An optimised protocol for isolation of RNA through laser capture microdissection of leaf material

Laser Capture Microdissection is a powerful tool that allows thin slices of specific cells types to be separated from one another. However, the most commonly used protocol, which involves embedding tissue in paraffin wax, results in severely degraded RNA. Yields from low abundance cell types of leaves are particularly compromised. We reasoned that the relatively high temperature used for sample embedding, and aqueous conditions associated with sample preparation prior to microdissection contribute to RNA degradation. Here we describe an optimized procedure to limit RNA degradation that is based on the use of low melting point wax as well as modifications to sample preparation prior to dissection, and isolation of paradermal, rather than transverse sections. Using this approach high quality RNA suitable for down-stream applications such as quantitative reverse transcriptase polymerase chain reactions or RNA-sequencing is recovered from microdissected bundle sheath strands and mesophyll cells of leaf tissue.


Introduction
Multicellularity has evolved repeatedly across the tree of life, and is a defining feature of land 32 plants. Not only does multicellularity solve size and lifespan limitations caused by diffusion and 33 ageing of individual cells respectively, it also allows increased complexity through the 34 differentiation of cell types that become specialized for particular functions. Thus, to understand how 35 a multicellular organism is built and then its structures maintained, analysis of its constituent cell 36 types is desirable. 37 Various methods have been developed to isolate and study specific cell types in plants. In some 38 cases, different tissue types can be separated relatively easily. For instance, in some plant species 39 with C4 leaf anatomy bundle sheath strands can be separated from the adjoining mesophyll by  However, it is not always possible to generate transgenic plants, or identify a promoter that drives 51 strong expression in the cell type being studied. In the case of leaves, the process of protoplasting 52 is known to generate a significant stress response and de-differentiation (Sawers et al. 2007) such 53 that this approach is compromised if the aim is to better understand photosynthesis. In principle, 54 laser capture microdissection provides an orthogonal method to these approaches, enabling highly 55 purified cell populations to be harvested without requiring the generation of transgenic lines (Nelson 56 Although histological details are well preserved using this method, considerable RNA degradation 69 can take place (Gomez et al. 2009;Roux et al. 2018). We found this to be a particular problem with 70 low abundance cell types of leaves. To address this issue, we sought to modify existing protocols to 71 increase RNA yield and integrity during sample processing as well as the laser capture 72 microdissection procedure itself. By adopting a low-melting point wax, as well as modifying sample 73 preparation prior to microdissection and isolation of paradermal rather than transverse sections, we 74 provide a simple and robust method to allow high quality RNA to be obtained from specific cells of 75 leaves that are not accessible using existing methodologies.

Plant materials and growth 78
Seeds of Arabidopsis thaliana ecotype Columbia were sown in 1:1 mixture of Levington M3 high 79 nutrient compost and Sinclair fine Vermiculite soil, vernalized for 3 days and then transferred to a 80 controlled environment room set at 22°C with a photoperiod of 16h light and 8h dark, a photon flux 81 density of 200 μmol photons m -2 s -1 . Rice (Oryza sativa ssp. indica IR64) was germinated and grown 82 in 1:1 mixture of top soil and sand for two weeks in a controlled environment growth room set at 83 28 °C day 25 °C night, a relative humidity of 60%, a photoperiod of 12h light and 12h dark, and a 84 photon flux density of 300 μmol m −2 s −1 . 85 86

Sample preparation 87
To evaluate the effect of fixative on RNA integrity, fully expanded leaves of Arabidopsis or rice 88 were sampled and fixed in ice-cold 100% (v/v) acetone or Farmer's fixative (75% (v/v) ethanol, 25% 89 (v/v) acetic acid) for 2 hours and 4 hours on ice, respectively, prior to immediate RNA extraction. To 90 conduct laser capture microdissection rice leaves were cut into 5-8 mm pieces with RNAZap treated 91 scissors, and fixed under vacuum for two 10 minutes periods in ice-cold 100% (v/v) acetone and 92 then left with gentle stirring for 3 hours. Arabidopsis leaves were treated in the same way, but to 93 maintain tissue structure not subjected to vacuum infiltration. Leaf tissue was then dehydrated 94 through an ice-cold series of 70%, 85%, 95% and 100% (v/v) ethanol for 1 hour each. Samples were 95 incubated in 100% (v/v) ethanol overnight at 4°C, prior to being placed in 25%, 50%, 75% and then 96 100% Steedman's wax at 37°C for 2 hours. This final solution of 100% Steedman's wax was replaced 97 twice every 2 hours. Tissue was embedded in a 9-cm petri-dish, and after wax had solidified it was 98 cut into 1 cm 3 blocks and stored in 50 ml falcon tubes with self-indicating silica gel at -80°C. 99 Steedman's wax was prepared as described (Vitha et al. 2000), 1000 g polyethylene glycol 400 100 distearate and 111 g 1-hexadecanol were melted at 60 °C and mixed thoroughly, prior to being 101 aliquoted into 50 ml RNase-free Falcon tubes. Tissue embedded in paraffin wax was also processed 102 of cytosolic and chloroplastic ribosomal RNA peaks, and quantitatively using the common metrics of the 25S to 18S ribosomal RNA ratio, and the RNA Integrity Number (RIN) (Schroeder et al. 2006). 115 116

Sectioning and laser capture microdissection 117
Paradermal sections were prepared using a microtome. Paraffin embedded sections were placed 118 onto a dry polyethylene naphthalate (PEN) membrane slide (Arcturus) and then floated on diethyl 119 pyrocarbonate (DEPC)-treated water at 42°C to expand the sections and ensure they were flat. 120 Water was then removed and slides dried at 42°C for 20 min. Steedman's wax embedded sections 121 were similarly expanded on DEPC-treated water at room temperature on a PEN membrane slide, 122 before the slide was dried using tissue paper at room temperature. Before laser capture 123 microdissection, paraffin wax was removed by incubating slides in 100% (v/v) Histoclear for 5 124 minutes, whilst Steedman's wax was removed by incubating slides in 100% (v/v) acetone for 1 125 minute at room temperature. Laser capture microdissection was performed on an Arcturus Laser 126 Capture Microdissection system using Capsure macro caps to collect bundle sheath strands and 127 mesophyll cells.

RNA integrity is maintained with non-crosslinking fixatives 130
To limit RNA degradation, tissue fixation needs to be rapid. Compared with cross-linking fixatives, 131 precipitative fixatives such as acetone and Farmer's fixative have commonly been used for laser 132 capture microdissection sample preparation because they retain good histological detail as well as 133 reasonable RNA yields (Kerk et al. 2003). However, to our knowledge, a quantitative analysis of the 134 effect of these fixatives on RNA integrity has not been reported. We therefore fixed Arabidopsis 135 leaves using acetone or Farmer's fixative on ice for 2 and 4 hours, extracted RNA and found that 136 yield and integrity were similar after 2 hours and 4 hours fixation using either fixative (Supplemental 137 Figure 1). This suggests that RNA was preserved well by each of these precipitative fixatives. 138 However, it was noticeable that leaf tissue sank more rapidly in acetone than in Farmer's fixative, 139 suggesting a faster penetration into leaf tissue. We therefore subsequently used acetone for sample 140 preparation. 141

RNA integrity is improved after Steedman's wax infiltration 143
The most commonly used embedding medium for laser capture microdissection studies of plants 144 is paraffin wax, presumably due to its ease of handling and good preservation of histological details. 145 Therefore, initially we embedded rice leaves using paraffin wax and transverse sections were 146 prepared to isolate mesophyll and bundle sheath strands ( Figure 1A&B). However, even when a cap 147 was fully loaded with tissue, which takes around 2 hours of continuous microdissection, very low 148 quantities of RNA were obtained from bundle sheath strands ( Figure 1E). We therefore tested 149 whether sampling from paradermal sections ( Figure 1C&D) improved yields. About ten paradermal 150 sections could be prepared from one leaf, and in approximately one hour, nearly all of the bundle 151 sheath strands in these sections could be dissected. This yielded significantly greater amounts of 152 RNA ( Figure 1E). Thus, paradermal sectioning resulted in more tissue being captured per slide 153 ( Figure 1A and 1C), was about twice as quick, and so reduces the risk of RNA degradation. However, 154 consistent with reports on other tissues (Roux et al. 2018), we also found that RNA quality from 155 paraffin embedded tissue was low. Since the fixation process itself appeared not to have a 156 deleterious effect on RNA quality (Supplemental Figure 1), we reasoned that losses in RNA integrity representing the cytosolic and chloroplastic ribosomal RNAs (Figure 2A-D). Both the RNA Integrity Number (RIN) and ribosomal 28S:18S RNA ratio were statistically significantly higher when 168 Steedman's wax was used to embed Arabidopsis leaves ( Figure 2E&F). Although the RIN values 169 from rice leaves were not increased significantly, it was noticeable that the ribosomal RNA peaks 170 were more defined, and that the ratio of the cytosolic ribosomal 25S to 18S RNAs was significantly 171 higher when Steedman's embedding medium was used ( Figure 2E&F). Taken together, these 172 findings indicate that RNA recovered from leaves embedded in Steedman's wax was of higher quality 173 than that isolated after embedding in paraffin wax. Whilst we are not able rule out other effects, the 174 simplest explanation is that the lower temperature used during infiltration of Steedman's wax leads 175 to less RNA damage. 176 To determine whether the duration of infiltration in Steedman's wax affects RNA quality, we 177 extracted RNA from rice leaf tissue after 1 hour, 3 hours, or 6 hours of infiltration in Steedman's wax. 178 Both the RIN value and ribosomal 28S to 18S RNA ratio remained essentially unchanged over this 179 time-course, suggesting that in species that require longer infiltrations for good sectioning, extending 180 the infiltration time in wax could be used without compromising RNA quality (Supplemental Figure  181 2). 182 183 A procedure to minimize RNA degradation during slide preparation 184 Subsequent to wax embedding, but prior to laser capture microdissection, there are further 185 opportunities for RNA to be degraded. For example, during slide preparation, sections are typically 186 expanded by floating on warm RNase-free water at 42°C to ensure they lie flat on the microscope 187 slides. Water is then removed and samples dried at 42°C for 20-30 minutes. Consistent with RNA 188 degradation during this process, after slide preparation from paraffin-embedded rice tissue the 189 cytosolic and chloroplastic ribosomal RNA peaks were less defined, yields were lower, and the 28S 190 to 18S ribosomal RNA ratio was lower compared with to that isolated from freshly-cut sections 191 ( Figure 3A-D). In contrast, after slide preparation using Steedman's wax for embedding, the cytosolic 192 and chloroplastic ribosomal RNA peaks were clearly defined, and the 28S to 18S ribosomal RNA 193 ratio was maintained ( Figure 3E-3H). We also found that sections embedded in Steedman's 194 expanded immediately on water at room temperature (20°C), and that the water could be removed 195 rapidly by absorption onto soft tissue paper. Adhesion of thin sections to the slide was enhanced by 196 providing gentle pressure with dry tissue paper (Supplemental Figure 3). This rapid process avoids 197 the prolonged exposure of sections to higher temperatures and so preserves RNA integrity during 198 slide preparation. Examination of tissue integrity using light microscopy after embedding in 199 Steedman's wax showed that histological details were as good as those seen after embedding in 200 paraffin wax (Supplemental Figure 4). 201

203
To assess the combined impact of the modifications documented above on the final RNA quality 206 obtained from laser capture microdissection, we compared RNA quality from microdissected tissues 207 embedded in either paraffin or Steedman's wax. For this purpose, bundle sheath strands and 208 mesophyll cell sections were captured from both Arabidopsis and rice leaf tissues. RNA isolated by 209 laser capture microdissection from paraffin embedded sections of either species showed either no 210 clear, or compromised ribosomal RNA peaks ( Figure 4A-D). This was particularly noticeable for the 211 bundle sheath strands. Moreover, the baseline was high ( Figure 4A-D) indicating that the RNA was 212 severely degraded, and quantitation confirmed these qualitative assessments ( Figure 4I-L). In 213 contrast, RNA isolated from either Arabidopsis or rice tissue embedded in Steedman's wax showed 214 less elevated baselines, more defined ribosomal RNA peaks ( Figure 4E-H), and higher RIN values 215 and 28S to 18S ribosomal RNA ratios ( Figure 4I-L). 216 With the advances and reduced cost of next-generation sequencing, RNA-SEQ has become a 217 common tool for profiling transcript abundance. However, a high quality RNA input is important for 218 reliable and reproducible results. For example, it has been reported that RNA degradation can have 219 a broad effect on quality of RNA-SEQ data, including 3' bias in read coverage, quantitation of 220 transcript abundance, increased variation between replicates, and reductions in library complexity 221 wax was used during sample preparation. In contrast, our optimised sample preparation method 226 using low-melting temperature wax, led to most RNA isolated after microdissection having RIN 227 values >5. We therefore conclude that these simple modifications allow tissue to be prepared such 228 that different cell types in the leaf can still be identified, and that the quality of the RNA available for 229 sampling is improved. We anticipate that this approach will greatly facilitate the analysis of gene