Whole genome analysis of spontaneous antimicrobial resistance in Liberibacter crescens suggests long-term efficacy for antimicrobial treatment of citrus greening disease

Currently, oxytetracycline and streptomycin are being applied to citrus groves in Florida for the control of citrus greening disease caused by the unculturable bacterium ‘ Candidatus Liberibacter asiaticus’. Here, the closest cultured relative, L. crescens , was used to estimate the frequency of spontaneous antimicrobial resistance of, as yet, uncultured ‘ Ca . Liberibacter asiaticus’ . Results yielded thirteen streptomycin and zero oxytetracycline mutants after exposing 13 billion cells to the antimicrobials. These low rates, alongside the restrictive habitats of the vector and pathogen, suggest resistance may develop very slowly, if at all. Thus, the treatments will likely remain useful long enough before pathogen-resistant or -tolerant genotypes are deployed.


Introduction
Only three antimicrobials are registered by the Environmental Protection Agency (EPA) for use in plant agriculture: oxytetracycline (OTC), streptomycin, and kasugamycin.Prior to 2016, streptomycin comprised 90% of the antimicrobials used, largely to control plant diseases on temperate fruit trees such as apples, pears, and peaches (Stockwell and Duffy 2012).Nowadays, the use of antimicrobials in plant agriculture has increased dramatically with streptomycin and OTC registered by the EPA and Food and Drug Administration (FDA) for use on citrus canker disease (FDA 2013).Recently, the EPA's emergency exemption for use of streptomycin sprays for citrus greening disease was vacated (9th Cir 2023).Meanwhile, under FIFRA Section 24(c), OTC trunk injections were approved for citrus in Florida in 2022 (Archer et al. 2023).Currently, field tests have shown OTC trunk injections are more effective than sprays on 'Candidatus Liberibacter asiaticus' (CLas)-infected trees because of the systemic nature of CLas.(Hu and Wang 2016;Hu et al. 2018;Hu et al. 2018;Li et al. 2019;Killiny et al. 2020;Archer et al. 2022;Vincent et al. 2022).In addition, these antibiotics are stable in citrus tissues for months after injection (Hu and Wang 2016;Al-Rimawi et al. 2019).
Low or moderate levels of streptomycin resistance (up to 200 µg/ml) are often conferred by lateral transfer of the strA-strB gene pair which inactivates streptomycin by phosphorylation catalyzed by an aminoglycoside phosphotransferase (Chiou and Jones 1995).High levels of resistance (up to 2000 µg/ml) are conferred by single base mutations in rpsL gene which codes for ribosomal protein S12, a constituent of the 30S ribosomal subunit (Chiou and Jones 1995;Finken et al. 1993).Less commonly observed mechanisms include 16S rRNA gene modifications and the lack of N-7 methylation in 16S rRNA at position 527 caused by a mutation in rsmG (Nishimura et al. 2007;Okamoto et al. 2007;Demirci et al. 2014).Since lateral gene transfer is unlikely in CLas given its restrictive habitat, and the lack of strA-strB in the CLas and L. crescens genomes, we focused on rpsL and 16S rRNA mutations as potential streptomycin resistance factors in CLas.
Among plant pathogens, streptomycin resistance has been studied mostly in Erwinia amylovora, the causative agent of fire blight disease on apple, pear, and other plants.Resistance in this pathogen has been reported in Michigan, New York, and California (Chiou and Jones 1995;Förster et al. 2015;McGhee et al. 2011;McGhee and Sundin 2012;Tancos et al. 2016;Tancos and Cox 2016) (Chiou and Jones 1995), while a separate report showed 32 of 34 resistant strains possessed the strA-strB gene pair and the remaining 2 rpsL mutations (Tancos et al. 2016).Consequently, rpsL mutations and presence of strA-strB were examined here.
Oxytetracycline carries a single hydroxyl group present on the B ring of the naphthacene core that is missing in tetracycline (Nguyen et al. 2014).Tetracycline class antibiotics prevent protein synthesis by the binding of aminoacyl-tRNAs to the A site of the 30S subunit of the ribosome (Schnappinger and Hillen 1996).As tetracycline and oxytetracycline share similar structure and mode of action, their activity is blocked by the same resistance mechanisms.The four mechanisms of tetracycline resistance include: ribosomal protection, efflux, degradation, and 16S rRNA mutations (Nguyen et al. 2014).
The objective of this work was to determine which of these resistance mechanisms, if any, are developed spontaneously in large populations of L. crescens cells by comparing the genomes of resistant mutants to wild-type sensitive strains.L. crescens was chosen because it is the closest cultured bacterial relative of phloem-limited uncultured CLas and very similar physiologically and genetically (Fagen et al., 2014;Leonard et al., 2012).These results, along with our knowledge of the etiology of the disease, anticipate a slow pace for the appearance of antimicrobial resistance in CLas.

Materials and Methods
L. crescens BT-1 was cultured in liquid BM7 medium at 28°C with shaking (125 rpm) as described previously (Fagen, Leonard, McCullough, et al., 2014) to an OD 600 of 0.3-0.4.Fifty mL of this culture was plated on 3 sets of 12 BM7 agar plates; each set was supplemented with either 2.5 mg/mL streptomycin, 0.25 mg/mL oxytetracycline, or 2.5 mg/mL streptomycin and 0.25 mg/mL oxytetracycline.These concentrations were chosen as they were the minimum concentrations required for inhibition of L. crescens.Serial dilutions of the same culture were done from 10 -4 to 10 -9 in liquid BM7 medium and 50 mL of each dilution were plated on BM7 agar in triplicates with and without the appropriate antimicrobial to determine the rate of spontaneous mutation.Plates were incubated at 28°C for about 25 days.Colonies resistant to the antimicrobials were selected and grown in liquid BM7 medium supplemented with the appropriate antibiotic and concentration for about seven days.The antimicrobial-resistant bacterial cultures were stored in 25% glycerol at 80°C.
For whole genome sequencing, the wild-type strain (M1) and the mutant strains (A2, A3, K1, K2, K3, K4, and K5) were cultured from the frozen glycerol stocks on BM7 plates supplemented with 2.5 mg/mL streptomycin.After incubating for about fourteen days at 28°C, DNA from these subcultures was extracted using the E.Z.N.A.® Bacterial DNA Kit (D3350) and sent out for PacBio RS II whole genome shotgun sequencing, as described previously (Leonard et al., 2012), using the PacBio platform with the P6-C4 chemistry (Pacific Biosciences, Menlo, CA, USA).A single SMRT cell was used for each genome and the reads obtained were assembled to generate a single contig using the HGAP_Assembly.3 protocol on the PacBio SMRT Analysis server.
Once rpsL was shown to be the only gene with mutations in seven mutants through whole genome sequencing, the remaining five streptomycin resistant mutants (A5, A6, A7, A8, and A9) were subjected to polymerase chain reaction.
The L. crescens rpsL gene was amplified using the primer set (F246: 5'GTGTGGTGGTCGTGTAAAAGA3') and (R363:5'ACGCTTTGCCCCATACTTGG3') targeting a region spanning the 43 rd and 88 th codons.The amplification conditions were set as follows: initial denaturation at 95°C for 2min, 30 cycles of 95°C for 20s, 55°C for 30s, and 72°C for 30s, and final elongation step of 72°C for 30s.The amplicons were submitted for Sanger sequencing through Azenta Life Sciences.

Results
Twelve streptomycin resistant L. crescens colonies were isolated after incubation and the frequency of spontaneous resistance was one in 500 million cells or 2 x 10 -9 , substantially lower than the expected 1 x 10 -8 typically found in resistant bacteria from a single base mutation (Martinez & Baquero, 2000).No oxytetracycline resistant colonies were recovered despite exposing over 13 billion cells to oxytetracycline, yielding a frequency of spontaneous resistance lower than one in 1.3 x 10 -10 cells.
The full genome was sequenced for seven of the twelve streptomycin-resistant spontaneous mutants.Each of the seven mutant genomes was assembled into a single contig and analyzed for the presence of single nucleotide polymorphisms in the rpsL, 16S rRNA, and rsmG genes from L. crescens.Only rpsL mutations were observed in the streptomycin resistant mutants.Each mutant possesses one of these three single base mutations in L. crescens rpsL: A128G, G129T, or A263G which result in codon mutations K43R, K43N, and K88R respectively.Amino acid change K43R was present in genomes A1, K3, and K4 and K88R in genomes A3, K1, K2, and K5; these represent the two most common forms of streptomycin resistance in Escherichia coli (Pelchovich et al., 2014).Any rpsL gene mutations in the remaining five strains not subjected to WGS were identified by PCR and Sanger sequencing of the rpsL gene.These resulted in rpsL codon changes K43R in A5 and A6 mutants or K88R in A7, A8, and A9 mutants (Table 1).
In field studies, the minimum OTC concentration to suppress CLas in field citrus was 0.86 μg/g and the minimum inhibitory concentration against L. crescens in artificial media was 0.01 μg/mL (Li et al., 2019).Within the recommended range, at 0.72g of OTC per tree injections, the levels of OTC in the leaves were found to be 7 μg/g FWT (Vincent et al., 2022).Here, L. crescens was inoculated in BM7 media supplied with 0.25 μg/mL of OTC for which no cells were recovered.All taken together, the frequency of spontaneous resistance to OTC applications by CLas is expected to be insignificant.
The frequency of spontaneous resistance was very low for streptomycin in L. crescens.Single nucleotide changes in the rpsL gene were found in all streptomycin resistant mutants; this is not surprising given the relatively small L. crescens genome does not contain the classical streptomycin resistance genes, including strA/B.rpsL genes of L. crescens and CLas have 84% sequence homology, and since rpsL mutations confer streptomycin resistance in a diverse group of bacteria, it's very likely it does so in CLas as well.Yet, spontaneous resistance via rpsL mutations in CLas is expected to be negligible.
Certain mutations within the 16S rRNA gene can lead to a modest level of resistance to oxytetracycline and streptomycin; however, in L. crescens all three of its copies would need to be mutated to observe any robust resistance.Several of the appropriate mutations within a single copy of 16S rRNA are additive towards increased tetracycline resistance (Nguyen et al., 2014); but, this is not expected to provide protection at the level of selective pressure expected in the field.One can argue that once one of these alleles is mutated, resistance can be conferred to the other two by additional mutation or by homologous recombination; however, the slow growth rate of Liberibacter and multiple 16S rRNA gene copies likely mean a very slow appearance of resistance by this mechanism.Accordingly, it is not surprising that none of the L. crescens mutants isolated here contained 16S rRNA mutations predictive of resistance.
Antibiotic resistance in CLas in citrus groves is unlikely.
First, CLas has a predicted slow growth ratecomparable to L. crescens (Cruz-Munoz et al., 2018;Fagen, Leonard, Coyle, et al., 2014) meaning it will take a very long time for spontaneous resistance to dominate in the population compared to a typical bacterium.Second, CLas small genome size of 1.2 Mb shows a lack of genes commonly used by other bacteria for resistance, such as ribosome protection, tetracycline efflux and degradation, or the strA/B gene pair.And third, it is unlikely that CLas will acquire antimicrobial resistance genes through lateral transfer because of its very restrictive habitats; the citrus phloem, where only CLas has been found in infected trees (Tyler et al., 2009), and the Asian citrus psyllid gut, where only four bacteriawith reduced genomes and lack of antimicrobial resistance (AMR) genesreside (Fagen et al., 2012;Petrone et al., 2022).This leaves spontaneous antimicrobial resistance as the only means by which Liberibacter could obtain resistance.However, CLas lacks nine out of ten factors that contribute to increased frequency of AMR in bacteria reviewed in Martinez & Baquero, (2000). of living plant tissue and 10 kg of living tissue per tree, there are 10 million CLas cells per tree.This is 50-fold below L. crescens cells required (1 in 500 million) to isolate one streptomycin-resistant colony in the lab.Thus, if spontaneous resistance occurs at the same frequency in CLas, any set of 50 heavily infected trees would harbor one spontaneously resistant CLas mutant before antimicrobial selective pressure begins.The CLasresistant cell number is expected to increase with selective pressure; however, the pace a cell can spread in a set of 50 trees through psyllid feeding is likely very slow, especially in groves with psyllid control.The frequency of spontaneous resistance to oxytetracycline is 1 in 13 million cells, meaning it would take more than 1300 infected trees to observe one resistant CLas cell.Nevertheless, with sufficient antibiotic exposure, significant resistance in the field will likely occur but we expect economically important levels of resistance will occur much faster with streptomycin compared to oxytetracycline.As antibiotic resistance in an uncultured pathogen is difficult to measure directly, the easiest means to assess resistance may be qPCR measurements of the pathogen as signs of decreased antibiotic efficacy.

Initial levels of
(McGhee et al. 2011)in resistance was conferred by strA-strB genes; the remaining 1.3% was conferred by a single base mutation in rpsL(McGhee et al. 2011).In contrast, other reports highlight 102 strains of streptomycin-resistant E. amylovora possessed mutations in rpsL