Plant growth and development are strongly affected by small differences in temperature
1
. Current climate change has already altered global plant phenology and distribution
2, 3
, and projected increases in temperature pose a significant challenge to agriculture
4
. Despite the important role of temperature on plant development, the underlying pathways
are unknown. It has previously been shown that thermal acceleration of flowering is
dependent on the florigen, FLOWERING LOCUS T (FT)
5, 6
. How this occurs is however not understood, since the major pathway known to upregulate
FT, the photoperiod pathway, is not required for thermal acceleration of flowering
6
. Here we demonstrate a direct mechanism by which increasing temperature causes the
bHLH transcription factor PHYTOCHROME INTERACTING FACTOR4 (PIF4) to activate FT. Our
findings provide a new understanding of how plants control their timing of reproduction
in response to temperature. Flowering time is an important trait in crops as well
as affecting the lifecycles of pollinator species. A molecular understanding of how
temperature affects flowering will be important for mitigating the effects of climate
change.
Arabidopsis thaliana, like many higher plants, responds to warmer ambient temperatures
by increasing its growth rate and accelerating the floral transition
1, 5, 7
. Arabidopsis is a facultative long day plant, and plants grown under short photoperiods
are dramatically delayed in flowering. Interestingly, late flowering in short days
can be overcome by growth at higher temperatures
6
. The underlying mechanism is however unknown. The flowering response to temperature
is dependent on the floral pathway integrator gene FT
6
indicative of a thermosensory pathway that upregulates FT expression independently
of daylength. Since the bHLH transcription factor PHYTOCHROME INTERACTING FACTOR4
(PIF4) has been shown to regulate architectural responses to high temperature
8, 9
, we tested if PIF4 is required for the induction of flowering at high temperature
in short photoperiods. While pif4-101 is slightly delayed in flowering at 22 °C, pif4-101
mutants show a striking loss of thermal induction of flowering at 27 °C (Fig. 1a and
b). To test if pif4-101 perturbed floral induction by affecting FT expression, we
examined the thermal induction of FT in Col-0 and pif4-101. While FT expression is
strongly thermally inducible in Col-0, this response is largely abolished in pif4-101
at 27 °C (Fig. 1c), indicating that PIF4 is necessary for the thermal acceleration
of flowering in short days. By contrast, PIF4 is not required for the thermosensory
induction of flowering under continuous light
8
, suggesting that the photoperiod pathway also interacts with the ambient temperature
sensing pathway. The reduced role of PIF4 under continuous light likely reflects the
instability of PIF4 in light
10
coupled with the fact that the output of the photoperiod pathway, CONSTANS (CO) protein,
is stabilised by light
11
, shifting the balance of floral induction from PIF4 to the photoperiod pathway. Since
PIF4 is necessary for the thermal induction of flowering in short days, we tested
if it is sufficient to trigger flowering when overexpressed. 35S::PIF4 causes extremely
early flowering (Fig. 1d and 1e), similarly to the effect of overexpressing a related
gene, PHYTOCHROME INTERACTING FACTOR5
12
, suggesting that PIF4 is limiting for the acceleration of flowering at lower temperature
in short photoperiods. Consistently, 35S::PIF4 plants show elevated levels of FT (Fig.
1f). Furthermore, 35S::PIF4 ft-3 shows a complete suppression of the early flowering
phenotype, showing that the induction of flowering by 35S::PIF4 is dependent on FT
(Fig. 1g and h). This activation of FT appears to be independent of the established
photoperiod pathway since CO does not change in response to 35S::PIF4 (Fig. 1f). Finally,
while co-9 mutants are late flowering
6, 13
, we find 35S::PIF4 co-9 plants are early flowering, indicating that PIF4 acts largely
independently of CO (supplementary information Fig. S1), consistent with the thermal
induction of flowering being independent of the photoperiod pathway (Fig. 1f)
6
.
Although PIF4 has been shown to be important for high temperature responses, long-term
increases in either PIF4 transcript or PIF4 protein levels in response to higher ambient
temperature that can account for the observed growth responses have not been detected
8, 9
. To examine if variation of PIF4 transcription under our experimental conditions
might account for the increases in PIF4 activity with temperature, we measured PIF4
transcript levels at 12, 17, 22 and 27 °C in seedlings (Fig. 2a). PIF4 transcript
levels increase from 12 °C to 22 °C, while the difference between 22 °C and 27 °C
is not statistically significant. Plants at 27 °C, compared to 22 °C, show a very
large PIF4-dependent response, suggesting that variation in the PIF4 transcript is
not sufficient to account for the acceleration of flowering at 27 °C compared to 22
°C. To test whether temperature-mediated changes in PIF4 transcription are rate-limiting
for the biological response, we analysed the behaviour of plants constitutively expressing
PIF4. While 35S::PIF4 plants at 22 °C are extremely early flowering, this phenotype
can be largely suppressed at 12 °C (Fig. 2b and Fig. S2), indicating that even when
PIF4 transcript is abundant, lower temperatures are inhibitory for PIF4 activity.
A possible explanation for this difference is that PIF4 protein is destabilised by
low temperature. Indeed, PIF4 protein levels have already been shown to be strongly
regulated by light
10
, and growth in red and blue photocycles destabilises PIF4 protein at low temperatures
14
. To test this, we examined the levels of PIF4:HA protein at 12 °C, 17 °C, 22 °C and
27 °C under the same light conditions used for our flowering time assays. Consistent
with previous studies
10
we see a strong accumulation of PIF4 at the end of the night period, which is subsequently
degraded during the day. Despite the suppression of early flowering in 35S::PIF4 at
12 °C compared to 22 ° (Fig. 2b), we do not observe an appreciable difference in PIF4
protein levels at these two temperatures that is likely to account for these different
phenotypes (Fig. 2c, Fig. S3). Slightly higher levels of PIF4:HA appear to be present
at 27 °C (Fig. 2c), suggesting high temperature stabilisation of PIF4 may also contribute
to higher PIF4 activity at 27 °C.
Taken together, these data indicate that PIF4 regulates FT in a temperature dependent
manner. To determine if this is likely to be the case in planta, we analysed the spatial
expression of FT and PIF4. FT has a distinctive pattern of expression in the vasculature
of the leaf
15, 16
, and significantly PIF4 is expressed in the same domain (Fig. 3a). Since the regulation
of FT by PIF4 could be either direct or indirect, we used chromatin immunopurification
(ChIP) to analyse if PIF4 binds directly to the FT promoter proximal to the transcriptional
start site (TSS). This region of the promoter was chosen since it has been shown to
be both phylogenetically conserved and the site for light mediated regulation of FT
expression
16, 17
. We observe robust enrichment of PIF4 near to the TSS (Fig. 3b), indicating that
PIF4 binds this region in vivo to activate FT expression.
Given the striking effect of ambient temperature on PIF4 activity, which occurs even
when PIF4 is constitutively expressed, we hypothesised that the ability of PIF4 to
bind the FT promoter may be temperature dependent. To test this, we performed ChIP
experiments using 35S::PIF4 plants grown at 12 °C and 27 °C with primers flanking
an E-box in the FT promoter (Fig. 3c). Strikingly, we observe a very strong temperature
dependence for this binding, with an approximately 5-fold increase in binding at 27
°C compared to 12 °C (Fig. 3d). This indicates that the later flowering of 35S::PIF4
at 12 °C is caused by a decrease in PIF4 binding to FT. Since the 35S promoter causes
strong ectopic expression of PIF4, we sought to confirm that PIF4 protein expressed
at endogenous levels displays similar temperature dependent binding to the FT promoter.
We therefore performed ChIP experiments on a pif4-101 line complemented with PIF4pro::PIF4:ProteinA
(Fig. S4). Consistent with the overexpression studies, we observe a strong increase
in PIF4 binding to FT as a function of temperature. Reduced binding is observed at
17 °C, consistent with the very late flowering of plants under short days at low temperature,
but this binding increases at 22 °C and is even higher at 27 °C (Fig. 3e). The temperature
dependent binding of PIF4 to FT could be due to growth temperature influencing the
affinity of the PIF4 transcription factor for its binding site or the efficiency of
the ChIP could be affected by the temperature that tissues were grown at. To test
these possibilities, we analysed another recently described PIF4 target locus
18
, CYP79B2 (At4g39950), which is up-regulated in 35S::PIF4 (Fig. S5a). We find PIF4
binding to occur constitutively at both 12 and 27 °C at a region in the first exon
(Fig. S5b). Another region further upstream in the promoter shows a temperature dependent
binding of PIF4, and in both cases, no enrichment is seen for a control locus (Fig.
S5b). This indicates that the abundant PIF4 protein we observe at 12 °C is active
and able to bind target sites and confirms that the ChIP method per se is not influenced
by the temperature at which the sample is grown, consistent with other studies
19
. The ability of PIF4 to bind loci in a more temperature independent-manner might
explain why 35S::PIF4 at 12 °C maintains hypocotyl and petiole elongation, while early
flowering is strongly suppressed. We do not exclude that temperature may also influence
PIF4 activity post-translationally.
Temperature signals are mediated through H2A.Z-nucleosomes in Arabidopsis
20
, suggesting that temperature may be increasing the accessibility of the PIF4 binding
site at the FT promoter. Consistent with this hypothesis, we find that H2A.Z-nucleosomes
are present at the PIF4 binding site in the FT promoter. Furthermore, we find that
the levels of H2A.Z-nucleosomes at the FT promoter decrease with higher temperature
(Fig. 3f). These results suggest that the presence of H2A.Z nucleosomes is limiting
for PIF4 binding to FT, and that the PIF4 binding we observe at higher temperature
is due to the greater accessibility of chromatin containing H2A.Z-nucleosomes at higher
temperature. This suggests that in the absence of H2A.Z-nucleosomes, PIF4 should bind
FT more strongly. We therefore compared the ability of PIF4 expressed under its own
promoter to bind to the FT promoter in wild-type compared to arp6-1, a background
lacking incorporation of H2A.Z-nucleosomes. Interestingly, we observe considerably
greater binding of PIF4 in arp6-1 (Fig. 3g), indicating that H2A.Z-nucleosomes are
rate-limiting for PIF4 to activate FT expression. The eviction of H2A.Z-nucleosomes
by higher temperature therefore provides a direct mechanism for the temperature regulated
expression of FT (Fig. 4c). Consistent with our previous results and the established
role of H2A.Z in regulating temperature dependent gene expression, we find that there
is increased PIF4 mRNA in arp6-1 background (Fig. S6). However, our results for 35S::PIF4
suppression by 12 °C indicate that transcriptional up-regulation of PIF4 is not the
rate-limiting step in regulating PIF4 mediated flowering at higher temperatures.
Our results indicate that the temperature dependent regulation of FT by PIF4 is controlled
at the level of chromatin accessibility of the FT promoter and possibly at the level
of PIF4 protein activity. PIF4 activity is controlled through the repressive activity
of DELLA proteins that prevent PIF4 binding DNA
21, 22
. Consistently, plants having reduced or absent DELLA function are early flowering
23
. We hypothesised that delay in flowering at lower temperatures might at least in
part be due to DELLA mediated repression of PIF4 activity. If so, it would be expected
that absence of DELLAs should cause accelerated flowering at lower temperatures. In
accord with this expectation, we found that a mutant lacking DELLAs flowers much earlier
than wild-type when grown at 12 °C (Fig. 4a and 4b). The phytohormone gibberellin
(GA) triggers DELLA protein degradation, and plays a key permissive role for FT induction,
since in a GA deficient background, GA application increases FT expression 15-fold
24
. While it was proposed more than 50 years ago that gibberellins are upstream of florigen
25
, the mechanism has not been clear. As DELLA proteins have been shown to be key regulators
by which GA influences PIF4, our finding that PIF4 is able to directly activate FT
suggests a possible mechanism by which changes in GA levels may influence flowering.
Climate change has already caused measurable changes in plant phenology and behaviour
2
, and plants that incorporate temperature information into their lifecycles appear
to be able to adapt to warmer conditions more effectively than those plants that primarily
rely on photoperiod to synchronise their lifestyles
3
. The importance of the effects of climate change on yield are highlighted by the
significant detrimental effects of increasing temperatures on yield
4
. PIF4 is a central integrator of environmental information in the plant and our finding
that it activates FT at higher temperatures suggests it will be a key node for breeding
crops resilient to climate change. This importance is suggested by the recent discovery
that natural variation at PIF4 plays a major role in key ecological traits
26
.
ONLINE METHODS
1. Plant material and growth conditions
All plant lines used were in Col-0 background unless otherwise specified. pif4-101
mutant was provided by C. Fankhauser, HA tagged 35S::PIF4 by S. Prat
12
All references to “35S::PIF” and “PIF:HA” refer to this line, i.e. 35S::PIF4:HA. FT::GUS
was obtained from K.Goto
15
. phyb-9 ft-10 double mutants were generated by crossing respective single mutants.
35S::PIF4:HA co-9 was obtained by crossing. The crosses were genotyped for presence
of the 35S::PIF4 construct by PCR on genomic DNA using primers 2362 and 2363, resulting
in two products of different size representing the cDNA transgene and the genomic
DNA fragment, respectively. ft-10 was genotyped using primers 1580 and 1582 for the
insertion, or 1580 and 1581 to detect the wildtype fragment. co-9 was genotyped with
primers 3650 and 3652 for insertion, and 3291 and 3292 for the wildtype fragment.
For genotyping phyb-9, DNA was amplified using oligos 2137 and 2138 followed by Mnl
I digestion to distinguish between wildtype and mutant alleles. The global della mutant
is in the Ler background and was described previously
8
. PIF4::PIF4:PROTEINA and PIF4::PIF4:GUS were constructed by amplifying the genomic
fragment of PIF4 including the promoter with oligos 1534 and 1535. The PCR product
was cloned into pENTR/D-TOPO (Invitrogen) and inserted into the binary plasmids PW889
(C-terminal PROTEIN A) and PW395 (C-terminal GUS), respectively, using Gateway technology
(Invitrogen). Transgenic plants were obtained by transforming pif4-101 by floral dip.
For hypocotyl measurements seeds were surface sterilized, sown on ½ MS media, stratified
for 2 days at 4°C in the dark and germinated for 24 h at 22°C. The plates were then
transferred to short day conditions (8/16 h photoperiod) at 22°C and 27°C respectively
and grown vertically for 10 days before being imaged and hypocotyl length measured
using the ImageJ software (http://rsbweb.nih.gov/ij/). Oligonucleotide sequences are
provided in Table S1.
2. Transcript analysis
Samples from plants grown in long days (16/8 h photoperiod) were harvested and total
RNA was extracted using Trizol Reagent (Invitrogen). 2 μg of RNA were treated with
DnaseI (Roche) and used for cDNA synthesis (First strand cDNA synthesis kit, Fermentas).
cDNA was diluted 1:8 and used for quantitative PCR using a Roche Lightcycler 480 and
the corresponding Sybr Green master mix. To detect FT transcript levels, oligos 3180
and 3181 were used, for CO oligos 2951 and 2952. PIF4 transcript levels were analyzed
using oligos 3952 and 3953. Oligos 3247 and 3408 amplifying TUB6 (At5g12250) were
used for normalization.
3. Immunoblot analysis
For analysing the possible effect of temperature on PIF4 protein stability, Plants
overexpressing PIF4:HA (35S::PIF4:HA) were used. Seven day old 35S::PIF4:HA seedlings
grown in short days at 17 °C were transferred to 12, 17, 22 and 27 °C in short days
for 2 days. Samples were collected at end of night (EON) and thereafter 30 min, 1
hour and 4 hours under illumination. Protein samples were separated by SDS-PAGE and
transferred on to nitrocellulose membrane. PIF4:HA was detected using HRP conjugated
anti-HA antibody (Miltenyi Biotech) and visualised by chemiluminscent detection using
Immobilon Chemiluminescent HRP substrate (Millipore).
4. GUS histochemical assay
For GUS-staining plants were grown on ½ MS plates in long days (16/8 h photoperiods)
for 10 days and kept in the dark for 24 h before harvesting. Plants were stained in
buffer containing 100 mM phosphate buffer, pH 7, 10 mM EDTA, 0.1% Triton-X100, 0.5
mM K-ferrocyanide and 1 mM X-gluc at 37°C for 24 h before destaining in ethanol.
5. Chromatin immunoprecipitation (ChIP)
ChIP was performed as described
20
with minor modifications. 35S::PIF4:HA seedlings were grown on ½ MS plates for 10
days and kept in the dark for 24 h at respective temperatures before harvesting. 1.5
g of plant tissue and 4 μg of antibody (HA-tag antibody ab9110 from Abcam) were used
for ChIP. To analyse the dynamics of PIF4::PIF4:PROTEINA, plants were grown in respective
temperatures under short day conditions for four weeks. Aerial parts of the plants
were collected and cross-linked before being used for chromatin preparations. ChIP
was done using magnetic beads (Dynabeads M-270 Epoxy, Invitrogen) coated with rabbit
IgG (Sigma, I5006) as described (http://www.ncdir.org/protocols/Rout/Conjugation%20of%20Dynabeads.pdf).
To analyse H2A.Z dynamics at the FT locus in response to temperature, we used 3 week
old seedlings of HTA11::HTA11:GFP grown at 17 °C and 27 °C. ChIP was done using anti-GFP
antibody (Abcam, ab290). To analyze PIF4 binding in Col-0 and arp6-1 backgrounds,
respective genotypes with PIF4::PIF4:3XFLAG were grown on soil at 22 °C under short
photoperiods for three weeks before samples fixed by formaldehyde cross-linking. ChIP
was performed using anti-FLAG M2 affinity gel (Sigma A2220). Immuno-complexes were
eluted using 3X FLAG peptide (Sigma F4799) according to manufacturer’s instructions.
Immunoprecipitated DNA was eluted after reverse cross-linking by boiling at 95 °C
for 1 min in presence of 10 % Chelex (BioRad laboratories) followed by treatment with
Proteinase K. Oligonucleotides 3255 and 3256 for FT-15 region, 3613 and 3614 for FT-c1,
3607 and 3608 for FT-c and 3261 and 3262 for FT-f were used for detecting PIF4 binding
to the FT locus. As a positive control for PIF4 binding, At5g45280 was analysed using
oligos 2857 and 2958. HSP70 was used as a negative control using oligos 1862 and 1865.
For analysing PIF4 binding at At4g39950 oligonucleotides 4240 and 4241 was used for
region 1; and oligonucleotides 4246 and 4247 were used for region 2. Oligos 1860 and
1861 were used for HSP70 as a negative control. Oligonucleotide sequences are provided
in Table S1.
Supplementary Material
1