Functional divergence enabled Cissus adaptation to aridity
In Cissus , 675 orthogroups were remarkably expanded (P < 0.05) and 232 were diminished (P < 0.05) compared to other representative eudicots (Figure S3). The expanded orthogroups were mainly enriched in the abiotic/biotic stress-responsive pathways, metabolism of carbohydrates, and hormone biosynthesis (Figure 3a).Cytochrome P450 , found in dramatic proliferation (Table S14), could contribute to the foliar wax deposition in Cissus (Shepherd and Wynne Griffiths, 2006). Some polysaccharide-related genes, such as pectate lyase, pectinesterase, and polysaccharide biosynthesis genes, also displayed an increased paralogous number, probably attributable to the succulent leaf formation through the modification of pectin and other polysaccharides in cells (Griffiths and Males, 2017). Apart from the genes that directly contributed to the leaf character, transcription factors like MYB , WRKY , AP2/ERF , GRAS , andLEA were strongly expanded (Figure 3a), suggesting that the probable enhanced abiotic stress resistance and secondary metabolism inCissus (Dubos et al., 2010; Gao and Lan, 2016; Jiang and Rao, 2020). To further address the functional divergence of Cissusreferring to adaptation, we compared its gene repertoire to grape which indicated that selective amplification of genes belonging to plant immunity had occurred in these two species (Figure S6, Table S15). Among respective orthogroups, nucleotide-binding site leucine-rich repeat (NBS-LRR ) genes were found in favor of expansion in both species but showed preference for different subclasses (e.g., orthogroup 12 in the grape; orthogroup 4,7 in Cissus ). The co-abundance ofR genes would represent the basic objective of an organism to protecting itself against the surging threats from microbial pathogens (Plomion et al., 2018; Tobias and Guest, 2014). The significant copy number variation of paralogous genes (orthogroup 13: terpenoid cyclase, orthogroup 2: TMV resistance protein N-like) likely suggested the different responses to pathogens induction (Mestre and Baulcombe, 2006; Warren et al., 2015). Additionally, small heat shock proteins (sHSPs),HSP20-like were found particularly amplified in Cissus and upregulated in its shoot and leaves compared with root (Figure S6, Table S16). This would fairly reflect the increased ability of Cissus ’s vegetative organs to deal with heat shock and promote resistance to environmental stress factors (Bondino et al., 2012; Guo et al., 2020). The enrichment pattern of the gene family in Cissus led us to investigate if a similar preference for gene proliferation occurred in other succulent species. To this end, we took another four typical succulent plants (Ananas comosus , Hylocereus undatus ,Kalanchoe fedtschenkoi, and Kalanchoe laxiflora ) into account on the gene family comparison. We found that 88 of the 97,335 orthogroups demonstrated as succulent-specific expansion, which significantly (P -adjust < 0.05) enriched in ‘terpene synthase ’, ‘HSP20 ’ (Figure 3b, Tables S17-S19). However, 178 orthogroups GO termed mainly as serine/threonine-protein kinase receptor precursor (SKR ), cysteine-rich receptor-like protein kinase (CRK ), wall-associated receptor kinase (RLK ) were observed in co-expansion in the other 13 non-succulent plant genomes investigated (Figure 3b, Tables S17 and S18, Table S20). The diverged preference for functional gene families would reflect a specialized convergent mechanism in succulent plants dealing with high temperatures and water deficiency (Griffiths and Males, 2017). On the other hand, we identified 1,878 tandemly duplicate (TD) arrays of two or more genes in Cissus , and the total number of genes in such arrays is 4,746, slightly higher than 3,958 genes in 1,524 TD arrays in grape (Table S21). There are 2,582 TD genes shared in two species, whose functional classification is mainly enriched in 134 GO terms (e.g., oxidoreductase activity, oxidation-reduction process, and response to auxin), and a total of 2,164 TD genes are species-specific inCissus (Table S22). Functional bias in TD retention was observed encompassing different periods of evolution in Cissus (Figure 3c). An overrepresented number of genes in the Cissus lineage were enriched in cell wall-related pathways (e.g., cell wall modification, cell wall organization and xyloglucan metabolic activity), probably, conferred to its succulent leaves or stems (Ahl et al., 2019). In contrast, functional categories specific to grapes were mainly associated with stress responses (Figure 3c, Table S23). The result is consistent with the notion that TD genes would have a lineage-specific selection (Freeling, 2009). Nevertheless, earlier studies inArabidopsis and rice demonstrated that an elevated probability of retention of stress-responsive TD is preferential for adaptive evolution after speciation (Hanada et al., 2008; Rizzon et al., 2006). The functional bias of TD in Cissus indicates genes referred to as morphological innovation for adaptation might be particularly selected and expanded via local duplication. It would be interesting to check if a similar profile of lineage-specific TD is exhibited in other morphology-specialized plants. We found that lineage-TD genes categorized as ‘cellular component-related’ and ‘resistance’ were overrepresented in gross tandem duplicated genes in succulent species, in contrast to the discrete pattern that occurred in the other non-succulent plants (P = 0.03 and P = 0.007) (Figure 3d, Table S24). Local gene amplification with a high frequency of gene birth/death plays a critical role in plants’ adaptive responses to environmental stimuli and is mostly attributable to gene copy number and allelic variation within a population (Hanada et al., 2008; Jiang and Rao, 2020). The succulent fashion of TD expansion observed here would suggest another pattern of functional bias in TD retention during seed plant evolution. We speculated that the intense environment change provided multi-options for plants on morphological innovation and rapid expansion of resistance genes.
CAM photosynthesis in C. rotundifolia
CAM photosynthesis is a recurrently evolved strategy for high water use efficiency (WUE), enabling plants to survive in water-limited environments (Silvera et al., 2010). In CAM plants, the carbon dioxide (CO2) is fixed in the cytosol and stored as malic acid in the vacuole during the night (Figure 4c). The stomata remained closed during the daytime to decrease water loss by evapotranspiration, and the stored malic acid is decarboxylated to release CO2 that could be re-fixed through the Calvin-Benson cycle (Borland et al., 2014). Such a feature of CO2 uptake was ubiquitous inCissus lineage, which may have facilitated the spread of the genus from wet into arid tropics (DeSanto and Bartoli, 1996).
To investigate the CAM evolution in Cissus , we determined the pattern of diurnal oscillation of titratable acidity in the leaves ofC. rotundifolia in growth chambers with a climate close to the dry seasons in Kenya (Figure 4a). The amount of titratable acid reached its maximum (150 μeq g-1 FW) early in the dawn (~ 6:00 a.m.) and dropped to its minimum (20 μeq g-1 FW) later in the day (~ 6:00 p.m.), which qualified C. rotundifolia as a CAM species (Nelson and Sage, 2008; Sayed, 2001). We identified 47 candidate CAM pathway genes based on their orthologs in pineapple (Ananas comosus L. Merr., CAM plant) (Ming et al., 2015), maize (Zea mays L., C4 plant) (Schnable et al., 2009), rice (Oryza sativa L., C3 plant),Kalanchoe fedtschenkoi (CAM plant) (Yang et al., 2017), andPhalaenopsis equestris (CAM plant) (Cai et al., 2015). Further, these genes were well categorized into nine gene families that characterized the core network of carboxylation and decarboxylation pathways (Ming et al., 2015) (Table S25). These gene families showed no significant expansions in C. rotundifolia compared with other plants, as shown in Table S26, implying that CAM photosynthesis might evolve through the re-organization of existing enzymes (Chen et al., 2020).
The diurnal expression patterns of these CAM genes were interrogated by transcriptome comparison of leaves during 3-hour intervals over a 24-hour period. In general, the expression of 21 genes showed typical circadian patterns as defined via a polynomial regression (Figure 4b). The transcripts of enzymes involved in carbon assimilation such ascarbonic anhydrase (CA ), phosphoenolpyruvate carboxylase kinase (PPCK ), and malate dehydrogenase(MDH ) were highly accumulated at night. As the core gene involved in CO2 fixation, four PEPC genes of Cissus have extremely high expression in the daytime rather than at nighttime (Figure S7). Similar expression patterns were also found in other CAM plants, such as Kaladp0095s0055.1 in Kalanchoe andSal_001109 in Sedum album (Abraham et al., 2020; Wai et al., 2019; Yang et al., 2017; Zhang et al., 2016). Correspondingly, the enzymes that participate in decarboxylation processes, such asMDH , ME-NADP , and phosphoenolpyruvate carboxykinase(PEPCK ) were highly expressed during the day (Figure 4b). Interestingly, as a major protein for carbon fixation, previous studies in pineapple have shown that only βCA subfamily is expressed at nighttime and early morning in green leaf tissues (Ming et al., 2015). We observed all five CAs including α (1), β (3), and γ (1) expressed at night in C. rotundifolia (Figure 4b). The expression of βCA1 in Cissus and pineapple (Rcr_ac> 0.8) increased during the night, and a peak occurred at 9:00 in the morning. While its orthologs in Arabidopsis(Rcr_at < 0.5) showed stable and lower expression during the diurnal cycle (Figure S8, Table S4) (Mockler et al., 2007). Beyond that, members of the MDH also showed diverged expression patterns as MDH2 was more active at night while the other four MDHs were upregulated during the day, consistent with other CAM plants (Ming et al., 2015; Yang et al., 2017; Wickell et al., 2021) (Figure 4b). This may be associated with their different roles in decarboxylation processes since MDH catalyzes the reversible reaction between oxaloacetic acid and malic acid.
We constructed the gene co-expression network based on the transcriptome data from nine mature leaf samples collected every three hours over a 24-hour period. Among 27 modules identified, MEbrown2 (2,020 genes that were highly expressed during the night) was significantly (P< 0.05) related to the night period (Figure S9). We found thatβCA2 , βCA3, and γCA were also found in this MEbrown2 module. Pathways such as response to organonitrogen compound and root meristem growth in this module were significantly enriched in this module (Table S27). MEdarkorange2 module (311 genes that were highly expressed in the day) was found to be significantly associated with the day period. We found PEPCK , PPDK , MDH6,and ALMT s in this module. Biological processes such as response to abiotic stimulus were enriched in this module (Table S27).
Moreover, transcripts in leaf with time-course diel expression patterns were classified into 9 clusters (Figure S10, Table S28). The highly connected hubs genes identified by network construction for each cluster were associated with CAM genes. For example, Cluster 4 containedPPCK2 (CRGY0218762) and γCA (CRGY0214246) and had patatin-like phospholipase as the hub (Figure 4d, Table S29). Heat shock protein, which played important roles during stress responses in many plants, was the hub in Cluster5 and connected withPEPC1 and PEPC5 (Figure 4e, Table S29).
The promoters of the diurnally expressed photosynthetic genes were enriched in circadian clock-related cis -elements (Chen et al., 2020; Michael et al., 2008) (Figure 4b). Comparative analysis betweenCissus , pineapple, rice, maize, and sorghum showed that onlyβCA1 with typical circadian patterns in Cissus had one EE (Table S30) (Ming et al., 2015), suggesting its contribution to CO2 fixation via combination EE motif during nighttime (Wai et al., 2019). Additional comparison within CAM genes indicated that EE and G-box elements were mainly enriched in the subgroups of highly expressed genes at night (Table S30).
The higher WUE in CAM plants relied on the appropriate control of stomatal movement during day and night. We identified the stomata open/close related genes in the C. rotundifolia genome based on their homologs in Arabidopsis (Chen et al., 2020) (Table S31). A subset of genes that are responsible for the stomata opening or closing were uniquely expressed either at night or during the daytime, which implied the coincidental organization of stomata movement and CAM genes (Figure S11). The expression patterns of stomata movement genes were compared to their orthologs in Arabidopsis (Table S4). We identified 86 out of 141 stomata movement genes with diurnal expression patterns in Cissus (Table S31). The diurnal expression of 64 genes showed a low correlation with its orthologs in Arabidopsis(Figure S11), suggesting their putative roles during stomatal movement in Cissus . OST1 (Stomatal opening factor1), which plays a vital role in abscisic acid (ABA) triggered stomatal closure (Mustilli et al., 2002), was found to be highly expressed at 9:00 (Figure S11), compared with accumulated transcription of its orthologs at night inArabidopsis . The result was also consistent with diel expression patterns of OST1 in A. americana , K. laxiflora, andKalanchoë (Boxall et al., 2020; Abraham et al., 2016; Abraham et al., 2020). Interestingly, the MOE and G-box motifs were enriched in the promoter of OST1 in C. rotundifolia but not in V. vinifera and Arabidopsis (Table S31). These results indicate coordinated transcriptional regulation of circadian rhythm and stomatal movement-related genes with evolved CAM in C. rotundifolia .