Developmental changes in Scots pine transcriptome ...

Plant Physiology Preview. Published on September 6, 2016, as DOI:10.1104/pp.16.01082

1

Short title: Gene expression in pine heartwood transition zone

2 3

Corresponding author:

4

Teemu H. Teeri

5

Department of Agricultural Sciences

6

University of Helsinki

7

Latokartanonkaari 7

8

00790 Viikki, Finland

9

Phone: +358 2941 58380

10

Email: [email protected]

11 12

Research areas:

Biochemistry and Metabolism

13

Secondary areas: Genes, Development and Evolution

14 15 16 17 18 19 20 21 22 23 24 25 26 27 28

1 Downloaded from www.plantphysiol.org on January 5, 2017 - Published by www.plantphysiol.org Copyright © 2016 American Society of Plant Biologists. All rights reserved.

Copyright 2016 by the American Society of Plant Biologists

29

Developmental changes in Scots pine transcriptome during heartwood

30

formation

31 32

Kean-Jin Lim1, Tanja Paasela1, Anni Harju2, Martti Venäläinen2, Lars Paulin3,

33

Petri Auvinen3, Katri Kärkkäinen4 and Teemu H. Teeri1*

34 35

1

36

00014, University of Helsinki, Finland

37

2

38

Punkaharju, Finland

39

3

Institute of Biotechnology, P.O. Box 56, 00014, University of Helsinki, Finland

40

4

Natural Resources Institute Finland (Luke), P.O. Box 413, 90014 University of

41

Oulu. Finland

Department of Agricultural Sciences, Viikki Plant Science Centre, P.O. Box 27, Natural Resources Institute Finland (Luke), Finlandiantie 18, 58450

42 43

One-sentence Summary

44

The Scots pine heartwood formation takes place in the summer and is marked

45

by programmed cell death.

46 47

Footnote:

48

List of author contributions: THT, KK, AH and MV conceived and designed

49

the research; AH and MV obtained the plant material; LP and PA were

50

responsible for designing the library phase and conducting SOLiD sequencing;

51

KJL and TP designed, planned and carried out the experiments; KJL performed

52

data analysis; THT contributed in interpreting the data; KJL and TP wrote and

53

THT revised the manuscript, which all authors read, commented and approved.

54 55

Funding information: This work was funded by the National Technology

56

Agency of Finland (TEKES).

57 58

* Corresponding author: Teemu H. Teeri ([email protected])

59


60

Abstract

61 62

Scots pine (Pinus sylvestris L.) wood is desired in woodworking industries due

63

to its favourable timber characteristics and natural durability that is contributed

64

by heartwood extractives. It has been discussed whether the Scots pine

65

heartwood extractives (mainly stilbenes and resin acids) are synthesized in the

66

cells of the transition zone between sapwood and heartwood, or if they are

67

transported from the sapwood. Timing of heartwood formation during the yearly

68

cycle has also not been unambiguously defined. We measured steady state

69

mRNA levels in Scots pine transition zone and sapwood using RNA

70

sequencing. Year round expression profiles of selected transcripts were further

71

investigated by quantitative RT-PCR. Differentially accumulating transcripts

72

suggest that, of the Scots pine heartwood extractives, stilbenes are synthesized

73

in situ in the transition zone and gain their carbon-skeletons from sucrose and

74

triglycerides. Resin acids, on the other hand, are synthesized early in the spring

75

mainly in the sapwood, meaning that they must be transported to the heartwood

76

transition zone. Heartwood formation is marked by programmed cell death that

77

occurs during the summer months in the transition zone.

78 79 80 81 82 83 84 85 86 87 88 89


90

Introduction

91 92

Wood biosynthesis is a complex process that involves several developmental

93

activities such as cell division, cell expansion, cell wall thickening, and

94

programmed cell death (Plomion et al., 2001). In many, but not all tree species,

95

the final stage in wood development is formation of heartwood (HW). HW is

96

traditionally defined as “the inner layers of wood, which, in the growing tree,

97

have ceased to contain living cells, and in which the reserve materials (e.g.

98

starch) have been removed or converted into heartwood substances”, while

99

sapwood (SW) is defined as “the portion of the wood that in the living tree

100

contains living cells and reserve materials” (IAWA committee, 1964). Heartwood

101

can also be defined as the region where extractives have accumulated and by

102

this definition it can be metabolically active (Beekwilder et al., 2014; Celedon et

103

al., 2016) although all changes associated with heartwood formation may not

104

have yet occurred (Taylor et al., 2002).

105 106

The biological role of HW is still unclear, but it may be important in long-term

107

resistance against pathogens (Venäläinen et al., 2004). Nevertheless, HW has

108

a clear economic importance due to natural decay resistance of HW timber in

109

many tree species (Singh and Singh, 2011) and due to increasing

110

pharmaceutical and biocide applications of heartwood substances (Kampe and

111

Magel, 2013).

112 113

Heartwood can be found in both angiosperm (e.g. Acacia, Catalpa, Juglans and

114

Robinia) and gymnosperm (e.g. Cryptomeria, Larix, Picea and Pinus) tree

115

species. It is physiologically inactive, drier compared with the SW and filled with

116

HW extractives (Taylor et al., 2002). SW, on the other hand, is physiologically

117

active, has high moisture content, and functions in water, nutrient and sugar

118

transport between roots and the foliage (Taylor et al., 2002).

119 120

As the tree grows, SW is converted to HW. This takes place in the narrow

121

transition zone (TZ), the living cells in which reserve materials like starch are


122

consumed. TZ is drier than SW, but not as dry as HW (Bergström, 2003; Hillis,

123

1987). The width of the TZ varies both within and between tree species and

124

could be influenced by seasonal as well as other environmental aspects. In

125

general, the TZ is about one to three annual rings wide (Hillis, 1987), in Scots

126

pine usually one to two annual rings wide (Gustafsson, 2001).

127 128

It is still unclear when and how the initiation of HW formation occurs.

129

Involvement of the plant hormones ethylene and auxin in regulation of the HW

130

formation have been suggested (Nilsson et al., 2002; Shain and Hillis, 1973;

131

Hillis, 1968). Spicer (2005) described HW formation as a form of programmed

132

cell death (PCD) where the organelles of ray parenchyma cells in the TZ are

133

gradually disintegrated. Several studies have shown that the cell wall structure

134

of ray parenchyma cells change and the cells gradually die towards HW at the

135

boundary between SW and HW (Yang et al., 2003; Nakaba et al., 2008; Magel

136

et al., 2001; Nakada and Fukatsu, 2012; von Arx et al., 2015). In addition to

137

plant hormones, Nakada and Fukatsu (2012) suggested that desiccation at the

138

boundary between SW and HW initiates HW formation.

139 140

The specific timing of the HW formation is unclear and may certainly be species

141

specific. Different studies have described the process taking place at different

142

times of the year. HW formation in black locust (Robinia pseudoacacia) took

143

place in autumn, while in radiata pine (Pinus radiata) and black walnut (Juglans

144

nigra) it occurred during the dormancy period in winter (Taylor et al., 2002;

145

Kampe and Magel, 2013). Bergstöm et al. (1999) concluded that there is no

146

specific time of the year for HW formation in Scots pine. Indeed, the timing of

147

HW formation is not clearly understood nor has been satisfactorily described up

148

to date.

149 150

Two types of HW formation were distinguished by Magel (2000). In type I, or

151

Robinia-type of HW formation, HW extractives are biosynthesised in situ and

152

accumulate in the TZ. Precursors to the extractives are not present in the

153

ageing SW. In type II, or Juglans-type of HW formation, HW extractives are


154

formed through transformation of phenolic precursors by hydrolysis, oxidation

155

and polymerization in the TZ. In type II HW formation, HW extractive precursors

156

are gradually accumulating in the ageing SW (Magel, 2000; Kampe and Magel,

157

2013). Deposition of HW extractives and PCD marks the end of the HW

158

formation (Spicer, 2005). Many HW extractives are synthesised also in living

159

tissues when trees are under biotic or abiotic stress and thus contribute also to

160

the active defence of SW and other living tissues (Chiron et al., 2000b).

161 162

In Scots pine, HW extractives consist of the stilbenes pinosylvin (PS) and

163

pinosylvin monomethyl ether (PSME), resin acids and free fatty acids

164

(Saranpää and Nyberg, 1987). Variation of Scots pine HW phenolic extractive

165

content between tree individuals is wide and strongly correlates with decay

166

resistance of HW (Harju and Venäläinen, 2006; Leinonen et al., 2008). Although

167

extractive content is influenced by the environment (Magel, 2000), it is highly

168

inherited (Fries et al., 2000; Partanen et al., 2011).

169 170

It is still unclear if Scots pine HW extractives are synthesized in the cells of TZ

171

between SW and HW, or if they are transported from the SW. Furthermore, the

172

timing of HW formation during the yearly cycle has not been unambiguously

173

defined. In this work we used a transcriptomic approach to characterise

174

changes in gene expression during HW formation, i.e. in TZ compared to SW.

175

Using a set of indicator genes, defined through transcriptomics, we further

176

addressed the question of timing of HW formation by quantitating gene

177

expression in SW and TZ throughout the year in two Scots pine individuals.

178 179


180

Results

181

Localisation of the transition zone and transcript sampling

182

Increment cores were pencil marked in the field for the border between the wet

183

and the dry zone and photographed under UV illumination in the laboratory to

184

localise the fluorescing stilbenes. Fluorescence was observed in the whole of

185

the dry region, except for the outermost about one annual ring wide zone

186

(Figure 1A).

187 188

In order to determine the TZ in molecular terms, increment cores from two

189

trees, collected in June 2010, were sectioned per year ring around the pencil

190

mark (Figure 1A) and total RNA was extracted. The RNA yield, based on

191

absorbance at 260 nm, was comparable from zones 5-8 and somewhat less

192

from zone 4 (Figure 1B and Table S1). Zones 1-3 yielded much less material

193

absorbing at 260 nm, except for the June sample of tree 11 (Table S1). These

194

samples were analysed using Bioanalyzer and RT-PCR with primers designed

195

for a Scots pine actin gene. Results indicated that the absorbance was not due

196

to RNA (Figure S1), in line with the observations by Nakaba and colleagues

197

(2008) and with the definition of HW (IAWA committee, 1964). Semiquantitative

198

RT-PCR for the pine stilbene synthase (STS) encoding transcript was

199

performed for each sample. The outermost dry zone (zone 4) showed strongest

200

STS expression in all sampled series (Figure S2) and was designated as the

201

TZ, well in accordance of the definition by Bergström (2003) and Hillis (1987).

202

The SW sampling position was specified as 4 or 5 annual rings outwards from

203

the TZ (Figure 1A).

204 205

To choose an ideal sampling time, we harvested increment cores from two trees

206

during the summer months June, July and August of 2010. Semiquantitative

207

RT- PCR for the STS transcript was carried out from SW and TZ (Figure S2).

208

The pattern was consistent through these months, showing STS activity in TZ

209

but not in SW. As the earliest sampling month (June) tended to show stronger

210

activity than samples towards the end of summer, it was chosen as the full

211

scale transcriptome sampling month for the following year.


212 213

Scots pine transcriptomes

214

About thirty increment cores (per tree) were drilled from four 46 years old

215

unrelated Scots pine trees at breast height at the Leppävirta progeny trial stand

216

in June 2011. SW and TZ samples from each tree were pooled for RNA

217

extraction, tested for STS expression (Figure S3) and sequenced using the

218

SOLiD platform. 12-17 million paired-end reads were obtained from each RNA

219

sample. All transcriptome libraries were mapped against the Pinus EST

220

collection version 9.0 (The Gene Index Databases, 2014), the mapping rate

221

being 64-66 % for each library (Table S2). The Pinus EST collection is a


222

combined collection of expressed sequence tags from 36 pine species (P. taeda,

223

P. sylvestris, P. radiata etc.). Sequences from different species show high

224

similarity and their assembly into “tentative consensus” (TC) sequences has

225

been done as if they originated from a single species (Quackenbush et al.,

226

1999). Therefore, the collection of 77 326 TC sequences in the Pinus EST

227

collection is not an indicator of the pine gene number in any sense, nor do they

228

represent necessarily assemblies that are species specific, although many

229

assemblies match one-to-one with cDNA molecules isolated from individual

230

species. For example, while TC154538 is 98 % identical to the Scots pine STS

231

mRNA sequence S50350, there are altogether 25 STS encoding TCs matching

232

S50350 for 10-100 % in length and 88-98 % for sequence identity. By the

233

nature of the Pinus EST collection, the number of similar TCs sharing

234

differential expression has no meaning (only their annotation does), although

235

they may sometimes represent gene family members in Scots pine.

236 237

We also mapped the reads to an annotated Trinity (Haas et al., 2013) assembly

238

generated from the RNA sequencing (RNA-Seq) reads themselves after

239

converting them first to nucleotides. While the mapping rate was often higher

240

and fold changes for differentially expressed transcripts in some cases larger,

241

the expression data obtained was very similar to that obtained with Pinus EST

242

collection as reference. We present here results with the Pinus EST collection

243

since it includes also transcripts that are not expressed in our samples (e.g.,

244

transcripts for resin acid biosynthesis, see below) and the assemblies (TCs) are

245

longer. Although the full genome of Pinus taeda was recently published with

246

draft gene models (Wegrzyn et al., 2014), it performed very poorly as a target

247

for mapping our reads and the mapping rate was only approximately 6 % for

248

each library (Table S2).

249

Differential expression analysis was carried out using edgeR (see materials and

250

methods). A total of 1673 transcripts had statistically significant (false discovery

251

rate, FDR < 0.05) differential expression in TZ compared to SW, 1021 were

252

upregulated (Table S3) and 652 were downregulated (Table S4). Gene ontology

253

(GO) functional enrichment analysis (Table S5) indicated that GO categories 9 Downloaded from www.plantphysiol.org on January 5, 2017 - Published by www.plantphysiol.org Copyright © 2016 American Society of Plant Biologists. All rights reserved.

254

like biosynthetic and secondary metabolic processes and response to stimulus

255

were significantly over represented in the TZ. We focussed on pinosylvin

256

biosynthesis (secondary metabolism) and on its upstream pathways, then

257

extended our interest to other processes such as plant hormonal signalling,

258

dehydration, programmed cell death, transcription factors, cell wall modification

259

and defence.

260 261

Metabolic processes activated during heartwood formation

262

The biosynthesis of stilbenes is initiated by conversion of phenylalanine to

263

cinnamic acid, catalysed by the enzyme phenylalanine ammonia lyase (PAL).

264

STS converts the activated cinnamic acid (cinnamoyl-CoA) into pinosylvin. We

265

defined the TZ as the zone where STS encoding transcripts are most highly

266

expressed. Congruently, we could observe 5 to 13 fold upregulation of STS

267

encoding TCs in TZ compared to SW. Our transcriptome data also showed that

268

PAL encoding transcripts were 7 to 12 fold upregulated in the TZ. The next

269

enzyme, presumably an acyl-CoA ligase acting on cinnamate, has not been

270

characterized from pine, but some 4-coumarate:CoA ligase (4CL) transcripts

271

were upregulated with a 3 to 13 fold change. Lignin biosynthesis related

272

transcripts (including 4CL) were also upregulated in TZ. In the end of the

273

pathway, pinosylvin is in part methylated to PSME. Chiron et al. (2000a)

274

characterised in Scots pine an O-methyl transferase catalysing this reaction,

275

however, TC180420 that matches the published PMT sequence (99 % identity)

276

was not upregulated (in fact, very few reads mapped to it). Instead, another O-

277

methyltransferase encoding TC (TC166778) was upregulated with a 5 fold

278

change in TZ compared to SW (Figure 2, Table S6).

279 280

Upstream of stilbene pathway is the shikimate pathway that provides

281

phenylalanine for stilbene biosynthesis, as well as for the phenylpropanoid

282

pathway in general (including lignin and lignan biosynthesis) and for protein

283

synthesis. The shikimate pathway starts with the conversion of

284

phosphoenolpyruvate and erythrose 4-phosphate to 3-deoxy-D-arabino-

285

heptulosonate-7-phosphate (DAHP) by the enzyme DAHP synthase (DAHPS).


286

DAHPS is thought to be the rate-controlling enzyme of the shikimate pathway

287

(Tzin et al., 2012), and we observed 4 to 7 fold upregulation in TZ compared to

288

SW for TCs encoding DAHPS. Several other TCs encoding enzymes of the


289

shikimate pathway were upregulated in TZ, including those for 3-

290

dehydroquinate synthase (4 fold) and shikimate kinase (4-7 fold) and finally

291

chorismate mutase and prephenate dehydratase, 5 and 10-13 fold, respectively

292

(Figure 2, Table S6).

293 294

Further upstream, transcripts encoding phosphoenolpyruvate carboxykinase

295

was upregulated 5 to 8 fold in TZ. This enzyme converts oxaloacetate to

296

phosphoenolpyruvate in gluconeogenesis. TCs for enzymes involved in

297

glycolysis and the oxidative pentose phosphate pathway also showed

298

upregulation and finally transcripts for sucrose synthase were induced with a 2

299

to 5 fold change (Figure 2, Table S6). Activity of sucrose synthase is often taken

300

as indication of metabolic activity (Winter and Huber, 2000), showing that

301

metabolite production is taking place in the TZ before it is programmed for cell

302

death (see below). It further indicates that breakdown of sugar is then

303

channelled to the downstream processes like stilbene biosynthesis both for

304

building the carbon backbone and providing metabolic energy for the process.

305

Our transcriptome data show that stilbene biosynthesis definitely takes place in

306

situ in the TZ.

307 308

Plant hormones in the transition zone

309

The plant hormones ethylene and auxin have been suggested to be involved in

310

HW formation particularly their ratio being important (Hillis, 1968; Shain and

311

Hillis, 1973). Indeed, for TCs encoding of 1-aminocyclopropane-1-carboxylate

312

oxidase (ACO) we observed the highest differential expression ratio between

313

SW and TZ, an induction of 36-61 fold (Table S6). ACC synthase (ACS) was

314

not induced, but S-adenosylmethionine (SAM) synthase was expressed 3-9

315

times more strongly in TZ than in SW. Note that SAM synthase provides

316

substrate also for methylation of pinosylvin. Transcripts relating to auxin

317

biosynthesis or transport were not up or downregulated in TZ, but a group of

318

auxin response and auxin related or induced transcripts were downregulated

319

with an average of 2 fold change between TZ and SW (Table S6). Gibberellins

320

have not been suggested to be involved in HW formation. Beauchamp (2011)


321

was able to detect the inactive precursor GA24 in both SW and TZ, but not the

322

active forms or the 2-hydroxylated inactive forms. Still, one of the induced

323

transcripts encoding GA 2-oxidase (TC171916) was 16 times upregulated in TZ

324

compared to SW.

325 326

Dehydration and programmed cell death during heartwood formation

327

Moisture content in the TZ is low but still higher than in the HW. We observed

328

that a group of desiccation-related (DPR) transcripts was upregulated with fold

329

change of 23 to 31 in TZ (Table S6). DPR transcripts have been suggested to

330

be involved in tolerance of desiccation in plants (Zha et al., 2013; Piatkowski et

331

al., 1990). A group of aquaporin-like transcripts, on the other hand, were

332

downregulated (Table S6). Aquaporins are involved in regulation of cell-to-cell

333

water transport and in maintaining water homeostasis in plants (Bienert and

334

Chaumont, 2011; Hachez et al., 2006). Upregulation of the DPR and

335

downregulation of aquaporin transcripts may relate to the fact that water is

336

withdrawn from the TZ during HW formation. Transcripts encoding an S1-like

337

nuclease and a bifunctional nuclease (BFN) were highly induced in TZ with 13

338

and 20 fold change (Table S6) compared to SW. Both nucleases have been

339

described to be involved in PCD (Bollhöner et al., 2012; Farage-Barhom et al.,

340

2008). Upregulation of these transcripts indicated that PCD was taking place in

341

TZ.

342 343

Cell wall modification during heartwood formation

344

Cell wall modification and lignification related enzymes were upregulated during

345

HW formation. Lignification has been shown to be part of the HW formation

346

process in Scots pine (Bergström, 2003). Transcripts encoding enzymes

347

involved in lignification were upregulated in TZ at an average of 4 to 16 fold

348

(Figure 2, Table S6). Transcripts relating to cell wall carbohydrate modification

349

were expressed at an average 12 fold higher in TZ than in SW (Table S6).

350


351

Transcription factors in the transition zone

352

Two TCs encoding transcription factors, a MYB-like transcription factor and a

353

NAC domain-containing protein, were expressed 12 and 6 fold higher,

354

respectively, in TZ compared to SW (Table S6). Transcription factors of the

355

MYB family are often key regulators of secondary metabolite pathways (Dubos

356

et al., 2010). NAC domain proteins are involved in secondary cell wall

357

biosynthesis and PCD (Nakano et al., 2015; Bollhöner et al., 2012).

358 359

Resin acids and plant defence

360

Resin acids are conifer defence barriers to biotic challenges. They can be found

361

in the bark, in SW and in HW. According to Harju and colleagues (2003), HW

362

contains four times more resin acids on a volume basis than SW. It was

363

therefore surprising that transcripts involved in terpenoid biosynthesis and its

364

upstream pathways were neither up nor downregulated in the June

365

transcriptome data. In fact, very few reads mapped to these TCs (with an

366

average 95 % (and up to 99 %) identity with published diterpene synthase (Ro

367

and Bohlmann, 2006) and abietadienol/abietadienal oxidase (CYP720B)

368

(Geisler et al., 2016) sequences, indicating that constitutive synthesis of resin

369

acids was not taking place either (Table S7). On the other hand, a set of

370

pathogenesis-related transcripts were upregulated 5 to 20 fold (Table S6).

371 372

A set of unknown transcripts was also upregulated and downregulated in our

373

transcriptome data, sometimes up to 12 fold (Table S6).

374 375

Year round expression profiles

376

The Scots pine RNA-Seq data collected in June shed light to the metabolic

377

activities in the TZ during active growth of the tree. It did not answer the

378

question of when, during the yearly cycle, HW formation is most active, and left

379

completely open both the time and place of resin acid biosynthesis. Based on

380

the June results, we selected transcripts upregulated in TZ and apparently

381

related to HW formation as well as to resin acid biosynthesis for a year round

382

study using quantitative RT-PCR (qPCR) (Table S8).


383 384

We studied the expression profiles in the SW and TZ of increment cores of two

385

trees, harvested monthly in Punkaharju from March 2011 to March 2012. The

386

January 2012 increment cores were lost due to fragmentation of the frozen

387

tissue upon removal from the drill core.

388 389

PCR efficiency was determined for each primer pair used in the analysis, so the

390

qPCR results reflect relative transcript levels also between different genes.

391

Total RNA yields of the test trees varied, but not systematically throughout the

392

year (Table S9). Level of actin and histone transcripts compared to ribosomal

393

RNA (i.e., total RNA amount) in the extracted RNA samples varied remarkably

394

little (Table S10). Level of each selected transcript was expressed relative to the

395

geometric mean of the actin and histone transcripts and is shown throughout

396

the year in Figures 3 and 4.

397 398

We determined by qPCR expression levels of the selected transcripts also in

399

the SW and TZ RNA samples of the four trees from which the RNA-Seq data

400

was generated and differential expression was calculated, The two data sets

401

are in good agreement (Table S11).

402 403

Pinosylvin biosynthesis pathway transcripts showed a concerted

404

expression profile

405

The year round expression profiles of the two analysed trees varied, so they

406

were not averaged but are shown separately. Still, many processes important

407

for heartwood formation shared correlated patterns (Figures 3 and 4). The

408

expression of the stilbene biosynthesis pathway transcripts (PAL, 4CL, STS,

409

OMT) showed statistically significant correlation (Figure 5). Shikimate pathway

410

(DAHPS) followed the stilbene pathway closely, as did sucrose synthase

411

transcript levels. Transcript levels for all these markers in TZ increased in the

412

spring (Figure 3). Subsequently, they all dropped in June and showed recovery

413

during the summer months. PAL, 4CL, STS and OMT transcripts were

414

expressed at very low level in the SW throughout the summer months. The


415

expression level of these transcripts differed between the two trees in autumn

416

months and, interestingly, the SW followed the pattern although at a several fold

417

lower level. The level of all of these transcripts showed increase during the


418

winter months in both the TZ and the SW, even when the temperature was

419

constantly below the freezing point (Figure 4H).

420 421

Expression profiles of plant hormone related transcripts varied

422

Many studies have concluded that ethylene would play an important role in HW

423

formation. In the June 2011 RNA-Seq data we observed that ACO was the

424

number one differentially expressed transcript in TZ, compared to SW. The year

425

round qPCR data showed that ACO very strongly reacted to as yet undefined

426

stimuli throughout the sampling scenes, sometimes more strongly in SW and

427

very differently in the two trees sampled (Figure 4A). The observed differential

428

expression of ACO in the RNA-Seq samples seems not to be a typical pattern

429

during heartwood formation.

430 431

The expression level of the auxin responsive (AUX) transcript in the TZ was

432

generally very low, and ceased completely in the winter (Figure 4B). This

433

transcript was chosen since it was downregulated in TZ in the RNA-Seq data,

434

and it indeed showed lower expression in TZ throughout the year.

435


436

The two transcription factors shared a similar profile

437

Expression of the MYB-like and NAC domain-containing protein encoding

438

transcripts were higher in TZ in both trees throughout the year (Figure 4C and

439

D). In both trees their expression pattern correlated very clearly with the stilbene

440

synthase pathway transcripts as well as those for DAHPS and sucrose

441

synthase (Figure 5).

442 443

Programmed cell death

444

A very clear feature of the PCD related BFN transcript was that it was absent in

445

SW. In both trees, the transcript in TZ showed increased accumulation in May,


446

a trough in June and gradual increase through the late summer, in synchrony

447

with the stilbene pathway. BFN was not following the increased stilbene

448

biosynthesis peaks in late autumn and winter, suggesting that different signals

449

were inducing the stilbene pathway at that time (Figure 4E).

450 451

Expression profiles of resin acid biosynthesis genes

452

Transcripts involved in terpenoid biosynthesis were expressed at low level in

453

the June transcriptome data, both in SW and TZ. Excitingly, we noticed that the

454

transcript levels of diterpene synthase and abietadienol/abietadienal oxidase

455

(CYP720B) encoding transcripts (Figure 4F and G) in both SW and TZ started

456

to rise in early spring, dropped in late spring or early summer, and remained low

457

during the summer, autumn and winter. The expression level for these

458

transcripts was greater in the SW than in the TZ. The pattern was similar in both

459

trees.

460 461

Correlation between months of the year

462

We also calculated correlations between each sampling month for the

463

expression pattern of the selected genes (Figure S4). We expected that highest

464

correlation would be observed between successive months, and the TZ

465

samples of both trees followed this pattern roughly. On the contrary, the SW

466

samples showed strong correlations across the months, i.e. between months

467

that are far apart in the yearly cycle (Figure S4). Interestingly, this means that

468

the TZ follows seasonal changes for its gene expression patterns much more

469

strongly than the SW does.

470 471


472

Discussion

473 474

Scots pine (Pinus sylvestris L.) wood is extensively used as sawn timber

475

because of its strength, straightness, and visual appearance. Particularly the

476

HW of Scots pine is valued for its properties such as dimensional stability and

477

natural durability. Extractives of HW give, e.g., colour and low hygroscopicity to

478

HW timber and are key determinants of the passive defence, i.e., resistance

479

against degradation of wood by microorganisms (Nakada and Fukatsu, 2012;

480

Harju and Venäläinen, 2006; Leinonen et al., 2008). Sampling of Scots pine

481

transcriptomes in the SW and TZ to HW by RNA sequencing in June and qPCR

482

throughout the year allowed us to identify the place and time during the yearly

483

cycle when these important components are synthesised and HW is formed.

484 485

From carbon skeletons to the heartwood phenolics

486

During HW formation, reserves of the SW are consumed, in Scots pine mainly

487

starch and triglycerides (Magel, 2000; Bergström, 2003; Nakada and Fukatsu,

488

2012). Magel (2000) concluded that the most important contributor for sucrose

489

cleavage was sucrose synthase followed by invertase. Upregulation of

490

transcripts in TZ for both of these enzymes was observed in our data. In

491

addition, transcripts relating both to glycolysis and oxidative pentose phosphate

492

pathway were expressed at an average 2 fold more in TZ than SW, suggesting

493

that the breakdown of sugars is channelled downstream for synthesis of

494

precursors to the shikimate pathway, and finally to the pinosylvin pathway.

495 496

Phosphoenolpyruvate carboxykinase, an enzyme that converts oxaloacetate to

497

phosphoenolpyruvate (Bowsher et al., 2008), was found upregulated strongly in

498

TZ. In plants, oxaloacetate is generated from malate by malate dehydrogenase

499

in the cytosol. Malate is synthesised from breakdown of plant triglycerides via

500

the glyoxylate pathway in plant fat storage tissues (glyoxysomes). Study of

501

Saranpää and Höll (1989) and Bergstöm (2003) showed that amount of free

502

fatty acids was increased in the TZ and HW but not in SW. In addition, lipophilic

503

droplets were observed in cells of TZ associated with HW formation (Taylor et 20 Downloaded from www.plantphysiol.org on January 5, 2017 - Published by www.plantphysiol.org Copyright © 2016 American Society of Plant Biologists. All rights reserved.

504

al., 2002; Kampe and Magel, 2013). Upregulation of phosphoenolpyruvate

505

carboxykinase may indicate that triglycerides are used not only as a direct

506

source of heartwood extractives (free fatty acids), but also as energy and

507

carbon source for stilbenes.

508 509

During HW formation, the whole shikimate pathway is expressed at an average

510

of 2 fold higher in TZ and the rate limiting enzyme DAHPS up to 7 fold higher.

511

This pathway connects the primary metabolism (aromatic amino acid

512

biosynthesis) to the phenylpropanoid pathway, including the stilbene and lignin

513

biosynthesis pathways. Both of these branches are active in TZ, the stilbene

514

pathway being responsible for generation of the pine specific heartwood

515

extractives pinosylvin and its monomethyl ether. Pinosylvin lacks hydroxylation

516

of the phenylalanine derived aromatic ring, similar to cinnamic acid.

517

Furthermore, the Scots pine STS prefers cinnamoyl-CoA to 4-coumaroyl-CoA

518

(Fliegmann et al., 1992; Kodan et al., 2002). Only few plant 4CLs activate

519

cinnamic acid (Kumar and Ellis, 2003) and in some plant species activity of

520

enzymes with a more distant relationship to 4CL have been shown to be

521

involved in cinnamoyl-CoA formation (Gaid et al., 2012; Klempien et al., 2012).

522

We observed a possible 4CL-like candidate from our transcriptome data that

523

might be specific to pine pinosylvin biosynthesis during HW formation.

524

Nevertheless, characterization its enzymatic specificity is needed to address

525

this hypothesis.

526 527

The end product of pine stilbene biosynthesis is PSME, catalysed by PMT. A

528

Scots pine PMT was purified and partially characterised by Chiron et al.

529

(2000a). However, no transcript encoding this PMT was upregulated in our

530

transcriptome data, but instead another O-methyltransferase encoding

531

transcript was strongly induced in TZ. This finding suggests that the PMT

532

characterised by Chiron and colleagues might not be the correct Scots pine

533

PMT for pine HW extractive biosynthesis.

534 535

The strong upregulation of transcripts encoding enzymes leading to stilbene


536

biosynthesis in TZ indicates that these HW extractives are synthesised in situ in

537

the TZ, gaining the carbon skeleton from sucrose. In this respect, Scots pine

538

heartwood formation is of type I, or the Robinia-type (Magel, 2000).

539 540

Biosynthesis of heartwood resin acids

541

In addition to stilbenes and free fatty acids, Scots pine HW contains abundantly

542

resin acids (Harju et al., 2002; Fries et al., 2000). Contrary to expectations,

543

resin acid biosynthesis was neither up nor downregulated in TZ compared to

544

SW. Lorio (1986) proposed the concept of a balance between growth and

545

differentiation for resin acid biosynthesis in southern pines. He suggested that

546

during the growing season, the sink of photosynthates is preferentially the

547

growth process, and only after the growing period, resources are available for

548

resin acid production. The year round expression profile of transcripts encoding

549

diterpene synthase and abietadienol/abietadienal oxidase (CYP720B) showed

550

very low expression during summer months but a strong peak in early spring.

551

Throughout the year, resin acid biosynthesis related transcripts accumulated

552

more in SW than in TZ, still resin acids accumulate four times higher in HW than

553

in SW (Venäläinen et al., 2003). Resin acids do not seem to be biosynthesised

554

in situ in the TZ like stilbenes are, but instead they get loaded from SW to HW

555

during HW formation. Therefore, from the resin acid biosynthesis point of view,

556

Scots pine HW is of type II, or the Juglans type (Magel, 2000).

557 558

Balance between ethylene and auxin

559

The plant hormones ethylene and auxin have been suggested to be closely

560

related to HW formation. A recent study by Beauchamp (2011) reported that

561

auxin concentration was either reduced or below level of detection in the TZ, in

562

line with Hillis’s (1968) description that depletion of auxin content is associated

563

with HW formation. Shain and Hillis (1973) suggested that the balance between

564

ethylene and auxin during HW formation is important. Indeed, transcripts

565

encoding ACO were most strongly differentially expressed in TZ compared to

566

SW. On the other hand, transcripts for ACS, the rate limiting enzyme for

567

ethylene biosynthesis in most plants (Woeste et al., 1999; Hudgins et al., 2006),


568

was not induced. This may be a reflection of the fact that ACS regulation takes

569

predominantly place at protein level (Chae and Kieber, 2005), which we did not

570

address in our experiments. Transcripts for SAM synthase, the second

571

substrate for ACS (Yoon and Kieber, 2013), were induced in TZ, but SAM is

572

also needed for pinosylvin biosynthesis. The importance of ethylene in HW

573

formation, at least as a sufficient inducer, can however be questioned since

574

ACO was expressed erratically with little correlation between trees or zones.

575

Although we designed repeated sampling of the trees to avoid wound induction

576

(Figure S5), we cannot exclude that this caused a reaction in ACO. Transcripts

577

related to auxin biosynthesis were not differentially expressed between SW and

578

TZ. Again, regulation of auxin levels prominently take place after biosynthesis of

579

the hormone (Woodward and Bartel, 2005). Nevertheless, several auxin

580

response transcription factors were downregulated in TZ.

581 582

Transcriptional control in the transition zone

583

Two transcription factor encoding transcripts, for a NAC domain and a MYB-like

584

protein, were found to be significantly upregulated in the TZ and their

585

expression correlated with the stilbene pathway transcripts throughout the year

586

in both sampled trees. NAC domain transcription factors have been reported to

587

be involved in developmental processes, for example, in secondary cell wall

588

biosynthesis as well as biotic and abiotic stresses, acting as master switches to

589

regulate downstream genes (Nakano et al., 2015). The upregulated NAC

590

domain protein is however not similar to the white spruce (Picea glauca)

591

secondary cell wall biosynthesis NAC transcription factors, encoded by

592

PgNAC4 and PgNAC7 (Duval et al., 2014). Also the MYB-like transcription

593

factor is not particularly similar to the Pinus taeda secondary cell wall

594

biogenesis related MYB transcription factors that were isolated from

595

differentiating xylem, PtMYB1 (Patzlaff et al., 2003b) and PtMYB4 (Patzlaff et

596

al., 2003a). Still, as MYB transcription factors are in many cases main

597

regulators of secondary metabolism in plants, involvement of the upregulated

598

MYB transcript as well as the NAC transcript in stilbene or lignin biosynthesis

599

deserves a closer look.


600 601

Cells in the transition zone prepare for death

602

Pine heartwood is dead tissue and during HW formation the cells in the TZ not

603

only are responsible for filling the HW with extractives, but they also prepare for

604

their own death.

605 606

Moisture content in the TZ is lower than in SW (Hillis, 1987; Bergström, 2003).

607

The upregulation of transcripts encoding desiccation-related proteins in TZ

608

could result from the fact that water is withdrawn from the cells. The role of plant

609

DPR proteins may help the cells to tolerate the dehydration condition and allow

610

them to continue with the still necessary physiological processes during HW

611

formation. Downregulation of a group of aquaporin-like proteins may also relate

612

to dehydration in the TZ.

613 614

Transcripts encoding enzymes necessary for lignification, as well as groups of

615

cell wall polysaccharide modifying enzymes were expressed in the TZ during

616

HW formation. The cell walls of ray parenchyma cells in the TZ are indeed

617

undergoing lignification, as observed by Bergström (2003) and Yamamoto

618

(1982).

619 620

A chitinase and set of plant pathogenesis-related transcripts were upregulated

621

in the TZ. In general, these plant defence related proteins are induced in

622

response to biotic and abiotic stresses (Van Loon, 1997). Chitinases have been

623

reported to be involved not only in plant defence but also in developmental

624

processes, such as secondary cell wall lignification, and they are induced by

625

plant hormones (Grover, 2012). Yoshida et al. (2012) and Yang et al. (2004)

626

also reported that a set of pathogenesis-related proteins are highly expressed in

627

the TZ of Cryptomeria and Robinia, respectively.

628 629

The storage compounds in the TZ are depleted and consumed and the

630

parenchyma cell walls are undergoing modification. At the end of the process,

631

all physiological processes will eventually cease.


632 633

Programmed cell death is the key for heartwood formation

634

HW formation can been seen as a form of tissue senescence (Spicer, 2005;

635

Kampe and Magel, 2013). The BFN is an endonuclease similar to S1-like

636

nucleases (Farage-Barhom et al., 2008), is induced during senescence and is

637

involved in plant developmental PCD (Farage-Barhom et al., 2008; Bollhöner et

638

al., 2012). Farage-Barhom et al. (2008) demonstrated that the arabidopsis

639

BFN1 moves to the nucleus during leaf senescence and the protein was

640

detected from fragmented nuclei in the late stage of leaf senescence (Bollhöner

641

et al., 2012). Degradation of nuclei of ray parenchyma cells in the TZ has been

642

observed in past studies (Taylor et al., 2002; Spicer, 2005; Nakada and

643

Fukatsu, 2012). Thus, the upregulation of the BFN transcript in TZ, compared to

644

SW, is in line with the fact that PCD is involved in HW formation. Expression of

645

the PCD transcript BFN in TZ appears to be a key marker to distinguish

646

developmentally programmed HW formation in Scots pine from stress induced

647

extractive biosynthesis.

648 649

Timing of heartwood formation in Scots pine

650

The year round expression profiles of selected transcripts related to primary and

651

secondary metabolic pathways, the pinosylvin biosynthesis pathway, plant

652

hormones, transcription factors, programmed cell death and resin acid

653

biosynthesis pathway were investigated with the qPCR approach in order to

654

understand if there is a preferential time during the year when HW is formed.

655 656

As a general observation, we noticed that the expression levels in each tested

657

trees varied in the autumn and in the winter, but shared a pattern during the

658

summer months. It is apparent that the trees are responding both to shared

659

developmental programmes (e.g., HW formation), as well as to local challenges

660

that may be different for each tree individual responded to differently by different

661

genotypes. The two trees analysed were growing close to each other but were

662

not related.

663


664

The year round expression profiles of the two transcripts related to resin acid

665

biosynthesis were well in synchrony. The genes were expressed, and

666

presumably synthesis proceeded, in early spring and was practically shut down

667

for the summer months, in accordance to the concept suggested by Lorio

668

(1986). Resin acid biosynthesis transcripts were expressed in both SW and TZ,

669

but more strongly in SW, indicating that the compounds loaded in HW are

670

mainly not synthesised in situ in the TZ.

671 672

The ACO transcripts, strongly differentially expressed in TZ compared to SW

673

and related to the suggested ethylene involvement in HW formation, were

674

expressed at very high levels in both SW and TZ in one of the two trees. This

675

may reflect a challenge this tree met in the late summer. Usually, however,

676

regulation of ethylene synthesis is addressed to protein level control of ACS, not

677

ACO. The auxin responsive transcript was expressed in the spring and summer

678

in SW, with low levels in TZ throughout the year.

679 680

Transcripts responsible for stilbene biosynthesis, our signature pathway for HW

681

formation, were always more strongly present in TZ than in SW. There seems to

682

be a shared peak of activity in early summer, somewhat later than that for resin

683

acid biosynthesis. In late summer and in winter the two trees were behaving

684

nonsynchronously, but the pathway, including its upstream components of

685

sugar metabolism and aromatic acid biosynthesis, were reacting in concert.

686

Expression of the two transcription factor transcripts investigated strongly

687

correlated with expression of the stilbene pathway genes. An interesting

688

peculiarity is that transcripts for the stilbene pathway enzymes, and for the two

689

transcription factors, accumulated during the winter months even when the

690

maximum daily temperature kept well below zero for nearly two months (Figure

691

3 and 4H). This does not seem to be an artefact, other transcripts did not show

692

this and RNA yields from the tissues do not explain this phenomenon, nor does

693

the fraction of actin or histone transcripts compared to total RNA and used as

694

reference.

695


696

In a summary, we believe that we can answer the question when HW formation

697

takes place in Scots pine: Resin acids are formed early in the spring, followed

698

by stilbenes, whose synthesis is carried out through the summer. Programmed

699

cell death of TZ cells, i.e., their transformation into HW, takes place in autumn,

700

when growth has ceased. In November, everything is over and the outermost

701

dry zone of the tree trunk contains cells that prepare for the final active period in

702

their life, taking place the next summer.

703 704

Conclusion

705

Past studies have suggested that ethylene plays an important role in HW

706

formation. Upregulation of ACO in TZ seemed first to support this, but since

707

ACO was found to react strongly also in SW during the growth season, the

708

evidence is not conclusive.

709 710

The carbon skeleton for the HW stilbene biosynthesis is obtained by breaking

711

down sucrose and is channelled to pinosylvin biosynthesis through the

712

shikimate pathway. The result demonstrates that the Scots pine HW stilbenes

713

are synthesised in situ in the TZ.

714 715

Two transcription factors follow closely phenylpropanoid biosynthesis in the TZ,

716

leading not only to stilbenes but also to lignins. Finally, expression of a PCD

717

regulator distinguishes developmentally regulated heartwood stilbene

718

production from other activating signals (stilbenes are also stress metabolites).

719 720

Stilbene biosynthesis clearly takes place in TZ cells, but resin acids, another

721

prominent group of heartwood extractives are mainly synthesised in the

722

sapwood, based on steady state levels of transcripts encoding enzymes needed

723

for their biosynthesis. Therefore, heartwood formation is of type I from the

724

stilbene point of view, but type II from the resin acids point of view.

725 726

Our transcriptome data indicates that heartwood formation takes place during


727

the growth period, with different timing for resin acid and stilbene biosynthesis.

728 729

Materials and methods

730

Plant material for RNA Sequencing and qPCR studies

731

Increment cores for RNA-Seq were sampled in June 2011 from four 46 years

732

old Scots pine trees (T285, T328, T335 and T415) at Leppävirta progeny trial

733

stand number 30902 (62°25′N, 27°45′E), established by the Finnish Forest

734

Research Institute, METLA. On average, thirty increment cores (5 mm in

735

diameter) were randomly drilled through each tree trunk with a battery driven

736

increment core borer approximately at chest height level (130 cm) avoiding the

737

knots. The border between wet and dry wood was marked with a pencil upon

738

collection, after which the increment cores were frozen in dry ice and

739

transferred to the -80 °C laboratory freezer.

740 741

Increment cores for initial tests (including STS RT-PCR) throughout the summer

742

months June to August in 2010 were sampled from 35 years old Scots pine

743

trees at Punkaharju progeny trial stand number 62201 (61°49′N, 29°20′E) with

744

the aforementioned method (trees 11 and 20). Increment cores for the year

745

round expression study were sampled from other two 35-36 years old Scots

746

pine trees (trees C and D) at the same progeny trial in Punkaharju from March

747

2011 to March 2012. Samples were harvested in the middle of each month.

748

Four increment cores were drilled through the pith each month, about one cm

749

distance apart. During each sampling date, the sampling position was moved

750

laterally and longitudinally around the trunk (Figure S5). By this procedure we

751

minimised the effects of the previous drills on the samples. The sampling took

752

place from several interwhorls starting from around 180 cm downwards. The

753

increment cores were frozen in liquid nitrogen and stored at -80 °C.

754 755

Localisation of the transition zone

756

PS and PSME give strong fluorescence under UV excitation (Harju et al., 2009),

757

allowing straightforward visualisation of HW extractives. The frozen increment

758

cores were illuminated at wavelength 365 nm and photographed. The TZ


759

annual ring of each increment core was identified according both to the pencil

760

mark made in the field and the fluorescence image made in the laboratory. We

761

discarded increment cores where the pattern was not regular, for example when

762

stilbenes were present outside of the dry zone in SW, indicating a reaction to a

763

biotic or abiotic challenge.

764 765

Total RNA extraction

766

Annual rings from TZ and SW were excised with a scalpel while keeping the

767

increment core frozen. SW sample consisted of four to five pairs of annual rings

768

located at approximately four or five annual rings away from the TZ. Samples

769

were ground to fine powder in liquid nitrogen using first mortar and pestle and

770

subsequently a milling machine (Qscillating Mill MM400 Retsch GmbH,

771

Germany).

772 773

Total RNA isolation for SW and TZ samples was carried out as described by

774

Chang et al. (1993) with minor modifications. After overnight lithium chloride

775

precipitation, the RNA pellet was collected by centrifugation at 10 000 ×g for 30

776

minutes at 4 °C, washed with 70 % cold ethanol, air-dried for 3 minutes and

777

dissolved in 50 µL nuclease-free Milli-Q water. Genomic DNA was digested with

778

DNase I, followed by purification with RNeasy Plant Mini Kit (Qiagen, Germany)

779

according to the manufacturer’s protocol. The concentration and the quality of

780

the genomic DNA-free total RNA was assessed with spectrophotometer

781

(GeneQuant 1300 Eppendorf, Germany) and agarose gel electrophoresis.

782 783

Semiquantitative STS RT-PCR

784

Semiquantitative STS RT-PCR was performed for localisation of the TZ and

785

screening for an ideal of sampling time. The annual ring sampling and total RNA

786

isolation were performed as described above. First strand cDNA was

787

synthesised using 50 ng of total RNA with SuperScript III reverse transcriptase

788

(Invitrogen, USA) with oligo(dT)20 primers according to manufacturer’s

789

instructions in a 20 µL reaction volume. The STS RT-PCR was carried out with

790

the programme 94 °C for 2 min, 24 cycles of 94 °C for 30 s, 57 °C for 1 min and


791

72 °C for 1 min, and 72 °C for 7 min. Primers (Table S12) were designed to

792

span an exon-intron boundary with Primer3 software (Untergasser et al., 2007).

793 794

Sample preparation for RNA sequencing

795

Eight micrograms of genomic DNA free total RNA was selectively depleted from

796

ribosomal RNA (rRNA) according to the procedures described in RiboMinus

797

Plant Kit for RNA-Seq (Invitrogen, USA). 500 ng of rRNA-depleted RNA was

798

used for transcriptome library preparation. After the ligation of SOLiD adaptor

799

mix and reverse transcription, size fractionation of the transcriptome library was

800

performed with gel size selection of 150-200bp cDNA, followed by amplification

801

and purification according to the protocols described in SOLiD Total RNA-Seq

802

Kit (Invitrogen, USA). Next, paired-end sequencing was conducted with SOLiD

803

5500xl at Institute of Biotechnology, University of Helsinki.

804 805

Mapping of reads and data handling

806

The paired-end colour space reads of SW and TZ RNA-Seq libraries were

807

mapped against the Pinus EST collection, version 9.0 (The Gene Index

808

Databases, 2014) using the SHRiMP2 alignment tool (David et al., 2011) with

809

parameters as shown in Protocol S1. Results were obtained in the Sequence

810

Alignment/Map (SAM) format (Li et al., 2009) and then converted to sorted

811

Binary Alignment/Ma (BAM) format, duplicates were removed with Picard tools

812

(Wysoker et al., 2013). The abovementioned tasks were carried out on the

813

Finnish Grid Infrastructure clusters and supercluster servers in CSC-IT Center

814

for Science Ltd (Espoo, Finland).

815 816

Differential expression analysis

817

Tables of mapped read counts for SW and TZ RNA-Seq libraries were imported

818

to edgeR R session to build a DGEList object for differential analysis (Robinson

819

et al., 2010). Low expression transcripts of Pinus ESTs collection were filtered,

820

only transcripts that achieved at least eight counts per million (CPM) in at least

821

four libraries were kept. The trimmed mean of M-value normalization, dispersion

822

estimation and differential expression analysis was then carried out. List of the


823

statistically significant differentially expressed transcripts at false discovery rate

824

(FDR) of less than 0.05 were then exported to a text file for downstream

825

analysis.

826 827

The Pinus EST collection has annotations but as some of them are not

828

informative or are missing, we compared the sequences to databases using the

829

blastx algorithm (Altschul et al., 1997) and obtained annotation updates for

830

many cases. Original and updated annotations are shown in Tables S3 and S4.

831 832

Gene Ontology Enrichment analysis

833

Gene Ontology (GO) terms of Pinus EST collection was obtained using

834

Blast2GO (Conesa et al., 2005). GOSlim plant ontology mapping was then

835

carried out in order to reduce complexity of obtained GO annotations. GO term

836

enrichment analysis was performed for upregulated and downregulated

837

transcripts with GO annotated Pinus EST collection as referral using Fisher's

838

exact test with multiple testing correction of FDR function implemented in

839

Blast2GO. General enriched GO terms were filtered from result list at FDR