Technical Comments

Comment on “A promiscuous intermediate underlies the evolution of LEAFY DNA binding specificity”

+ See all authors and affiliations

Science  06 Feb 2015:
Vol. 347, Issue 6222, pp. 621
DOI: 10.1126/science.1255437


Sayou et al. (Reports, 7 February 2014, p. 645) proposed a new model for evolution of transcription factors without gene duplication, using LEAFY as an archetype. Their proposal contradicts the evolutionary history of plants and ignores evidence that LEAFY evolves through gene duplications. Within their data set, we identified a moss with multiple LEAFY orthologs, which contests their model and supports that LEAFY evolves through duplications.

Gene and genome duplication followed by functional divergence is a major source of genetic diversity and is especially important for transcription factor (TF) evolution (1). Applying this predominant model, most research has focused on how multiple gene copies of TFs undergo selective sub/neo/nonfunctionalization after a duplication event. Sayou et al. used the plant-specific TF LEAFY (LFY) to propose a new model for the evolution of TF binding sites, not through gene duplication but instead through a “promiscuous,” gain-of-function intermediate form (2). By reconciling the phylogeny of LFY orthologs with the prevailing and widely accepted history of plant evolution, however, we argue that the evolution of LFY through gene duplications and ensuing changes in function is more consonant with the findings of Sayou et al. than their proposed “promiscuous” pathway.

Sayou et al. rely exclusively on a gene tree of LFY to make inferences about LFY evolution, suggesting that “[t]he hornwort LFY lineage diverges from a phylogenetic node that lies between the type III [algal] and type I-II [moss, liverwort, and polysporangiophyte] binding specificities.” This proposed “phylogenetic node,” however, is not supported by the literature. Over the past decade, plant biologists have used genomic methods combined with morphological and fossil data to build a strong and widely accepted consensus about the evolutionary relationship among land plants, with very little dissent (37). Liverworts were the first to diverge and are therefore sister to all other embryophytes. Mosses were the second clade to split off. Finally, the ancestors of hornworts and polysporangiophytes (“tracheophytes”) diverged more than 400 million years ago.

It is not surprising that the LFY gene tree does not reflect the consensus species tree of land plants: Any tree built on the sequences of a single gene is most likely to yield evolutionarily incorrect phylogenies (8). Instead, gene evolution must be considered within the constraints of species evolution (9). Sayou et al. do offer a second model of LFY evolution in this context: The promiscuous form evolved before the divergence of liverworts and was then lost twice (in liverworts and again in mosses), remaining only in the hornworts, and then lost again in land plants [Fig. 1A, a reproduction of figure S9 in (2)]. This second model (Fig. 1A) would be less parsimonious than their proposed LFY phylogeny [figure 4 in (2)] because it requires the convergent evolution of type I binding specificities in liverworts and polysporangiophytes. Although this is an intriguing hypothesis, the most parsimonious explanation—and one that follows plant evolutionary history—is that type I binding specificity arose in the early ancestors of land plants, and then new binding specificities evolved in mosses and hornworts (Fig. 1B).

Fig. 1 The LEAFY gene tree within the constraints of plant evolution.

(A) Although not their preferred model, Sayou et al. provide a scenario of LFY evolution that considers the evolutionary history of plants (reproduced here). This phylogeny assumes that the promiscuous LFY form found in hornworts is “intermediate,” evolving after the divergence of algae, and then that new forms evolved separately in liverworts and mosses, were retained in hornworts, and evolved again in land plants. This model requires three changes in LFY binding specificity as well as convergent evolution of type I binding specificities in liverworts and polysporangiophytes. (B) We show that the most parsimonious explanation for LFY evolution is that type I binding arose in the early ancestors of land plants, and then new binding specificities evolved in mosses and hornworts. DNA binding specificity is color-coded: type III (blue), promiscuous (red), type II (green), type I (orange). Each change in binding site specificity is numbered.

Sayou et al. then argue that “the LFY gene probably evolved new DNA binding modes independently of changes in copy number,” a provocative hypothesis. There is substantial evidence that LFY evolves through gene duplications, however. First, as noted by the authors, two LFY homologs are found in gymnosperms: LFY and NEEDLY. These genes are differentially expressed and regulate partially distinct genes, although we do not yet have the genetic tools in gymnosperms to clearly demonstrate their independent functions (10, 11). Second, developmental studies of maize have revealed that its two paralogs of LFY (ZFL1 and ZFL2), the products of a recent whole-genome duplication (5 to 12 million years ago) (12), also have partially nonredundant functions in regulating flowering time and morphology (13, 14). Third, even stipulating that LFY is often a single-copy gene in extant species does not indicate that LFY duplications did not occur in ancestral species. In fact, nonfunctionalization and eventual gene loss of one of the two paralogs is the most common outcome for duplicated genes (15); therefore, it is not surprising to find only one copy of LFY in many species. Furthermore, without thoroughly searching for evidence that LFY has not undergone duplication and then evolved new binding site specificities, which would require including more than two taxa from each of the diverse bryophyte clades and probing genomes for nonfunctional LFY homologs, Sayou et al. cannot make conclusions about whether or not duplications play a role in LFY evolution.

To further test their hypothesis, we mined the same transcriptome databases used by the authors ( (16) to find algal and hornwort orthologs of LFY and identified a moss that has both type I and type II LFY orthologs. We used polymerase chain reaction (PCR) and Sanger sequencing to confirm that Polytrichum commune, the common haircap moss, expresses at least three genes that encode LFY-like proteins (Fig. 2A) (17). After aligning the predicted amino acid sequences with other LFY orthologs, we found that two of the copies have the type II binding specificity (D312-C345-H387), but a third copy has type I binding specificity (H312-R345-H387). Thus, at least in the lineage of P. commune (and, most parsimoniously, in the ancestors of all mosses), LFY duplicated and then evolved two distinct binding specificities (Fig. 2B). This directly contradicts the inference that changes in the mode of LFY binding specificity do not involve gene duplication.

Fig. 2 Revised hypothesis of LEAFY evolution.

(A) Alignment of predicted LEAFY amino acid sequences representing each clade of land plants. Key residues 312 (blue), 345 (magenta), and 387 (yellow) confer distinct DNA binding specificities. P. commune gametophytes express three paralogs of LEAFY, two with type II binding specificity (D312-C345-H387) and one with type I binding specificity (H312-R345-H387). (B) Constrained by the accepted evolutionary history of plants, LEAFY most likely evolved type I binding site specificity (yellow) in the ancestor of all land plants, later gaining new binding site specificities (type II, green) in mosses (through gene duplication) and in hornworts (“promiscuous” type, red) by an unknown mechanism, but likely through gene duplication as well. Tree follows the style of figure 4 in (2).

Considering the current consensus on the history of plant evolution, the process of gene duplications, and the impressive results presented by Sayou et al., the most well-supported hypothesis is that land plants evolved a new form of LFY (type I) sometime around their divergence from algae (Fig. 2B). In mosses, type II LFY evolved after gene duplication, with the type I LFY homolog conserved in some extant species. Separately, hornworts evolved a relatively “promiscuous” type of LFY. We do not know how this form evolved, because this last statement is based solely on evidence from two hornwort species. Additional species sampling may yield insights into how hornworts evolved a novel form of LFY, possibly through the promiscuity pathway suggested by Sayou et al., but probably through ancestral gene duplication.

We commend Sayou et al. for their extraordinary study of LFY binding site specificity in algae, bryophytes, and polysporangiophytes. This type of analysis highlights the importance of including divergent clades when investigating gene/genome evolution and raises new questions about how TFs sub/neo/nonfunctionalize after duplication.

References and Notes

  1. P. commune RNA was extracted (Spectrum Plant Total RNA Kit, Sigma-Aldrich), treated with deoxyribonuclease I (New England Biolabs), used to synthesize cDNA with oligo(dT)20 primers and SuperScript III (Invitrogen), and finally treated with ribonuclease H (Invitrogen). PcLEAFY orthologs were amplified using gene-specific primers (PcLFY1: 5′GGCATCAAAGCAGCAATAAGGG3′ and 5′CCAGCCCAACTGAATCAAAAAGA3′; PcLFY2: 5′GAAGACAAAAAAGTTGCAAAGAAGAAACAG3′ and 5′TGAAACAACCTGCTTGGATTGG3′; PcLFY3: 5′CTTGAAGAAGGAGCTCAGGGTA3′ and 5′CGTAATTGAGAGCTACCAAAATCA3′), and PCR products were sequenced using internal primers (PcLFY1: 5′AAAGCACACAGGAGCTAGTTACA3′ and 5′TGAGCGCATTTGATTGATCT3′; PcLFY2: 5′CTTGTTGATATAACCAGCTCCTG3′ and 5′GCAAATTGCGAAAGAGAAGG3′; PcLFY3: 5′CAAAACATACAGGAGCAGGC3′ and 5′TGATTGGTCTAAATCAAGGCAG3′). Predicted protein translations of sequences were aligned to the LEAFY data set of Sayou et al. (2) with Multiple Alignment using Fast Fourier Transform (MAFFT).
  2. Acknowledgments: We thank Howard M. Goodman for critical discussions of this work. J.O.B. and A.M.R. are supported by predoctoral fellowships from the National Science Foundation, grant DGE 1106400.
View Abstract

Related Content


Navigate This Article