Next-generation sequencing (NGS) offers numerous research opportunities to the plant systematics and evolutionary biology communities (e.g., Godden et al., 2012; Strickler et al., 2012; Soltis et al., 2013). To apply NGS to whole transcriptome sequencing, high-quality RNA must be reliably obtained, often from diverse taxa. However, high levels of compounds such as flavonoids, tannins, waxes, and other secondary metabolites found in the tissues of aromatic, woody, and aquatic plants can make it difficult to extract RNA in sufficient quantity and quality for NGS. To circumvent these problems, many different methods have been developed for RNA extraction from plant tissues (e.g., Johnson et al., 2012; Yockteng et al., 2013; Zhu et al., 2013).

During our involvement in the 1KP project (One Thousand Plants Initiative, see  www.onekp.com), we developed two RNA isolation protocols that were used to obtain high-quality and -quantity RNA from a diverse set of plant species. Here we present two protocols (Options 1 and 2) in detail and compare their success rate with a third protocol (Option 3) that is standardly used for many species (e.g., Buggs et al., 2009; Johnson et al., 2012). An earlier version of our Option 1 can be found in the appendix of Johnson et al. (2012) in Protocol 14, but is presented here with further refinements that improve RNA quality and quantity. Option 2 is presented for the first time.

In this manuscript, we quantify RNA extraction success among the methods with Bioanalyzer metrics, not transcriptome sequencing success as was done in Johnson et al. (2012), by quantifying number of scaffolds and resulting assemblies. An extraction was considered successful if it met the quantity and quality set by the 1KP consortium agreement: ≥30 µg of total RNA, RNA integrity number (RIN) higher than 5, and rRNA ratio (26S/18S) greater than 1. Although Johnson et al. (2012) conclude that transcriptome assemblies were mostly dependent on the NGS sequencing platform, RNA quality (RIN and optical density [OD] ratios) and quantity also had significant effects. The two preferred protocols presented here result in RNA isolations of the quality and quantity required for transcriptome sequencing in a wide range of plant taxa.

METHODS AND RESULTS

Selection of material —The plants selected for the 1KP project were meant to circumscribe the entire green plant clade (Viridiplantae), including chlorophyte algae (see Johnson et al., 2012; Matasci et al., 2014). RNA from the majority of taxa in the 1KP project were extracted with protocols outlined in Johnson et al. (2012). The protocols presented here were developed to handle samples for which those methods had failed to provide good-quality RNA extractions. Most are angiosperms and many have characteristics (aromatic, woody, aquatic) known to present challenges to RNA extraction (Appendix 1).

Fig. 1.

Flow chart describing the iterative process to successfully obtain high-quality and -quantity RNA using Option 1 and Option 2. (Option 3 is excluded, as it is not recommended.)

Fig. 2.

Examples of Agilent 2100 Bioanalyzer spectra of total RNA showing improvement with Options 1 and 2 compared to Option 3. Each graph shows the intensity of the peaks of the ribosomal RNA subunits: nuclear large-28S, small-18S, cytoplasmic, mitochondrial, and chloroplastic (smaller subunits). The electrophoretic gel for each sample is shown to the right indicating the subunit bands or degradation (i.e., smear). nt = number of estimated nucleotides based on ladder; FU = fluorescence unit (i.e., intensity of peak). (A) Degraded ribosomal RNA subunits of Canella winterana (L.) Gaertn. extracted with Option 3 that resulted in an estimation of 27 µg of RNA, but the subunits are degraded. (B) A second example using Option 3, Muntingia calabura L., also shows an inflated quantity reading with degraded subunits. (C) Canella winterana, extracted with Option 1, indicating peaks for intact ribosomal RNA subunits. (D) Muntingia calabura extracted with Option 2.

Overview of protocols —Option 1 is the manufacturer's protocol for RNA extraction with TRIzol (Ambion, Life Technologies, Carlsbad, California, USA), with the addition of sarkosyl to the extraction solution. Option 2 combines traditional cetyltrimethylammonium bromide (CTAB) extraction (Doyle and Doyle, 1987) followed by the use of TRIzol with sarkosyl (i.e., CTAB followed by Option 1). Both Option 1 and 2 used the TURBO DNA-free kit (Ambion) for DNA digestion, which proved to be superior for RNA stability compared to the on-column digestion (Appendix 2 for taxa that degraded [gray cells]). Option 3 combines the CTAB extraction method followed by the use of half of the QIAGEN RNeasy Plant Mini Kit for on-column DNA digestion (QIAGEN, Valencia, California, USA). Option 3 was the protocol used during the initial phase of the 1KP project, and in other projects in our laboratory (e.g., Buggs et al., 2009), but after multiple failures, Options 1 and 2 were developed to deal with difficult plant materials. These methods are briefly outlined below, and detailed bench-ready protocols can be found in Appendices 3–6). A flow chart can be found in Fig. 1. Option 1 is recommended as a starting point when working with unstudied taxa because it has fewer steps. If this fails, researchers should then move to Option 2. The protocols here can be adapted to processing few samples or relatively high-throughput extractions of 12 to 24 samples at a time. The use of 2-mL microcentrifuge tubes and an automatic shaker for tissue pulverization is ideal (see Appendix 3 for details on tissue collection and processing). These methods can still be used with tissue pulverization in a mortar and pestle if an automatic shaker is not available or the tissue will not pulverize by beads alone, then transferred to the 2-mL tubes (i.e., some Poaceae and succulent species).

The addition of sarkosyl in the TRIzol extraction step in either Option 1 or 2 was integral for improving quality and yield in some taxa, but was not required for all taxa (data not shown). The addition of sarkosyl to lysis solutions is not new; it has been used in both plant and animal nucleic acid and protein isolations since the 1970s (Kingston, 2010) and is still used today for various applications (Huang et al., 2012). It is commonly recommended as an addition to isolation buffers containing guanidine (Kingston, 2010), an ingredient in the TRIzol solution. The addition of sarkosyl to the Ambion TRIzol extraction kit was used to deal with plants with high amounts of organic compounds or other complexities (typically indicated by a brown color in the aqueous layer). After centrifugation, the lightweight sarkosyl layer rested at the interface between the upper aqueous solution and the lower organic layer. Following TRIzol and sarkosyl extraction, a 100% chloroform extraction was always done once, sometimes twice, depending on whether a whitish interface was present between the organic and the aqueous layers.

RNA precipitation and pelleting —In all of the options, the method of RNA precipitation was similar (in Option 3, 5 M NaCl was also added). The tube containing the aqueous phase was filled with 100% isopropanol at room temperature, gently inverted, then incubated at −20°C for 10 min for Options 1 and 2, and 20 min for Option 3. Overnight incubation did not increase the yield, and the best RNA was obtained when no cloudiness or white precipitate was formed after addition of isopropanol. After incubation, in Options 1 and 2 the samples were centrifuged at 4°C for 20 min to pellet the RNA, typically producing pellets of 0.25–1 cm long (the expected size as indicated by Ambion). The RNA was not pelleted in Option 3. The pellet was washed with 75% ethanol and airdried for no longer than 10 min (otherwise resuspension was difficult), and resuspended in RNase-free water by incubation at 37°C on an orbital shaker for 10 min. If the pellet was not dissolved by the end of this incubation period, in most cases, it was not worth proceeding, and the process was started from the beginning with a new sample (Fig. 1). For Option 3, if the NanoDrop (Thermo Scientific, Waltham, Massachusetts, USA) gave poor results (data not shown), the sample was also discarded. Therefore, these failed attempts do not have Bioanalyzer metrics as others listed in Appendix 2.

Extraction success: Quality, quantity, and stability —Extraction success was measured from Bioanalyzer metrics for the quality and quantity of RNA isolated (Agilent Technologies, Santa Clara, California, USA) (Fig. 2, Appendix 2, and  Appendix S1 (apps.1400105_s1.pdf) for those that were analyzed with the Bioanalyzer). Over the course of this study, 382 separate RNA extractions were attempted from leaf tissues, of which 138 were successful (Table 1). Many of the failures were due to the iterative process of developing the protocols presented here and multiple unsuccessful attempts with Option 3 before alternatives were sought. Table 1 summarizes the levels of extraction success of each method during this process, and the number of taxa for which RNA was successfully extracted in the end. Using our methods, we finally successfully extracted RNA from 77 unique taxa of 91 attempted (85%). Option 1 was tested on 74 of the 91 taxa in this data set, Option 2 was tested on 41 of the 91 taxa, Option 3 was used on 68 out of 91 taxa. Not all methods were tried on all taxa, as the goal of our work was to obtain pure RNA for the 1KP project as quickly as possible, not to test all methods across all taxa. Options 1 and 2 often succeeded with material that failed to yield RNA with Option 3 (Appendix 2). Nine taxa were unsuccessful in extraction with any of the three methods, perhaps due to the chemical composition of the plant, or perhaps tissue quality (Appendix 2). Two taxa yielded good-quality and -quantity RNA, but the library construction failed at BGI (Shenzen, China) (for details see Johnson et al. [2012] and Appendix 2). Most of the extractions that were partially successful with Option 3 were of low RNA quality, and often degraded (gray cells in Appendix 2), but the nondegraded isolations from Option 3 were sometimes ultimately suitable for transcriptome sequencing (Appendix 2).

Table 1.

Success of RNA extraction for each method. Success is defined by the Bioanalyzer results and final quantity of pure RNA that was sequenced. Linal concentration was estimated from the Bioanalyzer and the known volume of the final extraction. Some samples were extracted multiple times and pooled (i.e., “repeats”). Note that most samples were initially tried with Option 3; those that failed were then attempted with Options 1 and 2. Thus, not all samples were extracted with all options.

Bioanalyzer metrics for the total quantity of RNA isolated, RIN, and OD ratio were not the same across the different extraction options (Fig. 3). Given that the data set was compiled as trial and error, but not formally designed for statistical measures, only the distribution of the data is shown ( Appendix S1 (apps.1400105_s1.pdf)). For the purposes of the 1KP project, 30 µg of total RNA was requested for each taxon from BGI for a full transcriptome sequence (Johnson et al., 2012). However, today, 30 µg of RNA is not necessarily needed to generate transcriptomes. For the 18 taxa successfully extracted with Option 3 in this study, multiple extractions (2–4) had to be done and pooled to reach the desired amount (using ∼100 mg of frozen tissue per extraction). Out of the successful extractions made with Options 1 and 2, only 33% of the taxa gave less than 30 µg in the first extraction attempt, and at most two extractions were needed to obtain the quantity desired. The majority of the time for Option 2 (67% of the extractions), the quantity averaged 77 µg of RNA from only 100 mg of tissue in the first extraction attempt, so no pooling was needed. The maximum amount was 250 µg of RNA in one extraction.

To NanoDrop or not —NanoDrop or similar equipment for RNA (or DNA) concentration measures can be used to obtain an estimate for total amount of nucleic acids. The NanoDrop cannot detect the presence of intact ribosomal-RNA subunits, and therefore is not able to detect if the RNA is degraded. Additionally, the NanoDrop cannot reliably measure the concentration in impure samples from chemically complex plant extractions, giving an inflated reading. However, as most laboratories do not have easy access or funds to use a Bioanalyzer, the use of the NanoDrop (or Qubit, Life Technologies) is still a laboratory necessity. We advise that using spectrophotomic methods such as NanoDrop or fluorescent-dye methods (e.g., Qubit) are rough and easy ways to assess if the samples are “on track,” but should not be used as an absolute measure for concentration and cannot give a reading of the quality of the ribosomal RNA subunits to prepare for transcriptome sequencing. Low-quality RNA can still be sequenced, but the resulting sequences will be poor, and only determined so after the transcriptome is assembled (R. Cronn, personal communication).

RNA stability: Storing and shipping pure RNA —As with most molecular biological materials, a freeze/thaw process can damage a sample. We found that RNA samples that were not digested immediately with DNase could be stored at 4°C for 2 to 3 d before digestion with no apparent change in quality (data not shown). Once the DNA was digested, a 3-µL aliquot was run on the Agilent 2100 Bioanalyzer, and the remaining samples were placed at −80°C for storage until it was mailed. The pure RNA was sent in the mail for library construction and sequencing after drying down onto specially coated tubes (i.e., Gen Vault, now renamed as GenTegra; IntegenX, Pleasanton, California, USA) that inhibit RNase activity and stabilize the RNA at room temperature. In general, samples extracted with Options 1 and 2 were more likely to make it through storage, Gen Vault shipping, and resuspension at BGI than were samples extracted with Option 3. The Option 3 samples were often degraded beyond use upon resuspension at the BGI sequencing facility (Fig. 2A, B and Appendix 2 [gray cells]).

Fig. 3.

Distribution of RNA quantity and quality. Box plots illustrate (A) total micrograms of RNA isolated by each method, (B) RNA integrity number resulting from each method, and (C) rRNA (OD) ratio resulting from each method. The error bars and the boxes indicate the quantiles of the data (JMP Pro 11.0, SAS Institute).

CONCLUSIONS

Use of the TRIzol supplemented with sarkosyl followed by removal of DNA with the TURBO DNA-free kit (Option 1) is an efficient and effective means of extracting RNA from a diverse array of plants, especially those that are woody, aromatic, or aquatic. With the addition of the traditional CTAB method prior to the TRIzol (Option 2), even the most stubborn taxa were mostly successful and gave consistent RNA quality measures. Option 3, which has been used successfully in many laboratories (including our own, e.g., Buggs et al., 2009; Johnson et al., 2012) for RNA isolation, is not the most efficient or robust method for obtaining high-quantity and -quality RNA in transcriptomics across the Embryophyta. Despite the success of the protocols described here, our methods were not successful for some plants that contain high amounts of mucilage, such as Opuntia sp.