In the leaves of most C4 plants, mesophyll (M) and bundle sheath (BS) cells develop and maintain highly differentiated biochemical networks. Separation and analysis of M and BS cells has greatly influenced our understanding of the C4 pathway. A number of approaches including mechanical separation, digestion with cell wall-degrading cocktails, laser-capture microdissection, and leaf rolling have been used to isolate these cell types. Although leaf rolling is conceptually and practically the simplest method, to date it has only been used to assess the metabolite content of M cells from C4 leaves of maize. This study reports an adapted leaf-rolling method for the isolation of high-quality RNA from M cells of sorghum. Analysis of leaf cell structure, RNA integrity, and transcript abundance of marker genes demonstrated that the sap collected by leaf rolling was from M cells and had no significant contamination. It was concluded that leaf rolling is a fast, cheap, and efficient method of measuring transcript abundance in M cells of sorghum.
Isolation of specific cell types is fundamental for analysis of complex traits in multicellular organisms. For example, in leaves of C4 plants, photosynthetic machinery is partitioned between mesophyll (M) and bundle sheath (BS) cells (Brown et al., 2005). The ability to selectively isolate M and BS cells from leaves of C4 plants has been instrumental in characterizing the compartmentation of proteins (Kanai and Edwards, 1973; Ku and Edwards, 1975; Harrison and Black, 1982; Nakano and Edwards, 1987; Meierhoff and Westhoff, 1993; Taniguchi et al., 2004; Majeran et al., 2005, 2008, 2010; Brautigam et al., 2008) and metabolites (Chollet and Ogren, 1973; Jenkins and Boag, 1985; Leegood, 1985; Stitt and Heldt, 1985) associated with the C4 leaf. Furthermore, the underlying patterns of mRNA abundance in these cells can also be investigated (Sheen and Bogorad, 1986, 1988; Langdale et al., 1991; Westhoff et al., 1991; Kubicki et al., 1994; Roth et al., 1996; Wyrich et al., 1998; Sawers et al., 2007; Covshoff et al., 2008; Jiao et al., 2009; Li et al., 2010).
Typically, M cells are isolated as protoplasts by enzymatic digestion, whilst BS strands, which consist of BS cells attached to the vasculature, are isolated by mechanical means using shearing forces. These methods yield abundant quantities of M protoplasts and BS strands suitable for a variety of downstream applications, as described above. However, the prolonged period required for protoplast isolation is known to induce stress in leaf tissue and alter transcript patterns in M cells (Sawers et al., 2007). Noise in the data introduced by stress can be normalized using a ‘stress’ control (Hall et al., 1998; Brutnell et al., 1999; Rossini et al., 2001; Markelz et al., 2003; Sawers et al., 2007; Covshoff et al., 2008). To allow this, four samples are created: (i) total leaf tissue; (ii) total leaf tissue that has undergone a mock protoplast digestion without cell wall-degradation enzymes; (iii) total leaf tissue that is digested to release protoplasts; and (iv) total leaf tissue that is mechanically shredded and filtered to collect BS strands. M cell transcripts are normalized for stress by subtraction, using the mock protoplast and total leaf samples. Afterwards, M and BS transcript profiles are compared in order to determine cell specificity. However, this method adds cost and complexity to experiments. Alternatively, experiments can be designed such that only BS:BS or M:M comparisons are made between differing samples, such as mutant and wild-type plants (Covshoff et al., 2008). Relationships between M and BS transcript profiles can then be inferred but not compared directly.
Recently, laser-capture microscopy (LCM) has simplified M and BS comparative transcriptomics. LCM allows the collection of pure samples of specific cells, without stress-induced transcript changes (Nakazono et al., 2003; Jiao et al., 2009; Li et al., 2010). Thus, RNA profiles of M and BS cells can be compared directly. However, LCM is expensive, available in only a few laboratories, and the quantity of RNA extracted is low, reducing the technique’s current impact on C4 studies. A simple, cost-effective technique to rapidly isolate RNA from M cells, without inducing stress, is therefore required.
In maize, enzymes and metabolites can be extracted specifically from M cells by pressing the leaf with a specialized rolling device (Furbank et al., 1985; Leegood, 1985). This method takes advantage of C4 monocot vein structure, where tightly packed veins run parallel to the leaf axis forming long files that are resistant to mechanical disruption. We were interested in whether this leaf rolling method would allow isolation and analysis of mRNA abundance in M cells of C4 leaves and so represent a complementary approach to those described above. Here, we report a simplified version of this method to mechanically isolate M cell contents from sorghum. A wallpaper seam roller was used to press M cell contents from the sorghum leaf. This simple technique allowed the rapid extraction of high-quality mRNA from M cells and therefore direct comparison of M cell and BS strand RNA profiles.
Materials and methods
Plant materials and growth
In April and May, 150 seeds of sorghum (Sorghum bicolor BTx623) were imbibed in water for 2 d in a growth chamber at 28 °C with a 16h photoperiod, 400 µmol m–2 s–1 photon flux density in 40% humidity. Seeds were then sown onto soil (approx. 3:1 compost to medium-sized vermiculite) in a 5×10 grid on each of three trays and allowed to grow in the same conditions until the third leaf emerged. Plants were transferred to a greenhouse and grown until the fourth leaf was approximately half the height of the third leaf. This occurred at least 3 d after transfer. At this stage, the third leaf was fully emerged from the stalk. At this developmental stage, the third leaf was harvested for RNA extraction by cutting just above the ligule.
Standardization of tissue collection
Sorghum leaves were harvested between 4 and 6h after dawn. Each biological replicate of total leaf tissue was prepared by combining the third leaf from two plants in a single sample and flash freezing in liquid nitrogen. Biological replicates for M protoplast isolation and M rolling were prepared using the third leaf of 16 plants. M protoplasts were prepared as described previously (Markelz et al., 2003) with the following modifications: leaves were cut into 1–2mm2 and the first filtration was performed with a 135 µm mesh. Pelleting was performed at 200g for 10min. After the final spin, the wash buffer was decanted and the pellet resuspended in residual supernatant. Biological replicates of the mock protoplast stress control were prepared using the third leaf of eight plants (Markelz et al., 2003) with the following modifications: leaves were cut into 1–2mm2 and filtered through 60 µm mesh. Biological replicates of BS strands were prepared using the third leaf of 16 plants (Markelz et al., 2003) with the following modifications: high speed shredding was performed three times for 1min each.
A glass plate was treated with 0.1N NaOH for at least 30min, rinsed with deionized water, fitted onto a box full of ice, and allowed to cool. A 2ml round-bottomed tube containing 250 µl mirVana Lysis/Binding buffer from the mirVana miRNA Isolation kit (Ambion) was also placed on ice to cool. The roller (Wallpaper Seam Roller; Homebase) was washed with RNase AWAY (Life Technologies) prior to use and then kept on clean paper towels. Using a fresh razor blade, the third leaf was harvested just above the ligule, the mid-rib removed, and the lamina cut into 3–5cm sections (see Supplementary Video 1 at JXB online). These sections were laid flat on the glass plate and quickly rolled. Fluid that emerged from the pressed leaf was quickly removed by filter pipette from the plate or roller by adding 2 µl of mirVana buffer and dispensing into the 2ml tube containing the stock Lysis/Binding buffer on ice. The leaf roller and glass plate were washed with RNase AWAY before the next leaf was harvested.
RNA extraction from total leaf, mock protoplast stress control, protoplast, and BS tissues was performed using a mirVana miRNA Isolation kit (Ambion). Tissue was ground to a fine powder in liquid nitrogen. Depending on the volume of tissue that resulted, either the whole sample was added to 3ml of Lysis/Binding buffer or it was split between two tubes. Conical snap cap tubes (14ml volume) were used. The remaining steps followed the Organic Extraction protocol (section E) and Total RNA Isolation Procedure (section F.I.) supplied with the kit. RNA was eluted with 100 µl nuclease-free water (not DEPC-treated) from Ambion. Two of the BS samples required additional processing because of poor ratios after analysis of absorption at 260 and 230nm. After passing through the mirVana column, RNA from these BS samples was run through the mirVana protocol a second time, as above, to remove contaminants. RNA quality was assessed on a NanoDrop 1000 Spectrophotomer (Thermo Scientific) and then with an Agilent RNA 6000 Nano Kit on a 2100 Bioanalyzer (Agilent Technologies).
RNA (0.5 µg) was treated with RQ1 RNase-Free DNase (Promega) in 10 µl at 37°C for 30min. The reaction was stopped with 1 µl of RQ1 DNase Stop solution at 65 °C for 10min. Reverse transcription was performed with Superscript II (Invitrogen). Each reaction was diluted to 300 µl upon completion. Quantitiative PCR (qPCR) was performed using SYBR Green JumpStart Taq ReadyMix and the protocol from Sigma-Aldrich. In each 10 µl reaction, 4 µl cDNA and 4 µM of primers were used.
qPCR primers were designed to either bridge an exon–intron boundary or to anneal to different exons in the same gene. Primers were as follows: PEPC (Sb10g021330) forward 5’-TGGGAAGCAGCTCAGGGACAA ATA and reverse 5’-TGCTTCAGGTAAGGATCGCCTTCA-3’, MDH (Sb07g023920) forward 5’-TGCAGATCGAAGGGTGATGGTGAT-3’ and reverse 5’-ATGGGCAACGCATTTCTTCTCAGC-3’, ME (Sb03g003230) forward 5’-TGTTATTTCTGGAGCCGTCCGTGT-3’ and reverse 5’-TGATGTTGGTGAAGGGTGGGAAGA-3’, RBCS (Sb05g003480) forward 5’-AGCCTCGCCAAAGTCAGCAA-3’ and reverse 5’-ACGACAGCGTCTCGAACTTCTTGT-3’, CAB-m9 (Sb09G028720) forward 5’-GAGCTCAAGGTGAAGGAGCTCAAG AA-3’ and reverse 5’-GTGACGATGGCCTGGACGAAGAAT-3’, and ACTIN (Sb01g003250) forward 5’-TGTGACTGCAGAGGATGT CCAGAA-3’ and reverse 5’-TCTTGCTCTTCACCTTGGCAGAGT-3’. The threshold cycle (Ct) was normalized against ACTIN and ΔCt was used to compare RNA accumulation between samples. All P values were determined by t-test. The sorghum Cab-m9 homologue (Sb09G028720) was identified by a BLASTp search against the sorghum genome using the maize protein.
Results and discussion
RNA quality and yield
The quality of RNA extracted from total leaf, M sap isolated after leaf rolling, BS strands, M protoplasts, and the mock protoplast stressed control was assessed (Fig. 1 and Supplementary Fig. 1 at JXB online). RNA from the untreated total leaf sample separated in a pattern typical to sorghum (Fig. 1A). Cytosolic and plastid rRNA separated into four clear peaks, and transfer and small RNAs were observed as a well-defined short broad peak. No degradation products were observed. The same RNA peaks were clearly visible in RNA isolated from sap of rolled sorghum leaves, also with no degradation products (Fig. 1B), indicating that leaf rolling was able to isolate good-quality RNA. The same analysis of RNA quality was performed on BS strands isolated mechanically (Fig. 1C), mesophyll protoplasts (Fig. 1D), and the mock protoplast stressed control (Fig. 1E). In all cases, this analysis showed that good-quality RNA had been isolated.
Quantification of RNA concentration and yield from each sample showed that M rolling generated the lowest amounts (Fig. 1F). This was despite the fact that the quantity of starting material for M protoplast isolation and leaf rolling was the same. It is therefore recommended that both the quantity of starting material and final elution volume is optimized for downstream applications. BS strand RNA may also require additional processing, possibly due to starch contamination, resulting in reduced yield. In all cases, a sufficient amount of high-quality RNA was available for quantification of mRNA abundance by qRT-PCR.
RNA from leaf rolling is highly M enriched
Before rolling, M and BS cells, arranged in long files along the leaf axis, had clear cell membranes and visible chloroplasts (Fig. 2A). After rolling, whilst the BS cells remained intact, the M cells appear empty (Fig. 2B). This is consistent with the results of maize leaf rolling prior to analysis of metabolites (Leegood, 1985). These data were used as preliminary evidence that the wallpaper seam roller is an effective tool to extract M cell contents from sorghum leaves.
To test leaf rolling for M cell specificity, transcripts known to be expressed abundantly in M or BS cells, and not affected by stress (Sawers et al., 2007), were measured by qRT-PCR. PEPC and MDH were used as markers for M cells (Fig. 2C, 2D), while ME and RBCS were used for BS cells (Fig. 2E, 2F).
A comparison of total leaf and stress samples confirmed that PEPC, MDH, ME, and RBCS transcript accumulation was not affected by stress, allowing a direct comparison of protoplast and rolling profiles. M protoplasts and rolled leaves shared similar expression profiles, with no transcripts showing significant statistical differences. Transcripts of PEPC and MDH were abundant in both protoplasts and M rolled samples, whilst ME and RBCS transcripts were barely detectable. In contrast, BS strands had low amounts of PEPC and MDH but high amounts of ME and RBCS. Differences in transcript accumulation between rolled leaves/protoplasts and BS strands were statistically significant (P <0.01 for PEPC, MDH, and ME and P <0.05 for RBCS). These data indicate that leaf rolling allows isolation of mRNA from M cells of sorghum and that contamination from other cell types is not significant.
Leaf rolling limits stress-related alterations in transcript abundance
To test whether leaf rolling limits stress-related changes in transcript levels associated with M protoplast isolation, the abundance of the chlorophyll a-b binding protein M9 transcript (Cab-m9) was determined by qRT-PCR (Fig. 3). In maize, the abundance of Cab-m9 transcripts is statistically reduced in stress-treated compared with non-stressed total leaf tissue (Sawers et al., 2007). This implies that the abundance of Cab-m9 transcripts is underestimated after the production of M protoplasts. We identified a sorghum homologue to maize Cab-m9 and used qRT-PCR to determine the abundance of sorghum Cab-m9 transcripts after leaf rolling. The Cab-m9 transcripts showed a 68% reduction in abundance in stressed versus untreated total leaf tissue (P=0.01) and reduced accumulation in M protoplast samples relative to M rolling (Fig. 3). These data indicate that RNA obtained from M rolling is likely to more accurately reflect the transcript accumulation in M cells than that obtained from M protoplast preparations.
Applications of leaf rolling for RNA analysis
Because C4 crop plants are high yielding (Brown, 1999; Sage and Zhu, 2011), there is much interest in understanding the transcriptional networks that lead to and maintain these specializations (Sheen, 1999; Hibberd and Covshoff, 2010). An initial step in this sort of analysis is to define the presence and also the cellular location of mRNAs. Such knowledge may allow us to greatly improve C3 crop production (Covshoff and Hibberd, 2012). For example, installing the C4 pathway in rice could lead to an increase in grain yield of 50% (Mitchell and Sheehy, 2006; Hibberd et al., 2008; Sage and Zhu, 2011). However, to understand the C4 system, we need to define components that are restricted to each cell type. Protocols to isolate M and BS have been developed and vary in their ease of implementation (Edwards et al., 2001). Isolating M cells by protoplast isolation results in altered transcript levels due to stress (Sawers et al., 2007), complicating experimental design and analysis. LCM allows the collection of pure M and BS cells (Jiao et al., 2009; Li et al., 2010), but the technique is currently not feasible for most laboratories. The present data showed that leaf rolling can be used to collect high-quality M RNA from fully expanded leaves of sorghum. The method is easy to establish and can be used in a regular laboratory setting, or even in outreach activities. The protocol allows direct comparison of M cells and BS strands in downstream applications.
Supplementary data are available at JXB online.
Supplementary Video 1. An instructional video on leaf rolling for isolation of M cell RNA from sorghum leaves. When following this protocol, sap should be added to the Lysis/Binding buffer as quickly as possible after rolling in order to prevent RNA degradation. Speeds in this video are for demonstration purposes only.
Supplementary Fig. S1. Bioanalyzer traces showing the mRNA integrity of additional biological replicates from: total leaf (A, B), mock protoplast stress control (C, D), M protoplasts (E, F), M leaf roll (G, H), and BS strands (I, J). From left to right, the peaks represent the Bioanalyzer gel marker at 25 nt, then tRNA/sRNAs (broad multipeak), followed by a typical pattern for sorghum of four peaks representing rRNAs. The peaks were well defined, indicative of high-quality RNA that has not been degraded. FU, fluorescence units.
We thank the C4 Rice Consortium for discussions, the International Rice Research Institute for sorghum seed, and the Bill and Melinda Gates Foundation for funding the C4 Rice Project.