Precision Delivery of Multiscale Payloads to Tissue‐Specific Targets in Plants

Abstract The precise deployment of functional payloads to plant tissues is a new approach to help advance the fundamental understanding of plant biology and accelerate plant engineering. Here, the design of a silk‐based biomaterial is reported to fabricate a microneedle‐like device, dubbed “phytoinjector,” capable of delivering a variety of payloads ranging from small molecules to large proteins into specific loci of various plant tissues. It is shown that phytoinjector can be used to deliver payloads into plant vasculature to study material transport in xylem and phloem and to perform complex biochemical reactions in situ. In another application, it is demonstrated Agrobacterium‐mediated gene transfer to shoot apical meristem (SAM) and leaves at various stages of growth. Tuning of the material composition enables the fabrication of another device, dubbed “phytosampler,” which is used to precisely sample plant sap. The design of plant‐specific biomaterials to fabricate devices for drug delivery in planta opens new avenues to enhance plant resistance to biotic and abiotic stresses, provides new tools for diagnostics, and enables new opportunities in plant engineering.


Analysis of interaction between Cs and silk fibroin
Cs is family of highly water soluble, negatively charged peptides extracted from silk fibroin heavy chain with a MW between 2-10kDa ( Figure S1) and a primary structure that accounts for only 10-15% of hydrophobic amino acids. We used Cs to enhance silk fibroin solubility for in planta application to build on the biodegradability and non-toxic nature of silkbased materials. Silk fibroin used in this study has an average MW of 100-150 kDa ( Figure S1) and we fabricated blends with a weight ratio between 0 to 40% Cs. By molarity, this means that in the blends, the number of Cs molecules is larger than the amount of silk fibroin. For example, for Cs20SF80 blends, we have roughly five times more Cs molecules than silk fibroin ones in the final material. Cs is incorporated in silk materials during the assembly process, when hydrogen bonds between silk nanomicelles and water are replaced with intermolecular hydrogen bonds.
During this step, nanomicelles coalesce and form a monolithic material. Cs would then participate in this assembly process as it is made by a portion of the silk fibroin primary structure.
However, being of smaller MW, the incorporation of Cs results in the weakening of the interactions/entanglement between large silk fibroin molecules, ultimately enhancing material disassembly upon exposure to water. The intermolecular and intramolecular interaction of hydrophobic amino acid domains may also be weakened. To further explore this mechanism, we have conducted several investigations of silk fibroin-Cs interactions both in water suspension and in solid, monolithic materials (i.e. film format).
In aqueous suspension, Cs does not show noticeable influence on silk nanomicelle size and on the secondary structure of the protein, as supported by DLS and CD measurements ( Figure   S1 and Figure 2b, respectively). Additionally, SDS-PAGE analysis of Cs-silk fibroin blends shows no aggregation or dimerization of Cs exposed to silk fibroin ( Figure S1b). Investigation of the Cs-silk fibroin blends in the solid format was conducted using WAXS, SAXS, TGA, DSC, ATR-FTIR and Raman. WAXS and SAXS showed no difference between silk fibroin and Cs20SF80 samples since the materials are not crystal dominant. Given the low impact of this study to the manuscript we did not incorporate the results of crystallography analysis in SI.
ATR-FTIR spectra of silk fibroin mixed with various content of Cs from 0% up to 40% were collected and showed no significant difference ( Figure S2); all the spectra depicted a wide peak centered around 1645 cm -1 , corresponding to random coil. Self-deconvolution and peak fitting were carried out for all the spectra collected to quantify the secondary structure content in each sample. Incorporation of increasing concentrations of Cs in the blends did not result in a change of beta sheet content, showing that Cs did not drive a random coil to beta-sheet transition during silk fibroin assembly. Turns increased slightly as the Cs content increases, which may be attribute to the intrinsic properties of Cs, which serves as hydrophilic linkers. To further investigate the interactions between silk fibroin and Cs in solid state, Raman spectra were collected for Cs, silk fibroin, and Cs20SF80 before (solid line) and after (dotted line) methanol treatment ( Figure S3). In particular, in this study we focused on the Amide I and III shifts and on the Fermi doublet peaks of the tyrosyl phenolic ring at 853 and 829 cm -1 . [1] In all the samples analyzed, analysis of the Amide bands showed that exposure to methanol resulted a random coil to beta-sheet transition of the silk materials, indicating Cs does not hinder polymorphic changes of the structural protein. The intensity ratio I853/I829 has been used to study the hydrogen bonding formed by the tyrosyl phenolic-OHa more hydrophobic tyrosine environment (i.e., reduction of structural water in the protein and of hydrogen bonding) corresponds to higher I853/I829 ratio. As shown in Table S1, the inclusion of Cs in silk fibroin materials results in an increased I853/I829 ratio, which corroborates the proposed mechanism that Cs reduces the formation of intermolecular hydrogen bonds.
Thermal analysis ( Figure S4) showed decomposition at about 180°C for Cs, 225°C for silk fibroin and 205°C for Cs20SF80. Calorimetric analysis depicted a Tg for Cs at 60°C, for silk fibroin at 77°C and at 75°C for Cs20SF80. In literature, this is referred to as the first Tg, i.e. Tg (1) of water-containing silk materials and corresponds to the removal of free water molecules entrapped between silk fibroin molecules during the random coil to beta sheet transition of the material. An exothermic peak was depicted at 125°C for silk fibroin only, followed by a large endothermic process. The exothermic peak is described in literature as formation of more stable structures in silk where water is present and acts as a plasticizer. The endothermic process is present in SF and Cs20SF80 samples and it corresponds to the release of some of the bound water molecules as free water and subsequent evaporation. The lack of the exothermic peak in the Cs20SF80 blend may be used as an evidence that Cs weakens the entanglement of silk fibroin molecules and reduces the formation of new, stable conformations between adjacent silk molecules upon water release. Both silk fibroin and Cs20SF80 blend showed an exothermic peak at 222°C and 214°C, respectively, which corresponds to a non-isothermal crystallization peak of silk material. [2] Payload release profiles from SF and Cs20SF80 Payload release profiles in silk fibroin constructs have been studied extensively in controlled drug release applications, [3,4] with most studies indicating that diffusion, swelling, and proteolytic degradation are primary drivers in this process. As targeted plant tissues are not protease-rich, we used simulated sap to investigate payload release profile. Rhodamine 6G, azoalbumin, and GFP-expressing Rhizobium tropici CIAT 899 (GFP-CIAT 899) were used as representative models for small molecules, large proteins, and bacteria, and their release profiles in SF and Cs20SF80 were investigated. GFP-CIAT 899 was used in the release study in lieu of Agrobacterium as several attempts of staining Agrobacterium were inconclusive due to interaction between silk fibroin and the dyes used for live/dead assays. Silk fibroin and Cs20SF80 were found to have negligible effects on fluorescence and absorbance signal. The release profile of all three payloads for both silk fibroin and Cs20SF80 follow a power law ( Figure S7a) described by the semi-empirical model developed by Ritger and Peppas, [4] = ∞ = , (S1) which can be rewritten as lg ( ) = ( ) + ( ), where is the fraction of released payload at time t, is the amount of released payload over time t (unit: hour), ∞ is the amount of released payload at infinity time, (i.e., the total payloads loaded), k denotes the release velocity constant determined by the structural and geometric characteristic of the system, and n denotes the exponent of release indicating the release mechanism. Parameters for the power law were obtained by linear fitting, shown in Table S2. Figure  protrusions, which display similar morphology to GFP-CIAT 899. All three payloads loaded into Cs20SF80 possessed a Super Case II release mechanism (n>1). This is likely due to the hydrophilicity of Cs, which dissolves easily in simulated sap and expedites the rate of sample degradation. These results show that Cs20SF80 allows for faster payload release profiles than SF, from small molecules, to large proteins, and to bacteria.

Release and transport model in xylem
The velocity of xylem sap flow is at the order of 10 -3 m s -1 although it varies a lot according to the condition of measured plants during the day [5] . However, the velocity we got here is at the order of 10 -5 to 10 -4 m s -1 , which may due to the influence of injection. This gives a Péclet Since we focus on the longitudinal transport along xylem, Equation S1 can be simplified to one dimensional (1D) condition as The initial condition (IC) and boundary conditions (BCs) are as follow Once a phytoinjector is injected into xylem, the payload is released following the power law, contributing to the concentration change at x=0 at time t 0 ( ) (Schematic S1). Mass conservation, i.e. payload released equals to that in the xylem, can be used to determine 0 ( ).
Schematic S1. Schematic of the model.
To solve this problem, let (S9) The concentration thus is Thus the concentration for the whole field is (S12) In addition, the concentration must meet mass conservation This integral equation determines boundary condition (0, ) = 0 ( ) and thus ( , ). While it is hard to explicitly solve the integral equation, we can solve it numerically. By Taylor series, we have Where n denotes time t and i is position x.
The power law release describes well the first 60% payload release but not for 100%. Thus, our model well describes the release and transport in the first 5 minutes only. For longer time period, the payload loaded to other parts of the phytoinjector may also be released and contributes as payload source at the injection site, which invalidates the mass conservation assumption used here.

Lucas-Washburn model for phytosampler
Reswelling of the phytoinjectors and diffusion of metabolite and catabolite in silk phytosampler was modeled with a Lucas-Washburn equation. [6] The fitting was carried out in MATLAB R2019a Curve Fitting Toolbox on collected data of penetration depth of water frontier in a phytosampler over time.

The fitting equation is
where H is the penetration depth, t is time (unit second). The adjusted R 2 =0.9932. The time t0=54.32 s may attribute to the cone shape of the phytosampler, which does not match the 1D case for Lucas-Washburn model.

Estimation of the amounts of payloads delivered by phytoinjectors
We have estimated the amount of cargo molecules delivered for a payload equivalent to 10wt% and compared it with the functional amount found in several plant tissues. In particular, we found that the deliverable weight of cargo molecules is in the order of 10s of ng per phytoinjector.
The total volume of xylem and phloem phytoinjector (Vphyt) is 18.741.05 nl and 9.111.83 nl, respectively. Given that Cs-silk fibroin blends have a density of 1.40 g cm -3 (which is equal to  (Table S3).
Plant hormones level is usually in the range of 0.1-50 ng g -1 of fresh weight. [7] As hormones are found in specific tissues such as shoot apical meristem and leaves, which have a weight in the order of tens to hundreds of milligrams, the delivered level of hormones by phytoinjectors would provide the plant with physiologically relevant quantities of hormones.
Micronutrients are present in plant tissues at concentration of ppm per dry weight, which approximately equals to 100ng g -1 fresh weight. This makes phytoinjectors suitable for delivering a wide range of micronutrients, including Cu, Mo, and Ni. (Table S4 [8] ). Note that micronutrients deficiency does not mean we need to deliver adequate concentration of micronutrients to plants. In addition, according to our experience, less than 1 ng of siRNA per leave of Nicotiana benthamiana result in the suppression of chlorophyll synthesis, indicating a very low functioning quantity of iRNA.

Experimental Section
Extraction of silk fibroin: The aqueous silk fibroin solution was prepared from Bombyx mori cocoons as described with modification. [9] Briefly, dime size cocoon pieces were boiled for 45 Cs preparation: Cs was prepared following the method described previously with modification. [10] Alpha-chymotrypsin was added to aqueous silk fibroin solution by an enzyme to substrate weight ratio 1:100, followed by incubation at 37 ˚C for 24 h. The gel formed was OriginPro 2017 software (OriginLab Corporation, Northampton, MA), following the previously described method. [11] Raman spectroscopy: Raman spectra were obtained with a Renishaw inVia Raman Microscope  Figure S11.
Bacteria culture: Rhizobium tropici CIAT899 expressing bacterial GFP was obtained from Miguel Lara. [12] R. tropici was cultured at 30 °C to OD600 of 1 following the instructions before use. GFP gene was cloned into pEAQ-HT vector and transformed into A. tumefaciens strain (LBA4404). Transformants were cultivated and selected at 30 °C for 24-36 h to OD600 of 1.5 in YM medium (0.4 g L -1 yeast extract, 10 g L -1 mannitol, 0.1 g L -1 NaCl 0.2 g L -1 MgSO4·7H2O, 0.5 g L -1 K2HPO4·3H2O, 15 g L -1 agar, pH 7) supplemented with 50 µg mL -1 rifampicin, 50 µg mL -1 kanamycin, and 50 µg mL -1 streptomycin. Payloads release: Simulated sap was prepared according to the xylem exudate. [13] Rhodamine 6g and azoalbumin were added to SF and Cs20SF80 (6% w/v of dry materials) to get a final concentration of 0.1 mM and 2 mg ml -1 , respectively. R. tropici was centrifuged at 3000 × g for 30 minutes and resuspended by SF and Cs20SF80 to get an OD600 of 1. The solutions were then cast on PDMS and dried overnight in a hood. The films were then cut into discs and attached to the bottom of a well of a 48 well plate, enabling only one side of the disc exposed to simulated sap. 1 ml of fresh simulated sap was added after the previous solution was collected for measurement. Released rhodamine 6g and GFP-expressing R. tropici were monitored based on fluorescence intensity (excitation at 524 nm and 499 nm, emission at 550 nm and 520 nm).

Preservation of Agrobacterium tumefaciens: A. tumefaciens
Released azoalbumin was monitored based on absorbance at 410 nm. At least three samples were tested for each case. The standard curve is in Figure S11. was cast over Al master in a 60 mm petri dish, degassed, and finally incubated at 70 ˚C for 2 h.
Phytoinjector fabrication: The desired amount of payloads were mixed with Cs20SF80 solution and added to negative PDMS molds, followed by centrifuge at 1200 ×g for 15 minutes. Molds were then kept in a fume hood to dry at room temperature overnight. The phytoinjector array was then cut into smaller arrays by a razor blade for tissue application. Pure Cs solution has a hydrodynamic radius below 1 nm.       c and d, photograph of phytoinjectors for xylem and phloem, respectively. Scale bar 1 mm.
Error bar means s.d.  HRP, respectively. R 2 is adjusted R-squared. Error bar means s.d.
Supporting Tables   Table S1.  The unit for time t is hour for paramater k.