Molecular Ecology Resources (2011)

doi: 10.1111/j.1755-0998.2011.03010.x

Next-generation sequencing of transcriptomes: a guide to RNA isolation in nonmodel animals PHILIPPE GAYRAL, LUCY WEINERT*, YLENIA CHIARI, GEORGIA TSAGKOGEORGA, MARION BALLENGHIEN and N I C O L A S G A L T I E R Institut des Sciences de l’Evolution, CNRS UMR 5554, Universite´ Montpellier 2, Place E. Bataillon, 34095 Montpellier, France

Abstract Next Generation Sequencing technologies (NGS) are rapidly invading many evolutionary and ecological fields, such as phylogenomics, molecular evolution, population genomics and molecular ecology. Among the potential targets of NGS is transcriptome sequencing, a fast and relatively cheap way to generate massive amounts of coding sequence data, offering promising perspectives for the analysis of molecular diversity in the wild. A number of molecular ecology research groups therefore may switch from DNA-based to RNA-based typing in the near future. Sample preparation from natural populations, however, requires specific care and protocols when RNA is the target. Furthermore, NGS sequencing of transcriptome requires high amount of good-quality RNA. Here we present the results of RNA extraction experiments from various samples of 39 animal species caught in the wild. We compared tissue preparation and storage conditions, evaluated and improved standard RNA extraction protocols, and achieved RNA yield and quality suitable for NGS in all cases. We derive general guidelines for the production of ready-to-sequence RNA in nonmodel animals sampled in the field. Keywords: high-throughput sequencing, mRNA, NGS, nonmodel organisms, RNA extraction, transcriptomics Received 18 November 2010; revision received 27 January 2011; accepted 4 February 2011

Introduction Next Generation Sequencing technologies (NGS) make genomic data available for everyone (Hudson 2008). So far, genome projects have been restricted to a relatively small number of model organisms, handled by large consortiums. Thanks to 454 sequencing technology, Illumina or SOLiD System, and the several orders of magnitude of reduction in sequencing cost that these technologies bring, a single research group can now explore the population genomics of its pet species, which is a new situation for many laboratories. Among the potential targets of NGS in biodiversity research, the transcriptome is a promising candidate (Vera et al. 2008). Transcriptomes are typically 10–100 times smaller than genomes and are therefore 10–100 times cheaper to sequence, for a given level of coverage. Transcriptomes include the coding sequence and some of the regulatory regions of genes (3¢ and 5¢ UTR), which presumably represent a large fraction of the functional Correspondence: Philippe Gayral, Fax: +33 4 67 14 36 10; E-mail: [email protected] *Present address: Department of Infectious Disease Epidemiology, MRC Centre for Outbreak Analysis and Modelling, Imperial College, Faculty of Medicine, St Mary’s Campus, Norfolk Place, London W2 1PG, UK.

 2011 Blackwell Publishing Ltd

content of a genome. From an evolutionary genomic perspective, transcriptomes provide hundreds of singlenucleotide polymorphism (SNPs) scattered throughout the genome, with applications in the study of population structure, history and speciation (Bers et al. 2010; Renaut et al. 2010). From an evolutionary genomic point of view, they offer large numbers of orthologous coding sequences across species, useful for phylogeny and studies of molecular adaptation (Elmer et al. 2010; Ku¨nstner et al. 2010). Finally, transcriptome data contain information about the expression levels of genes in distinct tissues, which is of primary interest to functional ecological and evolutionary studies (Renaut et al. 2010; Wolf et al. 2010). Targeting transcriptomes, however, requires isolating total RNA from living material or properly preserved samples. RNA extraction is routinely achieved for model organisms in laboratory conditions (reviewed in Vomelova´ et al. 2009). Standard protocols, however, are not clearly suitable for population or evolutionary studies of nonmodel species, when animals are to be captured in the field. Unlike DNA, RNA is prone to very rapid degradation after animal death or tissue sampling, and therefore specific procedures of sample preservation before RNA extraction are mandatory (Sambrook et al. 1989). Furthermore, optimal sampling, preservation and extraction techniques can vary depending on the size and

2 P. GAYRAL ET AL. nature of the analysed organisms or tissues. Compared to other downstream applications, NGS also require a larger amount of good-quality RNA, especially for 454 sequencing which requires five times more RNA than does Illumina sequencing. Many ecological and population genetics laboratories will switch from DNA-based to RNA-based molecular typing in the near future and might therefore be in search of appropriate protocols. Here we report on RNA extraction experiments carried on a taxonomically and histologically diversified set of samples in animals. We explored, modified and validated existing protocols in a variety of conditions, and derived general guidelines for RNA isolation singly or simultaneously with DNA isolation, in nonmodel animals.

Methods Species used in the study We sampled a variety of tissues in 39 species from six distinct animal phyla (Table 1), including both invertebrate and vertebrate species. Protostomes were represented by four groups of insects—butterflies, ants, bees and termites—one crustacean, one nematode, three molluscs—gastropods, cephalopods and bivalves—and one nemertian. Deuterostomes were represented by mammals, birds, reptiles, dipnoi and one tunicate (Table 1). RNA and DNA extractions were performed on a broad range of tissues: whole organisms in arthropods, nematodes and cephalopods, blood in turtles and tit (see also Chiari & Galtier 2011), muscle in penguin, jaw (bone, teeth and muscle) in caiman and lungfish (dipnoi), and various separately treated tissues in oyster, tunicate, hare and lizard. A total of 16 tissues were tested in this study (see details in Table 1). Samples were flash-frozen or preserved in appropriate buffer. RNA was extracted using guanidinium thiocyanate–phenol–chloroform extraction (referred as GTPC extraction in the text, with TRIzol or TRI Reagent BD), or silica matrix (SM, RNeasy columns), or a combination of the two approaches (see Methods). The quality of these extractions was assessed using spectrophotometry (NanoDrop), agarose gel electrophoresis and capillary electrophoresis (Agilent), and compared across sample preparation procedures and RNA extraction methods.

Sample preparation Microbiology and aseptic techniques were followed to create and maintain an RNase-free environment and to avoid RNase contamination, during or after extraction. Shell and digestive tube were removed for gastropods. Living animals were brought back to the laboratory,

frozen in liquid nitrogen and stored at )80 C or directly processed for RNA extraction. When laboratory facilities were not immediately available, samples were immediately stabilized in TRI Reagent BD (Molecular Research Center, Cincinnati, OH, USA) for blood samples (see Chiari & Galtier 2011), and in RNAlater for all other tissues following manufacturer’s instructions. Whole organisms and hard tissues were cut into pieces <5 mm thick, or kept entire if smaller than 5 mm. Unless specified, animal tissues were finely ground for efficient cell lysis in liquid nitrogen using a ceramic mortar and pestle. Because field conditions are not always compatible with the manufacturer’s recommended storage, further testing and improvement were thus done beforehand on several species. We tested whether dissecting hare tissues (muscle, heart, lung, spleen, kidney and liver from Lepus granatensis, L. americanus, L. timidus) in thin pieces (slices of approximately 3 mm thick) offered a better stabilization by RNAlater compared to the sizes recommended by the manufacturer (approximately 5 mm thick), when RNA extractions are performed with a constant weight of tissue. Furthermore, we compared whether the presence of the shell in small fresh water gastropods and whether the effect of long storage at room temperature could interfere with RNA integrity by submerging the animals directly in RNAlater and keeping them for 3 days at room temperature. Similarly, for insects we tested whether RNA integrity is affected by cutting the animals into small pieces or by immersing them directly in the stabilization buffer. We then tested whether insect segments could be efficiently stabilized in RNAlater for 10 days at room temperature (imitating field conditions), i.e., a longer period than what is recommended by RNAlater manufacturer (7 days at room temperature), and without the need of a refrigerator (efficient stabilization at 4 C for up to 28 days, according to manufacturer). Unless specified (tests on temperature of incubation), recommendations of RNAlater manufacturer for optimal stabilization were followed (initial incubation overnight at 4 C before freezing the sample). For butterflies, we also compared the latter to immediately freezing the animals in liquid nitrogen. We finally tested whether butterfly tissue grinding using a plastic pestle in a 1.5-mL polypropylene tube in TRIzol reagent could provide an effective means for efficient RNA stabilization and a subsequent RNA extraction using TRIzol reagent (homogenization was here performed by room temperature grinding using ceramic pestles and mortars).

RNA extractions We tested two distinct RNA extractions methods that are currently widely used (Tan & Yiap 2009). RNA

 2011 Blackwell Publishing Ltd

 2011 Blackwell Publishing Ltd

Melitaea cinxia, M. parthenoides, Euphydryas desfontainii, Pieris spp. Galleria mellonella

Species

Vertebrata, Mammalia

Mollusca, Cephalopoda Nemertea, Anopla

Insecta, Hymenoptera

Hare

F

Ribbon worm L

Lineus ruber, L. longissimus, L. lacteus Lepus granatensis, L. americanus, L. timidus

L

R

Ni

Ni

Adult

20 mg 20 mg 20 mg

Kidney Liver

75 mg

Spleen

Embryo (latest develop. stage) Sub-adult Whole

20 mg 150 mg 150 mg 150 mg 150 mg 230 mg

Whole body§ Mantle Gill Digestive gland Abductor muscle Whole Sub-adult Adult

Ni RG

L LF

Sea squid

6 mg

Whole body§

Adult

Ni R

LF

Sepia officinalis

20 mg

Whole body§

Adult

Ni

L

112 mg 58 mg 6 mg 75 lL

Whole body Whole body Whole body

Imago-worker Imago-nymph Adult

44 mg 4 mg

60 mg

75 mg

Pellet of clones or descents

G

L

Whole body Whole body

Whole body

Imago Imago-worker Imago-worker

Whole body

Imago

Tissue

Mean weight ⁄ volume

Adult

Ni Ni Ni

L L L

Ni Ni Ni

L

Greater wax moth Ant Ant

Ni R G

L L

LF

Development Sampling* Stabilization† stage

Butterfly

Common name

Messor barbarus Crematogaster scutellaris Halictus scabiosae Bee Insecta, Isoptera Reticulitermes grassei Termite Crustacea, Artemia franciscana Brine shrimp Branchiopoda Nematoda, Caenorhabditis Soil Chromadorea brenneri, C. sp10, roundworm C. sp16, Oscheius tipulae, O. sp2, O. sp3 Mollusca, Physa acuta, Physa Gastropoda P. hendersoni Galba truncatula, Limneid sails Lymnaea sp. Radix peregra Limneid sails Mollusca, Ostrea edulis, Oyster Bivalvia O. chilensis, Ostreola stentina

Insecta, Lepidoptera

Taxonomic group

Table 1 Comparison of RNA and DNA extraction protocols in nonmodel animal species

GTPC (3) SM (20) GTPC (5) SM (22) GTPC (2)

Combined (5)

GTPC (6) SM (19) GTPC (8) SM (10) SM (5) Combined (28) Combined (28) Combined (29) Combined (31) SM (1)

GTPC (6) GTPC (7) Combined (5) GTPC (1) GTPC (2) GTPC (2) SM (3) Combined (34)

Combined (2)

GTPC (11) SM (4)

Extraction method (No. extractions)

179.8–8.99 39.5–1.98 58.6–2.93 25.3–1.27 198.7–9.94

19.8–0.26

72.9–3.65 31.4–1.57 13.1–2.18 18.2–3.03 26.5–1.33 23.5–0.16 50.5–0.34 16.8–0.11 8.4–0.06 56.8–0.25

9.9–0.23 3.2–0.80 0.8–0.20 47.4–0.42 1.8–0.03 9.9–1.65 3.5–0.58 58.1–0.77

30.4–0.51

121.2–1.62 1.1–0.01

1.90–2.19 2.03–2.02 1.98–2.00 2.04–2.10 1.99–2.00

2.14–1.55

1.88–1.26 2.12–2.08 1.88–1.26 2.12–2.08 2.10–2.03 2.06–2.00 2.08–2.07 2.05–1.60 1.96–1.27 2.17–2.14

1.96–0.84 1.94–0.50 1.85–2.54 2.16–1.46 2.13–2.37

1.73–0.52 1.59–0.23

2.10–1.97

1.84–0.99 2.25–1.07

RNA yield‡ per extraction— Mean RNA per lg (or ll) 260 ⁄ 280-260 ⁄ 230 tissue Abs. ratio

RNA EXTRACTION FOR NEXT-GEN MRNA SEQUENCING 3

Protopterus annectens Ciona intestinalis

Podarcis liolepis

African lungfish Vase tunicate

Ni Ni R

LF

G

Ni

G

R

G

Stabilization†

L

L

L

Spectacled caiman Catalonian wall lizard

F F

King penguin

Aptenodytes patagonicus Emys orbicularis, Trachemys sp. Caiman crocodilus

F

Sampling*

Turtles

Blue tit

Parus caeruleus

Species

Common name

Adult

Adult

Adult

Adult and sub-adult Juvenile

Adult and chick Adult

Development stage

20 20 20 20 20 40

Muscle

Muscle + Bone + Teeth Tail Skin Liver Spleen Muscle + Bone + Teeth Gonads + Muscle

Blood

Muscle

Testis Ovary Blood

20 mg

Heart

40 mg

30 mg 10 mg 10 mg 3 mg 20 mg

20 mg

100 lL

2 mg

mg mg mg mg mg lL

20 mg

Mean weight ⁄ volume

Lung

Tissue

Sample preparation at the laboratory (L) or on the field (F). Stabilization with liquid nitrogen (Ni), RNAlater  (R) or guanidinium thiocyanate–phenol solution (G). ‡ Mean NanoDrop measurement, in lg. § Shell and digestive tube removed. ¶ RNeasy columns with Proteinase-K treatment. GTPC, guanidinium thiocyanate–phenol–chloroform; SM, Silica matrix.

†

*

Vertebrata, Dipnoi Urochordata, Ascidiacea

Vertebrata, Reptilia

Vertebrata, Aves

Vertebrata, Mammalia

Taxonomic group

Table 1 Continued

SM (20) GTPC (5) SM (17) GTPC (1) SM (2) GTPC (6) SM (6) SM¶(6) GTPC (1) SM (1) Combined (3)

GTPC (11) SM (43)

Combined (4) Combined (2) Combined (2) Combined (1) Combined (4)

Combined (8)

Combined (7)

40.4–1.01 31.6–0.79

3.3–0.11 0.5–0.05 7.0–0.70 9.0–3.00 10.8–0.54

9.9–0.50

7.9–0.08

4.7–2.35

23.7–1.19 28.3–1.42 9.9–0.50 1.2–0.06 3.6–0.18 13.6–0.68 0.7–0.04 4.2–0.21 124.7–6.24 23.0–1.15 8.2–0.21

Extraction method (No. extractions)

SM¶ (7)

RNA yield‡ per extraction— per lg (or ll) tissue

1.89–1.83 2.02–1.16

2.02–1.22 1.88–0.23 2.10–1.69 2.09–2.24 2.04–2.00

1.99–1.55

2.19–1.79

2.06–1.56

2.04–2.10 1.84–1.72 2.08–1.96 1.61–0.31 2.16–1.62 1.85–1.34 2.41–0.48 1.99–1.84 2.04–2.12 2.06–1.94 1.98–1.32

Mean RNA 260 ⁄ 280-260 ⁄ 230 Abs. ratio

4 P. GAYRAL ET AL.

 2011 Blackwell Publishing Ltd

RNA EXTRACTION FOR NEXT-GEN MRNA SEQUENCING 5 extractions steps are tissue disruption, lysate homogenization, isolation and purification of RNA. In guanidinium thiocyanate–phenol–chloroform (GTPC) extraction, RNA is separated from DNA by centrifugation in an acidic solution. Most DNA and proteins remain in the interphase or the organic phase, while total RNA stays in the upper aqueous phase. RNA is then precipitated with isopropanol (Chomczynski & Sacchi 1987, 2006). GTPC extraction was performed in our experiments with TRI Reagent BD for blood samples (Chiari & Galtier 2011) and TRIzol for the other tissues following manufacturer’s instructions, unless specified. For ants and termites, we tested the efficacy of RNase-free glycogen (Invitrogen) as a nucleic acid carrier for increasing the RNA yield, as suggested in the TRIzol manufacturer’s instructions. Prior to precipitation of the RNA with isopropanol, 10 lg of glycogen was added to the aqueous phase. Silica matrix (SM) RNA extraction is the method found in most commercial extraction kits available on the market. RNA is adsorbed to the silica matrix of a spin column with the aid of the high pH and salt concentration of the binding solution. Contaminants (e.g. proteins) are removed by using washing buffer containing a competitive agent and operated under centrifugal force. Elution of purified RNA is performed with DEPC-treated water or commercial RNase-free water. SM extraction was performed using RNeasy Mini Spin Columns (RNeasy plus kit; Qiagen, Chatsworth, CA, USA). Unless specified, the manufacturer’s protocols were followed, such as the use of gDNA Eliminator Mini Spin Columns (Qiagen) to remove contaminant genomic DNA. Muscle RNA extraction from hares and penguins was performed with the RNeasy Fibrous Tissue kit (Qiagen), with 25 min of incubation in proteinase K for penguins instead of the recommended 10 min. For hares muscle, an additional step was performed after proteinase K incubation: the lysate was passed through gDNA Eliminator Mini Spin Columns, instead of the DNAse treatment step suggested by the manufacturer. This change allowed assessing the direct effect of proteinase K, because gDNA Eliminator column was used in both standard (RNeasy plus) and specific (RNeasy Fibrous Tissue) kits. We adapted a combined SM ⁄ GTPC protocol for RNA extraction (see Data S1), initially designed for human cells RNA extraction (Baelde et al. 2001) and then used in honey bees (Koywiwattrakul et al. 2009) and fish embryo (Triant & Whitehead 2009). This method was tested here on five animal phyla (Table 1). For nematode, nemertea and insects, whole bodies were ground in a ceramic pestle and mortar using liquid nitrogen and mixed with TRIzol (1 mL per 50–100 mg tissue) by pipetting. Tubes were shaken for 15 s and incubated for 5 min at room temperature for efficient lysis. For oyster and lizard,

 2011 Blackwell Publishing Ltd

tissue lysis was performed by incubation in TRIzol following a modification of the protocol described by Montagnani et al. (2001). Tissues were finely cut into small pieces in sterile cell culture plates and incubated in TRIzol reagent (1 mL ⁄ 100 mg of tissue) for 10 h at 4 C under shaking conditions. For lizards, plastic pestles (described above) were used to help tissue grinding. Size of tissue parts was visibly reduced and texture changed when lysis by incubation was complete. This lysis solution was then used for RNA extraction after a short centrifugation step. Then, for all species, chloroform (200 lL per mL of TRIzol) was added and tubes were shacked by inversions for 30 seconds and incubated for 5 minutes at room temperature. Tubes were then centrifuged at 4 C for 15 min at 12 000 g. An equal volume of 70% ethanol was mixed by pipetting to the RNA-containing upper aqueous phase. The solution was passed through an RNeasy column; then, all remaining steps were carried out according to RNeasy protocols. A careful homogenization of the sample after disruption is necessary with all RNA extraction methods to achieve high RNA yields and reliable results in terms of relative abundance of mRNA. According to kits and products manufacturers, this was done by intensive (20 times) pipetting of the lysate, or by passing the lysate five times into a 0.9-mm needle attached to a syringe, when the solution was very viscous. For insects, extensive centrifugation and ⁄ or use of QIAshredder columns (Qiagen) helped removing insoluble material (cuticle, scales) to prevent clogging the SM column or interfering with upper phase removal in the GTPC method.

DNA extractions When total RNA was extracted with RNeasy columns, DNA was simultaneously extracted using DNA Mini Spin Columns (AllPrep DNA ⁄ RNA mini kit; Qiagen) following the manufacturer’s instructions. After butterflies (Melitaea sp.), ants, oysters, sea squid and nemertian total RNA extraction with the GTPC method, DNA was extracted using a back extraction buffer (BEB) (4 M guanidinium thiocyanate; 50 mM sodium citrate; 1 M Tris, pH 8.0), which was obtained from a slightly modified protocol of Triant & Whitehead 2009. After phase separation using chloroform, all the residual aqueous phase (and interphase for oysters) was removed and BEB was added (1 volume of BEB per volume of TRIzol used for tissue lysis). Tubes were mixed for 30 s by vortexing and centrifuged for 30 min at 12 000 g at room temperature. DNA was precipitated by adding 0.4 mL isopropanol per initial mL of TRIzol to the upper phase. Tubes were incubated for 5 min at room temperature and then centrifuged for 15 min at 12 000 g at 4 C. The supernatant was removed, and the DNA pellet was washed by

6 P. GAYRAL ET AL. inverting in 70% ethanol (0.5 mL per initial mL of TRIzol). After a final centrifugation of 15 min at 4 C, ethanol was thoughtfully removed; DNA was then dried for 5–10 min at room temperature and finally dissolved in 50–100 lL of DNAse-free water depending on pellet size. Complete DNA dissolution was achieved by slow pipetting and incubating overnight at 4 C.

Quantity and quality control of total RNA Quantity and quality (purity and integrity) was assessed by three methods. We first measured the absorbance at different wavelengths using a NanoDrop spectrophotometer (Thermoscientific, Wilmington, DE, USA). Readings at 260 nm indicate the concentration of total nucleic acids—RNA and DNA. A 260 ⁄ 280 ratio estimates the purity of RNA regarding compounds absorbing UV light, such as proteins (260 ⁄ 280 approximately 1.85 in nuclease-free water—pH 6–7 for pure RNA) (Wilfinger et al. 1997). A 260 ⁄ 230 ratio was also used to estimate the presence of other contaminants (salts, such as guanidinium thiocyanate, and carbohydrates, peptides and phenol). Expected 260 ⁄ 230 values are commonly in the range of 2.0–2.2 for pure RNA. Second, integrity was approximately assessed using electrophoresis of 0.5 lg total RNA on 1% agarose gel (RNase-free grade) and ethidium bromide staining. A capillary electrophoresis in RNA 6000 Nano LabChip (Agilent) was then performed using an Agilent Bioanalyzer 2100 system (Van de Goor 2003). This method allows a finer integrity analysis and a more precise estimate of dosage because DNA contaminants do not disturb RNA dosage. Integrity of RNA is summarized by two main statistics: RNA integrity number (RIN) (range 1–10, score 10 when no degradation) and rRNA ratio (28S ⁄ 18S) (score 2 for a very high-quality RNA) (Schroeder et al. 2006).

Statistical analyses The Wilcoxon rank test was used to assess the difference in distribution of two samples to be compared. Ties, if any, were resolved randomly and independently before computing the P-value of the test. The P-value of 50 tie resolutions was calculated, and to be conservative, the maximum value was reported here. Type-III ANOVA was used to assess the effect of glycogen treatment on total RNA yield on three insect species, with and without the interaction term ‘treatment · species’. Both tests were performed with the R software (RTeam 2010).

NGS Next-generation sequencing was contracted out to GATC Biotech (Konstanz, Germany). For 454 sequencing, poly-

A RNA was isolated from 25 lg total RNA, a random primed cDNA synthesis was performed and prepared for sequencing. This size-selected cDNA was sequenced for half a run using a Genome Sequencer (GS) FLX Titanium Instrument (Roche Diagnostics) using a standard protocol (Margulies et al. 2005). For Illumina sequencing, 5 lg of total RNA was reverse-transcribed using SMART cDNA library Construction kit (Clontech, Mountain View, USA). An oligo(dT)-primed first-strand synthesis and cap-primed second-strand synthesis was performed. Adapters were ligated after cDNA was colligated and nebulizated, and sequencing (five tagged individuals per lane) was carried out on a Genome Analyzer II (Illumina, Inc.).

Results RNA quality control RNA is considered to be of good quality when on an agarose gel or on an Agilent profile both 18S and 28S rRNA bands or peaks are well defined, without strong smearing towards a smaller size, and when the 28S peak is stronger than the 18S peak (Fig. 1). In several protostomes, the 28S peak was not clearly visible or absent. The denaturative hydrogen bond breaking nature of the reagents used in RNA extraction is known to dissociate the 28S rRNA into two fragments that migrate in a similar manner to the 18S rRNA (Fig. 1d). Consequently, the 28S ⁄ 18S rRNA ratio and sometimes the RIN (both measures of RNA integrity, see Methods) can unfortunately not be calculated when dissociation occurs. When extracting RNA from nucleated blood cells, 28S ⁄ 18S ratios may be below 0.8 without observing RNA degradation, as a result of the an incorrect concentration of glacial acetic acid on the total volume of blood and buffer used for the extraction (see Chiari & Galtier 2011 for additional information). For all those latter cases, RNA integrity was quantified by looking at Agilent profiles. On the whole, based on NanoDrop 260 ⁄ 280 and 260 ⁄ 230 ratios (measures of RNA purity, see Methods) and on Agilent RIN and rRNA ratios, we achieved good-quality RNA extractions from large numbers of diversified samples as described in detail below.

Optimal stabilization of samples collected in the field Field conditions are not always compatible with the recommended storage of tissues in stabilization buffers for preserving RNA. We therefore assessed the influence of organism ⁄ tissue dissection, preservation temperature and length of preservation time on RNA integrity in a panel of vertebrate and invertebrate species. In hare, we tested whether dissecting tissues in thin pieces offered a better stabilization in RNAlater than

 2011 Blackwell Publishing Ltd

RNA EXTRACTION FOR NEXT-GEN MRNA SEQUENCING 7 (a)

(b)

(c)

(d)

Fig. 1 Example of quantification and integrity analysis from Agilent Bioanalyzer. Panel (a) very good-quality RNA. Panel (b) very degraded RNA. Integrity of the mRNA is assessed by integrity of the abundant and clearly visible 18S and 28S rRNA (Alberts et al. 1994). As degradation results in cutting rRNA into smaller RNAs, RNA degradation is visualized by smears towards small sizes and by a drop in the 18S peak. Panel (c) slightly degraded RNA: moderate drop in 28S peak and little smear of lower sizes. Panel (d) very goodquality RNA (absence of smear of lower sizes) with 28S peak not visible (see main text). Panels a, c and d are suitable for Next Generation Sequencing technology.

with sizes usually recommended by manufacturers of stabilization buffers. Table 2 shows the two statistics used for RNA integrity control, RIN and rRNA ratio, for the two size classes in hares. The RIN ratio was increased by 7.8% (not significant) and the rRNA ratio by 54.5% (P-value below 0.024, Wilcoxon rank) when small-sized pieces of tissues were used, which suggested greater stabilization and lowered RNA degradation than in samples of recommended sizes. Table 2 Effect of tissue size in RNA quality from RNAlater stabilized tissues of hares

Thick Thin N Max p-value

RIN*

rRNA ratio*

N

7.7 8.3 57 0.276

1.1 1.7 59 0.0235

76 40

*

Mean RNA integrity number (RIN) values and rRNA ratio are from Agilent bioanalyzer.

 2011 Blackwell Publishing Ltd

We also tested for small gastropods whether the presence of a shell during incubation for 3 days at room temperature significantly decreased RNA integrity, by preventing access of the stabilization buffer to tissues. Figure 2 shows that there was degradation of RNA when the shell was not removed exemplifying the importance of sample preparation when using stabilization buffers. A similar result was observed in insects when they were plunged entirely or after the main segment has been cut (data not shown). We inspected Agilent profiles from 31 RNA TRIzol extractions from fresh water gastropods and did not observe any differences in RNA quality profiles between flash-frozen tissues (N = 18), and tissues were incubated over-night in RNAlater at 4 C and subsequently stored at )20 C (N = 12). This shows that when manufacturer’s protocols were strictly followed, this stabilization buffer allowed efficient protection of RNA from degradation in fresh water gastropods. Furthermore, when we compared different length of preservation times and temperature (to simulate field conditions), our tests showed that

8 P. GAYRAL ET AL. observed that grinding these animals in TRIzol and storage for several days at 4 C, followed by an extraction using the same buffer, again, provided good-quality RNA (Fig. 4).

Comparison GTPC, SM and Combined RNA extractions methods

Fig. 2 Effect of sample preparation on RNA stabilization. Total RNA extracted from Galba truncatula loaded on an ethidium bromide-stained 1% agarose gel. Lines 1–4: snails were incubated with shell during 3 days at room temperature in 1.5-mL RNAlater, lines 5–8: shells of the snails were removed and then incubated the same way. M: 10-kb marker Smart Ladder (Eurogentec, Belgium).

only little RNA degradation occurred in butterfly segments in RNAlater during 10 days at room temperature, compared to 10 days at 4 C or after immediate flash freezing in liquid nitrogen (Fig. 3). Finally, we

Based on results presented in Table 1, RNA yield was in most cases higher in GTPC than in SM extractions. In hare, yield per extraction was doubled for kidney and lung, moderately increased for muscles (+18.6%), liver (+8.4%) and spleen (+4.6%), and slightly decreased for heart when using GTPC, as compared to SM extraction. GTPC extraction also increased yield by a factor of 2.8 in artemia and 2.3 in physa freshwater snails, whereas it slightly decreased yield for the smaller limneid snails compared to SM extraction. When comparing the purity of RNA from 300 GTPC and 326 SM RNA extractions (Table 1), we observed that the SM method yielded RNA with less protein contaminant than GTPC extractions (Wilcoxon rank test on A260 ⁄ 280 values, max P-value = 1.2 · 10)37) and less salt

(a)

(b)

(c)

(d)

Fig. 3 Effect of RNA later storage condition on RNA stabilization. Agilent Bioanalyzer Chromatography shows butterfly RNA quality. Animals were cut into pieces, put in RNA later and incubated 10 days at room temperature (a and b), 10 days at 4 C (c), or immediately flash-frozen without incubation in RNAlater (d). TRIzol extraction was performed afterwards following manufacturer’s instructions.

 2011 Blackwell Publishing Ltd

RNA EXTRACTION FOR NEXT-GEN MRNA SEQUENCING 9

Fig. 4 Effects of storage temperature on RNA integrity after grinding in TRIzol. Butterflies (Pieris spp.) were ground in TRIzol and either incubated 3 days at room temperature (line 1 and 2) or incubated 3 days at 4 C (lines 3 and 4). Total RNA was isolated using TRIzol extraction (following manufacturer’s instructions) and loaded on an ethidium bromide-stained 1% agarose gel. M: 10-kb marker Smart Ladder (Eurogentec, Belgium).

content (Wilcoxon rank test on A260 ⁄ 230 values, max P-value = 2.93 · 10)15) (Table 3). Mean 260 ⁄ 280 and 260 ⁄ 230 absorbance ratios for all extractions were 1.96 and 1.34, respectively, which are considered suitable values for NGS. The combined RNA extraction method was applied to 6 genera: wax moth, nematode, oyster, nemertea, caiman and lungfish, in a total of 169 extractions. The mean purity of these extractions clearly outperformed those from GTPC, but also from SM alone. Mean 260 ⁄ 280 and 260 ⁄ 230 absorbance ratios for the 169 combined extractions were 2.06 and 1.85, respectively. Mean yield was also very suitable (30 lg); this value can however not be directly compared to GTPC and SM yield because the species chosen for Combined extractions were not the same than those for GTPC and SM extractions.

Simultaneous extraction of RNA and DNA Because many samples are difficult to obtain, it is oftentimes desirable to obtain both DNA and RNA from a single precious sample. We thus tested simultaneous DNA and RNA extractions from several species. When RNA Table 3 Comparison of RNA purity with GTPC or SM extraction Extraction method

260 ⁄ 280*

260 ⁄ 230*

N

GTPC SM N Max P-value

1.84 2.07 313 1.19 · 10)37

0.98 1.66 313 2.93 · 10)15

300 326

* Absorbance reading at 230, 260 and 280 nm are from NanoDrop. GTPC, guanidinium thiocyanate–phenol–chloroform; SM, Silica matrix.

 2011 Blackwell Publishing Ltd

was extracted using the SM method, simultaneous DNA isolation performed with DNA Mini Spin Columns produced high yield and quality DNA in all species tested (data not shown). For RNA extracted with GTPC, however, simultaneous DNA isolation following the TRIzol manufacturer’s manual resulted in low yield and lowquality DNA (data not shown). We therefore applied a modification of the protocol for DNA extraction by using BEB after GTPC-based RNA extraction. This method was applied to butterfly, ant, oyster, sea squid and nemertian, and mean yield (lg) was 452.3 (N = 4), 242.7 (N = 2), 422.1 (N = 2), 92.5 (N = 28), 3.1 (N = 1), 35.2 (N = 6), respectively. Despite the fact that the mean 260 ⁄ 280 absorbance ratio (NanoDrop) was relatively low (1.50), and agarose gel electrophoresis of DNA stained with ethidium bromide did not reveal any visible genomic DNA (a result also observed in Triant & Whitehead 2009), this method was nevertheless chosen because it yielded a large amount of DNA, and because PCR amplifications were successfully achieved from diluted samples obtained this way (data not shown).

Organism-specific considerations Although in the majority of cases, the GTPC and SM methods, or their combination, gave high RNA yield and good-quality RNA, extraction remained problematic in a couple of specific tissues and organisms (Table 1). Low RNA yield from skeletal muscle, heart and skin tissue may be because of the abundance of contractile proteins, connective tissue and collagen. Using the standard SM method, heart and muscle tissue gave low RNA yield in hares (3.6 and 0.7 lg per extraction, respectively, from 20 mg tissue) (Table 1). However, the additional use of proteinase K in SM extractions (RNeasy Fibrous Tissue kit) improved yield in hare and penguin muscle (4.2 lg RNA from 20 mg muscle, N = 6 extractions and 4.7 lg RNA from 2 mg tissues, N = 7, respectively). GTPC extractions of hare muscle produced higher yield (13.6 lg from 20 mg tissue, N = 6), but quality was lower because of the presence of contaminants (260 ⁄ 280 ratio = 1.85 and 260 ⁄ 230 ratio = 1.34). Purity in this case was similar to that of all 300 GTPC RNA extractions, which is shown in Table 3. Silica matrix RNA extractions from insects produced erratic yield, and in most cases, no RNA was isolated, even from relatively large animals such as Melitaea butterflies (Table 1). Based on our results, GTPC extraction was the optimal extraction method for insects. Given the low yield obtained with either method for very small insects, we also tested whether the use of glycogen as a nucleic acid carrier could significantly increase the yield of RNA recovery during GTPC extraction. This modification produced an increase in mean RNA yield by a factor

10 P . G A Y R A L E T A L . of 4.1 and 1.5 in the ant species Crematogaster sp. and Messor sp., respectively, and by a factor 2 in termite (Table 4). Glycogen effect on RNA yield was statistically significant (ANOVA P-value = 0.0058). This effect remained significant (P-value = 0.0119) when the interaction term ‘treatment · species’ (P-value = 0.4091) was added in the analysis. Box-Cox transformation of the data to better approximate normality (Box & Cox 1964) did not change this result (data not shown). Finally, we tested whether incubation of very finely cut tissues from oysters and lizards in TRIzol could replace the laborious task of tissue grinding, and produce good-quality and quantity RNAs with less bench work. In combination with purity assay (NanoDrop) and Agilent chromatography (data not shown), our results indicate that the incubation in TRIzol provides a good tissue stabilization and an efficient cell lysis. The resulting RNA is therefore thoroughly suitable for mRNA sequencing. The only exception was lizard skin, which did not produce a satisfactory yield (Table 1).

NGS applications from total RNA recovered from this study cDNA library construction and 454 sequencing was successfully achieved from RNA extracted with one of the above indicated methods for each of Ciona intestinalis, Emys orbicularis, Lepus granatensis, Physa acuta, Galba truncatula and Ostrea edulis. Mean number of reads per sample was 551 353, mean length was 325 nucleotides and mean quality scores was 31. For Illumina sequencing, cDNA library construction was successfully achieved for 74 samples of 75 processed. Sequencing was performed in 10 individuals of Ciona intestinalis, 10 individuals of Ostrea edulis and 10 individuals of Lepus granatensis so far. Mean number of reads per sample was 4 319 358, mean length was 89 nucleotides and mean quality scores was 29.

Discussion The use of RNA as genetic material for producing genomic data in biodiversity studies is on the increase (Elmer Table 4 Effects of glycogen on mean total RNA yield (lg ⁄ extraction) in small insects Crematogaster Messor Reticulitermes scutellaris barbarus grassei N Standard TRIzol 3.085 9.922 1.848 protocol TRIzol + glycogen 12.562 14.576 3.801 N 7 16 4 ANOVA P-value 0.0058 (effect of glycogen factor)

12 15

et al. 2010; Nabholz et al. 2010; Renaut et al. 2010; Wheat 2010; Wolf et al. 2010). However, RNA extraction from wild-caught nonmodel animals using standard SM and GTPC protocols is far from being a routine technique, as these protocols have been typically designed for cell cultures or tissues of model organisms (Chomczynski & Sacchi 2006; Triant & Whitehead 2009; Vomelova´ et al. 2009). Ecology and population genetic researchers have to cope with field conditions, nonmodel animals and heterogeneous samples. RNA extractions in such conditions can therefore be challenging. RNA extraction protocols adapted to nonmodel animals are scarce. For half of the genera used in this study, no published protocols or results from RNA extractions were available in the literature. For all other genera used here, only a single study used RNA extraction for NGS (Vera et al. 2008). RNA extractions results were otherwise designed for single gene expression analysis using RTPCR or Northern blot analyses (Montagnani et al. 2001; Iijima et al. 2008; Agrawal et al. 2009; Garant & MacRae 2009; Hung et al. 2009; Navet et al. 2009; Shaik & Sehnal 2009; Tarver et al. 2010), for phylogenomics using ESTs (Struck & Fisse 2008) or for phylogeny (Kiontke et al. 2004). The distinction is important because NGS require a large amount of starting material, and sample, organism size and RNA quality can therefore be limiting. Here we illustrate how manufacturer’s instructions or published protocols can be combined and modified to successfully extract RNA from a variety of nonmodel animals sampled in the field. Based on a battery of experiments, we draw general guidelines for sample preparation, stabilization and RNA extraction: 1 Our tests confirmed that recommendations of use and storage of stabilization buffer are to be adapted to nonmodel species and field conditions. Thickness of tissue pieces, presence of exoskeletons, storage time and temperature are the major factors impacting RNA integrity prior to RNA extraction. Although time-demanding, the dissection of tissues into small pieces appears worthwhile. In hares, thin tissues samples offered greater stabilization and significantly lowered RNA degradation than samples of standard sizes. 2 Differences between SM and GTPC, the main two extraction methods described in literature (Tan & Yiap 2009), are shown in Table 5. This should help the researchers to choose the preservation and extraction approach most appropriate to their specific problem. Combination of SM and GTPC into a single protocol provides efficient extraction of good quality RNA for many animals. 3 For GTPC extractions, tissue incubation in TRIzol enables convenient lysis with less work than grinding and can be adapted to soft and hard tissues.

 2011 Blackwell Publishing Ltd

R N A E X T R A C T I O N F O R N E X T - G E N M R N A S E Q U E N C I N G 11 Table 5 Comparison of the two main RNA extraction techniques Advantages

Drawbacks

GTPC

SM

Very efficient lysis High yield Less expensive Harmful chemicals (Phenol, Chloroform, Guanidinium thiocyanate) Longer procedure Possible residual contaminants (Phenol, salts) Removal of protein and DNA depends on pipetting skills (disturbing the phases leads to contamination)

Fast procedure Purer RNA regarding to unwanted compounds Harmful chemicals (2-mercaptoethanol, Guanidinium thiocyanate) Lower yield (loss of material because of incomplete elution of RNA) Expensive

GTPC, guanidinium thiocyanate–phenol–chloroform; SM, Silica matrix.

Table 6 Common problems and possible solutions for RNA extraction from nonmodels animals Organisms types ⁄ tissues

Symptoms

Experimental result

Fibrous tissues

Heart, skeletal muscle, skin

Inefficient homogenization and cell lysis

Low yield

Nucleic acid-rich tissues Nucleated red blood cells Cuticle-rich tissues

Spleen, thymus

Low yield, strong DNA contamination

Small (< 5 mg) samples

Insects, crustacean

Viscous lysate, column is saturated High DNA and nuclease content Lysate difficult to clear by centrifugation No visible pellet using GTPC extractions

Tissue characteristics

Birds and reptiles blood Insects

Low or no yield using column extraction Low or no yield

Procedure modification ⁄ protocol choice Thorough disruption, perform GTPC extractions or use Proteinase K in SM extractions Use less tissue or dilute lysate in lysis buffer See Chiari & Galtier 2011 Perform GTPC extractions Perform GTPC extractions, use a nucleic acid carrier (glycogen)

GTPC, guanidinium thiocyanate–phenol–chloroform; SM, Silica matrix.

4 Extraction protocols can be adapted to species, tissue type and sample size, as summarized in Table 6. SM extraction, for instance, is less appropriate for very small samples, or insects. 5 DNA and RNA can easily be simultaneously isolated, and the use of BEB provides satisfactory results in GTPC RNA extractions.

Escoubas, Nathalie Volkoff and Viviane Boulo for technical advice. Data used in this work were partly produced through molecular genetic analysis technical facilities of the SFR ‘‘Montpellier Environnement Biodiversite´’’, thanks to Erick Desmarais, Fre´de´rique Cerqueira and Philippe Clair (UM2-Montpellier GenomiX). This work has been supported by a European Research Council (ERC) grant to Nicolas Galtier (ERC PopPhyl 232971). This is publication number ISEM 2011-010.

These guidelines, derived from our own experiments, should assist biologists in obtaining high-quality RNA for NGS-based transcriptomics in nonmodel organisms.

References

Acknowledgements We thank many colleagues for their help with sampling, Vincent Cahais for support in bioinformatics, John Welch and Julien Dutheil for useful advice in statistics, and Marie-Ka Tilak-Jean for her valuable comments on the article. We thank Jean-Michel

 2011 Blackwell Publishing Ltd

Agrawal R, Wessely O, Anand A, Singh L, Aggarwal RK (2009) Male-specific expression of Sox9 during gonad development of crocodile and mouse is mediated by alternative splicing of its proline-glutamine-alanine rich domain. FEBS Journal, 276, 4184–4196. Alberts B, Bray D, Lewis J et al. (1994) Molecular Biology of the Cell, 3rd edn, p. 1361. Garland Publishing Inc., New York. Baelde HJ, Cleton-Jansen A-M, van Beerendonk H et al. (2001) High quality RNA isolation from tumours with low cellularity and

12 P . G A Y R A L E T A L . high extracellular matrix component for cDNA microarrays: application to chondrosarcoma. Journal of Clinical Pathology, 54, 778–782. Bers NEMV, Oers KV, Kerstens HHD et al. (2010) Genome-wide SNP detection in the great tit Parus major using high throughput sequencing. Molecular Ecology, 19, 89–99. Box GEP, Cox DR (1964) An analysis of transformations. Journal of the Royal Statistical Society: Series B, 26, 211–252. Chiari Y, Galtier N (2011) RNA extraction from sauropsids blood: evaluation and improvement of methods. Amphibia-Reptilia, 32, 136–139. Chomczynski P, Sacchi N (1987) Single-step method of RNA isolation by acid guanidinium thiocyanate-phenol-chloroform extraction. Analytical Biochemistry, 162, 156–159. Chomczynski P, Sacchi N (2006) The single-step method of RNA isolation by acid guanidinium thiocyanate-phenol-chloroform extraction: twenty-something years on. Nature Protocols, 1, 581–585. Elmer KR, Fan S, Gunter HM et al. (2010) Rapid evolution and selection inferred from the transcriptomes of sympatric crater lake cichlid fishes. Molecular Ecology, 19, 197–211. Garant KA, MacRae TH (2009) Cloning and sequencing of tubulin cDNAs from Artemia franciscana: evidence for differential expression of alphaand beta-tubulin genes. Biochemistry and Cell Biology, 87, 989–997. Hudson ME (2008) Sequencing breakthroughs for genomic ecology and evolutionary biology. Molecular Ecology Resoures, 8, 3–17. Hung CYC, Galvez F, Ip YK, Wood CM (2009) Increased gene expression of a facilitated diffusion urea transporter in the skin of the African lungfish (Protopterus annectens) during massively elevated post-terrestrialization urea excretion. Journal of Experimental Biology, 212, 1202– 1211. Iijima M, Takeuchi T, Sarashina I, Endo K (2008) Expression patterns of engrailed and dpp in the gastropod Lymnaea stagnalis. Development Genes and Evolution, 218, 237–251. Kiontke K, Gavin NP, Raynes Y et al. (2004) Caenorhabditis phylogeny predicts convergence of hermaphroditism and extensive intron loss. Proceedings of the National Academy of Sciences of the United States of America, 101, 9003–9008. Koywiwattrakul P, Thompson GJ, Sitthipraneed S, Oldroyd BP, Maleszka R (2009) Effects of carbon dioxide narcosis on ovary activation and gene expression in worker honeybees, Apis mellifera. Journal of Insect Science, 5, 1–10. ¨ M N et al. (2010) Comparative genomics Ku¨nstner A, Wolf JBW, BackstrO based on massive parallel transcriptome sequencing reveals patterns of substitution and selection across 10 bird species. Molecular Ecology, 19, 266–276. Margulies M, Egholm M, Altman WE et al. (2005) Genome sequencing in microfabricated high-density picolitre reactors. Nature, 437, 376–380. Montagnani C, Le Roux F, Berthe F, Escoubas JM (2001) Cg-TIMP, an inducible tissue inhibitor of metalloproteinase from the Pacific oyster Crassostrea gigas with a potential role in wound healing and defense mechanisms. Febs Letters, 500, 64–70. Nabholz B, Jarvis ED, Ellegren H (2010) Obtaining mtDNA genomes from next-generation transcriptome sequencing: a case study on the basal Passerida (Aves: Passeriformes) phylogeny. Molecular Phylogenetics and Evolution, 57, 466–470. Navet S, Andouche A, Baratte S, Bonnaud L (2009) Shh and Pax6 have unconventional expression patterns in embryonic morphogenesis in Sepia officinalis (Cephalopoda). Gene Expression Patterns, 9, 461–467.

Renaut S, Nolte AW, Bernatchez L (2010) Mining transcriptome sequences towards identifying adaptive single nucleotide polymorphisms in lake whitefish species pairs (Coregonus spp. Salmonidae). Molecular Ecology, 19, 115–131. RTeam (2010) R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria. Sambrook J, Fritsch EF, Maniatis T (1989) Molecular Cloning: A Laboratory Manual, 2nd edn, 1659 p. Cold Spring Harbor Laboratory Press, US. Schroeder A, Mueller O, Stocker S et al. (2006) The RIN: an RNA integrity number for assigning integrity values to RNA measurements. BMC Molecular Biology, 7, 3. Shaik HA, Sehnal F (2009) Hemolin expression in the silk glands of Galleria mellonella in response to bacterial challenge and prior to cell disintegration. Journal of Insect Physiology, 55, 781–787. Struck TH, Fisse F (2008) Phylogenetic position of nemertea derived from phylogenomic data. Molecular Biology and Evolution, 25, 728–736. Tan SC, Yiap BC (2009) DNA, RNA, and protein extraction: the past and the present. Journal of Biomedicine Biotechnology, 2009, 574398. Tarver M, Zhou X, Scharf M (2010) Socio-environmental and endocrine influences on developmental and caste-regulatory gene expression in the eusocial termite Reticulitermes flavipes. BMC Molecular Biology, 11, 28. Triant DA, Whitehead A (2009) Simultaneous extraction of high-quality RNA and DNA from small tissue samples. Journal of Heredity, 100, 246– 250. Van de Goor TA (2003) The principle and promise of labchip technology. Pharma Genomics, 3, 16–18. Vera JC, Wheat CW, Fescemyer HW et al. (2008) Rapid transcriptome characterization for a nonmodel organism using 454 pyrosequencing. Molecular Ecology, 17, 1636–1647. Vomelova´ I, Vanı´ckova´ Z, Sedo A (2009) Methods of RNA purification. All ways (should) lead to Rome. Folia Biologica (Praha), 55, 243–251. Wheat C (2010) Rapidly developing functional genomics in ecological model systems via 454 transcriptome sequencing. Genetica, 138, 433– 451. Wilfinger WW, Mackey K, Chomczynski P (1997) Effect of pH and ionic strength on the spectro-photometric assessment of nucleic acid purity. BioTechniques, 22, 474–481. Wolf JBW, Bayer T, Haubold B et al. (2010) Nucleotide divergence vs. gene expression differentiation: comparative transcriptome sequencing in natural isolates from the carrion crow and its hybrid zone with the hooded crow. Molecular Ecology, 19, 162–175.

Supporting Information Additional supporting information may be found in the online version of this article. Data S1 Supplementary material. Please note: Wiley-Blackwell are not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.

 2011 Blackwell Publishing Ltd