSNPs (25 mg, 2.5 mg/mL) and their eluent in ethanolic citric acid (20% w/v) exposed to 500 mg hairy root compared to control (error bars are standard deviation from quadruplicate measurements and statistically significant difference in mean (p ≤ 0.05) from unpaired t-test shown by ‘asterisk’).

MSNPTAs exposed to hairy roots show a high degree of displacement of ³H-MLA, but nanoparticles that were not exposed to roots or exposed to only leftover solution show no statistically significant displacement of ³H-MLA compared to control (Fig. 6b), suggesting that polyphenolic flavonoids active towards α7 nicotinic receptors are present on only the particle surface that was able to enter into and bind metabolites from hairy roots. The displacement is the same as a solution containing 0.16mg/mL of root extract (Fig. 6a). Similarly, the average displacement by compounds recovered in solution (ethanolic CA) from nanoparticles exposed to roots was higher compared to that of the nanoparticles not exposed to roots or exposed to root left-over solution only (Fig. 6b), although the differences are not as significant as when the activity is measured directly from the particles. The single washing step by ethanolic CA may not have been sufficient to remove all the flavonoids from the particle surface [46].

3.4. Flavonoid re-synthesis potential of roots after nanoharvesting

We have shown that S. nemoralis hairy roots remain viable and capable of regrowth after exposure to 2.5 mg/mL MSNPs, irrespective of the type of functionalization (TiO₂2 and/or amine). To determine whether nanoharvested roots have the ability to synthesize new flavonoids, we made extracts of nanoharvested roots after 2, 10 and 17 days of re-culturing. Extracts from fresh roots were prepared to measure concentration-dependent RSA, which shows RSA increases with increasing extract concentration (data not shown). RSAs of roots exposed to nanoparticles are presented in Fig. 7, where roots exposed to MSNPT (TiO₂ only) and MSNPTA (TiO₂ and amine) along with control roots (no particles) were re-cultured. After 2 days of exposure, roots exposed to MSNPTA (but not MSNPT) gave an extract showing less RSA activity, suggesting that a fraction of the metabolites were nanoharvested. This again supports our hypothesis that both amine and TiO₂ functional groups are required for flavonoid removal from plant cells. The root extract activity remained roughly the same after 10 days, which probably is due to the readjustment of roots to the new culturing environment. 17 days after exposure, roots exposed to MSNPTA regained their original activity, and their RSA was statistically the same when compared to control roots and roots exposed to MSNPT. The regaining of RSA activity in the root extracts indicates flavonoid re-synthesis after nanoharvesting using MSNPTAs. Thus, in situ product removal from plant culture, hypothesized as one way to increase the rate of production of secondary metabolites by biosynthesis in plants [20], has been demonstrated.

Fig. 7.

Comparative radical scavenging activity (RSA) of root extracts after exposure to control (no nanoparticles), only TiO₂ functionalized MSNPs and TiO₂-amine functionalized MSNPs (error bars are standard deviation from triplicate measurements and statistical difference in mean (p ≤ 0.05) from unpaired t-test represented by ‘asterisk’).

4. Conclusions

The use of engineered mesoporous silica nanoparticles in a nanoharvesting process, which is broadly applicable for several classes of biomolecules, was demonstrated for Solidago nemoralis plant hairy root cultures. MSNPs with uniform, well-defined pores were successfully synthesized and functionalized with amines and TiO₂ to promote both cellular nanoparticle uptake and polyphenolic compound binding, respectively.

The first requirement for nanoharvesting, the uptake of functionalized nanoparticles in hairy roots was demonstrated by fluorescence microscopy of RITC-tagged MSNPTA. To further demonstrate that particles are iinternalized within root cells, the fluorescence of extracellular particles was quenched by trypan blue (TPB) solution. The next requirements for nanoharvesting were demonstrated by showing that the particles could be readily recovered after hairy root uptake, and that the roots were viable and capable of synthesis of new flavonoids after nanoharvesting, as demonstrated by post-uptake growth studies and plant extract activity analysis. Finally, to demonstrate that the compounds recovered on the surface of the MSNPTAs are active flavonoids, the increase in radical scavenging activity (RSA) was analyzed after nanoharvesting. Nanoparticles exposed to roots and their eluents in ethanolic citric acid solution were also tested for biological activity, where both particles and recovered compounds were able to displace radiolabeled ³H-MLA from rat hippocampal cell membrane nicotinic receptors. ³H-MLA displacement from nicotinic receptors proves that the nanoharvested compounds are polyphenolic flavonoids that exhibit the type of bioactivity found in plant extracts.

The ability of particle-bound flavonoids from S. nemoralis to bind to nicotinic receptors on synaptic membranes suggests a new direction for treatment, where particles after nanoharvesting can be directly applied for biomolecule delivery to their cellular targets [66]. Furthermore, mesoporous silica layers can be coated onto magnetic nanoparticles to create functional magnetic core-shell particles, for enhanced recovery [35,104]. The magnetic core will facilitate particle recovery from living plant cells, while the TiO₂ functionalized mesoporous silica coating can act as a high surface area adsorbent in separation processes. The synthesized engineered particles are expected to represent a platform technology for the isolation and delivery of a broad range of therapeutic biomolecules in living tissues.