Unique photosynthetic phenotypes in Portulaca (Portulacaceae): C3-C4 intermediates and NAD-ME C4 species with Pilosoid-type Kranz anatomy

Portulacaceae shows great diversity in C4 photosynthetic phenotypes: all species in clade Cryptopetala are C3-C4 intermediates, while clade Pilosa has a unique anatomical form of Kranz with diversity in C4 biochemistry.


Introduction
Most terrestrial plants have C 3 -type photosynthesis, in which there is direct fixation of atmospheric CO 2 via ribulose bisphosphate carboxylase oxygenase (Rubisco) in all chloroplast-containing leaf mesophyll (M) tissues. Only ~3% of terrestrial plant species are C 4 , in which atmospheric CO 2 is fixed by phosphoenolpyruvate carboxylase (PEPC) located in M cells, with the production of initial compounds containing four carbon atoms. Through the CO 2 concentrating mechanism and repression of photorespiration, C 4 plants are recognized as having an increased capacity for carbon assimilation and higher efficiency in nitrogen and water use in warm climates. For this reason, there is interest in genetically modifying major C 3 crops with C 4 traits to reduce photorespiration (Brown, 1999;Sheehy et al., 2007). The main structural difference associated with this pathway in most C 4 species is the specialized leaf anatomy (called Kranz type) with close coordination of function between two types of cells surrounding vascular bundles (VB), the enlarged chlorenchymatous bundle sheath (BS) cells and the radially arranged M cells (Edwards and Walker, 1983;Hatch, 1987;Kanai and. C 4 photosynthesis has been found in 19 families of angiosperm plants, 16 of which correspond to dicot lineages (Sage, 2004).
In the order Caryophyllales there are 23 families (Thorne and Reveal, 2007), eight of which have species with C 4 photosynthesis: Aizoaceae, Amaranthaceae, Caryophyllaceae, Gisekiaceae, Molluginaceae, Nyctaginaceae, Polygonaceae and Portulacaceae (Sage, 2004). Portulacaceae has 29 genera as traditionally circumscribed (Eggli, 2002), although it is currently considered as a family with only one genus, Portulaca (Nyffeler and Eggli, 2010;Hernández-Ledesma et al., 2015), which is known to have C 4 species. Previous suggestions that some species of Anacampseros and Grahamia (once considered as members of Portulacaceae but now circumscribed in Anacampserotaceae) may be C 4 are not supported by a recent study (Guralnick et al., 2008), which indicates the occurrence of Crassulacean acid metabolism (CAM) rather than C 4 photosynthesis in these genera.
For many years it was accepted that the genus Portulaca includes only species having Kranz-type anatomy and C 4 photosynthesis with two clear groups. One group was defined as having species with NAD-malic enzyme (NAD-ME)-type C 4 cycle and well-developed grana in BS cell chloroplasts, as in P. oleracea L. (Laetsch, 1971(Laetsch, , 1974Gutierrez et al., 1974;Carolin et al., 1978;Sprey and Laetsch, 1978). Another group was described as having NADP-malic enzyme (NADP-ME) subtype species with agranal BS chloroplasts, with P. grandiflora Hook. being a well-known representative species (Gutierrez et al., 1974;Carolin et al., 1978). This diversity was a stimulus for more complex studies of representative species in the genus. A structural and functional analysis showed the existence of four clearly distinct C 4 types of leaf anatomy, which differ in the way Kranz anatomy is formed with respect to the position of the VB . Phylogenetic analyses of the suborder Cactineae, which included genus Portulaca, showed the existence of two clearly defined clades; the opposite-leaved clade (OL), and the alternate-leaved clade (AL), mostly corresponding to two previously recognized subgenera, Portulacella and Portulaca Columbus, 2010, 2012). There are two welldefined clades within the OL clade representing Australian and African-Asian species (Ocampo and Columbus, 2012). All representatives which have been studied in the OL clade have NADP-ME-type biochemistry, and a unique form of leaf anatomy (Portulacelloid type) where Kranz is formed around individual VB, which are located towards the adaxial side of the leaf, with several layers of water storage (WS) cells located towards the abaxial side Ocampo et al., 2013). In the AL clade, from studies of representative species, there are four clades with differences in anatomy and forms of photosynthesis: Oleracea, Umbraticola, Pilosa and Cryptopetala. The Umbraticola clade is reported to have NADP-ME species with Atriplicoidtype anatomy where Kranz forms around individual veins in planar leaves. The Oleracea clade has NAD-ME-type C 4 species with a specific variant of Atriplicoid-type leaf anatomy . The Pilosa clade is reported to have NADP-ME-type C 4 species with Pilosoid-type anatomy in which Kranz tissue encloses individual peripheral VB with WS tissue located in the center of the leaf. In the Cryptopetala clade, which is sister to the Oleracea clade, one species, P. cryptopetala, was shown to be a C 3 -C 4 intermediate species ; P. hirsutissima and P. mucronata are two other species in this clade which have C 3 -type carbon isotope composition, and from examination of leaf lamina of herbarium specimens, an apparent lack of Kranz anatomy (Ocampo et al., 2013). Since evolution of C 4 from C 3 species is considered a stepwise process with intermediate states (Sage et al., 2012), the goal of the present work was to fully characterize photosynthesis in species in clade Cryptopetala, and to analyze forms of photosynthesis in representative species in the under-studied Pilosa clade, which has a form of Kranz only found in Portulacaceae  and Aizoaceae (Bohley et al., 2015).

Plant material
The sources of seeds and plants of species in this study are provided in Table 1. All seeds were stored at 3-5 ºC prior to use and were germinated on the surface of potting soil (Sunshine LC-1 from SUNGRO Horticulture, Bellevue, WA, USA) at 25 ºC and a photosynthetic photon flux density (PPFD) of 100 µmol quanta m -2 s -1 . The seedlings were then transplanted to soil in 10 cm diameter pots (one seedling per pot). After ~ 1 week established plants were transferred to a greenhouse with day/night temperatures 26/18 °C and a maximum mid-day PPFD of 1000 µmol photosynthetic quanta m −2 s −1 . Plants were fertilized once per week with Peter's Professional (20:20:20;. For microscopy and biochemical analyses, samples of mature leaves were taken from ~2-month-old plants. Cotyledons were fixed ~2 weeks after germination when the first leaves were already established.

Light and electron microscopy
Hand cross sections of fresh leaves were placed in water and studied under UV light (with DAPI filter) on a Leica Fluorescence Microscope Leica DMFSA (Leica Microsystems Wetzlar GmbH, Germany).
For structural studies, two to three samples were taken from three plants for each species from cotyledons and the middle part of the leaf. They were fixed at 4 o C in 2% (v/v) paraformaldehyde and 2% (v/v) glutaraldehyde in 0.1 M phosphate buffer (pH 7.2), postfixed in 2% (w/v) OsO 4 and then, after a standard acetone dehydration procedure, embedded in Spurr's epoxy resin. Cross sections were made on a Reichert Ultracut R ultramicrotome (Reichert-Jung GmbH, Heidelberg, Germany). For light microscopy, semi-thin sections were stained with 1% (w/v) Toluidine blue O in 1% (w/v) Na 2 B 4 O 7 and studied under the Olympus BH-2 (Olympus Optical Co., Ltd.) light microscope equipped with LM Digital Camera & Software (Jenoptik ProgRes Camera, C12plus, Jena, Germany). Ultra-thin sections were stained for transmission electron microscopy with 4% (w/v) uranyl acetate followed by 2% (w/v) lead citrate. FEI Tecnai G2 (Field Emission Instruments Company, Hillsboro, OR, USA) equipped with Eagle FP 5271/82 4K HR200KV digital camera and Hitachi H-600 (Hitachi Scientific Instruments, Tokyo, Japan) transmission electron microscopes were used for observation and photography.
Observations of vascular pattern were obtained from fully expanded leaves. The samples were cleared in 70% ethanol (v/v) until chlorophyll was removed, treated with 5% (w/v) NaOH overnight and then rinsed three times in water. At least five leaves from two or three different plants were used. The leaves were mounted in water and examined under UV light (with DAPI filter) on a Fluorescence Microscope Leica DMFSA (Leica Microsystems Wetzlar GmbH, Germany) using autofluorescence of lignified tracheary elements of the xylem. The density of the venation (mm per mm 2 of the leaf surface area) was determined as the length of minor or peripheral veins per leaf area which were measured using the image analysis program ImageJ 1.37v (Wayne Rasband, National Institutes of Health, USA). Standard errors were determined and analysis of variance (ANOVA) was performed with Statistica 7.0 software (StatSoft, Inc.). Tukey's HSD (honest significant difference) test was used to analyze differences between vein density values in Portulaca species. All analyses were performed at the 95% significance level.
In situ immunolocalization Leaf samples (two to three samples from three plants for each species) were fixed at 4 o C in 2% (v/v) paraformaldehyde and 1.25% (v/v) glutaraldehyde in 0.05 M PIPES buffer, pH 7.2. The samples were dehydrated with a graded ethanol series and embedded in London Resin White (LR White, Electron Microscopy Sciences, Fort Washington, PA, USA) acrylic resin. The antibody used (raised in rabbit) was against the P subunit of mitochondrial glycine decarboxylase (GDC) IgG from Pisum sativum L. (courtesy of Dr David Oliver). Preimmune serum was used as a control.
For transmission electron microscopy immunolabeling, thin sections (~70-90 nm) on Formvar-coated nickel grids were incubated for 1 h in TBST+BSA to block non-specific protein binding on the sections. They were then incubated for 3 h with either the preimmune serum diluted in TBST+BSA or anti-GDC (1:10) antibodies. After washing with TBST+BSA, the sections were incubated for 1 h with Protein A-gold (15 nm) diluted 1:100 with TBST+BSA. The sections were washed sequentially with TBST+BSA, and TBST and distilled water, and then post-stained with a 1:3 dilution of 0.5% (w/v) potassium permanganate and 2% (w/v) uranyl acetate. Images were collected using a FEI Tecnai G2 transmission electron microscope. The density of labeling was determined by counting the gold particles on electron micrographs and calculating the number per unit area (µm 2 ) in the mitochondria, versus background labeling density in the rest of the cell. For each cell type, replicate measurements were made on parts of cell sections (n=10-60 images from at least two different experiments).

Measurements of rates of photosynthesis
Gas exchange was measured using the LI-6400XT portable photosynthesis system equipped with broadleaf chamber (LI-6400-02B) in response to varying CO 2 at 1000 µmol quanta m −2 s −1 PPFD and 25 ○ C and in response to varying light intensity at 400 µbar CO 2 and 25 ○ C. Measurements were performed in greenhouse with enclosing of one leaf or leaf portion inside of chamber. For each experiment the leaves were illuminated with 1000 PPFD under 400 µbar CO 2 and 25 ○ C until a steady state rate of CO 2 fixation was obtained (generally 30 min). For varying CO 2 experiments the CO 2 level was decreased, and then increased up to 2000 µbar at 5 min intervals. For varying light experiments, measurements were made beginning at 2000 PPFD, with decreasing levels at 4 min intervals. The CO 2 compensation point (Г) was determined by extrapolating the initial slope of CO 2 response curve through the x-axis and taking the zero intercept. The leaf area was calculated from digital image of the leaf portion enclosed in the chamber, using an image analysis program (ImageJ 1.37v).
δ 13 C values δ 13 C values, a measure of the carbon isotope composition, were determined at Washington State University on leaf samples taken from plants using a standard procedure relative to PDB (Pee Dee Belemnite) limestone as the carbon isotope standard (Bender et al., 1973). Plant samples were dried at 80 °C for 24 h, milled to a fine powder and then 1-2 mg were placed into a tin capsule and combusted in a Eurovector elemental analyzer. The resulting N 2 and CO 2 gases were separated by gas chromatography and admitted into the inlet of a Micromass Isoprime isotope ratio mass spectrometer (IRMS) for determination of 13 C/ 12 C ratios. δ 13 C values were calculated where δ 13 C=1000×(R sample /R standard )-1, where R= 13 C/ 12 C.

C 4 biochemical-type evolution
A combined data matrix of chloroplast and nuclear DNA sequences was prepared from the data used by Columbus (2010, 2012) and Ocampo et al. (2013) to obtain an evolutionary framework to estimate C 4 variant diversification within Portulacaceae. Sampling included species of Portulaca with known C 4 biochemical types. Sequences were aligned using MUSCLE version 3.7 (Edgar, 2004), followed by manual alignment in MEGA version 7.0.14 (Kumar et al., 2016). The combined data matrix was partitioned by locus and analyzed under maximum likelihood (ML; Felsenstein, 1973) in RA×ML version 7.2.6 (Stamatakis, 2006) under the GTRGAMMA model. Clade support was calculated by nonparametric bootstrapping (Felsenstein, 1985) from 10 000 replicates performed simultaneously with the ML search using the '-f a' option. The phylogenetic analysis was executed using the High Performance Computing Cluster at the California Academy of Sciences. Evolution of C 4 biochemical types was estimated by ML over the ML tree using Mesquite version 3.04 (Maddison and Maddison, 2016) under the Markov k-state 1 parameter model (Lewis, 2001).

Carbon isotope composition of leaves
The carbon isotope values of leaves of the three Portulaca species in the Crytopetala clade, P. cryptopetala, P. hirsutissima and P. mucronata, were all C 3 -like (δ 13 C −26.6 to −28.6 o / oo ). In all species in the Pilosa clade that were analyzed the values were in the range of C 4 plants (−12.1 to −17.4 o / oo ) ( Table 1). Figure 1 shows the general view of representative Portulaca species that were analyzed in the most detail in this study. Two of the species, P. cryptopetala (Fig. 1A) and P. mucronata ( Fig. 1D) have flattened leaves, while P. hirsutissima (Fig. 1G) has thicker and more succulent terete (terete means succulent circular or distorted circle shape) leaves. All three species clearly have C 3 -like dorsoventral type of anatomy with vascular bundles (VB) situated under the palisade layers, in the medium part of leaf lamina between palisade and spongy parenchyma ( Fig. 1B, C, E, F, H, I). In two species, P. cryptopetala and P. mucronata there are two layers of palisade parenchyma on the adaxial side and three to four layers of spongy parenchyma cells on the abaxial side; the VB are surrounded by BS cells (Fig. 1C, F). The general leaf anatomy in P. hirsutissima ( Fig. 1I) is similar except for the intensive development of spongy water storage (WS) parenchyma in the middle part of the leaf (Fig. 1I). In all three species the BS cells around lateral VB are rather large; but, their size is similar to the surrounding M cells (Fig. 1C, F, I).

General view of plants and leaf structure
In P. cryptopetala and P. mucronata, the main vein is located in the same plane as all lateral VBs (not shown); however; in P. hirsutissima the main vein is well below the lateral veins and it is surrounded by spongy WS parenchyma (Fig. 1H).
Representative species from the Pilosa clade (P. biloba, P. cf. gilliesii, P. elatior, P. halimoides, P. smallii and P. suffrutescens) have lanceolate semi-terete or cylindrical fleshy leaves with Kranz anatomy of Pilosoid type. As illustrated with P. elatior (Fig. 1K, L), in this type of anatomy the VB are distributed around the leaf periphery with each individual vein surrounded by two specialized chlorenchyma layers characteristic of Kranz-type anatomy. The main vein is located more or less in the center of the leaf and is surrounded by WS tissue.

Transmission electron microscopy
In three species, P. cryptopetala, P. hirsutissima and P. mucronata, the BS cells contain a significant number of organelles in the centripetal position, adjacent to VB and along the radial cell walls ( Fig. 2A, D, G). M and BS chloroplasts in all three species have a similar level of grana development with medium-sized grana (Fig. 2B, C, E, F, H, I). BS cells contain numerous centripetally arranged enlarged mitochondria ( Fig. 2A, D, G).
In all six species with Pilosoid Kranz-type leaf anatomy, the BS cells surrounding the minor veins contain numerous organelles in a centripetal position (illustrated in P. smallii and P. elatior, Fig. 2J, M). A more detailed study of M and BS ultrastructure revealed significant differences. In five species (P. biloba, P. cf. gilliesii, P. halimoides, P. smallii and P. suffrutescens), the BS cells contain nearly agranal chloroplasts while M chloroplasts have rather well-developed grana as shown for P. smallii in Fig. 2K, L. However, one species, P. elatior, has a reverse type of chloroplast ultrastructure; the chloroplasts in the BS cells have numerous mid-sized grana (Fig. 2N) while the M chloroplasts have a more extensive single thylakoid system (Fig. 2O). This species also has numerous large mitochondria with specific tubular cristae in BS cells (Fig. 2N).

Vein density
The vein density in the three species in the Cryptopetala clade (P. cryptopetala, P. hirsutissima and P. mucronata) which lack Kranz anatomy was analyzed in comparison to Kranz-type species including P. umbraticola with Atriplicoid leaf anatomy from Umbraticola clade, two species in the Oleracea clade with modified Atriplicoid-type anatomy (P. molokiniensis and two varieties of P. oleracea) and three species in the Pilosa clade with Pilosoid-type anatomy (P. amilis, P. biloba and P. grandiflora) (Fig. 3). Among these, in general species in the Cryptopetala clade had the lowest vein density while the highest vein densities were in the two species in the Oleracea clade (vein density ~1.8-fold higher than the Cryptopetala species). In the species in the Pilosa clade, the vein densities on the adaxial side of leaves were on average ~1.3-fold higher than the Cryptopetala species. Portulaca umbraticola had high vein density, appearing to be intermediate between Pilosa and Oleracea clades. The three species with Pilosoid Kranz-type leaf anatomy showed similar vein densities on the adaxial leaf side while the densities in the two species with flattened leaves (P. amilis and P. biloba) were 1.2 and 1.7 times lower on the abaxial leaf side.

Cotyledon anatomy
The species in the Cryptopetala clade, P. cryptopetala, P. hirsutissima and P. mucronata, have cotyledons with C 3 -like dorsoventral anatomy, with one to two layers of palisade-like M cells on the adaxial side and several layers of spongy M cells towards the abaxial side ( Fig. 4A-C). The number of spongy parenchyma layers varies between species: from one or two layers in P. cryptopetala and P. mucronata to mostly three in semi-terete cotyledons of P. hirsutissima with a little thicker mid-section. VB are distributed in a lateral plane in the central part of the lamina in P. cryptopetala and P. mucronata (Fig. 4A, C) and they are located closer to the adaxial side of the lamina under the palisade parenchyma in P. hirsutissima (Fig. 4B). In all three species the VB are surrounded by rather large BS cells with a higher density of organelles (chloroplasts and mitochondria) in the centripetal position or along the radial cell walls, similar to that described for leaves (structural analysis by TEM is not shown).
The cotyledons in five studied species in the Pilosa clade with Kranz-type anatomy in leaves (P. biloba, P. elatior, P. halimoides, P. smallii and P. suffrutescens) all have Atriplicoidlike Kranz anatomy with VB distributed in one lateral plane (illustrated only for P. elatior and P. smallii, Fig. 4D, E). The VB surrounded by two layers of Kranz chlorenchyma are located under the adaxial epidermis, with variable number of spongy parenchyma layers towards the abaxial epidermis, from one layer in P. elatior to mostly two layers in P. biloba, P. halimoides and P. suffrutescens. In P. smallii the VB are located closer to the adaxial side of the lamina with three/ four layers of spongy parenchyma cells on the adaxial side of the cotyledon (Fig. 4E).

Immunolabeling for GDC
In situ immunolabeling for GDC was performed for P. cryptopetala, P. hirsutissima and P. mucronata in the Cryptopetala clade in comparison to the C 4 species P. oleracea ( Fig. 5; Supplementary Fig. S1 at JXB online). Analysis of the density of immunogold particles for anti-GDC antibody shows strong selective localization in BS mitochondria (~10-fold higher than in M mitochondria) in P. oleracea. Similarly, in all three species, P. cryptopetala, P. hirsutissima and P. mucronata, there is clear selective labeling for GDC in BS mitochondria (Fig. 5). Supplementary Fig. S1 shows electron microscopy images of the immunolabeling for anti-GDC antibody, where it is selectively localized in BS mitochondria in P. oleracea and the C 3 -C 4 intermediates, while in the C 3 species Sesuvium portulacastrum there is similar density of labeling in mitochondria in M and BS cells (Fig. 5).

Western blot analysis
Immunoblots are shown for photosynthetic enzymes of the C 4 cycle, PEPC, PPDK, NAD-ME and NADP-ME, from total proteins extracted from leaves of Portulaca species (Fig. 6). In the three species in the Cryptopetala clade, (P. cryptopetala, P. hirsutissima and P. mucronata) the labeling for C 4 cycle enzymes PEPC and PPDK was much lower than in the other species all of which have Kranz anatomy. Analyses of the relative band densities (measured by Image J) showed the density of labeling in the intermediates for PEPC was 7-15% and for PPDK was 14-20% compared to the mean value for the seven C 4 species. The three species from the Cryptopetala clade had lower labeling for NAD-ME-type enzyme compared to the NAD-ME species P. elatior and P. oleracea, and no detectable labeling for NADP-ME (Fig. 6). Among the other species, five are in the Pilosa clade (P. biloba, P. cf. gilliesii, P. elatior, P. halimoides, P. smallii and P. suffrutescens), while the other, P. oleracea, is in the Oleracea clade. With respect to C 4 decarboxylases, all species from the Pilosa clade have high labeling for NADP-ME with the exception of For two species, P. amilis and P. biloba, vein density is shown for both sides of the leaf, adaxial and abaxial (ada and aba on graph). For P. oleracea, two accessions were studied: *, seeds from Pullman, WA, USA; and **, seeds from USDA PI 0121921. n≥10. Different letters indicate significant differences between species and leaf sides, P≤0.05. P. elatior; it has high labeling for NAD-ME and no detectable labeling for NADP-ME, which is similar to P. oleracea.

Gas exchange analysis
Gas exchange analyses were made to compare photosynthetic features of species in the Cryptopetala clade with a representative C 4 species, P. oleracea. The response of photosynthesis to varying intercellular levels of CO 2 were measured under 1000 PPFD, 25 ºC and atmospheric O 2 (21%). With increasing levels of CO 2 , there was a rapid increase in photosynthesis in P. oleracea with near saturation at ~200 μbar CO 2 , whereas P. cryptopetala, P. hirsutissima and P. mucronata show strong a continual increase in photosynthesis up to 800-1000 µbar CO 2 with very similar rates under high CO 2 (Fig. 7A). When photosynthesis was measured under near current atmospheric levels of CO 2 , 400 μbar (with resulting C i values ~200-250 µbar), the rate of photosynthesis was higher in P. oleracea than in the other three species. The CO 2 compensation points (Γ) determined from extrapolation of the initial slopes of the response curve, was 2.6 μbar for P. oleracea, while values in the Cryptopetala species P. cryptopetala and P. hirsutissima were ~ 30 μbar, and in P. mucronata Γ was 20 μbar (Fig. 7B). In light response curves, at 400 µbar CO 2 and 25 ºC, the response of photosynthesis was similar among the species (Fig. 7C). In the Cryptopetala species photosynthesis reaches saturated rates at ~1100 PPFD while there was some marginal increase in rates in P. oleracea up to 2000 PPFD. The maximum rates at 2000 PPFD in P. hirsutissima and P. mucronata were similar, and significantly (P<0.05) lower than in P. cryptopetala and P. oleracea (Fig. 7C).

Evolution of C 4 biochemical types
The sampling for the phylogenetic analysis included 16 Portulaca species with known C 4 biochemical types (see Voznesenskaya et al., 2010;Ocampo et al., 2013;this study) and three representatives of close-related taxa as outgroup (Table 2). Relationships within Portulaca follow those obtained by Ocampo and Columbus (2012), where the genus and the relations among major clades are well supported (Fig. 8). The analysis shows the NADP-ME biochemical variant is the ancestral C 4 type for Portulaca. It predicts that there were two independent switches to the NAD-ME type, one in the ancestor of the Oleracea clade and one in P. elatior (earliest divergent lineage of the Pilosa clade). The C 3 -C 4 condition originated only once in the ancestor of the Cryptopetala clade.

Discussion
In this study on species which belong to the AL clade of Portulaca, the focus was on clades Pilosa and Cryptopetala. Portulaca biloba, P. cf. gilliesii, P. elatior, P. halimoides, P. smallii and P. suffrutescens are from the Pilosa clade, and they all have C 4 -type carbon isotope ratios (this study; Ocampo et al., 2013). The discovery that this clade has not only NADP-ME species, but also an NAD-ME type C 4 (P. elatior, which is widely distributed from the Caribbean to Venezuela and Brazil) provides insight into diversity in evolution of forms of C 4 in the genus. The Cryptopetala clade, which is composed of P. cryptopetala, P. hirsutissima and P. mucronata, is the only lineage that has species with C 3 -type carbon isotope ratios (Table 1, Ocampo et al., 2013); in this study all three species are shown to be C 3 -C 4 intermediates.

Evolutionary trends in anatomy Leaf anatomy
Among the three intermediate Portulaca species in the Cryptopetala clade, P. cryptopetala and P. mucronata are sister species, while P. hirsutissima is an early-divergent (see Fig. 8). The former two species have rather succulent, flattened leaves, while P. hirsutissima has terete leaves with a thicker central part. This is consistent with differences in structure: P. cryptopetala and P. mucronata have broad leaves with VB   6. Western blots for four C 4 pathway enzymes from total proteins extracted from leaves of ten Portulaca species. Blots were probed with antibodies raised against PEPC, PPDK, NAD-ME and NADP-ME. Numbers at the right indicate molecular mass in kilodaltons. The originals were modified to group species according to the presentation (vertical lines); there were no selective changes in positions or densities of bands on the membrane. The reason for labeling of two bands in some species with the antibody for NAD-ME (prepared against A. hypochondriacus α-NAD-ME) is not known. It may represent the expression of two isoforms of the α-NAD-ME, or the antibody may have low reactivity with the smaller subunit β-NAD-ME in some species (see Murata et al., 1989;Tronconi et al., 2008;Maier et al., 2011). distributed in one plane, while P. hirsutissima differs in having semi-terete leaves with all lateral VBs distributed under the adaxial epidermis and the main vein underneath, surrounded by WS. In P. cryptopetala and P. mucronata the distribution of VB is similar to the C 4 species in the Umbraticola clade having Atriplicoid Kranz-anatomy, while in P. hirsutissima the leaf anatomy resembles the Portulacelloid-type Kranzanatomy found in the Australian clade C 4 species (see Fig. 8 for anatomical types).
Most of the C 3 -C 4 intermediate species found in different dicot lineages have a general leaf structural pattern that resembles closely related C 4 species. Most of them have flattened dorsoventral or isopalisade leaves with enlarged BS cells. This type of anatomy in flattened leaves is characteristic for C 3 -C 4 intermediates in genera Flaveria (Holaday et al., 1984b) and Parthenium (Moore et al., 1987) in family Asteraceae, in Euphorbia in family Euphorbiaceae , Heliotropium in family Boraginaceae (Muhaidat et al., 2011), and Alternantera in family Amaranthaceae (Rajendrudu et al., 1986). They have Kranz-like Atriplicoidtype anatomy as the most advanced stage of evolution from intermediates towards C 4 . This includes several species with C 4 -like photosynthesis (Flaveria brownii, Cheng et al., 1988) and two intermediate Heliotropium species with Kranz-like (C) Rates of CO 2 fixation in response to varying light intensities at 21% O 2 , 25 ºC and 400 µbar CO 2 . For the CO 2 response curves (panel A), the results represent the average of 3-5 replications (for each replication, mean values were taken from CO 2 response curves measured from high to low, followed by low to high levels of CO 2 ). In panel C, the results represent the average of 2-4 replications from measurements made on different leaves (averages were taken from results obtained with changes from high to low light intensity).
anatomy (Muhaidat et al., 2011). Genera Moricandia (Apel and Ohle, 1979;Holaday and Chollet, 1984a) and Diplotaxis (Ueno et al., 2003) also have intermediate species with similar type of anatomy in flattened leaves; but there are no known C 4 relatives in family Brassicaceae. In family Cleomaceae, the С 3 -С 4 intermediate Cleome paradoxa has flattened leaves with isopalisade anatomy with all VB distributed in one plane, Atriplicoid-like (Voznesenskaya et al., 2007). It was suggested to be closer phylogenetically to C 4 C. angustifolia which has Glossocardioid Kranztype leaf anatomy; nevertheless, the phylogenetic position of this species is not far from C 4 Cleome (Gynandropsis) gynandra with Atriplicoid-type leaf anatomy (Feodorova et al., 2010).
In considering intermediates and forms of Kranz in families Chenopodiaceae and Portulacaceae, there is large  diversity in C 4 from Atriplicoid type to forms of Kranz with VB distributed around WS tissue in succulent leaves. In Chenopodiaceae, the type of leaf anatomy in the C 3 -C 4 intermediate Sedobassia sedoides resembles the Kranz Kochioid type in C 4 Bassia in the positioning of veins with respect to WS tissue and the positioning of BS cells with respect to VB where they are distributed only on the outer side of the lateral VB (Freitag and Kadereit, 2014). Also, Kranzlike Sympegmoid and Kranz-like Salsoloid are C 3 -C 4 intermediate types of leaf anatomy found in Salsola s.l. species, which resemble Kranz Salsoloid-type in peripheral distribution of M and positioning of minor veins in a circular pattern around WS tissue in succulent leaves Schüßler et al., in press). In the present study, the leaf anatomy in C 3 -C 4 intermediate Portulaca species follow this trend in showing Atriplicoid-like and Portulacelloid-like types of anatomy with related C 4 species having these forms of Kranz anatomy.

Cotyledon anatomy
The three Cryptopetala species, which are shown to have C 3 -C 4 intermediate type of anatomy in leaves, all have a similar dorsoventral type of anatomy in cotyledons with centripetally arranged organelles in BS cells. In one species, P. hirsutissima, the VB are distributed close to the adaxial epidermis, with greater number of WS spongy parenchyma layers towards the abaxial side, compared to other species with VB distributed in the central lateral plane of the lamina. In closely related families Cleomaceae and Brassicaceae, the C 3 -C 4 intermediates Cleome paradoxa and Moricandia arvensis also have cotyledons with C 3 -C 4 intermediate anatomy (Rylott et al., 1998;Koteyeva et al., 2011). In other intermediate species (e.g. Salsola and Sedobassia sedoides, family Chenopodiaceae), the cotyledons have C 3 -type anatomy (Voznesenskaya, Koteyeva, unpublished data). In species where the cotyledons also have intermediate phenotype, this could be advantageous in climates where seedlings are more prone to losses due to photorespiration (warmer climates). In general, intermediate species having C 3 -versus C 3 -C 4 -type photosynthesis in cotyledons may be an example of heterobathmy (Takhtajan, 1959), a phenomenon which results in unequal levels of specialization of different parts of one organism or taxon, achieved during the process of biological evolution.
In five species analyzed in the Pilosa clade, which have Pilosoid Kranz-type in leaves, the cotyledons all have Atriplicoid-like Kranz anatomy with some differences in shape and number of WS spongy layers on the abaxial side. Thus cotyledons in P. elatior, P. biloba, P. halimoides and P. suffrutescens have classical Atriplicoid-type anatomy with all Kranz-units (VB surrounded by two layers of chlorenchyma, characteristic for C 4 species, see Peter and Katinas, 2003) distributed in a central plane and one or two layers of spongy parenchyma on the adaxial side. However, in cotyledons of P. smallii, all Kranz-units are distributed under the adaxial epidermis with several layers of WS spongy parenchyma underneath. It is easy to imagine that this type of structure could represent a transitional step between flat-leaved cotyledons with Atriplicoid anatomy to the Pilosoid anatomy characteristic of that in leaves of P. smallii. In a previous study, P. amilis and P. grandiflora in the Pilosa clade, P. umbraticola in the Umbraticola clade and P. oleracea in the Oleracea clade were shown to have Atriplicoidlike anatomy in cotyledons; however, P. cf. bicolor in the OL clade, has, as in leaves, Portulacelloid type anatomy in the cotyledons . In general, all Portulaca species examined having Kranz-type anatomy in leaves, also have Kranz anatomy in the cotyledons. As noted by Voznesenskaya et al. (2010), in C 4 dicots that have been analyzed in other families, the anatomy and type of photosynthesis in cotyledons has often been found to be similar to leaves. However, there is diversity with some Kochia species having Kochioid-type anatomy in leaves and Atriplicoid-type anatomy in cotyledons and with some Salsola species having Salsoloid-type anatomy in leaves with Atriplicoid-type anatomy in cotyledons (Pyankov et al., 1999a(Pyankov et al., , 2000; and some chenopods having Kranz anatomy and C 4 photosynthesis in leaves, with cotyledons having C 3 -type anatomy and biochemistry (Pyankov et al., 1999b(Pyankov et al., , 2000Voznesenskaya et al., 1999;Lauterbach et al., 2016). These variations are of interest in considering genetic factors that control differences in forms of anatomy and biochemistry that develop in C 4 dicot leaves versus cotyledons.

Trends in venation density in leaves
Adaptation of plants to arid environments is often accompanied by an increase in vein density; at the same time it was shown that in most cases vein density decreased with increasing succulence, which is the second structural and/or functional way of plant adaptation to aridity (Ogburn and Edwards, 2013). Both forms of adaptation likely occurred in evolution of types of Kranz anatomy in C 4 species of families Chenopodiaceae and Portulacaceae. Together with the tendency to decrease leaf surface area to volume ratio, most succulents have subterete to terete leaves (more or less roundish in cross section). This leaf shape is often accompanied by development of a 3D type of venation. This pattern helps to maintain the most balanced hydraulic pathway to support photosynthesis by a closer position of veins to chlorenchyma, with a decrease in the path from the vein to leaf surface in thick succulent leaves Edwards, 2009, 2013). Increased vein density is considered to be one of the crucial factors in the evolution of C 4 photosynthesis in many dicot and grass lineages (Sage et al., 2012). However, studies of venation in representative Salsola species with different types of photosynthesis showed there is no significant difference between C 3 , C 3 -C 4 intermediates and C 4 species in vein density, while there is a clear increase in volume of WS tissue .
All Portulaca species have succulent leaves that differ in thickness, shape and type of anatomy. In C 4 Portulaca lineages there are differences in vein density along with variation in the pattern of vein distribution in different anatomical types, with the highest density in species having Atriplicoid anatomy.
In intermediate Portulaca species, the vein density is about 2-2.3 times lower than in some other intermediate species including Cleome paradoxa (Marshall et al., 2007), several Flaveria intermediates (McKown and Dengler, 2007) and Salsola divaricata . This may be explained by the Portulaca intermediates having much larger cells in succulent leaves.
In this study, most of the C 4 Portulaca species have higher vein density compared to three intermediate species. The species with Atriplicoid type of anatomy from both clades Umbraticola and Oleracea, have the highest vein density among all studied Portulaca types (see Fig. 3). Thus in the С 4 Oleracea species, which have modified zig-zag Atriplicoid pattern of venation, the vein densities are ~1.8-fold higher than in the Cryptopetala intermediates (Fig. 3) and are very close to that published for flat-leaved C 4 Cleome gynandra (Marshall et al., 2007); however, the densities in the Oleracea species is ~1.5 times lower than in C 4 Flaveria (McKown and Dengler, 2007). In C 4 Pilosa species, which have peripheral vein distribution around terete or slightly flattened leaves with WS tissue in the middle, the vein densities were lowest among the C 4 types analyzed. Compared to the intermediates, the vein densities on the adaxial side of the Pilosa species were ~1.3-fold higher than in the Cryptopetala intermediates, while the vein densities on the abaxial side of Pilosa species was similar to that in Cryptopetala intermediates. The Pilosa species may have a lower vein density than the Oleracea species, by having a higher volume of WS tissue and decrease in chlorenchyma tissue.
In contrast to the absence of clear correlation between vein density and the type of photosynthesis in succulent Salsola species , in most C 4 Portulaca lineages the vein density is higher than in the intermediates. However, among C 4 lineages the vein density depends on the leaf anatomical type, with different patterns of minor veins distribution and/or forms of leaf succulence. In addition, in family Anacampserotaceae, which is closely related to the Portulacaceae, there is a C 3 species Anacampseros (Grahamia) coahuilensis (with weak Crasulacean acid metabolism, Guralnik et al., 2008) that has the same degree of succulence as P. cryptopetala, while its vein density is similar to that of C 4 species in the Umbraticola and Pilosa (adaxial side) clades (5.0 ± 0.1 mm mm −2 ; Voznesenskaya and Koteyeva, unpublished data). This further supports the understanding that vein density in succulent lineages is more related to species-specific adaptations than to a progression change from C 3 to C 4 . Further analyses of leaf structure of different photosynthetic phenotypes in Portulaca, including vein density, volume of WS tissue and chlorenchyma tissue is needed to consider anatomical adaptations to arid environments.

Evolutionary trends in types of photosynthesis
In the study of species in clade Cryptopetala, a variety of P. cryptopetala from Uruguay was analyzed (Table 1) in comparison to a variety from Argentina that was previously shown to be a C 3 -C 4 intermediate species . While there are differences in plant morphology (the size of the plant, thickness and size of the stems and leaves, the size of flowers), the variety used in the current study also has a similar dorsoventral leaf anatomy. Analysis of the other species in the Cryptopetala clade, P. hirsutissima and P. mucronata, showed they also have C 3 -C 4 anatomy with dorsoventral distribution of chlorenchyma. Microscopy studies show that BS cells in all three species have a substantial number of organelles with chloroplasts and enlarged mitochondria located mainly in a centripetal position. Measurements by gas exchange showed the values of Г in these three Cryptopetala species (20-32 μbar CO 2 ) indicate functionally they are intermediates compared to the representative C 4 species P. oleracea, which has a C 4 -type Γ value (2.6 μbar). Also, P. cryptopetala (variety from Argentina) has a Γ, which is intermediate to C 3 (outgroup species Sesuvium portulacastrum) and C 4 Portulaca . C 4 species typically have Г values of 0-5 μbar, C 3 species ~60 μbar, and C 3 -C 4 intermediates of 9-35 µbar depending on the species, e.g. Ku et al. (1991), Vogan et al. (2007).
The model for evolution of C 4 plants developed from studies of species having reduced photorespiration consists of a stepwise progression of structural, biochemical and functional changes from C 3 through stages of development of intermediates having a C 2 cycle, followed by intermediates having increased acquisition of the C 4 cycle, to fully functional C 4 photosynthesis (Sage et al., 2012;Voznesenskaya et al., 2013;Schulze et al., 2016). In the C 2 cycle photorespiratory glycine produced in M cells is shuttled to BS cells for decarboxylation by GDC where photorespired CO 2 is concentrated, enhancing its capture by BS Rubisco. In all three Cryptopetala species, GDC is selectively localized in mitochondria of BS cells, which is a distinguishing biochemical feature of intermediate species, while in C 3 species the labeling is distributed about equally between M and BS mitochondria (Rawsthorne et al., 1988). Analyses by western blots of C 4 pathway enzymes showed that PEPC, PPDK and NAD-ME were detectable in all three intermediate species in the Cryptopetala clade, but in much lower levels compared to the C 4 species. Also, Voznesenskaya et al. (2010) found the levels of these C 4 enzymes in P. cryptopetala (variety from Argentina) were low, similar to the C 3 species S. portulacastrum. The results suggest the capacity for C 4 cycle activity in intermediates in this clade is low, which is consistent with leaf biomass having C 3 -like carbon isotope values. Thus, when CO 2 is limiting they may have an increase in their efficiency of photosynthesis primarily by refixing photorespired CO 2 in BS cells by the C 2 cycle, with restricted contribution by a C 4 cycle (Edwards and Ku, 1987;Ku et al., 1991;Rawsthorne and Bauwe, 1998). Whether there is any C 4 cycle activity in these species could be more directly analysed by the method of Alonso-Cantabrana and von Caemmerer (2016) via online measurements of photosynthesis and carbon isotope discrimination. The lack of identification of C 3 and other species that could represent the proposed progression in C 4 evolution in the genus Portulaca could suggest they have become extinct or have not yet been discovered.
All the Portulaca species in the study from the Pilosa clade were shown to have Kranz-type Pilosoid anatomy. One of them, P. biloba, has flattened subterete leaves while P. cf. gilliesii, P. elatior, P. halimoides and P. smallii have cylindrical terete leaves. In these species, the lateral VB with Kranz anatomy are distributed around the leaf periphery; while the main vein, which lacks a chlorenchyma sheath, is enclosed by WS tissue located in the central part of the leaf. That type of structure was classified as having multiple simple Kranz units (with many separate VB surrounded by two layers of Kranz tissue) according to classification by Peter and Katinas (2003). A similar leaf anatomy was shown for other representatives of this clade Ocampo et al., 2013). Some species in this clade were previously shown to have NADP-ME type of C 4 biochemistry ; in the present study additional species were shown to be NADP-ME type with one exception. Portulaca elatior, with Pilosoid leaf anatomy, was shown to have NAD-ME type of biochemistry, and it represents the earliest-divergent lineage of the Pilosa clade. This was surprising, since NAD-ME subtype was previously associated only with the Oleracea clade while all other C 4 Portulaca were considered NADP-ME subtype . The results suggest NAD-ME type of C 4 biochemistry evolved twice within Portulaca, once in the Oleracea clade and once in P. elatior (Pilosa clade). It suggests evolution of different forms of C 4 is more complex than previously thought. This discovery is consistent with the hypothesis that all photosynthetic modifications within Portulaca, including C 3 -C 4 intermediates, were probably derived from C 4 NADP-ME type based on phylogenetic analyses ( Fig. 8 and Ocampo et al., 2013). Previously it was shown the dominant form of photosynthesis in Portulaca is C 4 , which was possibly lost in species in the Cryptopetala clade. Similar cases, where intermediates are nested within C 4 species, were shown for some clades in Aizoaceae (Bohley et al., 2015) and Salsoleae (Schüßler et al., in press); which might be other cases where reversions from C 4 to intermediates or C 3 occurred. This is in contrast to studies in some families, which support evolution from C 3 -C 4 intermediates to C 4 photosynthesis (see Leaf anatomy in evolutionary trends section, this paper, and Sage et al., 2012Sage et al., , 2014. Other studies suggest that a C 3 -C 4 ancestor may have given rise to different C 4 biochemical forms independently in separate lineages (e.g. Christin et al., 2011). This may have been in response to specific environmental conditions, although the adaptive advantages of the different C 4 variations are not well known. Careful largescale analyses of the distribution of C 3 -C 4 species and closely related C 3 and C 4 lineages showed it is impossible to define a universal ecological C 3 -C 4 niche, nevertheless between different ecological factors temperature is one of the most important driving forces (Lundgren and Christin, in press). Likewise, the diversity within Portulaca could be facilitated by development of different photosynthesis strategies in separate lineages from unknown C 3 -C 4 intermediate ancestors; although there is no evidence to support this. Gas exchange analyses indicate the C 3 -C 4 intermediates (which have the CO 2 concentrating C 2 cycle) would be less capable of maintaining their photosynthesis under CO 2 limited conditions than C 4 species, which have the CO 2 concentrating C 4 cycle. Both intermediates and different Kranz forms of C 4 Portulaca have been found to grow together in some locations (Ocampo et al., 2013). Further studies may help to clarify photosynthetic diversification in the different lineages of Portulaca, including the potential adaptive advantage of each phenotype based on habitat and seasonal growth cycles.