The unique evolution of the programmed cell death 4 protein in plants
© Cheng et al.; licensee BioMed Central Ltd. 2013
Received: 29 May 2013
Accepted: 13 September 2013
Published: 16 September 2013
Skip to main content
© Cheng et al.; licensee BioMed Central Ltd. 2013
Received: 29 May 2013
Accepted: 13 September 2013
Published: 16 September 2013
The programmed cell death 4 (PDCD4) protein is induced in animals during apoptosis and functions to inhibit translation and tumor promoter-induced neoplastic transformation. PDCD4 is composed of two MA3 domains that share similarity with the single MA3 domain present in the eukaryotic translation initiation factor (eIF) 4G, which serves as a scaffold protein to assemble several initiation factors needed for the recruitment of the 40S ribosomal subunit to an mRNA. Although eIF4A is an ATP-dependent RNA helicase that binds the MA3 domain of eIF4G to promote translation initiation, binding of eIF4A to the MA3 domains of PDCD4 inhibits protein synthesis. Genes encoding PDCD4 are present in many lower eukaryotes and in plants, but PDCD4 in higher plants is unique in that it contains four MA3 domains and has been implicated in ethylene signaling and abiotic stress responses. Here, we examine the evolution of PDCD4 in plants.
In older algal lineages, PDCD4 contains two MA3 domains similar to the homolog in animals. By the appearance of early land plants, however, PDCD4 is composed of four MA3 domains which likely is the result of a duplication of the two MA3 domain form of the protein. Evidence from fresh water algae, from which land plants evolved, suggests that the duplication event occurred prior to the colonization of land. PDCD4 in more recently evolved chlorophytes also contains four MA3 domains but this may have resulted from an independent duplication event. Expansion and divergence of the PDCD4 gene family occurred during land plant evolution with the appearance of a distinct gene member following the evolution of basal angiosperms.
The appearance of a unique form of PDCD4 in plants correlates with the appearance of components of the ethylene signaling pathway, suggesting that it may represent the adaptation of an existing protein involved in programmed cell death to one that functions in abiotic stress responses through hormone signaling.
Following transcription and processing of an mRNA, the ribosome is responsible for performing protein synthesis. Although the bacterial 30 S ribosomal subunit can identify the initiation codon through base-pairing between the 3′-end of its16 S ribosomal RNA subunit and the Shine-Dalgarno sequence upstream of the initiation codon, the 40 S ribosomal subunit of the eukaryotic 80 S ribosome requires several translation initiation factors (eIFs) for its binding to an mRNA and to identify the initiation codon [1–3]. eIF4F, which is composed of eIF4E, eIF4A, and eIF4G, is required to promote 40 S subunit binding to an mRNA. While eIF4E binds to the 5′-cap structure, eIF4A is an ATP-dependent RNA helicase that hydrolyzes ATP in order to unwind secondary structure present in the 5′-leader of an mRNA that would otherwise inhibit 40S subunit scanning during its search for the initiation codon . The helicase activity of eIF4A is enhanced by eIF4B which interacts directly with eIF4A [4–6]. eIF4G is a scaffolding protein that interacts with eIF4E, eIF4A, eIF4B, eIF3 (also required for 40S binding to an mRNA), and the poly(A) binding protein (PABP) [2, 4, 7–9]. The interaction of eIF4G with eIF4E bound to the 5′-cap and PABP bound to the poly(A) tail circularizes an mRNA and stimulates translation by promoting 40 S subunit recruitment [10, 11].
eIF4A binds to two regions of eIF4G that fold into HEAT (Huntington, Elongation Factor 3, PR65/A, TOR) domains, characterized by the presence of antiparallel α-helical hairpins known as HEAT repeats [12, 13]. One HEAT domain to which eIF4A binds is present in the middle region of eIF4G (HEAT-1/MIF4G, MIG) which is required for translation while the second is C-distal (eIF4G-MA3, HEAT2/MA3) and serves a regulatory function [14–16]. A third, functionally distinct HEAT domain, present in animal eIF4G but absent from plant eIF4G, does not bind eIF4A but does bind Mnk kinase which phosphorylates eIF4E during active translation . eIF4B and PABP also bind to a site within the central region of eIF4G that partially overlaps the HEAT-1/eIF4G-MIG but they do not bind the HEAT-2/eIF4G-MA3 domain , demonstrating the functional diversity of the HEAT-1/eIF4G-MIG and HEAT-2/eIF4G-MA3 domains in their interactions with partner proteins. In addition to eIF4G, plants express an eIF4F isoform called eIFiso4F, which is composed of eIFiso4E, eIF4A, and eIFiso4G . The eIF4B and PABP interaction sites overlap with the eIF4A binding site in the HEAT-1/eIF4G-MIG domain of eIFiso4G more extensively than in eIF4G . As a consequence, eIF4B and PABP compete with eIF4A for binding to the HEAT-1/eIF4G-MIG domain of eIFiso4G in the absence of the HEAT-2/eIF4G-MA3 domain but not in its presence .
In addition to eIF4G, other proteins containing an MA3 domain have been described. The programmed cell death 4 (PDCD4) protein is characterized by the presence of two tandem MA3 domains that fold into a subtype of HEAT domains. The N- and C-terminal MA3 domains of human PDCD4 contain four and three helical hairpins, respectively, while eIF4G-MA3 contains five helical hairpins . From a basal level, PDCD4 expression is induced upon programmed cell death in several cell types in mice, including lymphoid and neuronal cells . Overexpressing PDCD4 was unable to induce programmed cell death, suggesting no casual relationship between PDCD4 and apoptosis . Increasing PDCD4 expression, however, was sufficient to inhibit tumor promoter-induced neoplastic transformation while reducing PDCD4 expression resulted in a transformation-sensitive phenotype and the promotion of tumor invasion [22–24]. Animal PDCD4 binds eIF4A which inhibits the latter from binding to eIF4G-MA3 [25–27]. PDCD4 inhibits eIF4A RNA binding and helicase activity and inhibits translation in vivo[20, 26, 28]. PDCD4 binding to eIF4A also disrupts the ATP-binding site and prevents its conformational transition from a nonproductive open state to a productive closed state . PDCD4 binding to eIF4A is required for its ability to inhibit translation and transformation as disruption of eIF4A binding to PDCD4 abolishes its effect on both . The PDCD4 C-terminal MA3 domain contacts the N-terminal domain of eIF4A using structural features conserved with eIF4G-MA3 thereby preventing the interaction between eIF4A and eIF4G-MA3 and inhibiting translation initiation [29, 30]. The two MA3 domains of PDCD4 have similar secondary and tertiary structures and either can compete with eIF4A in binding eIF4G-MA3 . Although both MA3 domains of PDCD4 are structurally similar as are their eIF4A-binding surfaces, the two domains function synergistically to bind eIF4A, resulting in a more stable complex with eIF4A . A single PDCD4 MA3 domain is sufficient to inhibit translation but both domains are required to compete effectively with eIF4G-MA3 for binding to eIF4A . PDCD4 also binds eIF4G-MIG without affecting eIF4A binding as they bind to diametrically opposite sides of eIF4A . Binding of PDCD4 to eIF4G-MIG, however, may anchor the binding of eIF4A to eIF4G-MIG thereby preventing its cycling through eIF4G as part of its function during translation initiation .
PDCD4 homologs are present in lower eukaryotes and plants although no PDCD4 homolog has been reported in yeast. PDCD4 proteins in higher plants are unique in that they contain four tandem MA3 domains. Higher plants also appear to lack a two MA3 domain PDCD4 homolog. One such four MA3 domain PDCD4 protein (ECIP1) was reported to bind Arabidopsis thaliana ethylene receptors, ETR2 and EIN4, as well as EIN2, a downstream component of the ethylene signaling pathway required for the induction of ethylene responses . Loss of ECIP1 expression resulted in increased ethylene sensitivity and tolerance to salt . Thus, like the HEAT-2/ eIF4G-MA3 domain of eIF4G and eIFiso4G, the MA3 domains of ECIP1 are involved in protein-protein interactions. When the PDCD4 homolog containing four MA3 domains first appeared during plant evolution has not been examined. In this study, we have examined the evolution of plant PDCD4-like proteins from a two MA3 domain form in prasinophytes to the appearance of a four MA3 domain form in charophytes and land plants and the likely independent appearance of a four MA3 domain form in Chlamydomonadales. We also examine the expansion and divergence of the PDCD4-like gene family during plant evolution and identify how one distinct gene family member appeared following the evolution of basal angiosperms. How the expansion and divergence of the plant PDCD4 gene family might relate to their function as regulators of ethylene responses is also discussed.
PDCD4 in animals and many lower eukaryotes, including several algal species, contains two MA3 domains while the PDCD4 homolog in other algal species and land plants contains four MA3 domains. This raises the question of when duplication of the two MA3 domain protein into a four MA3 domain protein may have occurred. Examination of the available genome sequences for algae and algal relatives revealed the presence of four MA3 domain PDCD4 homologs in more recently evolved green algae such as Chlamydomonas reinhardtii and Volvox carteri representing the Chlamydomonadales but only two MA3 domain PDCD4 homologs in algal species of the Mamiellales such as Micromonas and Ostreococcus (Additional file 2). Only two MA3 domain PDCD4 homologs were observed in heterokonts such as the brown alga Ectocarpus siliculosus and in algal relatives such as Aureococcus anophagefferens, Phaeodactylum tricornutum, Thalassiisira pseudoonana, Phytophthora species, and Albugo laibachii (Additional file 2).
Phylogenetic analysis of MA3 domains 1–2 and MA3 domains 3–4 of C. reinhardtii and V. carteri PDCD4 with the two MA3 domain PDCD4 homologs from other algae and algal relatives revealed that C. reinhardtii and V. carteri MA3 domains 1–2 were more similar to the two MA3 domain PDCD4 homologs of prasinophytes or heterokonts than were MA3 domains 3–4 (Additional file 3). MA3 domains 1–2 and MA3 domain 3 of a Chlorella variabilis partial PDCD4 protein sequence also clustered with the respective MA3 domains of C. reinhardtii and V. carteri PDCD4 (Additional file 3). These data suggest that the four MA3 domain PDCD4 homolog in more recently evolved chlorophytes arose from the duplication of a more ancestral, two MA3 domain protein and that, following duplication, MA3 domains 3–4 diverged to a greater extent than did MA3 domains 1–2.
As land plants did not evolve from Chlamydomonadales but rather from charophytes (i.e., fresh water algae), the appearance of a four MA3 domain PDCD4 homolog in C. reinhardtii and V. carteri likely represents an independent event from the appearance of the four MA3 domain PDCD4 homolog in the land plant lineage. This is because charophytes and chlorophytes (which include C. reinhardtii and V. carteri) are thought to have derived from a common ancestor that likely contained a two MA3 domain PDCD4 form of the protein that was maintained in older algal lineages (e.g., those in Mamiellales) but underwent duplication during later chlorophyte evolution.
List of plant species and PDCD4 sequences used with the number of MA3 domains present in each protein
Number of MA3 domains
Micromonas sp. RCC299
Gene divergence often follows gene duplication. To determine how the plant PDCD4 gene family members may have diverged during the evolution of land plants, phylogenetic analysis of the gene family was performed. Because of the poor sequence conservation of the N-terminus, only the region containing the four MA3 domains was used for the analysis. Although the C. globosum sequence is not full-length, it was included to represent a progenitor sequence to land plants. Similarly, one of the P. abies PDCD4 homologs (MA_10693g0010) is not full-length but was included along with the second P. abies homolog (MA_2413g0010), which is full-length, to represent the two PDCD4 gene members of a more recent gymnosperm species.
These data suggest that the two genes produced from the duplication of the original PDCD4 gene during early higher plant evolution might have been the progenitor genes for MAT8 and the MAT5/6/7 subgroups and the latter subgroup underwent expansion in some species. However, the considerable similarity of the two PDCD4 homologs in P. patens suggests that gene duplication occurred relatively recently in this species. In contrast, the two homologs in S. moellendorffii have diverged suggesting an earlier gene duplication event. The two gene members of P. patens and S. moellendorffii are sufficiently different from the MAT8 and MAT5/6/7 subgroups that it is not possible to assign them to either subgroup. Moreover, the gymnosperm PDCD4 homologs cluster in a subclade that is separate from the MAT8 and MAT5/6/7 subgroups, including the two homologs of P. abies, suggesting that the gymnosperm PDCD4 homologs have diverged from angiosperms homologs. No MAT8 homolog was identified in basal angiosperm species as the PDCD4 homologs in the basal angiosperms Amborella trichopoda, Austrobaileya scandens, and Illicium floridanum are more similar to the MAT5/6/7 subgroup than the MAT8 subgroup (Figure 6). A MAT8-like homolog was identified in Aristolochia elegans and Eschscholzia californica in addition to MAT5/6/7-like homologs (Figure 6). These observations suggest that MAT8 subgroup is not present in basal angiosperms but appeared at or prior to the evolution of species within the Piperales such as A. elegans.
As these deletions are not present in the early land plants P. patens and S. moellendorffii but appear in the early lineage of both gymnosperms and angiosperms only to disappear again during subsequent evolution of each lineage, the most parsimonious explanation is that the deletions in the MA3 domains of early evolved gymnosperm and angiosperm species likely occurred independently in the evolution of these species as they branched from the ancestral gymnosperm and angiosperm lineages. It is also formally possible that the deletions appeared subsequent to early land plant evolution but prior to the appearance of the distinct gymnosperm and angiosperm lineages only to disappear again during the subsequent evolution of each lineage although this possibility would seem less likely. Despite the deletions present in the basal angiosperm PDCD4 homologs, they are more similar to the MAT5/6/7 subgroup than the MAT8 subgroup (Figure 6), supporting the conclusion that the MAT8 member of the PDCD4 gene family evolved following basal angiosperm evolution and is specific to more recently evolved angiosperm species.
These results suggest that the domain organization of PDCD4 homologs in most lower plants, i.e., algae and algal relatives, is similar to those in animals in that it contains two MA3 domains but it underwent duplication to a four MA3 domain form in some recently evolved chlorophytes such as C. reinhardtii and V. carteri and, independently, in at least one charophyte such as C. globosum. That land plants did not evolve from chlorophytes, particularly from those species that contain a four MA3 domain PDCD4 homolog, supports the possibility of an independent duplication event during charophyte evolution. The sequence similarity between higher plant and C. globosum PDCD4 homologs at the fusion site between MA3 domain 2 and 3 and the significant sequence difference between higher plant and C. reinhardtii and V. carteri PDCD4 homologs in this same region further supports this conclusion. In addition to the duplication of PDCD4 from a protein containing two MA3 domains to a protein containing four MA3 domains, the PDCD4 gene family expanded from a single member in algae to two or more members upon colonization of land and the evolution of higher plants. The expansion of the gene family occurred early during higher plant evolution as evidenced by the two member gene family present in bryophytes and lycophytes with additional expansion of the gene family in some higher plant species that appears to have occurred in a species-specific manner.
The PDCD4 homologs of the angiosperm species examined have diverged into MAT8-like and MAT5/6/7-like subgroups. No MAT8 homolog was identified in gymnosperm or basal angiosperm species, but appeared at or prior to the evolution of species within the Piperales such as A. elegans, suggesting that the MAT8 homolog is an angiosperm-specific member that evolved after the appearance of the most basal angiosperm species. Although no further expansion of the gene family appears to have occurred within the MAT8 subgroup as only one gene encoding a MAT8-like homolog was observed in those species whose PDCD4 gene family included a MAT8-like member, additional expansion did occur in the MAT5/6/7 subgroup in some dicot and monocot species in a species-specific manner resulting in the paralogs observed in this subgroup. For example, orthologs of A. thaliana MAT7 are found in A. lyrata and C. rubella but not in T. halophila and B. rapa, suggesting it appeared following the separation of Arabidopsis/Capsella from Brassica. Other recent gene duplication events may have occurred in P. patens and G. raimondii. In contrast, the expansion of the MAT5/6/7 subgroup in some monocots such as O. sativa, S. bicolor, and B. distachyon predated their speciation as each contains two genes within the MAT5/6/7 subgroup which cluster into two distinct clades. All PDCD4 homologs identified in basal angiosperm species were MAT5/6/7-like, supporting the conclusion that this form of PDCD4 evolved prior to the appearance of MAT8-like homologs.
Interestingly, the PDCD4 homologs in early evolved gymnosperm and angiosperm species possess two deletions within each MA3 domain relative to early land plants. The observation that these deletions are present in both early evolved gymnosperm and angiosperm species supports the notion that they occurred prior to the appearance of the distinct gymnosperm and angiosperm lineages. However, the fact that these deletions are not observed in angiosperm species since (and including) the evolution of A. elegans and are not present in more recently evolved gymnosperm species such as P. abies and P. taeda, suggests that these deletions in the PDCD4 homologs of early evolved gymnosperm and angiosperm species likely occurred independently during the evolution of these species after they branched from the ancestral gymnosperm and angiosperm lineages. If they did occur independently in early gymnosperm and angiosperm species, it would suggest convergent evolution which may have been driven by the deletion of structural or functional elements within each MA3 domain that may have conferred an advantage to these species. The importance of the deleted regions within each MA3 domain is unknown but the elucidation of their structural or functional impact on the MA3 domains will provide insight into the adaption of PDCD4 proteins in early evolved gymnosperm and angiosperm species.
These observations collectively raise the question of what advantage does a four MA3 domain PDCD4 protein confer to recently evolved algal species and land plants and what advantage does expansion of the PDCD4 gene family confer to plants as they transitioned from an aquatic to a terrestrial environment? Loss of expression of just one of the four PDCD4 proteins in A. thaliana was sufficient to alter ethylene sensitivity and tolerance to salt . The ability of MAT5 (ECIP1) to bind ethylene receptors and the EIN2 signaling component, suggests that plant PDCD4 may regulate ethylene signaling which in turn regulates responses to abiotic stresses such as salt . Whether plant PDCD4 may interact also with eIF4A to inhibit protein synthesis as has been reported for animal PDCD4 remains to be determined. Moreover, its function in programmed cell death in plants has not been examined. An interesting correlation, however, is the presence in land plants and their closest algal relatives of a PDCD4 protein containing four uninterrupted MA3 domains with components involved in ethylene biosynthesis and signaling including homologs to the ETR2 and EIN4 ethylene receptors and the EIN2 downstream signaling component to which MAT5 binds [31, 34, 35]. This raises the intriguing possibility that the appearance of the four MA3 domain protein may represent the evolution of a protein involved in programmed cell death to one involved in abiotic stress-related hormone signaling similar to the evolution of ethylene receptors from two-component environmental sensor regulators . The expansion of the PDCD4 gene family may have provided specificity of function if different PDCD4 isoforms are involved in different pathways or interact with different receptors within a single pathway such as in ethylene signaling. This would be particularly important following colonization of land which presents a more diverse array of stress conditions, such as desiccation, UV radiation, and temperature fluctuations, than would be present in many aquatic environments. Arabidopsis species contain five distinct ethylene receptors and MAT5 (ECIP1) interacts with just two of these, i.e., ETR2 and EIN4 (as well as EIN2). Whether other members of the PDCD4 gene family exhibit different specificities in their interactions with ethylene receptors, EIN2, or other proteins remains to be determined but the divergence of the MAT8 subgroup and the MAT5/6/7 subgroup might provide the basis for any functional specificity that exists within this family.
MA3-containing sequences obtained from Arabidopsis thaliana were used to query genome-wide analysis and comparative studies. MA3-containing sequences were obtained by BLAST searches of the NCBI  Phytozome v9.1 , and the Spruce Genome Project  protein, genome, and EST databases where appropriate. The Phytozome BLAST search implements NCBI Blast (v2.2.13). Reiterative searches of a particular species were performed using initial MA3-containing sequences from a species or from closely related species. Basal angiosperms and early evolved gymnosperm homologs were obtained from http://www.onekp.com. Predicted protein sequences from genomic and EST sequences were obtained using the ExPASy Translate tool . Protein alignments were performed by Clustal Omega  with manual adjustments. Sequences queried included dicot and monocot plant genomes representing a diverse array of plants groups such as Arabidopsis relatives (Capsella rubella, Thellungiella halophila, and Brassica rapa), legumes (Glycine max and Medicago truncatula), the castor oil plant (Ricinus communis), cereals and grasses (Zea mays, Sorghum bicolor, Oryza sativa, Hordeum vulgare, and Brachypodium distachyon), fruits and vegetables (Vitis vinifera and Solanum lycopersicum), cotton (Gossypium raimondii), trees (Populus trichocarpa and Theobroma cacao), basal angiosperms (Amborella trichopoda, Austrobaileya scandens, and Illicium floridanum), gymnosperms (Cycas micholitz, Sundacarpus amarus, and Picea abies), bryophytes and lycophytes (Physcomitrella patens and Selaginella moellendorffii), green marine and fresh water algae (Chlamydomonas reinhardtii, Volvox carteri, Chlorella variabilis, Micromonas species, Ostreococcus species, and Chaetosphaeridium globosum), and stramenopiles or algal relatives (Ectocarpus siliculosus, Aureococcus anophagefferens, Phaeodactylum tricornutum, Thalassiisira pseudoonana, Phytophthora species, and Albugo laibachii). A list of protein sequences and cognate genes used for the comparative analysis is provided in Table 1. Maximum-likelihood phylogenetic trees were constructed using the PhyML software (v3.1)  with 1000 bootstrap replicates. The default LG amino acid replacement matrix  was used. Numbers included on each branch denote percentages of bootstrap support. Aligned sequences used for the phylogenetic analysis of Figure 2 are presented in Additional file 4; for the phylogenetic analysis of Additional file 3 in Additional file 5; and for the phylogenetic analysis of Figure 6 in Additional file 6.
EIN2 C-terminal interacting protein 1
Translation initiation factor
Huntington, Elongation Factor 3, PR65/A, TOR
Poly(A) binding protein
Programmed cell death 4.
The authors thank D. Soltis, J. Leebens-Mack, Tao Chen, T. Kutchan for sequences from the Onekp project. This work was funded by a grant from the National Science Foundation (DBI-0820047) and the University of California Agricultural Experiment Station.
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.