Seeds of 66 rice varieties (56 landraces and 10 improved), both Brazilian and imported (Electronic Supplemen- tary Material), were disinfested with sodium hypochlorite (2%) and germinated on gauze in distilled water. Five days after germination (DAG), the seedlings were collected and used to extract total genomic DNA [6] . For the polymerase chain reactions (PCRs), 26 primers were selected from those described by Araújo et al. [7] , Areias et al. [8] , Bhuyan et al. [9] and Rabbani et al. [10] .

After tests and amplification condition adjustments, seven primers (OPA 04, OPA 07, OPA 19, OPD 10, OPE 17, OPH 05, OPP 02) were chosen based on the number of polymorphic bands and band reproducibility profile. The PCRs were performed in 20 µL of material, containing 25 ng of genomic DNA, 2.3 mM of MgCl₂, 0.2 mM of dNTP, 0.2 µM of each primer, one unit of Taq recombinant polymerase DNA (Invitrogen) and 2 µL of the enzyme’s buffer, with the volume completed by adding ultrapure water (Invitrogen), at temperatures according to Areias et al. [8] .

The PCR products were revealed by electrophoresis in 1.5% agarose gel, prepared using TBE 1/2 X, for two hours at 100 V and then stained in an ethidium bromide solution (10 µg∙mL⁻¹). The gel images were analyzed using the GelCompar II software (Applied Maths) and the band profiles were used to generate a genetic similarity matrix using the coefficient of Jaccard [11] . Finally, a genetic similarity dendrogram was plotted using the unweighted pair group method with arithmetic mean algorithm (UPGMA).

To analyze the expression of the nitrate transporter genes with low (OsNRT1.1) and high affinity (OsNRT2.1 and OsNRT2.2) and to determine the uptake kinetics parameters, four genetically contrasting rice varieties were se- lected (IAC-47, Bico Ganga, Arroz de Revenda and Manteiga) from the genetic similarity dendrogram obtained by Random Amplified Polymorphic DNA (RAPD) analysis (Figure 1). The seeds were disinfested and germinated as described previously. Fourteen days after sowing, the seedlings, measuring about 10 cm in height, were transferred to plastic pots (four seedlings each) containing 600 mL of the nutrient solution of Hoagland and Arnon [12] modified to 2 mmol∙L⁻¹ of -N and other nutrients, with one-fourth the normal ionic strength (IS). Four days later, the nutrient solution was modified to 1/2 IS, while the -N was maintained at 2 mmol∙L⁻¹. The solution was renewed every three days thereafter. On 30 DAG, the plants were subjected to 72 h of total N deprivation, while maintaining the other nutrients at 1/2 ionic strength. After the deficiency period, the N supply was resupplied to one group of plants, with 0.2 mmol∙L⁻¹ of nitrate, while the other group was kept in the solution without N. Root samples were collected 3, 6 and 9 h after resupply and stored at −80˚C for subsequent total RNA extraction.

The total RNA was extracted according to Gao et al. [13] and the synthesis of the first cDNA strand was carried out from 0.5 µg of RNA, in a 96-well thermocycler (Eppendorf Mastercycler) using the High-Capacity cDNA Reverse Transcription Kit according to the manufacturer’s recommendations. The PCRs in real time were performed in duplicate in a Step One Plus Real Time PCR System using the SYBR^® Green PCR Master Mix kit

(Applied Biosystems), according to the manufacturer’s recommendations. The sequences of the primers defined by Sperandio et al. [14] were used to amplify the nitrate transporters genes. Actin gene (OsAct) was used as an endogenous control [15] . The relative expressions were calculated according to Livak and Schmittgen [16] by relativization of the expression levels of the plants after resupply in function of the levels of those without resupply, so as to measure the levels of induction by the ion.

For the nitrate uptake kinetics assay, the Manteiga variety was excluded due to the absence of viable seeds. The germination of seeds of the other three varieties and the cultivation conditions in the growth chamber were the same as for the gene expression analysis. Immediately after N resupply and then each 30 minutes for the next 8 hours, 0.5-mL aliquots of nutrient solution were collected from each pot. At the end of this period, the volume of solution remaining in each pot was measured and the plants were collected and placed to dry at 60˚C in a forced-air chamber, after which the dried material was weighed.

The nitrate concentrations in the solution were determined according to Cataldo et al. [17] and were used to plot ion depletion curves and to determine the nitrate uptake kinetics parameters (K_m and V_max) through the mathematical graphing method proposed by Ruiz [18] , using the Cinética Win software developed by Ruiz and Fernandes Filho [19] . The values of K_m and V_max obtained for each replicate were submitted to analysis of variance and were used to estimate the influx rates, using the Michaelis-Menten equation, as described by Epstein and Bloom [20] .

The genetic similarity dendrogram obtained from the RAPD analysis presented two large groups (I and II) (Figure 1). Group I was highly heterogeneous regarding the origin of the varieties analyzed, including improved Brazilian and imported cultivars, new accessions to germplasma banks and a few from crops grown in the state of Maranhão.

The Piauí variety showed 88% similarity with IAC-600, a well-known variety called “black rice” developed by the Agronomic Institute of Campinas (São Paulo, Brazil). Although heterogeneous in its magnitude, Group I contained successive subdivisions that significantly separated the varieties included in it. Subgroup I.1a contained varieties with 100% similarity: Dourado Precoce and IAC-4440; Dobradinho and Zebu Branco. The improved variety IAC-47 was 95% similar to Dourado Precoce and IAC-4440. All the improved varieties were clustered in Group I. Most of the varieties with the name Lajeado from germplasma banks were also in Group I (Figure 1). It should be highlighted that Lageado Liso-220,029 and Lageado-220,006 had similarity greater than 80%, as also observed by Areias et al. [8] .

In turn, Group II presented the particular characteristic that it only contained varieties from crops grown in the county of Arari, Miranda do Norte, Penalva, Vitória do Mearim and Viana in Maranhão, suggesting a close genetic base for the varieties grown in that region (Figure 1).

Therefore, the similarity dendrogram allowed selecting four genetically contrasting rice varieties (IAC-47, Bico Ganga, Arroz de Revenda and Manteiga) for analysis of expression of the genes OsNRT1.1, OsNRT2.1 and OsNRT2.2 and determination of the nitrate uptake kinetics parameters.

Three hours after the resupply of nitrate, increases were observed of between 40 and 250 times in the expression of the OsNRT2.1 and OsNRT2.2 genes in relation to the plants maintained without nitrogen (Figure 2(a) and Figure 2(b)), while for OsNRT1.1 the expression levels were two times higher (Figure 2(c)). In general, OsNRT2.1 and OsNRT2.2 were strongly induced with the resupply following the suppression after 3 hours, while interestingly OsNRT1.1 presented higher expression in some varieties 6 and 9 hours later, suggesting a slower induction response (Figure 2(c)).

In the Arroz de Revenda and Manteiga varieties, the induction of expression of the OsNRT2.1 and OsNRT2.2 genes was greater with the resupply and was longer lasting (Figure 2(a) and Figure 2(b)), suggesting better adaptation to conditions of low soil nitrate levels.

Due to the large differences regarding the induction of the NRTs, especially observed 3 hours after resupply, we also expected to see large differences in the nitrate uptake capacity between the varieties. However, no difference was observed between the varieties with respect to the curves of nitrate depletion in the nutrient solution and the uptake isotherms (Figure 3). There were no significant differences of V_max between the treatments. The K_m value of the Arroz de Revenda variety was higher than that of IAC-47, which had the lowest value of this parameter (Table 1).