rapid acclimation of juvenile corals to co2‐mediated...

Rapid acclimation of juvenile corals to CO2-mediatedacidification by upregulation of heat shock protein andBcl-2 genes

A. MOYA,*† ‡ L. HUISMAN,*§ S . FORET, *¶ J . -P . GATTUSO,† ‡ D. C. HAYWARD,¶ E. E . BALL*¶and D. J . MILLER***

*ARC Centre of Excellence for Coral Reef Studies, James Cook University, Townsville, Qld 4811, Australia, †Laboratoired’Oc�eanographie de Villefranche, INSU-CNRS, 181 Chemin du Lazaret, 06230 Villefranche-sur-mer, France, ‡SorbonneUniversit�es, UPMC Univ. Paris 06, Observatoire Oc�eanologique, 06230 Villefranche-sur-mer, France, §Section of Computational

Science, Universiteit van Amsterdam, Science Park 904, 1098 XH, Amsterdam, The Netherlands, ¶Evolution, Ecology and

Genetics, Research School of Biology, Australian National University, Bldg. 46, Canberra, ACT 0200, Australia, **School of

Pharmacy and Molecular Sciences, James Cook University, Townsville, Qld 4811, Australia

Abstract

Corals play a key role in ocean ecosystems and carbonate balance, but their molecular

response to ocean acidification remains unclear. The only previous whole-transcriptome

study (Moya et al. Molecular Ecology, 2012; 21, 2440) documented extensive disruption of

gene expression, particularly of genes encoding skeletal organic matrix proteins, in juve-

nile corals (Acropora millepora) after short-term (3 d) exposure to elevated pCO2. In this

study, whole-transcriptome analysis was used to compare the effects of such ‘acute’ (3 d)

exposure to elevated pCO2 with a longer (‘prolonged’; 9 d) period of exposure beginning

immediately post-fertilization. Far fewer genes were differentially expressed under the

9-d treatment, and although the transcriptome data implied wholesale disruption of

metabolism and calcification genes in the acute treatment experiment, expression of most

genes was at control levels after prolonged treatment. There was little overlap between

the genes responding to the acute and prolonged treatments, but heat shock proteins

(HSPs) and heat shock factors (HSFs) were over-represented amongst the genes respond-

ing to both treatments. Amongst these was an HSP70 gene previously shown to be

involved in acclimation to thermal stress in a field population of another acroporid coral.

The most obvious feature of the molecular response in the 9-d treatment experiment was

the upregulation of five distinct Bcl-2 family members, the majority predicted to be anti-

apoptotic. This suggests that an important component of the longer term response to ele-

vated CO2 is suppression of apoptosis. It therefore appears that juvenile A. milleporahave the capacity to rapidly acclimate to elevated pCO2, a process mediated by upregula-

tion of specific HSPs and a suite of Bcl-2 family members.

Keywords: Acropora millepora, Bcl-2, caspases, climate change, corals, heat shock proteins

Received 5 September 2014; revision received 17 November 2014; accepted 20 November 2014

Introduction

Whilst there is no doubt that coral reefs are in global

decline, it is not yet clear how much climate change

will influence their fate. One widely propagated view is

that the dual impacts of continuing high levels of CO2

emission—elevated seawater temperatures and decreas-

ing ocean pH—will, to use a phrase coined in a some-

what different context (Pandolfi et al. 2005), inevitably

push coral reefs down the ‘slippery slope to slime’. Sea

surface temperatures have risen by a global average of

0.6 °C over the last century (Roemmich et al. 2012) and

Correspondence: Aurelie Moya, Fax: +61 7 4781 6722;

E-mail: [email protected] and David J. Miller,

Fax: +61 7 4781 6078; E-mail: [email protected]

© 2014 John Wiley & Sons Ltd

Molecular Ecology (2015) 24, 438–452 doi: 10.1111/mec.13021

are projected to rise by a further 3.0 °C by the end of

the 21st century under the RCP8.5 scenario (Collins

et al. 2013). In addition, ocean pH has decreased by

0.1 unit since the 1950s and is projected to decrease a

further 0.2–0.3 units by the end of this century (Orr

2011). Similar projections based on earlier studies gave

rise to the publication of an influential study (Hoegh-

Guldberg et al. 2007), predicting the inevitable demise

of coral reefs under ‘business as usual’ CO2 emission

scenarios, and these predictions have been supported

by some recent mesocosm experiments (Dove et al.

2013). However, other studies, based on a diversity of

sources and a longer term perspective of hundreds of

million years, suggest that change will be slower and

less dramatic than suggested by the studies mentioned

above (Pandolfi et al. 2011). In view of these continuing

uncertainties, it is therefore crucial to understand the

response of corals to ocean acidification (OA) at the

molecular level.

A major limitation in predicting the fate of coral reefs

is our lack of understanding of the molecular responses

of corals, even to individual stressors, and this is partic-

ularly the case with respect to OA. Most research to

date on the impact of OA on corals has focused on

growth and rates of survival. Considerably less atten-

tion has been paid to the molecular mechanisms

involved in the response to elevated CO2, which is pre-

dicted to decrease calcification rates in corals and other

calcifying organisms, as more energy should be

required due to the lower availability of carbonate ions

(Gattuso et al. 1999; Doney et al. 2009).

In contrast with the paucity of molecular data avail-

able, the broader biological literature detailing coral

responses to OA is large and variable in both methodol-

ogies and results, even for early developmental stages.

The recent literature on this restricted topic is summa-

rized by Albright (2011) and is tabulated in Chua et al.

(2013a) and Cumbo et al. (2013), the latter listing

approximately 20 studies dealing with the effects of

warming and acidification, either alone or in combina-

tion. Although the experiments are heterogeneous in

design, length and execution, the general picture that

emerges is that warming has a much greater immediate

effect on larval development and survival than CO2

concentration does. In general, acidification alone led to

few and small effects. Amongst the exceptions are the

results of Albright et al. (2010) on Acropora palmata and

Albright & Langdon (2011) on Porites asteroides. In both

studies, OA had a significant effect on survival and

growth post-settlement. However, the reduction in

growth reported in these two studies is not necessarily

inconsistent with other studies which reported little or

no effect, as the former experiments lasted considerably

longer (50 d for A. palmata and 49 d for P. asteroides).

This much greater length of experiment would allow

small growth effects, not apparent in shorter experi-

ments, to summate into larger ones. Different experi-

mental designs were also the cause of some of the

variable responses of early developmental stages to OA,

with some unanticipated variables proving important.

For example, whether the tiles used for settlement were

conditioned under ambient or acidified conditions

resulted in quite different bacterial communities and

settlement successes (Doropoulos et al. 2012).

Few experiments have tested the effects of OA on

gene expression, and they involved a limited group of

candidate genes. These studies are summarized in

Table 1, with further discussion below when warranted.

Kaniewska et al. (2012) used EST microarrays (Grasso

et al. 2008) to investigate the impact of chronic (28 d)

exposure to elevated pCO2 on adult colonies of Acropora

millepora. They concluded that the most significant

impacts were on genes involved in respiration and sym-

biosis rather those directly involved in calcification. The

most comprehensive molecular investigation of the

impacts of elevated CO2 on corals (Moya et al. 2012)

studied the ‘acute’ (3 d) response of juvenile A. mille-

pora to 380 and 750 ppm CO2(air) (448 and 677 latm) on

the whole transcriptome and revealed considerable dis-

ruption in levels of expression of metabolic genes as

well as of many genes implicated in skeleton deposi-

tion. In that earlier study, primary polyps were exposed

for 3 d to elevated CO2 starting 1 day after settlement,

when calcification was just beginning. To investigate

what happens over a longer 9-day period, but with

exposure starting at fertilization, this study was carried

out using the same species, facilities and equipment. In

both cases, the experiment was terminated at 4 d after

settlement (total exposure time of 3 and 9 d for the

‘acute’ and ‘prolonged’ treatment experiments, respec-

tively). In each year, levels of gene expression were

compared between the two treatments (control and

acidified treatments) after mapping the individual reads

onto the reference transcriptome (Moya et al. 2012). Pre-

liminary analysis led us to focus specifically on the heat

shock and apoptotic repertoires, which have also been

implicated in the responses to thermal stress (Kenkel

et al. 2013; Polato et al. 2013) and in thermal stress toler-

ance of natural populations (Barshis et al. 2013; Palumbi

et al. 2014). This study is not only the first to document

the response of a complete coral gene set to longer term

CO2 stress, but also the first to dissect the responses of

individual heat shock protein (HSP) and Bcl-2 isoforms

in what are complex gene repertoires. This latter point

is critical, because these are typically large gene families

in metazoans, only a few members of which may

respond to specific stressors. For example, the Acropora

digitifera genome encodes 23 distinct HSP70 family


RAPID ACCLIMATION OF JUVENILE CORALS TO OA 439

members (Shinzato et al. 2012a), only a small number of

which are likely to be stress responsive (Richter et al.

2010).

The transcriptomic response of A. millepora juveniles

to prolonged CO2 stress is quite different to that

observed under acute CO2 stress. Whereas many genes

responded under acute exposure to CO2, suggesting

that growth and skeletogenesis would be severely dis-

rupted, expression data implied that most metabolic

processes had adjusted to 750 ppm CO2(air) (677 latm)

after 9 d of exposure. The acclimation to elevated pCO2

implied by these results appears to be facilitated by

higher levels of expression of a core set of HSPs and

Bcl-2 family members. Consistent with predictions, only

a small number of HSP70 family members respond to

CO2 stress; one of the two that are up-regulated under

prolonged CO2 stress is also implicated in thermal

stress tolerance in natural populations of Acropora hya-

cinthus (Barshis et al. 2013; Palumbi et al. 2014) and may

thus prove useful as a molecular marker for stress in

general. Although at least some Acropora species are

apparently able to rapidly acclimate to either CO2 or

thermal stress, it would be premature to immediately

revise the predictions concerning the future of coral

reefs, as multiple stressors may have more damaging

impacts, acclimation alone may not equip natural popu-

lations to survive frequent and severe thermal stress,

and species may differ in their ability to acclimate.

However, given the dominance of Acropora in the Indo-

Pacific, the ability of at least some species to rapidly

acclimate to stress constitutes a glimmer of hope for the

survival of corals and the maintenance of coral reefs.

Methods

Maintenance of coral material

Prior to the spawning event in November 2010, fifteen

adult colonies of Acropora millepora were collected off the

coast of Orpheus Island, Queensland, Australia (under

GBRMPA Permit No G10/33174.1). These were trans-

ported to the aquaculture facilities at James Cook Univer-

sity in Townsville where they were maintained in outdoor

flow-through aquaria filled with 5 lm filtered seawater.

Eggs and sperm were collected from each of these 15 colo-

nies individually, immediately after spawning, and used

to make three independent crosses (each cross containing

eggs and sperm from five colonies) corresponding to three

biological replicates. Each biological replicate was split

between a control and an experimental tank.

Experimental design

For the 9-d exposure experiment, embryos were trans-

ferred immediately after fertilization to seawater that

had been equilibrated with air containing 380 ppm

Table 1 Summary of published data on the effects of ocean acidification on gene expression in juvenile corals

Authors Species Genes

Length of

experiment (d)

Atmospheric

CO2 level (ppm) Result

Nakamura

et al. (2012)

Acropora digitifera HSF1

HSP70

HSP90

7 1000 No significant difference

No significant difference

No significant difference

Putnam et al. (2013) Pocillopora

damicornis

HSP70 9 600 Up-regulated

Ogawa et al. (2013) Acropora aspera CoCA2

CoCA3

Glycogen synthase

Glycogen

phosphorylase

GADPH

1, 4, 6, 14 194–756 General pattern: up-regulated

at 6 d and down-regulated

at 14 d

Kaniewska

et al. (2012)

Acropora millepora Microarray

containing 8606

unigene clusters

(Grasso et al. 2008)

0, 1, 28 600–790and 1010–1350

Discussed in the

present paper

Moya et al. (2012) Acropora millepora Entire transcriptome 3 750 and 1000 Discussed in the

present paper

Moya et al.

(present study)

Acropora millepora Entire transcriptome

(Moya et al. 2012)

3, 9 750

The Kaniewska et al. (2012) study is included as it was on Acropora millepora and is thus directly relevant to the present work. HSPs,

heat shock proteins; HSFs, heat shock factors.


440 A. MOYA ET AL.

(pHNBS~8.16) or 750 ppm (pHNBS~7.96) CO2, reflecting

the control condition and medium CO2 scenario for

the 21st century. Note that, in the case of the short-

term (‘acute’) exposure experiment, larvae were culti-

vated under ambient conditions (380 ppm pCO2(air),

pHNBS~8.16) and transferred to the experimental

tanks 24 h after settlement for 3 d of CO2 exposure

(Moya et al. 2012). As in the previous experiment,

settlement was induced 5 d after fertilization by the

introduction of unglazed terracotta tiles that had

been conditioned with crustose coralline algae under

ambient seawater conditions and then autoclaved.

Figure 1 summarizes the experimental design used

for both the earlier (Moya et al. 2012) and present

experiments. Exposure to 750 ppm of CO2 had no

obvious effect on embryonic development, larval sur-

vivorship and metamorphosis of A. millepora larvae

(Chua et al. 2013a,b).

The desired CO2 concentration was produced using a

CO2 mixing system developed by Munday et al. (2009).

pH was measured on the National Bureau of Standards

(NBS) scale with a portable metre (Hach HQ11D) cali-

brated daily with NBS buffers (pH 4 and 7). Tempera-

ture, pH, oxygen concentration and total alkalinity were

monitored daily at 11:00. The seawater chemistry was

estimated using CO2SYS (Lewis & Wallace 1998). Average

pCO2(seawater) was estimated to be 448 and 677 latm for

the 9-d experiment. Seawater carbonate parameters are

shown in Table 2.

Sampling and RNA extraction

The primary polyps (~50 per biological sample and per

condition) were carefully removed from the tiles (1–3tiles per condition) with a sterile scalpel, pooled and

immediately snap-frozen in liquid nitrogen and stored

at �80 °C until further processed. Total RNA was

extracted from each sample of 50 using TRIzol Reagent

(Invitrogen) according to the Chomczynski method

(Chomczynski & Sacchi 1987) and dissolved in RNase-

free water. RNA quantity and quality were assessed

using a NanoDrop ND-1000 spectrometer and denatur-

ing gel electrophoresis using standard methods (Sam-

brook & Russell 2001). Before being shipped on dry ice

to the Macrogen sequencing facilities in Seoul, South

Korea, each RNA sample was precipitated in ethanol

and sodium acetate (29 and 0.19 sample volume,

respectively) and stored at �80 °C.

High-throughput sequencing and data analysis

mRNA isolation and library construction were per-

formed by Macrogen (South Korea). The libraries were

sequenced using the Illumina HiSeq2000 platform, pro-

ducing an average across the six samples of 19 million

reads per sample (100 bp) for the 9-d CO2 exposure.

Sequencing reads were mapped onto the A. millepora

transcriptome assembly (Moya et al. 2012) using the BOW-

TIE mapping software version 0.12.7 (Langmead et al.

2009). Differential gene expression was inferred based on

C1 C2 C3

Ac1 Ac2 Ac3

C1 C2 C3

Pr1 Pr2 Pr3

380 ppm CO2 (air)

750 ppm CO2 (air)

3 day elevated CO2

ACUTE

Sampling

380 ppm CO2 (air)

750 ppm CO2 (air)

Sampling

PROLONGED

Fertilization

PlanulaEgg

9 day exposure to elevated CO2

Basal platedeposition

Fully developedcalcification

Settlementday 5

Fig. 1 Summary of the experimental design. For the prolonged (9-d) exposure to pCO2, embryos were transferred immediately after

fertilization to seawater that had been equilibrated with air containing 380 ppm (pH~8.16) or 750 ppm (pH~7.96), reflecting the con-

trol condition and medium CO2 scenario for the 21st century. For the short-term (‘acute’) exposure described in Moya et al. (2012),

larvae were cultivated under ambient conditions and transferred to the experimental tanks 24 h after settlement for 3 d of CO2 expo-

sure. In both cases, settlement was induced 5 d after fertilization by the introduction of unglazed terracotta tiles conditioned with

crustose coralline algae under ambient seawater conditions. Four days later, primary polyps (~50 per condition, three biological

replicates per condition) were snap-frozen in liquid nitrogen prior to RNA extraction for RNA-seq analysis. C, Control; Ac, Acute;

Pr, Prolonged.



these mapping counts using the edgeR package (Robin-

son et al. 2010). Gene ontology (GO) enrichment analysis

was performed using the GOseq package (Young et al.

2010) that accounts for selection bias such as gene size or

expression level. The RNA-seq reads used in this study

have been submitted to the NCBI Gene Expression Omni-

bus (GEO) database under Accession no. GSE61114. The

corresponding data for the ‘acute’ treatment experiment

are available from the same source under Accession no.

GSE33016. Note that, rather than a common reference for

both the 3- and 9-d experiments, a distinct set of controls

was employed for the 9-day experiment.

For each gene category of interest, BlastP and

HMMER domain searches (e-value cut-off = 1e�5) were

performed on the A. millepora transcriptome (Moya et al.

2012). An additional blast onto the NCBI nr database

confirmed the identification of each sequence.

Results

Whole-transcriptome analysis was used to compare Acro-

pora millepora juveniles that had been exposed to 380 or

750 ppm CO2(air) (414 or 669 latm) for a period of 9-d

(‘prolonged’ exposure—this experiment), as shown sche-

matically in Fig. 1, and these data were compared with

results of a previous 3-d ‘acute’ exposure treatment, car-

ried out using the same species and experimental format

(Moya et al. 2012). The previous (Moya et al. 2012) and

present experiments allow comparison of gene expres-

sion levels in primary polyps at the same time point, rela-

tive to their respective controls, but after different

periods of exposure to elevated pCO2. In the acute treat-

ment, elevated CO2 stress was initiated 24 h after settle-

ment (Moya et al. 2012), whereas in the case of prolonged

exposure, fertilized eggs were immediately exposed to

elevated CO2, with the resulting polyps then harvested at

the same age as in the acute experiment.

The transcriptomic responses to the acute and 9-d CO2

treatment are largely discrete

The transcriptional response to 9-d CO2 exposure dif-

fers both quantitatively and qualitatively to that under

acute exposure. Using a cut-off of adjusted P < 0.05,

20% of A. millepora transcripts were differentially

expressed in the acute treatment, whereas only 4% of

transcript clusters responded in the 9-d treatment group

(Fig. 2A). The prolonged exposure resulted in a pre-

dominantly positive transcriptional response, whereas

similar numbers of transcript clusters were up- and

down-regulated under acute exposure to CO2. More-

over, the 3- and 9-d responses were largely discrete

Table 2 Summary of seawater parameters in the control and acidified treatments

pCO2(air) (ppm) pHNBS Total alkalinity (lmol/kg) Temperature (°C) Ωaragonite pCO2(seawater) (latm)

Control 8.16 � 0.01 2028.8 � 5.0 27.5 � 0.2 2.87 � 0.08 414.2 � 2.0

pCO2 acute 7.96 � 0.01 1876.7 � 4.0 28.0 � 0.1 1.80 � 0.05 669.2 � 2.0

Control 8.12 � 0.07 2076.2 � 140.0 26.5 � 0.3 2.80 � 0.48 447.5 � 64.5

pCO2 chronic 7.96 � 0.06 2015.5 � 129.0 26.5 � 0.2 1.99 � 0.28 677.4 � 90.8

The saturation states of aragonite (Ωa) and pCO2 were estimated from pHNBS and total alkalinity using the computer package CO2SYS.

Values are mean � SD. NBS, National Bureau of Standards.

Up-regulated Down-regulated Non-regulated

11%

9%

80%

Acute

3 day, 750 ppm CO2 (air)

<1%3%

97%

Prolonged

9 day, 750 ppm CO2 (air)

305

Ac

5334 Pr1422

Shared up-regulatedgenes

36

Shared down-regulatedgenes

Ac

4812125

Pr

(A)

(B)

Fig. 2 Effects of the elevated CO2 treatments on the Acropora

millepora transcriptome. Part (A) of the figure shows the per-

centages of transcripts up- (green) and down-regulated (red) in

response to acute (3-d) and prolonged (9-d) CO2 treatments

(adjusted P < 0.05). The Venn diagrams shown in (B) indicate

the numbers of up-regulated (green) and down-regulated (red)

transcripts in the acute and prolonged CO2 treatments

(adjusted P < 0.05), as well as those common to both. Ac,

Acute; Pr, Prolonged.


442 A. MOYA ET AL.

(Fig. 2B); only 341 transcript clusters (0.65% of clusters;

305 and 36 up- and down-regulated, respectively)

responded under both acute and prolonged exposure to

CO2. Both sets of differentially expressed genes were

enriched in genes of unknown function (no blast hit

with a cut-off of 1e�5, possibly coral-specific genes),

compared to the transcriptome database. The percent-

ages of genes with unknown function were 56% and

47% in the 3- and 9-d treatments, respectively, com-

pared to 35% in the transcriptome database. Although

this result suggests the involvement of coral-specific

mechanisms, the fact that these genes lack annotation

precludes speculation concerning their nature.

A major characteristic of the acute response of

A. millepora primary polyps to elevated pCO2 is distur-

bance of the expression of membrane-associated or

secreted carbonic anhydrases (CAs) and genes associ-

ated with the skeletal organic matrix (SOM; Moya et al.

2012). This response appears to be specifically associ-

ated with acute CO2 stress, however, as levels of

expression of all but a few SOM proteins did not differ

significantly from controls after 9-d CO2 stress (Fig. 3).

Likewise, after 9-d CO2 exposure, no evidence was

found for the kind of general metabolic suppression

implied from previous studies of the acute transcrip-

tomic response.

As a preliminary approach to data analysis, over-rep-

resentation of GO terms in the different sets of differen-

tially expressed genes was examined using the Gene

Ontology database (Ashburner et al. 2000; Table S1,

Supporting information). Specific protein families iden-

tified as over-represented on the basis of this prelimin-

ary GO analysis were subjected to more comprehensive

annotation; note that, whereas the initial analyses used

the 56 000 transcript clusters as reference, in these fol-

low-up analyses, transcript clusters were consolidated

on the basis of sequence similarity and reference to gen-

ome data available for both Acropora digitifera (Shinzato

et al. 2011) and A. millepora (Foret et al., in preparation).

More comprehensive analyses of candidates belonging

to the categories ‘heat shock proteins’ and ‘apoptosis’

were undertaken, as described below.

Prolonged (9-d) exposure to elevated pCO2 inducesexpression of Bcl-2 family members

Amongst those genes responding specifically to 9-d ele-

vated pCO2, the Bcl-2 gene family was clearly over-rep-

resented. Eight transcriptome clusters encoding Bcl-2-

related proteins were differentially regulated in the 9-d

exposure experiment, but none of these showed signifi-

cant changes in expression after acute treatment. The

eight differentially regulated clusters are likely to corre-

spond to five Bcl-2 genes (for comparison, the A. digitif-

era genome encodes 8 Bcl-2 family members; Shinzato

et al. 2011), several of which fall into the vertebrate Bcl-

2 subfamilies based on sequence comparisons. Proper-

ties of these A. millepora Bcl-2 family members are sum-

marized in Table 3.

Each of the proteins encoded by four of these five dif-

ferentially expressed Bcl-2 genes contains four BH

domains; the exception, Cluster015074, is a particularly

interesting case because although both the A. millepora

and A. digitifera proteins appear to lack BH4 domains,

the Nematostella vectensis orthologue (XP_001628866)

clearly contains a BH4 domain. Although Bcl-2 family

members have a range of functions (Hardwick & Soane

2013), the BH4 domain composition and sequence

Bcl-2proteins

Fluorescentproteins

USPs Carbonicanhydrases

SOMproteins

SCRiPs3 day 9 day 3 day 9 day 3 day 9 day 3 day 9 day 3 day 9 day 3 day 9 day

100%

90%

80%

70%

60%

50%

40%

30%

20%

10%

0%

(n = 15) (n = 12) (n = 12) (n = 22) (n = 35) (n = 8)

Up-regulated Down-regulated Non-regulated

Fig. 3 Changes in levels of expression of

various families of genes. The bar graphs

summarize the impact of acute and pro-

longed pCO2 treatments on the total

complement of Acropora millepora tran-

scripts in various families of genes (Bcl-2,

fluorescent proteins, USPs, carbonic an-

hydrases, skeletal organic matrix proteins

—excluding SCRiPs, and SCRiPs). Tran-

scripts whose levels changed significantly

(adjusted P < 0.05) are indicated as fol-

low: significant upregulation shown in

green, significant downregulation in red

and nonregulated in blue. Total number

of transcripts in each category is indi-

cated above the corresponding bars. USP,

universal stress protein; SOM, skeletal

organic matrix proteins; SCRiP, small

secreted cysteine-rich protein.



similarity with specific mammalian proteins (Table 3)

imply that the majority of the Bcl-2s that are up-regulat-

ed under pCO2 stress are likely to be anti-, rather than

pro-apoptotic.

Elevated pCO2 induces caspase expression

The finding that eight Bcl-2 transcripts were up-regulat-

ed during 9-d CO2 stress prompted us to conduct a

more extensive survey of the key components of the

apoptotic machinery in Acropora and to examine their

expression levels in our experiments.

Amongst the approximately 20 transcriptome clusters

encoding caspase domains (PFAM domain Pepti-

dase_C14) present in the A. millepora transcriptome, two

were up-regulated during acute CO2 stress, and three

transcripts that most likely correspond to a single locus

were more highly up-regulated in the 9-d stress treat-

ment (Table 4).

The A. millepora caspase encoded by Cluster012253 is

up-regulated approximately 1.8-fold under acute expo-

sure to CO2; in addition to the caspase domain, both

this gene and its likely A. digitifera orthologue ADIG_-

Casp20 (aug_v2a.09611.t1) encode a CARD domain

(PF00619). The presence of a CARD domain suggests

that Cluster012253 encodes an initiator caspase (i.e. a

caspase which cleaves an effector caspase to activate it).

The second A. millepora caspase that was up-regulated

under acute CO2 stress (Cluster025634, 2.1-fold) is likely

to be orthologous with A. digitifera Casp13 (aug_-

v2a.01975.t1). Its function is uncertain as it contains

features of both initiator and effector caspases. The

A. millepora caspase that was highly up-regulated (~4.8-fold) under 9-d exposure to CO2 (Cluster003711p/Clus-

ter004256p/Cluster004868p; A. digitifera orthologue

ADIG-Casp3: aug_v2a.05413.t1) is most similar to mam-

malian effector caspases (Caspases 3, 6 and 7) rather

than initiator caspases (caspase 8 and 10) on the basis

of comparison of caspase domain sequences.

Although there was no evidence for differential

expression of the Acropora homologues of apoptotic pro-

tease-activating factor-1 and the mammalian inhibitors

of apoptosis, three CARD domain-only proteins were

up-regulated during acute CO2 stress (Table 4). The

significance of this finding is unclear; in mammals,

CARD-only proteins have diverse roles as regulators of

apoptosis.

Oxidative stress response proteins show short-termdownregulation but longer term upregulation

Previous work (Kaniewska et al. 2012) implied that an

oxidative stress response occurred in colonies of A. mille-

pora exposed to elevated pCO2, so the responses of key

genes potentially responding to oxidative stress were

investigated. In the experiments described here, none of

the superoxide dismutase or peroxiredoxin isoforms was

differentially expressed, and the expression of only a sin-

gle aldehyde dehydrogenase changed significantly (Clus-

ter015813 was up-regulated under acute exposure;

Table S2, Supporting information). However, both gluta-

thione peroxidase (GPX) and glutathione S-transferase

Table 3 Bcl-2 genes responsive to acute (3-d) and prolonged (9-d) exposure to elevated levels of CO2

Transcript ID Best blast hit e-Value BH domains

Fold-change versus control

Acute exposure Prolonged exposure

Cluster022467 NP_001279536.1

Bcl-2-related ovarian killer protein-like

protein [Callorhinchus milii]

6.00E�26 BH 1, 2, 3, 4 — 2.57

Cluster041039 — 2.64

Cluster016692 ABX61040.1

Bcl-like protein [Acropora millepora]

2.00E�134 BH 1, 2, 3, 4 — 1.96

Cluster025914 XP_002740789.2

Apoptosis regulator R1-like [Saccoglossus

kowalevskii]

2.00E�25 BH 1, 2, 3, 4 — 4.89

Cluster034949 — 5.26

Cluster024735 — 5.62

Cluster015074 XP_001628866.1

Predicted protein [Nematostella vectensis]

5.00E�87 BH 1, 2, 3 — 2.58

Cluster011480 EKC30554.1

Bcl-2-like protein 1 [Crassostrea gigas]

2.00E�26 BH 1, 2, 3, 4 — 2.18

Transcript clusters were consolidated on the basis of sequence similarity. The last two columns denote the fold-change in expression

compared to the control. BH, Bcl-2 homology.


444 A. MOYA ET AL.

isoforms showed differential expression during acute

CO2 stress. Five distinct GPX isoforms were identified in

A. millepora, two of which were down-regulated under

acute CO2 stress (Cluster018808 and Cluster011976,

�1.66 and �3.39-fold, respectively, Table S2, Supporting

information). Both Omega and Pi isoforms of GST

responded in the same way, suggesting coordinated

downregulation of at least some antioxidant defences

under acute CO2 stress, potentially facilitating a more

appropriate stress response, such as apoptosis.

Under 9-d exposure to CO2, the expression of both

the GPX and GST isoforms was at control levels, but

there was upregulation of both catalase (about twofold)

and both subunits of NADPH oxidase (1.9- and 2.2-fold,

respectively, Table S2, Supporting information) suggest-

ing induction of enzymatic antioxidant defences.

Two other multigene families, those encoding the GFP-

related (FPs) and universal stress proteins (USPs), were

over-represented amongst the genes differentially

expressed in the present analysis, but the functions of

these proteins remain unclear. A total of 12 distinct mem-

bers of the FP family were identified as encoded by the

A. millepora transcriptome (Table S3, Supporting infor-

mation). Comparison with data for A. digitifera (Shinzato

et al. 2012b) combined with BlastP analysis implies that

the 12 A. millepora FP transcripts encode two cyan FPs,

five green FPs, two chromoproteins and three red FPs.

With the exception of the two chromoproteins, all of the

FP genes identified were down-regulated specifically

under acute exposure to CO2 (on average 4.4-fold, see

Fig. 3 and Table S3, Supporting information).

The USPs were originally identified in the context of

bacterial stress response (Nystr€om & Neidhardt 1992;

reviewed in Kvint et al. 2003). In Hydra, those USP

genes for which expression data are available are

expressed in the endodermal epithelium, which acts as

a barrier for protection against intruding microbes,

suggesting potential roles of some USPs in this defen-

sive barrier (Bosch et al. 2009). Thirteen USP genes have

been identified in A. millepora (Foret et al. 2011), five of

which, including the calcification candidate Amil10,

were down-regulated in response to acute exposure to

CO2 by an average of 1.8-fold (Fig. 3 and Table S4, Sup-

porting information).

Differential expression of heat shock proteins iscommon to both acute and prolonged exposure toelevated pCO2

Although the responses to acute (3-d) and prolonged

(9-d) CO2 stress were largely discrete, HSPs and their

associated cochaperones (HSP40 and sacsin family

members) and transcription factors (the heat shock fac-

tors; HSFs) were over-represented amongst the rela-

tively few genes that comprised the common response

(Fig. 2B, Table 5 and Table S5, Supporting informa-

tion). A total of nine HSP-related transcripts (4 HSP20,

3 HSP70 and 2 HSP40 transcriptome clusters, corre-

sponding to 3 HSP20, 2 HSP70 and 1 HSP40 genes)

showed strong upregulation under both acute and 9-d

pCO2 stress (Table 5). All of the five A. millepora HSF

transcriptome clusters identified were likewise up-

Table 4 Caspases and CARD-only genes responsive to acute (3-d) and prolonged (9-d) exposure to elevated levels of CO2

Yellow background shading indicates caspases; white background indicates CARD-only proteins. Transcript clusters were consoli-

dated on the basis of sequence similarity. The last two columns denote the fold-change expression compared to the control.



Table 5 Heat shock protein genes responsive to acute (3-d) and prolonged (9-d) exposure to elevated levels of CO2

Transcript ID HMMER Best blast hit e-Value

Fold-change versus

control

Acute

exposure

Prolonged

exposure

HSP20 Cluster009259 HSP20 ABA42877.1

Small heat shock protein, partial [uncultured

cnidarian]

3.00E�155 6.81 3.66

Cluster017128 7.36 5.77

Cluster022217 HSP20 9 2 ABA42878.1

Small heat shock protein [uncultured cnidarian]

4.00E�178 8.07 7.88

Cluster020497 HSP20 9 2 ABA42878.1

Small heat shock protein [uncultured cnidarian]

2.00E�81 9.50 3.97

HSP40 Cluster010480 DnaJ

CTDII

XP_005112000.1

dnaJ homologue subfamily B member 1-like

[Aplysia californica]

5.00E�112 1.78 —

Cluster015487 DnaJ ABC84495.1

heat shock protein 40 [Locusta migratoria]

9.00E�111 8.87 9.00

Cluster018170 CTDII 8.66 9.02

Cluster008833 DnaJ

CTDII

XP_005296130.1

dnaJ homologue subfamily B member 1

[Chrysemys picta bellii]

2.00E�95 — 3.63

Cluster008505 DnaJ XP_007537836.1

PREDICTED: dnaJ homologue subfamily C

member 17 [Erinaceus europaeus]

3.00E�11 2.00 —

Cluster009800 DnaJ

zf-C2H2_jaz

EKC33886.1

DnaJ-like protein subfamily C member 21

[Crassostrea gigas]

3.00E�117 — 4.98

HSP70 Cluster007731 HSP70 XP_002731913.1

heat shock cognate 71 kDa protein-like

[Saccoglossus kowalevskii]

3.00E�137 5.96 7.19

Cluster005163 1.82 2.06

Cluster005613 HSPA4_like_NDB BAD89541.1

heat shock protein 70 [Pocillopora damicornis]

0.00E+00 4.48 2.72

Cluster006034 HSP70 XP_006823582.1

heat shock 70 kDa protein 12A-like [Saccoglossus

kowalevskii]

8.00E�179 3.73 —

HSP90 Cluster010802 HATPase_c_3 NP_999808.1

heat shock protein gp96 precursor

[Strongylocentrotus purpuratus]

0.00E+00 2.19 —Cluster005735 HSP90 1.60 —

Cluster022600 HSP90 2.60 —

Sacsin Cluster005249 HEPN XP_002740585.1

sacsin-like [Saccoglossus kowalevskii]

0.00E+00 1.83 —

Cluster010514 3.22 —Cluster017656 5.94 —

Cluster003566 HEPN 0.00E+00 3.27 —Cluster007958 3.65 —


446 A. MOYA ET AL.

regulated under both acute and prolonged exposure to

elevated CO2 (Table 5 and Table S5, Supporting infor-

mation).

In addition to the common response, other HSPs

responded to either 3- or 9-d elevated CO2. Six HSP

transcripts (2 HSP40, 1 HSP70 and 3 HSP90,

corresponding to 2 HSP40, 1HSP70 and the single

HSP90B1/gp96 gene) were specifically up-regulated

under 3-d CO2 stress, and 2 HSP40 (2 genes) were

up-regulated only after 9-d treatment (see Table 5). A

large proportion (14) of the 22 sacsin transcriptome

clusters (8 of the 18 genes) identified were up-regulated

specifically, and a single transcript (1 gene) was

down-regulated under acute CO2 stress.

Discussion

The observed extensive differences in gene expression

patterns between the ‘acute’ (3-d) and ‘prolonged’ (9-d)

exposure treatments were unexpected and surprising

given that both reflect relatively short-term responses.

Whereas acute CO2 stress suppressed metabolism and

altered the expression of many genes encoding CA and

SOM proteins (Moya et al. 2012), expression of most

genes in those categories was at control levels in the

prolonged treatment experiment. Ogawa et al. (2013)

reported that two CAs in adult Acropora aspera were

up-regulated at 6 d, but down-regulated at 14 d. As the

length of the present experiment was intermediate

Table 5 Continued

Transcript ID HMMER Best blast hit e-Value

Fold-change versus

control

Acute

exposure

Prolonged

exposure

XP_008106211.1

PREDICTED: sacsin isoform X3 [Anolis

carolinensis]

Cluster005325 3.62 —

Cluster001033 — XP_004631088.1

PREDICTED: sacsin isoform X1 [Octodon degus]

0.00E+00 2.61 —

Cluster001972 HEPN XP_003200093.2

sacsin-like isoform X1 [Danio rerio]

0.00E+00 2.24 —

Cluster009368 HEPN XP_005516322.1

sacsin [Pseudopodoces humilis]

9.00E�66 �3.71 —

Cluster004016 HEPN EOA99087.1

sacsin, partial [Anas platyrhynchos]

0.00E+00 2.08 —Cluster010311 1.85 —

Cluster009514 2.08 —Cluster013781 3.74 —

Cluster013781 HEPN XP_007234568.1

sacsin isoform X2 [Astyanax mexicanus]

0.00E+00 3.74 —

Cluster007316 HEPN XP_006927228.1

sacsin [Felis catus]

0.00E+00 2.15 —

Cluster000374 HEPN XP_007234567.1

sacsin isoform X1 [Astyanax mexicanus]

0.00E+00 — —

Cluster003923 1.87 —Cluster002694 — —

HSF Cluster007375 HSF-type

DNA-binding

ABR15461.1

HSF [Haliotis asinina]

3.00E�65 3.07 4.45

Cluster007882 3.19 4.20

Cluster007957 3.06 4.27

Cluster009578 2.95 4.13

Cluster003398 3.12 7.58

Transcript clusters were consolidated on the basis of sequence similarity. The last two columns denote the fold-change expression

compared to the control. HSPs, heat shock proteins; HSFs, heat shock factors.



between those two time points, there is no basis for

commenting on the consistency/inconsistency of the

results of the two experiments.

Upregulation of multiple Bcl-2 family members was

the single most obvious feature of the 9-d exposure

treatment group. Five distinct Bcl-2 proteins were

up-regulated in the 9-d treatment group, whereas no

proteins of this type were differentially expressed in the

acute treatment group. Based on comparisons with the

well-characterized mammalian proteins, four of the five

Bcl-2 proteins are likely to be anti-apoptotic, indicating

that suppression of apoptosis is a key component of

acclimation to elevated pCO2. Conversely, two caspase

genes were up-regulated after acute exposure to ele-

vated pCO2, suggesting that apoptosis may be activated

in this treatment group. However, in apparent contra-

diction to the above interpretation of the Bcl-2 data, a

distinct caspase gene showed significant upregulation

specifically in the prolonged treatment group. Interpret-

ing the significance of the caspase data is complicated

by the ambiguous relationships of the cnidarian pro-

teins with the well-defined functional groups of verte-

brate caspases. In the case of the Bcl-2 family, several of

the coral proteins showed convincing sequence similar-

ity with specific classes from bilaterians, but the coral

caspase repertoire is far more difficult to interpret. For

example, the Acropora protein Cluster002669p has cas-

pase 8-like specificity despite having caspase 3-like resi-

dues at what are otherwise diagnostic positions

(Sakamaki et al. 2014). Although phylogenetic analyses

were equivocal, the presence of a CARD domain in one

of the caspases (Cluster012253) up-regulated during

acute CO2 stress suggests that it is likely to be an initia-

tor caspase. The caspase induced under 9-d CO2 stress

has clear counterparts in other anthozoans, and despite

equivocal relationships with the vertebrate CASP clas-

ses, these are likely to be effectors rather than initiators.

Whilst the caspase data are difficult to interpret at

this stage, the elevated expression of a suite of anti-

apoptotic Bcl-2 proteins implies suppression of apopto-

sis in the 9-d treatment group. Based on four coral spe-

cies, Tchernov et al. (2011) presented evidence of the

ability to survive moderate bleaching when the caspase-

mediated apoptotic cascade is down-regulated. Under

prolonged elevated CO2, the activation of anti-apoptotic

molecules could serve as an important mechanism

moderating the apoptotic response, as suggested by

Kvitt et al. (2011).

Fluorescent proteins (FPs) show great diversity in cor-

als (Alieva et al. 2008; Shinzato et al. 2012b) and can be

amongst the most abundant proteins (Oswald et al.

2007); however, our understanding of the role of this pro-

tein family is far from complete. One proposal is that FPs

function as a component of the oxidative stress response

through superoxide quenching (Bou-Abdallah et al. 2006)

or hydrogen peroxide scavenging (Palmer et al. 2009).

Previous work has shown that some FPs are strongly

down-regulated by heat stress (Dove et al. 2006; Smith-

Keune & Dove 2008; Rodriguez-Lanetty et al. 2009). A

similar pattern is shown in this study under acute expo-

sure to pCO2, supporting downregulation of antioxidant

defences in the short-term; however, functional studies

of individual FPs are needed to clarify their roles.

Although responses to the acute and prolonged treat-

ments were in large part mutually exclusive, HSPs and

related cochaperones and transcription factors were

over-represented amongst the small number of genes

constituting the common response (Table 5). Beyond

the common response, other members of the HSP reper-

toire responded specifically to either acute or prolonged

CO2 stress (Table 5). Nakamura et al. (2012) for Acropora

digitifera and Putnam et al. (2013) for Pocillopora damicor-

nis report no differential response by members of the

HSP70 family to acidification or acidification plus heat-

ing, respectively (note though that these authors stud-

ied effects only in presettlement stages), but we cannot

determine whether those responses are consistent with

those reported here for Acropora millepora. In the case of

A. digitifera, the sequences are not available, whilst in

the case of Pocillopora, it is unclear to which of four

A. millepora isoforms (Cluster005613, Cluster005163,

Cluster011524, Cluster007731) the HSP70 from P. dami-

cornis studied in Putnam et al. (2013; BAD89541.1) corre-

sponds (all had >70% identity and e-value = 0). The

first two of these A. millepora transcripts (Cluster005613,

Cluster005163) were up-regulated in both acute and

prolonged treatments, whereas the others were not dif-

ferentially regulated in either case.

There are a number of differences between the behav-

iour of individual genes in the study of Kaniewska et al.

(2012) compared to the present study. For example, the

downregulation of two proteins that are likely to have

anti-apoptotic activity (Bcl-2 and API-5) reported by

Kaniewska et al. (2012) contrasts with the upregulation

of a suite of anti-apoptotic Bcl-2 proteins reported here.

However, there are many differences in the design of

the two studies. Perhaps the most significant of these is

that Kaniewska et al. (2012) were studying adult

A. millepora containing symbionts, which greatly com-

plicates both the analysis and its interpretation, while

the present study was of early developmental stages

lacking symbionts. Only further study will clarify the

reason for such differences.

Whereas few studies have addressed gene expres-

sion changes in corals under elevated pCO2, thermal

stress-induced changes have received much greater

attention. The categories of genes responding to

thermal stress include HSPs and members of the apop-


448 A. MOYA ET AL.

totic repertoire, and as a result, models proposed to

account for the observed changes in gene expression in

response to acidification and heating are similar.

Indeed, Polato et al. (2013) state ‘In larval and adult

acroporids, expression of heat shock proteins is typi-

cally characterized by short-term upregulation followed

by a decline in expression after prolonged exposure to

high temperatures’. However, the superficial similarity

in molecular responses may be an artefact of not tak-

ing into account the complexity of the stress response/

apoptotic repertoire—for example, at least twelve dis-

tinct HSP70 isozymes and over 20 HSP40/DNAJ pro-

teins are present in A. millepora. Here we demonstrate

that, although a core set of HSPs responds to both

short and longer term exposure to elevated pCO2, dis-

tinct HSPs respond specifically to either duration of

exposure.

Heat shock proteins have been extensively explored as

potential stress markers in corals, but with limited con-

sideration of their functional diversity (Morris et al.

2013). A preliminary survey of the HSP and stress

response repertoire of Acropora has been published

(Shinzato et al. 2012a), but the available annotation is lim-

ited. The available data can best be interpreted in the

context of conservation of function more broadly, at the

class level. Typically, the HSP70 and HSP90 classes each

consist of constitutive and inducible variants, the former

dealing with the instability of many cellular proteins

even under normal physiological conditions. On the

other hand, the HSP40/DNAJ and other small HSP pro-

teins—sometimes referred to as ‘holdases’, whereas

HSP70/HSP90 are ‘foldases’ (Richter et al. 2010)—are the

first line of the chaperone response to proteins in non-

native conformations and are typically stress induced.

The cochaperone ‘holdases’ stimulate the ATPase activity

of the ‘foldases’ (HSP70, HSP90), stabilizing the ‘fol-

dase’/substrate interaction (Richter et al. 2010).

A substantial proportion of the small HSP isoforms

identified in A. millepora were up-regulated under acute

CO2 stress, including 15 of the 22 sacsin transcripts

(Table 5). However, most of the small HSPs up-regulated

under acute CO2 stress were at control levels in the 9-d

treatment group, although some (all four differentially

expressed HSP20 transcripts; two HSP40 transcripts)

remained at higher than control levels. The ‘foldase’

activity elevated in both the 3- and 9-d treatment groups

was limited to only three of the seventeen HSP70 tran-

scripts identified, the ‘holdase’ component consisting of

four HSP20 and two HSP40 types (Table 5).

The consensus view is that the chaperone-based

stress response is largely generic—because the HSPs

respond to the presence of unfolded protein, rather than

to specific stress sensors, the expectation is that the

same foldases should respond to CO2 stress as to any

other type of stress. However, this expectation is not

consistently fulfilled when our results are compared to

previous studies. For example, Leggat et al. (2011)

reported strong upregulation of HSP90 and HSP70 in

heat-stressed A. aspera, but the A. millepora orthologues

of these isoforms (the HSP90A1 protein DY584045.1—

Cluster009669 and GO000475.1—Cluster010972, respec-

tively) were not differentially regulated in our experi-

ment. The HSP70 variant up-regulated in heat-stressed

A. millepora larvae (Rodriguez-Lanetty et al. 2009) and

in immune challenged adults of the same species

(Brown et al. 2013) corresponds to the A. millepora

Cluster006117 (a likely gp78 orthologue), which was not

differentially expressed in response to CO2. DeSalvo

et al. (2008) reported upregulation of an HSP90 variant

in Montastraea faveolata during heat stress; whilst this

sequence is most similar to A. millepora HSP90A1, the

EST is too short to be unequivocally identified.

Although the data discussed above do not match the

expectations of a common response to all stressors,

other results, including those from the most comprehen-

sive field-based study on coral thermal tolerance con-

ducted to date (Barshis et al. 2013), are consistent with a

generic foldase response. The A. millepora HSP90B1/

gp96 variant, which is up-regulated under acute CO2

stress (Table 5), has also been shown to be up-regulated

in an acute thermal stress experiment on larvae of

A. millepora (Rodriguez-Lanetty et al. 2009). A clear

gp96 orthologue (CAC38753.1) was likewise up-regulated

in the octocoral Dendronephthya klunzingeri subjected to

heat stress (Wiens et al. 2000). Analyses of field popula-

tions of Acropora hyacinthus revealed that orthologues of

the A. millepora HSF (contig145533 and contig209263)

and one of the HSP70 types (contig187488 in A. hyacin-

thus, corresponding to Cluster 007731 and Cluster

005163 in A. millepora) were consistently more highly

up-regulated in more thermotolerant than in less ther-

motolerant colonies (Barshis et al. 2013). As it is also a

component of the common CO2 stress response of

A. millepora, this HSP70 variant, rather than others, is a

good candidate general stress response marker for cor-

als. The second HSP70 component of the common CO2

stress response in A. millepora (Cluster005613) was also

up-regulated in heat-stressed Acropora palmata larvae

(Contig_13281 and Contig_23857 in the microarray

analysis of Polato et al. 2013), which is again consistent

with a generic foldase stress response.

Taken together, these results imply that, despite ini-

tial metabolic disruption, A. millepora juveniles have

remarkable capacity for acclimation to elevated pCO2

with a short response time. Although the initial

response to elevated pCO2 involves suppression of

metabolism and disrupted expression of genes

involved in skeleton deposition, after 9-d exposure,



these categories of genes had returned to control lev-

els. Acclimation to pCO2 involves up-regulated expres-

sion of a suite of anti-apoptotic Bcl-2 and chaperone

genes. In this context, the inducible HSP70 variant

(Cluster187488) may be particularly significant, because

it is not only a core component of the CO2 stress

response repertoire, but also is implicated in acclima-

tion to thermal stress in field populations of A. hyacin-

thus (Palumbi et al. 2014).

It remains uncertain whether coral populations can

adapt or shift in distribution on timescales that will

allow them to survive projected rates of climate change

(Hoegh-Guldberg et al. 2007; Pandolfi et al. 2011; Bars-

his et al. 2013; Palumbi et al. 2014). The short timescale

for acclimation to pCO2 observed in this study is there-

fore particularly significant. Clearly, field populations

can also rapidly acclimate to thermal stress (Palumbi

et al. 2014), suggesting that climate change impacts on

coral reefs may not be as fast as projected by some

authors (Hoegh-Guldberg et al. 2007; Dove et al. 2013).

These findings provide a glimmer of hope given the

ecological significance of species of the genus Acropora

throughout the Indo-Pacific. However, a number of

caveats apply to such extrapolations (Moya et al. 2012).

First, although these experiments imply that rapid accli-

mation to single stressors may occur, the combination

of changes in several stressors, such as OA and warm-

ing, could lead to larger impacts and less potential for

acclimation. Second, a healthy coral colony comprises

not only the animal host, but also one or more Symbi-

odinium strains and a complex bacterial community,

any one of which may be the weakest link in determin-

ing the survival of the colony. Third, Doropoulos et al.

(2012) have shown that crustose coralline algae, which

harbour bacterial epiphytes providing cues for coral

settlement, are very sensitive to OA. A decline in their

abundance, along with the decline in abundance of

their associated bacteria, could greatly reduce coral set-

tlement. Finally, coral communities will face not only

gradual changes in environmental parameters, but also

extreme thermal anomalies (Depczynski et al. 2013). It

is unclear whether acclimation alone can enable coral

populations to survive more frequent extreme events.

We are in the midst of an ongoing experiment on a

complex system that exhibits some built-in resilience

and adaptability. Whether this will be adequate for the

survival of coral reefs or whether they will disappear

within this century, as some have predicted, remains to

be seen.

Acknowledgments

This research was supported by the Australian Research Council

through Discovery Grant DP1095343 to D.J.M., E.E.B. and S.F.,

and via the Centre of Excellence for Coral Reef Studies, and by a

Marie Curie International Outgoing Fellowship (grant agree-

ment # PIOF-GA-2008-235142 project title AMICAL) to A.M.

This work is a contribution to the ‘European Project on Ocean

Acidification’ (EPOCA), which received funding from the

European Community’s Seventh Framework Program (FP7/

2007–2013) under grant agreement # 211384. The authors thank

Andrew Baird and William Leggat for the use of aquaculture

facilities, Gergely Torda, Chia-Miin Chua and Peter Cross for

their help with field trips and experiments, and Philip Munday

for developing the CO2-mixing system. The authors are also

grateful to Andrew Negri, Eneour Puill-Stephan, the SEASIMS

aquaculture facilities at AIMS for providing larvae for conduct-

ing pilot experiments, and Kazuhiro Sakamaki (Kyoto Univer-

sity) for commenting on the manuscript.

References

Albright R (2011) Reviewing the effects of ocean acidification

on sexual reproduction and early life history stages of reef-

building corals. Journal of Marine Biology, 2011, 1–14.Albright R, Langdon C (2011) Ocean acidification impacts mul-

tiple early life history processes of the Caribbean coral Porites

astreoides. Global Change Biology, 17, 2478–2487.Albright R, Mason B, Miller M, Langdon C (2010) Ocean acidi-

fication compromises recruitment success of the threatened

Caribbean coral Acropora palmata. Proceedings of the National

Academy of Sciences, USA, 107, 20400–20404.Alieva NO, Konzen KA, Field SF et al. (2008) Diversity and

evolution of coral fluorescent proteins. PLoS One, 3, e2680.

Ashburner M, Ball CA, Blake JA et al. (2000) Gene Ontology:

tool for the unification of biology. Nature Genetics, 25, 25–29.Barshis DJ, Ladner JT, Oliver TA, Seneca FO, Traylor-Knowles

N, Palumbi SR (2013) Genomic basis for coral resilience to

climate change. Proceedings of the National Academy of Sciences,

110, 1387–1392.Bosch TC, Augustin R, Anton-Erxleben F et al. (2009) Uncover-

ing the evolutionary history of innate immunity: the simple

metazoan Hydra uses epithelial cells for host defence. Devel-

opmental and Comparative Immunology, 33, 559–569.Bou-Abdallah F, Chasteen ND, Lesser MP (2006) Quenching of

superoxide radicals by green fluorescent protein. Biochimica

et Biophysica Acta (BBA)-General Subjects, 1760, 1690–1695.Brown T, Bourne D, Rodriguez-Lanetty M (2013) Transcrip-

tional activation of C3 and HSP70 as part of the immune

response of Acropora millepora to bacterial challenges. PLoS

One, 8, e67246.

Chomczynski P, Sacchi N (1987) Single-step method of RNA

isolation by acid guanidinium thiocyanate–phenol–chloro-form extraction. Analytical Biochemistry, 162, 156–159.

Chua CM, Leggat W, Moya A, Baird AH (2013a) Temperature

affects the early life history stages of corals more than near

future ocean acidification. Marine Ecology Progress Series, 475,

85–92.Chua CM, Leggat W, Moya A, Baird AH (2013b) Near-future

reductions in pH will have no consistent ecological effects

on the early life-history stages of reef corals. Marine Ecology

Progress Series, 486, 143–151.Collins M, Knutti R, Arblaster J et al. (2013) Long-term climate

change: projections, commitments and irreversibility. In: Cli-

mate Change 2013: The Physical Science Basis. Contribution of


450 A. MOYA ET AL.

Working Group I to the Fifth Assessment Report of the Intergovern-

mental Panel on Climate Change (eds Stocker TF, Qin D, Plattner

G-K, Tignor M, Allen SK, Boschung J, Nauels A, Xia Y, Bex V,

Midgley PM), pp. 1029–1136. Cambridge University Press,

Cambridge.

Cumbo VR, Fan TY, Edmunds PJ (2013) Effects of exposure

duration on the response of Pocillopora damicornis larvae to

elevated temperature and high pCO2. Journal of Experimental

Marine Biology and Ecology, 439, 100–107.Depczynski M, Gilmour JP, Ridgway T et al. (2013) Bleaching,

coral mortality and subsequent survivorship on a West Aus-

tralian fringing reef. Coral Reefs, 32, 233–238.DeSalvo MK, Voolstra CR, Sunagawa S et al. (2008) Differential

gene expression during thermal stress and bleaching in the

Caribbean coral Montastraea faveolata. Molecular Ecology, 17,

3952–3971.Doney SC, Fabry VJ, Feely RA, Kleypas JA (2009) Ocean acidi-

fication: the other CO2 problem. Annual Review of Marine Sci-

ence, 1, 169–192.Doropoulos C, Ward S, Diaz-Pulido G, Hoegh-Guldberg O,

Mumby PJ (2012) Ocean acidification reduces coral recruit-

ment by disrupting intimate larval-algal settlement interac-

tions. Ecology Letters, 15, 338–346.Dove S, Ortiz JC, Enriquez S et al. (2006) Response of holosym-

biont pigments from the scleractinian coral Montipora monas-

teriata to short-term heat stress. Limnology and Oceanography,

51, 1149–1158.Dove SG, Kline DI, Pantos O, Angly FE, Tyson GW, Hoegh-

Guldberg O (2013) Future reef decalcification under a busi-

ness-as-usual CO2 emission scenario. Proceedings of the

National Academy of Sciences, USA, 110, 15342–15347.Foret S, Seneca F, De Jong D et al. (2011) Phylogenomics

reveals an anomalous distribution of USP genes in metazo-

ans. Molecular Biology and Evolution, 28, 153–161.Gattuso J-P, Allemand D, Frankignoulle M (1999) Photosynthe-

sis and calcification at cellular, organismal and community

levels in coral reefs: a review on interactions and control by

carbonate chemistry. American Zoology, 39, 160–183.Grasso LC, Maindonald J, Rudd S et al. (2008) Microarray

analysis identifies candidate genes for key roles in coral

development. BMC Genomics, 9, 540.

Hardwick JM, Soane L (2013) Multiple functions of BCL-2 family

proteins. Cold Spring Harbor Perspectives in Biology, 5, 1–24.Hoegh-Guldberg O, Mumby PJ, Hooten AJ et al. (2007) Coral

reefs under rapid climate change and ocean acidification. Sci-

ence, 318, 1737–1742.Kaniewska P, Campbell PR, Kline DI et al. (2012) Major cellular

and physiological impacts of ocean acidification on a reef

building coral. PLoS One, 7, e34659.

Kenkel CD, Goodbody-Gringley G, Caillaud D, Davies SW,

Bartels E, Matz MV (2013) Evidence for a host role in

thermotolerance divergence between populations of the mus-

tard hill coral (Porites astreoides) from different reef environ-

ments. Molecular Ecology, 22, 4335–4348.Kvint K, Nachin L, Diez A, Nystr€om T (2003) The bacterial

universal stress protein: function and regulation. Current

Opinion in Microbiology, 6, 140–145.Kvitt H, Rosenfeld H, Zandbank K, Tchernov D (2011) Regula-

tion of apoptotic pathways by Stylophora pistillata (Anthozoa,

Pocilloporidae) to survive thermal stress and bleaching. PLoS

One, 6, e28665.

Langmead B, Trapnell C, Pop M, Salzberg SL (2009) Ultrafast

and memory-efficient alignment of short DNA sequences to

the human genome. Genome Biology, 10, R25.

Leggat W, Seneca F, Wasmund K, Ukani L, Yellowlees D, Ains-

worth TD (2011) Differential responses of the coral host and

their algal symbiont to thermal stress. PLoS One, 6, e26687.

Lewis E, Wallace DWR (1998) Program Developed for CO2 Sys-

tem Calculations. ORNL/CDIAC-105. Carbon Dioxide Infor-

mation Analysis Center, Oak Ridge National Laboratory U.S.

Department of Energy, Oak Ridge, Tennessee.

Morris JP, Thatje S, Hauton C (2013) The use of stress-70 pro-

teins in physiology: a re-appraisal. Molecular Ecology, 22,

1494–1502.Moya A, Huisman L, Ball EE et al. (2012) Whole transcriptome

analysis of the coral Acropora millepora reveals complex

responses to CO2-driven acidification during the initiation of

calcification. Molecular Ecology, 21, 2440–2454.Munday PL, Dixson DL, Donelson JM et al. (2009) Ocean acidi-

fication impairs olfactory discrimination and homing ability

of a marine fish. Proceedings of the National Academy of Sci-

ences, 106, 1848–1852.Nakamura M, Morita M, Kurihara H, Mitarai S (2012) Expres-

sion of HSP70, HSP90 and HSF1 in the reef coral Acropora

digitifera under prospective acidified conditions over the next

several decades. Biology Open, 1, 75–81.Nystr€om T, Neidhardt FC (1992) Cloning, mapping and

nucleotide sequencing of a gene encoding a universal

stress protein in Eschericha coli. Molecular Microbiology, 6,

3187–3198.Ogawa D, Bobeszko T, Ainsworth T, Leggat W (2013) The

combined effects of temperature and CO2 lead to altered

gene expression in Acropora aspera. Coral Reefs, 32, 895–907.Orr JC (2011) Recent and future changes in ocean carbonate

chemistry. In: Ocean Acidification (eds Gattuso J-P, Hansson

L), pp. 41–66. Oxford University Press, Oxford.

Oswald F, Schmitt F, Leutenegger A et al. (2007) Contributions

of host and symbiont pigments to the coloration of reef cor-

als. FEBS Journal, 274, 1102–1122.Palmer CV, Modi CK, Mydlarz LD (2009) Coral fluorescent

proteins as antioxidants. PLoS One, 4, e7298.

Palumbi SR, Barshis DJ, Traylor-Knowles N, Bay RA (2014)

Mechanisms of reef coral resistance to future climate change.

Science, 344, 895–898.Pandolfi JM, Jackson JBC, Baron N et al. (2005) Are US Coral

reefs on the slippery slope to slime? Science, 307, 1725–1726.Pandolfi JM, Connolly SR, Marshall DJ, Cohen AL (2011) Pro-

jecting coral reef futures under global warming and ocean

acidification. Science, 333, 418–422.Polato NR, Altman NS, Baums IB (2013) Variation in the tran-

scriptional response of threatened coral larvae to elevated

temperatures. Molecular Ecology, 22, 1366–1382.Putnam HM, Mayfield AB, Fan TY, Chen CS, Gates RD (2013)

The physiological and molecular responses of larvae from

the reef-building coral Pocillopora damicornis exposed to near-

future increases in temperature and pCO2. Marine Biology,

160, 2157–2173.Richter K, Haslbeck M, Buchner J (2010) The heat shock

response: life on the verge of death.Molecular Cell, 40, 253–266.Robinson MD, McCarthy DJ, Smyth GK (2010) edgeR: a Bio-

conductor package for differential expression analysis of dig-

ital gene expression data. Bioinformatics, 26, 139–140.



Rodriguez-Lanetty M, Harii S, Hoegh-Guldberg O (2009) Early

molecular responses of coral larvae to hyperthermal stress.

Molecular Ecology, 18, 5101–5114.Roemmich D, Gould WJ, Gilson J (2012) 135 years of global

ocean warming between the Challenger expedition and the

Argo Programme. Nature Climate Change, 2, 425–428.Sakamaki K, Shimizu K, Iwata H et al. (2014) The apoptotic ini-

tiator caspase-8: its functional ubiquity and genetic diversity

during animal evolution. Molecular Biology and Evolution, 31,

3282–3301.Sambrook J, Russell DW (2001) Molecular Cloning: A Laboratory

Manual. Cold Spring Harbor Laboratory, Cold Spring Har-

bor, New York.

Shinzato C, Shoguchi E, Kawashima T et al. (2011) Using the

Acropora digitifera genome to understand coral responses to

environmental change. Nature, 476, 320–323.Shinzato C, Hamada M, Shoguchi E, Kawashima T, Satoh N

(2012a) The repertoire of chemical defense genes in the

coral Acropora digitifera genome. Zoological Science, 29, 510–517.

Shinzato C, Shoguchi E, Tanaka M, Satoh N (2012b) Fluores-

cent protein candidate genes in the coral Acropora digitifera

genome. Zoological Science, 29, 260–264.Smith-Keune C, Dove S (2008) Gene expression of a green

fluorescent protein homolog as a host-specific biomarker of

heat stress within a reef-building coral. Marine Biotechnol-

ogy, 10, 166–180.Tchernov D, Kvitt H, Haramaty L et al. (2011) Apoptosis and

the selective survival of host animals following thermal

bleaching in zooxanthellate corals. Proceedings of the National

Academy of Sciences, 108, 9905–9909.Wiens M, Ammar MS, Nawar AH et al. (2000) Induction of

heat-shock (stress) protein gene expression by selected nat-

ural and anthropogenic disturbances in the octocoral Den-

dronephthya klunzingeri. Journal of Experimental Marine

Biology and Ecology, 245, 265–276.Young M, Wakefield M, Smyth G, Oshlack A (2010) Gene

ontology analysis for RNA-seq: accounting for selection bias.

Genome Biology, 11, R14.

A.M., L.H., S.F., E.E.B. and D.J.M. analysed the data.

A.M. and L.H. performed the experiments. A.M., L.H.,

S.F., J.P.G. and D.J.M. conceived and designed the

experiments. A.M., S.F., J.P.G., D.C.H., E.E.B. and

D.J.M. contributed to reagents, materials and analysis

tools, A.M., L.H., S.F., J.P.G., D.C.H., E.E.B. and D.J.M.

wrote the study and designed figures.

Data accessibility

The RNA-seq reads used in this study have been

submitted to the NCBI Gene Expression Omnibus

(GEO) database under Accession no. GSE61114.

Sequences of all genes mentioned in this study can

be found in the transcriptome of Acropora millepora

(Moya et al. 2012).

Supporting information

Additional supporting information may be found in the online ver-

sion of this article.

Fig. S1. Level of agreement between biological replicates.

Table S1. Over-representation of Gene Ontology (GO) terms in

the set of differentially expressed genes responding to acute (3-

d) and prolonged (9-d) exposure to elevated levels of CO2.

Table S2. The response of oxidative stress proteins to acute (3-

d) and prolonged (9-d) exposure to elevated levels of CO2.

Table S3. The response of fluorescent proteins (FPs) to acute

(3-d) and prolonged (9-d) exposure to elevated levels of CO2.

Table S4. The response of universal stress proteins (USP) to

acute (3-d) and prolonged (9-d) exposure to elevated levels of

CO2.

Table S5. The response of heat shock proteins (HSP) and heat

shock factors (HSF) to acute (3-d) and prolonged (9-d) expo-

sure to elevated levels of CO2.


452 A. MOYA ET AL.

rapid acclimation of juvenile corals to co2‐mediated...

Documents