Bundle sheath chloroplast volume can house sufficient Rubisco to avoid limiting C4 photosynthesis during chilling

The volume of bundle sheath chloroplasts available for Rubisco investment in the leaves of four C4 grasses could potentially support much greater photosynthetic activity than is typically observed, even at chilling temperature.


Introduction
C 4 photosynthesis involves a biochemical CO 2 concentrating mechanism. In mesophyll cells, the enzyme phosphoenolpyruvate carboxylase assimilates CO 2 into oxaloacetate, which is then metabolized into further C 4 compounds that are transferred to, and decarboxylated in, bundle sheath (BS) cells to raise [CO 2 ] around the enzyme Rubisco (von . Rubisco then fixes this CO 2 via the Calvin-Benson cycle in the BS. In C 4 plants, Rubisco is therefore predominantly localized to the chloroplasts of BS cells, where the increased [CO 2 ] greatly improves photosynthetic efficiency because it effectively eliminates photorespiration, the energetically costly process initiated when O 2 instead of CO is fixed by Rubisco (Hatch, 1987). The BS cells of C 4 leaves are arranged radially around veins and isolated from internal leaf air spaces by surrounding mesophyll cells (Dengler and Nelson, 1999). Relative to the leaves of C 3 plants, C 4 leaves achieve greater overall BS tissue area via a combination of higher vein density, enlarged BS cells, and more numerous BS cells (Christin et al., 2013;Lundgren et al., 2014).
The enhanced efficiency of C 4 photosynthesis under warm conditions is evident in the high productivity of the Andropogoneae grass crops maize (Zea mays L.), sorghum (Sorghum bicolor (Lu.) Moench), and sugarcane (Saccharum officinarum L). However, photosynthesis in the majority of C 4 grasses is characterized by poor chilling tolerance, limiting them to warmer environments (Long, 1983;Sage, 2002;Long and Spence, 2013). Improving chilling tolerance could therefore expand the growing region and lengthen the growth seasons of C 4 crops (Głowacka et al., 2016). Such tolerance of low temperatures has evolved many times in wild C 4 grasses, enabling them to shift their niches into cooler alpine or temperate environments (Watcharamongkol et al., 2018).
The mechanisms conferring chilling tolerance to C 4 grasses have been especially well studied in the grass Miscanthus × giganteus Greef et Deu. because of its importance for cellulosic biomass production (Heaton et al., 2010). For example, Z. mays leaves developing at 14 °C have less than 10% of the photosynthetic capacity of Z. mays leaves developing at 25 °C, while leaves of M. × giganteus are unaffected by this temperature difference (Long and Spence, 2013). Another study found that M. × giganteus achieved 59% greater biomass than Z. mays by producing photosynthetically competent leaves earlier in the year and maintaining them several weeks after Z. mays senesced in side-by-side trials in the US Corn Belt (Dohleman and Long, 2009). This growth advantage may be even more pronounced in the near future, as anthropogenic climate change may cause more frequent and intense springtime chilling events across the US Corn Belt (Kim et al., 2017). Understanding and harnessing the potential of chilling-tolerant C 4 photosynthesis could provide crucial improvements to the yield and robustness of key C 4 crops Zhu et al., 2010;Yin and Struik, 2017).
Chilling tolerance in C 4 grasses may be linked to leaf anatomy. Because C 4 leaves restrict Rubisco to BS cells, the space potentially available to house this enzyme is roughly halved relative to C 3 leaves, which can accommodate the enzyme in all photosynthetic cells (Pittermann and Sage, 2000). Under moderate temperatures, flux analysis points to Rubisco as a major control point on the rate of CO 2 assimilation in C 4 leaves, as it is in C 3 leaves (Furbank et al., 1997). Since catalytic rate declines with temperature, Rubisco becomes an even greater limitation under chilling, unless its amount is increased (Sage et al., 2011;Long and Spence, 2013).
It has been proposed that BS chloroplast volume would limit acclimatory increases in Rubisco in C 4 plants at chilling temperatures (<15 °C), so disadvantaging them relative to their C 3 counterparts (Pittermann and Sage, 2000;Kubien et al., 2003;Kubien and Sage, 2004;Sage and McKown, 2006;Sage et al., 2011). This hypothesis is supported by the observation that leaves of chilling-tolerant C 3 plants often increase Rubisco content during acclimation, whereas this is rarely seen in C 4 leaves (Sage and McKown, 2006;Long and Spence, 2013). Net photosynthetic CO 2 uptake (A sat ) in C 4 leaves correlates with Rubisco content (Pearcy, 1977) and activity (Pittermann and Sage, 2000;Kubien and Sage, 2004;Friesen and Sage, 2016) at low (<15 °C), but not high (>25 °C), temperatures. Rubisco's flux control coefficient over photosynthetic CO 2 assimilation reaches 0.99 (i.e. near-total control) at 6 °C in Flaveria bidentis L. Kuntze (Kubien et al., 2003). These observations raise important questions: does Rubisco limit photosynthesis in all C 4 plants at low temperatures, and is this limitation specifically imposed by the restricted space available in the BS to house the enzyme?
Under chilling conditions, the chilling-tolerant M. × giganteus maintains photosynthetic capacity and, unusually, maintains or slightly increases leaf Rubisco content per unit leaf area, while showing large increases in pyruvate P i dikinase (PPDK) expression (Naidu et al., 2003;Wang et al., 2008b;Long and Spence, 2013). Accessions of M. sacchariflorus, one of the parent species of M. × giganteus, achieved some of the highest light-saturated rates of leaf CO 2 uptake (A sat >16 µmol m −2 s −1 ) recorded for any plant grown and measured at 15 °C (Głowacka et al., 2015), showing that this species must accumulate sufficient Rubisco to support such high photosynthetic rates. Of course, there is the possibility that these Miscanthus genotypes are exceptional in providing unusually large bundle sheath chloroplast volumes.
Coinciding with the acclimation of C 4 cycle enzymes in Miscanthus, the up-regulation of key photoprotective mechanisms reduces damage to photosystem II (Farage et al., 2006). This suggests that decreased photosynthetic rates in most C 4 grasses at low temperature have multiple causes rather than arising from one inherent limitation. Indeed, comparative transcriptomics has suggested that the chilling tolerance of photosynthesis in M. × giganteus corresponds to the up-regulation of genes encoding several photosynthetic proteins (Spence et al., 2014). Miscanthus × giganteus maintains the linear relationship between operating photochemical efficiency of photosystem II and the quantum efficiency of CO 2 assimilation during chilling, suggesting that the balance of C 3 and C 4 cycles is not compromised (Naidu and Long, 2004). In total, these findings suggest that Rubisco is not the sole limitation to C 4 photosynthesis at chilling temperatures, and that any volume limitation imposed by restriction of the enzyme to the bundle sheath can be overcome, at least in the case of M. × giganteus and related species (Long and Spence, 2013).
Because most Rubisco in C 4 leaves is confined to BS chloroplasts, a measure of the total volume of chloroplasts in the BS is required to determine if there is enough space available to increase Rubisco content in C 4 leaves. However, most attempts at chloroplast quantification have not documented 3D measurements, but rather chloroplast counts and 2D planar area (Pyke and Leech, 1987;Brown and Hattersley, 1989;Stata et al., 2014Stata et al., , 2016. With confocal laser scanning microscopy, it is possible to measure chloroplast volume directly from an optically produced 3D image (Park et al., 2009;Coate et al., 2012). Chloroplast measurements have previously been made on fixed, dehydrated samples in accordance with TEM imaging procedures (Sage and Williams, 1995). While this method is adequate for relative comparisons of chloroplast size and number between plant taxonomic clades or functional types (Stata et al., 2016;Stata et al., 2014), it may distort chloroplast shape and prevent accurate estimation of absolute chloroplast volume in vivo. Cryo-sectioning and analysis of fresh plant material may prevent this type of distortion.
To test the hypothesis that BS chloroplast volume restricts the capacity for Rubisco to the extent that it would limit photosynthesis in C 4 grasses, chloroplast volume and associated leaf anatomical characteristics were measured, and used to calculate the amount and activity of Rubisco that could be supported on a leaf area basis. The focus of the study was on grasses of the Andropogoneae: since M. × giganteus appears to escape the low temperature limitation observed in most C 4 grasses, its BS chloroplast volumes were compared to two chilling-intolerant crop species of the same tribe (Z. mays and S. officinarum). The unrelated, non-Andropogoneae, non-crop and chilling-intolerant C 4 grass (Alloteropsis semialata J. Presl) was also included in the study .

Plant material
Measurements were taken on Z. mays cv. FR1064, S. officinarum hybrid complex cultivar cv. CP88-1762, a C 4 lineage of A. semialata originating from South Africa , and the 'Illinois' clone of M. × giganteus. Miscanthus × giganteus was grown in the field and the other species were grown in a controlled-environment greenhouse, maintained between 25 and 30 °C with high pressure sodium lamps ensuring an average photon flux of 450 μmol m −2 s −1 over a 12 h photoperiod.
Miscanthus × giganteus was grown at the University of Illinois Agricultural Research Station farm near Champaign, IL, USA (40°02′N, 88°14′W, 228 m above sea level). Soils at this site are deep Drummer/ Flanagan series (a fine silty, mixed, mesic Typic Endoaquoll) with high organic matter typical of the central Illinois Corn Belt. Fertilizer was not applied. As in previous studies, the youngest fully expanded leaf of M. × giganteus plants, as judged by ligule emergence, was sampled in July (Dohleman et al., 2012;Arundale et al., 2014a,b;Pignon et al., 2017).
Alloteropsis semialata and Z. mays seeds were germinated on moist filter paper in a growth chamber maintained at 25 °C with an average photon flux of 200 μmol m −2 s −1 . They were then transferred to pots of soil-less cultivation medium (LC1 Sunshine Mix, Sun Gro Horticulture, Agawam, MA, USA), with additional coarse sand and perlite mixed into pots for A. semialata. Single stem segments of S. officinarum were planted directly into pots of a second soil-less cultivation medium (Metromix 900: SunGro Horticulture). All pots were watered daily to field capacity. Zea mays was initially fertilized with granulated fertilizer (Osmocote Plus 15/9/12, The Scotts Company LLC, Marysville, OH, USA) followed by general nutrient solution (Peter's Excel 15-5-15, Everris NA Inc., Dublin, OH, USA) and iron chelate supplement (Sprint 330, BASF Corp. NC, USA) added to the watering regime once every week. Alloteropsis semialata and S. officinarum were fertilized with granulated fertilizer (Osmocote Classic 13/13/13, The Scotts Company LLC), and A. semialata supplemented with iron chelate (Sprint 330, BASF Corp.). Plants were grown until at least the fifth leaf was fully expanded, as judged by ligule emergence, and the youngest fully expanded leaf was sampled.

Sample preparation and measurement
On sampling, leaves were immediately immersed in a glycol and resin based cryostat embedding medium (Tissue-Tek O.C.T. Compound, Sakura Finetek, Torrance, CA, USA), which provides solid sectioning support on dry ice. Transverse sections of 40 µm were cut (Leica CM3050 S, Leica Biosystems, Wetzlar, Germany) and mounted on glass slides. Slides were then immersed for 15 min in a cell membrane and wall dye solution (FM 1-43FX, Thermo Fisher Scientific, Waltham, MA, USA), and diluted to 3.6 mM in dimethylsulfoxide (Thermo Fisher Scientific) and water, in order to image cell walls. Samples were imaged with a confocal laserscanning microscope (LSM 700, Carl Zeiss AG, Oberkochen, Germany). Images were acquired through a ×63 oil-immersion objective (×63 Plan-Apochromat, Carl Zeiss AG) for M. × giganteus. It was determined that reduced magnification could be used to widen the field of view while still providing accurate estimates of chloroplast volume. Therefore a ×40 oil-immersion objective (×40 Plan-Apochromat, Carl Zeiss AG) was used for Z. mays, S. officinarum, and A. semialata.
The fluorescence of dye-labelled cell walls was analysed by excitation at 555 nm, and emission was detected at a bandpass of 405-630 nm. Chlorophyll was excited at 633 nm, and its fluorescence emission was detected at a bandpass of 630-700 nm. Serial optical sections were obtained at 1-µm depth intervals, i.e. in the z-axis (Zen software, Carl Zeiss AG). Although sampling depth (8-15 µm in the z-axis) was insufficient to capture whole BS cells, each leaf section contained a random sampling of cells, which avoided the risk of biasing measurements due to non-homogeneous chloroplast distribution through the length of the cell.
Supplementary Video S1 at JXB online illustrates how the delineation of BS and mesophyll compartments, and the chloroplasts within them, was achieved within a 3D optical section. BS and mesophyll compartments were identified from the fluorescence of dye-labelled cell walls, using image segmentation software (IMARIS 7.0.0 software, BitPlane, Inc., Zürich, Switzerland). These segments were used to determine the volume of BS (vol BS ) and mesophyll (vol M ) per unit leaf area. The chlorophyll fluorescence signal within the BS and mesophyll was then used to determine total chloroplast volume per unit leaf area within each compartment (vol BS,cp and vol M,cp , respectively) and the percentage occupancy of each compartment by chloroplasts (% BS,cp and % M,cp , respectively). Although chlorophyll fluorescence from out-of-focus planes was typically visible in individual optical slices, the surface-finding algorithm of the image segmentation software was able to accurately delineate chloroplast volumes when processing the overall 3D optical section. As a result, individual 2D slices appear to overestimate chloroplast content of cells, but the 3D sections actually used to produce measurements do not; this can be seen by comparing Fig. 1C with Supplementary Video S1.
Leaf thickness was measured in a single location per image, across the mesophyll between two veins, and inter-veinal distance (IVD) was measured as the average distance between the centers of all the adjacent vascular bundles visible in each image.

Calculating photosynthetic capacity
An important goal of this study was to determine the theoretical maximum amount of Rubisco that C 4 BS chloroplasts could contain, in order to calculate the corresponding theoretical maximum level of Rubiscolimited photosynthetic CO 2 uptake (A max,cp ) that could be achieved by a given leaf. Calculated values for A max,cp could then be compared to achieved values for light-saturated photosynthetic CO 2 uptake (A sat ). Because A max,cp is a measure of theoretical, and not achieved, Rubiscolimited CO 2 uptake, factors such as leaf N content and incident light intensity could be ignored. Instead, A max,cp was determined from the volume of BS chloroplasts available for Rubisco investment (vol BS,cp ), the amount of Rubisco that could be contained within these chloroplasts, and the carboxylation activity of Rubisco. Although there is evidence of C 4 subspecies of A. semialata expressing Rubisco in chloroplasts outside of the BS (Ueno and Sentoku, 2006), here it was assumed in all species that only BS chloroplasts contained Rubisco.
vol BS,cp was determined experimentally in this study as described above. A Rubisco carboxylation rate per site at 25 °C (k cat ) of 3.3 mol CO 2 mol site −1 s −1 had been determined previously for both Z. mays and M. × giganteus (Wang et al., 2008a). This value was reduced by 15%, reflecting the Rubisco activation state at 25 °C of 85%, reported for M. × giganteus (Wang et al., 2008a). This gives an estimated carboxylation rate of 41.6 µmol CO 2 g −1 Rubisco s −1 at 25 °C. Rubisco content per unit chloroplast volume was assumed to be 2.2 × 10 5 g Rubisco m −3 chloroplast based on measurements for mesophyll chloroplasts of several genotypes of the hexaploid bread wheat Triticum aestivum L. (Pyke and Leech, 1987). Combining the carboxylation rate per gram Rubisco calculated with a molecular mass of 540 kDA, with the grams of Rubisco per unit volume of chloroplast, leads to a theoretical maximal photosynthetic rate of 9.2 mol CO 2 m −3 chloroplast s −1 at 25 °C. In the Results, this factor is combined with measured BS chloroplast volume (vol BS,cp ) to determine the potential photosynthetic rate that could theoretically be supported given the measured chloroplast volume (A max,cp ).

Statistical analysis
Replication was: Z. mays (n=7), S. officinarum (n=5), A. semialata (n=6), and M. × giganteus (n=6). Statistical analysis was performed on the following parameters: leaf thickness, IVD, vol BS , vol M , vol BS,cp , vol M,cp , % BS,cp , and % M,cp . The fixed effect of species on each parameter was tested by one-way ANOVA (PROC GLM, SAS v8.02; SAS Institute Inc., Cary, NC, USA), with homogeneity of variances tested by Levene and normality of residuals tested by Shapiro-Wilk (PROC UNIVARIATE, SAS v8.02) at a P=0.05 threshold. A Tukey test was performed alongside the ANOVA at a P=0.05 threshold in order to identify significant pairwise differences between species. When no significant differences were found, the test was repeated at a P=0.1 threshold to reduce the risk of a type II error given the relatively low replication for each species.

Results
The average volume of chloroplasts per unit leaf area ranged from 6 × 10 -6 to 10 × 10 -6 m 3 m −2 in the BS and from 10 × 10 -6 to 14 × 10 -6 m 3 m −2 in the mesophyll (Figs 1, 2, 3E, F). There was no evidence of greater BS chloroplast volume available per unit leaf area (vol BS,cp ) in the chilling-tolerant M. × giganteus compared with the chilling-sensitive species. On the contrary, M. × giganteus had the smallest BS chloroplast volume per unit leaf area, at ca. 40% less than the wild and chilling-sensitive A. semialata. Although there were no significant differences between species in vol BS , significantly greater occupancy of the BS by chloroplasts (% BS,cp ) resulted in greater vol BS,cp overall in A. semialata (Fig. 3C, E, G).
Across the four study-species, chloroplasts occupied 15-30% of the BS (% BS,cp ), and 8-14% of the mesophyll (% M,cp ) (Figs 1, 3G, H; Supplementary Video S1). Between species, % BS,cp and % M,cp were significantly highest and lowest, respectively, in A. semialata. Leaf thickness ranged from 100 to 250 µm, with veins spaced 100-140 µm apart on average (Figs 1, 3A, B). Alloteropsis semialata leaves at ca. 225 µm were nearly twice as thick as those of M. × giganteus at ca. 125 µm. The distance between veins (IVD) in the two crops (Z. mays and S. officinarum) was ca. 40% greater than in the two wild species (M. × giganteus and A. semialata) (Fig. 3B). Across the species, the volume of mesophyll per unit leaf area (vol M ) generally mirrored leaf thickness, though due to a thick epidermis the significantly greater leaf thickness of A. semialata did not result in a substantially greater vol M (Fig. 3D). BS volume per unit leaf area (vol BS ), however, was conserved across species at ca. 40 × 10 -6 m 3 m −2 (Fig. 3C).
When the maximal theoretical photosynthetic capacity of the leaf (A max,cp ) was estimated from vol BS,cp , values ranged from ca. 60 to 90 µmol m −2 s −1 at 25 °C. This was substantially greater than published values of light-saturated net photosynthetic CO 2 uptake (A sat ) for these species at this temperature (Fig. 4). However, at lower temperatures A sat was closer to A max,cp , with A sat being 20-90% of A max,cp at 5 °C.

Discussion
In all four of the C 4 grass species studied here, the volume of BS per unit leaf area available for Rubisco (vol BS ) was not a limitation for observed rates of photosynthesis, even at chilling temperatures. This conclusion is based on two key findings, derived from 3D confocal microscopy and analysis of leaf structure (Fig. 2). First, the chilling-tolerant M. × giganteus (Long and Spence, 2013) has a smaller BS chloroplast volume per unit leaf area (vol BS,cp ) than the chilling-sensitive C 4 grasses S. officinarum, A. semialata, and Z. mays (Fig. 3). Second, the theoretical maximum level of Rubisco-limited photosynthetic CO 2 uptake (A max,cp ) that could be achieved by each species was greater than realized levels of A sat , even at chilling temperatures (Fig. 4). This study focused on closely related C 4 grasses of the Andropogoneae clade, which contain the major C 4 crops as well as candidate bioenergy crops. Even A. semialata, which descends from a separate evolutionary lineage in the Paniceae, would not suffer from limitation of photosynthesis by vol BS during chilling.
Several leaf structural characteristics, including leaf thickness, IVD, vol M , % BS,cp , and % M,cp , varied significantly between species (Figs 1, 3A, B, D, G, H). Indeed, the vol BS,cp was actually greatest in the chilling-sensitive A. semialata and lowest in the chilling-tolerant M. × giganteus (Fig. 3E). This clearly demonstrates that vol BS,cp does not determine chilling tolerance in C 4 plants, and therefore that the volume of BS chloroplast available for leaf Rubisco investment is unlikely to meaningfully restrict C 4 photosynthesis at low temperatures.
Based on 2D leaf profiles, the percentage occupancy of the total mesophyll volume by chloroplasts varies significantly between photosynthetic types and taxonomic clades of diverse C 4 plants, with an average occupation of ca. 12.2% (Stata et al., 2014), which is similar to the 8-14% range seen here (Figs 1,  3H). In various species of the eudicot genus Flaveria that use the NADP-ME subtype of C 4 photosynthesis, chloroplasts occupied 12-18% of the total BS volume (Stata et al., 2016), which is somewhat lower than the range of 15-25% seen in our grasses (Figs 1, 3G); this may reflect differences due to taxonomy or specimen preparation. Alloteropsis semialata, which belongs to the Paniceae tribe, had the greatest volume of chloroplast in the BS (% BS,cp ) (Figs 1, 3G, H). This may reflect this species' need to house grana in their BS chloroplasts, while the other three studied grasses of the Andropogoneae tribe have little to no BS chloroplast grana (Ueno and Sentoku, 2006). Alloteropsis semialata's high BS chloroplast volume may also result from the very recent development of C 4 anatomy in this species, which might not have evolved the faster Rubisco kinetics of other, older C 4 lineages and could therefore require relatively more Rubisco in the BS to compensate (Lundgren et al., 2015;Dunning et al., 2017).
While chloroplasts across the entire mesophyll tissue are available for Rubisco investment in C 3 plants, there is clearly less space available in the BS tissue of C 4 leaves. However, in the mesophyll of C 3 species, CO 2 must diffuse from the air space to Rubisco in the chloroplast, and chloroplasts must be adjacent to the cell wall to maximize mesophyll conductance to CO 2 and facilitate access of Rubisco to CO 2 (Evans and Loreto, 2000;Flexas et al., 2008). In the BS of C 4 species, CO 2 results from decarboxylation of C 4 -dicarboxylates in the chloroplast or the cytosol, and the effective chloroplast volume is therefore not limited by the area of wall adjacent to air space. In effect, this can allow larger and more numerous chloroplasts, and may explain the greater proportion of the BS cell occupied by chloroplasts, relative to mesophyll (Figs 1, 3G, H).
The comparison of A max,cp to published values for A sat is directly dependent on terms used to calculate A max,cp : for instance, a 20% lower value for k cat will result in 20% lower A max,cp . At lower temperatures this could lead to A max,cp much closer to published values for A sat (Fig. 4A, B). However, the values used in this study were generally conservative. In a survey of Rubisco k cat in 14 grasses using different subtypes of C 4 photosynthesis (Ghannoum et al., 2005), all seven NADP-ME grasses and five of the seven NAD-ME grasses registered values greater than, and up to two times, the k cat value used here; i.e. 3.3 mol CO 2 mol site −1 s −1 (Wang et al., 2008a).
Another important term in the calculation of A max,cp is the Rubisco content per unit volume chloroplast. Here, we used a published value of 0.41 mol Rubisco m −3 chloroplast, derived from T. aestivum mesophyll chloroplasts (Pyke and Leech, 1987). This value is conservative, as it is at the lower end of the 0.4-0.5 mol Rubisco m −3 chloroplast range predicted from measurements in C 3 chloroplasts (Jensen and Bahr, 1977). Furthermore, C 4 plants generally produce larger chloroplasts than C 3 plants, particularly in the BS (Brown and Hattersley, 1989;Stata et al., 2014) and these chloroplasts likely contain more Rubisco per unit volume, since NADP-ME C 4 grasses, including Z. mays, S. officinarum, and M. × giganteus, typically show few or no stacked thylakoids in the BS. This arrangement leaves more space available for stroma, and therefore Rubisco, in comparison with bread wheat chloroplasts (Furbank, 2011;Voznesenskaya et al., 2006Voznesenskaya et al., , 2007Bellasio and Griffiths, 2014).
Despite the use of conservative terms to calculate A max,cp , this parameter was greater than published light-saturated photosynthetic rates (A sat ) for all four studied species (Fig. 4) (Long, 1983;Naidu et al., 2003;Naidu and Long, 2004;Osborne et al., 2008;Głowacka et al., 2014Głowacka et al., , 2015Głowacka et al., , 2016Spitz, 2015;Friesen and Sage, 2016). This was even true at low temperatures, where Rubisco has been predicted to be a strong limitation to C 4 photosynthesis (Pearcy, 1977;Pittermann and Sage, 2000;Kubien et al., 2003;Kubien and Sage, 2004). Therefore, we conclude that while the quantity of Rubisco may be limiting, this is not an inherent result of the smaller proportion of cells that can contain the enzyme in C 4 leaves with Kranz anatomy. Further supporting our conclusion that BS chloroplast space does not limit Rubisco comes from the fact that Rubisco content does increase in M. × giganteus on chilling (Long and Spence, 2013). Additional evidence comes from a recent transgenic up-regulation of Rubisco content by >30% above wild type in leaves of Z. mays (Salesse et al., 2018).
the C 4 grass clades (Watcharamongkol et al., 2018). Radiation into temperate climates has therefore involved solving the challenges of chilling and freezing temperatures faced by all tropical plants, regardless of photosynthetic type, as well as any additional restrictions added by the C 4 cycle and associated anatomy. The literature has already addressed these additional restrictions and the evolution of chilling-tolerant C 4 photosynthesis (Long, 1983(Long, , 1999Long and Spence, 2013).
Several C 4 grasses, including Muhlenbergia glomerata (Kubien and Sage, 2004), Spartina anglica (Long et al., 1975), and Cleistogenes squarrosa (Liu and Osborne, 2008) can achieve rates of CO 2 assimilation at chilling temperatures that equal or exceed rates achieved by temperate and even arctic/alpine C 3 grasses. Notably, the C 4 grass M. × giganteus appears exceptional in its ability to acclimate its photosynthetic apparatus to chilling temperatures. Comparison with the chilling-intolerant Z. mays suggests that chilling tolerance in M. × giganteus results from its ability to maintain and increase the expression of the enzymes PPDK and Rubisco, as well as increase leaf xanthophyll content, in particular zeaxanthin, to harmlessly dissipate excess absorbed light energy under chilling conditions and protect photosystem II from oxidative damage (reviewed in Long and Spence, 2013). Gene expression analyses suggest that these increases are part of a syndrome of acclimative changes that allow efficient C 4 photosynthesis under chilling conditions (Spence et al., 2014), and in turn the exceptional productivities achieved by M. × giganteus in temperate climates (Dohleman and Long, 2009). Therefore, while Rubisco content clearly co-limits photosynthesis in many C 4 species under chilling conditions, the findings here show that this does not directly result from restricting Rubisco to the BS in C 4 grasses.
In conclusion, while the volume of the cells that can hold Rubisco in C 4 grass leaves is lower than in their C 3 counterparts, measurements of BS chloroplast volume show that space per se does not present a physical, and in turn intrinsic, limitation on photosynthesis at chilling temperatures. Therefore, restriction of leaf Rubisco content by the volume of BS chloroplasts does not inherently limit the adaptation of C 4 grasses to cold environments.

Supplementary data
Video S1. Video of the full 3D image of leaf, bundle sheath cells, mesophyll cells, and chloroplasts seen in Fig. 2.