In 2018, Mexico was the fourth largest sweet orange (Citrus sinensis) producer in the world, with 4.7 million tons, 98% of which was used for domestic consumption (^{SIAP, 2019}; ^{Rivera-López et al., 2020}). However, historical production does not show the national productive potential and represents a highly heterogenous citrus production, determined by technological diversity, water regimes, edaphic factors and phytosanitary problems. Several federal and/or state programs have been implemented to boost productivity and even promote the modernizing of the citrus sector. Particularly in the last two decades, national phytosanitary campaigns have been established for the management of the Citrus tristeza virus (CTV) and Candidatus Liberibacter asiaticus (CLas) (^{Flores-Sánchez et al., 2017}; ^{Mora-Aguilera et al., 2014b}). These pathogens have a high epidemic potential, due to the dispersal capacity and agroecological adaptation of their vectors, the systemic infection condition and their genetic variability (^{Flores-Sánchez et al., 2017}; ^{Domínguez-Monge et al., 2014}; ^{Rivas-Valencia et al., 2010}). The CTV, and recently CLas, have caused important economic losses in Brazil, Argentina, USA, Venezuela and other countries, and have modified the world’s citrus production models (^{Flores-Sánchez et al., 2015}; ^{Dowson et al., 2015}). In Mexico, CLas, which causes the diseased named huaglongbing or HLB, is a bacterium, restricted to the phloem efficiently transmitted by Diaphorina citri (Hemyptera: Liviidae). In this country, the symptoms expression in C. sinensis is influenced by the rootstock and the bacterium sequence variant, and can even be asymptomatic under field condition (^{Martínez-Bustamante et al., 2015}). In contrast, the symptoms expression in sour citrus species is more intense, yet also dependent on the rootstock and pathogen genetics. Thus, C. aurantifolia/C. macrophylla displays asymmetric foliar chlorosis, seed abortion and fruit deformities, typically reported in orange trees in Brazil and Florida, but it also presents a generalized yellowing of the foliage (^{Esquivel-Chávez et al., 2015}; ^{Robles-González et al., 2013}; ^{Esquivel-Chávez et al., 2012}).

CTV, causal agent of the ‘citrus tristeza’ disease, is a closterovirus related to the phloem and transmitted semi-persistently by a vector aphid complex (Hemyptera: Aphididae), including Aphis gossypii and Toxoptera citricida, the most important in Mexico (^{Dowson et al., 2015}; ^{Loeza-Kuk et al., 2011}). In this country, as opposed to the epidemic scenario of the 1930´s en South America, in which wilting (‘tristeza’) and quick decline and death of about 15 million C. sinensis/C. aurantium trees became a notorious characteristic, the predominant condition has been asymptomatic and subendemic since pathogen detection in the 1980’s (G. Mora-Aguilera, 2018. Personal communication; ^{Rivas-Valencia et al., 2010}). Simulation models show that moderate races selection has operated on this virus in detriment of severe races, maintaining a relatively stable population structure with limited genotypic variability (G. Mora-Aguilera, 2018. Personal communication). However, this apparent equilibrium may be temporarily altered with the introduction of new races and imbalances due to climate and microbiological factors, such as those implied by the recent outbreak reported in central Veracruz (^{Martínez-Bustamante et al., 2020}). The difficulty of controlling these pathogens with conventional strategies (^{Mora-Aguilera et al., 2014a}; ^{Loeza-Kuk et al., 2011}), along with the need to produce resistant genetic materials in the least time possible (CTV) or the lack of feasible resistance sources against these pathogens (i.e. CLas) led to successfully obtain transgenic citrus material in the 1990’s, first on CTV and later on CLas (^{Soares et al., 2020}; ^{Loeza-Kuk et al., 2011}; ^{Gutiérrez-Espinosa et al., 1997}). Nevertheless, biosafety regulations have limited their extensive commercial implementetion. New genomic technologies have favored the study of endogenous hosts genes, as opposed to an exogenous gene implicit in transgenic technology, exploiting the contrasting expression of symptoms based on the pathogen genetic and the graft/rootstock composition. This natural phenomenon, known as systemic acquired resistance may be strengthened through transcriptomic studies by differential expression analyses, at the level of mRNA, of genes that respond to a pathogenic process respect to reference genes from the host or ‘housekeeping’ (^{Rawat et al., 2017}; ^{Fu et al., 2016}; ^{Mafra et al., 2013}). The latter genes are essential for the plant physiology; therefore, its expression is not dependent, in theory, of the infections process (^{Yan et al., 2012}; ^{Mafra et al., 2012}; ^{Xu et al., 2008}; ^{He et al., 2006}). An earlier investigation cloned three endogenous genes that may eventually provide resistance from pathogens in citrus species, including CLas and CTV, through its sustained expression in high concentrations in the cisgenic plants, providing resistance with its own genes (^{Ventura-Medina et al., 2019}). A crucial stage in this cisgenic approach is the quantification of the expression of a potentially endogenous gene in response to the plant-pathogen interaction, which depends on the correct selection of a reference gene (^{Yan et al., 2012}; ^{Mafra et al., 2012}). In this context, the aim of this investigation was to develop an in-house protocol to evaluate the expression of five reference genes based on efficiency and stability criteria in the amplification process with RT-qPCR regarding the endogenous CDR13 gene transcribed in response to CLas and CTV infection in C. sinensis/C. aurantium for future systemic acquired resistance studies under the Mexican citrus production and phytosanitary condition.

Plant material. In May 2019, nine C. sinensis plants, grafted onto C. aurantium, were chosen from the COLPOS-INIFAP collection established in the Tecoman experimental station of the National Institute of Forestry, Agriculture and Livestock Research (INIFAP-CE Tecomán). This material was kept during the investigation under greenhouse conditions in COLPOS Montecillo. The experiment included three plants for each sanitary condition: a) positive to CTV, b) positive to CLas and c) healthy. The diseased and healthy conditions were confirmed by qPCR in the Fruiticulture laboratory, COLPOS. The diseased plants were inoculated in 2014 with infected citrus tissue from Jalisco and Puebla commercial orchards. The healthy plants were 18-months-old. The plant sample used for the extraction of nucleic acids consisted of eight mature leaves per plant from different orientation and strata.

Extraction, quantity and quality of RNA. For extraction of total RNA, 0.1 g of midrib leaves (C. sinensis) were used with the CTAB protocol at 2%. Yield and quality of total nucleic acids were quantified using the NanoDrop 2000 (Thermo Fisher Scientific). The optic density rate at 260/280 and 260/230 nm wavelengths was evaluated to assured a 1.8-2.0 range to guarantee the absence of proteins or phenols contamination. The extraction of total nucleic acid allowed to obtain concentrations of 200-1000 ng µL^-1 (Table 1). In all samples, the amounts obtained were enough to homogenize a subsample at 50 ng µL^-1. This stage was crucial for the correct comparison of differential expression of the evaluated genes, avoiding over- or underestimating them. The optical density rate (260/280) was in the 1.93 - 2.01 range (Table 1), meaning that the quality and purity of the nucleic acids were within an optimal range. Values below 1.8 indicate the presence of proteins, phenols or other contaminants (^{Taylor et al., 2015}).

Retrotranscription. For the cDNA synthesis, 50 ng µL^-1 RNA were used after homogenizing with nuclease-free water. Retrotranscription with dT oligos was performed in two stages using an in-house protocol. Five trials were carried out with the purpose of optimizing the cDNA concentration (Table 2). The reaction mixture at the first stage was incubated at 85 °C for 3 min in a Bio-Rad T100™Thermal Cycler, thereafter the samples were placed in ice for 5 min.

Further, the volume corresponding to the reaction mixture of the second stage was added into each tube. Trials 1 and 2 were incubated at 44 °C for 60 min, whereas trials 3-5 were incubated at 44 °C for 100 min. In all trials, retrotranscriptase was deactivated at 92 °C for 10 min. The cDNA obtained was kept at -20 °C and quantified by spectrophotometry using the NanoDrop 2000 to evaluate the yield obtained.

The cDNA synthesis was carried out successfully by incubating at 44 °C for 100 min with the trial 5 configuration. This trial extended the time of enzymatic synthesis, which helped obtain average concentrations of 1069 ng µL^-1. The remaining trials produced lower average yields with 666-680 ng µL^-1 (trials 1 and 4) and 865-880 ng µL^-1 (trials 2 and 3). An optimum yield guarantees enough amounts of cDNA to implement simultaneous RT-qPCR with several primers related to multiple genes. It also reduces reagents and laboratory materials used for every sample. On the other hand, Oligo-dT is used in 40% of RT-qPCR trials, since it hybridizes to mRNA through the poly-A tail in comparison with an alternative random primer. However, the combination of both primer types is considered the best optimization option (^{Taylor et al., 2015}).

Reference genes. Five reference or housekeeping genes were bibliographically selected for their highly conserved condition in eukaryotic organisms. These genes have a continuous expression, independent to a phytosanitary condition (Table 3). We later evaluated its feasibility to normalize the expression of potentially associated genes with systemic acquired resistance under the experimental conditions of this investigation. For this purpose, the endogenous gene CDR13 was used, which has been related to this type of resistance in citrus plants (^{Rawat et al., 2017}).The primers used to detect and quantify the transcribed product of this gene were F: CAAGCTGATATAATACCCAATATCGGAG, and R: GAGGCTCGCACTGCGT (^{Rawat et al., 2017}; ^{Fu et al., 2016}).

RT-qPCR optimization. RT-qPCR was used to amplify and quantify the primary transcript (in its complementary cDNA form) of the five reference genes and the CDR13 endogenous gene. This technique was performed on a C1000 Touch™ (Bio-Rad) equipment. For every gene, 20 and 10 µL of the final reaction volume were tested. Both mixtures contained 1X of SsoAdvanced Universal SYBR^® Green Supermix (2X). For each primer, 200, 250 and 500 nM were evaluated, along with 50 ng of cDNA (Table 3). The configuration of the experiment consisted of three replications per sample obtained for each plant and the phytosanitary condition. Each replication was used to amplify the transcript of each reference gene and the endogenous gene. The same number of ‘not target controls’ were included for each primer. In total, the experiment included a total of 135 RT-qPCR tests. The amplification efficiency for each primer was evaluated with a dynamic range of 1x10³ - 1x10^-1 cDNA with a 1:10 dilution factor. Two thermocycling programs were evaluated: the first one consisted of denaturalization at 95 °C for 30 s, followed by 40 cycles at 95 °C for 5 s and an annealing temperature gradient in the range of 57-60 °C assessed for 30 s. The second program included denaturalization at 95 °C for 3 min, 30 cycles at 94 °C for 30 s and annealing at 57 °C for 40 s. For both amplification programs, three temperature ranges were evaluated to generate melting curves with increases of 0.5 °C/cycle: A) 70-85 °C; B) 60-95 °C, and C) 65-95 °C.

The curve shape and low values of cycling threshold (Ct) per sample, as well as the identification of the optimum melting temperature (T_m) with the formation of only one amplification peak were considered to optimize the RT-qPCR protocol. In addition to T_m, the specific amplification of a single expected fragment was verified, without the formation of dimers, by electrophoresis with agarose gel at 1.5% prepared in a TBE 1X buffer solution and using 110 volts for 40 min. It was finally viewed in the UVP Bioimaging Systems photodocumenter.

The optimum monoplex RT-qPCR protocol included a final reaction volume of 10 uL with 500 nM per primer and 1X of SsoAdvanced Universal SYBR^® Green. This configuration reduces the amount of primers without affecting the Ct or T_m. However, due to the accuracy required in the quantification, the use of probes may be another alternative to SYBR^® Green. The thermal profile included initial denaturalization at 95 °C for 30 s, followed by 40 amplification cycles at 95 °C for 5 s and 60 °C for 30 s.

The T_m was estimated with the temperature related to the single peak of the melting curve for each primer of GAPDH (81°C), F-BOX (83.5) and ACTIN (84) transcripts (Figure 1A, B). The primer associated with 18S rRNA gene produced a moderate peak at different temperatures for healthy plant samples and those infected with CTV and CLas, as opposed to what would be expected of a reference gene (Figure 1B). In the case of COX, a melting curve was not produced, and therefore, they were both discarded for use as reference genes. Putative resistance gene CDR13 had a T_m of 81 °C (Figure 1C). The single-peak melting curve indicated no problems regarding specific hybridization, primer concentration, or inefficient primer reaction conditions (^{Taylor et al., 2015}).

The amplification curves with the primers related to ACTIN and GADPH displayed Ct constants between 22.5-26 with 3500-4200 RFU (Relative fluorescence units) for the healthy plant samples and those infected with CTV and CLas (Figure 1D), whereas gene F-BOX displayed Ct between 25-27 with a similar RFU (Figure 1E). The primer related to 18S rRNA gene had Ct´s in the range of 35-37 with 600 RFU (Figure 1E), congruent with the inadequate melting curve, confirming its elimination. Contrary to expected, in all genes cases, samples of citrus infected with CTV showed the higher Ct´s (25-27) followed by health samples (24-25), indicating possible infection effect on the expression of these genes. The amplification curve with the primer relative to CDR13 gene had a 23 Ct for the three replicates (Figure 1F). The visualization in agarose gel of the fragments amplified per repetition was sampled with products obtained with the primers associated to ACTIN and GADPH genes. A simple band, without secondary products or dimers, confirmed the specificity of the primers used (Figure 1D).

RT-qPCR standard curve. The dynamic range was determined with four dilution factors, base 10 of cDNA (1x10³ - 1x10⁰ ng µL^-1), each with three replicates, in order to establish the optimum amount of cDNA required for the RT-qPCR reaction and to estimate the efficiency (E) (^{Taylor et al., 2015}). This process is illustrated with the primers related to ACTIN and CDR13 genes (Figure 2A-D). For these genes, the cDNA concentration had an inverse relation with Ct values (Figure 2 A, C), allowing the adjustment to a simple linear regression model with negative slopes of -3.33 (r²= 0.99) and -3.49 (r² = 0.99), and to estimate the efficiency (E) in 99.5 and 93.3 %, respectively. With the line adjusted, 1 ng µL^-1 was established as the minimum cDNA concentration threshold required to detect the ACTIN transcription with Ct = 32-33 and of CDR13 with Ct = 33.5-34 (Figure D, E). The estimated reaction efficiencies (E) were within the 90-110% range, considered acceptable (Taylor et al., 2015). These results are illustrative of the procedure, therefore the dynamic range must be estimated for each gene of interest, since the genic expression is not homogenous (^{Mafra et al., 2012}; ^{Yan et al., 2012}).

Figure 1 Melting curves (A, B, C) and amplification curves (D, E, F) with RT-qPCR using SYBR® Green for primers taget ing products transcribed from GAPDH and ACTIN (A, D), F-BOX and 18S rRNA (B, E) and CDR13 (C, F) gen using composed samples of leaves of C. sinensis/C. aurantium with three phytosanitary conditions: healthy (green and brown), positive to CLas (red) and positive to CTV (blue). Agarose gel at 1.5 % with amplified fragments with primers related to genes GAPDH and ACTIN (D, top section). 

Selection of reference genes. Based on the relative quantification obtained using the optimum RT-qPCR protocol, an analysis was carried out on the stability of the expression of GAPDH, ACTIN and F-BOX using the program CFX Manager™ ver 3.1 by Bio-Rad. GAPDH and ACTIN had the lowest M value (0.06), therefore the highest stability was achieved with 2.84 (ln 1/M). This indicates that expression of these genes had variability attributable mostly to the experimental process, but not to the infection by CLas or CTV, which was expected from a reference gene (Figure 3). Nevertheless, M < 0.5 is considered acceptable for normalization purposes. Due to the synchrony of GAPDH and ACTIN (Figure 3), the use of both genes is recommended for normalization purposes. ^{Mafra and collaborators (2012}) evaluated a set of genes, including ACTIN, GAPDH, F-BOX and 18S rRNA in different citrus species and rootstocks. They conclusively identified only F-BOX as the most stable. 18S rRNA and ACTB genes have also been reported effective (^{Yan et al., 2012}). These results indicate that the selection of reference genes, before a specific study of systemic acquired resistance, requires evaluation in the specific experimental conditions and with the biological material of interest. Likewise, it is crucial to optimize the RT-qPCR protocol with the designed or chosen primer(s), the plant sample and experimental goal (^{Taylor et al., 2015}). This investigation also identified the importance of standardizing the experimental units as much as possible, since the expressions of constitutive genes may be influenced by abiotic factors such as water stress and temperature.

Figure 2 Standardization curves and adjustment to a simple linear regression model of the RT-qPCR reactions obtained from four dilution factors (1x103 - 1x100) of cDNA used as a template for primers related to the expression of ACTIN (A, B) and CDR13 (C, D). E= reaction efficiency. 

Figure 3 Stability and average of expression variation between pairs of reference genes (M) obtained from quantifying by RT-qPCR central leaf nervation taken from orange trees (C. sinensis/C. aurantium), both healthy and infected with CLas and CTV. 

In conclusion, an in-house, non-commercial RT-qPCR protocol was successfully developed to normalize the endogenous gene CDR13, associated with systemic acquired resistance in sweet orange. The GAPDH, ACTIN and F-BOX reference genes presented melting curves expressing no dimers formation, Ct ≤ 28 and reaction efficiencies in the 90-110% range. However, GAPDH and ACTIN had the most stable genic expression (2.83) and are therefore proposed for the normalization of the expression of potentially endogenous genes in resistance studies of C. sinensis/C. aurantium in response to infections by CTV and CLas under experimental conditions, analogous to those used in this investigation.