PERSpECTIvES

metabolism of a plant. In this context, OPINION it might be supposed that maladaptive mutations would be quite common and The genetics of convergent evolution: that the evolutionary trajectory of each CCM might be exceedingly narrow and rarely stumbled upon24,25. On the other insights from plant photosynthesis hand, the sheer number of CCM origins argues precisely the opposite — these traits Karolina Heyduk , Jose J. Moreno-Villena,​ Ian S. Gilman, Pascal-Antoine Christin must be relatively simple to evolve and and Erika J. Edwards might be assembled via multiple and varied trajectories and, as such, are evolutionarily Abstract | The tree of life is resplendent with examples of convergent evolution, accessible (sensu Maynard Smith26,27). whereby distinct species evolve the same trait independently. Many highly In this Perspective, we highlight evidence convergent adaptations are also complex, which makes their repeated emergence that supports the view that CCMs are surprising. In plants, the evolutionary history of two carbon concentrating evolutionarily accessible, with a focus on the mechanisms (CCMs) — C and crassulacean acid metabolism (CAM) photosynthesis genetic elements that might give rise to their 4 repeated origins. We first review shared key — presents such a paradox. Both of these modifications of ancestral C3 components of and differences between C4 photosynthesis require the integration of multiple anatomical and biochemical and CAM photosynthesis, both of which use components, yet together they have evolved more than one hundred times. similar biochemical pathways. We highlight The presence of CCM enzymes in all plants suggests that a rudimentary CCM recent comparative studies in well-sampled​ might emerge via relatively few genetic changes in potentiated lineages. Here, we clades that suggest that a newly emerging, rudimentary CCM may be less complex propose that many of the complexities often associated with C4 and CAM than many of the highly optimized C4 and photosynthesis may have evolved during a post-​emergence optimization phase. CAM species that are commonly studied. The ongoing development of new model clades for young, emerging CCMs is We briefly review the roles of CCM enzymes enabling the comparative studies needed to test these ideas. in C3 species to emphasize that, at the broadest scale, all plants routinely express the essential biochemical building blocks of a (ref.8) Convergent evolution has been pervasive CO2 and O2 , and the processing of fixed functional CCM pathway. In this context, we throughout the history of life. Even very O2 releases CO2 and results in energetically discuss new evidence that some lineages may complicated adaptations, such as camera wasteful reactions9. To circumvent this be genetically predisposed to evolve CCMs. eyes in animals1, sex determination systems problem, land plants have repeatedly evolved Throughout the article, we emphasize how in eukaryotes2 and eusociality in insects3, carbon concentrating mechanisms (CCMs) the growing field of comparative genomics is have evolved multiple times. In plants, known as C4 photosynthesis and crassulacean already catalysing discovery and improving convergence has led to repeated transitions acid metabolism (CAM), which consist of our understanding of CCM evolution. in flower colour4,5 and scent profiles6 to both anatomical and biochemical adaptations attract pollinators, multiple origins of that internally concentrate CO2 before its C4 and CAM photosynthesis 7 (Fig. 1) parasitic lifestyles and a multitude of other fixation by Rubisco , thereby making Both C4 and CAM photosynthesis work to traits. Many convergent traits are genetically photosynthesis more efficient10,11. reduce photorespiration, which occurs when simple or perform secondary functions that Over the past 25 years, improvements Rubisco binds to O2 instead of CO2. Plant are not linked to core metabolism. A notable in our understanding of plant phylogenetic photorespiration increases with temperature exception is the repeated modification of relationships have been instrumental owing to the decreased solubility of CO2 one of the most fundamental processes in assessing the number and timing of and the reduced specificity of Rubisco at on Earth, photosynthesis. Although the CCM origins12,13, their relationship with higher temperatures28. Photorespiration machinery involved in the sequestration environmental factors14–17 and the history also increases with abiotic stresses, such of light energy is mostly conserved from of associated phenotypic changes18–21. as drought, that force stomata to close cyanobacteria to flowering plants, the Both C4 and CAM photosynthesis represent and prevent CO2 from entering the leaf and biochemical pathways involved in the textbook examples of convergent evolution, reaching Rubisco29. It is thought that capture of atmospheric carbon vary quite each having likely evolved independently photorespiratory stress is the major driver 22,23 25,30 widely across photosynthetic organisms. more than 60 times . On one hand, it is of C4 and CAM evolution , and many plant

In C3 photosynthesis, which is used by most remarkable that CCMs could have evolved species inhabiting hot and dry habitats use plants, atmospheric CO2 is directly fixed by so frequently because of their seeming one of these CCMs. Both CCMs require that the Calvin cycle via the enzyme Rubisco. complexity and the fact that some relevant acquisition of atmospheric CO2 is separated

However, Rubisco has affinities for both mutations would affect the primary from CO2 fixation via the Calvin cycle

NATure RevIewS | GEnEtiCs volume 20 | AUGUST 2019 | 485 Perspectives

abFig. 1 | Biochemistry of C3 and CCM pathways.

a | The CO2 that C3 plants acquire from the atmos- phere is largely fixed by Rubisco in the Calvin cycle.

Small amounts of CO2, mainly from respiration, can be used by anaplerotic pathways that replenish – PEP HCO3 β-CA intermediates in the tricarboxylic acid (TCA) cycle.

This process involves the conversion of CO2 to – PEPC bicarbonate (HCO3 ) by a carbonic anhydrase β-CA Rubisco (β-​CA) and then subsequent carboxylation of OAA phosphoenolpyruvate (PEP) by a carboxylase HCO – PEP Calvin 3 (PEPC) to a four-​carbon acid, oxaloacetate (OAA). cycle MDH The OAA is further converted to malate by malate dehydrogenase (MDH); finally , malate is fed into PEPC Mesophyll cell the TCA cycle. b | The C4 carbon concentrating mechanism (CCM) is largely composed of these Sugar OAA Bundle sheath cell same enzymes that are spatially separated into distinct compartments. In the most common type MDH Chloroplast Chloroplast of C photosynthesis, CO diffuses into the meso- ME 4 2 Mitochondrion phyll cells, where it is converted to malate via the PYR same pathway (β-​CA , PEPC and MDH). Malate is then transported into adjoining bundle sheath cells, where it is decarboxylated through one or TCA Rubisco Calvin Sugar cycle cycle two different malic enzymes (MEs), releasing CO2 for efficient carboxylation via Rubisco. This pro- cess also releases pyruvate (PYR), which can be used to regenerate the PEP substrate or for carbo- hydrate production. Although not shown here, in

some C4 lineages, different intermediates are c formed instead of malate (for example, aspartate), and other decarboxylases exist, which require slightly different biochemistry and transporters. c | Crassulacean acid metabolism (CAM) species

are nearly identical to C4 plants in their biochemi­ Chloroplast cal pathway of carbon acquisition and fixation, – although separation of uptake and fixation occurs PEP HCO3 β-CA Day temporally rather than spatially. At night, CO2 is PEPC Night Sugar taken up and converted to malate in the same

manner as in C4 plants but then stored in the OAA vacuoles. During the day , the stomata close and malate is decarboxylated in the cytosol, again MDH Calvin cycle flooding Rubisco with high concentrations of CO2. Stomata Stomata Decarboxylation can occur in a variety of loca- tions within the cell, including the cytoplasm, open closed Rubisco Vacuole mitochondria and chloroplasts.

32,33 BS cells . In C4 plants, spatial separation of reactions results in a tenfold increase in 34,35 CO2 in the BS cells relative to M cells ME and enables efficient carbon fixation in

conditions where CO2 is limiting, such PYR as warm, open habitats in the low-CO​ 2 atmosphere that has prevailed since the Oligocene12,36,37. Typically occupying CO2 Malate high-light​ habitats, C4 plants have the highest measured photosynthetic rates and use the same set of enzymes to do so. bundle sheath (BS) cells, respectively of all plants owing to the extremely high (Fig. 1) However, C4 and CAM plants fundamentally , although variations exist across carboxylation rates of the Calvin cycle (Fig. 1) differ in how this separation is achieved . independent origins of C4, and in several when it is not limited by either CO2 or 38 cases CO2 assimilation and reduction take light reaction products . Examples of C4 31 Spatial separation of CO2 uptake and place in distinct regions of a single cell . plant species include maize (Zea mays), fixation in C4 species. In C4 plants, CO2 Specific leaf properties are required to sugarcane (Saccharum officinarum) and uptake and fixation occur in spatially sustain synchronized C4 reactions across sorghum (Sorghum bicolor) in the grass distinct compartments within the leaf. two distinct cell types, including high leaf family (Poaceae), as well as members of the

In most C4 species, uptake and fixation vein density, high BS:M cell volume ratio sunflower family (Flaveria, Asteraceae) and occur in distinct mesophyll (M) cells and and the localization of most chloroplasts to amaranths (Amaranthus, Amaranthaceae).

486 | AUGUST 2019 | volume 20 www.nature.com/nrg Perspectives

Temporal separation of CO2 uptake and Insights from comparative genomics In general, comparisons between groups fixation in CAM species. In CAM species, The regulation of cellular and temporal chosen because they differ in one particular the carbon uptake and fixation reactions are compartmentalization of CCM biochemistry aspect of their phenotype will also capture temporally distinct. CO2 is primarily acquired is complicated, and, unsurprisingly, the many unrelated differences, and the number at night when transpiration rates are low and genomics era has facilitated a new surge of of these irrelevant differences will increase is converted to malate and stored in vacuoles research into the genetics of these complex with the evolutionary distance separating as malic acid. During the day, stomata close traits. The past few years have seen the the groups being compared. Therefore, and malic acid is decarboxylated, resulting rapid production of genomic resources when traits are measured that differ between in increased CO2 concentrations around for an increasing number of species. The CCM and non-CCM​ lineages, the captured 39 Rubisco . The large amount of malic acid first genome of a C4 plant (S. bicolor) was changes are not necessarily required for stored in CAM plants demands bigger assembled in 2009 (ref.49), and the first CAM CCM evolution, as they may also include vacuoles and cells, and therefore CAM is genome (Phalaenopsis equestris, an orchid) those that have occurred before or after the generally associated with a succulent leaf was assembled in 2015 (ref.50), and genomes CCM origination (Box 1; Fig. 2b). Although phenotype. The temporal configuration and transcriptomes are now being sequenced the changes that occur after the origin of reactions, the inverted stomatal for a variety of plants spanning CCM — of the CCM are important for understanding behaviour and the low transpirational including Z. mays (maize) and Ananas CCM diversity, function and especially demand at night help CAM plants achieve comosus (pineapple)50,51 — and non-CCM​ optimization, they may complicate an the highest water use efficiencies in the species52. While fully sequenced reference evaluation of the relative ease or difficulty plant world40. As a result, many desert genomes are still relatively cost-prohibitive​ with which CCMs have evolved. species are CAM plants, including cacti for plants with large genomes, sequencing Historically, many genomic studies of (Cactaceae), agaves (Asparagaceae), expressed genes via transcriptomics can CCM evolution have relied on comparisons and euphorbs (Euphorbiaceae). Less be done at substantially lower cost and between distantly related C3 and CCM intuitively, many tropical forest plants such is much easier to perform in non-model​ taxa53,54, largely because genomic resources as orchids (Orchidaceae) and bromeliads systems. Although transcriptomes are were only available for model systems. (Bromeliaceae) also use CAM; these are unable to uncover regulatory regions and Such broad evolutionary comparisons mostly epiphytes (plants that grow on other non-coding​ sequences, they provide between C3 and CCM species have often other plants) and thus can also experience information on relevant changes in gene highlighted hundreds to thousands of substantial drought stress. expression as well as any changes to gene differentially expressed genes55–57. Although sequences and possible selection across the such studies have been foundational for Alternative and variant CCMs. genome. The growing taxonomic diversity of developing comparative techniques and

C4 and CAM are quite distinct from genomic resources will permit well-designed​ hypotheses, more recent work has focused

C3 photosynthesis, but a number of comparative analyses to better identify the on closely related C3 and CCM taxa intermediate photosynthetic forms also exist. genetic changes underlying the evolution of for finer-​scale resolution of the genetic 30,45,58,59 A third type of CCM, sometimes called C2, new phenotypes, such as a CCM. underpinnings of CCMs . In Flaveria concentrates CO2 in BS cells by restricting (Asteraceae), a lineage comprising closely elements of the photorespiratory cycle Broad comparisons likely overestimate related C3, C4, C2 and C3+C4 species, to those cells; C2 has long been supposed to CCM complexity. Examination of only extant comparative studies found hundreds to (refs30,41) be a precursor to C4 , although diversity, which is often an unavoidable over a thousand differentially expressed it is theoretically not necessary for C4 constraint, will likely overestimate the transcripts between different photosynthetic evolution42. A variety of other phenotypes number of changes required for an initial types51. Similarly, roughly 600 transcripts that are intermediate between C3 and transition from C3 photosynthesis to a CCM. were differentially expressed in closely

C4, termed C3+C4, have been identified, which have a limited C4 photosynthetic cycle, sometimes in addition to a C2 Glossary 41 cycle . CAM photosynthesis, unlike C4, is Anaplerotic Gene flow exceptionally phenotypically plastic because Chemical reactions that provide intermediates to various Movement of genetic information between populations. all photosynthetic cells in CAM plants metabolic pathways, including the tricarboxylic acid (TCA) still have a fully functional C3 cycle. As a cycle. Genome-wide association studies result, many CAM species can perform Analyses that correlate genetic markers from across the varying degrees of CAM or C carbon Carboxylation genome with a phenotype of interest in order to find loci 3 The addition of a carboxyl group to a substrate, often via underlying traits. fixation depending on their developmental a carboxylase enzyme. 43,44 stage or on environmental conditions . Lateral gene transfer For example, there are many so-called​ Co-option Movement of genes between individuals by mechanisms The recruitment of a gene, enzyme or other trait for an other than sexual reproduction. C3+CAM species, which predominantly alternative function. rely on C3 photosynthesis but fix a small Photorespiration amount of carbon via the CAM pathway Decarboxylated Fixation of oxygen by Rubisco, resulting in a loss of 45–47 nocturnally . Some plant species can use Pertaining to a molecule from which a carboxyl group has energy and a release of CO2 but no net gain in 48 been removed by a decarboxylase enzyme in a process carbohydrates. both C4 and CAM in a single individual that releases CO . (Fig. 2a), although these instances are very 2 Transpiration rare. In general, intermediate C3+CAM, like Duons The passive movement of water via stomata from the leaf C2 or C3+C4, is thought to be an evolutionary Portions of the genome that both code for amino acids intercellular airspace to the atmosphere along the water precursor to a full CAM metabolism23,46. and provide motifs that can regulate gene expression. concentration gradient.

NATure RevIewS | GEnEtiCs volume 20 | AUGUST 2019 | 487 Perspectives

a Caryophyllales C4 CAM

otaceae C4 and CAM (within family)

* C4 and CAM (within individual) Asterids Theaceae e A nacampser alinaceae Portulacaceae* A Cactaceae ste T Halophytaceae Montiaceae ctaginaceae Didiereaceae Basellaceae ophyllaceaeanthaceae raceae Molluginaceae Ny izoaceae Gisekiaceae A gonacea Cary Amar Poly A piaceae

ApocynRuaceabiaceae Boraginaceae

e Magnoliids PlantaginaceaGesner e Scr ophular iacea e Lamiaceaiaceaee aceae Piper

Acanthaceae

Welwitschiaceae Acrogymno- sperms Crassulaceae Zamiaceae Vitaceae Geraniaceae

Hydrocharitaceae* Alismataceae Araceae Poaceae Cy peraceae Bromeliaceae

Commelinaceae Aspar Asphodelaceae agaceae

Or chidaceae Monocots

Cleomaceae Passiflor gophyllaceae

Euphorbiaceae

Zy Clusiaceae

aceae

Rosids ucurbitaceae C

Oxalidaceae

b Complete taxon sampling A B True history A Reconstruction of D B phylogeny and E C† character evolution H D I E † K F L G† H M I N J† P K R L S M U N O† P Limited taxon sampling Q† R S T† A U B S U

488 | AUGUST 2019 | volume 20 www.nature.com/nrg Perspectives

Fig. 2 | Evolutionary patterns of CCMs. a | Lineages with carbon concentrating mechanisms (CCMs) 59 ◀ Salsola divaricata and in the C3+CAM 67 are distributed across the phylogeny of vascular plants. In most cases, C4 and crassulacean acid metab- intermediate hybrid Yucca gloriosa . olism (CAM) are found in distinct lineages (blue and orange family labels, respectively); however, fam- In addition, variation between accessions of ilies in which there are examples of both C and CAM photosynthesis exist (grey labels). Indeed, some 4 the C species Gynandropsis gynandra has individuals within some of these families (Portulacaceae and Hydrocharitaceae, indicated by asterisks) 4 can use both C and CAM photosynthesis. Only CCM lineages are labelled. Figure created using phylo­ been proposed as a system to identify the 4 68 geny from ref.123. b | Even with phylogenetic information, comparative studies can still inflate the genetic determinants of C4 photosynthesis . number of changes required for a CCM to evolve. For example, in a theoretical tree of extant and The development of model systems in a extinct (†) taxa labelled A to U, a single origin of a CCM occurred at the blue dot. All blue branches greater number of species, particularly those

indicate portions of the tree where the CCM exists. Black dots represent changes in ancestral C3 spe- with intermediate or variable CCM usage, cies that pre-​date the CCM origin, and yellow dots are changes that occurred after the CCM origin will help determine whether patterns seen in that may or may not be related to the CCM itself. If all extant taxa could be sampled, the number of Alloteropsis and Flaveria are representative changes inferred on the branch where the CCM occurred is relatively low; in a more realistic case, of CCM evolution more generally. where only a subset of extant species are sampled, a greater number of changes are assigned to the The generation of genomic resources branch with the CCM origin, even though many are unrelated to CCM evolution. for new and existing model species will enable genome-wide association studies and 58 related C3 and C4 members of Cleomaceae . but important variation. For example, it is other approaches to detect the determinants

Even by reducing evolutionary distance, already known that the strength of the C4 of CCMs and potentially identify the comparative transcriptomic studies between cycle varies between accessions of the precise genetic changes required for 66 closely related species are still likely inflating C3+C4 intermediate Mollugo verticillata , the initial emergence of a CCM and its the estimates of genetic changes required for potentially between populations of subsequent adaptation. CCM origins.

Emerging model systems to study CCM Box 1 | Refinement of a CCM pathway evolution. Model clades such as Flaveria Although all plant lineages possess the enzymatic machinery for a carbon concentrating will continue to be highly relevant to CCM mechanism (CCM), and some lineages appear to be potentiated for CCM evolution, phylogenetic evolution, but the problem of overestimating analyses indicate that a subset of changes have occurred after the origin of the CCM pathway and complexity can be reduced further. Ideally, are, therefore, likely to be involved in its refinement68. a spectrum of photosynthetic phenotypes, Refinement of C4 pathways

all within a single species, would be studied Much more is known about when key post-origin​ CCM traits evolved in C4 lineages than for to capture the emergence of a rudimentary crassulacean acid metabolism (CAM) lineages owing largely to more robust phylogenetic work in

CCM. Such a system is currently being important C4 lineages, predominantly the grasses.

developed in the grass species Alloteropsis Anatomical specialization. C4 species arising from a single origin of the CCM show variation in 60 19,106–108 semialata . The existence of C3 and C4 anatomical traits important for C4 function, such as vein spacing and bundle sheath cell size , genotypes of A. semialata has been known for which suggests that different lineages have different post-origin​ anatomical improvements for some time, but recent work has demonstrated CCM activity. a diverse spectrum of physiology that Gene evolution. Although the core set of CCM enzymes is present in all plants, analyses of their

spans C3 and C4 in important ways. amino acid sequence in C4 lineages suggest they undergo improvement after the origin of the Phenotypically intermediate populations of CCM. For example, some core CCM enzymes are known to vary in their amino acid sequence within C lineages98, molecular evolution studies indicate that certain residues are under positive selection A. semialata that perform a weak C4 cycle 4 61 47,109,110 have been identified , and genome-wide​ within C4 species , and convergent substitutions have been identified that are restricted to some C species within each group111–113. Additionally, phylogenetic analyses indicate that a number analyses have provided evidence of genetic 4 of gene duplication events that allowed for enzyme adaptation occurred after a CCM origin114. exchange between different photosynthetic 62 Refinement of CAM pathways types . C4 populations of A. semialata show incomplete segregation of Rubisco Post-origin​ changes are thought to be required for the refinement of constitutive CAM activity, but such changes are not necessarily required in C CAM species or in taxa that have recently into BS cells63 and have a photosynthetic 3+ 61 evolved a CCM. However, the timing of these changes is largely unknown owing to a lack of efficiency below that of other C4 grasses systematic comparisons between C3, C3+CAM and taxa with strong CAM activity. Candidates for and can, therefore, be considered as having modifications that might occur after the origin of a CAM pathway include the following. a rudimentary CCM. Comparative work Stomatal regulation. Although some C species are known to open their stomata at night115,116, has indicated that the initial emergence of 3 the inverse stomatal behaviour observed in CAM species (Fig. 1) likely requires large transcriptional C4 photosynthesis in the group involved reconfiguration. In particular, CAM species likely modify pathways involving abscisic acid and blue 64 upregulation of only a handful of genes . light sensing, two of the main pathways for stomatal signalling. Blue light in particular is a major Subsequent refinement of the CCM in signal for stomatal aperture, and some experiments on blue light sensitivity and transcriptomic Alloteropsis is supported by the presence of analyses have suggested that guard cells in CAM species are less sensitive to blue light than guard 117–120 genes better suited for the C4 context in only cells in C3 species . 121 some of the C4 populations of A. semialata Circadian regulation. Some CAM enzymes show clear circadian behaviour , and, indeed, and by evidence of sustained positive integration of carbon fixation and metabolism with an endogenous clock is thought to be critical 64,65 121,122 selection within the C4 group . for efficient CAM function . Comparative genomics between pineapple (Ananas comosus, Bromeliaceae) and unrelated C and C species showed an increase in clock-​regulated regulatory Although distinct C3 and C4 populations 3 4 have not yet been reported in species other motifs upstream of canonical CAM genes (Fig. 1c), indicating circadian control of gene 87 than A. semialata, we predict that new expression . Although it is likely that fine-​scaled circadian control of CAM gene expression occurs only with or after the origin of CAM, further work is needed to show circadian regulation as a efforts to broadly screen wild populations refinement of the CAM cycle. of young CCM species will reveal subtle

NATure RevIewS | GEnEtiCs volume 20 | AUGUST 2019 | 489 Perspectives

All plants have key CCM elements frequent co-option​ into a variety of Genetic potentiation. Similar to The known components of the CCM cellular roles. environment and anatomy, genetic biochemical pathway are present in all characteristics could also predispose (Fig. 1) vascular plants , and the co-​option Potentiation for CCM evolution certain lineages to evolve a CCM. In C4 of existing enzymes is a straightforward CCM origins seem to be phylogenetically and CAM lineages, CCM enzymes need mechanism by which CCMs could initially clustered, with large regions of the land to be expressed at specific locations83–85, develop. The core mechanism of carbon plant phylogeny completely lacking any times86,87 and levels88,89. Non-CCM​ lineages 23 (Fig. 2a) concentration — the fixation of CO2 by the known C4 or CAM plants . Thus, that have regulatory pathways in place that coupled action of the enzymes carbonic although all plants have the biochemical could confer these properties on CCM anhydrase and phosphoenolpyruvate components for a CCM, some lineages may enzymes might be predisposed to evolving carboxylase (PEPC) — is part of an be more likely to evolve CCMs because CCMs. Indeed, genome-wide​ comparisons anaplerotic pathway that supplements malate of differential exposure to environmental and transcription factor binding assays to the tricarboxylic acid (TCA) cycle and pressures or the evolution of characteristics between C3 and C4 species have revealed exists in the cytosol of all plants69. In a — either anatomical or genomic — that later shared regulatory motifs90,91. For example, non-​CCM plant, a rise in the concentration facilitate the co-option​ of these enzymes into a simple 59 bp deletion upstream of a − of cytosolic bicarbonate (HCO3 ), for photosynthetic metabolic pathways. gene encoding the P-subunit​ of glycine example, from an increase in respired CO2, decarboxylase in the C3 species Arabidopsis could result in increased malate production. Anatomical potentiation. The idea thaliana is sufficient to confer BS cell-​ − If HCO3 levels are high enough, PEPC may that some lineages may be potentiated specific expression, which is required for 77 92 generate more malate than the maintenance (sensu Blount ) for CCM evolution has C2 biochemistry . Another A. thaliana pathways can accommodate. In this often been discussed28, but primarily in gene, NAD-me​ , which encodes a CCM scenario, the high malate concentration the context of leaf anatomy. The spatial decarboxylating enzyme that also has would typically feed back to inhibit PEPC70. separation of carbon assimilation and roles in mitochondrial and chloroplast

However, if excess malate instead became fixation via Rubisco in C4 plants is housekeeping, contains motifs within its sequestered in the vacuole or diffused generally realized across distinct cell coding sequence called duons that have 32,33 (Fig. 1) to adjacent cells along a concentration types . The typical C4 anatomical regulatory functions. These sequences are gradient, this feedback would not occur, traits, such as large BS:M cell volume also found in the C4 species G. gynandra, and a rudimentary CCM cycle could be ratio and small distance between veins, in which they are necessary and sufficient 93 established. Thus, the upregulation of a few have evolved before the C4 pathway in a for BS cell-specific​ expression . Thus, 19 key enzymes might be sufficient to initiate number of lineages including grasses , these sequences confer C4 activity through CCM evolution. Flaveria18 and Cleomaceae78, suggesting changes to gene regulation rather than

The non-​photosynthetic roles of that C4 origins are clustered in certain via alterations to the coding sequence 79 (Fig. 2a) some CCM enzymes in C3 species may regions of the plant phylogeny of the enzyme. Orthologues of C4 genes also support the view that CCMs are because some lineages happened to first in C3 species have also been shown to 84,94 evolutionarily accessible. For example, evolve a C4-like anatomy. Anatomical be transcriptionally induced by light non-photosynthetic​ CCM pathways that potentiation could have arisen either from and regulated by chloroplast-to-nucleus​ act in a tissue-specific​ manner have been environmental pressures or stochastic signalling95, two characteristics of many identified in several C3 species: low-level​ processes; regardless, the establishment of enzymes in the C4 pathway. The prevalence nocturnal CO2 fixation without malic acid a C4-like anatomy then allowed for frequent of shared regulatory DNA sequences 71 accumulation occurs in tobacco leaves transitions to C4 photosynthesis within and transcriptional cascades in C3 and 72 64 and in cotton ovules , and a C4-like these lineages . CCM species suggests that pre-existing​ mechanism is found in cells adjacent to The phylogenetic distribution of CAM regulatory mechanisms may facilitate the the vasculature of tobacco stems, whereby photosynthesis appears to be much less repeated evolution of C4 photosynthesis in photosynthetic cells use CO2 that has been clustered, and the evidence of anatomical certain lineages. respired by roots and transported through enabling is mixed46,80 (Fig. 2a). Many Genetic potentiation can also occur in 73 the xylem . Genes that give identity to root known C3+CAM species are only mildly certain groups as a result of genome-wide​ endodermal cells have also been shown to succulent, whereas species with strong patterns and processes. In some cases, the pleiotropically affect differentiation of BS CAM activity typically have large, tightly ability of plant genomes to undergo dramatic cells in the leaf, which are important for C4: packed cells and thick photosynthetic reconfiguration may have facilitated the 80,81 SCARECROW and SHORTROOT mutants tissues . In the Agavoideae, tissue evolution of a CCM. For example, in C4 have deformed endodermal cell layers in the succulence seemingly evolved before the maize, hundreds of DNA motifs associated 82 roots and a proliferation of BS cells in the emergence of strong CAM photosynthesis , with C4 have apparently been moved leaves74,75. Perhaps most notably, Camellia but there are few studies that combine around the genome by transposons96. oleifera (Theaceae, Ericales), a mesic forest anatomy, CAM activity and a well-resolved​ Lateral gene transfer has even played a role tree, has been shown to respond to leaf and sampled phylogeny. Although in CCM evolution in certain cases; genes fungal infection by strongly inducing tissue more studies are needed, the lack of optimized for C4 function were laterally succulence and a fully functional CAM identifiable anatomical specialization transferred to A. semialata from distantly 76 cycle . The upregulation of a complete in C3+CAM plants suggests that leaf related C4 grasses and are preferentially CAM cycle in a species so distantly related anatomical changes are likely necessary for expressed relative to the native gene copies64. from any known CCM origin emphasizes the development of strong CAM activity Hybridization between C3 and CAM species both the ubiquitous presence of functional but not for the evolution of a weak in Yucca gave rise to a new C3+CAM CCM enzymes in all plants and their CAM cycle. species67, and population gene flow between

490 | AUGUST 2019 | volume 20 www.nature.com/nrg Perspectives

C4 and non-C​ 4 genotypes in A. semialata patterns could result from a variety of of comparative genomics to identify the facilitated movement of C4-adapted processes. Perhaps these copies are more genetic potentiation of CCM evolution. alleles across the range of the species62. highly expressed because of a dosage effect Although the use of existing data sets such Additionally, in some CCM lineages, caused by local gene duplications (we did as 1KP for analysis of CCM evolution offers duplicated genes are associated with not quantify lineage-specific​ copy numbers exciting possibilities, better phylogenetic co-​option into CCM pathways97–100 and per PEPC paralogue), which could then resolution of CCM-evolving​ lineages will subsequent positive selection101,102. allow for subsequent specialization of continue to expand the availability of model By contrast, other studies have indicated paralogues100. Alternatively, the increased systems that are particularly powerful for that recruitment of genes into a CCM transcript abundance of one copy over the assessing initial genetic changes associated was not immediately preceded by a gene other might have been selected for because with CCM evolution. duplication event but instead relied on of a non-photosynthetic​ role of PEPC, which differentiation of cis-regulatory​ elements simultaneously made it amenable to co-​ Conclusions in ancient paralogues50,85,87. The evidence option by a CCM. Although our analysis is In this Perspective, we have outlined from such genomic comparative studies based on a broadly sampled transcriptome evidence supporting the idea that relatively suggests that re-wiring​ of transcriptional data set that was not designed for this few modifications are required to reach a cascades — whether through gene sort of question, the emergence of such a rudimentary CCM phenotype, at least in duplications, genomic rearrangements or striking pattern reveals the great potential groups that are potentiated. If correct, this molecular evolution — is critical for CCM evolution. Asteraceae 1E2 Future perspectives Molluginaceae As sequencing costs decrease and protocols become more streamlined, gene expression Cleomaceae data for non-model​ species are becoming Atriplex increasingly available, which allows relative gene expression to be compared across Boraginaceae hundreds or thousands of species in a single Aizoaceae 1E1 analysis. As a simple illustration of the types of studies we imagine will become Amaranthaceae s.s. increasingly common and insightful, we analysed the relative levels of expression Scrophulariaceae of major PEPC gene copies in leaves of Nyctaginaceae

C3 species across flowering plants (using 1KP data103,104, Supplementary Methods, Euphorbiaceae Supplementary Figure 1 and Supplementary Table 1). PEPC is the best studied enzyme Poaceae in the CCM pathway and is an exceptional system to explore the mechanics of gene Orchidaceae 1M1 recruitment into a novel function. Most Nolinoideae flowering plants have at least three main copies of genes encoding PEPC, with two Agavoideae 1M2 of those copies arising from independent duplication events in eudicots and 0.01 0.1 1 10 100 monocots48,105 (Supplementary Figure 1). Relative expression of the co-opted copy (TPM) Studies in grasses and orchids have Fig. 3 | Expression of PEPC in the leaves of C3 angiosperms belonging to CCM-​evolving lineages. suggested that the copy that is most highly We used the 1KP103 database to assess whether phosphoenolpyruvate carboxylase (PEPC) copies expressed in C3 relatives is the one recruited upregulated in C species predict which copy is co-opted​ in C or crassulacean acid metabolism (CAM) 88,89 3 4 into the CCM . We specifically focused origins within the same lineage. Gene expression for PEPC was assessed per species for the major on C3 species in flowering plant lineages clades of the gene family: PPC-1E1 and PPC-1E2 in eudicots (top panel) and PPC-1M1 and PPC-1M2 that have evolved CCMs, in which the copy in monocots (bottom panel) (Supplementary Figure 1). Expression was summed across all contigs from recruited into the CCM is known from the transcriptomes of species that, based on phylogenetic analysis, were within each of the major empirical studies. Because we relied on clades (Supplementary Methods and Supplementary Table 1). Expression of PPC-2, a separate para­ transcriptomic data rather than genomic logue of PEPC, was not considered here, as it is not commonly recruited to a carbon concentrating mechanism (CCM). The background colour indicates which copy is used by the C or CAM species data, we summed the expression of multiple 4 assembled transcripts per PEPC paralogue within a given lineage, and the box plots represent the log of the ratio of expression (in transcript per million (TPM)) of the co-​opted copy to the other copy for all species within that lineage. Clades for per species. Even so, in 12 of the 14 CCM-​ which the box plot falls to the right of the dashed line co-opt​ the more highly expressed copy of PEPC, evolving lineages, the copy recruited into whereas box plots to the left of the dashed line indicate clades that have co-​opted the lesser expressed the CCM is more highly expressed than copy of PEPC. For example, C4 Asteraceae species use PPC-1E2 in the CCM pathway , and their C3 (Fig. 3) other PEPC copies in the C3 taxa . relatives have higher expression of PPC-1E2 relative to PPC-1E1. In the majority of lineages assessed,

While the recruitment bias is likely driven the co-​opted copy is more highly expressed than the other copy in C3 species, suggesting that by expression levels of the two copies of this quality is important for recruitment into a CCM and may actually facilitate CCM evolution in PEPC, the differences in the expression certain lineages.

NATure RevIewS | GEnEtiCs volume 20 | AUGUST 2019 | 491 Perspectives

hypothesis would help to explain how such phenotypes across lineages of eusocial insects. 31. Voznesenskaya, E. V. et al. Kranz anatomy is not Mol. Biol. Evol. 32, 690–703 (2015). essential for terrestrial C4 plant photosynthesis. a seemingly complex trait has evolved so 4. Smith, S. D. & Kriebel, R. Convergent evolution of Nature 414, 543–546 (2001). frequently. Once a rudimentary CCM is floral shape tied to pollinator shifts in Iochrominae 32. Dengler, N. G. et al. Quantitative leaf anatomy of (Solanaceae). Evolution 72, 688–697 (2018). C3 and C4 grasses (Poaceae): bundle sheath and operating, selection to improve its efficiency 5. Larter, M. et al. Convergent evolution at the pathway mesophyll surface area relationships. Ann. Bot. 73, will be strong because it relates to the level: predictable regulatory changes during flower 241–255 (1994). color transitions. Mol. Biol. Evol. 35, 2159–2169 33. Hattersley, P. W. Characterization of C4 type leaf primary metabolism of plants. Over time, (2018). anatomy in grasses (Poaceae). Mesophyll: bundle this selection will lead to the accumulation 6. Knudsen, J. T. & Tollsten, L. Floral scent in bat-​ sheath area ratios. Ann. Bot. 53, 163–180 (1984). pollinated plants: a case of convergent evolution. 34. Hatch, M. D. in C4 Plant Biology (eds Sage, R. F. of additional traits that further optimize Bot. J. Linn. Soc. 119, 45–57 (1995). & Monson, R. K.) 17–46 (Academic Press, 1999). 7. Conn, C. E. et al. Convergent evolution of strigolactone 35. Furbank, R. T. & Hatch, M. D. Mechanism of C the new pathway and that strongly contrast 4 perception enabled host detection in parasitic plants. photosynthesis: the size and composition of the with those of their non-CCM​ relatives. Science 349, 540–543 (2015). inorganic carbon pool in bundle sheath cells. Plant Although these specializations are important 8. Tcherkez, G. G. B., Farquhar, G. D. & Andrews, T. J. Physiol. 85, 958–964 (1987). Despite slow catalysis and confused substrate 36. Christin, P.-A. et al. C4 eudicots are not younger than for understanding the nature of extant specificity, all ribulose bisphosphate carboxylases may C4 monocots. J. Exp. Bot. 62, 3171–3181 (2011). CCMs, they lead to an overestimation of the be nearly perfectly optimized. Proc. Natl Acad. Sci. 37. Christin, P.-A. & Osborne, C. P. The recurrent assembly USA 103, 7246–7251 (2006). of C4 photosynthesis, an evolutionary tale. Photosynth. initial changes required to evolve a CCM 9. Bauwe, H., Hagemann, M. & Fernie, A. R. Res. 117, 163–175 (2013). Photorespiration: players, partners and origin. Trends 38. Long, S. P. in C Plant Biology (eds Sage, R. F. & and, therefore, to an underestimation of its 4 Plant Sci. 15, 330–336 (2010). Monson, R. K.) 215–249 (Academic Press, 1999). evolutionary accessibility. 10. Hatch, M. D. C4 photosynthesis: a unique blend of 39. Cockburn, W., Ting, I. P. & Sternberg, L. O. We envision that future comparative modified biochemistry, anatomy and ultrastructure. Relationships between stomatal behavior and internal Biochim. Biophys. Acta 895, 81–106 (1987). carbon dioxide concentration in crassulacean acid genomic studies will continue to refine our 11. Osmond, C. B. Crassulacean acid metabolism: metabolism plants. Plant Physiol. 63, 1029–1032 understanding of CCM evolution in two a curiosity in context. Annu. Rev. Plant Physiol. 29, (1979). 40 important ways. First, the accumulation 379–414 (1978). . Borland, A. M. et al. Exploiting the potential of plants 12. Christin, P.-A. et al. Oligocene CO2 decline promoted with crassulacean acid metabolism for bioenergy of genomic data from an increasingly C4 photosynthesis in grasses. Curr. Biol. 18, 37–43 production on marginal lands. J. Exp. Bot. 60, (2008). 2879–2896 (2009). phylogenetically diverse species pool will 13. Arakaki, M. et al. Contemporaneous and recent 41. Sage, R. F. et al. Some like it hot: the physiological facilitate robust comparative analyses across radiations of the world’s major succulent plant ecology of C4 plant evolution. Oecologia 187, lineages. Proc. Natl Acad. Sci. USA 108, 8379–8384 941–966 (2018). multiple CCM origins — in other words, we (2011). 42. Heckmann, D. et al. Predicting C4 photosynthesis imagine a proliferation of refined versions of 14. Edwards, E. J. & Still, C. J. Climate, phylogeny and the evolution: modular, individually adaptive steps on a ecological distribution of C4 grasses. Ecol. Lett. 11, Mount Fuji fitness landscape. Cell 153, 1579–1588 the preliminary 1KP analyses we presented 266–276 (2008). (2013). here. Second, we predict that new data and 15. Edwards, E. J. & Smith, S. A. Phylogenetic analyses 43. Winter, K., Garcia, M. & Holtum, J. A. M. On the reveal the shady history of C4 grasses. Proc. Natl nature of facultative and constitutive CAM: analytical tools in population genomics Acad. Sci. USA 107, 2532–2537 (2010). environmental and developmental control of CAM will enable a renaissance of intraspecific 16. Horn, J. W. et al. Evolutionary bursts in Euphorbia expression during early growth of Clusia, Kalanchöe, (Euphorbiaceae) are linked with photosynthetic and Opuntia. J. Exp. Bot. 59, 1829–1840 (2008). studies of CCM variation and will uncover pathway. Evolution 68, 3485–3504 (2014). 44. Winter, K. & Holtum, J. A. M. Induction and reversal additional A. semialata-like systems with 17. Crayn, D. M. et al. Photosynthetic pathways in of crassulacean acid metabolism in Calandrinia which to explore the evolutionary emergence Bromeliaceae: phylogenetic and ecological significance polyandra: effects of soil moisture and nutrients. of CAM and C3 based on carbon isotope ratios for Funct. Plant Biol. 38, 576–582 (2011). of a nascent CCM. Taken together, these 1893 species. Bot. J. Linn. Soc. 178, 169–221 45. Heyduk, K. et al. Shifts in gene expression profiles are (2015). associated with weak and strong crassulacean acid efforts will disentangle the genetic changes 18. McKown, A. D. & Dengler, N. G. Key innovations in the metabolism. Am. J. Bot. 105, 587–601 (2018). that are required for refinement of CCM evolution of Kranz anatomy and C4 vein pattern in 46. Silvera, K., Santiago, L. S. & Winter, K. Distribution Flaveria (Asteraceae). Am. J. Bot. 94, 382–399 (2007). of crassulacean acid metabolism in orchids of Panama: evolution from those responsible for the 19. Christin, P.-A. et al. Anatomical enablers and the evidence of selection for weak and strong modes. origin of a CCM pathway. From a rich evolution of C4 photosynthesis in grasses. Proc. Natl Funct. Plant Biol. 32, 397 (2005). history of biochemical and genetic studies, Acad. Sci. USA 110, 1381–1386 (2013). 47. Goolsby, E. W. et al. Molecular evolution of key 20. Silvera, K. et al. Crassulacean acid metabolism and metabolic genes during transitions to C4 and CAM our knowledge of plant CCM evolution epiphytism linked to adaptive radiations in the photosynthesis. Am. J. Bot. 105, 602–613 (2018). may already be far ahead of other classical Orchidaceae. Plant Physiol. 149, 1838–1847 (2013). 48. Christin, P.-A. et al. Shared origins of a key enzyme 21. Crayn, D. M., Winter, K. & Smith, J. A. C. Multiple during the evolution of C4 and CAM metabolism. examples of convergence. In time, we predict origins of crassulacean acid metabolism and the J. Exp. Bot. 65, 3609–3621 (2014). that a growth of genomic resources in non-​ epiphytic habit in the Neotropical family Bromeliaceae. 49. Paterson, A. H. et al. The Sorghum bicolor genome Proc. Natl Acad. Sci. USA 101, 3703–3708 (2004). and the diversification of grasses. Nature 457, model systems has the potential to provide 22. Sage, R. F., Christin, P.-A. & Edwards, E. J. The C4 plant 551–556 (2009). a fairly complete explanation of how C and lineages of planet Earth. J. Exp. Bot. 62, 3155–3169 50. Cai, J. et al. The genome sequence of the orchid 4 (2011). Phalaenopsis equestris. Nat. Genet. 47, 65–72 CAM photosynthesis evolved so often in 23. Edwards, E. J. & Ogburn, R. M. Angiosperm responses (2015). to a low-​CO world: CAM and C photosynthesis as 51. Yang, X. et al. The Kalanchoë genome provides plants, and perhaps it is not so far away. 2 4 parallel evolutionary trajectories. Int. J. Plant Sci. 173, insights into convergent evolution and building blocks Karolina Heyduk 1*, Jose J. Moreno-Villena​ 1, 724–733 (2012). of crassulacean acid metabolism. Nat. Commun. 8, 24. Kellogg, E. A. in C Plant Biology (eds Sage, R. F. 1899 (2017). Ian S. Gilman1, Pascal-Antoine Christin​ 2 and 4 & Monson, R. K.) 411–444 (Academic Press, 1999). 52. Goodstein, D. M. et al. Phytozome: a comparative 1 Erika J. Edwards 25. Sage, R. F. Environmental and evolutionary platform for green plant genomics. Nucleic Acids Res. 1Department of Ecology and Evolutionary Biology, preconditions for the origin and diversification of the 40, D1178–D1186 (2012). C photosynthetic syndrome. Plant Biol. 3, 202–213 53. Ding, Z. et al. Identification of photosynthesis-​ Yale University, New Haven, CT, USA. 4 (2001). associated C4 candidate genes through comparative 2Department of Animal and Plant Sciences, 26. Smith, J. M. et al. Developmental constraints and leaf gradient transcriptome in multiple lineages of PLOS ONE 10 The University of Sheffield, Sheffield, UK. evolution: a perspective from the Mountain Lake C3 and C4 species. , e0140629 (2015). Conference on Development and Evolution. Q. Rev. 54. Wang, L. et al. Comparative analyses of C4 and C3 *e-mail:​ [email protected] Biol. 60, 265–287 (1985). photosynthesis in developing leaves of maize and rice. 27. Blount, Z. D., Lenski, R. E. & Losos, J. B. Contingency Nat. Biotechnol. 32, 1158–1165 (2014). https://doi.org/10.1038/s41576-019-0107-5 and determinism in evolution: replaying life’s tape. 55. Brilhaus, D. et al. Reversible burst of transcriptional Published online 18 March 2019 Science 362, eaam5979 (2018). changes during induction of crassulacean acid 28. Jordan, D. B. & Ogren, W. L. The CO2/O2 specificity metabolism in Talinum triangulare. Plant Physiol. 170, 1. Fernald, R. D. Casting a genetic light on the evolution of ribulose 1,5-bisphosphate carboxylase/oxygenase. 102–122 (2016). of eyes. Science 313, 1914–1918 (2006). Planta 161, 308–313 (1984). 56. Xu, J. et al. Systems analysis of cis-​regulatory motifs in

2. Fraser, J. A. et al. Convergent evolution of 29. Farquhar, G. D. & Sharkey, T. D. Stomatal conductance C4 photosynthesis genes using maize and rice leaf chromosomal sex-​determining regions in the animal and photosynthesis. Annu. Rev. Plant Physiol. 33, transcriptomic data during a process of de-​etiolation. and fungal kingdoms. PLOS Biol. 2, e384 (2004). 317–345 (1982). J. Exp. Bot. 67, 5105–5117 (2016).

3. Berens, A. J., Hunt, J. H. & Toth, A. L. Comparative 30. Mallmann, J. et al. The role of photorespiration during 57. Gowik, U. et al. Evolution of C4 photosynthesis in the

transcriptomics of convergent evolution: different the evolution of C4 photosynthesis in the genus genus Flaveria: how many and which genes does it

genes but conserved pathways underlie caste Flaveria. eLife 3, e02478 (2014). take to make C4? Plant Cell 23, 2087–2105 (2011).

492 | AUGUST 2019 | volume 20 www.nature.com/nrg Perspectives

58. Bräutigam, A. et al. An mRNA blueprint for Agavoideae (Asparagaceae). Mol. Phylogenet. Evol. 108. Lundgren, M. R. et al. C4 anatomy can evolve via a

C4 photosynthesis derived from comparative 105, 102–113 (2016). single developmental change. Ecol. Lett. 22, 302–312

transcriptomics of closely related C3 and C4 83. Gowik, U. et al. cis-​Regulatory elements for mesophyll-​ (2019).

species. Plant Physiol. 155, 142–156 (2011). specific gene expression in the C4 plant Flaveria 109. Rosnow, J. J. et al. Kranz and single-​cell forms of C4

59. Lauterbach, M. et al. De novo transcriptome assembly trinervia, the promoter of the C4 phosphoenolpyruvate plants in the subfamily Suaedoideae show kinetic C4

and comparison of C3, C3–C4, and C4 species of Tribe carboxylase gene. Plant Cell 16, 1077–1090 (2004). convergence for PEPC and Rubisco with divergent Salsoleae (Chenopodiaceae). Front. Plant Sci. 8, 1939 84. Matsuoka, M. et al. The promoters of two amino acid substitutions. J. Exp. Bot. 66, 7347–7358

(2017). carboxylases in a C4 plant (maize) direct cell-​specific, (2015).

60. Ellis, R. P. Anomalous vascular bundle sheath light-regulated​ expression in a C3 plant (rice). Plant J. 110. Rosnow, J. J., Edwards, G. E. & Roalson, E. H.

structure in Alloteropsis semialata leaf blades. 6, 311–319 (1994). Positive selection of Kranz and non-​Kranz C4 Bothalia 11, 273–275 (1974). 85. Akyildiz, M. et al. Evolution and function of a cis-​ phosphoenolpyruvate carboxylase amino acids in 61. Lundgren, M. R. et al. Evolutionary implications of regulatory module for mesophyll-​specific gene Suaedoideae (Chenopodiaceae). J. Exp. Bot. 65,

C3–C4 intermediates in the grass Alloteropsis expression in the C4 dicot Flaveria trinervia. Plant Cell 3595–3607 (2014).

semialata. Plant Cell Environ. 39, 1874–1885 19, 3391–3402 (2007). 111. Besnard, G. et al. Phylogenomics of C4 photosynthesis (2016). 86. Yin, H. et al. Diel rewiring and positive selection of in sedges (Cyperaceae): multiple appearances and 62. Olofsson, J. K. et al. Genome biogeography reveals ancient plant proteins enabled evolution of CAM genetic convergence. Mol. Biol. Evol. 26, 1909–1919 the intraspecific spread of adaptive mutations for a photosynthesis in Agave. BMC Genomics 19, 588 (2009). complex trait. Mol. Ecol. 25, 6107–6123 (2016). (2018). 112. Christin, P.-A. et al. Evolutionary switch and genetic

63. Ueno, O. & Sentoku, N. Comparison of leaf structure 87. Ming, R. et al. The pineapple genome and the convergence on rbcL following the evolution of C4

and photosynthetic characteristics of C3 and C4 evolution of CAM photosynthesis. Nat. Genet. 47, photosynthesis. Mol. Biol. Evol. 25, 2361–2368 Alloteropsis semialata subspecies. Plant Cell Environ. 1435–1442 (2015). (2008).

29, 257–268 (2006). 88. Moreno-​Villena, J. J. et al. Highly expressed genes 113. Christin, P.-A. et al. C4 Photosynthesis evolved in

64. Dunning, L. T. et al. Introgression and repeated are preferentially co-​opted for C4 photosynthesis. grasses via parallel adaptive genetic changes.

co-option facilitated the recurrent emergence of C4 Mol. Biol. Evol. 35, 94–106 (2018). Curr. Biol. 17, 1241–1247 (2007). photosynthesis among close relatives. Evolution 71, 89. Silvera, K. et al. Multiple isoforms of 114. Besnard, G. et al. Herbarium genomics retraces the

1541–1555 (2017). phosphoenolpyruvate carboxylase in the Orchidaceae origins of C4-specific carbonic anhydrase in

65. Christin, P.-A. et al. Adaptive evolution of C4 (subtribe Oncidiinae): implications for the evolution of Andropogoneae (Poaceae). Bot. Lett. 165, 419–433 photosynthesis through recurrent lateral gene crassulacean acid metabolism. J. Exp. Bot. 65, (2018). transfer. Curr. Biol. 22, 445–449 (2012). 3623–3636 (2014). 115. Caird, M. A., Richards, J. H. & Donovan, L. A. 66. Sayre, R. T., Kennedy, R. A. & Pringnitz, D. J. 90. Brown, N. J. et al. Independent and parallel Nighttime stomatal conductance and transpiration

Photosynthetic enzyme activities and localization in recruitment of preexisting mechanisms underlying in C3 and C4 plants. Plant Physiol. 143, 4–10

Mollugo verticillata populations differing in the levels C4 photosynthesis. Science 331, 1436–1439 (2007).

of C3 and C4 cycle operation. Plant Physiol. 64, (2011). 116. Snyder, K. A., Richards, J. H. & Donovan, L. A.

293–299 (1979). 91. Borba, A. R., Serra, T. S. & Górska, A. Synergistic Night-time conductance in C3 and C4 species: do plants 67. Heyduk, K. et al. Gas exchange and leaf anatomy of binding of bHLH transcription factors to the promoter lose water at night? J. Exp. Bot. 54, 861–865 (2003).

a C3-CAM hybrid, Yucca gloriosa (Asparagaceae). of the maize NADP-ME​ gene used in C4 photosynthesis 117. Tallman, G. et al. Induction of CAM in J. Exp. Bot. 67, 1369–1379 (2016). is based on an ancient code found in the ancestral Mesembryanthemum crystallinum abolishes the

68. Reeves, G. et al. Natural variation within a species for C3 state. Mol. Biol. 35, 1690–1705 (2018). stomatal response to blue light and light-​dependent

traits underpinning C4 photosynthesis. Plant Physiol. 92. Adwy, W., Laxa, M. & Peterhansel, C. A simple zeaxanthin formation in guard cell chloroplasts. Plant

177, 504–512 (2018). mechanism for the establishment of C2-specific gene Cell Physiol. 38, 236–242 (1997). 69. O’Leary, B., Park, J. & Plaxton, W. C. The remarkable expression in Brassicaceae. Plant J. 84, 1231–1238 118. Lee, D. M. & Assmann, S. M. Stomatal responses to diversity of plant PEPC (phosphoenolpyruvate (2015). light in the facultative Crassulacean acid metabolism carboxylase): recent insights into the physiological 93. Reyna-​Llorens, I. et al. Ancient duons may underpin species, Portulacaria afra. Physiol. Plant. 85, 35–42

functions and post-​translational controls of non-​ spatial patterning of gene expression in C4 leaves. (1992). photosynthetic PEPCs. Biochem. J. 436, 15–34 Proc. Natl Acad. Sci. USA 115, 1931–1936 (2018). 119. Abraham, P. E. et al. Transcript, protein and (2011). 94. Kausch, A. P. et al. Mesophyll-​specific, light and metabolite temporal dynamics in the CAM plant

70. Smith, F. A. & Raven, J. A. Intracellular PH and its metabolic regulation of the C4 PPCZm1 promoter in Agave. Nat. Plants 2, 16178 (2016). regulation. Annu. Rev. Plant Physiol. 30, 289–311 transgenic maize. Plant Mol. Biol. 45, 1–15 (2001). 120. Heyduk, K. et al. Altered gene regulatory networks are

(1979). 95. Burgess, S. J. et al. Ancestral light and chloroplast associated with the transition from C3 to Crassulacean

71. Kunitake, G., Stitt, C. & Saltman, P. Dark fixation of regulation form the foundations for C4 gene acid metabolism in Erycina (Oncidiinae: Orchidaceae).

CO2 by tobacco leaves. Plant Physiol. 34, 123–127 expression. Nat. Plants 2, 16161 (2016). Front. Plant Sci. 9, 2000 (2019). (1959). 96. Cao, C. et al. Evidence for the role of transposons in 121. Hartwell, J. The co-​ordination of central plant 72. Dhindsa, R. S., Beasley, C. A. & Ting, I. P. the recruitment of cis-​regulatory motifs during the metabolism by the circadian clock. Biochem. Soc.

Osmoregulation in cotton fiber: accumulation of evolution of C4 photosynthesis. BMC Genomics 17, Trans. 33, 945–948 (2005). potassium and malate during growth. Plant Physiol. 201 (2016). 122. Ceusters, J. et al. Light quality modulates metabolic

56, 394–398 (1975). 97. Wang, X. et al. Comparative genomic analysis of C4 synchronization over the diel phases of Crassulacean

73. Hibberd, J. M. & Quick, W. P. Characteristics of C4 photosynthetic pathway evolution in grasses. Genome acid metabolism. J. Exp. Bot. 65, 3705–3714

photosynthesis in stems and petioles of C3 flowering Biol. 10, R68 (2009). (2014).

plants. Nature 415, 451–454 (2002). 98. Christin, P.-A. et al. Evolution of C4 phosphoenolpyruvate 123. Harris, L. W. & Davies, T. J. Data from: A complete 74. Cui, H., Kong, D., Liu, X. & Hao, Y. SCARECROW, SCR-​ carboxykinase in grasses, from genotype to fossil-calibrated​ phylogeny of seed plant families as a LIKE 23 and SHORT-​ROOT control bundle sheath cell phenotype. Mol. Biol. Evol. 26, 357–365 (2009). tool for comparative analyses: an example testing the fate and function in Arabidopsis thaliana. Plant J. 78, 99. Wang, P. et al. Evolution of GOLDEN2-LIKE gene ‘time for speciation’ hypothesis. Knowledge Network

(319–327 (2014). function in C3 and C4 plants. Planta 237, 481–495 for Biocomplexity https://doi.org/10.5063/F13T9F5P 75. Slewinski, T. L. et al. Scarecrow plays a role in (2013). (2016). establishing Kranz anatomy in maize leaves. Plant Cell 100. Bianconi, M. E. et al. Gene duplication and dosage

Physiol. 53, 2030–2037 (2012). effects during the early emergence of C4 photosynthesis Acknowledgements 76. Yuan, M. et al. A single leaf of Camellia oleifera has in the grass genus Alloteropsis. J. Exp. Bot. 69, The authors thank members of the Edwards laboratory for

two types of carbon assimilation pathway, C3 and 1967–1980 (2018). their thoughtful discussions on this manuscript. K.H. is sup- crassulacean acid metabolism. Tree Physiol. 32, 101. Emms, D. M. et al. Independent and parallel evolution ported by a Donnelley Postdoctoral Fellowship through the

188–199 (2012). of new genes by gene duplication in two origins of C4 Yale Institute of Biospheric Studies. Additional support came 77. Blount, Z. D., Borland, C. Z. & Lenski, R. E. Historical photosynthesis provides new insight into the from US National Science Foundation awards DEB-1252901

contingency and the evolution of a key innovation in mechanism of phloem loading in C4 species. Mol. Biol. and IOS-1754662 to E.J.E. P.-A.C. is supported by a Royal an experimental population of Escherichia coli. Proc. Evol. 33, 1796–1806 (2016). Society Research Fellowship (grant number URF120119). Natl Acad. Sci. USA 105, 7899–7906 (2008). 102. Huang, P. et al. Cross species selection scans identify

78. Marshall, D. M. et al. Cleome, a genus closely related components of C4 photosynthesis in the grasses. Author contributions to Arabidopsis, contains species spanning a J. Exp. Bot. 68, 127–135 (2017). The authors contributed equally to all aspects of the article.

developmental progression from C3 to C4 103. Matasci, N. et al. Data access for the 1,000 Plants

photosynthesis: C4 photosynthesis in Cleome. Plant J. (1KP) project. Gigascience 3, 17 (2014). Competing interests 51, 886–896 (2007). 104. Wickett, N. J. et al. Phylotranscriptomic analysis of the The authors declare no competing interests.

79. Sage, R. F. A portrait of the C4 photosynthetic family origin and early diversification of land plants. Proc. on the 50th anniversary of its discovery: species Natl Acad. Sci. USA 111, E4859–E4868 (2014). Publisher’s note number, evolutionary lineages, and Hall of Fame. 105. Deng, H. et al. Evolutionary history of PEPC genes in Springer Nature remains neutral with regard to jurisdictional J. Exp. Bot. 67, 4039–4056 (2016). green plants: implications for the evolution of CAM claims in published maps and institutional affiliations. 80. Males, J. Concerted anatomical change associated in orchids. Mol. Phylogenet. Evol. 94 (Suppl. B), with crassulacean acid metabolism in the 559–564 (2016). Reviewer information Bromeliaceae. Funct. Plant Biol. 45, 681–695 (2018). 106. Lundgren, M. R., Osborne, C. P. & Christin, P.-A. Nature Reviews Genetics thanks R. VanBuren and the other

81. Zambrano, V. A. B. et al. Leaf anatomical traits which Deconstructing Kranz anatomy to understand C4 anonymous reviewer(s) for their contribution to the peer accommodate the facultative engagement of evolution. J. Exp. Bot. 65, 3357–3369 (2014). review of this work.

crassulacean acid metabolism in tropical trees of the 107. Freitag, H. & Kadereit, G. C3 and C4 leaf anatomy genus Clusia. J. Exp. Bot. 65, 3513–3523 (2014). types in Camphorosmeae (Camphorosmoideae, Supplementary information 82. Heyduk, K. et al. Evolution of CAM anatomy predates Chenopodiaceae). Plant Syst. Evol. 300, 665–687 Supplementary information is available for this paper at the origins of crassulacean acid metabolism in the (2014). https://doi.org/10.1038/s41576-019-0107-5.

NATure RevIewS | GEnEtiCs volume 20 | AUGUST 2019 | 493