Plant hairy roots enable high throughput identification of antimicrobials against Candidatus Liberibacter spp.

A major bottleneck in identifying therapies to control citrus greening and other devastating plant diseases caused by fastidious pathogens is our inability to culture the pathogens in defined media or axenic cultures. As such, conventional approaches for antimicrobial evaluation (genetic or chemical) rely on time-consuming, low-throughput and inherently variable whole-plant assays. Here, we report that plant hairy roots support the growth of fastidious pathogens like Candidatus Liberibacter spp., the presumptive causal agents of citrus greening, potato zebra chip and tomato vein greening diseases. Importantly, we leverage the microbial hairy roots for rapid, reproducible efficacy screening of multiple therapies. We identify six antimicrobial peptides, two plant immune regulators and eight chemicals which inhibit Candidatus Liberibacter spp. in plant tissues. The antimicrobials, either singly or in combination, can be used as near- and long-term therapies to control citrus greening, potato zebra chip and tomato vein greening diseases.

to present their system as a culturing system. In my opinion the advantage of their system is the high throughput capabilities for checking antimicrobioal compounds. But there is no culturing bacteria here at all. The authors can find many examples of infected plants with these two bacteria, and even there are a few manuscripts where antimicrobial compounds are tested in planta (as examples: Scientific Reports volume 8, Article number: 17288 (2018), https://doi.org/10.1371/journal.pone.0111032). The mentioned papers have the advantage of using grafting and testing the compounds actually in plants, and not only in roots. I see that the advantage and main contributions of the current Irigoyen manuscript is basically in the high throughput capabilities of their system. I suggest changing the title and overall angle of this manuscript to represent what it is about: a high throughput antimicrobial screening system. 1) As mentioned above, the great advantage of this system being high throughput, but has a disadvantage with other system in the fact that these compounds are only tested in roots, and is not clear at all how this will translate when these compounds are applied to fully developed plants in the leaves. One of the highest challenges for these Liberibacter pathogens is the fact that they are limited to the phloem, a niche hard to reach by spraying compounds. 2) Besides the technical advantage of this process presented here, authors should have shown as a proof of principle that at least one of the chemicals they discovered to be antagonistic, actually works in fully developed plants when added (even if only under controlled conditions). In other words, the new systems needs to be validated to have a big impact in the scientific community. Without validation, the impact of this research will be limited. I understand this will take time, but it should be doable to test a few of the microbial compounds in planta in the greenhouse (especially in the potato or tomato system that is faster than citrus) 3) I was surprised for the lack of discussion on root infection by CLas and CLso. There are several papers published by E. Johnson (UF) and other groups in Brazil that studied the root infection of citrus by CLas. This manuscript is exceptionally well written and describes a marked breakthrough in testing therapeutic approaches for non-culturable plant pathogens. My comments are mainly quite minor, mainly dealing with typos, but I suggest that they merit the authors attention. All my comments below are also highlighted in the text and have associated comment balloons.
The biggest issue is the inconsistency and errors that appear to be abundant in the References. I only read through the first 20, and 6 of them had irregularities. The entire References section should be carefully edited. I did not verify linkage of References and numbers in the manuscript, but suggest this also be carefully reviewed.
Line 41: t in tuberosum not italicized Line 48: need to indicate that HLB is haunglongbing and a synonym for citrus greening Fig 1: last sentence has apparent redundancy Line 133: italicize Arabidopsis? Line 180: Likely should be "diverse" rather than "diversity" Line 210: "Academy" not "academy" Line 212: "for" would be standard, not "to" Line 234: "gene knockout" would be more accurate Line 303: Linnaeus did not describe grapefruit. I THINK this should be Macfad. Line 337: g/molecule? as written could also be g/ mole Line 370: rephrase. Maybe "all the most-effective compounds" Someone is going to read this as "all the Pb compounds" 1. First, we measured the expression of three defense-related marker genes (the pathogenesisrelated, or PR, genes PR-1 like, PR-3 like and WRKY6-like) in the SlNPR1-and AtNPR1expressing potato hairy roots, along with empty vector controls in both healthy and CLsoinfected conditions. PR gene expression was significantly higher in both SlNPR1-and AtNPR1-expressing hairy roots in healthy conditions, when compared to empty vector controls (Fig. 2E). The induction of PR genes was greatly amplified upon CLso infection in both SlNPR1-and AtNPR1-expressing hairy roots, when compared to vector-alone controls ( Fig. 2E). Together, these results suggest that SlNPR1 and AtNPR1 function as transcriptional activators of PR genes in potato to mediate defense responses against 'Candidatus Liberibacter spp.' 2. Next, we measured the levels of salicylic acid (SA). Interestingly, despite the activation of PR and WRKY defense-related genes, SA levels were significantly lower in SlNPR1-and AtNPR1-expressing hairy roots, in both healthy and CLso-infected conditions, when compared to empty vector controls (Fig. 2F). We propose three scenarios that could explain the lower SA levels in SlNPR1-and AtNPR1-expressing potato hairy roots: a. SA accumulation in NPR1 overexpressors is directly (positively) associated with the levels of CLso, i.e., lower CLso, lower SA. b. SA-mediated signaling is far more potent in the NPR1 overexpressors relative to controls, and thus less SA is needed to mediate the defense responses. This hypothesis posits that in empty vector controls, NPR1 concentration/activity is less than optimal for triggering SA-mediated defenses, and so more SA is produced to compensate for relatively lower amounts of NPR1. c. A negative feedback loop in the NPR1 overexpressors could suppress SA levels in order to maintain defense homeostasis. There is some evidence to support this scenario. Although we did not find previous studies that determined SA  We thank the reviewer for suggesting these experiments, which led to interesting insights into the mechanism whereby NPR1 inhibits 'Candidatus Liberibacter sp.' and expanded our understanding of the utility of the microbial hairy root system for conducting fundamental studies. We have added these results and discussion to the revised manuscript. Fig. 2E, the appropriate control should be R. rhizogenes containing the vector without the NPR1 gene to exclude the putative inhibitory effect from R. rhizogenes or the empty vector.

In addition, for
In all genetic experiments, our controls were indeed hairy roots transformed with the empty vector without a target gene. We clarified this further in the figure legend and the methods.

For Figs. 2C-D, the appropriate control should be R. rhizogenes containing the vector without the coding sequences for antimicrobial peptides to exclude the putative inhibitory effect from R. rhizogenes or the empty vector. In addition, roots can easily absorb antimicrobial peptides. The authors need to test the inhibitory effect dipping the hairy roots into solutions containing antimicrobial peptides.
In all genetic experiments, our controls were hairy roots transformed with the empty vector without a target gene. We clarified this further in the figure legend and the methods.
We thank the reviewer for leading us toward a new approach to deliver peptides in the microbial hairy root cultures. Unlike small molecules, peptide uptake/absorption into intact tissues does not occur readily when roots are dipped in a solution. There are several constraints on this process, such as the physical barriers of cell walls, the hydrophobicity of the cell membrane, and size-exclusion limits, as well as significant proteolytic activity in the apoplast/cell membrane 13 However, we explored other ways to deliver smaller peptides into root tissues. For this, we performed vacuum infiltration, in a manner similar to delivery of small molecules in vitro (Fig. 4A). As proof of concept, we selected two peptides corresponding to AMP2 and AMP5 and evaluated their use in CLas-citrus hairy root cultures. Both the peptides showed good efficacy in inhibiting CLas and CLso when overexpressed in the microbial hairy roots via T-DNA vectors (Fig. 2G). Briefly, the two peptides were produced recombinantly, and vacuum-infiltrated into the hairy roots at 5 and 10 µg/ml concentration. After 72 h, molecular diagnostics was performed to determine levels of CLas. Both peptides showed statistically significant (P ≤ 0.05) dose-dependent inhibition of CLas, and the results parallel with those of the genetic-based overexpression assays (Fig. 2G). Although this is yet another useful, and probably faster, approach to test AMPs, it would be challenging to deliver proteins as large as NPR1 or protein complexes such as that required for CRISPR-Cas9. Another caveat to the use of direct protein delivery is the need for appropriate folding/post-translational modifications/native activity of the peptides/proteins when produced by a synthetic or recombinant route. In these situations, genetic-based expression/delivery into plant tissues would be appropriate. Nevertheless, we plan to further explore the upper size limits of proteins that can be delivered by this approach in new studies and thank the reviewer again for leading us into this area of investigation. This extended the in vitro assay system for highthroughput screening of AMPs, in a manner similar to that for small molecules. We have added these new results to the revised manuscript (Fig. S5). 6. For the genome editing experiments, the authors seems to generate some chimeric or low efficacy mutations. To make the mutations work in disrupting gene function, the mutation should be biallelic or homozygous. The wild type is more than 50%. It is not convincing that such low efficacy gene editing will have the intended effect. The authors need to test whether NPR3 gene expression is changed, its protein level is changed, whether it has the intended antagonistic effect in suppressing plant defense by testing the expression of immunity related genes.
The main reason for the observed chimeric/low rate (~50%) of mutations in the SlNPR3experiment (Fig. 3) is the endogenous copy number (ploidy). Potato is tetraploid (2n = 4x = 48). This reduces the chances of obtaining biallelic, homozygous mutations, particularly in transient transformation assays, such as with hairy roots in which each root is affected by an independent transformation event. For instance, in CRISPR experiments where we attempted to edit a singlecopy GFP transgene, we observed editing rates of ~86-100% (Fig. S3e,f), suggesting that the low rate of mutations seen in editing endogenous genes is inevitable in potato given its polyploidy. Nevertheless, from the perspective of biological significance, even knocking out ~50% of NPR3 in the hairy root population was sufficient to promote the systemic immune response against CLso, as indicated by the significant decrease in CLso titers in NPR3-edited hairy roots when compared to Cas9-alone vector controls (Fig. 3D). Please also note that the intent of this hairy-root-based assay is to expedite testing of loss-of-function of potential targets. If promising results are attained, stable CRISPR lines with the preferred mutations can be generated for further studies, aspects that are beyond the scope of this study. We added the CRISPR discussion points to the revised manuscript.
As for the NPR3 protein/activity, since we do not have antibodies against the endogenous potato NPR3, and to the best of our knowledge there are no other reports of anti-NPR3 antibodies in other systems, we measured the expression of downstream target marker genes (PR-1 like, PR-3 like), as well as WRKY6-like and NPR1, which are transcriptional co-activators in SAR responses. Expression of all four SAR markers was significantly higher in StNPR3 edited hairy roots than in Cas9 vector control (Fig. 3e), again suggesting that NPR3 activity is sufficiently impaired by the CRISPR editing. We have added these new data to the revised manuscript (Fig. 3e).
7. For this in vitro multi-well plat assay, the authors need to conduct minimum Inhibitory Concentration (MIC) and minimum Bactericidal Concentration (MBC) to generate convincing and meaningful results.
MBC and MIC are defined as the lowest concentration of an antibiotic/chemical that can kill or inhibit target bacteria to a point that there is no growth in culture or plates (>99.9% reduction) as determined visually or by optical density measurements. Unfortunately, it is not technically feasible to estimate MBC and MIC for fastidious pathogens like 'Candidatus Liberibacter spp.' 18 , even in the ex vivo hairy root cultures, since they are not axenic cultures. However, one can determine biologically active concentrations by doing dose-response assays in the hairy root system. To demonstrate this, we selected three compounds (#3, #8 and #9) that inhibited both CLso and CLas (Fig. 4) and conducted new dose-response assays with 0, 5, 10, 25 and 50 µM concentrations (Fig.  S6). Furthermore, using the dose-response results as a guide, we formulated dosages for subsequent in planta foliar spraying experiments. The results showed good consistency between the hairy root and in planta trials (Fig. 5). These new results are added to the revised manuscript in Fig. S6 and Fig. 5.

In the abstract, the authors mentioned a susceptibility gene. If it is referring to NPR3, please check the term susceptibility gene. NPR3 does not qualify as a susceptibility gene based on current nomenclature.
Good point. In the revised manuscript, we referred to this gene as a negative immune regulator.  propagating the viruses on demand, the process is referred to as 'culturing' or 'cultivation' 19, 20, 21,   22, 23, 24, 25, 26 . Furthermore, since the above strategies required host tissues removed from the host organism and experimented on or maintained in an external environment, they are classified as ex vivo (Latin: "out of the living") approaches.

• https://www.cdc.gov/coronavirus/2019-ncov/about/grows-virus-cell-culture.html • Enders JF, Weller TH, Robbins FC (1949) Cultivation of the Lansing strain of poliomyelitis virus in cultures of various human embryonic tissues. Science 109: 85-87. • Steinhardt E, Israeli C, Lambert RA (1913) Studies on the cultivation of the virus of vaccinia. The Journal of Infectious Diseases: 294-300. • Cox HR (1952) Growth of viruses and rickettsiae in the developing chick embryo. Annals of the New York Academy of Sciences 55: 236-247 • Litwin J (1957) A simple method for cultivation of viruses and rickettsiae in the chorioallantoic ectoderm of the chick embryo by inoculation via the air sac. The Journal of infectious diseases: 100-108 • McClelland L (1946) Simultaneous cultivation of typhus Rickettsiae and Influenza virus in the developing chick embryo. Proceedings of the Society for Experimental Biology and Medicine 63: 427-431. • Yoshino K (1967) One-day egg culture of animal viruses with special reference to the production of anti-rabies vaccine. Japanese Journal of Medical Science and Biology 20
: 111-125. • Pyrc, K., Sims, A.C., Dijkman, R., Jebbink, M., Long, C., Deming, D., Donaldson, E., Vabret, A

., Baric, R., and van der Hoek, L. (2010). Culturing the unculturable: human coronavirus HKU1 infects, replicates, and produces progeny virions in human ciliated airway epithelial cell cultures. J. Virol. 84, 11255-11263. • Pelzek AJ, Schuch R, Schmitz JE, Fischetti VA (2013) Isolation, culture, and characterization of bacteriophages. Current Protocols Essential Laboratory Techniques 7: 4.4. 1-4.4. 33
Inspired by the above studies, we formulated our hypothesis that fastidious bacteria such as CLso and CLas are conceptually akin to the obligate viruses, thus ex vivo plant (host) tissues would be suitable to culture them in the laboratory. Also, since CLso and CLas are vascular-limited pathogens, we hypothesized that hairy root matrices (with intact vasculature) would be ideal support for their growth. Note: Although we are using the infected plant tissues as source, the produced hairy root cultures are artificially induced by employing R. rhizogenes and maintained further in the laboratory. In the revised manuscript, to conform with the terminology of the classical ex vivo approaches to cultivate viruses, we replaced culturing with "ex vivo cultivation", and added the above rationale/hypothesis in our discussion.
1) As mentioned above, the great advantage of this system being high throughput, but has a disadvantage with other system in the fact that these compounds are only tested in roots, and is not clear at all how this will translate when these compounds are applied to fully developed plants in the leaves. One of the highest challenges for these Liberibacter pathogens is the fact that they are limited to the phloem, a niche hard to reach by spraying compounds.
We agree with the reviewer that the Liberibacter spp. are hard to reach in planta by foliar spraying, as they reside deep in the phloem tissues. As part of any drug-discovery pipeline, the leads will need to be further tested in planta, provided that the delivery systems are improved. The latter is indeed an active area of research, especially in the citrus-HLB community, with several groups evaluating alternative approaches to foliar spraying, such as trunk injections and nanoparticlebased systems 27,28 . Indeed, the Citrus Disease Sub-committee (CDS) of the National Agricultural Research, Education, Extension and Economics (NAREEE) Advisory Board made "delivery systems for therapeutics, nutrition and other HLB solutions" a #1 priority for the research community to tackle in the FY2020 Emergency Citrus Disease Research and Extension Program. Hopefully, the citrus community will soon find better ways to deliver the active ingredients into citrus trees.

https://nifa.usda.gov/sites/default/files/rfa/FY2020-RFPA-Emergency-Citrus-Preapplication.pdf
As such, the above in planta issues have no bearing on the hairy root bioassays. Conversely, the hairy root system overcomes the in-planta delivery problems in regard to the screening of new compounds, since the compounds and small peptides can be effectively vacuum infiltrated into hairy root tissues. We suggest that hairy root bioassays are thus an ideal pre-screening system for large-scale AI screening and drug discovery pipelines, to narrow down potential new leads before pursuing in planta experiments/trials.

2) Besides the technical advantage of this process presented here, authors should have shown as a proof of principle that at least one of the chemicals they discovered to be antagonistic, actually works in fully developed plants when added (even if only under controlled conditions).
In other words, the new systems needs to be validated to have a big impact in the scientific community. Without validation, the impact of this research will be limited. I understand this will take time, but it should be doable to test a few of the microbial compounds in planta in the greenhouse (especially in the potato or tomato system that is faster than citrus) We agree that it would be useful to compare the results of the hairy root bioassays in planta, but again bear in mind the constraints of in planta delivery as discussed above, especially for citrus. Note that we can already assess this by comparing the efficacy of tetracycline in the hairy root bioassays. Several published reports have established that tetracycline derivatives inhibit CLas in planta when delivered appropriately via trunk injections, but not by foliar spraying 27 . In the hairy root bioassays, we consistently observed that tetracycline significantly inhibited CLas and CLso in hairy roots, thus suggesting that the HR assay data parallels well to the in-planta studies.
Nevertheless, as suggested by the reviewer, we selected three new compounds that showed inhibitory activity against both CLso and CLas (#3, #8 and #9) in the hairy root assays (Fig. 4) and tested them in planta (in potatoes). The three compounds were applied to CLso-infected potatoes by foliar spraying twice a week, at two different dosages (10 µM and 25 µM), and disease symptoms were monitored periodically. Disease progression was monitored for 28 days post infection (dpi), by which point untreated plants showed typical foliar disease symptoms of chlorosis, necrosis, leaf curling and wilting, and were close to dying (Fig. 5a). By contrast, potatoes sprayed with any of the three molecules showed clear tolerance, in a dose-dependent manner, as plants sprayed with 25 µM showed the fewest disease symptoms, on par with those treated with tetracycline (Fig. 5a). The attenuated symptoms were associated with lowered CLso titers in the various treatments, when compared to untreated controls (Fig. 5B). Together, these experiments demonstrate that the new compounds inhibit 'Candidatus Liberibacter spp.' in planta, and substantiate the results obtained in the hairy root bioassays. We added the new data and results to the revised manuscript (Fig. 5). We added the missing citations to the revised manuscript.

5) Single quotation marks re needed every time they refer to 'Candidatus Liberibacter spp.'.
We edited this throughout the manuscript. Fig. 1 Good question. We had noticed that too. Given the documented broad-spectrum activities of these AMPs against other bacteria and/or fungi, it is very much possible that in these specific instances, the AMPs could be inhibiting other competitive microbes that are present in the CLso and CLas hairy root cultures.

Review of : Plant Hairy Roots Enable ex vivo Culturing 1 of Fastidious Pathogens and Identification of New Antimicrobials
This manuscript is exceptionally well written and describes a marked breakthrough in testing therapeutic approaches for non-culturable plant pathogens. My comments are mainly quite minor, mainly dealing with typos, but I suggest that they merit the authors attention. All my comments below are also highlighted in the text and have associated comment balloons.
Thank you for the positive feedback.
The biggest issue is the inconsistency and errors that appear to be abundant in the References. I only read through the first 20, and 6 of them had irregularities. The entire References section should be carefully edited. I did not verify linkage of References and numbers in the manuscript, but suggest this also be carefully reviewed. Line 180: Likely should be "diverse" rather than "diversity"