cDNA microarrays were used to identify genes that were differentially expressed in genotypes contrasting for sucrose content. The arrays preparation, validation and analysis were done as previously described [31
]. Multiple crossings were performed for twelve years among S. officinarum
and S. spontaneum
(Population 1) and between commercial varieties SP80-180 and SP80-4966 (Population 2) to generate genotypes with extreme values of sugar content. The simplest way to access phenotypic differences with a high degree of confidence is to measure sucrose in the culm juice. This can be done in the field using a simple refractometer that evaluates Brix (soluble solids content). In sugarcane most of the soluble solids in the juice (70 to 91%) correspond to sucrose. Using this approach, thousands of genotypes can be phenotyped and contrasting individuals among the populations can be selected for further agronomic evaluation. Brix measurements were taken from 500 individuals of each population and the extreme clones in this population were selected and evaluated for sucrose content (see Additional file 1
). To evaluate gene expression samples were collected from single individuals as well as from pools of seven or eight plants grown for seven, ten and elevenmonths.
Two experimental designs were used to perform transcriptome comparisons: (I), internodes 1, 5 and 9 from high Brix plants were compared to the same internodes from low Brix plants (HB vs LB) in both populations or (II), mature internodes 9 were compared to immature internodes 1 from plants with high or low Brix in population 2 [33
]. Twenty six hybridizations were performed revealing 239 genes associated with sucrose content and regulated during culm development (see Additional file 2
and Figure ).
Figure 1 Comparison of differential gene expression associated with sucrose content, culm development and drought responses in sugarcane. Genes were identified as associated with sucrose content if they were differentially expressed when high Brix or low Brix (more ...)
A total of 117 genes were found to be differentially expressed in at least one comparison between high and low Brix genotypes (internodes 1, 5 or 9), and ten genes (SCCCLR1048F03.g, SCCCLR2003E10.g, SCCCRZ1001F02.g, SCCCRZ1001H05.g, SCCCRZ1002E08.g, SCEZST3147A10.g, SCJFRZ2007F10.g, SCAGLR1043E04.g, SCSBHR1050B11.g and SCVPCL6041F01.g) were found to be differentially expressed in both populations analyzed (see Additional file 2
). Among these SAS, we found three transcription factors, two aquaporins and two transcripts related to development. The gene expression comparison between mature and immature internodes showed a total of 173 differentially expressed genes (see Additional file 2
and Figure ).
Table lists a selection of the differentially expressed genes along with the number of biological samples that displayed altered expression when high and low Brix pools of plants were compared (HB vs LB) and when mature and immature internodes were compared (MI vs II). The expression data sets were compared to those obtained for plants exposed to drought conditions or ABA treatment [31
] (see Additional file 2
). Comparison to ABA treated plants yielded eleven differentially expressed genes in common, including the ScPKABA1-3
(SCRFLR1034G06.g) and the ScMAPK-4
(SCSBAM1084E01.g), which were both more expressed in high Brix and repressed by ABA, and a PP2C-like protein phosphatase (SCEPRZ1010E06.g) which showed the opposite profile. Comparison to drought-regulated genes showed an extensive overlap in differential expression between the two datasets. Between 117 and 173 genes associated with high sucrose content and internode development, respectively, 43.6% and 28.3% were previously shown to be altered by drought while twenty-two genes were altered in all conditions analyzed (Figure ).
Selection of SAS showing differential expression when high and low Brix plants were compared or when mature and immature internodes were compared.
Expression data of forty-two genes was also obtained using qRT-PCR. We determined gene expression differences for pools of extreme individuals from both populations (Figure ), in mature and immature internodes (Figure ) and in response to drought and ABA treatment (Figure ). The significance of the data obtained by qRT-PCR was inferred statistically by calculating values of P for expression differences against the reference sample (see Methods for details). Overall gene expression data obtained using cDNA microarrays was confirmed in qRT-PCR experiments for over 80% of the genes tested, even when the target RNA derived from a distinct biological replicate. We also investigated, using qRT-PCR, how the expression levels varied among the individual genotypes from Population 1 (Figure ). In this case, the value of P was calculated against the average expression level across genotypes and the validation rate was around 58%. Additional file 3
lists all the values of P for the validated genes.
Figure 2 Real Time PCR (qRT-PCR) analysis of Populations gene expression. The y axis refers to the relative expression ratio between target mRNA versus the reference mRNA (polyubiquitin-PUB SCCCST2001G02.g). The relative expression levels were determined in Internode (more ...)
Figure 3 Real Time PCR (qRT-PCR) analysis of internode developmental gene expression. The y axis refers to the relative expression ratio between target mRNA versus the reference mRNA (polyubiquitin SCCCST2001G02.g). The relative expression levels were determined (more ...)
Figure 4 Real Time PCR (qRT-PCR) analysis of drought and ABA-responsive gene expression. The y axis refers to the relative expression ratio between target mRNA versus the reference mRNA (polyubiquitin SCCCST2001G02.g; GAPDH Gene ID: 542367; UBE2 SCBGLR1002D06.g) (more ...)
Figure 5 Real Time PCR (qRT-PCR) analysis of individual genotypes gene expression. The y axis refers to the relative expression ratio between target mRNA versus the reference mRNA (polyubiquitin SCCCST2001G02.g). The relative expression levels were determined (more ...)
In order to unravel signaling aspects of sucrose accumulation, we asked whether genes differentially expressed in contrasting Brix genotypes or in mature-versus-immature internodes could represent direct sucrose- and/or glucose-regulated genes and, therefore, be part of the sucrose- and glucose-response pathways. To this end, sugarcane seedlings were treated with 3% sucrose or 3% glucose for 4 h and the expression of thirty-four genes was analyzed by qRT-PCR. The expression of thirty of these genes was affected by sucrose, of which six were also found to be regulated by 3% manitol (osmotic control) and thus, were not considered as true sucrose-responsive genes (see Additional file 3
). Figure shows the expression pattern of fifteen of these genes. Among the twenty-four sucrose-regulated genes, nineteen were also found to respond to glucose, indicating a significant overlap between these two signaling pathways (see Additional file 3
and Figure ). This is not unexpected since sucrose can be readily converted to glucose and sucrose-specific responsive pathways have been identified previously. The five genes, identified here as genuine sucrose-regulated genes, include three SNF1-like kinases, a pathogen-response related protein and a multidrug resistance ABC transporter (see Additional file 3
). A weak overlap with ABA signaling was detected, since only three sucrose/glucose-regulated genes were also modulated by ABA (Table ). Finally, we noticed that thirteen of the twenty-four genes exhibited opposite regulatory responses in high Brix genotypes and/or mature internodes as compared to the short-term sugar-induced regulation in seedlings (data not shown). Together, these data establish the existence of a correlation between high sucrose content and early sucrose and/or glucose-responsive genes, some of which may be relays of signal transduction pathways triggered by these sugars.
Figure 6 Quantitative PCR (qRT-PCR) analysis of sucrose and glucose responsive genes. The y axis refers to the relative expression ratio between target mRNA versus the reference mRNA (tubulin SCCCRZ1002H03.g) for 3 different experiments in sugarcane thirteen-old (more ...)
In addition, we sought to obtain some insight into the extent to which the short term sucrose and/or glucose regulatory cascade is conserved between sugarcane, a monocot and Arabidopsis thaliana
(Arabidopsis), a model eudicot organism. Therefore, we compared the data obtained in this study on sugarcane seedlings with results described for Arabidopsis
seedlings under similar experimental conditions (3% glucose [30
] or 0,5% sucrose [34
]). Among the twenty-four sugar-regulated sugarcane genes, six of them, along with their eight orthologues in Arabidopsis
(forming five groups of orthologues) were found to be similarly regulated by glucose and/or sucrose (see Additional file 4
and Additional file 5
). The groups of orthologues correspond to SNF1-like kinases (SCRFLR1034G06.g and SCACLR2007G02.g – At1g78290
), two calreticulin genes (SCRFLR2037F09.g – At1g56340
), an auxin/IAA transcription factor gene (SCCCRZ1001G10.g – At3g04730
), a defense and cell wall-related gene encoding a phenyl ammonia-lyase (SCEQRT1024E12.g – At2g37040
) and a dehydrin gene (SCQGLR1085F11.g – At3g50980
) (see Additional file 4
). Furthermore, two Arabidopsis genes, the CUC1/NAC-type transcription factor (At3g1550
) and a wound-responsive gene (At4g10270
) and their closely related respective sugarcane homologues (SCCCLR2003E10.g and SCCCLR2C01F06.g) were found to be similarly regulated by sugars (Table ).