retrograde signalling caused by heritable mitochondrial ... · retrograde controls in chloroplasts...

1

Retrograde signalling caused by heritable mitochondrial dysfunction is partially mediated by ANAC017 and improves plant performance

Olivier Van Aken§, Ethan Ford, Ryan Lister, Shaobai Huang and A. Harvey

Millar

§ Corresponding author

ARC Centre of Excellence in Plant Energy Biology, Faculty of Science, Bayliss

Building M316, The University of Western Australia, 35 Stirling Highway, Crawley

6009, Western Australia, Australia.

Running title: ANAC017 mediates endogenous mitochondrial signalling

Keywords: Arabidopsis thaliana, mitochondria, retrograde signalling, stress

signalling

RNA-seq ArrayExpress Accession Number: E-MTAB-4655

The author responsible for distribution of materials integral to the findings

presented in this article is

Olivier Van Aken

Tel +61 8 6488 2112

Fax +61 8 6488 4401

[email protected]

Email addresses: [email protected]; [email protected];

[email protected]; [email protected]

mailto:[email protected]




2

Word count: 8368

3

Summary

Mitochondria are crucial for plant viability and are able to communicate

information on their functional status to the cellular nucleus via retrograde

signalling, thereby affecting gene expression. It is currently unclear if retrograde

signalling in response to constitutive mitochondrial biogenesis defects is

mediated by the same pathways as those triggered during acute mitochondrial

dysfunction. Furthermore, it is unknown if retrograde signalling can effectively

improve plant performance when mitochondrial function is constitutively impaired.

Here we show that retrograde signalling in mutants defective in mitochondrial

proteins RNA polymerase rpotmp or prohibitin atphb3 can be suppressed by

knocking out the transcription factor ANAC017. Genome-wide RNA-seq

expression analysis revealed that ANAC017 is almost solely responsible for the

most dramatic transcriptional changes common to rpotmp and atphb3 mutants,

regulating classical marker genes such as alternative oxidase 1a (AOX1a) and

also previously-uncharacterised DUF295 genes that appear to be new retrograde

markers. In contrast, ANAC017 does not regulate intra-mitochondrial gene

expression or transcriptional changes unique to either rpotmp or atphb3

genotype, suggesting the existence of currently unknown signalling cascades.

The data show that ANAC017 function extends beyond common retrograde

transcriptional responses and affects downstream protein abundance and

enzyme activity of alternative oxidase, as well as steady state energy metabolism

in atphb3 plants. Furthermore, detailed growth analysis revealed that ANAC017-

dependent retrograde signalling provides benefits for growth and productivity in

plants with mitochondrial defects. In conclusion, ANAC017 plays a key role in

both biogenic and operational mitochondrial retrograde signalling, and improves

plant performance when mitochondrial function is constitutively impaired.

4

Introduction

Energy metabolism in plants is highly adaptive and requires the optimal

functioning of cellular organelles including mitochondria, chloroplasts and

peroxisomes. It is well established that changes in the environment and

application of stress transiently affect mitochondrial function at the transcriptional,

translational and post-translational levels (Giraud et al., 2012). To optimise

mitochondrial function in response to a specific set of circumstances, two-

directional communication between the mitochondria and the nucleus is required

(Rhoads and Subbaiah 2007). The feedback that the mitochondrion provides to

the nucleus is commonly known as retrograde signalling, and results in

transcriptional changes that are presumed to address the initial defects (Leister

2012, Schwarzländer et al., 2012, Ng et al., 2014).

Two main means for inducing and studying mitochondrial retrograde signalling

have been reported. On one hand, studies have used respiratory inhibitors to

impair mitochondrial functions and thus induce short-term activation of

mitochondrial retrograde pathways; these include antimycin A (complex III),

rotenone (complex I) and monofluoroacetate (aconitase inhibitor) (Sweetlove et

al., 2002, Dojcinovic et al., 2005, Giraud et al., 2009, Schwarzländer et al., 2012,

Umbach et al., 2012, Ng et al., 2013b). On the other hand, mutants with genetic

defects in mitochondrial functions that exhibit responses in a similar set of

mitochondrial retrograde responsive genes have been studied (Van Aken et al.,

2007, Kuhn et al., 2009, Meyer et al., 2009, Van Aken and Whelan 2012). While

marker genes such as alternative oxidase 1a respond to both chemical and

genetic inductions of retrograde signalling, it is unclear if similar or overlapping

signalling systems are in control of this response. For instance treatment with

antimycin A results in transient gene expression changes that return to basal

levels within 24h (Ivanova et al., 2014), while retrograde marker genes in

mitochondrial mutants remain induced for weeks and may remain elevated

throughout the lifetime of the plants (Van Aken et al., 2007).

Retrograde signalling in plant organelles has previously been categorised in

either operational control (adaptation of organellar function to acute changes and

5

malfunctions) or biogenic control (response to defects in the biogenesis of the

organelle during development) (Pogson et al., 2008). Operational and biogenic

retrograde controls in chloroplasts have been shown to be regulated by a range

of different pathways (Chan et al., 2015). Whether operational and biogenic

retrograde controls of mitochondrial function are overlapping or completely

separate is currently poorly understood.

Regulatory factors that are known to control mitochondrial retrograde signalling

have been identified using chemical treatments such as antimycin A. Several

transcription factors of the WRKY family (AtWRKY15, AtWRKY40, AtWRKY63),

NAC family (ANAC013, ANAC017) and AP2/ERF family (ABI4) have been

identified in this way (Vanderauwera et al., 2012, De Clercq et al., 2013, Ng et

al., 2013a, Ng et al., 2013b, Van Aken et al., 2013, Ivanova et al., 2014).

However, to our knowledge, no studies have examined if such transcription

factors are also involved in transmitting signals from constitutively dysfunctional

mitochondria. Furthermore, it is unknown whether mitochondrial retrograde

signalling has tangible benefits for plant growth or performance. Solving these

questions is important to define if retrograde responses result in long term benefit,

or are largely non-constructive distress signals of short-term malfunctions.

Here we address the role and regulation of retrograde signalling in response to

constitutive mitochondrial dysfunctions using mutants that are defective in

mitochondrial proteins. One mutant is defective in prohibitin 3 (atphb3), an inner

mitochondrial membrane protein that is important for mitochondrial function and

morphology, most likely by acting as a scaffold for membrane protein complexes,

such as the electron transport chain (ETC) (Van Aken et al., 2007, Van Aken et

al., 2010, Piechota et al., 2015). The second mutant is defective in organellar

RNA polymerase (rpotmp), resulting in reduced abundance and activity of

mitochondrial ETC complexes I and IV (Kuhn et al., 2009). Besides mitochondrial

defects, both atphb3 and rpotmp mutants show significant growth retardation and

altered (retrograde) gene expression profiles. These mutants were selected

because previous studies showed that retrograde transcriptional regulation was

induced strongly in both mutants, but the underlying defects within the

6

mitochondria were clearly different (RNA polymerase activity vs. protein

scaffolding). In this study, we show that the transcription factor ANAC017 is of

key importance for these mitochondrial retrograde responses, and regulates not

only gene expression, but also plant growth and respiratory metabolism.

Results

Loss of ANAC017 in mitochondrial mutants impairs retrograde gene expression

Previous studies described constitutive changes in gene expression in a variety

of mutants with nuclear mutations resulting in mitochondrial defects (Van Aken

and Whelan 2012). This study aimed to assess whether the transcription factor

ANAC017 was involved in those retrograde transcriptional responses.

Arabidopsis lines with mutations in important nucleus-encoded mitochondrial

proteins (prohibitin 3 atphb3 and organellar RNA polymerase rpotmp) were

crossed with mutants defective in ANAC017 caused by a premature STOP codon

(anac017, rao2-1) (Van Aken et al., 2007, Kuhn et al., 2009, Ng et al., 2013b).

Double mutants were selected with homozygous T-DNA insertions in the genes

encoding the mitochondrial protein and homozygous non-sense mutations in

ANAC017 (atphb3 anac017 and rpotmp anac017). To minimise any indirect

effects caused by background mutations, single mutants with two wild-type

ANAC017 alleles were selected from the same F2 populations as the double

mutants (atphb3 ANAC017 and rpotmp ANAC017). These fraternal single

mutants were used as controls for experiments.

Transcript levels of several previously described mitochondrial retrograde

signalling marker genes (Van Aken and Whelan 2012) were measured by qRT-

PCR in two week-old plants in the different wild type (Col-0) and mitochondrial

mutant combinations (Figure 1). As expected, the fraternal single mutants

(rpotmp ANAC017 and atphb3 ANAC017) showed significantly induced transcript

levels of indole-3-butyric acid UDP-glycosyltransferase UGT74E2 (Tognetti et al.,

2010), mitochondrial stress marker gene At2g41730 and alternative oxidase

7

AOX1a (Zarkovic et al., 2005). Overall, the atphb3 mutation caused higher

induction levels of marker genes than the rpotmp mutation. In contrast, the

increased expression levels were almost completely abolished in both double

mutant lines (Figure 1). Statistically significant reductions in gene expression

between the respective single and double mutants were measured in each case.

No more significant differences could be observed between the rpotmp anac017

mutants and Col-0, indicating an almost complete reversal of this retrograde

molecular phenotype. Also atphb3 anac017 mutants showed a near-reversal of

the transcript inductions, although slightly elevated transcript levels remained for

At2g41730 and AOX1a compared to Col-0. Some redundancy of ANAC017 with

closely related transcription factors such as ANAC013 and ANAC053 has been

reported previously (De Clercq et al., 2013, Van Aken et al., 2016). This partial

redundancy likely explains the strong, but not complete, repression of retrograde

gene expression in atphb3 anac017 mutants.

To verify that the absence of functional ANAC017 was responsible for the loss of

retrograde signalling, rpotmp anac017 mutant plants were complemented with a

functional ANAC017 expression construct. Transcript levels of mitochondrial

retrograde signalling were tested in independent complemented lines and

compared to Col-0, rpotmp ANAC017 and rpotmp anac017 plants (Figure 1B).

Transcript levels of the retrograde marker genes were significantly induced in

rpotmp ANAC017 and in the complemented lines compared to Col-0, as opposed

to rpotmp anac017 plants that showed no significant induction of these marker

genes. These results indicate that ANAC017 plays a constitutive role in

transmitting information from dysfunctional mitochondria to the nucleus.

Genome-wide analysis of transcript changes in mitochondrial mutants

To examine the genome-wide effects on transcript levels, RNAseq analysis was

performed on Col-0, rpotmp ANAC017, rpotmp anac017, atphb3 ANAC017 and

atphb3 anac017. Transcripts were considered to be significantly differentially

8

expressed between genotypes when padj<0.05 (after multiple testing correction)

and fold change >1.5x (DEGs: differentially expressed genes). As a starting point,

the differentially expressed transcripts between the rpotmp ANAC017 vs Col-0

and atphb3 ANAC017 vs Col-0 were determined (Supplementary Table I). 579

DEGs were found in rpotmp ANAC017 vs Col-0, of which 276 were up-regulated

and 303 were downregulated. These included 32 mitochondrially-encoded genes

(24 up-, 8 downregulated) and 1 plastidially-encoded gene (RPS18, 1.55x). In

atphb3 ANAC017 vs Col-0, 559 DEGs were found, of which 277 were up- and,

282 were downregulated. These included 8 upregulated mitochondrially-encoded

genes.

In total 98 DEGs were common to both rpotmp ANAC017 and atphb3 ANAC017

genotypes vs Col-0 (64 consistently up-, 20 consistently downregulated) (Figure

2, Supplementary Table II). These 98 transcripts are highly enriched for

transcripts encoding mitochondrial proteins: 6 mitochondrially-encoded

transcripts, 20 nucleus-encoded transcripts encoding verified or predicted

mitochondrial proteins (e.g. AOX1a, NAD(P)H dehydrogenases NDB3 and

NDB4, stomatin-like proteins AtSLP1 and AtSLP2, mitochondrial heat shock

proteins sHSP23.5, mtHsc70-1 and GrpE1, prohibitin AtPHB1, mitochondrial

import component AtTIM17-1) (Van Aken et al., 2007, Van Aken et al., 2009,

Smith et al., 2011, Gehl et al., 2014, Wang et al., 2014). Furthermore 10

transcripts were mapped to a pericentromeric region of chromosome 2, which

contains a 99 % identical duplication of the Arabidopsis mtDNA (Stupar et al.,

2001). For instance, the nuclear pseudogene At2g07712 is >99% identical to the

full-length AtM00520 Type-II intron maturase MatR, which is also differentially

expressed in both mitochondrial mutants. Given the relatively recent occurrence

of this duplication and low mutation rate, it was very difficult to differentiate

between the nuclear and mitochondrial forms. It seems more likely that these

virtually indistinguishable differentially expressed transcripts are mitochondria-

derived, rather than originating from chromosome 2.

Several genes linked to energy metabolism are also commonly differential

including glycolytic enzymes (fructose-bisphosphate aldolase FBA6 and 2,3-

9

biphosphoglycerate-independent phosphoglycerate mutase IPGAM2), amino

acid metabolism enzymes (mitochondrial alanine aminotransferase 2 and proline

dehydrogenase, aspartate aminotransferase 2). Other groups of genes

represented in the DEGs are related to anthocyanin production (2), aquaporins

(3), transport (6), cell wall composition (2), transcription factors (2, including

ANAC017-related ANAC013) and auxin homeostasis (UGT74E2 and Pin-likes 4).

RNAseq analysis reveals DUF295 transcripts as novel markers for mitochondrial dysfunction

Overall the DEGs identified by RNAseq are similar to previously published results

obtained with Affymetrix ATH1 microarrays (Van Aken et al., 2007, Kuhn et al.,

2009). For example, of the 443 DEGs in atphb3 mutants found by microarray

analysis (padj <0.05; FC>1.5x), 105 are in common with the new RNAseq

analysis. 44 of the 105 overlapping genes are also differential in rpotmp mutants

based on the RNAseq data, and thus represent a highly repeatable set of core

genes affected by mitochondrial dysfunction (Supplementary Table III). However,

of the 98 common DEGs between rpotmp ANAC017 and atphb3 ANAC017

(Figure 2A), 41 were uniquely identified by RNAseq and not previously described

as differential by microarray analysis (Van Aken et al., 2007, Van Aken and

Whelan 2012). Disregarding transposable elements and chromosome 2 mtDNA-

duplicated transcripts, 27 newly-described DEGs responding to heritable

mitochondrial dysfunctions were identified (Table 1, Supplementary Table IV).

These included 6 mitochondrially-encoded transcripts, transcription factor

ANAC044, and a transcript encoding a predicted mitochondrial outer membrane

protein similar to optic atrophy 3 (OPA3), which is involved in mitochondrial

fragmentation in humans (Ryu et al., 2010).

13 DEGs uniquely identified by RNAseq were annotated as unknown nuclear-

encoded proteins (Table 1). Strikingly, 4 out of the 10 most strongly-induced

common DEGs in atphb3 and rpotmp plants belong to a group of proteins

containing the domain of unknown function 295 (DUF295). In total 8 transcripts

encoding DUF295 proteins were differentially expressed in atphb3 ANAC017,

including the most strongly-induced transcript found in this study genome-wide

10

(FC 561x), and another four transcripts with FC >85x (Figure 2B). Six out of 8

were also more than 2x induced in rpotmp ANAC017, but only four were retained

after multiple testing correction. Although no experimental protein-targeting

information has been published for any of the 8 genes, 7 are predicted as

mitochondrially-targeted by the consensus classifier SUBAcon that combines a

large number of location predictions for these proteins (Hooper et al., 2014).

Loss of ANAC017 in mitochondrial mutant backgrounds has genome-wide transcriptional impact

To assess the impact of ANAC017 loss on the hundreds of transcripts

differentially expressed in these mitochondrial mutant backgrounds, the DEGs

between single and double mutants were examined. 59 transcripts were

differentially expressed between rpotmp ANAC017 and rpotmp anac017 plants

(padj<0.05, FC<1.5x) (Supplementary Table V). 28 of 59 transcripts were initial

DEGs in rpotmp ANAC017 single mutants vs Col-0, and 23 of these 28 were no

longer significantly differential in rpotmp anac017 vs Col-0, indicating that

ANAC017 is required for their respective up- or down-regulation (Figure 2A). Four

more transcripts showed approx. 3x lower induction levels in ropotmp anac017

than in rpotmp ANAC017 mutants, indicating a partial requirement for ANAC017.

The remaining 31 of the 59 differential transcripts were not initial DEGs in rpotmp

ANAC017 single mutants (including ANAC017 itself), and can be considered as

knock-on consequences of lacking ANAC017-dependent retrograde signalling.

Among the most-strongly induced genes specifically in rpotmp anac017 were

genes involved in nutrient storage such as oleosin 2 (lipid storage, FC 21x),

cruciferina 1 (nutrient storage, FC 13.5x) and a late embryogenesis abundant

protein (At3g15670, FC 6.1x). Other genes specifically induced in rpotmp

anac017 included chloroplast aldehyde reductase (detoxification of lipid

peroxidation products, FC 13x) and 7-hydroxymethyl chlorophyll a reductase

(chlorophyll a biosynthesis, FC 1.74x). Two gluconeogenesis/glycolysis-related

enzymes were downregulated in rpotmp anac017: fructose bisphosphate

aldolase FBA8 (FC -1.5x) and pyruvate decarboxylase PDC1 (FC -2x).

11

The loss of ANAC017 in rpotmp mutants thus appeared to block or attenuate the

retrograde response of 27 transcripts and triggered a number of downstream

knock-on transcript changes. Nevertheless, 351 transcripts were still commonly

differentially expressed in both rpotmp ANAC017 and rpotmp anac017 mutants

vs Col-0, and were not differentially expressed between rpotmp ANAC017 and

rpotmp anac017. A gene ontology analysis of these 351 genes reveals significant

overrepresented and downregulated transcripts encoding peroxidases (7), and

transcripts involved in cell wall/casparian strip/suberin synthesis and transport

(ions, iron and nitrate).

A similar comparative analysis was performed on atphb3 ANAC017 and atphb3

anac017 mutants. 964 transcripts were differentially expressed between atphb3

ANAC017 and atphb3 anac017 plants (padj<0.05, FC<1.5x) (Supplementary

Table VI). 215 of 964 transcripts were initial DEGs in atphb3 ANAC017 single

mutants vs Col-0, and 129 of these 215 were no longer significantly differential in

rpotmp anac017 vs Col-0, indicating that ANAC017 is required for their respective

up- or down-regulation (Figure 2A). 71 additional transcripts of 215 showed

approx. 2.5x lower induction/repression levels in atphb3 anac017 than in atphb3

ANAC017 mutants, indicating a partial requirement for ANAC017. The remaining

676 of 964 transcripts that were not initial DEGs in atphb3 ANAC017 single

mutants vs Col-0 (including ANAC017 itself) can also be viewed as knock-on

consequences of lacking ANAC017-dependent retrograde signalling in atphb3

mutants. A GO analysis of these 676 atphb3 anac017 specific transcripts

revealed a downregulation of peroxidases (12), and transcripts involved in cell

wall/casparian strip/suberin synthesis and transport (ions, iron and nitrate).

Significant overrepresentation was found in defence-related genes among

transcripts uniquely upregulated in atphb3 ANAC017 mutants.

The loss of ANAC017 in atphb3 mutants thus appeared to block or attenuate the

retrograde response of 200 transcripts and triggered a great number of

downstream knock-on transcript changes (20x more than in rpotmp mutants),

indicating that loss of ANAC017 had more dramatic effects in atphb3 mutants.

Nevertheless, 140 transcripts were still commonly differentially expressed in both

12

atphb3 ANAC017 and atphb3 anac017 mutants vs Col-0, and were not differential

between atphb3 ANAC017 and atphb3 anac017. A further gene ontology analysis

of these 140 genes revealed significant overrepresentation of upregulated

transcripts related to mitochondria (including 8 transcripts mapped to the mtDNA

duplication on nuclear chromosome 2, and ABC transporter ABCI4 putatively

involved in complex IV assembly). Downregulated transcripts were

overrepresented in transcripts involved in transport (ions, carboxylic acids, amino

acids, iron and nitrate) and response to starvation.

At first glance, our results suggested that ANAC017 regulates only a subset of

constitutive expression changes caused by heritable mitochondrial dysfunction,

and significant transcript changes persist regardless of ANAC017’s absence in

both rpotmp and atphb3 backgrounds. These ANAC017-independent changes

are however very background-specific (rpotmp vs atphb3): only 21 transcripts

were common between the 351 (rpotmp) and 140 (atphb3) ANAC017-

independent expression changes. After removing chromosome 2 mtDNA

duplications and transposable elements from this list of 21, only 6 nuclear

transcripts (3 up- and 3 downregulated) and 5 mitochondrial transcripts (4

upregulated, 1 variable) that were differentially expressed across backgrounds in

an ANAC017-independent manner remain (Figure 2C). It was also clear that

within each genetic background ANAC017 was largely responsible for the most

dramatically induced transcripts (Figure 2A). For example, of 48 transcripts that

were >10x induced in atphb3 ANAC017 vs Col-0, 37 transcripts were (co-

)regulated by ANAC017 (86.5%). Similarly, of 15 transcripts >10x induced in

rpotmp ANAC017 vs Col-0, 10 (at padj<0.06) were (co-)regulated by ANAC017.

When examining all 98 transcripts that were differentially expressed in both

rpotmp ANAC017 and atphb3 ANAC017 mutants (including 13 transposable

elements, pseudogenes and chromosome 2 mtDNA duplications), the role of

ANAC017 becomes clearer (Figure 2A, Supplementary Table II). Out of 98

transcripts, 63 are differentially expressed between atphb3 ANAC017 and atphb3

anac017, and 24 between rpotmp ANAC017 and rpotmp anac017 (22 common

with atphb3 anac017). As can be judged from Figure 2A, the fold changes are

13

generally higher in atphb3 ANAC017 mutants than in rpotmp ANAC017 mutants,

making any ANAC017-dependent effects more obvious and discernible by

statistical testing. In conclusion, of 98 transcripts commonly differentially

expressed due to heritable mitochondrial defects, 65 are affected by ANAC017,

while only 6 (nuclear, protein-coding) transcripts are ANAC017-independent.

Intra-organelle transcription is largely ANAC017-independent

The above analysis clearly showed that ANAC017 has a large impact on nuclear

genes that commonly respond transcriptionally to mitochondrial dysfunction. The

RNAseq analysis also allowed detailed analysis of mitochondrial and chloroplast

transcript abundance. In rpotmp ANAC017 mutants 32 mitochondrial transcripts

are significantly differential compared to Col-0, while only one chloroplast

transcript RPS18 was slightly differential (FC 1.55x). Overall, the transcriptional

changes in rpotmp ANAC017 found by RNAseq were very similar to previously

reported mitochondrial transcript levels in rpotmp mutants as determined by qRT-

PCR (Kuhn et al., 2009). In atphb3 ANAC017 mutants 8 mitochondrial transcripts

were significantly differential compared to Col-0. Chloroplast RPS18 was

significantly differential based on p-value (padj=0.0158), but the fold change was

very low (FC 1.19x). When comparing the differential organellar transcripts in

each of the single mutants to their respective anac017 double mutants, virtually

no statistically significant differences were observed (Figure 3). Only Ribosomal

RNA 18 was differential in rpotmp anac017 vs rpotmp ANAC017, and was in fact

even more downregulated (FC -3.65x vs -2.07x). Furthermore, no mitochondrial

transcripts that were normally expressed in the single mutants became

differential in the respective double mutants. Interestingly, 26 chloroplast

transcripts that were normally expressed in the atphb3 ANAC017 single mutant

became differential in atphb3 anac017 vs atphb3 ANAC017 (FC >1.3x, 2

transcripts with FC >1.5x), further underlining the primary effects of the loss of

ANAC017, particularly in the atphb3 mutants. These findings suggest that the

role of ANAC017 is to adapt nuclear gene expression in response to

14

mitochondrial dysfunction, but not (even indirectly) to affect mitochondrial gene

expression.

Retrograde signalling targets specific changes in energy metabolism

RNAseq analysis revealed that several of the core ANAC017-dependent

retrograde marker genes were involved in primary metabolism (Figure 2A). These

included glycolytic enzymes (fructose-bisphosphate aldolase FBA6 and 2,3-

biphosphoglycerate-independent phosphoglycerate mutase IPGAM2) and amino

acid metabolism enzymes (mitochondrial alanine aminotransferase 2 and proline

dehydrogenase, aspartate aminotransferase 2). To determine if these

transcriptional changes address actual changes in metabolism upon

mitochondrial dysfunction, metabolite profiles of the different mutants were

measured by gas-chromatography mass spectrometry (GC-MS) (Figure 3A,

Supplemental Table VII). Metabolites were extracted from the same tissue

samples that were used for the RNAseq analysis (growth stage 1.04), allowing

reliable comparisons to be made between transcript and metabolite levels.

Overall, the mitochondrial dysfunctions studied here resulted in altered relative

abundance compared to Col-0 of a broad range of metabolites with the metabolic

profiles of the four different mutant lines proving to be relatively similar. In

particular, 9 amino acids accumulated to significantly higher levels in all four

mutants compared to Col-0. L-threonine was significantly reduced only in rpotmp

ANAC017 and rpotmp anac017 mutants, while L-aspartic acid accumulated

significantly in both atphb3 ANAC017 and atphb3 anac017 mutants. L-glutamine

and L-lysine were only significantly accumulated in atphb3 anac017 mutants. L-

glutamine was not detectable in Col-0, and overall its measurement was highly

variable and close to the detection limit in these samples.

A number of carbohydrates (D-Mannose, D-galactose and trehalose) were

significantly more abundant in all four mutants, but D-fructose was reduced only

in atphb3 anac017 double mutants. Glycolytic intermediates (glucose-6-

phosphate and fructose-6-phosphate) appeared to be more abundant in all four

15

mutants, though in rpotmp anac017 they did not pass statistical testing.

Detectable organic acids were relatively similar between all five genotypes, with

only malate significantly induced in the single mutants only. A mild but significant

increase in fatty acids (palmitic and stearic acid) was observed uniquely in atphb3

anac017 mutants. In a few cases significant differences were observed between

the respective single and double mutants (Figure 3A), but for none of the

measured compounds a consistent significant difference could be observed

between both pairs of single/double mutants. Only atphb3 anac017 plants display

unique changes in steady state metabolite levels compared to atphb3 ANAC017,

with significant accumulations of fatty acids, citric acid, lactic acid, lysine and

glutamine, but a >50% depletion of D-fructose.

In conclusion, ANAC017-dependent retrograde signalling specifically targets

transcription of several enzymes that directly or indirectly use substrates affected

by mitochondrial defects (Figure 3B). However, the steady state levels of primary

metabolites acted on by these enzymes were mostly unaffected by the presence

or absence of ANAC017, except in anac017 atphb3 plants.

Loss of ANAC017 abolishes induction of alternative oxidase protein levels and activity

As ANAC017 was responsible for the induction of the transcript level of AOX1a,

it was then investigated if the lack of transcript induction in the double mutants

translated into steady-state differences in AOX protein abundance and activity.

Intact mitochondria were isolated from developmentally matched Col-0 and the

respective double and single mutants for rpotmp anac017 and atphb3 anac017.

Protein levels of the alternative oxidase were estimated by Western Blot analysis

and compared to mitochondrial ATP synthase beta subunit (ATP-B) as a

reference (Figure 4A). AOX protein was more abundant in both rpotmp ANAC017

and atphb3 ANAC017 single mutants compared to Col-0, as was reported

previously for rpotmp ANAC017 (Kuhn et al., 2015). However, AOX protein levels

in rpotmp anac017 and atphb3 anac017 mutants were closer to the level in Col-

0.

16

The lack of AOX protein induction also suggested that respiratory pathway activity

could be altered in the different mutants. To estimate the relative contribution of

the cytochrome and alternative pathways, oxygen consumption rates were

measured in the isolated mitochondria (Figure 4B). The State III (succinate +

ADP) oxygen consumption using succinate as a substrate was not significantly

altered between the genotypes. Also the State III to State II (succinate only) ratio

was the same in all five genotypes, so no significant differences were observed

in coupling of respiration to ATP synthesis (p>0.05; n=3-4). KCN, DTT and

pyruvate was added during state III respiration to inhibit cytochrome c oxidase-

dependent oxygen consumption and reveal maximum AOX-dependent oxygen

consumption. It was clear from these data that rpotmp ANAC017 and atphb3

ANAC017 mitochondria had a higher capacity for alternative respiration than Col-

0, for instance maximal AOX-dependent rates were as high as state III total

respiration in atphb3 ANAC017 plants. However, the ratio of AOX capacity to

State III respiration had reverted completely to wild type levels when ANAC017

function was lost in these mutants (Figure 4B).

Mitochondrial mutants defective in ANAC017 displayed exacerbated growth defects

The single rpotmp and atphb3 mutant plants are viable and fertile, but display

significant growth defects (Van Aken et al., 2007, Kuhn et al., 2009, Van Aken et

al., 2010). The double mutants lacking ANAC017-dependent signalling are thus

interesting tools to study the physiological importance of retrograde signalling in

mutants with constitutive mitochondrial defects. After 15 days of growth, rpotmp

anac017 plants were of similar size as the rpotmp ANAC017 single mutants

(Figure 5A and 5B). However, atphb3 anac017 plants were significantly smaller

than the atphb3 ANAC017 single mutants (approximately 40-50% smaller).

Moreover, atphb3 anac017 mutants displayed abnormal and incomplete leaf

shapes (Figure 5A). As the plants developed further, obvious differences in plant

biomass were observed between both pairs of single and double mutants, with

the double mutants being consistently smaller than the comparable single

17

mutants (Figure 5B). After 4 weeks of growth rpotmp anac017 were significantly

smaller than rpotmp ANAC017 plants, and furthermore showed punctate lesions

that were absent in the single mutants (Figure 5C). Single anac017 mutant plants

appeared to have comparable rosette sizes to Col-0 throughout development

(Supplementary Figure 1). Complemented rpotmp anac017 35S::ANAC017

plants showed significantly larger rosette area than rpotmp anac017 plants, and

lesions were no longer observed on the leaves (Supplementary Figure 1). Atphb3

anac017 plants remained at approximately 50% of the rosette size of atphb

ANAC017 until 31 days, but increased to 61% after 36 days. In contrast to the

malformations observed in the oldest leaves, younger leaves in the rosettes had

normal shapes (Figure 5C).

Analysis of relative growth rates (RGR) of the different mutants at a range of time

intervals showed that from 15 days onwards the single mutant showed no or only

mildly lower growth rates compared to Col-0 (Supplementary Figure 2). This

indicates that the growth defects are mainly caused by delays in germination in

the single mutants as described previously (Van Aken et al., 2007, Kuhn et al.,

2009), while relative growth rates later in the life cycle are largely unaffected. In

contrast, both double mutants showed significantly reduced RGR vs Col-0 and

the respective single mutants at different times in development. For rpotmp

anac017 the reduced RGR occurred between 19 and 31 days of age, while for

atphb3 anac017 this was found up until day 19. Interestingly, RGR of both double

mutants showed a slower decline over time than of Col-0 and the single mutants,

resulting in significantly higher RGR of the double mutants vs single mutants and

Col-0 at the latest time points. This is likely because single mutants and Col-0

had reached maturity and reproductive age, resulting in reduced investment in

rosette growth. Remarkably, maximal RGR of atphb3 anac017 stayed nearly

constant over time and was more than twofold higher than in Col-0 over the 31-

36 day interval, indicating a prolonged growth spurt in this mutant. This also

underscores the apparent increase in % biomass in atphb3 anac017 observed at

36 days (Figure 5B).

18

Besides reductions in rosette size, double mutants also showed delayed

flowering compared to the respective single mutants (Figure 5D). Both the

percentage of plants that had started flowering (inflorescence height >3 mm) at

36 days after sowing and the actual height of the inflorescences was significantly

lower for atphb3 anac017 vs atphb3 ANAC017 and for rpotmp anac017 vs rpotmp

ANAC017. In agreement, the dry weight of fully senesced inflorescences in

rpotmp anac017 plants was significantly lower than in rpotmp ANAC017 (Figure

5E). Inflorescence dry weight in atphb3 ANAC017 was already significantly

reduced compared to Col-0 by 35-40%, but was not further reduced in atphb3

anac017 plants. Finally, total seed production per plant was significantly lowered

by >40% in both rpotmp ANAC017 and atphb3 ANAC017 plant compared to Col-

0. In both rpotmp anac017 and atphb3 anac017 total seed production was

significantly further reduced compared to the single mutants, indicating a further

reproductive penalty (Figure 5E). Furthermore, complemented rpotmp anac017

35S::ANAC017 plants showed restored seed production, similar to rpotmp

ANAC017 plants (Supplementary Figure 1). Seed production in anac017 single

mutants was not significantly reduced compared to Col-0.

Root lengths after 7 days of growth were measured and, in line with previous

reports (Van Aken et al., 2007, Kuhn et al., 2009), atphb3 ANAC017 and rpotmp

ANAC017 mutants showed 70% reduction in root length (Figure 5F). Similarly to

what was observed for rosette sizes during early development, no significant

reduction in root length was observed between rpotmp ANAC017 and rpotmp

anac017 plants. However, the roots of atphb3 anac017 plants were very short

after 7 days (only 2.8 mm), being 60% shorter than the already shorter roots of

atphb3 ANAC017 plants. To assess whether the loss of ANAC017 impacted more

severely on root growth when mitochondrial function was further limited, root

length was observed after 7 days of growth in the presence of antimycin A (Figure

5F). Rpotmp anac017 plants were significantly more susceptible than rpotmp

ANAC017 plants to the addition of antimycin A. Also atphb3 anac017 were

extremely susceptible to antimycin A, with average root lengths around 1 mm

after 7 days. Overall, these results demonstrate that ANAC017-dependent

19

retrograde signalling is important to minimise growth- and productivity penalties

caused by mitochondrial malfunctions.

Discussion

The aim of this study was to determine if a transcription factor responsible for

retrograde signalling in responses to short term treatments with chemical

inhibitors was also involved in retrograde signalling during constitutive

developmental defects in mitochondrial biogenesis. Based on the work presented

here, it is evident that ANAC017 does play this dual role and is responsible for

transmitting signals from mitochondria that have ongoing dysfunctions caused by

a genetic fault. As the mitochondrial malfunctions are present from seed

germination onward, they can be considered as biogenic defects. It thus appears

that ANAC017 is a crucial signalling component of both operational (e.g. inhibitor-

induced short term responses) as well as biogenic retrograde control

(Supplementary Figure 3A)(Pogson et al., 2008, Ng et al., 2013b). A comparison

of the previously published microarray data addressing the role of ANAC017

during antimycin A treatment (Ng et al., 2013b) with the currently presented

RNAseq data on mitochondrial mutants, reveals that the chemical treatment has

much broader effects, with 7399 probe sets significantly responding to antimycin

A after 3h. This is a >12x larger number than the 579 (rpotmp ANAC017) and 559

(atphb3 ANAC017) transcripts responding to genetic mitochondrial dysfunction.

In total, 40 genes (39 probe sets) were common between 3h antimycin A

treatment, rpotmp ANAC017 and atphb3 ANAC017 (Supplementary Figure 3B).

Strikingly, 32 of the 40 were ANAC017-regulated (e.g. AOX1a), and are thus core

ANAC017-regulated genes that respond to both acute and constitutive

mitochondrial dysfunction (Supplementary Table VIII). The promoters of 9 of 92

nuclear-encoded ANAC017-regulated genes (Figure 2A) were experimentally

shown to bind ANAC017 and its close relative ANAC013 (De Clercq et al., 2013).

45 of 92 promoters contain the ANAC017 consensus binding site within 1 kb of

the transcriptional start site, with the median distance just 178 bp (Supplementary

Figure 3C, Supplementary Table IX). However, ANAC017 has little or no control

20

over transcription within the mitochondrion, in line with previous reports that

nuclear and mitochondrial transcription are largely regulated independently

(Giege et al., 2005).

When mitochondrial function was perturbed, the transcriptional changes that

occur could be classified as ANAC017-dependent and -independent changes.

ANAC017 is to a large extent responsible for the retrograde responses that are

commonly triggered by mitochondrial dysfunction, regardless of how the

dysfunction is caused. This group of transcripts also had the largest fold-changes,

so ANAC017 was responsible for the most dramatic changes. Based on the

above results, ANAC017 regulates the expression of approximately 200 genes.

The amplitude of transcript changes (fold change) appeared to depend on the

severity of the mitochondrial genetic defect (apparently greater in atphb3 than

rpotmp), so weaker defects may only have triggered statistically discernible

changes in a smaller set of genes. Gene expression of retrograde marker genes

(e.g. AOX1a and ANAC013) in single anac017 mutants is not significantly

different from Col-0 under normal conditions, indicating that the role of ANAC017

only becomes apparent when a mitochondrial stress is occurring (Ng et al.,

2013b, Van Aken et al., 2016).

Newly-identified ANAC017-dependent transcripts include a group of 8 DUF295

genes that putatively encode mitochondrial proteins, with currently unknown

functions. A number of other DUF295 proteins were previously characterised,

which are not predicted to be mitochondrially targeted. Rice MAIF1 also contains

an F-box domain and is proposed to play a negative role in abiotic-stress

resistance by promoting root growth (Yan et al., 2011). An Arabidopsis DUF295

F-box protein called Upward Curly Leaf 1 (UCL1) is part of an E3-ligase protein

complex with Curly Leaf (CLF) polycomb protein, involved in histone methylation

(Jeong et al., 2011). A T-DNA knockout of UCL1 did not display any phenotypical

defects. The DUF295-containing C-terminal of UCL1 could bind the CLF protein,

suggesting that the DUF295 domain is an uncharacterised protein-protein

interaction domain. The N-terminal F-box domain (which is also a protein-protein

interaction domain) may co-determine the binding specificity. It will thus be of

21

interest to identify the potential binding-partners of the putatively mitochondrial

DUF295 proteins, and whether they contribute to recovery from mitochondrial

defects. An uncharacterised NAC transcription factor ANAC044 was also

uncovered by the RNAseq analysis as part of the ANAC017 regulon. Unlike

ANAC017 or ANAC013, ANAC044 does not contain a C-terminal transmembrane

domain, suggesting it might regulate a specific subset of ANAC017-dependent

genes that are not bound by ANAC017 itself.

Besides the transcripts commonly responding to mitochondrial dysfunction, each

specific mitochondrial defect has its own transcriptional signature, which is largely

unaffected by ANAC017. So ANAC017 regulates core mitochondrial retrograde

responses, but each mitochondrial defect causes unique downstream changes.

It will thus be a significant future challenge to determine the signalling networks

that underlie the perturbation-specific expression changes.

Loss of ANAC017 not only affects expression of genes it is directly binding and

regulating, but has further knock-on effects that could not be observed in the

single mutants with functional ANAC017 (i.e. genes that were differentially

expressed between respective single and double mutant, but not originally

differential in single mutant vs Col-0). For instance, rpotmp ANAC017 and rpotmp

anac017 transcript profiles are in general very similar, with only 59 significant

transcript changes, of which 27 are ANAC017-regulated. So only a handful of

genes are likely knock-on effects of losing ANAC017 in the rpotmp mutant

background. This is in line with the relatively mild phenotypical differences in both

rpotmp mutants in the first three weeks of growth. Strikingly, 676 knock-on

transcript changes were found in atphb3 anac017 mutants. This is reflected in the

strong phenotypic defects observed in atphb3 anac017 mutants, as well as in

several changes in important energy metabolism molecules (fatty acids-, lactate-

and citrate accumulation, fructose depletion). Somewhat surprisingly, the knock-

on effects caused by loss of ANAC017 are completely different in rpotmp (31

transcripts) and atphb3 (676 transcripts), with the only overlapping gene that is

consistently up- or down regulated being the causative agent itself, ANAC017

(two-fold downregulated in anac017 genotypes, but normally expressed in the

22

single mitochondrial mutants). Both genotypes thus seem to diverge in their

response to the absence of ANAC017-dependent retrograde signalling.

Despite completely different knock-on consequences caused by loss of

ANAC017 in rpotmp and atphb3 mutants, a remarkable similarity was found

between the 351 core rpotmp ANAC017-independent genes and the 676 atphb3

anac017 knock-on transcripts. 67 transcripts were in common, nearly 10x more

than the 7 transcripts expected by random selection (p<0.00001). GO analysis of

this list indicates an overrepresentation of downregulated transcripts encoding

peroxidases, and transcripts involved in cell wall/casparian strip/suberin

synthesis and transport (ions, iron and nitrate). Overall the targets are consistent

with the growth problems faced by the mitochondrial mutants: reduced

mitochondrial activity may result in less production of H2O2 (thus requiring less

peroxidase activity)(Huang et al., 2016), less cell growth and expansion (requiring

less cell wall biosynthesis; e.g. hemicelluloses), and an apparent overflow of

unutilised amino-acids and potentially lipids (Figure 3, reduced starvation

response and nitrogen transport).

Part of the core atphb3 ANAC017-independent gene response is downregulation

of auxin and tryptophan metabolism (again relating to growth and expansion), as

well as downregulation of nitrogen/metal transport and starvation response as

observed among rpotmp ANAC017-independent genes. However, the individual

genes involved in atphb3 are completely different to the ones targeting these

pathways in rpotmp mutants. This indicates that both atphb3 and rpotmp mutants

have alterations associated with similar cellular processes, but they are regulated

via different ANAC017-independent transcriptional pathways. When ANAC017-

dependent retrograde signalling is no longer active in atphb3 anac017 mutants,

an additive pathway is switched on that appears to be activated a priori in rpotmp

ANAC017 mutants.

In this study, the transcriptional and metabolic steady state levels of mutants

could be compared. Despite relatively divergent transcriptional responses (apart

from ANAC017-dependent signalling), the metabolic profiles of all four mutant

genotypes were quite similar. Constitutive loss of mitochondrial function and

23

slowed growth thus seems to result in a relatively stable altered steady state of

primary metabolism. It must be noted that metabolite pools might change at

subcellular levels that cannot be resolved by the used methods, and thus

metabolites involved in retrograde signalling might not have been identified here.

Furthermore, metabolic fluxes might be much more important in retrograde

signalling than steady state levels. The metabolic profiles found in this study

phenocopied the metabolite levels previously reported for complex I subunit

mutant ndufs4 (Meyer et al., 2009) and malate dehydrogenase mmdh1 mmdh2

(Tomaz et al., 2010), with accumulations of many amino acids such as alanine,

asparagine, glutamate, serine and proline. Also malate is commonly enriched in

all these mitochondrial mutants. A time course that studied acute mitochondrial

inhibition by rotenone in Arabidopsis cell culture showed nearly opposite patterns,

with rapid depletions of many of these amino acids, while organic acids and

sugars are nearly unchanged (Garmier et al., 2008). Therefore, it seems that

despite similarities in ANAC017-dependent transcriptional changes, the

metabolic results of short term operational and long term biogenic mitochondrial

dysfunctions are very different. There is no support from our data to indicate that

TCA cycle intermediates are the downstream triggers of ANAC017-dependent

mitochondrial retrograde signalling, as their levels are generally at Col-0 levels in

the mitochondrial mutants (Figure 3), while the retrograde signalling is clearly

active. Also, no transcripts are commonly differential between a microarray

experiment performed on citrate feeding (Finkemeier et al., 2013) and the

RNAseq data presented here.

Most of the metabolic changes in this study reflect accumulations of intermediates

(amino acids, glycolytic intermediates, carbohydrates) rather than depletions (no

metabolite is consistently depleted in the mitochondrial backgrounds). This could

suggest that the plants do not suffer from limited input of energy equivalents from

photosynthesis, but are slowed in processing these metabolites into growth-

supporting compounds. In terms of metabolic changes, the role of ANAC017 is

most obvious in atphb3 anac017 mutants, in agreement with the observation that

this mutant displays the most severe growth defects. Several important energy

metabolites are uniquely altered, with accumulation of fatty acids, citrate, lactic

24

acid, lysine and glutamine, but more than 50% depletion of fructose. Lactate

accumulation of >2.5x suggests a slowing of aerobic metabolism and a shift to

fermentation which is a classical sign of limited mitochondrial function.

In conclusion, this study shows that ANAC017 can play an important role in

helping plants with mitochondrial dysfunctions to establish readjusted metabolic

equilibriums. Whether this is causing a significant improvement to plant growth

apparently depends on the severity of the initial defect. Based on these results,

the atphb3 mutation is arguably more detrimental than the rpotmp mutation. The

fact that growth rates observed in atphb3 and rpotmp single mutants are largely

indistinguishable is likely due to the retrograde response orchestrated by

ANAC017. In other words, ANAC017 allows atphb3 mutants to grow at a rate

similar to that of the less-affected rpotmp mutant. In rpotmp anac017, problems

appear to only reach a noticeable threshold after 4 weeks of growth, with a

reduction in rosette size and the onset of chlorotic lesions. Similar chlorotic

lesions have been reported in rpotmp aox1a double mutants (Kuhn et al., 2015),

which appear to have stronger growth reductions than rpotmp anac017. The

difference is likely that rpotmp anac017 mutants still have basal AOX activity, they

just cannot increase it in response to retrograde signals. This is consistent with

AOX1a not only being a useful marker transcript to study retrograde signalling,

but a significant player that is beneficial to plants with mitochondrial dysfunctions

(Robson and Vanlerberghe 2002, Vanlerberghe et al., 2002).

Experimental Procedures

Plant growth conditions, transformation and materials

Arabidopsis thaliana Col-0 was used in all experiments. Seeds were sown on soil

mix or MS media with 2% sucrose and stratified for 2-3 days at 4°C, then grown

under long-day conditions (16 h light/8 h dark) at 22 °C and 100 µmol m-2 s-1.

Previously published transgenic lines were described in Ng et al., 2013 (anac017:

rao2.1), Kuhn et al., 2009 (rpotmp: SALK_132842) and Van Aken et al., 2007

(atphb3: SALK_020707). ANAC017 complemented plants were generated by

25

cloning the full length ANAC017 CDS into pB7WG2 (Karimi et al., 2002) using

standard Gateway (Life Technologies) cloning strategies, and transformed by

floral dipping as described previously (Clough and Bent 1998, Zhang et al., 2012).

Quantitative measurements of root lengths and rosette sizes were performed in

ImageJ. Statistical tests throughout the manuscript were performed using

ANOVA with subsequent post-hoc student’s t-test, except where indicated

differently. Relative growth rates for individual plants over time were calculated

by RGR = (lnRS2 – lnRS1)/(t2-t1), where RS is rosette surface area and t is time

in days (Hoffmann and Poorter 2002).

Quantitative RT-PCR

Seeds were sown on petri dishes containing MS media + 2% sucrose, stratified

for two-three days in the cold room and then incubated in long day growth

conditions for two (Col-0) or three (mitochondrial mutants) weeks. Pools of plants

(whole seedlings) were then collected on the same day in the morning and

immediately placed in liquid nitrogen for storage and further processing. RNA

isolation, cDNA generation and quantitative RT-PCR (qRT-PCR) was performed

as described in (Van Aken et al., 2013) using Spectrum RNA Plant extraction kits

(Sigma-Aldrich, Sydney), iScript cDNA synthesis kit (BIO-RAD) and a Roche

LC480 lightcycler using SYBRgreen detection assays. All primers for qRT-PCR

are shown in Supplementary Table X.

RNAseq analysis

Libraries for RNAseq analysis were prepared from 500 ng Ambion Turbo DNAse-

treated total RNA (as described above for RT-PCR analysis) using the Illumina

Ribo-zero Plant kit (RS-122-2401) following standard procedures. Libraries were

clustered at 19 pM on an Illumina cBot using Truseq SR Cluster Kit v3 cBOT HS

(GD-401-3001). Sequencing was then performed on an Illumina HiSeq 1500

using SBS kit v3 for 61 cycles (FC-401-3002). In total, between 19 to 30 M

clusters passed filter per sample. Reads were aligned to TAIR10 with Tophat2

and 17 to 27 million uniquely aligned reads were obtained for each sample (Kim

et al., 2013). Aligned reads were assigned to genes with htseq-count (Anders et

26

al., 2014). Differentially expressed genes were called with DEseq2 with no

independent filtering (Love et al., 2014) Transcripts were considered to be

significantly differentially expressed between genotypes when padj<0.05 (after

multiple testing correction) and fold change >1.5x. Data files are available at

ArrayExpress under accession number E-MTAB-4655.

Mitochondrial purification, respiration and Western Blot analysis

Mitochondria were isolated from developmentally similar seedlings grown on MS

agar plates (equivalent to 2 week old for Col-0) using differential centrifugation

as described previously (Lister et al., 2007). Mitochondrial proteins were

separated on BIO-RAD anyKD pre-cast gels, then blotted and detected as

previously described (Zhang et al., 2012). Oxygen electrodes (Oxytherm,

Hansatech, UK) were used for isolated mitochondrial respiratory assays as

previously described (Jacoby et al., 2015).

Metabolite Analysis

Approx. 20-30 mg of frozen tissue from pooled two-three week old plants grown

in vitro were analysed using GC-MS (as described for RT-PCR analysis). Full

description of metabolite analysis is available in Supplementary Methods section.

Data were imported and statistically analysed in Metabolome Express (Carroll et

al., 2010).

27

Acknowledgements

We thank Dorothee Hahne of Metabolomics Australia for performing the GC-MS

metabolite analysis and acknowledge the use of Metabolomics Australia facilities

funded by National Collaborative Research Infrastructure Strategy (NCRIS) and

the Government of Western Australia. The authors thank Muhamad Che Othman

for help in preparing metabolism diagram and Rosemarie Farthing for technical

assistance. We thank Dr Kristina Kühn (Humboldt University Berlin), Prof James

Whelan (Latrobe University), Dr Estelle Giraud and Prof Frank Van Breusegem

(Ghent University) for providing single mutants used in this study and useful

discussions. This work was supported by the facilities of the Australian Research

Council Centre of Excellence Program (CE140100008). O.V.A by Australian

Research Council APD fellowship and grants (DP110102868, DP160103573),

A.H.M by an ARC Future Fellowship and ARC grants (FT110100242;

DP130102918, DP160103573), S.H. was supported by ARC Future Fellowship

and ARC grant (FT130101338), R.L. was supported by an ARC Future

Fellowship (FT120100862) and Sylvia and Charles Viertel Senior Medical

Research Fellowship. The authors declare no conflict of interest.

28

Short Supporting Information Legends

Supplementary Figure 1. Complementation of rpotmp anac017.

Supplementary Figure 2. Relative growth rate analysis.

Supplementary Figure 3. The dual role of ANAC017 in constitutive and acute

mitochondrial retrograde signalling.

Supplementary Table I. RNAseq overview table.

Supplementary Table II. Core mitochondrial dysfunction responsive genes.

Supplementary Table III. RNAseq vs ATH1 microarray comparison atphb3.

Supplementary Table IV. RNAseq unique Differentially Expressed Genes.

Supplementary Table V. RNAseq rpotmp ANAC017 vs rpotmp anac017.

Supplementary Table VI. RNAseq atphb3 ANAC017 vs atphb3 anac017.

Supplementary Table VII. Metabolite analysis.

Supplementary Table VIII. Antimycin A microarray vs mutant RNAseq

Supplementary Table IX. ANAC017 binding sites in core mitochondrial

dysfunction regulon.

Supplementary Table X. qRT-PCR primers.

Supplementary Methods. Additional information on metabolite extraction and

analysis.

29

References

Anders, S., Pyl, P.T. and Huber, W. (2014) HTSeq—a Python framework to work with high-throughput sequencing data. Bioinformatics, 31, 166-169.

Carroll, A.J., Badger, M.R. and Harvey Millar, A. (2010) The MetabolomeExpress Project: enabling web-based processing, analysis and transparent dissemination of GC/MS metabolomics datasets. BMC bioinformatics, 11, 376.

Chan, K.X., Phua, S.Y., Crisp, P., McQuinn, R. and Pogson, B.J. (2015) Learning the Languages of the Chloroplast: Retrograde Signaling and Beyond. Annual review of plant biology.

Clough, S.J. and Bent, A.F. (1998) Floral dip: a simplified method for Agrobacterium-mediated transformation of Arabidopsis thaliana. The Plant journal : for cell and molecular biology, 16, 735-743.

De Clercq, I., Vermeirssen, V., Van Aken, O., Vandepoele, K., Murcha, M.W., Law, S.R., Inze, A., Ng, S., Ivanova, A., Rombaut, D., van de Cotte, B., Jaspers, P., Van de Peer, Y., Kangasjarvi, J., Whelan, J. and Van Breusegem, F. (2013) The membrane-bound NAC transcription factor ANAC013 functions in mitochondrial retrograde regulation of the oxidative stress response in Arabidopsis. The Plant cell, 25, 3472-3490.

Dojcinovic, D., Krosting, J., Harris, A.J., Wagner, D.J. and Rhoads, D.M. (2005) Identification of a region of the Arabidopsis AtAOX1a promoter necessary for mitochondrial retrograde regulation of expression. Plant molecular biology, 58, 159-175.

Finkemeier, I., Konig, A.C., Heard, W., Nunes-Nesi, A., Pham, P.A., Leister, D., Fernie, A.R. and Sweetlove, L.J. (2013) Transcriptomic analysis of the role of carboxylic acids in metabolite signaling in Arabidopsis leaves. Plant physiology, 162, 239-253.

Garmier, M., Carroll, A.J., Delannoy, E., Vallet, C., Day, D.A., Small, I.D. and Millar, A.H. (2008) Complex I dysfunction redirects cellular and mitochondrial metabolism in Arabidopsis. Plant physiology, 148, 1324-1341.

Gehl, B., Lee, C.P., Bota, P., Blatt, M.R. and Sweetlove, L.J. (2014) An Arabidopsis stomatin-like protein affects mitochondrial respiratory supercomplex organization. Plant physiology, 164, 1389-1400.

Giege, P., Sweetlove, L.J., Cognat, V. and Leaver, C.J. (2005) Coordination of nuclear and mitochondrial genome expression during mitochondrial biogenesis in Arabidopsis. The Plant cell, 17, 1497-1512.

Giraud, E., Van Aken, O., Ho, L.H. and Whelan, J. (2009) The transcription factor ABI4 is a regulator of mitochondrial retrograde expression of ALTERNATIVE OXIDASE1a. Plant Physiol, 150, 1286-1296.

Giraud, E., Van Aken, O., Uggalla, V. and Whelan, J. (2012) REDOX regulation of mitochondrial function in plants. Plant, cell & environment, 35, 271-280.

Hoffmann, W.A. and Poorter, H. (2002) Avoiding bias in calculations of relative growth rate. Annals of botany, 90, 37-42.

Hooper, C.M., Tanz, S.K., Castleden, I.R., Vacher, M.A., Small, I.D. and Millar, A.H. (2014) SUBAcon: a consensus algorithm for unifying the subcellular

30

localization data of the Arabidopsis proteome. Bioinformatics, 30, 3356-3364.

Huang, S., Van Aken, O., Schwarzländer, M., Belt, K. and Millar, A. (2016) The roles of mitochondrial reactive oxygen species in cellular signalling and stress response in plants. Plant Physiology, in press.

Ivanova, A., Law, S.R., Narsai, R., Duncan, O., Lee, J.H., Zhang, B., Van Aken, O., Radomiljac, J.D., van der Merwe, M., Yi, K. and Whelan, J. (2014) A Functional Antagonistic Relationship between Auxin and Mitochondrial Retrograde Signaling Regulates Alternative Oxidase1a Expression in Arabidopsis. Plant physiology, 165, 1233-1254.

Jacoby, R.P., Millar, A.H. and Taylor, N.L. (2015) Assessment of respiration in isolated plant mitochondria using Clark-type electrodes. Methods Mol Biol, 1305, 165-185.

Jeong, C.W., Roh, H., Dang, T.V., Choi, Y.D., Fischer, R.L., Lee, J.S. and Choi, Y. (2011) An E3 ligase complex regulates SET-domain polycomb group protein activity in Arabidopsis thaliana. Proceedings of the National Academy of Sciences of the United States of America, 108, 8036-8041.

Karimi, M., Inze, D. and Depicker, A. (2002) GATEWAY vectors for Agrobacterium-mediated plant transformation. Trends in plant science, 7, 193-195.

Kim, D., Pertea, G., Trapnell, C., Pimentel, H., Kelley, R. and Salzberg, S.L. (2013) TopHat2: accurate alignment of transcriptomes in the presence of insertions, deletions and gene fusions. Genome Biol, 14, R36.

Kuhn, K., Richter, U., Meyer, E.H., Delannoy, E., de Longevialle, A.F., O'Toole, N., Borner, T., Millar, A.H., Small, I.D. and Whelan, J. (2009) Phage-type RNA polymerase RPOTmp performs gene-specific transcription in mitochondria of Arabidopsis thaliana. The Plant cell, 21, 2762-2779.

Kuhn, K., Yin, G., Duncan, O., Law, S.R., Kubiszewski-Jakubiak, S., Kaur, P., Meyer, E., Wang, Y., Small, C.C., Giraud, E., Narsai, R. and Whelan, J. (2015) Decreasing electron flux through the cytochrome and/or alternative respiratory pathways triggers common and distinct cellular responses dependent on growth conditions. Plant physiology, 167, 228-250.

Leister, D. (2012) Retrograde signaling in plants: from simple to complex scenarios. Frontiers in plant science, 3, 135.

Lister, R., Carrie, C., Duncan, O., Ho, L.H., Howell, K.A., Murcha, M.W. and Whelan, J. (2007) Functional definition of outer membrane proteins involved in preprotein import into mitochondria. The Plant cell, 19, 3739-3759.

Love, M.I., Huber, W. and Anders, S. (2014) Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol, 150, 550.

Meyer, E.H., Tomaz, T., Carroll, A.J., Estavillo, G., Delannoy, E., Tanz, S.K., Small, I.D., Pogson, B.J. and Millar, A.H. (2009) Remodeled respiration in ndufs4 with low phosphorylation efficiency suppresses Arabidopsis germination and growth and alters control of metabolism at night. Plant physiology, 151, 603-619.

31

Ng, S., Giraud, E., Duncan, O., Law, S.R., Wang, Y., Xu, L., Narsai, R., Carrie, C., Walker, H., Day, D.A., Blanco, N.E., Strand, A., Whelan, J. and Ivanova, A. (2013a) Cyclin-dependent kinase E1 (CDKE1) provides a cellular switch in plants between growth and stress responses. The Journal of biological chemistry, 288, 3449-3459.

Ng, S., Ivanova, A., Duncan, O., Law, S.R., Van Aken, O., De Clercq, I., Wang, Y., Carrie, C., Xu, L., Kmiec, B., Walker, H., Van Breusegem, F., Whelan, J. and Giraud, E. (2013b) A membrane-bound NAC transcription factor, ANAC017, mediates mitochondrial retrograde signaling in Arabidopsis. The Plant cell, 25, 3450-3471.

Ng, S., De Clercq, I., Van Aken, O., Law, S.R., Ivanova, A., Willems, P., Giraud, E., Van Breusegem, F. and Whelan, J. (2014) Anterograde and retrograde regulation of nuclear genes encoding mitochondrial proteins during growth, development, and stress. Molecular plant, 7, 1075-1093.

Piechota, J., Bereza, M., Sokolowska, A., Suszynski, K., Lech, K. and Janska, H. (2015) Unraveling the functions of type II-prohibitins in Arabidopsis mitochondria. Plant molecular biology, 88, 249-267.

Pogson, B.J., Woo, N.S., Forster, B. and Small, I.D. (2008) Plastid signalling to the nucleus and beyond. Trends Plant Sci, 13, 602-609.

Rhoads, D.M. and Subbaiah, C.C. (2007) Mitochondrial retrograde regulation in plants. Mitochondrion, 7.

Robson, C.A. and Vanlerberghe, G.C. (2002) Transgenic Plant Cells Lacking Mitochondrial Alternative Oxidase Have Increased Susceptibility to Mitochondria-Dependent and -Independent Pathways of Programmed Cell Death. Plant Physiology, 129, 1908-1920.

Ryu, S.W., Jeong, H.J., Choi, M., Karbowski, M. and Choi, C. (2010) Optic atrophy 3 as a protein of the mitochondrial outer membrane induces mitochondrial fragmentation. Cellular and molecular life sciences : CMLS, 67, 2839-2850.

Schwarzländer, M., Konig, A.C., Sweetlove, L.J. and Finkemeier, I. (2012) The impact of impaired mitochondrial function on retrograde signalling: a meta-analysis of transcriptomic responses. J Exp Bot, 63, 1735-1750.

Smith, C., Barthet, M., Melino, V., Smith, P., Day, D. and Soole, K. (2011) Alterations in the mitochondrial alternative NAD(P)H Dehydrogenase NDB4 lead to changes in mitochondrial electron transport chain composition, plant growth and response to oxidative stress. Plant & cell physiology, 52, 1222-1237.

Stupar, R.M., Lilly, J.W., Town, C.D., Cheng, Z., Kaul, S., Buell, C.R. and Jiang, J. (2001) Complex mtDNA constitutes an approximate 620-kb insertion on Arabidopsis thaliana chromosome 2: implication of potential sequencing errors caused by large-unit repeats. Proceedings of the National Academy of Sciences of the United States of America, 98, 5099-5103.

Sweetlove, L.J., Heazlewood, J.L., Herald, V., Holtzapffel, R., Day, D.A., Leaver, C.J. and Millar, A.H. (2002) The impact of oxidative stress on Arabidopsis mitochondria. The Plant journal : for cell and molecular biology, 32, 891-904.

32

Tognetti, V.B., Van Aken, O., Morreel, K., Vandenbroucke, K., van de Cotte, B., De Clercq, I., Chiwocha, S., Fenske, R., Prinsen, E., Boerjan, W., Genty, B., Stubbs, K.A., Inze, D. and Van Breusegem, F. (2010) Perturbation of indole-3-butyric acid homeostasis by the UDP-glucosyltransferase UGT74E2 modulates Arabidopsis architecture and water stress tolerance. The Plant cell, 22, 2660-2679.

Tomaz, T., Bagard, M., Pracharoenwattana, I., Linden, P., Lee, C.P., Carroll, A.J., Stroher, E., Smith, S.M., Gardestrom, P. and Millar, A.H. (2010) Mitochondrial malate dehydrogenase lowers leaf respiration and alters photorespiration and plant growth in Arabidopsis. Plant physiology, 154, 1143-1157.

Umbach, A.L., Zarkovic, J., Yu, J., Ruckle, M.E., McIntosh, L., Hock, J.J., Bingham, S., White, S.J., George, R.M., Subbaiah, C.C. and Rhoads, D.M. (2012) Comparison of intact Arabidopsis thaliana leaf transcript profiles during treatment with inhibitors of mitochondrial electron transport and TCA cycle. PloS one, 7, e44339.

Van Aken, O., Pecenkova, T., van de Cotte, B., De Rycke, R., Eeckhout, D., Fromm, H., De Jaeger, G., Witters, E., Beemster, G.T., Inze, D. and Van Breusegem, F. (2007) Mitochondrial type-I prohibitins of Arabidopsis thaliana are required for supporting proficient meristem development. The Plant journal : for cell and molecular biology, 52, 850-864.

Van Aken, O., Zhang, B., Carrie, C., Uggalla, V., Paynter, E., Giraud, E. and Whelan, J. (2009) Defining the mitochondrial stress response in Arabidopsis thaliana. Molecular plant, 2, 1310-1324.

Van Aken, O., Whelan, J. and Van Breusegem, F. (2010) Prohibitins: mitochondrial partners in development and stress response. Trends Plant Sci, 15, 275-282.

Van Aken, O. and Whelan, J. (2012) Comparison of transcriptional changes to chloroplast and mitochondrial perturbations reveals common and specific responses in Arabidopsis. Frontiers in plant science, 3, 281.

Van Aken, O., Zhang, B., Law, S., Narsai, R. and Whelan, J. (2013) AtWRKY40 and AtWRKY63 modulate the expression of stress-responsive nuclear genes encoding mitochondrial and chloroplast proteins. Plant Physiol, 162, 254-271.

Van Aken, O., De Clercq, I., Ivanova, A., Law, S.R., Van Breusegem, F., Millar, A.H. and Whelan, J. (2016) Mitochondrial and chloroplast stress responses are modulated in distinct touch and chemical inhibition phases in Arabidopsis. Plant Physiol, PMID: 27208304.

Vanderauwera, S., Vandenbroucke, K., Inze, A., van de Cotte, B., Muhlenbock, P., De Rycke, R., Naouar, N., Van Gaever, T., Van Montagu, M.C. and Van Breusegem, F. (2012) AtWRKY15 perturbation abolishes the mitochondrial stress response that steers osmotic stress tolerance in Arabidopsis. Proceedings of the National Academy of Sciences of the United States of America, 109, 20113-20118.

Vanlerberghe, G.C., Robson, C.A. and Yip, J.Y.H. (2002) Induction of Mitochondrial Alternative Oxidase in Response to a Cell Signal Pathway Down-Regulating the Cytochrome Pathway Prevents Programmed Cell Death. Plant Physiology, 129, 1829-1842.

33

Wang, Y., Law, S.R., Ivanova, A., Van Aken, O., Kubiszewski-Jakubiak, S., Uggalla, V., van der Merwe, M., Duncan, O., Narsai, R., Whelan, J. and Murcha, M.W. (2014) The mitochondrial protein import component, TRANSLOCASE OF THE INNER MEMBRANE17-1, plays a role in defining the timing of germination in Arabidopsis. Plant Physiol, 166, 1420-1435.

Yan, Y.S., Chen, X.Y., Yang, K., Sun, Z.X., Fu, Y.P., Zhang, Y.M. and Fang, R.X. (2011) Overexpression of an F-box protein gene reduces abiotic stress tolerance and promotes root growth in rice. Molecular plant, 4, 190-197.

Zarkovic, J., Anderson, S.L. and Rhoads, D.M. (2005) A reporter gene system used to study developmental expression of alternative oxidase and isolate mitochondrial retrograde regulation mutants in Arabidopsis. Plant molecular biology, 57, 871-888.

Zhang, B., Carrie, C., Ivanova, A., Narsai, R., Murcha, M.W., Duncan, O., Wang, Y., Law, S.R., Albrecht, V., Pogson, B., Giraud, E., Van Aken, O. and Whelan, J. (2012) LETM proteins play a role in the accumulation of mitochondrially encoded proteins in Arabidopsis thaliana and AtLETM2 displays parent of origin effects. The Journal of biological chemistry, 287, 41757-41773.

34

Table 1. Genes responding to heritable mitochondrial function uniquely identified by RNA-seq. Numerical values indicate fold change gene expression values normalised to Col-0.

Gene Description mitochondrial protein? Col-0

rpotmp ANAC017

rpotmp anac017

atphb3 ANAC017

atphb3 anac017

AT5G54550 Protein of unknown function (DUF295) Predicted 1 34.0 14.5 250.0 68.5

AT4G12735 unknown protein 1 17.3 -1.3 222.3 10.8

AT2G38823 unknown protein 1 26.0 1.5 214.0 30.5





AT3G58150 Optic atrophy 3 protein (OPA3) Predicted 1 16.2 5.0 66.7 17.0

AT3G54530 unknown protein 1 8.3 -2.0 45.4 4.6

AT2G18193 P-loop containing AAA+ ATPase 1 4.6 -1.0 21.1 3.2

AT5G13210 Uncharacterised conserved protein UCP015417 1 2.0 1.2 13.5 1.9

AT3G01600 NAC domain containing protein 44 ANAC044 1 4.2 2.3 12.7 4.3

AT5G38005 unknown gene 1 2.9 3.9 6.6 3.7

AT1G24095 Putative thiol-disulphide oxidoreductase DCC 1 2.7 1.6 5.8 2.9

AT2G28830 PLANT U-BOX 12 1 1.7 1.3 4.6 2.1


AT5G08030 PLC-like phosphodiesterases superfamily 1 7.3 6.3 3.4 5.8


AT3G61198 unknown gene 1 1.6 1.5 2.4 2.4

ATMG01090 mitochondrial ATP synthase 9 Yes 1 4.1 4.6 2.4 2.1

ATMG00630 unknown protein ORF110B 1 4.0 5.0 2.0 2.2

AT1G76530 Auxin efflux carrier Pin-likes 4 (PILS) 1 5.3 2.9 1.9 2.5

ATMG00665 NADH dehydrogenase 5B Yes 1 6.3 5.3 1.9 1.7

AT4G22880 leucoanthocyanidin dioxygenase (LDOX) 1 2.8 3.0 1.8 4.4

ATMG00670 unknown mitochondrial ORF275 Yes 1 2.6 2.5 1.7 1.8

ATMG00520 Type-II intron maturase MatR Yes 1 -6.5 -7.1 1.7 1.8

ATMG00660 unknown mitochondrial ORF149 yes 1 2.9 2.7 1.6 1.6

35

Figure Legends

Figure 1. ANAC017 regulates endogenous mitochondrial retrograde signalling marker genes. (A) Developmentally-matched seedlings at stage 1.04

were collected in pools, and relative mRNA levels were measured by qRT-PCR

(n=3-8) and normalised to Col-0 samples (±SE). (B) Developmentally-matched

seedlings at stage 1.04 of different wild-type, mutant and complemented

genotypes were collected in triplicate pools, relative mRNA levels were measured

by qRT-PCR (n=3) and normalised to Col-0 samples (±SE). Statistically

significant changes between the 5 genotypes were tested by ANOVA (p<0.01),

changes compared to Col-0 or in between mutant genotypes were assessed by

post-hoc student’s t-test and are indicated (* p<0.05; ** p<0.01, *** p<0.001).

Figure 2. Genome-wide transcript analysis by RNAseq. (A) Heat map of 98

transcripts commonly differential in rpotmp ANAC017 vs Col-0 and in atphb3

ANAC017 mutants. Colour bar indicates linear fold change. (B) Heat map of

DUF295 domain containing transcripts differentially expressed in rpotmp

ANAC017 and/or atphb3 ANAC017 mutants. Colour bar indicates log 2 fold

change. (C) Heat map of ANAC017-independent differentially expressed

transcripts in both rpotmp ANAC017 and atphb3 ANAC017 mutants. Colour bar

indicates linear fold change. (D) Heat map of mitochondrially-encoded transcripts

differentially expressed in rpotmp ANAC017 and/or atphb3 ANAC017 mutants.

Asterisks indicate statistically significant differences between rpotmp ANAC017

and rpotmp anac017 mutants (*** p<0.001).

Figure 3. Retrograde signalling targets specific changes in energy metabolism. (A) Relative abundance of metabolites as determined by GC-MS

analysis in same sample set used for RNAseq analysis (developmental stage

1.04) (±SE). Statistically significant differences with Col-0 or in between mutant

genotypes are marked with asterisks. (n=3-7; # p<0.10; * p<0.05). (B)

Representative model of enzymatic pathways linked to mitochondrial

metabolism. Metabolites are marked as up or down if significantly altered in 3 out

36

of 4 mitochondrial mutant genotypes. Enzymatic reactions performed by proteins

for which a corresponding transcript is commonly differential in rpotmp ANAC017

and atphb3 ANAC017 mutants vs Col-0 are indicated. Abbreviations: G3P:

glyceraldehyde-3-phosphate; 1,3-BPG: 1,3-bisphosphoglycerate; 3PG: 3-

phosphoglycerate; 2PG: 2-phosphoglycerate; PEP: phosphoenolpyruvate;

GABA: gamma-aminobutyric acid; TCA: tricarboxylic acid cycle; UQ: ubiquinone;

UQH2: ubiquinole.

Figure 4. ANAC017 controls AOX protein abundance and activity in mitochondrial mutants. (A) Western blot analysis of protein abundance in

isolated mitochondria of indicated genotypes. (B) Oxygen consumption of purified

mitochondria was measured using a Clark-type oxygen electrode in the presence

of succinate and a range of activators and inhibitors. Ratio’s between different

rates were calculated as follows. III/II: State III respiration (succinate and

ADP)/State II respiration (succinate); KCN+DTT+Pyr/III: maximised KCN-

resistant respiration (succinate, ADP, KCN, dithiothreitol, pyruvate)/State III

respiration. Statistically significant between the 5 genotypes were tested by

ANOVA (p<0.01), changes compared to Col-0 or in between mutant genotypes

were assessed by post-hoc student’s t-test and are indicated (* p<0.05; ** p<0.01,

*** p<0.001; n=3-4).

Figure 5. Loss of ANAC017 in mitochondrial mutants results in developmental defects and reduced plant performance. (A) Comparison of

15-day old seedlings of different genotypes. (B) Rosette leaf surface area of

different genotypes over time. Black asterisks indicate statistically significant

differences of Col-0 against all four mutant genotypes (±SE, n=25-31), blue

asterisks indicate statistical differences between atphb3 ANAC017 and atphb3

anac017 mutants, orange asterisks indicate statistical differences between

rpotmp ANAC017 and rpotmp anac017 (*** p<0.001). (C) Phenotypes of 5-week

old single and double mutants, inset shows magnification of leaf with chlorotic

lesions. (D) Inflorescence height and percentage of flowering plants of different

genotypes at day 36 and day 39 after stratification Statistically significant between

the 5 genotypes were tested by ANOVA (p<0.01), changes compared to in

37

between mutant genotypes were assessed by post-hoc student’s t-test and are

indicated (±SE, n=25-31, *** p<0.001). (E) Dry inflorescence weight at maturity

and total seed production in mg/plant Statistically significant between the 5

genotypes were tested by ANOVA (p<0.01), changes compared to Col-0 or in

between mutant genotypes were assessed by post-hoc student’s t-test and are

indicated (* p<0.05; ** p<0.01, *** p<0.001; n=4-7 pools of 3-8 plants) (F) Primary

root length after 7 days of incubation on vertical MS media plates for different

genotypes. Plates were unsupplemented (untreated), or supplemented with 50

µM antimycin A (±SE, n=10-20). Two-way ANOVA was performed with the 5

genotypes and two treatments, and significant differences (p<0.001) were

observed for genotype, treatment and genotype x treatment. Post-hoc pairwise

student’s t-tests were then performed against Col-0 or in between single and

double mutants as indicated (*** p<0.001).

02468

10121416

Col-0 rpotmp rpotmpanac017

atphb3 atphb3anac017

rela

tive

mR

NA

abun

danc

e UGT74E2

05

1015202530354045

Col-0

At2g41730

00.5

11.5

22.5

33.5

44.5

5

Col-0

AOX1a

ANAC017 ANAC017rpotmp rpotmp

anac017atphb3 atphb3

anac017ANAC017 ANAC017rpotmp rpotmp

anac017atphb3 atphb3

anac017ANAC017 ANAC017

**

****

*

**

***

**

******

***

** ***

*

0

2

4

6

8

10

12

rela

tive

mR

NA

abun

danc

e UGT74E2

05

10152025303540 At2g41730

***

***

******

**

*

00.5

11.5

22.5

33.5

44.5

5 AOX1a

* **

***

B

A

Figure 1. ANAC017 regulates endogenous mitochondrial retrograde signalling marker genes. (A) Developmentally-matched seedlings at stage 1.04 were collected in pools, and relative mRNA levels were measured by qRT-PCR (n=3-8) and normalised to Col-0 samples (±SE). (B) Developmentally-matched seedlings at stage 1.04 of different wild-type, mutant and complemented genotypes were collected in tripli-cate pools, relative mRNA levels were measured by qRT-PCR (n=3) and normalised to Col-0 samples (±SE). Statistically significant changes between the 5 genotypes were tested by ANOVA (p<0.01), changes compared to Col-0 or in between mutant genotypes were assessed by post-hoc student’s t-test and are indicated (* p<0.05; ** p<0.01, *** p<0.001).

rpot

mp

ANAC

017

rpot

mp

anac

017

atph

b3 A

NAC

017

atph

b3 a

nac0

17

Col

-0

rpot

mp

ANAC

017

rpot

mp

anac

017

atph

b3 A

NAC

017

atph

b3 a

nac0

17

(A) (B)

(C)

(D)

Col-0

-4 40

rpot

mp

ANAC

017

rpot

mp

anac

017

atph

b3AN

AC01

7at

phb3

anac

017

***

-10 0 80

-2 1 10

-5 0 5

Figure 2. Genome-wide transcript analysis by RNAseq. (A) Heat map of 98 transcripts commonly differen-tial in rpotmp ANAC017 vs Col-0 and in atphb3 ANAC017 mutants. Colour bar indicates linear fold change. (B) Heat map of DUF295 domain containing transcripts differentially expressed in rpotmp ANAC017 and/or atphb3 ANAC017 mutants. Colour bar indicates log 2 fold change. (C) Heat map of ANAC017-independent differentially expressed transcripts in both rpotmp ANAC017 and atphb3 ANAC017 mutants. Colour bar indicates linear fold change. (D) Heat map of mitochondrially-encoded transcripts differentially expressed in rpotmp ANAC017 and/or atphb3 ANAC017 mutants. Asterisks indicate statistically significant differences between rpotmp ANAC017 and rpotmp anac017 mutants (*** p<0.001).

Col

-0

0

200

400

600

800

1000

1200

1400

1600

L-Glutam

ine0

20

40

60

80

100

120

Amino Acids

0123456789

10

rela

tive

abun

danc

e

00.2

0.40.60.8

11.21.41.61.8

2

rela

tive

abun

danc

e

TCA cycle

0

0.5

1

1.5

2

2.5

3Glycolysis

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6Fatty acids

*

*

* ** * *

**

* *

* *

*

*

*

*

** * * *

*

** *

* *

*

*

*

**

* ** *

* *

*

*

0

1

2

3

4

5

6Carbohydrates

* **

*

** *

**

*

*

*

*

** * *

*

*

* * **

##

#

##

#

#

#

#

#

#

Col-0

rpotmp ANAC017

rpotmp anac017

atphb3 ANAC017

atphb3 anac017

Pyruvate

GABA

Glutamate

Succinylsemialdehyde

AcetylCoA

Citrate

IsocitrateAconitate

2-Oxoglutarate

Succinate

FumarateMalate

OxaloacetateAspartate

Glutamate

GABA shuntTCA cycle

NADH

NADH

NADH

NADH

NADH

Mitochondrion

Cytosol

2-Oxoglutarate

FADH2

Proline

Asparagine

Lysine

Alanine

glucose

glucose-6-phosphate

fructose-6-phosphate

fructose-1,6-bisphosphate

G3P

3PG

2PG

1,3-BPGNADH

PEP

trehalose-6-phosphate trehalose

Glycerate

P5C

UQH2 UQ

SerineGlycine

metabolite upmetabolite down transcript down

transcript up

Lactic acid

A

B

*

**

**

*

**

*

*

*

0

0.5

1

1.5

2

2.5

3

3.5

Ferment.

*

#

*

metabolite up in atphb3 anac017 Figure 3. Retrograde signalling targets specific changes in energy metabolism. (A) Relative abundance of metabolites as determined by GC-MS analysis in same sample set used for RNAseq analysis (developmental stage 1.04) (±SE). Statistically signifi-cant differences with Col-0 or in between mutant genotypes are marked with asterisks. (n=3-7; # p<0.10; * p<0.05). (B) Representa-tive model of enzymatic pathways linked to mitochondrial metabolism. Metabolites are marked as up or down if significantly altered in 3 out of 4 mitochondrial mutant genotypes. Enzymatic reactions performed by proteins for which a corresponding transcript is commonly differential in rpotmp ANAC017 and atphb3 ANAC017 mutants vs Col-0 are indicated. Abbreviations: G3P: glyceralde-hyde-3-phosphate; 1,3-BPG: 1,3-bisphosphoglycerate; 3PG: 3-phosphoglycerate; 2PG: 2-phosphoglycerate; PEP: phosphoenolpy-ruvate; GABA: gamma-aminobutyric acid; TCA: tricarboxylic acid cycle; UQ: ubiquinone; UQH2: ubiquinole.

0

0.2

0.4

0.6

0.8

1

1.2

1.4



00.20.40.60.8

11.21.41.61.8

Col-0 rpotmprpotmpanac017

atphb3atphb3anac017

III/II

**

***

*

***

rpotm

p

anac

017

rpotm

p

ANAC017

atphb

3

anac

017

atphb

3

ANAC017

Col-0

AOX

ATP-B

B

A

Figure 4. ANAC017 controls AOX protein abundance and activity in mitochondrial mutants. (A) West-ern blot analysis of protein abundance in isolated mitochondria of indicated genotypes. (B) Oxygen consump-tion of purified mitochondria was measured using a Clark-type oxygen electrode in the presence of succinate and a range of activators and inhibitors. Ratio’s between different rates were calculated as follows. III/II: State III respiration (succinate and ADP)/State II respiration (succinate); KCN+DTT+Pyr/III: maximised KCN-resist-ant respiration (succinate, ADP, KCN, dithiothreitol, pyruvate)/State III respiration. Statistically significant between the 5 genotypes were tested by ANOVA (p<0.01), changes compared to Col-0 or in between mutant genotypes were assessed by post-hoc student’s t-test and are indicated (* p<0.05; ** p<0.01, *** p<0.001; n=3-4).

020406080

100120140160180



State III

nmol

O2

min

m

g p

rote

in

020406080

100120140160180



KCN+DTT+Pyr

KCN+DTT+Pyr/III

*

**

*

*

-1-1

nmol

O2

min

m

g p

rote

in-1

-1

ratio

ratio

0

100

200

300

400

500

600

dry

wei

ght (

mg)

Inflorescence weight

0

500

1000

1500

2000

2500

3000

3500

15 18 21 24 27 30 33 36

mm

2

Days

Rosette area

Col-0rpotmp anac017rpotmp ANAC017atphb3 anac017atphb3 ANAC017

rpotmpanac017

rpotmpANAC017

atphb3anac017

atphb3ANAC017

Col-0A

B

****

*** *** ******

***

***

******

***

******

***

***

***

***0

50

100

150

200

250

Col-0 rpotmpANAC017

rpotmpanac017

atphb3ANAC017

atphb3anac017

mm

Inflorescence height Day 36

Day 39

0

20

40

60

80

100

120

Col-0 rpotmpANAC017

rpotmpanac017

atphb3ANAC017

atphb3anac017

%

Flowering at day 36

D

***

***

***

***

C

1 cm

rpotmpanac017

rpotmpANAC017

atphb3anac017

atphb3ANAC017

0

5

10

15

20

25

30

prim

ary

root

leng

th (m

m)

Untreated

0

2

4

6

8

10

12

14 Antimycin A

*** *** ******

******

***

***

*** ******

rpotmp ANAC017

atphb3 ANAC017

atphb3 ANAC017

E F

0

20

40

60

80

100

120

140 Seed production

rpotmp ANAC017

atphb3 ANAC017

* ***

*

***

***

**

*

atphb3 ANAC017

rpotmp ANAC017

Figure 5. Loss of ANAC017 in mitochondrial mutants results in developmental defects and reduced plant performance. (A) Comparison of 15-day old seedlings of different genotypes. (B) Rosette leaf surface area of different genotypes over time. Black asterisks indicate statistically significant differences of Col-0 against all four mutant genotypes (±SE, n=25-31), blue asterisks indicate statistical differences between atphb3 ANAC017 and atphb3 anac017 mutants, orange asterisks indicate statistical differences between rpotmp ANAC017 and rpotmp anac017 (*** p<0.001). (C) Phenotypes of 5-week old single and double mutants, inset shows magnification of leaf with chlorotic lesions. (D) Inflorescence height and percentage of flowering plants of different genotypes at day 36 and day 39 after stratification Statistically significant between the 5 genotypes were tested by ANOVA (p<0.01), changes compared to in between mutant genotypes were assessed by post-hoc student’s t-test and are indicated (±SE, n=25-31, *** p<0.001). (E) Dry inflorescence weight at maturity and total seed production in mg/plant Statistically significant between the 5 genotypes were tested by ANOVA (p<0.01), changes compared to Col-0 or in between mutant genotypes were assessed by post-hoc student’s t-test and are indicated (* p<0.05; ** p<0.01, *** p<0.001; n=4-7 pools of 3-8 plants) (F) Primary root length after 7 days of incubation on vertical MS media plates for different genotypes. Plates were unsupplemented (untreated), or supplemented with 50 μM antimycin A (±SE, n=10-20). Two-way ANOVA was performed with the 5 genotypes and two treatments, and significant differences (p<0.001) were observed for genotype, treatment and genotype x treatment. Post-hoc pairwise student’s t-tests were then performed against Col-0 or in between single and double mutants as indicated (*** p<0.001).

rpotmp ANAC017

0

500

1000

1500

2000

2500

3000

3500

4000

18 23 28 33 38 43 48

mm

time (days)

Col-0anac017rpotmp ANAC017rpotmp anac017rpotmp anac017 35S::ANAC017.3

******

******

*

***

***

***

***

***

***

*******

rpotmp ANAC017 rpotmp anac017rpotmp anac01735S::ANAC017.3

A

B

Supplementary Figure 1. Complementation of rpotmp anac017. (A) Rosette leaf surface area of differ-ent genotypes over time and average seed production per plant. Statistically significant changes between the 5 genotypes were tested by ANOVA (p<0.01), changes compared to Col-0 or in between mutant geno-types were assessed by post-hoc student’s t-test and are indicated. Black asterisks indicate statistically significant differences of Col-0 against all three rpotmp mutant genotypes (±SE, n=14-16), green asterisks indicate statistical differences of indicated line with rpotmp anac017 35S::ANAC017, yellow asterisks indicate statistical differences between indicated line and rpotmp anac017 (* p<0.05; *** p<0.001). (B) Phe-notypes of 39-day-old mutant plants. Red arrows indicate occurrence of chlorotic lesions.

0102030405060708090 Seed produc on

35S::ANAC017.3

mg/

plan

t

*****

***

*****2

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35re

lativ

e gr

owth

rat

eRelative growth rate Col-0

rpotmp ANAC017rpotmp anac017atphb3 ANAC017atphb3 anac017

**

**

**

*

*** **

*

*

**

*** *****

**

Supplementary Figure 2. Relative growth rate analysis. Relative growth rates of soil-grown plants (rosette surface area) were measured for the different genotypes at different time intervals (n=13-31). Statistically significant changes between the 5 genotypes were tested by ANOVA (p<0.05), changes compared to Col-0 or in between mutant genotypes were assessed by post-hoc student’s t-test and are indicated (* p<0.05; ** p<0.01, *** p<0.001).

*

0

2

4

6

8

10

12

14

16

0 3

rela

tive

mRN

A ab

unda

nce

time (h)

AOX1aCol-0anac017rpotmp ANAC017rpotmp anac017

Antimycin A 3h in Col-0

rpotmp ANAC017 vs Col-0

atphb3 ANAC017 vs Col-0

Supplementary Figure 3. The dual role of ANAC017 in constitutive and acute mitochondrial retro-grade signalling. (A) 2-3 week-old in vitro grown seedlings were spray treated with 50 μM antimycin A. Samples were collected at the indicated times and mRNA levels were measured by qRT-PCR. Error bars indicate SE. Statistically significant changes between the 5 genotypes were tested by two-way ANOVA (p<0.001 for genotype, treatment, and genotype x treatment), changes compared to Col-0 or in between mutant genotypes were assessed by post-hoc student’s t-test and are indicated. Asterisks indicate statisti-cally significant changes vs Col-0 at 0h, or between indicated genotypes within the same time point accord-ing to student’s t-tests (* p<0.05, ** p<0.01). (B) Comparison of RNAseq data with antimycin A ATH1 microarray data. Venn diagrams representing common and unique transcripts differentially expressed by transient treatment with antimycin A (Ng et al., 2013), and constitutive responses to genetic mitochondrial defects in rpotmp ANAC017 and atphb3 ANAC017 plants vs Col-0 (this study). (C) Histogram representing the distance of ANAC017 binding motifs to the transcriptional start site (TSS) of 92 core ANAC017-regulat-ed nuclear-encoded genes responding to constitutive mitochondrial dysfunction.

A B

* *

**

*

*

*

**

Antimycin A treatment

0

2

4

6

8

10

12

Freq

uenc

y

Distance from TSS

C

ANAC017 promoter binding motifs

retrograde signalling caused by heritable mitochondrial ... · retrograde controls in chloroplasts...

Documents