Nonreciprocal complementation of KNOX gene function in land plants

Abstract

Class I KNOTTED-LIKE HOMEOBOX (KNOX) proteins regulate development of the multicellular diploid sporophyte in both mosses and flowering plants; however, the morphological context in which they function differs. In order to determine how Class I KNOX function was modified as land plants evolved, phylogenetic analyses and cross-species complementation assays were performed. Our data reveal that a duplication within the charophyte sister group to land plants led to distinct Class I and Class II KNOX gene families. Subsequently, Class I sequences diverged substantially in the nonvascular bryophyte groups (liverworts, mosses and hornworts), with moss sequences being most similar to those in vascular plants. Despite this similarity, moss mutants were not complemented by vascular plant KNOX genes. Conversely, the Arabidopsis brevipedicellus (bp-9) mutant was complemented by the PpMKN2 gene from the moss Physcomitrella patens. Lycophyte KNOX genes also complemented bp-9 whereas fern genes only partially complemented the mutant. This lycophyte/fern distinction is mirrored in the phylogeny of KNOX-interacting BELL proteins, in that a gene duplication occurred after divergence of the two groups. Together, our results imply that the moss MKN2 protein can function in a broader developmental context than vascular plant KNOX proteins, the narrower scope having evolved progressively as lycophytes, ferns and flowering plants diverged.

Keywords: Ceratopteris richardii; Physcomitrella patens; Selaginella kraussiana; Arabidopsis; KNOTTED homeobox genes; cross-species complementation; land plant evolution; phylogeny.

Figure 1

Maximum‐likelihood and Bayesian phylogenetic analysis of KNOTTED‐LIKE HOMEOBOX (KNOX) homologues. (a, b) Phylogenetic trees inferred with (a) RA x ML or (b) MrBayes, using a partitioned nucleotide dataset of 59 + 2 sequences (Supporting Information Fig. S1) evolving under the GTR + Γ + I model. Support values (a, bootstrap values; b, posterior probabilities) are indicated next to the corresponding branch. To avoid destabilization at the base of the Class I clade by Klebsormidium flaccidum sequences, kfl00113_0180 and kfl00118_0070 were added to the maximum‐likelihood tree after the phylogeny was inferred using the evolutionary placement algorithm of RAxML. As a consequence, K. flaccidum sequence branches are dashed and their associated likelihood weight ratios (Stamatakis, 2014) are indicated in italics. Each sequence was placed on its recipient tree branch with respect to the actual distance derived from a maximum likelihood tree inferred using all 61 sequences (Fig. S2). Colour‐coded boxes correspond to the species’ phyla. Both trees were rooted using the Chlorophyta clade as an outgroup.

Figure 2

Transgene integration and expression in Physcomitrella patens mkn triple mutants. (a) DNA gel blot analysis of NdeI digested genomic DNA from mkn2;mkn4;mkn5 mutants transformed with MKN2pro:KNOX constructs. Blots were hybridized with a fragment of the blasticidin resistance gene present in the transgene construct (see Supporting Information Fig. S2). Hybridizing fragments were not detected in DNA samples from nontransformed mutants (NT), whereas 11.8‐kb (MKN2pro:MKN2), 10‐kb (MKN2pro:BP), 11.59‐kb (MKN2pro:STM), 9.9‐kb (MKN2pro:CrKNOX1), 11.78‐kb (MKN2pro:CrKNOX2), 9.42‐kb (MKN2pro:SkKNOX1) and 9.72‐kb (MKN2pro:SkKNOX2) fragments were detected in transgenic lines. (b–h) reverse transcription polymerase chain reaction (RT‐PCR) showing transgene transcript accumulation in mkn2;mkn4;mkn5 mutant plants transformed with (b) MKN2pro:MKN2, (c) MKN2pro:BP, (d) MKN2pro:STM, (e) MKN2pro:CrKNOX1, (f) MKN2pro:CrKNOX2, (g) MKN2pro:SkKNOX1 and (h) MKN2pro:SkKNOX2. GAPC1 was used as an amplification and loading control in each case. Amplifications were performed for 29, 32 and 35 cycles (as indicated) to ensure amplification in the exponential phase. Negative control reactions were carried out in the absence of reverse transcriptase (‐RT) and with cDNA isolated from nontransformed triple mutants (NT).

Figure 3

Cross‐species complementation tests with Physcomitrella patens mkn triple mutants. (a–i) Representative sporophyte phenotype of (a) wild‐type (WT), (b) mkn2;mkn4;mkn5 mutant, and (c–i) mkn2;mkn4;mkn5 mutants transformed with (c) MKN2pro:MKN2, (d) MKN2pro:BP, (e) MKN2pro:STM, (f) MKN2pro:CrKNOX1, (g) MKN2pro:CrKNOX2, (h) MKN2pro:SkKNOX1 and (i) MKN2pro:SkKNOX2 constructs. Bars, 200 μm. (j, k) Schematic depiction of measurements taken (j) and average ratios of sporophyte surface area to the area occupied by spores (k) for the lines exemplified in (a–i). Thirty sporophytes from each of three independent transformed lines were analysed for each construct, with the exception of MKN2pro:CrKNOX2 for which only two independent lines were recovered. Error bars represent ± SEM.

Figure 4

Cross‐species complementation tests with Arabidopsis bp‐9 mutants. (a–h) Representative inflorescence phenotype of (a) wild‐type (WT), (b) bp‐9 mutant, and (c–h) homozygous bp‐9 mutant plants transformed with (c) BPpro:BP, (d) BPpro:MKN2, (e) BPpro:SkKNOX1, (f) BPproSkKNOX2, (g) BPpro:CrKNOX1 and (h) BPpro:CrKNOX2 constructs. Bars, 2 cm. (i) Boxplot showing range of silique angles on the primary inflorescence of lines exemplified in (a–h). Primary inflorescences from 10 plants of at least three independent lines per transgene were analysed. Red asterisks indicate lines that are significantly different (P < 0.05) from WT. All lines are significantly different (P < 0.05) from bp‐9 mutants. Note that blue points on the y‐axis are not to scale relative to those in black. (j–o) Quantitative reverse transcription polymerase chain reaction (qRT‐PCR) analyses of (j) BPpro:BP, (k) BPpro:MKN2, (l) BPpro:SkKNOX1, (m)BPpro:SkKNOX2, (n) BPpro:CrKNOX1 and (o) BPpro:CrKNOX2 transgene expression. Numbers on the x‐axis represent independent transgenic lines as indicated in (i). The y‐axis represents transcript levels (N ₀) normalized against the EF1Balpha2 and UBP6 genes. Three technical replicates were performed for three different biological samples of each independent line. Scatter blots represent values of individual biological replicates; bars represent mean of three biological replicates.

Figure 5

Maximum‐likelihood and Bayesian phylogenetic analysis of BELL homologues. (a, b) Phylogenetic trees inferred with (a) RA x ML or (b) MrBayes, using a protein alignment evolving under the JTT + Γ + I model. Support values (a, bootstrap values; b, posterior probabilities) are indicated next to the corresponding branch. In order to avoid destabilization of the early divergent clades, the identified BELL homologue in Marchantia polymorpha was placed using the evolutionary placement algorithm of RA x ML and the maximum‐likelihood tree as recipient. As a consequence, the M. polymorpha BELL sequence branch is dashed and the associated likelihood weight ratio (Stamatakis, 2014) is indicated in italics. Colour coded boxes correspond to the species phyla. (a) The maximum‐likelihood tree was rooted using the M. polymorpha BELL homologue and (b) the Bayesian tree was rooted using the moss clade.