Abstract
The control of starch granule number and morphology in plastids is poorly understood. Here, we demonstrate that AtFZL, a protein involved in thylakoid membrane organisation, is required for correct starch granule morphology in Arabidopsis. Leaves of mutants lacking AtFZL had the same starch content as wild-type leaves, but their starch granules were smaller and had a distinct, uneven surface morphology. Most chloroplasts in the mutant were larger than those of the wild type.
However, the difference in chloroplast size could not explain the difference in granule size and shape in the Atfzl mutants, since other mutants with larger chloroplasts than the wild type (arc mutants) had granules that were similar in size and shape to wild-type granules. As reported previously, the Atfzl mutant had aberrant thylakoid organisation. We found that this phenomenon was particularly pronounced in regions surrounding starch granules. The location of the thylakoid-bound granule initiation protein MFP1 was unaffected in the Atfzl mutant. We propose that AtFZL affects starch granule size and shape by influencing thylakoid organisation at the periphery of starch granules. Our results are consistent with an important role for thylakoid architecture in determining granule morphology.
Introduction
Starch is the primary form of carbohydrate storage in plants. Plants use photoassimilates to synthesise semi-crystalline insoluble starch granules within plastids of leaves and storage organs. In leaves, starch granules are synthesized in chloroplasts during the day then degraded during the night to supply metabolic energy required for maintenance and growth in the absence of photosynthesis (Smith and Zeeman, 2020). In most plants, leaf starch granules are relatively small and have a flattened lenticular shape. In Arabidopsis, each chloroplast is reported to contain 5-7 starch granules that are approx. 1-2 µm in diameter at the end of the day (Crumpton-Taylor et al., 2012) and are formed within stromal pockets between thylakoid membranes (Seung et al., 2018; Bürgy et al., 2021). However, factors involved in determining the number, shape and size of starch granules within leaf chloroplasts are still poorly understood.
Recent work has advanced our understanding of starch granule initiation in chloroplasts, which determines both the number and size of granules. In Arabidopsis, the key enzyme that mediates granule initiation is STARCH SYNTHASE 4 (AtSS4) – a glucosyltransferase that may perform the initial glucan elongation steps that prime granule formation (Roldán et al., 2007). Mutants defective in AtSS4 have drastic reductions in the number of starch granules in chloroplasts, most containing zero or one granule (Roldán et al., 2007; Bürgy et al., 2021; Lu et al., 2018). Several additional proteins are proposed to act in granule initiation, by influencing AtSS4 activity and/or localisation. These include PROTEIN TARGETING TO STARCH2 (AtPTST2), a Carbohydrate Binding Module 48 (CBM48)-containing protein proposed to deliver short maltooligosaccharide (MOS) substrates to AtSS4 for further elongation (Seung et al., 2017). AtPTST2 interacts with the MAR1-BINDING FILAMENT PROTEIN (AtMFP1) (Seung et al., 2018). Most chloroplasts in mutants defective in AtPTST2 or AtMFP1 have a single, large granule, but the granules retain a normal flattened disc shape (Seung et al., 2018; Thieme et al., 2022). Interestingly, in Atss4 mutants, starch granules are round rather than lenticular, suggesting AtSS4 is also required for correct granule morphology (Roldán et al., 2007; Bürgy et al., 2021; Lu et al., 2018).
It has also been speculated that the flattened shape of leaf starch granules may be influenced by space available for granule growth within the chloroplast, in between layers of thylakoid membranes (Zeeman et al., 2002). Indeed, recent work demonstrated that granules form within stromal pockets between thylakoids, in which they initiate as multiple small granules that eventually fuse into a single granule (Bürgy et al., 2021). There is also evidence for associations between granule initiation proteins and thylakoid membranes. AtMFP1 is exclusively bound to the stromal side of thylakoid membranes (Jeong et al., 2003; Seung et al., 2018). This allows some AtPTST2 to attach to thylakoids through interaction with AtMFP1 (Seung et al., 2018). AtMFP1 and AtPTST2 co-locate to numerous, discrete puncta within chloroplasts, and it is hypothesised that they define areas in which granules initiate and form (Seung et al., 2017; Seung et al., 2018; Seung and Smith, 2019). AtMFP1 may also associate with plastid nucleoids (Jeong et al., 2003), although the role of this interaction is not known.
These previous findings raise the possibility that the number and shape of starch granules are influenced by the nature and extent of stromal pockets that form between thylakoids. Relatively little is known about the nature of these pockets and the factors involved in their formation, but important clues come from the phenotypes of Arabidopsis mutants deficient in a chloroplastic dynamin-like protein, FZO-LIKE (FZL). AtFZL is located in punctate structures associated with both thylakoids and envelope membranes of chloroplasts. In mutants lacking this protein, chloroplast thylakoids are disorganised and the ratio of granal to stromal thylakoids is reduced (Gao et al., 2006). It has been proposed that AtFZL may fuse membrane compartments together during thylakoid biogenesis, with a particular role in fusing developing grana stacks to stromal thylakoids (Gao et al. 2006; Liang et al., 2018). Other studies of Arabidopsis and Chlamydomonas fzl mutants have confirmed the importance of FZL for chloroplast structure (Findinier et al., 2019; Landoni et al., 2013).
Here, we have taken advantage of the Atfzl mutant to analyse whether disruption of thylakoid organisation affects starch granule formation and morphology. We demonstrate that AtFZL is required for correct starch granule morphology in Arabidopsis. Chloroplasts of the Atfzl mutant produce starch granules that are reduced in size and have uneven surface morphology. We propose that these changes in granule morphology result from disorganised thylakoids at the periphery of starch granules in the mutant, indicating that thylakoid architecture influences starch granule growth.
Results
Loss of AtFZL results in increased starch granule number per chloroplast due to increased chloroplast size
To investigate whether the thylakoid-organising protein AtFZL influences starch granule formation in Arabidopsis leaves, we obtained three independent T-DNA mutants from the Eurasian Arabidopsis Stock Centre (uNASC). Two of these lines have been described previously (Gao et al., 2006): Atfzl-1 (SALK_009051) has two insertions in the AtFZL gene (one in the intron between exons 9 and 10 that encodes the second transmembrane domain, and another in the 3’ UTR), while Atfzl-2 (SALK_033745) has an insertion in exon 2 (Fig. 1a). An additional line, Atfzl-3 (SALK_152584C) has a T-DNA insertion in exon 4, which encodes part of the GTP-binding domain. The presence of each T-DNA insertion in the respective mutants was confirmed using PCR. All plants were grown in controlled environment chambers under a 12 h photoperiod.
(a) Schematic illustration of the gene model for the primary transcript of AtFZL (At1g03160.1). Exons are represented by grey boxes and UTRs are represented by white boxes. Grey and yellow arrows indicate T-DNA insertion sites (SALK_009051: Atfzl-1, SALK_033745: Atfzl-2, SALK_12584C: Atfzl-3). Regions encoding protein domains are shown as black and grey arrows (CC: coiled coil, TenI: thiamine phosphate synthase, TM: transmembrane).
(b) Photographs of 30-day-old Atfzl mutant and wild-type (Col) rosettes. Bar = 1 cm.
(c) Fresh weight of 28-day-old rosettes. Dots represent fresh weight of an individual rosette (n = 8 per genotype). The bottom and top of the box represent the lower and upper quartiles respectively, and the band inside the box represents the median. The ends of the whiskers represent values within 1.5x of the interquartile range. Outliers are values outside 1.5x the interquartile range. No significant differences (including outliers) were observed under Kruskal-Wallis One-Way Analysis of Variance on the Ranks (P = 0.437).
(d) Iodine-stained, 33-day-old Atfzl mutant and wild-type rosettes (Col-0). Bar = 1 cm.
(e) Starch content of Atfzl mutants at the end of the day (ED) and the end of the night (EN). Bars and error bars represent the mean±SEM of n = 8 plants, while dots represent individual data points. Values indicated with an asterisk are significantly different from Col-0 at each timepoint under a one-way ANOVA and Holm-Šídák multiple comparisons procedure (P ≤ 0.001).
Although previous studies reported pale leaves, chlorotic lesions and delayed flowering in Atfzl mutants (Gao et al., 2006; Landoni et al., 2013), rosette morphology of the Atfzl-1 and Atfzl-3 mutants was indistinguishable from the wild type (WT) under our growth conditions (Fig. 1b).
The fresh weights of the Atfzl-1 and Atfzl-3 rosettes were not significantly different from those of the WT (Fig. 1c). We then assessed the effect of the Atfzl mutations on transient starch accumulation in leaves. Iodine staining for starch in whole rosettes harvested at the end of the day showed no obvious differences in staining pattern or intensity between the Atfzl mutants and the WT (Fig. 1d). Quantitative starch assays confirmed that there were no differences in starch content between the mutants and the WT at the end of the day (Fig. 1e). However, there was a significant reduction in the end-of-night starch content in both mutants compared to the WT.
To determine if Atfzl mutations affect the number and size of starch granules per chloroplast, we examined sections of rosette leaves using light microscopy. Consistent with earlier reports (Gao et al., 2006), chloroplasts in mesophyll cells in the Atfzl mutants were larger than those of the WT (Fig. 2a-d). The total chloroplast area per cell section was significantly larger in the Atfzl-3 mutant and slightly larger in the Atfzl-1 mutant, when compared to the WT (Fig. 2a). Additionally, the total area of individual mesophyll cells was significantly larger in the Atfzl-3 mutant and slightly larger in the Atfzl-1 mutant than in the WT (Fig. S2d). Moreover, the shape of the chloroplasts in the Atfzl mutants was drastically altered and not uniform - many were flat and elongated, while some were WT-like and round (Fig. 2b-d).
(a) Total chloroplast section area per cell section. Values represent total area of chloroplast sections per cell for ten cells per biological replicate. Values that are significantly different from Col-0 under a Kruskal-Wallis One Way Analysis of Variance on Ranks followed by a post-hoc ukey’s est (P ≤ 0.0) are indicated with an asterisk.
(b-d) Light microscopy images of leaf sections of Atfzl mutants and the wild type (Col-0). A recently fully expanded leaf was harvested from 29-day-old plants at the end of day. Sections were stained ith toluidine lue and chi ‘s stain. Bar = 10 μm.
(e) Number of granules expressed relative to the chloroplast section area. No significant differences between genotypes were observed under Kruskal-Wallis One-Way Analysis of Variance on the Ranks (P = 0.140).
(f-h) Starch granule number per chloroplast section. Histograms show the frequency of chloroplasts containing a given number of granule sections relative to the total number of chloroplasts analysed (200 chloroplasts per replicate). Three biological replicates were analysed per genotype (labelled I-III), where each was prepared from a different plant.
Chloroplasts in the Atfzl mutants had visibly more starch granules than those of the WT. We therefore quantified the number of starch granules per chloroplast section using light microscopy images. The number of starch granules, when expressed relative to the chloroplast section area, was identical between the Atfzl mutants and the WT (Fig. 2e). Therefore, given the heterogeneity in chloroplast size observed in the mutant, the distribution of granule numbers per chloroplast section for the Atfzl-1 and Atfzl-3 mutants was significantly different from that of the WT – with the mutants having a broader distribution that was shifted towards higher granule numbers (Fig. 2f-h). We fitted a negative binomial regression to these distributions and extracted the mean count of granules per chloroplast section. Atfzl-1 and Atfzl-3 had a predicted mean count of 5.69 and 5.68 granules per chloroplast section, respectively; compared to the WT which had a predicted mean of 3.10 granules per chloroplast section. A similar trend towards increased numbers of granules per chloroplast was observed for the Atfzl-2 mutant (Fig. S1). When we correlated the number of starch granules per chloroplast section to the section area of the chloroplasts, we observed a distinct tendency of larger chloroplasts containing more starch granules in the Atfzl mutants (Fig. S2a-c), suggesting that chloroplast size was the primary factor underpinning the differences in the number of granules observed per chloroplast in the mutants.
AtFZL is required for normal starch granule morphology
We then examined starch granule morphology in the Atfzl-1 and Atfzl-3 mutants. Leaf starch was purified from whole rosettes harvested at the end of the day. Analysis of starch granule size distribution on the Coulter counter showed that Atfzl-1 and Atfzl-3 mutants had smaller granules than the WT (Fig. 3a). In the Atfzl-1 and Atfzl-3 mutants, the average granule diameter was 1.765±0.032 µm and 1.817±0.004 µm respectively. In the WT, the mean granule diameter was 2.120±0.024 µm. The smaller granule size in the mutants was also evident when purified starch granules were observed using scanning electron microscopy (Fig. 3 b-h). Interestingly, starch granules of the Atfzl mutants had distinctive uneven surfaces, as opposed to the smooth surfaces of WT granules. This feature was more pronounced in granules of Atfzl-3 than in those of Atfzl-1.
(a) Size distributions of starch granules purified from 34-day-old rosettes harvested at the end of the day, pooling 60 rosettes per genotype. The volume of granules at each diameter relative to the total granule volume was quantified using a Coulter counter. Values represent mean (solid line) ± SEM (shading) of three technical replicates.
(b-h) SEM images of starch granules purified from 34-day-old rosettes. Bar = μm. Panels (f) and (n) are close-up images of Atfzl-1 and Atfzl-3, showing the uneven granule morphology.
To determine whether the altered starch granule size and morphology in the Atfzl mutants was an inevitable consequence of larger chloroplast size in the mutants, we also examined purified starch granules from Atarc3 and Atarc6 mutants. Both these mutants are defective in plastid division and have larger chloroplasts (chloroplast cross-sectional areas are about two and four times larger than WT in Atarc3 and Atarc6, respectively), and like our Atfzl mutants, the number of granules per chloroplast volume in the Atarc3 and Atarc6 mutants is identical to the WT (Crumpton-Taylor et al., 2012).
However, the arrangement of the thylakoid membranes in Atarc3 and Atarc6 mutants remains largely unaltered (Maple et al., 2007; Pyke et al., 1994; Robertson et al., 1995; Shimada et al., 2004; Vitha et al., 2003). Coulter counter analysis of the purified starch granules of the Atarc3 and Atarc6 mutant showed that their sizes were not statistically different from the WT, with mean granule diameters of 2.349±0.031 and 2.149±0.010 µm, respectively (Fig. 3a). Atarc3 and Atarc6 starch granules also had similar morphology to WT granules (Fig. 3b). This indicates that the altered size and shape of starch granules in Atfzl mutants is not a function of increased chloroplast size, but may be attributable to altered thylakoid structure and organisation in these mutants.
The Atfzl-2 and Atfzl-3 mutants have altered thylakoid organisation around starch granules
We next examined the organisation of stromal pockets and starch granules within chloroplasts of the Atfzl mutants. Consistent with previous reports (Gao et al., 2006), transmission electron microscopy (TEM) showed disorganised thylakoids in chloroplasts of the Atfzl mutants. Granal lamellae appeared less stacked and uniform, and spaces between the thylakoid membranes were larger and more randomly distributed. Starch granules appeared to be less evenly distributed throughout the chloroplast in the Atfzl mutants, with some clusters of granules close to each other and some granules occurring unusually close to the plastid envelope. Most strikingly, membrane disorganisation was particularly pronounced around starch granules, often appearing as detached membrane structures, likely thylakoid membranes, in unusually large pockets surrounding the starch granules in the Atfzl-3 mutant (Fig. 4). These thylakoid membrane phenotypes were also seen in the Atfzl-2 mutant (Fig. S3), but not in the Atfzl-1 mutant, possibly reflecting different degrees of phenotype severity depending on the location of the T-DNA insertion site within the FZL gene (Fig. 1a).
TEM images of thin sections prepared from recently fully expanded leaves harvested from 29-day-old rosettes at the end of the day. Arrowheads indicate disor anised, ‘detached’ thylakoid layers observed at the periphery of starch granules in Atfzl-3. Asterisk (*) indicates granule clusters. Hash (#) indicates granules close to chloroplast envelope. Bar = 1 μm
Given the unusual membrane organisation around starch granules, we also examined thylakoid organisation and plastid ultrastructure in leaves of the Atfzl mutants harvested at the end of the night, after starch had been largely depleted (Fig. 5). Starch granules were almost completely degraded in the Atfzl mutants and the pockets previously occupied by granules were greatly reduced in size, but the detached thylakoid membranes observed around granules in Atfzl-3 chloroplasts now occupied the stromal pocket. By contrast, in the WT, the thylakoid pockets around starch granules had shrunk to accommodate the reduced granule size but the thylakoids remained organised.
TEM images of thin sections prepared from recently fully expanded leaves harvested from 32-day-old rosettes at the end o the ni ht. rro heads indicate disor anised, ‘detached’ thylakoid layers running through stromal pockets. “s” indicates starch ranules. Bar = 1 μm
The punctate localisation of AtMFP1 is not dependent on AtFZL
Since thylakoid organisation was severely disrupted in the Atfzl mutants, we investigated whether there is functional link between AtFZL and MAR BINDING FILAMENT-LIKE PROTEIN 1 (MFP1), a thylakoid-associated protein that is important for starch granule initiation in Arabidopsis.
Localisation experiments with AtFZL transiently expressed in Nicotiana benthamiana with a C-terminal YFP tag (AtFZL-YFP) showed that it localises to small puncta in chloroplasts (Fig. 6a-c), as has been reported previously (Gao et al., 2006). AtMFP1 also localises to discrete puncta (Seung et al., 2018), but co-expression of AtFZL-YFP and AtMFP1-RFP did not show a clear co-localisation pattern between the punctae of the two proteins (Fig. 6d-h). A pairwise immunoprecipitation assay between the AtFZL-YFP and AtMFP1-RFP proteins co-expressed in N. benthamiana leaves also failed to reveal an interaction between the two proteins (Fig. S4). We then tested whether AtFZL is necessary for the correct localisation of AtMFP1 in Arabidopsis. We created transgenic Arabidopsis lines expressing AtMFP1-YFP in the WT and Atfzl-3 mutant background. AtMFP1-YFP formed discrete puncta in both mesophyll and epidermal chloroplasts of the mutant as well as the WT. Interestingly, although it was visible from the chloroplast autofluorescence that the Atfzl-3 mutant had drastically increased chloroplast size in the mesophyll cells compared to the WT, the chloroplasts in the epidermal pavement cells were indistinguishable in size and shape from those of the WT (Fig. 6i-n). Taken together, it is unlikely that AtFZL affects starch granule morphology by affecting AtMFP1 function.
Images were obtained using confocal laser-scanning microscopy.
(a-h) Images of AtFZL-YFP and AtMFP1-RFP expressed in N. benthamiana epidermal cells. The YFP- and RFP fluorescence are shown in yellow and cyan, while chlorophyll autofluorescence is shown in red.
(i-n) Z-projection of image stacks of AtMFP1-YFP stably expressed in Arabidopsis wild-type (Col-0) and Atfzl-3 backgrounds. Recently fully expanded leaves from 5-week-old rosettes were imaged. Examples of chloroplasts of epidermal pavement cells are indicated with an arrowhead, and those of mesophyll cells are marked with an asterisk. Bars = 5 μm.
Discussion
Correct thylakoid organisation is required for normal starch granule morphology in chloroplasts
We demonstrate that AtFZL, a chloroplast protein required for correct organization of thylakoid membranes (Gao et al., 2006; Liang et al., 2018), is also necessary for correct starch granule morphology. Starch granules in the Atfzl mutants are smaller and have less even surfaces than those of the wild type (WT)(Fig. 3).
Starch granules in Arabidopsis chloroplasts orm ithin de ined ‘stromal pockets’ et een thylakoid membranes, via the fusion of multiple, separately-initiated starch granules (Bürgy et al., 2021). The alterations in thylakoid structure in Atfzl mutants appear to affect the size and structure of these stromal pockets. Consistent with previous reports (Gao et al., 2006), we found that Atfzl chloroplasts had disorganised thylakoids with less uniformly stacked grana than WT chloroplasts, and more randomly distributed intermembrane spaces (Fig. 4). In the Atfzl-2 and Atfzl-3 mutants, we also observed disorganised or detached membranes, likely thylakoid membranes, particularly within the stromal pockets around the starch granules (Fig. 4). These membranes were still visible within stromal pockets at the end of the night following starch degradation (Fig. 5). Interestingly, detached membranes were not observed in the Atfzl-1 mutant (Fig. 4 and Fig. S3). In Atfzl-1, the T-DNA insertion is integrated in the region encoding the C-terminal end of AtFZL after the predicted transmembrane domains. Insertions in both Atfzl-2 and Atfzl-3 are closer to the start of the coding sequence, either before or within the region encoding the GTP-binding domain (Fig. 1). It is plausible that the position of the Atfzl-1 insertion towards the end of the coding sequence permits the translation of a partly functional protein, resulting in a weaker phenotype. Notably, the Arabidopsis gene annotation for FZL suggests four different splice variants at the locus, two of which (At1g03160.2 and At1g03160.3) encode a truncated protein lacking the two C-terminal transmembrane domains. The fzl-1 mutation does not disrupt the coding sequence of these two splice variants, and may allow these truncated proteins to be expressed.
We propose that the aberrant starch granule size and shape observed in the Atfzl mutants are consequences of altered thylakoid structure. Disrupted structure of the stromal pockets may limit either the space available for fused granules within the pocket to grow, or the number of initiation events that can occur within a stromal compartment – resulting in smaller starch granules. The distinct uneven surface of the mutant starch granules, which was more pronounced in the Atfzl-3 mutant than in Atfzl-1, might also be influenced by the less organised thylakoid membranes around the stromal pocket, which may interfere with starch granule growth and prevent the formation of a smooth granule surface. It is also possible that restricted stromal space around growing starch granules might introduce tension within the matrix of growing granules, leading to deformation.
Starch granule morphology is in addition influenced by components that control the non-random deposition of new material (Bürgy et al., 2021). Arabidopsis STARCH SYNTHASE 4 (AtSS4), a key enzyme in starch granule initiation (Roldán et al., 2007), is required for anisotropic growth of starch granules in Arabidopsis (Bürgy et al., 2021). Knock-out of AtSS4 results in uniform deposition of newly synthesised starch onto the granule surface, leading to spherical granules in place of the lenticular granules in the WT (Bürgy et al., 2021; Roldán et al., 2007). AtSS4 and PROTEIN TARGETING TO STARCH2 (AtPTST2) are associated with thylakoid membranes (Gámez-Arjona et al., 2014). AtPTST2 co-localises in discrete puncta within Arabidopsis chloroplasts with the MAR1-BINDING FILAMENT PROTEIN 1 (AtMFP1) that is exclusively bound to thylakoid membranes (Seung et al., 2018). In Atmfp1 and Atptst2 mutants, most chloroplasts contain only a single, large starch granule, but this retains the WT-like lenticular shape and smooth surfaces (Seung et al., 2018; Thieme et al., 2022).
We investigated whether the effects of loss of AtFZL on starch granule morphology result from its interactions with or displacement of the MFP1 thylakoid-binding component of granule initiation complexes. Our results provided no evidence for a functional link between AtFZL and AtMFP1. Although we were able to confirm that both AtMFP1 and AtFZL localise to puncta within Arabidopsis chloroplasts (Gao et al., 2006; Seung et al., 2018), co-expression of AtFZL-YFP and AtMFP1-RFP in N. benthamiana chloroplasts revealed distinct punctate patterns for each protein, and the puncta overlapped only occasionally (Fig. 6). In addition, we did not detect a direct interaction of AtFZL with AtMFP1 (Fig. S4). Moreover, AtMFP1 localisation was still punctate in the Atfzl mutants (Fig. 6). These results support the idea that the defective starch granule morphology in Atfzl chloroplasts is caused by altered chloroplast architecture, rather than by direct interaction with AtMFP1.
Starch granule number in the Atfzl mutants correlates with chloroplast area
In common with plastid division mutants (Gao et al., 2006; Maple et al., 2007; Pyke et al., 1994; Shimada et al., 2004; Vitha et al., 2003), mesophyll cells of the Atfzl-1 and Atfzl-3 mutants contained larger chloroplasts with a broad variety of shapes. This resulted in the increase of the total chloroplast area per cell, when compared to the WT (Fig. 2). However, the size of the mesophyll cells was also increased in the Atfzl mutants (Fig. S2), leading to a similar chloroplast compartment size per cell as the WT. Interestingly, it seems that loss of AtFZL does not affect the chloroplast size of epidermal pavement cells (Fig. 6).
Although the Atfzl mutants had a significantly broader distribution of starch granule numbers per chloroplast section than the WT (Fig. 2), the number of starch granules per unit chloroplast area in the mutants was not different from WT (Fig. 2). This result mirrors that of a previous study of Arabidopsis arc mutants defective in plastid division, where chloroplasts had approximately the same number of granules per unit volume of stroma as the wild type regardless of the size and number of chloroplasts per cell (Crumpton-Taylor et al., 2012). Thus, AtFZL itself is not involved in the control of the number of starch granules per plastid volume. Likewise, the increased chloroplast size by itself is unlikely to cause the smaller starch granule size in the Atfzl mutants, since starch granule size is unaltered in the plastid division mutants Atarc3 and Atarc6 - which have enlarged chloroplasts comparable in size (or larger) to those of Atfzl mutants and their thylakoid membrane organisation remains largely unchanged (Maple et al., 2007; Robertson et al., 1995; Shimada et al., 2004; Vitha et al., 2003). More likely, the amount of accessible stromal volume or thylakoid membrane space might be involved in determining the number of initiation events per plastid. Future work using high-resolution 3D-imaging of mesophyll cells and chloroplasts in relation to the whole leaf will be necessary to further elucidate the relationships between cell and plastid size and starch granule formation.
Leaf starch content is unaffected by the Atfzl mutations
The loss of proper thylakoid organisation in Arabidopsis can lead to a range of growth phenotypes – from severely retarded growth and development (e.g., mutants of VESICLE-INDUCING PROTEIN IN PLASTIDS 1-Gupta et al., 2021; Kroll et al., 2001) and variegated leaf patterns (e.g., mutants of THYLAKOID FORMATION 1 – Wang et al., 2004), to almost no defects in growth or leaf colour (CURVATURE THYLAKOID 1 – Armbruster et al., 2013). Arabidopsis Atfzl mutants were previously reported to have pale leaves, chlorotic lesions, reduced growth and delayed flowering (Gao et al., 2006; Landoni et al., 2013). However, under our growth conditions, they showed no differences in growth or development from the WT (Fig. 1 b-c). One major difference in growth conditions between our study and previous studies is that we used a 12 h rather than a 16 h photoperiod. The total starch content of Atfzl rosettes in our study was indistinguishable from the WT at the end of the day (Fig. 1 d-e). The absence of differences in both rosette fresh weight and starch content indicates that the Atfzl mutation is unlikely to have disrupted general photosynthetic efficiency and carbon fixation under our growth conditions. Interestingly, we observed a nearly complete degradation of starch granules in the Atfzl mutants during the night, whereas in the WT, small granules remained. It is possible that the reduced starch granule size allows more complete starch degradation in the mutants (Fig. 1e, 3). The opposite trend has been observed in mutants with increased granule size, such as Atptst2, Atmrc and Atmfp1, which all have significantly higher starch content than the WT at the end of the night (Seung et al., 2017; Seung et al., 2018).
In summary, we discovered that starch granules in Arabidopsis Atfzl mutants are smaller and have a more uneven surface morphology than the WT. While the experiments with AtMFP1 seem to indicate that there is no direct association of AtFZL with starch granule initiation proteins, our results suggest a strong influence of chloroplast ultrastructure and specifically thylakoid organisation on starch granule morphology.
Materials and Methods
Plant material and growth conditions
Arabidopsis thaliana plants were grown on soil in controlled environment chambers at constant 20°C and 60% relative humidity, and 12h light (80-150 µmol m-2 s-1)/12h dark cycles. Nicotiana benthamiana plants were grown in the glasshouse set to a minimum of 16 h light at 22°C.
Arabidopsis T-DNA insertion lines SALK_009051 (Atfzl-1), SALK_033745 (Atfzl-2) and SALK_152584C (Atfzl-3) were obtained from the Eurasian Arabidopsis Stock Centre (uNASC). Mutants SALK_009051, SALK_152584C were homozygous and mutant SALK_033745 was heterozygous as determined by genotyping PCR using the primers listed in Supplementary Table S1. Homozygous SALK_033745 plants were selected in the next generation by genotyping PCR using the primers listed in Supplementary Table S1.
Cloning and plant transformation
To produce the AtFZL-YFP construct, AtFZL (without its stop codon) was amplified from total cDNA prepared from Arabidopsis leaves using the attB-flanked primers listed in Supplementary Table S1. The amplicon was recombined into the Gateway entry vector pDONR221 using BP clonase II (Thermo Fisher Scientific), and the sequence was verified by Sanger sequencing. The insert was recombined into the Gateway-compatible destination vector pUBC-YFP (Grefen et al., 2010) using Gateway LR clonase II (Invitrogen, Thermo Fisher Scientific). For AtMFP1-YFP and AtMFP-RFP constructs, we used the AtMFP1 pDONR221 from Seung et al. (2018) to recombine the AtMFP1 coding sequence into pUBC-YFP and pB7RWG, respectively.
Nicotiana benthamiana leaves were transiently transformed via infiltration of Agrobacterium tumefaciens (GV3101) carrying the respective constructs. The bacteria were grown at 28 °C for 48 h. Cultures were resuspended in MMA buffer (10 mM MES pH 5.6, 10 mM MgCl2, 0.1 mM acetosyringone) at OD600 = 1.0 for infiltration of leaves to be used for confocal microscopy, and OD600= 0.3 (0.2 for p19) for leaves to be used for protein extraction. Leaves were infiltrated into the abaxial side using a syringe and harvested for confocal microscopy and protein extraction 48-72 h after infiltration.
Arabidopsis thaliana Col-0 and Atfzl-3 mutant plants were transformed with AtMFP1:pUBC-YFP by floral dipping (Zhang et al., 2006). Transformants were selected using the BASTA resistance marker.
Starch content and iodine staining
Total starch content was quantified as described by Smith & Zeeman, 2006. Rosettes of 3-4 week old Arabidopsis plants were harvested at the end of day and end of night. The entire rosette was homogenised in 0.7 M perchloric acid and the insoluble fraction was collected by centrifugation. The pellet was washed in 80% ethanol three times and resuspended in water. Starch was gelatinised at 95°C and digested to glucose using a mix of amyloglucosidase (Megazyme) and ɑ-amylase (Megazyme). Glucose was then quantified by measuring NADH production using an assay based on hexokinase and glucose-6-phosphate dehydrogenase (Roche). Starch content was calculated in glucose equivalents.
For iodine staining, rosettes were harvested at the end of the day and chlorophyll was removed using 80% (v/v) ethanol prior to staining with Lugol’s solution (I2/KI solution).
Chloroplast visualisation using light- and electron microscopy
Leaf segments were taken 1 mm from the central axis of Arabidopsis leaves of 3-4 week old plants at about one third of the distance between the tip and the petiole. Preparation for microscopy was as described in Watson-Lazowski et al. (2022). Briefly, leaf segments were fixed in 2.5% (v/v) glutaraldehyde in 0.05 M sodium cacodylate, pH 7.3 at 4°C, post-fixed in 1% (w/v) osmium tetroxide (OsO4) in 0.05 M sodium cacodylate for 2 h at room temperature, dehydrated in ethanol and infiltrated with LR White resin (Agar Scientific, Stansted, UK), using a EM TP embedding machine (Leica, Milton Keynes, UK). LR White blocks were polymerised at 60°C for 16 h. For light microscopy, semi-thin sections (ca. 0.5 µm) were prepared and stained with 1% (w/v) toluidine blue and the Periodic Acid Schiff kit (ab150680, Abcam) as described in Hawkins et al. (2021) and mounted in Histomount. Sections were imaged using the Zeiss Axio Imager Z2 using a 100x oil immersion objective. For transmission electron microscopy (TEM) ultrathin sections (ca. 80 nm) were cut with a diamond knife and placed onto formvar and carbon coated copper grids (EM Resolutions, Sheffield, UK). The sections were stained using 2% (w/v) uranyl acetate for 1 h and 1% (w/v) lead citrate for 1 min, washed in water and air dried. Sections were imaged on a Talos 200C TEM (FEI) at 200 kV and a OneView 4K x 4K camera (Gatan, Warrendale, PA, USA). Images were processed and analysed using ImageJ software (http://rsbweb.nih.gov/ij/) and Adobe Photoshop 2020. Granule numbers per plastid section were compared by fitting Mixed Effect Models (negative binomial regression) as in Chen et al. (2022).
Starch granule purification, morphology and size analysis
Four-week-old plants (60 rosettes per genotype) were pooled and homogenised in 50 mM Tris-HCl, pH 8.0, 0.2 mM EDTA, 0.5% v/v Triton X-100 using an immersion blender. The suspension was sequentially filtered through Miracloth, a 60 µm nylon mesh and a 20 µm nylon mesh. Starch granules were separated by centrifugation at 2500g over a Percoll cushion (95% v/v Percoll, 5% v/v 0.5 M Tris-HCl, pH 8.0) then washed in water until clean, then washed twice in 0.5% w/v SDS in water and once in 100% ethanol to dry. Starch granule morphology was imaged using a Nova NanoSEM 450 (FEI) scanning electron microscope. Granule size distribution was analysed using a Beckman Multisizer 4e Coulter counter (Beckman Coulter) with a 30 µm aperture. A minimum of 50,000 particles were measured per replicate. Measurements were conducted with logarithmic bin spacing but are presented on a linear x-axis. Mean granule diameter was determined by the Multisizer 4e Coulter counter Software (Beckman Coulter).
Confocal microscopy
For imaging of fluorescent proteins in Arabidopsis and Nicotiana benthamiana leaves, images were acquired on either the Zeiss LSM880 or the Leica Stellaris 8 laser scanning confocal microscope using a 40x or 63x water-immersion objective. YFP signal was excited using an Argon or a white light laser set to 514 nm and emission was detected at 519 nm to 560 nm. RFP signal was excited using a white light laser set to 555 nm and emission was detected at 562 nm to 623 nm. Chlorophyll autofluorescence was excited using 561 nm laser or a white light laser set to 555 nm, 561 nm or 576 nm and emission was detected at 651 nm to 750 nm. Images were processed using ImageJ software (http://rsbweb.nih.gov/ij/) and Adobe Photoshop 2020.
Protein extraction and Immunoblotting
For the pairwise immunoprecipitation assay between AtFZL-YFP and AtMFP1-RFP, proteins were extracted from transiently transformed Nicotiana benthamiana leaves. One or two leaf discs (1 cm diameter) were harvested from each of three transformed leaves and homogenised in extraction buffer (50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 1% v/v Triton X-100, 1x protease inhibitor cocktail (Roche), 1 mM DTT). The homogenate was centrifuged at 20,000g for 10 min, and the supernatant with the proteins was used as the input for immunoprecipitations. Immunoprecipitation was performed using the µMACs GFP Isolation Kit and µMacs Columns (Miltenyi Biotec). For immunoblotting, antibodies were used in the following concentrations: anti-YFP (Torrey Pines TP401 – 1:500) and anti-HA (Abcam ab9110 – 1:5000). Bands were detected using chemiluminescence, using the Anti-rabbit IgG (whole molecule)-Peroxidase (Sigma A0545 – 1:20000) and the SuperSignal West Femto Kit (Thermo Scientific).
Author contributions
L.E. conceived the study. L.E. and D.S. designed the research. L.E., Q.Y.N. and J.E.B. performed the research and analysed data; L.E. and D.S. wrote the article with input from all authors.
Figure Legends
Figure S1: Starch granule number per chloroplast section in the Atfzl-2 mutant.
(a-b) Light microscopy images of Atfzl-2 and wild type (Col-0) leaf sections. A young leaf was harvested from 29-day-old plants at the end of day. Sections were stained with toluidine blue and chi ‘s stain. Bar = 10 μm.
(c-d) Starch granule number per chloroplast section. Histograms show the frequency of chloroplasts containing a given number of granule sections relative to the total number of chloroplasts analysed (200 chloroplasts in one replicate).
Figure S2: Starch granule number per area of chloroplast section and cell area.
(a-c) Granule number per chloroplast plotted against chloroplast area for Atfzl-1, Atfzl-3 and the wild type (Col-0). Starch granule number per chloroplast section (data in Figure 2d-f), and area of each chloroplast section was determined for 200 chloroplasts per biological replicate (I-III) per genotype.
(d) Area of cell section of 10 cells for each of three biological replicates. Values indicated with an asterisk are significantly different from Col-0 at each timepoint under a one-way ANOVA and Holm-Šídák multiple comparisons procedure (P ≤ 0.0).
Figure S3: Chloroplast ultrastructure and starch granule placement in Atfzl-2 mutant.
TEM images of thin sections prepared from young leaves harvested from 29-day-old rosettes at the end o the day. rro heads indicate disor anised, ‘detached’ thylakoid layers o ser ed at the periphery of starch granules in Atfzl-2. Bar = 1 μm
Figure S4: Pairwise immunoprecipitation of AtFZL-YFP and AtMFP1-RFP.
Immunoprecipitation (IP) assay using anti-GFP beads for AtFZL-YFP and AtMFP1-RFP transiently co-expressed in tobacco leaves. Note that anti-GFP binds to AtFZL-YFP but not AtMFP1-RFP. Immunoblots with GFP and RFP antibodies were used to detect the AtFZL-YFP and AtMFP1-RFP proteins.
Acknowledgements
The authors thank the John Innes Centre (JIC) Horticultural Services for providing growth facilities and maintenance of plant material, JIC Bioimaging for providing access to microscopes, and Dr Andrew Breakspear for performing the LR-reaction to construct AtMFP1 in pB7RWG and transforming it into Agrobacterium. We thank Prof. Alison Smith for critically reading this manuscript. This work was funded through a Leverhulme Trust Research Project grant RPG-2019-095 (to D.S), a John Innes Foundation (JIF) Chris J. Leaver Fellowship (to D.S), and BBSRC Institute Strategic Programme grants BBS/E/J/000PR9790 and BBS/E/J/000PR9799 (to the John Innes Centre).
References
