Transgenic tomato plants with altered expression of the gene were prepared in order to carry out a functional characterization of TM8. In particular, in the TM8 over-expressing plants macroscopic anomalies were found in whorl 3, and they basically consisted of splayed out stamens with poorly viable pollen. Regarding MADS-box genes, splayed out and sterile stamens were first reported by Pnueli et al. [28] in antisense tomato plants with down-regulated expression of TM5 which is an E-function gene. In these plants other MADS-box genes, TM6 and TAG1 among them, were unaffected in their expression. Also Ampomah-Dwamena et al. [14] obtained splayed out and sterile stamens in tomato plants where TM29, another E-function gene, had been ectopically down-regulated, and such down-regulation did not affect the expression of the TM5 and TAG1 MADS-box genes. Finally, splayed out stamens were found in tomato plants with either decreased or missing expression of various B-function genes [12],[29].
The above findings are particularly interesting because they show that in the same species both the morphology and the functionality of stamens could be affected in an apparently similar manner by a decreased expression of either one or another of different types of MADS-box genes. Therefore, since the down-regulation of a single gene could cause the same anomalous phenotype, it appears that in tomato the protein products of both the B-function genes and the two E-function genes TM5 and TM29 interdependently participate in the process that leads to the differentiation of the third whorl.
The results obtained in this work add further complexity to the molecular network involved in the differentiation of whorl 3 in tomato. In fact, besides the four B-function and the two E-function genes already mentioned, also SlMBP21, another E-function gene, and both TAG1 and TAGL1, two C-function genes, and MACROCALYX, a putative A-function gene, appeared to be significantly down-regulated in the anomalous splayed out stamens produced by the TM8 over-expressing tomato plants. Since all the above genes had significantly reduced expression in the anomalous stamens, this finding suggests that in whorl 3 there must be a specific combination and dose equilibrium of various MADS-box proteins in order to have a correct differentiation of stamens. In petals (whorl 2) all the four B-function genes and MACROCALYX appeared significantly up-regulated in the TM8 over-expressing petals. On the contrary, a mixed situation was found in the TM8:SRDX expressing petals: the SlGLO1 and SlGLO2 genes appeared similarly up-regulated and the TM6 and TAP3 genes showed no variations in expression, while MACROCALYX showed a decreased expression. Since no significant morphologic difference could be evidenced in petals of both types of transgenic plants, it appears that the observed changes in the expression of both MACROCALYX and the B-function genes is not sufficient for significantly altering the petal morphology.
However, the data regarding the expression of MC in our transgenic plants suggest that this gene must play some role in the differentiation of the whole tomato flower. In fact, even though MC is considered a putative A-function gene, its expression in the fully differentiated flowers was not restricted to the first two whorls, as one would expect on the basis of the canonical ABC model [1]. On the contrary, MACROCALYX was expressed in all 4 whorls, and had a significantly changed expression in both the splayed out anthers and the anomalous ovaries, respectively.
The identity of flower organs is specified by various MIKC MADS-box transcription factors which act in a combinatorial manner [1]. The molecular networks formed by these proteins have been extensively explored using yeast two hybrid assays [26],[30]. Such studies have been performed also in tomato [12],[15],[29], however the possible interactions of the TM8 protein with other MADS-box transcription factors was never examined. We therefore decided to use this technique to identify those tomato MADS box proteins able to form heterodimers with TM8.
Unlike Arabidopsis, tomato has two AP3-like proteins, TAP3 and TM6 [3],[11], and two PI-like proteins, SlGLO1 and SlGLO2 [12], which represent the B-class function. None of the B type MADS box proteins was able to form dimers with TM8 in our yeast two hybrid assays. Yeast two hybrid assays also excluded that TM8 is able to homo-dimerize. TM8 did not interact with TAGL1 and TAG1 either. Regarding TAG1, we recorded a weak activation of the ADE2 reporter gene, but we were not able to observe growth on media lacking histidine, which suggests that HIS3 was not activated. Therefore it appears unlikely that TM8 and TAG1 may form dimers in vivo.
TM8 did not physically interact in yeast with those tomato SEPALLATA-like MADS box protein [TM5, TM29 and SlMP21 [31]] that had altered expression patterns in our transgenic plants, and also with JOINTLESS [19]. Interestingly, TM8 was able to interact in yeast with MACROCALIX, and the dimer TM8-MC could promote the transcription of both reporter genes, ADE2 and HIS3. This confirms that the chimeric TM8 protein used for yeast assays is properly folded.
Twenty years after the proposal of the ABC model, a modification was introduced by Causier et al. [32] in the review “Floral organ identity: 20 years of ABCs” to account for the absence of an A-function in most plant species. Schwarz-Sommer and co-workers introduced a new (A)-function [33] important to define the floral meristem identity and to produce the sepals that are considered as the ground state of floral organs. Our data seem to suggest that this model might apply also to the tomato flower, and that the activity of the TM8 protein might be mediated by interactions with the MACROCALYX protein.
The over-expression of the TM8:SRDX repressor chimera had macroscopic effects on both reproductive and vegetative structures. Although it has been shown in tomato [5] and in two Gymnosperms [34] that TM8-like genes are expressed also in leaves, the latter finding was unexpected because Lifschitz et al. [7] had reported anomalies only for the reproductive structures in their TM8 antisense tomato plants, while in our transgenic plants the leaves showed a marked epinasty and were greener compared to the untransformed ones. Also the flower peduncles were different compared to wild-types since they did not differentiate a correct abscission zone. To the latter purpose, it is known that a correct expression of the MADS-box gene JOINTLESS is necessary for the differentiation of a normal abscission zone in the tomato flower peduncle [25] and, as expected, in the anomalous abscission zones of the TM8:SRDX flower peduncles also the expression of JOINTLESS appeared significantly reduced, in agreement with the defective abscission zones.
In plants the physiological activity of a given hormone may also depend on its interactions with other hormones present in the same tissue, and this has been shown several times for ethylene and auxin [35],[36]. In tomato it was demonstrated that SlIAA3, a gene coding for an Aux/IAA protein, can be positively regulated by both auxin and ethylene, and antisense tomato plants for this gene had a reduced epinastic response compared to wild-type ones when treated with exogenous ethylene [36]. Interestingly, in the TM8:SRDX expressing plants the epinastic leaves had a significantly increased expression of the SlIAA3 gene, while a significantly reduced transcript amount was found in the lengthened flower peduncles. Therefore, the inability to measure the ethylene produced by the epinastic leaves might simply reflect the need for ethylene to just activate the expression of the SlIAA3gene, therefore the hormone had not to be produced in enormous amounts. Suggestions about a possible auxin involvement came also from other phenotypic characteristics of the 35S:TM8:SRDX plants, like the elongated fruits, their parthenocarpy and the deep green color of the foliage. In fact, in these anomalous situations other genes involved in the signal transduction pathway of auxin showed an expression that was altered as expected on the basis of their demonstrated function [22]-[24]. However, the possible connection between TM8 and auxin remains elusive.
As regards the phenotypic anomalies of the reproductive structures observed in the TM8:SRDX plants, they appeared to affect only whorl 4. In particular, all the fruits were parthenocarpic, a characteristics already described by Lifschitz et al. [7] for their TM8 antisense tomato plants. On the contrary, the stamens had a normal appearance and the pollen viability was comparable with that of the wild-types, therefore the anomaly was evidently due to problems in the carpel whorl. Actually, transgenic ovaries had an elongated shape that was maintained till the end of their development so that also ripe fruits had an ellipsoidal shape instead of being roundish like the untransformed ones. The shape of tomato fruits is under the control of various genes [37], in particular a low expression of the OVATE gene has been shown to be responsible for the formation of pear-shaped tomatoes [25]. Recently, Rodriguez et al. [37] evidenced that OVATE may also be involved in the formation of ellipsoidal tomatoes, which appears to be the case also for the35S:TM8:SRDX fruits since the OVATE gene had a significantly decreased expression in ovaries and very young fruits, that is when the fruit shape is established.
The C-function TAG1 gene was shown by Pnueli et al. [38] to be expressed in stamens and carpels, and to be of basic importance for a correct differentiation of these two organs. In particular, they found that a down-regulated expression of the gene caused the appearance of relevant malformations, among which both male and female sterility were reported. In tomato TAG1 and TAGL1 are the genuine C function genes while TAGL11 and SlMBP3 are D-function genes. It is interesting to note that the expression of both TAG1 and TAGL1 was consistent with the role played by them during the differentiation of reproductive structures. In the TM8 over-expressing plants the two genes had significantly decreased expression in the anomalous stamens but not in the normal ovaries, on the contrary in the TM8:SRDX expressing plants the two genes had normal expression levels in stamens and significantly reduced expression levels in the anomalous ovaries. Since in the latter ovaries both the D-function and the E-function genes did not show any significantly varied expression compared to wild-type, the above data reinforce the role played by the TAG1 and TAGL1 genes in the development of tomato carpels [17],[38],[39].
