Artificial nanovesicles for dsRNA delivery in spray induced gene silencing for crop protection

Summary Spray-Induced Gene Silencing (SIGS) is an innovative and eco-friendly technology where topical application of pathogen gene-targeting RNAs to plant material can enable disease control. SIGS applications remain limited because of the instability of dsRNA, which can be rapidly degraded when exposed to various environmental conditions. Inspired by the natural mechanism of cross-kingdom RNAi through extracellular vesicle trafficking, we describe herein the use of artificial nanovesicles (AVs) for dsRNA encapsulation and control against the fungal pathogen, Botrytis cinerea. AVs were synthesized using three different cationic lipid formulations, DOTAP + PEG, DOTAP, and DODMA, and examined for their ability to protect and deliver dsRNA. All three formulations enabled dsRNA delivery and uptake by B. cinerea. Further, encapsulating dsRNA in AVs provided strong protection from nuclease degradation and from removal by leaf washing. This improved stability led to prolonged RNAi-mediated protection against B. cinerea both on pre- and post-harvest plant material using AVs. Specifically, the AVs extended the protection duration conferred by dsRNA to 10 days on tomato and grape fruits and to 21 days on grape leaves. The results of this work demonstrate how AVs can be used as a new nanocarrier to overcome dsRNA instability in SIGS for crop protection.

arabidopsidis 16 , an animal fungal pathogen Beauveria bassiana of mosqiuito 17 , a bacterial pathogen 19 and even symbiotic microbes, such as ectomycorrhizal fungus Pisolithus microcarpus 20 and Rhizobium (Bradyrhizobium japonicum) 21 , also utilize host Argonaute proteins for crosskingdom RNAi. The biogenesis of these pathogen sRNAs involved in cross-kingdom RNAi are also largely dependent on pathogen DCL proteins [7][8]10,15,[22][23][24][25][26] . These studies demonstrate crosskingdom RNAi is a conserved mechanism across different species, including plant and animal hosts. Consequently, SIGS RNAs which target genes involved in RNAi machinery and in sRNA biogenesis have proven to be quite effective [7][8]10,23,[25][26][27][28] , making them ideal biological pathways to target in other organisms. Furthermore, because RNAi can tolerate multiple mismatches between sRNAs and target RNAs 29 , fungal pathogens are less likely to develop resistance to SIGS RNAs than to traditional fungicides, which mostly bind to and inhibit a specific enzyme or protein. To date, SIGS has effectively been used to control a wide range of insect pests 30-33 , viruses 34-35 and pathogenic fungi including Fusarium graminearum infection in barley 9,10 , and gray mold disease on fruits, vegetables and flowers 7,11,[13][14]36 .
One major drawback of SIGS is the relative instability of RNA in the environment, particularly when subjected to rainfall, high humidity, or UV light 33 . Thus, improving environmental RNA stability is critical for successful SIGS applications. One strategy is to dock RNAs in synthetic inorganic materials. Specifically, dsRNAs targeting plant viruses have been loaded into layered double hydroxide (LDH) nanosheets to protect dsRNA from nuclease degradation 34-35 and increase the stability and the durability of the RNAi effect. Ultimately, this can provide RNAi-based systemic protection against several plant viruses for at least 20 days after topical application 34,37 . This nanotechnology was recently proven to be effective for plant protection against fungal pathogens 27 .
In nature, plants and animals encapsulate RNAs in extracellular vesicles (EVs) for safe . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted January 6, 2023. In this study, we demonstrate that dsRNAs packaged in AVs can be successfully utilized in crop protection strategies. Three types of AVs were synthesized and found to confer protection to loaded dsRNA, which remained detectable in large amounts on plant surfaces over a long period of time. When applied to plants, AV-dsRNA can extend the length of fungal control conferred by fungal gene-targeting dsRNA to crops. Overall, this work demonstrates how organic nanoparticles can be utilized to strengthen SIGS-based crop protection strategies.
Furthermore, since pathogens have been shown to hijack host Argonaute proteins using sRNA effectors for cross-kingdom RNAi in multiple systems, , Dicer-like proteins are strong candidates for inhibiting pathogenicity and virulence. Exogenous treatment of Bc-DCL1/2-dsRNA, a dsRNA integrating fragments of the Dicer-like 1 (252 bp) and Dicer-like 2 (238bp) sequences from Botrytis cinerea, on the plant leaf surface can efficiently inhibit fungal disease 7 . When loading AVs, the loading ratio is determined by the N:P ratio, where N = # of positively charged polymer nitrogen groups and P = # of negatively charged nucleic acid phosphate groups. Here, DOTAP has one positively charged nitrogen per molecule, whereas the double stranded Bc-DCL1/2 RNA, which is 490 nucleotides in length, has 980 negatively charged nucleic acid groups per molecule. Thus, loading ratios between the AVs and the Bc-DCL1/2-dsRNA, from 1:1 to 4:1, were examined to identify the minimum amount of AVs required to bind all the dsRNA present in the solution. We concluded that a 4:1 (AV:dsRNA) ratio was the optimal ratio needed for dsRNA loading as Bc-DCL1/2-dsRNA loaded into AVs at this ratio could not migrate from the loading well due to complete association with the AVs ( Figure 1A). The average size, polydispersity index (PDI), and zeta potential (ZP) of the loaded AVs was also determined. The AV-Bc-DCL1/2-dsRNA lipoplexes had an average size of 365.3 nm with PDI = 0.455 and ZP = +47.17 mV (Table S1).
The ability of the AVs to prevent nuclease degradation was then validated under different enzymatic hydrolysis conditions. Naked and AV-loaded Bc-DCL1/2-dsRNA were both treated with Micrococcal Nuclease (MNase). As seen in Figure 1B, the naked-Bc-DCL1/2-dsRNA exhibited . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted January 6, 2023. ; https://doi.org/10.1101/2023.01.03.522662 doi: bioRxiv preprint complete degradation after MNase treatment as compared to the AV-protected Bc-DCL1/2-dsRNA released from the AV-Bc-DCL1/2-dsRNA after the MNase treatment using 1% Triton X-100 that had no obvious degradation. Thus, the AVs provide protection for dsRNA against nuclease degradation.
Finally, we assessed the ability of the AVs as an efficient vehicle for dsRNA delivery to B. cinerea fungal cells. Previously, we discovered that naked dsRNA is effectively taken up by B. cinerea 7 . Here, we compared fungal uptake of naked and AV-encapsulated Fluorescein-labeled dsRNA using confocal laser scanning microscopy (CLSM). To do this, we placed PDA media directly onto microscope slides, and inoculated with 4 L of 1x10 5 spores/mL Botrytis cinerea. 4 L of fluorescein-labeled dsRNA at a concentration of 80 ng/ L was then applied to the slides.
Fluorescent dsRNA was detected inside the fungal cells after application of either naked-or AV-Bc-DCL1/2-dsRNA to B. cinerea spores cultured on PDA plates ( Figure 1C). The CLSM analysis was carried out after MNase treatment to eliminate any fluorescent signals coming from dsRNA or AV-dsRNA not inside the fungal hyphae. Under these conditions, a strong fluorescent signal was found on and within the hyphaes after AV-dsRNA application, suggesting that the AV-dsRNA were taken up by the fungal cells ( Figure 1C).

External AV-dsRNA application triggers RNAi in B. cinerea
After demonstrating that the AVs could be loaded with dsRNA and taken up by fungal cells, we next examined if external AV-dsRNA application triggered RNAi in B. cinerea. Naked-and AV-dsRNA were externally applied to a variety of agriculturally relevant plant materials, including tomato and grape fruits, lettuce leaves and rose petals, and a reduction of B. cinerea virulence was observed ( Figure 2A). These plant materials were treated with 20 µL droplets of the dsRNA . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted January 6, 2023. ; https://doi.org/10.1101/2023.01.03.522662 doi: bioRxiv preprint treatments, at a concentration of 20 ng/µL. After RNA treatment, 10 µL droplets of fungal suspension (concentration varies and is dependent on different plant material, see methods) were added to the plant material. Two fungal-gene targeting dsRNA sequences were used, both of which have previously been successfully used in SIGS applications against B. cinerea 7,8 . One was the above-mentioned BcDCL1/2 sequence, for which we had previously reported that fungal disease Consequently, three dsRNAs were generated by in vitro transcription for loading into AVs: two of them specifically targeting B. cinerea virulence-related genes (Bc-DCL1/2 and Bc-VPS51+DCTN1+SAC1 (Bc-VDS)), while the third one was a non-specific target sequence (YFP) used as a negative control. All plant materials treated with naked-or AV-fungal gene targeting-dsRNA (Bc-DCL1/2 or -VDS) had obvious reduced disease symptoms in comparison to the water treatment and YFP-dsRNA controls (Figure 2A, 2B). The relative lesion sizes were reduced 75-90%. Further, both naked-and AV-Bc-VDS treatments decreased expression of the three targeted fungal virulence genes ( Figure S1). Taken together, these results demonstrate how externally applied AV -dsRNA can inhibit pathogen virulence by suppression of fungal target genes. As the SIGS efficacy of Bc-DCL1/2 and Bc-VDS dsRNAs are similar, we used only the Bc-VDS dsRNA for the subsequent analysis.
. CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted January 6, 2023. ; https://doi.org/10.1101/2023.01.03.522662 doi: bioRxiv preprint AV-dsRNA extends RNAi-mediated protection against gray mold disease due to increased dsRNA stability and durability The instability of naked dsRNA currently limits the practical applications of SIGS. Though we demonstrated that AVs can protect dsRNA from nuclease degradation, environmental variables can also influence RNA stability, such as leaf washing caused by rainfall events. Thus, we were interested in evaluating if using the AV-dsRNA would prolong and improve the durability of the RNAi effect on B. cinerea.
To assess the influence of washing on the stability and adherence of the AV-dsRNA to plant leaves, we analyzed the intact dsRNA content on the leaf surface using Fluorescein-labeled Bc-VDS-dsRNA and Northern blot analysis after water rinsing. The same concentration of Fluoresceinlabeled naked-or AV-Bc-VDS-dsRNA (20 ng/ L) was applied to the surface of Arabidopsis leaves.
After 24 h of incubation, the treated leaves were rinsed twice with water by vigorous pipetting.
Immediately after, we found that the naked-dsRNA treated leaves showed a drastic decrease in fluorescence compared with AV-dsRNA treated leaves ( Figure 3A). These results suggest that most of the naked-dsRNA was washed off, whereas the AV-dsRNA largely remained on the leaves after rinsing ( Figure 3A). The effect of the AVs on dsRNA stability over time was also assessed.
To do this, we treated 4-week-old Arabidopsis plants with Fluorescein-labeled dsRNA or AV-dsRNA, then incubated the plants in dark conditions for 1 or 10 days. We observed a strong fluorescence signal after 10 days on Arabidopsis leaves that were treated with Fluorescein-labeled AV-dsRNA, indicating that AVs confer stability to dsRNA ( Figure 3B). By contrast, the naked-dsRNA application showed an undetectable fluorescent signal ( Figure 3B) and a weak hybridization signal on the Northern blot analysis, compared to AV-Bc-VDS-dsRNA treated leaves, which retained Bc-VDS-dsRNA ( Figure 3C). We further examined whether the AV-dsRNA remained biologically active over time and prolonged protection against B. cinerea compared to . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted January 6, 2023. ; https://doi.org/10.1101/2023.01.03.522662 doi: bioRxiv preprint naked dsRNA. To this end, Arabidopsis leaves were inoculated with B. cinerea 1-, 3-, and 10-days post RNA treatment (dpt). Both naked and AV-Bc-VDS-dsRNA treatments led to a clear reduction in lesion size over the time points assessed ( Figure 3D). However, the efficacy of the naked-VDS-dsRNA was reduced at a much faster rate than that of the AV-VDS-dsRNA, demonstrating that AVs can enhance the longevity of the RNAi effect of the loaded dsRNAs ( Figure 3E).
To examine if AV-dsRNAs could be similarly effective on economically important crops, we repeated these experiments using tomato fruits, grape berries and grape (V. vinifera) leaves. We applied naked-or AV-Bc-VDS-dsRNA on the surface of tomato and grape berries and on the surface of grape leaves. The dsRNA was applied via drop inoculation on the fruits, and via spray application on the grape leaves, always at a concentration of 20 ng/ L. Both the naked and AV-Bc-VDS-dsRNA applications led to reduced disease symptoms on tomato and grape berries at 1, 5 and 10 dpt, as well as on detached grape leaves at 1, 7, 14 and 21 dpt, compared to the water or empty AV treatments ( Figure 4A). As we had observed in the Arabidopsis interactions, the AV-Bc-VDS-dsRNA applications greatly prolonged and improved the RNAi activity as compared to the naked-dsRNA over time for all plant materials ( Figure 4B). While the naked treatment lost the majority of its efficacy at 5-dpt in tomato fruits, 10-dpt in grape berries, and 21-dpt in grape leaves, the AV-dsRNA treatments reduced lesion sizes by at least 75% across all timepoints and plant material tested ( Figure 4B). These trends were also reflected in experiments on rose petals after the nakedand AV-Bc-VDS-dsRNA treatments ( Figure S2). The enhanced reduction in lesion size observed specifically at the longer time points (i.e. 5, 10, 14, and 21 dpt) after AV-Bc-VDS-dsRNA application clearly demonstrates how AVs protect loaded dsRNA from degradation to extend the duration of plant protection against B. cinerea. Together, these results strongly support the ability of AVs to confer higher RNAi activity over time, effectively enhancing dsRNA stability for SIGS applications.
. CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted January 6, 2023. ; https://doi.org/10.1101/2023.01.03.522662 doi: bioRxiv preprint

Cost-effective AV formulations also provide strong RNAi activity
Our discovery that AVs can lengthen dsRNA mediated plant protection opens the door for its practical use in agricultural applications. Cost is a critical consideration for any crop protection strategy, so we next tested if more cost-effective AV formulations could be used for dsRNA delivery and RNAi activity. First, we removed the PEG, an expensive reagent in the formula, from our original DOTAP+PEG formulation, resulting in DOTAP AVs composed only of DOTAP and cholesterol in a 2:1 ratio. Additionally, we used a cheaper cationic lipid, 1,2-dioleyloxy-3dimethylaminopropane (DODMA), in a 2:1 ratio with cholesterol to form DODMA AVs.
DODMA has previously been utilized in drug delivery formulations, but has a tertiary amine and is an ionizable lipid compared to DOTAP, which could result in changes in RNA loading and activity. The DOTAP AVs were fully loaded with Bc-VDS dsRNA at a 1:1 N:P ratio ( Figure 5A Figure 5C). The size distribution data for each AV formulation can be found in Table S1. As expected, the z-average sizes of the DOTAP-derived AVs are similar, while the use of DODMA increases the z-average size (Table S1).
Next, we examined if the different AV formulations influenced fungal dsRNA uptake or RNAi activity. After application of the different AV formulations, the fungal dsRNA uptake was tracked over 16 hours using CLSM. After 16 hours, all three AV formulations showed a similar amount of fungal RNA uptake, however, the uptake of DOTAP+PEG and DODMA AVs was slightly faster . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted January 6, 2023. All AV-VDS-dsRNA treatments were also able to reduce expression of the targeted B. cinerea genes at all time points ( Figure S4). Further, no phytotoxicity on Arabidopsis plants was observed for any AV formulation, indicating that they can likely be used to protect plants in addition to fruits ( Figure S5). Overall, these experiments demonstrate how new AV formulations that are more economical, but equally as effective, can be developed.

Discussion
Liposomes have been extensively researched for their applications in clinical contexts 47 , in fact, they have been utilized for drug delivery to human fungal pathogens 51-52 and are able to transit across the fungal cell wall 53 . Here, we provide the first demonstration that lipid-based nanovesicles can also be used in agricultural contexts, to deliver dsRNA to plant pathogens. The primary advantage that AV-dsRNA offers for SIGS over naked dsRNA is increased dsRNA stability. This is crucial for extending the shelf-life of dsRNA products, since extracellular RNases and other ribonucleases have been identified on fruits and the leaves of important economic crops such as tomato or tobacco 54-55 , and for increasing the length of time needed between RNA applications. In . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted January 6, 2023. ; https://doi.org/10.1101/2023.01.03.522662 doi: bioRxiv preprint fact, utilizing AV-dsRNAs could extend necessary treatment intervals up to a few weeks, as we demonstrated on both grapes and tomatoes (Figure 4), making SIGS a much more agriculturally feasible crop protection strategy. This is similar to the extended protection provided by inorganic dsRNA complex formulations against viruses on Nicotiana tabacum cv. Xanthi leaves and fungal pathogens on tomato plants 27,34 . Another key advantage of utilizing liposome technology for crop protection, especially post-harvest products, is that the success of similar formulations in clinical applications 47 suggests that the AVs will be safe for human consumption.
With agricultural applications in mind, we tested two more cost effective AV formulations.
By removing the PEG from DOTAP-AVs, we can reduce the cost of AV synthesis. PEG is used in liposome preparations in clinical contexts to protect liposomes from immune cell recognition and prolonged circulation time 56 , however, this is not a concern in agricultural applications.
Additionally, in our DODMA formulation, we used the lipid DODMA in place of DOTAP, which can further reduce costs. Surprisingly, our DOTAP formulation was able to load dsRNA at a 1:1 N:P ratio, in comparison to a 4:1 N:P ratio observed in other formulations. At this lower loading ratio, the cost of DOTAP AV formulations can be even further reduced. The decreased costs of the DODMA and DOTAP AVs potentially make these formulations more suitable for agricultural use.
In summary, we have provided the first example of utilizing a lipid-based nanoparticle, AVs, for the delivery of dsRNAs in SIGS applications. The AV organic formulations used here confer protection to dsRNA that results in an effective and more durable RNAi effect against the fungal pathogen B. cinerea in a wide range of plant products, overcoming the main limitation of SIGS to date. This is one key step forward in the development of RNAi-based fungicides which will help reduce the volume of chemical fungicides sprayed on fields and offer a sustainable option to limit the impact of fungal pathogens on crop production and food security.    (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted January 6, 2023. ; https://doi.org/10.1101/2023.01.03.522662 doi: bioRxiv preprint of 10 samples, and three technical repeats were conducted for relative lesion sizes. Statistical significance (Student's t-test) compared to water: *, P < 0.05.  Statistical significance (Student's t-test) relative to water: *, P < 0.05.  Error bars indicate the SD. Statistical significance (Student's t-test) relative to water: *, P < 0.05.
(C) Relative fungal biomass was quantified by qPCR. Fungal RNA relative to tomato RNA was measured by assaying the fungal actin gene and the tomato tubulin gene by qPCR using RNA extracted from the infected fruits at 5 dpi. Statistical significance (Student's t-test) relative to water: *, P < 0.05; **, P < 0.01.
. CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted January 6, 2023.

Synthesis and Characterization of Artificial Vesicles
. CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted January 6, 2023. ; https://doi.org/10.1101/2023.01.03.522662 doi: bioRxiv preprint PEGylated artificial vesicles were prepared following previously established protocols 49 . In brief, PEGylated artificial vesicles were prepared by mixing 260 μl of 5% dextrose-RNase free dH2O with the lipid mix and re-hydrating overnight on a rocker at 4°C. The re-hydrated lipid mix was then diluted 4-fold and extruded 11 times using a Mini-Extruder with a 0.4 μm membrane (https://avantilipids.com/divisions/equipment-products/mini-extruder-extrusion-technique). respectively. All lipids were sourced from Avanti Polar Lipids. The average particle size and zeta potential of the artificial vesicles was determined using dynamic light scattering. All measurements were conducted at 25°C using a Zetasizer Advance instrument (Malvern Instruments Ltd, Malvern, Worcestershire, UK) after 10-fold dilution in filtered Milli-Q water. Data reported is the average of three independent measurements.

In Vitro Synthesis of dsRNA
In vitro synthesis of dsRNA was based on established protocols 7 . Following the MEGAscript® RNAi Kit instructions (Life Technologies, Carlsbad, CA), the T7 promoter sequence was introduced into both 5' and 3' ends of the RNAi fragments by PCR, respectively. After purification, the DNA fragments containing T7 promoters at both ends were used for in vitro transcription.
Primers for each DNA fragment can be found in Table S2.
. CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted January 6, 2023.

External Application of RNAs on the Surface of Plant Materials
All RNAs were adjusted to a final concentration of 20 ng μl -1 with RNase-free water before use. 20 μl of RNA (20 ng μl -1 ) were used for drop treatment onto the surface of plant materials, or, approximately 1 mL was sprayed onto grape leaves before inoculation with B. cinerea.

Stability of dsRNAs Bound to AVs
The potential environmental degradation of dsRNA was investigated by exposure of naked-Bc VPS51+DCTN+SAC1-dsRNA (200 ng) and AV-Bc-VDS-dsRNA (200 ng/2.5 µg) to Micrococcal Nuclease enzyme (MNase) (Thermo Fisher) treatment in four replicate experiments. Samples were treated with 0.2 U μL -1 MNase for 10 min at 37 °C, and dsRNAs were released using 1% Triton X-100. All samples were visualized on a 2% agarose gel. The persistence of sprayed naked-Bc VDS-dsRNAs and AV-Bc-VDS-dsRNAs (4:1) on leaves was assessed in two replicate experiments by extracting total RNA from leaves followed by a northern blot assay with probes specific to the . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted January 6, 2023. ; https://doi.org/10.1101/2023.01.03.522662 doi: bioRxiv preprint Bc-VDS-dsRNA. 4-week-old Arabidopsis plants were treated at day 0 with either a 20 μl drop of Bc-VPS51+DCTN1+SAC1-dsRNAs (20 ng µl -1 ) or AV-Bc-VDS-dsRNAs (400:100 ng µl -1 ) and maintained under greenhouse conditions. Single leaf samples were collected at 1, 3, 7, and 10 dpt.
Total RNA was extracted using TRIzol and subjected to northern blot analysis. Expression of targeted genes was also measured via qRT-PCR, using the same protocol as described above. Expression levels in both gene expression and relative biomass assays were determined via the 2 -ΔΔCt method 59 .

Total RNA extraction and qRT-PCR
. CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted January 6, 2023. ; https://doi.org/10.1101/2023.01.03.522662 doi: bioRxiv preprint . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted January 6, 2023. ; https://doi.org/10.1101/2023.01.03.522662 doi: bioRxiv preprint  (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted January 6, 2023. ; https://doi.org/10.1101/2023.01.03.522662 doi: bioRxiv preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted January 6, 2023.  . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted January 6, 2023. ; https://doi.org/10.1101/2023.01.03.522662 doi: bioRxiv preprint