Transcript profiling in M. truncatula lss and sunn-1 mutants reveals different expression profiles despite disrupted SUNN gene function in both mutants

Abstract

A novel autoregulation of nodulation locus in Medicago truncatula, lss, silences the SUNN gene thorough a cis-acting mechanism. Microarray analysis was performed on the Affymetrix Gene Chip^® Medicago Genome Array with cDNA isolated from seven-day-old seedlings of wild type, sunn-1 and lss plants. The results suggest that in lss plants expression of only a few dozen genes differs significantly from wild type while in sunn-1 plants expression of several hundred genes represented by over 800 probe sets is altered. These results suggest that the kinase domain modification caused by the sunn-1 mutation alters the receptor's influence on gene expression and that these differences are present even in the absence of nodulation.

Autoregulation of Nodulation (AON) in legumes relies on long distance signaling from the shoot to the root and vice versa. We recently identified a novel AON locus in Medicago truncatula, lss, that silences the SUNN gene thorough a cis-acting mechanism.¹ SUNN encodes a leucine-rich-repeart receptor kinase with homology to Arabidopsis CLV1,² and acts from the shoot to regulate nodule number in the root.³ The supernodulation phenotype of the lss mutant results from the silencing of SUNN, and we demonstrated that for multiple phenotypes associated with root growth and nodulation, the lss mutant behaves similarly to sunn-4, a putative null mutant of SUNN. Although there is very little SUNN expression in an lss mutant background, there is no difference in methylation in the SUNN promoter between plants carrying the lss mutation and wild type plants, including isolated natural revertants of the lss mutant. Thus the definitive cause of the silencing of SUNN in lss mutants is still undetermined.¹

In the original cloning of SUNN we identified four alleles. The sunn-1 allele was the first identified and the one most often used by other groups due to initial distribution of the mutant to others. The sunn-1 mutation results in an amino acid change in the kinase domain of the protein in a residue highly conserved in serine/threonine and tyrosine kinases.⁴ In contrast, the sunn-4 allele creates a stop codon which should result in truncation of the protein immediately after the initial signal peptide sequence and is presumed to act as a null mutation.² In our work comparing lss to sunn mutants, we discovered that for some phenotypes, the sunn-1 allele was significantly different from the sunn-4 allele and the lss mutation. As part of our characterization of lss, microarray analysis was performed on the Affymetrix Gene Chip^® Medicago Genome Array with cDNA isolated from seven-day-old seedlings of wild type (A-17 ecotype), sunn-1 and lss plants that had been mock-inoculated with buffer 24 hours prior to harvest (Table 1 and Fig. 1A).

Table 1.

Number of microarray probe sets with two fold or greater differential regulation compared to wild type in plants without rhizobia

Genotype	Tissue	Up ≥2x	Up ≥3x	Down ≥ 2x	Down ≥3x	Total ≥ 2x	Total ≥ 3x
sunn-1	Shoots	546	245	107	9	653	254
sunn-1	Roots	109	18	95	18	204	36
sunn-1	In common	18	4	4	1	22	5
sunn-1	Overall	637	259	198	285	835	285
lss	Shoots	8	0	24	18	32	18
lss	Roots	16	6	24	15	40	21
lss	In common	3	0	15	13	18	13
lss	Overall	21	6	33	20	54	26

Number of probe sets misregulated in sunn-1 and lss plants (reduced SUNN expression) compared to wild type A17 plants.

Figure 1.

(A) Comparison of the number of misregulated probe sets in the lss mutant versus the sunn-1 mutant. Data was obtained as in reference ⁷, except that data was normalized across ten independent experiments. DCHIP VERSION 1.3 (www.dchip.org) was used for primary data analyses. Data sets were filtered to eliminate probe sets with intensity values below 100 in both lines being compared and those with an absolute difference in intensity between the two lines of less than 100. Significantly different probe sets were then defined as those with an absolute difference of greater than two-fold and an absolute lower bound cutoff of 1.5. (B) Ribbon diagram of the predicted structure of the catalytic core of the Sunn kinase domain. Shown are the small N-terminal nucleotide binding lobe (yellow), the large C-terminal lobe (purple) and the predicted substrate binding pocket (red). The side chains of Glu857 and Arg950, forming a predicted ion pair between the substrate binding pocket and conserved kinase subdomain XI, are shown in blue. The sunn-1 allele has an R950K missense mutation that is predicted to disrupt the ion pair (see the text). The structure of the SUNN kinase domain was predicted by homology modeling using HHpred and MODELLER (Bioinformatics Toolkit, Max-Plank Institute for Developmental Biology, toolkit.tuebingen.mpg.de/) and visualized using DS ViewerPro 5.0 (www.accelrys.com).⁸,⁹

In these plants 653 probe sets were greater than two-fold misregulated in sunn-1 shoots compared to wild type shoots and approximately one third as many in roots, with 18 probe sets mis-regulated in common between tissues. We were therefore surprised to find that notably fewer probes were misregulated in lss plants: 32 probe sets in shoots and 40 in roots, with 18 in common. Furthermore, of the 54 total probe sets detected as differentially regulated in lss, only nine of these are among the 835 total probe sets detected as differentially regulated in sunn-1 roots and shoots. Among these nine are two probe sets for the SUNN gene, which is poorly expressed in all tissues of the lss mutant¹ and has been shown to have significantly reduced expression in the roots of the sunn-1 mutant.²

Thus our microarray results on plants inoculated with only buffer show that in non-nodulating plants, more genes are misregulated when a mutated kinase is present, as in sunn-1 plants, than when greatly reduced wild type SUNN message is present, as in lss plants. One interpretation is that SUNN may have a signal transduction role prior to (or in the absence of) nodulation. Alternatively, it may be that the kinase-dead mutant establishes aberrant interactions, affecting regulation outside of the usual context. Combined with our earlier finding that sunn-1 has a semi-dominant phenotype³ we interpret this as evidence that SUNN may be part of a larger signal transduction complex.

Structural studies of kinase domains show the highly conserved arginine altered in the sunn-1 kinase (R→K) forms an ion pair with a distant highly conserved glutamic acid located in the substrate binding loop (blue in Fig. 1B). Alteration of this arginine in human Jak2 tyrosine kinase abolished autophosphorylation activity.⁵ In contrast, disruption of the ion pair in the human EGFR kinase domain did not eliminate autophosphorylation but rather rendered the receptor less sensitive to kinase inhibitors and able to phosphorylate targets in the absence of signal.⁶ Either of these could be occurring in sunn-1, resulting in a large number of misregulated genes. In lss plants the molecule produced is normal, but it is produced in very small amounts. This would alter the strength of the signal, but not the nature of the signal. Thus the lss mutant might be expected to have fewer misregulated genes. If such is the case, sunn-1 would be the least suitable allele for determining SUNN autoregulatory function in nodulation because of the large number of genes affected by the altered signal. We propose that stronger alleles of SUNN should be used for future functional studies.

Addendum to: Schnabel E, Mukherjee A, Smith L, Kassaw T, Long S, Frugoli J. The lss supernodulation mutant of Medicago truncatula reduces expression of the SUNN gene. Plant Physiol. 2010;154:1390–1402. doi: 10.1104/pp.110.164889.