Independent and Parallel Recruitment of Preexisting Mechanisms Underlying C4 Photosynthesis

See allHide authors and affiliations

Science  18 Mar 2011:
Vol. 331, Issue 6023, pp. 1436-1439
DOI: 10.1126/science.1201248


C4 photosynthesis allows increased photosynthetic efficiency because carbon dioxide (CO2) is concentrated around the key enzyme RuBisCO. Leaves of C4 plants exhibit modified biochemistry, cell biology, and leaf development, but despite this complexity, C4 photosynthesis has evolved independently in at least 45 lineages of plants. We found that two independent lineages of C4 plant, whose last common ancestor predates the divergence of monocotyledons and dicotyledons about 180 million years ago, show conserved mechanisms controlling the expression of genes important for release of CO2 around RuBisCO in bundle sheath (BS) cells. Orthologous genes from monocotyledonous and dicotyledonous C3 species also contained conserved regulatory elements that conferred BS specificity when placed into C4 species. We conclude that these conserved functional genetic elements likely facilitated the repeated evolution of C4 photosynthesis.

Plants use CO2 as a substrate to make sugars through a complex biochemical process involving the Sun’s energy. The key photosynthesis enzyme RuBisCO (ribulose-1,5-bisphosphate carboxylase-oxygenase) fixes CO2 into a three-carbon molecule, but it does not completely distinguish between O2 and CO2 (1). Because of this, at least 45 lineages from 19 families of flowering plants have evolved C4 carbon–concentrating mechanisms that enhance the concentration of CO2 around RuBisCO and increase the efficiency of photosynthesis by reducing the energy wasted when RuBisCO fixes O2 (2).

All C4 lineages generate high concentrations of CO2 around RuBisCO by compartmentalizing photosynthesis, either within or between cells (3, 4), and normally carbon is fixed in mesophyll (M) cells to generate high concentrations of a four-carbon acid. The acid then diffuses through plasmodesmata into adjacent cells, often the cells of the bundle sheath (BS), where decarboxylases release CO2 around RuBisCO (3). In some cases, multiple C4 acid decarboxylases are used to deliver CO2 to RuBisCO (5, 6). High activities of these C4 acid decarboxylases around the veins of C3 species may have facilitated the polyphyletic evolution of C4 photosynthesis (7, 8).

The compartmentalization of photosynthesis results from substantial differences in gene expression in C4 leaves relative to those of C3 species (9, 10). Genes important for C4 photosynthesis have been recruited from orthologs present in C3 species, and in many cases this has involved alterations in spatial patterns of expression, restricting or enhancing the accumulation of proteins to specific cell types (11, 12). For example, in most cases, carbonic anhydrase, phosphoenolpyruvate carboxylase, and pyruvateorthophosphate dikinase accumulate preferentially in M cells, whereas a C4 acid decarboxylase and RuBisCO accumulate in BS cells (13, 14). As genes are recruited into C4 photosynthesis, alterations to the sequence of C3 genes can lead to cis-elements that allow accumulation of the cognate protein in M or BS cells (12). Despite large differences in gene expression between M and BS cells of C4 leaves (9, 10) and the recruitment of some genes belonging to multigene families into the C4 pathway, no cis-elements have been identified that regulate multiple genes required for C4 photosynthesis (1517).

NAD-dependent malic enzyme (NAD-ME) accumulates in BS cells of NAD-ME–type C4 plants, where it releases CO2 around RuBisCO (18). Because NAD-ME is encoded by two genes that generate a heterodimer (8), we used it as a model to understand how gene families are recruited into C4 photosynthesis. We isolated full-length transcript sequences for two NAD-ME genes from Cleome gynandra, which is the C4 plant most closely related to the model Arabidopsis thaliana (18). RNA hybridization in situ showed that CgNAD-ME1&2 transcripts are specific to BS cells (Fig. 1, A and B). Genomic sequences for the two genes were isolated, and a translational fusion between CgNAD-ME1 (promoter, exons, and introns) and uidA encoding the β-glucuronidase (GUS) reporter was generated. Stable transformants of C. gynandra expressing CgNAD-ME1::uidA accumulated GUS specifically in BS cells (Fig. 1C). When introns were removed and the CgNAD-ME1 promoter was replaced by the CaMV 35S promoter that directs expression in both M and BS cells (17, 19), GUS also accumulated specifically in BS cells (Fig. 1D). This result contrasted with the accumulation of GUS in both M and BS cells when the CgNAD-ME1 gene was not present (Fig. 1E and fig. S1A). These data indicate that sequences in the transcribed region of the gene are sufficient for BS-specific accumulation of CgNAD-ME1.

Fig. 1

(A and B) CgNAD-ME1&2 transcripts in C. gynandra BS cells. Scale bars, 50 μm. (C and D) GUS in the BS of C. gynandra transformants containing uidA fused to CgNAD-ME1 under control of the CgNAD-ME1 promoter (C) or the CaMV 35S promoter (D). (E) Leaf from C. gynandra transformant containing uidA under control of the CaMV 35S promoter. Scale bars in (C) to (E), 100 μm. (F to H) C. gynandra M and BS cells accumulate GUS after microprojectile bombardment of CaMV 35S driving uidA (F) or the spliced coding regions of CgNAD-ME1 (G) and CgNAD-ME2 (H). Construct details are shown in fig. S5.

Microprojectile bombardment of pCaMV 35S::CgNAD-ME1::uidA generated GUS specifically in BS cells (Fig. 1G and fig. S1B), in contrast to a CaMV 35S::uidA control in which M and BS cells accumulated GUS in roughly equal proportions (Fig. 1F). The full-length coding region of CgNAD-ME2 under control of the CaMV 35S promoter also resulted in BS-specific accumulation of GUS in C. gynandra (Fig. 1H). This suggests that for both CgNAD-ME genes the same mechanism generates BS-specific accumulation.

A 3′ deletion series of CgNAD-ME1 indicated that 240 nucleotides toward the 5′ end were sufficient for BS-specific accumulation of GUS, as was the equivalent region of CgNAD-ME2 (Fig. 2, A and B, and fig. S1, C to E). This BS specificity was due to defined sequence, because when we placed 240 nucleotides from the 3′ end of the gene in this 5′ position, GUS accumulated in both M and BS cells (fig. S1F). Replacing the CgNAD-ME1 5′ untranslated region (UTR) with that from the CaMV 35S promoter also maintained BS specificity (Fig. 2C), whereas removal of the CgNAD-ME1 coding region, leaving only the 5′UTR, led to GUS accumulation in both M and BS cells (Fig. 2D). These data indicate that the CgNAD-ME1 5′UTR is neither necessary nor sufficient for BS specificity.

Fig. 2

(A to F) C. gynandra M and BS cells accumulating GUS after microprojectile bombardment of 240 base pairs (bp) of CgNAD-ME1 (A) or CgNAD-ME2 (B), the CaMV 35S promoter with its own 5′UTR driving expression of 240 bp of CgNAD-ME1 (C) or of CgNAD-ME1 5′UTR-uidA alone (D), the 5′ 240 bp of CgNADME1 placed upstream of CaMV 35S (E), or the 5′ 240 bp of CgNAD-ME1 downstream of the uidA stop codon (F). Construct details are shown in fig. S6.

Placing the 240-nucleotide fragment upstream of the CaMV 35S promoter resulted in loss of BS specificity (Fig. 2E), indicating that these 240 nucleotides must be present in the transcript; this finding suggested a posttranscriptional mechanism. When the 240-nucleotide fragment was inserted downstream of the predicted stop codon, this also resulted in loss of BS specificity (Fig. 2F). A frameshift construct that maintained nucleotide sequence but translated into different amino acids still conferred BS specificity (fig. S1G). Thus, 240 nucleotides from the coding region of both CgNAD-ME1&2 genes is sufficient for BS-specific accumulation in C. gynandra. This region must be situated in both the transcribed and translated region of the gene, but amino acid sequence is not important. This conclusion contrasts with reports that cis-elements governing cell specificity are present in promoter regions or UTRs (11, 1517).

Analysis of orthologous NAD-ME genes from the closely related C3 species A. thaliana indicated that nucleotide sequences within the mature coding regions were on average 81% identical to those in C. gynandra. We introduced intact AtNAD-ME genes fused to uidA into C. gynandra. Stable transformants of C. gynandra expressing AtNAD-ME1::uidA accumulated GUS specifically in BS cells (Fig. 3A). Promoters of genes from C3 species that encode proteins recruited into the C4 cycle do not generate M or BS specificity in C4 species (20, 21). BS specificity was also observed when either AtNAD-ME1::uidA or AtNAD-ME2::uidA were used in microprojectile bombardment of C. gynandra (Fig. 3, B and C). The 5′UTRs and spliced coding regions of the A. thaliana NAD-ME1&2 genes under control of the CaMV 35S promoter were sufficient to generate GUS accumulation in C. gynandra BS cells (Fig. 3, D and E). A 3′ deletion series of each of the AtNAD-ME coding regions indicated that 240 nucleotides from the 5′ end of each led to accumulation of GUS specifically in the BS of C. gynandra (Fig. 3, F and G).

Fig. 3

(A) GUS accumulation in BS cells of C. gynandra transformant containing uidA fused to AtNAD-ME1. Scale bar, 100 μm. (B to G) C. gynandra M and BS cells accumulating GUS after microprojectile bombardment of AtNAD-ME1 (A) and AtNAD-ME2 (B), spliced coding regions of AtNAD-ME1 (D) and AtNAD-ME2 (E), or the 5′ 240 bp of AtNAD-ME1 (F) or AtNAD-ME2 (G). (H and I) Intact AtNAD-ME1 gene fused to uidA generates relatively constitutive accumulation of GUS in A. thaliana. Construct details are shown in fig. S7. Scale bars, 5 mm (H), 50 μm (I).

In A. thaliana, fusions between uidA and intact AtNAD-ME1&2 generated GUS in multiple cell types, including mid-vein, M, and BS cells (Fig. 3, H and I, and fig. S2), and promoter regions are responsible for these patterns (8). Thus, the promoter is sufficient to control spatial patterns of gene expression in C3 A. thaliana, whereas in C4 C. gynandra, elements in the coding region (also found in the C3 NAD-ME genes) have been recruited to recognize one or more trans-factors that generate BS specificity.

Domains of NAD-ME genes encoding the mitochondrial C4 acid decarboxylase and NADP-ME genes encoding the chloroplastic C4 acid decarboxylase show conservation among lineages of C4 plants. For example, NADP-ME from C4 maize has a 420-nucleotide region that is similar to the CgNAD-ME genes (fig. S3) (it is longer because the predicted chloroplast transit peptide is longer than that of the mitochondrial targeting sequence of the C. gynandra NAD-ME). Microprojectile bombardment of this region from ZmNADP-ME into C. gynandra led to BS-specific accumulation of GUS (Fig. 4A), as did the orthologous rice NADP-ME coding sequence (Fig. 4B).

Fig. 4

(A and B) C. gynandra M and BS cells accumulating GUS after microprojectile bombardment of 420 bp of maize ZmNADP-ME (A) or 429 bp of rice OsNADP-ME (B) under control of the CaMV 35S promoter. (C) Maize M and BS cells accumulating GUS after microprojectile bombardment of 240 bp of CgNADME1. (D) Representative image of maize BS cell accumulating GUS after microprojectile bombardment of 240 bp of CgNAD-ME1. Large rectangular BS cells lie adjacent to veins; M cells lacking GUS are circular. Scale bar, 50 μm. (E) The CaMV 35S promoter generates preferential expression of uidA in maize M cells after microprojectile bombardment. (F) A 420-bp sequence from ZmNADP-ME1 is sufficient for preferential accumulation in BS cells of maize. Construct details are shown in fig. S8.

Microprojectile bombardment of maize with the 240-nucleotide sequence from CgNAD-ME1 or AtNAD-ME1 fused to uidA also produced preferential accumulation of GUS in BS cells (Fig. 4, C and D, and fig. S4A). The CaMV 35S::uidA control generated preferential accumulation of GUS in M cells (Fig. 4E), which likely is associated with suberization of maize BS cell walls inhibiting entry of microprojectiles (19). However, for both ZmNADP-ME and OsNADP-ME genes, preferential BS expression was observed (Fig. 4F and fig. S4B). Promoter and 5′ UTR regions of maize NADP-ME generate preferential accumulation of GUS in the BS (22), and so both transcriptional and posttranscriptional mechanisms are likely responsible for ZmNADP-ME BS-specific expression.

Our results indicate that these separate lineages of C4 plants likely possess functionally equivalent mechanisms that control the accumulation of proteins important for C4 photosynthesis in BS cells. Because these sequences are present in C3 Arabidopsis and rice and are functional within C4 C. gynandra and maize, we conclude that these genes show a conserved functional latency. This evolutionary parallelism may be explained by the hypothesis that a number of trans-factors with similar properties (or possibly only one such factor) are responsible for the accumulation of NAD-ME or NADP-ME proteins in the BS. This hypothesis is supported by the fact that genes from C3 species are recognized by the regulatory circuitry responsible for generating expression in BS cells of multiple C4 lineages. Our data also indicate the importance of posttranscriptional control of NAD-ME genes. Loss of BS specificity when the 240-nucleotide element is present in the transcript but downstream of the stop codon argues against regulation by a small RNA, although it is possible that an RNA binding protein recognizes this 240-nucleotide sequence. Irrespective of the exact mechanism, it is noteworthy that separate lineages of C4 plants show similar regulation, and so the trans-factors required would have been repeatedly recruited into C4 photosynthesis.

Our data suggest that cell-specific accumulation of proteins in C4 leaves can be generated without sequence alterations to genes that encode them. This indicates that genes from C3 species can be directly recruited into C4 photosynthesis and implies that an initial step in this process involves modification of trans-factors to make use of existing cis-elements to generate cell specificity. Subsequent regulatory and functional alterations can follow (12, 23). Our data also indicate that cell specificity in the C4 leaf can be generated by elements in the coding region of genes. Regulation of cell type specificity in the C4 leaf by conserved elements in coding sequences may be a mechanism that facilitates the coordinate recruitment of multigene families into the pathway.

These results suggest that engineering C4 photosynthesis in rice to increase yield may be possible (24, 25). The fact that genes present in C3 species can be recruited into cell-specific functions in the C4 pathway without alterations to sequence indicates that the trans-factor(s) responsible for generating cell-specific accumulation of proteins must be identified for potential manipulation in rice.

Supporting Online Material

Materials and Methods

Figs. S1 to S9


References and Notes

  1. We thank the Leverhulme Trust and BBSRC for funding; C. Weil for B73; and M. Patel, J. Berry, S. Brockington, and D. Bradley for advice. N.J.B., J.E.C., and K.K. generated constructs; N.J.B., S.S., J.E.C., and A.J.P. performed microprojectile bombardment; C.A.N. and N.J.B. generated and analyzed stable transformants; N.J.B. and S.S. performed hybridization in situ; and N.J.B. and J.M.H. designed the experiments, analyzed the data, and wrote the paper. The complete study was supervised by J.M.H. All authors discussed the results and commented on the manuscript.
View Abstract

Navigate This Article