During the last decade, compelling models of plant–pathogen interactions have been established, mostly based on molecular studies of mesophyll cell-infecting pathogens (Dodds and Rathjen, 2010; Asai and Shirasu, 2015; Ceulemans et al., 2021). Local immune activation involves the activity of two classes of receptors: plasma membrane-localized receptors that typically recognize conserved microbial patterns in the apoplast, named as pathogen-associated-molecular-patterns (PAMPs). PAMPs detection constitutes the first tier of active plant immunity, which involves an influx of extracellular Ca₂ ⁺ into the cytosol, an oxidative burst, and a massive reprogramming of the transcriptome in the first hours after PAMP recognition. Some pathogens overcome this PAMP-triggered immunity (PTI) by secreting effectors that manipulate cellular processes and suppress PTI. In resistant plants, targeted recognition of these effectors by cytoplasmic receptors forms the second tier of active plant immunity, called effector-triggered immunity (ETI). Both responses are triggered very quickly, ensuring, for many interactions, the complete deployment of defenses withing the first days after pathogen detection. The ability of a pathogen to overcome these two defensive lines within this time frame determines its success in establishing the infection (Thomma et al., 2011; Bigeard et al., 2015; Remick et al., 2023).

Some of the most devastating plant pathogens, however, do not infect mesophyll cells. They are delivered into the sieve cells of the phloem by feeding insects using specialized mouth parts called stylets, which pierce the plant tissue and directly access the nutrient-rich phloem vessels. The stylets travel through the apoplast to reach the phloem, secreting saliva to form a hard sheath that seals off plant cell leaks caused by the penetration process, causing limited wounding to the plant. Thus, those pathogens are directly delivered to the sieve cell cytoplasm, which lacks a nucleus, and its potential to mount an effective immune response is uncertain (Jiang et al., 2019; Huang et al., 2020). In addition, the high genome reduction rate of these pathogens, which makes them very dependent on host metabolism for survival, implies a lack of typical transmembrane systems for secretion of effectors (Duan et al., 2009; Oshima et al., 2013; Huang et al., 2020; Rattner et al., 2021). Under this scenario, which aspects of the aforementioned plant immunity are applicable to these interactions remain unknown.

One of the most ravaging diseases caused by phloem-inhabiting prokaryotic pathogens is Huanglongbing (HLB), also known as citrus greening. Since 2004 and 2005, when it was respectively detected in São Paulo (Brazil) and in Florida (USA), two of the most significant citrus-growing regions worldwide, it has caused billions of dollars in losses. In Brazil, despite successful management strategies adopted in the last years, the HLB incidence in commercial groves continues to increase, now affecting more than 44% of sweet orange trees in the citrus belt of São Paulo and Minas Gerais states (Fundecitrus, 2024). In Florida, the citrus industry relies on the limited production of more than 99% of infected trees (Li et al., 2020b). The disease has spread to almost all citrus-producing regions worldwide, except Australia and Mediterranean Basin.

The predominant HLB pathogen is the non-culturable, gram-negative α-proteobacterium ‘Candidatus Liberibacter asiaticus’ (CLas), which is mainly vectored by the Asian citrus psyllid (ACP) Diaphorina citri (Kuwayama) (Sternorrhyncha: Psyllidae) (Burckardt et al., 2021). Foliar symptoms include asymmetrical chlorosis known as blotchy mottle, enlarged veins, and intense canopy defoliation (Bové, 2006), caused by disturbance of the transport system between source and sinks, with starch accumulation in plastids, chloroplast disruption, and plugging of sieve cells via callose deposition (Bové, 2006; Andrade et al., 2019; Pandey and Wang, 2019). In field trees, these symptoms appear within some months to years after infection, depending on the tree’s age and size (Gottwald, 2010; Lopes et al., 2009). During the asymptomatic phase, infected plants remain indistinguishable from CLas-free plants and are an important source for the disease to spread within the grove, which complicates disease management (Lopes et al., 2009; Lee et al., 2015). How CLas can evade immunity and remain undetected by its host for such a long period is still unknown. CLas does not possess the conserved secretion mechanism to deliver effectors into plant cells, although most components needed to generate a functional Sec-dependent secretion system are present in its genome (Duan et al., 2009). The Sec-dependent secretion system has been hypothesized to be essential for pathogens like CLas or phytoplasmas to infect plants (Sugio et al., 2011; Yan et al., 2013) and it has been reported that CLas genes belonging putatively to this Sec-dependent system are upregulated in planta, but not in CLas-infected psyllids (Yan et al., 2013).

Although sensitivity to CLas varies among Citrus species and varieties, no complete resistance has been documented in this genus (Folimonova et al., 2009). However, Oceanian Citrus species formerly included by Swingle and Reece (1967) within the Microcitrus and Eremocitrus genera have been reported to be resistant to CLas (Ramadugu et al., 2016; Alves et al., 2021b, Alves et al., 2022). Moreover, partial or complete resistance has been reported in certain related genera within the family Rutaceae subfamily Aurantioideae (Ramadugu et al., 2016; Beloti et al., 2018; Cifuentes-Arenas et al., 2019; Alves et al., 2021a, Alves et al., 2021b, Alves et al., 2022). The monoembryonic orange jasmine (Murraya paniculata (L.) Jack) and curry leaf (Bergera koenigii L.), are good hosts for ACP (Damsteegt et al., 2010; Teck et al., 2011; Westbrook et al., 2011), being even more attractive to ACP females than sweet orange (Beloti et al., 2017; Tomaseto et al., 2019; Eduardo et al., 2022). M. paniculata is considered a transient host for CLas while B. koenigii is CLas-immune (Damsteegt et al., 2010; Beloti et al., 2018; Cifuentes-Arenas et al., 2019), providing a unique opportunity to explore distinct molecular mechanisms potentially involved in HLB-resistance. Recently, in a simulation of the natural entry of CLas in the plant host, Alves et al. (2021a) studied CLas dynamics in sweet orange Citrus × sinensis (L.) Osbeck ‘Valencia’, in M. paniculata, and in B. koenigii in the first stages of these plant-bacteria interactions. Using a well-controlled challenge-inoculation system with infectious psyllids, critical time points for these plant species with different responses to CLas infection were identified. In all three experimental systems, CLas was successfully released into the phloem by ACP, and bacterial titers decreased during the first days after inoculation. Subsequent increases in bacterial titers were observed for each host: while an exponential increase occurred in C. × sinensis up to the 40^th day post-inoculation (DAI), from which a plateau was reached, CLas failed to successfully replicate in B. koenigii. In M. paniculata, CLas titers were generally lower than in C. × sinensis, and titer dynamics resembled that of C. × sinensis up to the 30^th DAI, from which bacterial titers progressively decreased to undetectable levels (Alves et al., 2021a). These observations suggest that the first weeks after bacterial release into the phloem by ACP are critical for the establishment of the bacterium in each host, providing an excellent system to investigate molecular mechanisms leading to either resistance or susceptibility.

Here, a thorough transcriptomic analysis of new flushes of C. × sinensis, B. koenigii and M. paniculata was performed to study host gene expression dynamics after the ACP-vectored transmission of CLas. Our results suggest that the different responses to the bacterium in the three species are not due to the deployment of classical immunity in the resistant species pointing to a mechanism in which primary metabolism and other basic cellular functions are rewired to limit bacterial infection.

The current model of plant resistance settles mostly on studies of mesophyll-infecting pathogens, and perhaps its most relevant feature is the activation of inducible responses by the plant upon pathogen recognition (Zhou and Zhang, 2020). Therefore, it is not surprising that many transcriptomic studies designed to evaluate the responses to CLas in susceptible or tolerant Citrus species centered the discussions on the deployment or not of complete defensive responses (Arce-Leal et al., 2020; Weber et al., 2022; Li et al., 2024). To our understanding, two essential aspects have been overlooked in many of these studies. On one hand, if CLas recognition by tolerant plants leads to PTI or ETI, a transcriptomic response should be activated very early after bacteria recognition. The foundation of the deployment of a robust PTI lies in the activation of stress-responsive transcription factors and the gene networks under their control, which begins soon after pathogen recognition (Li et al., 2020a). PTI is rapidly evaded by successful pathogens through the action of secreted effectors acting into the plant cells. When these effectors are recognized by resistant plans, ETI is activated, characterized by a marked transcriptional response within hours after pathogen recognition (Mine et al., 2018). Omics studies on HLB symptomatic samples, in which CLas infection took place months earlier, are not targeting the first molecular events after CLas recognition. In addition, CLas-infected material inoculated by grafting requires the earliest time point to be sampled weeks after inoculation to allow vascular recovery and CLas translocation to become detectable by qPCR (Raiol-Junior et al., 2021). Consequently, it is also an inadequate experimental system to evaluate early events of CLas infection. Simulation of natural entry by ACP-mediated inoculation using young shoots (Alves et al., 2021a; Lopes and Cifuentes-Arenas, 2021), as the one used by Wei et al. (2021) or in this work, seems to be more realistic to elucidate the molecular events taking place in this plant-pathogen interaction.

On the other hand, a complete immune response involves a massive transcriptional reprogramming of the cell, in the range of thousands of differentially-expressed genes (Lewis et al., 2015; Li et al., 2016; Mine et al., 2018). For example, in Arabidopsis, up to 25% of the genes presents altered gene expression after inoculation with the virulent Pseudomonas syringae (Tao et al, 2003), and in Citrus, more than 6000 genes were differentially expressed in the citrus canker resistant Kumquat 72 hours after recognition of Xanthomonas citri subs. citri, its causal agent (Ferrara et al., 2020). Therefore, up-regulation of some defense-related genes does not necessarily mean that a complete defense response has been deployed.

The relatively low number of DEGs found in the early stages after pathogen inoculation into the susceptible C. × sinensis observed in this study aligns with other reports analyzing the response to CLas in asymptomatic plants. For example, Fan et al. (2012) reported no significant changes in gene expression at 5 weeks after grafting inoculation to either C. × limon or C. × sinensis. Similarly, Ramsey et al. (2020) identified only 313 DEGs at 2 weeks after grafting inoculation to C. × limon, and Lombardi et al. (2024) found only 223 DEGs 4 weeks after inoculation to C. × sinensis, this last report using also ACP-vectored inoculation. Interestingly, a limited transcriptional reprogramming is also observed in response to other phloem-infecting pathogens of citrus, such as Spiroplasma citri (McNeil et al., 2023), suggesting that this might be a common feature of vascular pathogens that share similar infection pathways. By contrast, Wei et al. (2021) detected a high number of DEGs as early as 1 day after CLas inoculation in ‘Valencia’ sweet orange.

Our results suggest the idea that exposure of susceptible plants to CLas is not dramatically altering gene expression and indicate that a canonical and complete transcriptional immune response is not being activated in this susceptible host, although the possibility of waves of transcriptional changes occurring between the time intervals used in this experimental design cannot be discarded. Two non-exclusive hypotheses could explain these observations: a) the secretion of bacterial effectors suppress PTI, as it is proposed for several proteins putatively secreted by CLas to suppress PTI (Clark et al., 2018; Shi et al., 2019; Clark et al., 2020; Basu et al., 2022; Shi et al., 2023a, Shi et al., 2023b). So far, the mechanism for secretion is unknown, as CLas does not have the main machinery for effector secretion present in most plant pathogens (Cui et al., 2015; Jain et al., 2015). These effectors could be part of an orchestrated strategy in which CLas could also secrete other proteins that may interfere with specific plant metabolites produced by nucleated cells, altering plant cell homeostasis for its own benefit, and finally inducing symptoms, as observed for other phloem-limited pathogens (MacLean et al., 2011). b) CLas remains unnoticed for mesophyll cells, due to the apoplastic entry and exclusive vascular localization mediated by ACP, and the nuclei-free sieve element are unable to mount an effective immune response (Jiang et al., 2019; Huang et al., 2020; Lewis et al., 2022).

The use of resistant species such as M. paniculata and B. koenigii, in which CLas establishment is clearly limited (Alves et al., 2021a) opens the possibility to explore the molecular basis for resistance. Monoembryonic plants may exhibit genetic differences attributable to seed propagation, which might influence gene expression profiles. However, both Aurantioideae species were introduced into Brazil many decades ago, are seed-propagated and widely used as ornamentals, being highly homozygous (Nguyen et al., 2019; our unpublished results). In any case, as observed in the PCA, samples for each time point were closely grouped, indicating consistency among samples. Surprisingly, the limited number of DEGs observed in these two interactions in the first weeks after ACP-mediated bacterial inoculation, as well as the lack of immune defense categories enriched after functional analysis, suggest that canonical PTI or ETI responses are not the main mechanisms responsible for resistance in these two species. However, functional analysis revealed that the bacterium was not completely evading transcriptional responses in the plants. Both M. paniculata and B. koenigii tissues react to the presence of the bacterium by reprogramming the transcriptome of basic cellular functions. In both species, CLas presence provoked a subtle but consistent decrease in the expression of photosynthesis-related genes, photosynthesis being an enriched category upon less-expressed genes as soon as 48 hours after exposure to the bacterium. Enrichment of photosynthesis-related categories after transcriptomic analysis has been reported upon CLas infection in susceptible plants (Albrecht and Bowman, 2008; Zhong et al., 2015; Hu et al., 2017; Liu et al., 2019), but these experiments were performed in symptomatic plants, in which starch accumulation had already impaired photosynthetic processes and induced chloroplast degradation (Etxeberria et al., 2009). As mentioned before, this transcriptomic signature can be considered an effect of visual symptoms and therefore unrelated to the events investigated here. A decrease in photosynthesis in M. paniculata and B. koenigii at these very early stages after CLas inoculation, however, could be perceived as a mechanism triggered by the plants to hamper pathogen colonization. Curiously, this was also observed by Wei et al. (2021) in sweet orange at 5 days post inoculation with Clas-infected psyllids. Photosynthesis reduction has been reported in the incompatible interaction between tobacco and Phytophthora nicotianae (Scharte et al., 2005), as well as in resistant barley plants after inoculation with powdery mildew (Swarbrick et al., 2006). In M. paniculata or B. koenigii, a decrease in photosynthesis in the early stages of CLas-infection could be a mechanism to regulate source/sink relationships to limit pathogen spread (McIntyre et al., 2021). CLas moves from source to sink tissues through the phloem vessels. After inoculation, CLas remains in the young flushes until they are mature enough to become source tissues (Raiol-Junior et al., 2017; Alves et al., 2021a). In citrus, young flushes are nutritionally richer than mature flushes, with higher concentrations of total amino acids in the phloem sap, positively associated with CLas and ACP fitness (Setamou et al., 2016, Setamou et al., 2017). In this study, C. × sinensis flush maturation took longer than in M. paniculata and B. koenigii, and C. × sinensis flushes exposed to CLas-positive psyllids reached maturity slower than those exposed to CLas-negative psyllids. The combination of these two factors would increase the time window in which CLas have the available nutrients crucial for its establishment. However, in young, infected flushes of M. paniculata or B. koenigii, photosynthesis decay could translate into a retardation in the sink to source transition, compromising the establishment of CLas in hosts presenting nutrient inadequacy (Killiny, 2016). In this regard, the observation at t10 that two out of five B. koenigii samples exposed to CLas-positive psyllids cluster to samples exposed to t10 CLas-negative psyllids might be events in which the bacterium had failed to multiply in these plants.

Other biochemical processes altered after CLas exposure in B. koenigii could hamper bacterial survival and multiplication, such as the decrease in the biosynthesis of sulfur-containing amino acids. CLas is not able to synthesize cysteine de novo and must overcome its lack by acquiring it from the surrounding environment (Lai et al., 2016; Merfa et al., 2019). Interestingly, the expression of cysteine biosynthesis genes was also higher in non-inoculated C. × sinensis young flushes than in Bergera’s ones, suggesting that this metabolic constraint could be determinant for CLas multiplication in the resistant species. Although these cellular rearrangements are taking place upon CLas perception, the lack of nuclei in sieve elements suggests that these responses should be activated in adjacent nucleated cells. Similarly, in the case of C. × sinensis and other susceptible species, effector proteins, trafficking through the plasmodesmata (Cheval and Faulkner, 2018), could suppress immune signaling in nucleated cells and impact the production of systemic signals (Huang et al., 2024).

The lack of a bona fide defense response in M. paniculata and B. koenigii prompted us to investigate whether these species could also have physiological or metabolic constraints that could limit CLas establishment. Comparison of expression of orthologous genes provides an estimation of biological processes that are active in young flushes of any of the three species. Our analysis revealed that the three transcriptomes are very different, and consequently, their biochemical composition should also differ, as reported by Killiny (2016). The small genome content of CLas and the lack of core metabolic pathways indicate that this bacterium heavily depends on the host nutrition (Wang and Trivedi, 2013; Lai et al., 2016), which could impact on the suitability of different species as host. Our functional analysis suggest that the higher expression of genes related to phosphate transport in C. × sinensis or related to the response to phosphate starvation in B. koenigii, which is essential for Liberibacter optimal growth (Cruz-Munoz et al., 2019), differ between the three species. Moreover, C. × sinensis young flushes have a higher expression of genes of the tricarboxylic acid cycle (TCA) than B. koenigii. The ATP metabolism was also enriched upon genes more expressed in C. × sinensis than in M. paniculata. These findings likely reflect a higher activity of this metabolic pathway in the susceptible C. × sinensis. Citrate is the preferred carbon source for Liberibacter (Cruz-Munoz et al., 2018) and evidence suggests that CLas obtains energy by exploiting TCA cycle intermediates from the plant (Wang and Trivedi, 2013; Jain et al., 2017) and psyllid hosts (Killiny and Nehela, 2017). An ATP translocase has been identified in CLas, suggesting the potential to import ATP directly from their environment (Vahling et al., 2010). It could be hypothesized that a lower activity of the TCA cycle in B. koenigii and M. paniculata flushes could be determining their less suitability for CLas establishment. In the most extreme case of unsuitability for CLas survival, B. koenigii may be considered as a non-host of the bacterium.

Overall, our data support the hypothesis that most C. × sinensis cells do not recognize the presence of CLas after inoculation, likely due to the ACP strategy to introduce the bacterium directly into the phloem in less than two hours (Wu et al., 2016), which results in a very limited - transcriptional response. The fortuitous protection obtained by CLas in the vascular tissue would allow it to remain unnoticed for mesophyll cells, thus limiting the possibility to orchestrate a robust and effective defense response. This, together with the putative effector-mediated suppression of PTI would explain the long incubation period and the difficulty in managing this devastating disease. This model could also explain why foliar spray of inducers of pathogen resistance, such as salicylic acid, have proven to be ineffective in reducing bacterial titer in HLB-infected plants (Atwood and Brlansky, 2009; Li et al., 2021) as well as in other phloem-limited pathogens (Rodriguez-Saona et al., 2021). The same scheme would apply for B. koenigii and M. paniculata, although in these cases the presence of the bacterium in the phloem would induce biochemical rearrangements in the surrounding cells, suggesting the existence of signaling events initiated in the sieve elements upon CLas recognition.