During the cultivation, four independent samples were taken at a given time point for fresh and dry weight analysis, sugar analysis, and quantification of rrBChE and total soluble protein . The bioreactor run was terminated at day 5 following induction and the rice cell aggregates were allowed time to sediment . The culture medium was withdrawn from the bioreactor through a sampling port using a peristaltic pump and stored at 4 ◦C The bioreactor head plate was then opened. Rice cell biomass was collected and vacuum-filtered on Whatman Grade 1 . Fresh biomass was weighed and stored at −20 ◦C. Performing a media exchange using 1.25-times-concentrated sugar-free medium together with 1.25-times-reduced culture volume and addition of kifunensine prior to and after the media exchange resulted in increased total production levels of active rrBChE, volumetric productivity, and specific productivity by 1.5 times, 3.4 times, and 1.5 times, respectively, compared with a bioreactor run with same operating conditions but no kifunensine treatment. Moreover, kifunensine enhanced the excretion of recombinant rrBChE glycoprotein through the secretory pathway, leading to 44% of total rrBChE in the culture medium at day 5 following induction and increasing extracellular rrBChE purity to 1.6% rrBChE/TSP compared with 0.8% cell-associated rrBChE/TSP. Coomassie-stained SDS-PAGE and Western blot analyses showed different migration bands of rrBChE with and without kifunensine treatment due to different N-glycan structures. N-Glycosylation site-specific analysis revealed increased oligomannose glycans at site N57, N246, N341, and N455 in both purified cell-associated and culture medium-derived rrBChE in the presence of kifunensine, while the mass transfer limitation of kifunensine was thought to be the main reason for the weak inhibition of α-mannosidase I in this bioreactor study.
At the laboratory scale, we produced ~16 mg of rrBChE in a 2 L working volume during a 12-day batch run,vertical farming supplies corresponding to a volumetric productivity of 0.680 mg L−1 day−1 . A technoeconomic model developed for semicontinuous large-scale production of rrBChE at a higher volumetric productivity showed that the process could be cost-effective with a cost-of-goods sold of ~$660/gram, less than 3% of the estimated cost of plasma-derived hBChE at ~$25,000/gram. The addition of compounds to the culture medium to alter the function of glycan-modifying enzymes is the simplest method to modify N-glycan structure of a target glycoprotein compared to other methods. As a bio-processing approach, it does not require alteration of the primary amino acid sequence of the target protein , or time-consuming glycoengineering of the host that could impact cell growth or viability, yet still allows secretion of the product into the culture medium. For example, adding kifunensine for N-glycan modification is a simple and effective way of obtaining oligomannose glycoproteins with reduced plant-specific xylose and fucose moieties. However, this method may not be cost-effective in large-scale production depending on the production level and market price of the glycoprotein product, the amount and frequency of kifunensine addition , and the price of kifunensine in bulk quantities. Currently, at our laboratory-scale pricing for kifunensine , the addition of 5 µM kifunensine in NB-S increases induction medium costs by ~14-fold and contributes ~$225 in reagent costs for the 5 L bioreactor run. Although the cost of the growth and induction medium is still significantly lower than mammalian cell culture medium, and the price of kifunensine is likely to decrease more than 10-fold with larger demand and bulk pricing, it may be advantageous to reduce bioreactor working volume during the induction phase to minimize kifunensine cost, enhance mass transfer, and concentrate extracellular rrBChE. Employing current genomic editing techniques such as CRISPR/Cas9 to knock out XylT and FucT genes to remove plant-specific α-1,3 fucose and β-1,2 xylose in host rice lines, similar to what was done in N. tabacum BY-2 cell lines without negative impacts in terms of cell growth rate, would be worth investigating as an alternative to modify N-glycans of secreted glycoproteins, such as rrBChE, for large-scale operations.
Graminaceous plants, like other so-called metal-tolerant plants, mostly sequester metals in roots to protect reproductive and photosynthetic tissues . The ability to store metals in underground tissues is used in phytoremediation to reinstall a vegetation cover on heavily contaminated areas and limit the propagation of metals into the food chain . Panfili et al. showed that the grass species Festuca rubra and Agrostis tenuis accelerate the weathering of zinc sulfide when grown on contaminated dredged sediment, thus increasing Zn bio-availability in the rhizosphere. After two years of plant growth, micrometersized Mn-Zn black precipitates were observed at the surface of Festuca rubra roots, but not characterized . Zinc precipitation may be a bio-active tolerance mechanism in response to metal toxicity, or a passive mineralization at the soil-root interface . Clarifying this issue and determining the mineralogy and structure of this natural precipitate is important to enhance the effectiveness of using graminaceous plants in phytoremediation. These questions are addressed here with electron microscopy and synchrotron-based microanalytical tools, including X-ray fluorescence , extended Xray absorption fine structure spectroscopy and X-ray diffraction . Micro-XRD was employed to determine the nanocrystalline structure of the Mn-Zn precipitates and the nature of defects through modeling of their scattering properties . We show that the root precipitates are present in the root epidermis and consist of a poorly crystallized phyllomanganate with a constant Zn:Mn ratio higher than reported so far for any natural and synthetic manganate.The composition in major elements of the dredged sediment was 68.3 % SiO2, 6.9 % CaO, 4.8 % Al2O3, 2.4 % Fe2O3, 0.7 % P2O5, and 7.2% organic carbon, and the composition in a trace metals was 4700 mg.kg-1 Zn, 700 mg.kg-1 Pb, and ~270 mg.kg-1 Mn. Seeds of F. rubra were sown in plastic pots filled with 40 kg of either the untreated sediment, the sediment amended with 3 wt. % hydroxylapatite, or the sediment amended with 5 wt. % Thomas basic slag.
The pots were placed in a greenhouse without artificial lighting and daily irrigated with tap water in an amount similar to the mean rainfall in northern France. After two years of culture, the pots were dismantled to collect samples. The texture and color of the sediment in areas colonized by the roots were similar to a brown silty soil, whereas the initial sediment was black and compact. Roots of F. rubra from the treated and untreated pots were washed meticulously with distilled water to remove soil particles from the surface and then freeze-dried. The speciation of zinc in the initial sediment and in the rhizosphere of F. rubra after the two years of vegetation was described previously . Briefly, in the untreated and unvegetated sediment, Zn was distributed as ~50% sphalerite, ~40% Zn-ferrihydrite, and ~10 to 20% -hydrotalcite plus Zn-phyllosilicate. In the presence of plants, ZnS was almost completely dissolved, and the released Zn bound to phosphate and to Zn phyllosilicate plus -hydrotalcite . The coaddition of mineral amendment did not affect the Zn speciation in the vegetated sediment. The Zn:Mn and Ca:Mn ratios were measured with an Eagle III µ-XRF spectrometer equipped with a Rh anode and a 40 µm poly-capillary. The spectrometer was operated under vacuum at 20 kV and 400 µA,vertical lettuce tower and fluorescence was measured for 300 s per point. Micro XRF, XRD and EXAFS data were collected on beamline 10.3.2 at the Advanced Light Source . Short root fragments were attached to the tips of glass capillaries and cooled down to 110-150 K to minimize radiation damage . X-ray fluorescence maps were taken at 10 keV incident energy, with a beam size ranging from 5×5 µm to 16×7 µm . Fluorescence counts were collected for K, Ca, Mn, Fe and Zn with a seven-element Ge solid-state detector and a counting time of 100 ms per pixel. For µ-EXAFS measurements, the vertical beam size ranged from 5 to 7 µm. A maximum of two spectra per precipitate were taken at either the Mn or the Zn K-edge to prevent the reduction of tetravalent to divalent Mn and the increase of structural disorder under the beam . Diffraction data were collected with a CCD camera at 17 keV and exposure times of 120-240 seconds. At this energy, the incident flux and absorption cross-sections are low enough to make radiation damage during an exposure negligible even at room temperature. A background pattern was recorded next to each precipitate to subtract the scattering contribution from the root so as to obtain the precipitate pattern. Diffraction patterns collected on different precipitates were all statistically identical, and thus summed up to optimize data quality. Calibration of the energy and camera distance were obtained using an Al2O3 standard and Fit2D software . This software was also used to calculate the one-dimensional XRD traces from the radial integration of the two-dimensional patterns. Under the optical microscope, the Mn-Zn precipitates appear as black stains about ten to several tens of micrometers in size on the root surface . They are also observed in back scattered electron microscopy due to the presence of high-Z elements , but always are hardly noticeable in secondary electronimaging mode . This suggests that the precipitates are engulfed in the root epidermis and do not coat the root surface as iron and manganese plaques do . No differences were observed among precipitates from plants grown in the untreated and mineral amended sediments. This result, together with the compartmentation of the precipitates inside the roots, suggests a biological origin. This interpretation is supported also by the absence of Zn-rich phyllomanganates in the surrounding soil matrix . Elemental mapping of F. rubra roots shows that Zn is associated with Mn in localized spots, and uniformly distributed without manganese in the vascular cylinder as expected for this nutritive element .
All roots have Zn in their central stele, but not all are speckled with Mn-Zn precipitates. Some root fragments are partly covered by Zn-free Fe-rich plaques . These plaques are made of ferric oxyhydroxides, as indicated by their optical rusty color . In Zn-Mn-Ca tricolor representation all Mn-Zn precipitates generally have the same color , even among different roots , meaning that the relative proportions of Zn, Mn and Ca are about the same. The correlation coefficient between Zn and Mn counts for the precipitates is 0.8, with P-value < 0.0001 for the Anova F-test . The Zn:Mn atomic ratio was calculated from the relative absorption jumps measured at the Mn and Zn K-edges on four particles. For each particle, a pre-absorption edge background was removed first, and then a linear fit to the post-edge region was extrapolated back to the edge to measure edge jumps. The ratio of the Zn to Mn edge jumps is 0.310, which translates into a Zn:Mn ratio of 0.46 when taking into account the atomic absorptions of the two elements. A consistent 0.44 value was obtained independently with the Eagle III spectrometer. This analysis also confirmed that root precipitates have a constant Ca:Mn ratio. An atomic ratio of 0.41 was calculated after correction of the Ca-fluorescence from the root. Micro-EXAFS spectra were recorded in the vascular cylinder of four distinct roots, at spots containing little Mn. All spectra were indistinguishable, indicating that Zn speciation is uniform, and thus averaged. The resulting Root spectrum has the same frequency as the MnZn precipitate spectrum, which suggests that Zn is also mostly tetrahedral in the roots . However, in contrast to Mn-Zn precipitate, the second and third oscillations of the Root spectrum are not split, indicative of “light” back scatters from second-shell contributions. Consistently, the best spectral match to our organic and inorganic database of the Root spectrum was provided with Zn in a biofilm . This reference has 80 ± 10% Zn complexed to phosphoryl groups and 20 ±10% to carboxyl groups . Consistent with this other study, consideration of carboxyl and phosphate ligands alone, did not yield an optimal fit to the data. Zinc preferential binding to phosphate groups has been reported also in the roots of Arabidopsis halleri and A. lyrata grown hydroponically, on bacterial and fungi cells, and in biofilms . These studies have shown that Zn has a higher affinity for phosphate than for carboxyl groups, which is consistent with the predominance of the phosphate species in F. rubra roots. As the optimal fit to data was obtained using a trial-and-error approach, the sensitivity of the XRD simulations to key structural parameters needs to be assessed. A key parameter for birnessite’s ability to sorb trace metals is the origin of the layer charge.