Australian lime-derived hybrids were developed using conventional breeding and protoplast fusion techniques. For conventional breeding, the finger lime (C. australasica, PI 312872 accession from the USDA National Plant Germplasm System [NPGS]) was used as the male parent and crossed with red pulp pummelos (C. maxima (Burm.) Merr.) such as the Pum 5-1-99–2 and the Pum C2-5-12, mandarin UF304 (C. reticulata hybrid), or a C. inodora OP seedling ( Table 1 ). The hybrids generated through conventional breeding were visually selected based on the color and size of the leaves (Mahmoud et al., 2025). Allotetraploid finger lime-derived hybrid was developed through somatic cell fusion, where diploid ‘OLL8’ sweet orange callus-derived protoplasts were fused with finger lime (FL) mesophyll-derived protoplasts (Dutt et al., 2021). As previously reported, the tetraploid somatic hybrid was evaluated using flow cytometry and simple sequence repeat markers (Dutt et al., 2021). The selected hybrids were propagated in a greenhouse before being transplanted into a sandy-soil field in Lake Alfred, Florida, USA, where they were naturally exposed to CaLas for more than ten years through natural psyllid feeding. The trees were irrigated daily, and Florikan 12-4-8 360-day controlled-release fertilizer (FLORIKAN E.S.A. LLC, Sarasota, Florida) was applied annually in March. The hybrids used in this study were chosen based on their demonstrated tolerance for HLB and vigorous growth characteristics.

Six selected Australian lime-derived hybrids were propagated in the mist bed via single-node mature cuttings on a Pro - Mix BX general purpose potting mix and maintained under controlled conditions for one year. Following the propagation period, the plants were transplanted into 10 × 10 × 35 cm standard square pots (“citripots”) filled with the same Pro-Mix soilless medium (Premier Tech Horticulture, Quakertown, PA, USA). They were then placed in a 50% shaded greenhouse, exposed to natural daylight and maintained under a relative humidity of 70 ± 10% and a temperature of 32 ± 2°C. The clonally propagated trees were divided into two sets. The first set was top grafted with CaLas-free ‘Valencia’ sweet orange (Citrus × sinensis (L.) Osbeck) budwood. In contrast, the second set was grafted with CaLas-positive ‘Valencia’ budwood obtained from HLB-infected trees (Ct values of CaLas = 25 ± 3.5). The performance of those trees was compared to that of ‘Valencia’ grafted onto Swingle citrumelo, a common commercial rootstock.

The growth index was determined via measurements of scion length, scion diameter, and rootstock diameter using a Vernier caliper and metric tape measure. Scion length was measured from the tip of the main stem to the graft union, encompassing the length of both the spring and summer shoots. The scion and rootstock diameters were measured 5 cm above and below the graft union. The experimental design comprised twelve replicates for each graft combination. Graft-take, indicating graft compatibility, was also evaluated. Initial compatibility was determined by the presence of healed grafting unions and the observation of new viable shoots. The number of live grafts was recorded for each treatment, along with the time elapsed after grafting, and the graft-take percentage relative to the total number of grafted plants was calculated. Furthermore, three plants from each combination were longitudinally sectioned with a scalpel blade at the median region of the grafted region within a 5 cm long stem segment, splitting them into two halves for sampling. The samples were stained with a 5% ferric chloride solution for 2 hours to detect phenolic compounds. The ferric chloride solution binds to the phenolic compounds in the grafting area, resulting in a visible color change. The color was then observed under a microscope (Carl Zeiss Microscopy GmbH, Göttingen, Germany) equipped with a Zeiss AxioCam ICc1. ImageJ software was used to measure the optical intensity of staining (Ruzin, 1999; Marques et al., 2018).

To assess the CaLas titer in the infected leaves from the scion and the roots of the rootstock of greenhouse-grown trees, total DNA was periodically isolated from the leaf petioles and midveins of fully expanded leaves or root tissues. DNA extraction was performed using the GeneJET Plant Genomic DNA Purification Kit (Thermo Fisher Scientific, Massachusetts, USA) following the manufacturer’s instructions. The concentration of the extracted DNA was normalized to 25 ng/μL before conducting quantitative PCR (qPCR). Quantitative PCR was carried out using a QuantStudio™ 3 System (Thermo Fisher Scientific). The CaLas genomic DNA was tested using TaqMan™ Gene Expression Master Mix and the CQUL primers ( Supplementary Table 1 ), designed to amplify a segment of the CaLas rplJ/rplL ribosomal protein-encoding gene (Wang et al., 2006). CaLas multiplication was evaluated regularly over 24 MAI at 6, 12, 18, and 24 MAI with CaLas-infected or CaLas-free (control) trees.

Fifteen mature leaves from ‘Valencia’ plants were harvested, frozen, and ground in liquid nitrogen. Each hybrid was sampled in six biological replicates, and the ground leaf material was stored at -20°C until biochemical assays were performed. Chlorophyll a, chlorophyll b, and total chlorophyll levels were determined by measuring the absorbance at different wavelengths (665 nm for chlorophyll a (Chl a) and 653 nm for chlorophyll b (Chl b)). Chlorophyll a, chlorophyll b, and total chlorophyll contents were estimated (Lichtenthaler and Wellburn, 1983).

The starch content was determined as described previously by Rosales and Burns (2011) with slight modifications. Fresh tissues (100 mg) were ground to a powder, suspended, and homogenized in 700 μl of distilled water. Leaf samples and a standard solution were boiled in water for 10 min and then transferred to cold water to cool. The samples were vortexed and then centrifuged for 2 min at 6000 rpm. Aliquots of the supernatant were collected in new tubes. Three hundred microliters of the supernatant were subjected to extraction with 900 μl of 100% ethanol. The mixture was vortexed and centrifuged for 10 min at 10000 rpm. The supernatant was discarded, and 1 ml of distilled water was added to dissolve the pellet. Fifty microliters of KI: I2 (8 mM: 50 mM) was added. The starch content was quantified by monitoring the color change with a spectrophotometer at 594 nm. Rice starch (Sigma Aldrich, St. Louis, MO) was used as a standard.

The total phenolic compounds were estimated according to Singleton and Rossi (1965). In brief, 100 mg of fresh leaf tissue was subjected to extraction in 1 ml of 80% ethanol and then centrifuged for 20 min at 10000 rpm. The TPC extract was centrifuged, and 100 µL of Folin reagent (1:10) was mixed with the leaf extract. After vortexing and 5 min of incubation at room temperature, the reaction was initiated by adding 300 mL of 20% sodium carbonate (Na₂CO₃) to the extract. The color change was recorded after 60 min by measuring the absorbance at 650 nm. The standard curve for phenol was prepared by measuring 1 mL each of a series of gallic acid solutions in ethanol at different concentrations from 0 to 1.00 mg/mL. The phenol contents are expressed as mg of gallic acid per 100 g of tissue fresh weight.

The total flavonoid content in the leaf samples was estimated using a colorimetric assay with aluminum chloride (Aktumsek et al., 2013). The methanolic extract (50 µL) was diluted in 200 µL of distilled water, and the diluted samples were mixed with 60 µL of 5% sodium nitrite (NaNO₂) solution and 60 µL of 10% aluminum chloride (AlCl₃) solution. After a 5-minute incubation at room temperature, 400 µL of 1 M sodium hydroxide (NaOH) and 1.230 mL of distilled water were added before vortexing. The absorbance of the reaction mixture was measured at 510 nm. A standard curve was created using catechin standard solutions (0–200 ppm), and the content of total flavonoids is expressed as mg catechin equivalents per 100 mg of fresh weight.

RNA was isolated according to the manufacturer’s protocol from 100 mg of finely ground leaf tissue via a Direct-zol™ RNA Miniprep Kit. qPCR was performed with a final volume of 10 μl using the SYBR Green Kit according to the manufacturer’s instructions. Each sample was tested twice in twelve replicates. The Ct value of the PCR curve was analyzed and compared with that of Val/SW as a control. The relative expression levels of the selected genes were calculated by the 2^−ΔΔCT method (Livak and Schmittgen, 2001) and the calculation of Log2 fold change compared to ‘Valencia’ grafted onto Swingle rootstock using RStudio. The actin housekeeping gene was used as an endogenous control. The primer sequences of the genes evaluated in this study are outlined in Supplementary Table 2 .

To investigate the performance of ‘Valencia’ sweet orange grafted on the Australian lime-derived hybrids, analysis of variance (ANOVA) was conducted in JMP Pro version 17 (SAS Institute, Cary, NC, USA) and RStudio (version 2024.12.1 Build 563, Posit Software, PBC). The CaLas infection and scion/rootstock combinations were examined via a factorial experiment with two factors (CaLas– or CaLas + infection) and seven scion/rootstock combinations. For gene expression analysis, data collected under CaLas+ conditions were evaluated using one-way ANOVA to compare relative gene expression among graft combinations, with ‘Valencia’ on Swingle rootstock as the control. ANOVA followed by the post hoc Tukey–Kramer honestly significant difference (HSD) test was used. Differences were considered significant when the p value was less than 0.05%.

‘Valencia’ sweet orange exhibited graft compatibility with Australian lime-derived hybrids, with grafting success rates ranging from 80% to 100% when CaLas - budwood was used and 75% to 100% when CaLas+ budwood was used. However, two hybrids, O+M2–75 and UF SunLime, presented lower grafting success rates of 40% and 50%, respectively. When grafted with CaLas+ ‘Valencia’ budwood, the percentage of grafted plants further decreased to 35% and 40% in these hybrids, respectively ( Figure 1A ). To assess the accumulation of phenolic compounds at the graft union in infected trees, we applied a 5% ferric chloride solution to longitudinal sections of the grafting union, and the staining intensity was estimated in the infected trees. The mean optical density (OD) values varied among the hybrids (p < 0.0001) ( Figure 1B ). Val/PFL 2–61 and Val/MFL1–98 presented the lowest OD values (0.34 and 0.37, respectively), whereas PFL 1–3 presented the highest OD at 0.6467 ( Figure 1B ).

The growth performance of ‘Valencia’ orange grafted onto Australian lime-derived hybrids was assessed under both CaLas-free (HLB−) and CaLas-infection (HLB+) conditions at 12 and 24 months post grafting ( Table 2 ). The results revealed significant variations in plant vigor among the different rootstock combinations. At 12 months post-grafting, Val/PFL2–61 and Val/MFL1–98 presented the most significant growth across all the measured parameters, with a height of 67.17 cm, a scion diameter of 5.71 cm, and a rootstock diameter of 9.70 cm under CaLas-free (HLB−) conditions. Similarly, Val/MFL1–98 presented a comparable plant height of 67.13 cm, with a slightly larger scion diameter of 6.16 cm and a rootstock diameter of 9.14 cm. Under CaLas-infection (HLB+) conditions, a general decline in growth was observed across all rootstock combinations; however, Val/MFL1–98 maintained the highest growth among the infected trees, with a plant height of 65.32 cm, a scion diameter of 5.90 cm, and a rootstock diameter of 9.05 cm.

Other rootstock combinations, including Val/SW (Swingle), Val/PFL1-11, and Val/UF SunLime, also presented moderate vigor at 12 months. The plant height of Val/SW reached 55.67 cm (HLB−) and 47.83 cm (HLB+), whereas that of Val/UF SunLime reached 40.00 cm (HLB−) and 38.67 cm (HLB+). In contrast, Val/PFL1–3 presented the lowest growth in the first year, with heights of 32.33 cm (HLB−) and 30.87 cm (HLB+), along with the smallest scion and rootstock diameters among all the tested hybrids. At 24 months post grafting, Val/PFL2–61 and Val/MFL1–98 continued to exhibit the most vigorous growth. Val/PFL2–61 presented the greatest plant height (124.71 cm in HLB− and 94.05 cm in HLB+), along with substantial scion (12.55 cm, 10.46 cm) and rootstock (12.55 cm, 10.73 cm) diameters. Val/MFL1–98 maintained strong growth performance, with plant heights reaching 101.76 cm (HLB−) and 96.66 cm (HLB+), with scion diameters of 10.20 cm and 9.84 cm, respectively. Conversely, Val/O+M2–75 exhibited the lowest overall growth at 24 months, with plant heights of 59.29 cm (HLB−) and 57.79 cm (HLB+), as well as the smallest rootstock diameter among all combinations (7.94 cm and 7.26 cm, respectively). Additionally, Val/PFL1-3, despite showing some improvement in growth, remained among the least vigorous combinations, with plant heights of 71.77 cm (HLB−) and 64.37 cm (HLB+).

CaLas bacterial titers were evaluated in ‘Valencia’ grafted onto various rootstocks at four sampling periods (6, 12, 18, and 24 MAI) ( Table 3 ). At 6 MAI, the Ct values ranged from 21.1 to 23.3 in ‘Valencia’ leaves, with Val/PFL2–61 showing the lowest Ct value (21.1), whereas Val/SW presented the highest Ct value (23.3). At 12 MAI, the Ct values varied significantly (p < 0.0001), with Val/PFL1–3 exhibiting the highest Ct value (35.3), whereas Val/PFL1–11 had the lowest Ct value (22.5). Similarly, Val/UF SunLime, Val/O+M2-75, and Val/MFL1–98 displayed higher Ct values (33.5, 31.6, and 30.8, respectively) than Val/SW (24.2). At 18 MAI, the Ct values continued to reflect differences in the bacterial titer (p < 0.0001). Val/PFL1–3 maintained a high Ct value (33.0), whereas Val/PFL2-61, Val/PFL1-11, and Val/SW presented the lowest Ct values (26.4, 22.3, and 24.2, respectively). At 24 MAI, the Ct values remained significantly different across rootstocks (p < 0.0001). Val/PFL1–3 presented the highest Ct value of 35.1, followed by Val/MFL1-98 (30.6), Val/O+M2-75 (25.0), Val/PFL1-11 (22.4), and Val/UF SunLime (22.1). Additionally, the Ct values of the root samples from the hybrids ranged from 30.1 to 37.7, with all the hybrids testing negative for CaLas in the roots. However, PF1–11 and Swingle presented the lowest Ct values in the root samples (30.1 and 29.1, respectively) ( Figure 2 ).

Analysis of the chlorophyll content revealed significant differences across the hybrids, with both chlorophyll a (Chl a) and chlorophyll b (Chl b) levels significantly reduced in plants infected with CaLas (p < 0.05) ( Figure 3 ; Table 4 ). At 12 MAI, the infected plants exhibited consistently lower Chl a values, with the most pronounced reduction observed in Val/PF1-3 (9.27 mg/g), compared to its non-infected counterpart (15.26 mg/g). Similarly, reductions were recorded in Val/UF SunLime (from 15.40 to 10.40 mg/g), Val/PF2-61 (from 18.94 to 12.28 mg/g), and Val/O+M2-75 (from 16.10 to 11.64 mg/g). Among all genotypes, Val/PF2–61 had the highest Chl a content under non-infected conditions, while Val/PF1–3 showed the lowest Chl a content under infection. At 24 MAI, the trend continued, with CaLas-infected plants exhibiting further Chl a reduction in most genotypes. Val/PF1–3 again had an obvious drop from 13.91 mg/g(non-infected) to 8.92 mg/g (infected). Val/PF2–61 and Val/UF SunLime also showed significant declines (17.42 to 11.46 mg/g, and 14.88 to 10.07 mg/g, respectively). Val/Swingle and Val/O+M2–75 maintained relatively stable Chl a levels between 12 and 24 MAI but still exhibited a consistent decline in the presence of infection ( Figure 3A ).

Across all genotypes, CaLas infection consistently reduced Chl b content compared to uninfected controls. At 12 MAI, infected plants exhibited lower Chl b levels. For instance, Val/Swingle decreased from 7.22 mg/g (uninfected) to 4.11 mg/g (infected), Val/PF2–61 declined from 6.75 to 5.33 mg/g, and Val/MF1–98 dropped from 5.98 to 4.54 mg/g. The genotype with the lowest reduction was Val/O+M2-75, with infected plants retaining a relatively higher Chl b level (6.52 mg/g vs. 7.02 mg/g uninfected). In contrast, Val/PF1–3 and Val/UF SunLime showed a significant decline of over 1.8 mg/g. At 24 MAI, the downward trend in Chl b content continued in infected plants. Val/PF1–3 declined from 5.96 mg/g (uninfected) to 3.78 mg/g (infected), Val/PF2–61 from 6.25 to 4.23 mg/g, and Val/MF1–98 from 5.65 to 4.43 mg/g ( Figure 3B ).

Total chlorophyll (T Chl) content was significantly reduced in CaLas-infected plants compared to uninfected controls across all genotypes and both time points (12 and 24 MAI). The extent of chlorophyll loss varies depending on the genotype. At 12 MAI, the greatest reductions were observed in Val/PF2-61 (19.56 mg/g in infected vs. 24.79 mg/g in uninfected), Val/O+M2-75 (18.16 mg/g vs. 23.33 mg/g), and Val/PF1-3 (13.60 mg/g vs. 18.67 mg/g). In contrast, Val/MF1–98 showed minimal reduction, with 18.95 mg/g in infected plants compared to 19.53 mg/g in uninfected ones. By 24 MAI, chlorophyll loss became more pronounced in most genotypes. The most affected combinations included Val/PF2-61 (15.69 mg/g infected vs. 23.81 mg/g uninfected), Val/PF1-3 (12.92 mg/g vs. 20.22 mg/g), and Val/Swingle (14.25 mg/g vs. 21.50 mg/g). Val/MF1–98 maintained relatively higher chlorophyll levels under infection, with a slight difference between infected (19.81 mg/g) and uninfected (17.94 mg/g) plants ( Figure 3C ).

Starch content was elevated in CaLas-infected plants compared to uninfected controls at both 12 and 24 months after inoculation (MAI), with clear variation among genotypes. At 12 MAI, the highest starch accumulation among infected plants was observed in Val/PF1-11 (3.18 µg/mm²), followed by Val/Swingle (2.82 µg/mm²) and Val/PF1-3 (2.66 µg/mm²). In contrast, the lowest values were recorded in Val/PF2-61 (2.36 µg/mm²), Val/MF1-98 (2.42 µg/mm²), and Val/UF SunLime (2.41 µg/mm²). By 24 MAI, starch levels significantly increased in most infected combinations, with the highest content measured in Val/PF1-11 (3.53 µg/mg), Val/Swingle (3.28 µg/mg), and Val/O+M2-75 (3.04 µg/mg). Meanwhile, uninfected control plants maintained relatively stable starch levels across both time points ( Figure 4 and Table 4 ).

The ANOVA for total phenolic content (TPC) and total flavonoid content (TFC) revealed distinct effects of rootstock, time, and their interactions ( Table 4 ). For TPC, significant main effects were observed for rootstock (F = 13.042, p < 0.001) and time (F = 4.336, p < 0.05). A significant rootstock × time interaction was also detected (F = 6.238, p < 0.001). In contrast, CaLas infection alone did not significantly affect TPC levels (F = 0.778, p = 0.378), and none of its interactions with rootstock or time reached statistical significance.

For TFC, highly significant effects were also observed for rootstock (F = 37.988, p < 0.001) and time (F = 19.917, p < 0.001). Additionally, interaction terms were significant, including rootstock × infection (F = 3.956, p < 0.01), rootstock × time (F = 6.959, p < 0.001), and rootstock × infection × time (F = 5.591, p < 0.001). Similar to TPC, infection alone did not result in a significant main effect on TFC (F = 0.896, p = 0.344), indicating the importance of genotype-specific responses over infection status alone.

Despite the absence of significant main effects of infection, both TPC and TFC levels were higher in CaLas-infected plants compared to non-infected controls at both 12 and 24 months after inoculation (MAI) ( Table 4 ). At 12 MAI, TPC in infected plants ranged from 205.0 to 312.5 µg/g, with the lowest level observed in Val/Swingle and the highest in Val/PF1-3. In non-infected plants, TPC values ranged from 169.7 to 303.3 µg/g. By 24 MAI, TPC levels in infected plants ranged between 167.6 and 303.1 µg/g, with Val/Swingle exhibiting the lowest concentration and Val/PF2–61 the highest. In non-infected counterparts, the TPC range was 220.0 to 287.9 µg/g, with the lowest detected in Val/PF2–61 and the highest in Val/PF1-3 ( Figure 5A ). These results indicate that while infection may induce phenolic accumulation in some genotypes, the magnitude and direction of change are strongly dependent on the specific rootstock.

For TFC, non-infected plants showed a TFC range of 131.9 to 291.9 µg/g, with Val/PF1–11 and Val/Swingle showing the lowest and highest levels, respectively. However, the infected plants at 12 MAI displayed values ranging from 144.4 to 323.6 µg/g. The lowest level was recorded in Val/PF1-3, while the highest was in Val/Swingle. At 24 MAI, TFC content of the non-infected plants ranged between 149.5 to 279.96 µg/g, with Val/UF SunLime having the lowest value and Val/Swingle the highest ( Figure 5B ). Whereas the values in the infected plants ranged from 210.2 to 272.8 µg/g.

The relative expression levels of the selected genes were calculated using the 2^^−ΔΔCT method ( Figures 6 – 9 ), with ‘Valencia’ grafted onto Swingle rootstock under CaLas+ infection as the reference condition. To provide an integrated overview of gene expression responses, we additionally calculated and visualized the log2 fold change of selected genes involved in carbohydrate metabolism, cell wall synthesis, detoxification, and defense-related responses relative to the Swingle-grafted control ( Figure 10 ).

The relative expression of genes involved in carbohydrate metabolism and cell wall synthesis in ‘Valencia’ grafted onto the hybrid rootstocks was compared with ‘Valencia’ grafted onto Swingle. These included genes encoding Glycosyl hydrolase family 32 protein (orange1.1g008242m), Beta-Amylase 6 (orange1.1g010067m), and Chloroplast Beta-Amylase (orange1.1g017328m). We observed upregulation of Chloroplast Beta-Amylase in Val/PFL1-3, Val/MFL1-98, Val/MFL 1–11 and Val/MFL 2–62 compared with Val/SW. Additionally, Val/PFL2–61 showed significant upregulation of Glycosyl hydrolases compared with Val/SW ( Figures 6 , 10 ).

Val/PFL1–3 exhibited significant upregulation of the expression of 1,3-Beta-Glucan Synthase, followed by Val/PFL1–11 and Val/PFL2-61 ( Figures 7A, B , 10 ). Furthermore, we analyzed the relative expression of Callose Synthase 7 (orange1.1g037960m), Callose Synthase 10 (orange1.1g000180m), and Callose Synthase 11 (orange1.1g000258m). Val/PFL1–11 showed significant upregulation of Callose Synthase 7 compared with Val/SW. ‘Valencia’ grafted onto Australian Lime-derived hybrids exhibited downregulation of the expression of Callose Synthase 10. However, Valencia’ grafted onto PFL1–3 exhibited higher Callose Synthase 11 expression compared to Val/SW. In contrast, Val/MFL1–98 significantly reduced Callose Synthase gene expression compared with Val/SW ( Figures 7 , 10 ).

The expression patterns of genes involved in detoxification and stress responses varied when ‘Valencia’ plants grafted onto different rootstocks were compared. Glutathione S-Transferases (GSTs), including Glutathione S-Transferase Tau 7, and Glutathione S-Transferase Tau 8, showed slight differences in expression. Glutathione S-Transferase and Glutathione S-Transferase Tau 8 (orange1.1g046710m) were highly expressed in Val/PFL 1-11, Val/PFL 2-61, Val/PFL 1-3, and Val/MFL1–98 compared with Val/SW. Moreover, the Thioredoxin protein superfamily (Thioredoxin reductase) exhibited significantly higher expression in ‘Valencia’ grafted onto Australian Lime Derived Hybrids compared to Val/SW. Catalase expression exhibited higher expression in Val/PFL 1–11 compared to Val/SW ( Figures 8 , 10 ).

We investigated the relative expression of several defense-related genes, including PR1-like (CAP2), LCD69, SAM1, SAM2, DMR6, and MES2, across different graft combinations. CAP2 was highly upregulated in Val/PFL 1-11, Val/PFL 1-3, Val/MFL1-98, and Val/UF SunLime compared with Val/SW. The relative expression of LCD69, SAM2, and DMR6 was highly expressed in Val/MFL1-98, Val/PFL 1-3, Val/PFL 1-11, Val/PFL 2-6; however, its levels were significantly lower in Val/O+M2–75 and Val/UF SunLime compared with Val/SW. SAM1 and CsMES2 showed relatively lower expression compared to Val/SW ( Figures 9 , 10 ).