An Intact Krüppel-like factor 9 Gene Is Required for Acute Liver Period 1 mRNA Response to Restraint Stress

Abstract

The clock protein period 1 (PER1) is a central component of the core transcription-translation feedback loop governing cell-autonomous circadian rhythms in animals. Transcription of Per1 is directly regulated by the glucocorticoid (GC) receptor (GR), and Per1 mRNA is induced by stressors or injection of GC. Circulating GCs may synchronize peripheral clocks with the central pacemaker located in the suprachiasmatic nucleus of the brain. Krüppel-like factor 9 (KLF9) is a zinc finger transcription factor that, like Per1, is directly regulated by liganded GR, and it associates in chromatin at clock and clock-output genes, including at Per1. We hypothesized that KLF9 modulates stressor-dependent Per1 transcription. We exposed wild-type (WT) and Klf9 null mice (Klf9^-/-) of both sexes to 1 hour restraint stress, which caused similar 2- to 2.5-fold increases in plasma corticosterone (B) in each genotype and sex. Although WT mice of both sexes showed a 2-fold increase in liver Per1 mRNA level after restraint stress, this response was absent in Klf9^-/- mice. However, injection of B in WT and Klf9^-/- mice induced similar increases in Per1 mRNA. Our findings support that an intact Klf9 gene is required for liver Per1 mRNA responses to an acute stressor, but a possible role for GCs in this response requires further investigation.

A cell-autonomous circadian clock maintains cellular circadian rhythms in virtually all cells of animals. The core clock comprises a transcription-translation feedback loop with a positive and negative limb (1). In vertebrates, the positive limb comprises the transcriptional activators Circadian Locomotor Output Cycles Kaput (CLOCK) and brain and muscle ARNT-like 1 (BMAL1) that form heterodimers, bind to DNA at E-box motifs, and activate transcription of hundreds of genes (2). Targets of CLOCK-BMAL1 include the Period (Per) and Cryptochrome (Cry) genes whose protein products (PER and CRY) form the negative limb of the core clock. As cytoplasmic concentrations of PER and CRY increase, they form heterodimers, translocate to the nucleus, and disrupt CLOCK+BMAL1 actions (1). As the intracellular PER and CRY levels decline, CLOCK+BMAL1 activity increases, resulting in a reciprocal oscillation with a period of ~24 hours in the activity of these 2 protein complexes.

Whole-organism circadian rhythms in vertebrates are controlled by a master circadian pacemaker in the suprachiasmatic nucleus (SCN) of the hypothalamus (3), which controls oscillation in adrenal stress hormones (glucocorticoids [GCs]; e.g., cortisol, corticosterone [B]) (4, 5). Per1 transcription is induced by stress or GC treatment (6, 7), GCs may entrain peripheral circadian clocks, and GC regulation of Per genes may be part of a mechanism to synchronize peripheral cellular clocks with the SCN (8-11).

Some Krüppel-like factors (KLFs) are clock-output genes, and they modulate transcription of cellular circadian clock genes (12-18). KLF9 associates in chromatin in mouse hippocampus-derived cells (HT22) at several core circadian clock and clock-output genes (17), and can activate or repress transcription depending on the gene and cellular context (19). Like Per1, Klf9 transcription is induced by GCs via 2 evolutionarily conserved upstream glucocorticoid response elements (GREs) (20). Circadian variation in Klf9 mRNA levels in human keratinocytes correlated with changes in circulating cortisol, and KLF9 acted downstream of the core circadian transcription-translation feedback loop to modulate circadian variation in keratinocyte proliferation (12). KLF9, and perhaps other KLFs, can modulate cellular circadian clock activity, and may participate in the synchronization of peripheral cellular clocks by stress hormones (11).

Given that KLF9 associates in chromatin at Per1 (17, 19), and Per1 and Klf9 are induced in vivo by stress or GC injection (7, 11, 20-22), we hypothesized that KLF9 modulates stressor-dependent Per1 transcription. Because we previously found that KLF9 acts predominantly as a transcriptional repressor (19), we hypothesized that KLF9 counteracts GC induction of Per1, thereby preventing stressor-induced phase shifting of the circadian clock (23). We focused on the liver because stressor-dependent induction of Per1 in liver depends on a GRE in the Per1 promoter (7), Per1 and Klf9 mRNAs show circadian oscillation in liver (7, 17), and KLF9 is required for rhythmic activation of gene enhancers in hepatocytes (18). We subjected wild-type (WT) and Klf9^-/- mice of both sexes to restraint stress, then measured Per1 mRNA. We also investigated if Klf9 deficiency affected the Per1 mRNA response to exogenous GC by injecting WT and Klf9^-/- mice with B.

Materials and Methods

Animal care and use

We reared mice on 12L:12D with food and water ad libitum. We generated Klf9 null mice (Klf9^-/-) by breeding male homozygous Klf9^-/- mice (24) with WT female C57BL/6J mice (Jackson Laboratory, Bar Harbor, ME) and intercrossing heterozygous F1 offspring. We genotyped using forward primer: AGCGCGAGGTGACCAAGGAA; reverse primer: CGGGCTGTGGGAAGGACTCG. All procedures involving animals were conducted under an approved animal use protocol (PRO00006809) in accordance with the guidelines of the Institutional Animal Care and Use Committee at the University of Michigan.

We exposed mice to restraint stress (25) by placing animals (12-16 weeks of age) of both sexes and 2 genotypes headfirst into 50-mL conical tubes (Falcon) with 2 one-quarter-inch holes drilled into each side, and 1 hole in the cap. We placed 2.5-cm portions of 15-mL conical tubes over their tails to inhibit movement, then passed the tail through the hole in the cap while securing it. We restrained mice for 1 hour, then we either euthanized them immediately (restraint group) by rapid decapitation, or allowed them to recover for 2 hours in their home cage (restraint+recovery group) before euthanasia. Unhandled control animals were euthanized immediately after removal from their home cage. All animals were euthanized within a 1.5-hour time window between ZT0530 and ZT0700 (ZT, zeitgeber time; ZT0, lights-on time). We collected trunk blood into heparinized tubes and analyzed plasma B concentration by enzyme-linked immunosorbent assay (ELISA) kit (Arbor Assays, Ann Arbor, MI; RRID:AB_2877626). We conducted pilot assays to determine the appropriate dilution for the assay (1:200) so that samples fell on the linear part of the standard curve. Duplicate samples from both genotypes and sexes were run in the same assay. We dissected liver, snap-froze it in liquid nitrogen, and stored it at –80°C until RNA extraction.

For B injection, we divided sexually mature mice (12-16 weeks of age) of both sexes and genotypes into 3 treatments: unhandled (left in home cage until killed), vehicle-injected (corn oil), and B-injected (14 mg/kg body weight in corn oil; dose from previous work (20, 26)). At time 0, we administered IP injections (or left animals undisturbed), then euthanized after 2 hours between ZT0600 and ZT0700. We collected trunk blood for B ELISA and liver for RNA extraction.

RNA extraction and real-time quantitative PCR

We used TRIzol reagent (Invitrogen) to extract total RNA from mouse liver, treated with DNase 1 (20U; Roche) to remove genomic DNA and reverse-transcribed 1 μg of RNA using the High Capacity Reverse Transcription kit with ribonuclease inhibitor (Life Technologies Corp). We conducted real-time quantitative PCR (qPCR) with Absolute qPCR SYBR low ROX mix (ABgene). We constructed standard curves by pooling cDNA from all samples and making serial 10-fold dilutions. We normalized Per1, Fkbp5, and Klf9 mRNA levels to the mRNA level of the reference gene Ppia, which was unaffected by treatments. The oligonucleotide primer sequences used were:

• Per1 mRNA: Fwd 5′TGTGTCAAGCAGGTTCAG 3′; Rev

• 5′TGTCCTGGTTTCGAAGTGTG 3′

• Fkbp5 mRNA: Fwd 5′ GAAGTTTGATTCAGTCATGA CAGA 3′; Rev 5′CCAATGTCCCAGGCTTTGAT 3′

• Klf9 mRNA: Fwd 5′ GCACAAGTGCCCCTACAGT 3′; Rev 5′TGTATGCACTCTGTAATGGGCTTT 3′

• Ppia mRNA: Fwd 5′ TATCTGCACTGCCAAGACTGAATG 3′; Rev 5′CTTCTTGCTGGTCTTGCCATTCC 3′

Data analysis and statistics

We conducted factorial ANOVA and Fisher’s least significant difference post hoc test (α < 0.05) using Systat 13 (Systat Software). Values with studentized residuals >3 were flagged as outliers.

Results

KLF9 associates in chromatin at the Per1 locus, overlapping 2 GREs

We visualized KLF9 peaks at Per1 using the UCSC Genome Browser with data from a KLF9 chromatin-streptavidin sequencing (ChSP-seq) experiment on HT22 cells (19) and found KLF9 association in chromatin at the first intron and 5′ upstream region of Per1 (Fig. 1A) (17, 19). KLF9 ChSP-seq peaks overlap with 2 binding sites for GC receptor (GR; arrows in Fig. 1A) (7, 27).

Figure 1.

Krüppel-like factor 9 (KLF9) associates in chromatin at the Per1 locus. Restraint stress induced Per1 mRNA in liver of wild type, but not of Klf9^-/- mice. (A) Mapped ChSP-seq reads at the Per1 locus show 2 regions of KLF9 association in chromatin in HT22 cells (ChSP data from (19)). Gray boxes indicate computationally identified KLF9 peaks. Arrows indicate locations of glucocorticoid response elements (7, 27). (B) Outline of experimental protocol for restraint stress and recovery. Mice were either left unhandled (control), restrained for 1 hour and immediately killed (restraint), or restrained for 1 hour, allowed to recover in their home cage for 2 hours, and then killed (restraint+recovery). Treatments were timed such that all animals were killed at the same time of day (1 p.m., or zeitgeber 0500). Open bars indicate 1-hour periods in home cage; filled bars indicate 1-hour periods of restraint. (C) WT and Klf9^-/- mice show different Per1 mRNA responses to acute restraint stress, but no differences in changes in plasma B concentration. We exposed mice to restraint stress, analyzed Per1 mRNA RT-qPCR, and normalized it to the reference gene Ppia. We analyzed plasma B by ELISA. Top: female (left) and male (right) WT mice induce Per1 mRNA ~2-fold after 1 hour of restraint, whereas Per1 mRNA levels were unchanged in Klf9^-/- mice. Bottom: female (left) and male (right) WT and Klf9^-/- mice show equal responses in plasma B after acute restraint stress and return to baseline after 2-hour recovery. Bars represent the mean ± SEM, and sample sizes are given in parentheses. *Significantly different from control and restraint+recovery groups within a genotype, P < 0.05, Fisher’s least significant difference post hoc test. Abbreviations: B, corticosterone; ChSP-seq, chromatin-streptavidin sequencing; Per, Period; RT-qPCR, real-time qPCR.

Restraint stress increased liver Per1 mRNA level in WT but not in Klf9^-/- mice

We hypothesized that Klf9 deficiency enhances the Per1 response to an acute stressor. Exposure to 1-hour restraint stress caused ~2-fold increase in the mean Per1 mRNA level in WT animals of both sexes, but no change in Klf9^-/- mice (Fig. 1C). Three-factor ANOVA revealed statistically significant main effects of genotype, treatment, and sex, and significant genotype × treatment, and genotype × treatment × sex interactions (Table 1). Restraint caused statistically significant (P < 0.05) increases in Per1 mRNA level in WT mice of both sexes, but there were no differences between treatments in Klf9^-/- animals. The mean Per1 mRNA level returned to baseline by the end of the 2-hour recovery period in all animals.

Table 1.

Three-factor ANOVA results for restraint stress experiment

	Per1	Fkbp5	Plasma corticosterone
Geno	F(1,55) = 4.184, P = 0.046	F(1,52) = 0.040, P = 0.841	F(1,49) = 0.001, P = 0.979
Trt	F(2,55) = 8.574, P = 0.001	F(2,52) = 36.393, P < 0.0001	F(2,49) = 91.783, P < 0.0001
Sex	F(1,55) = 18.885, P < 0.0001	F(1,52) = 7.408, P = 0.009	F(1,49) = 5.828, P = 0.020
Geno × Trt	F(2,55) = 7.562, P = 0.001	F(2,52) = 0.563, P = 0.573	F(2,49) = 2.405, P = 0.101
Geno × Sex	F(1,55) = 0.756, P = 0.388	F(1,52) = 0.136, P = 0.713	F(1,49) = 0.083, P = 0.774
Trt × Sex	F(2,55) = 1.527, P = 0.226	F(2,52) = 0.255, P = 0.776	F(2,49) = 1.519, P = 0.229
Geno × Trt × Sex	F(2,55) = 3.377, P = 0.041	F(2,52) = 7.448, P = 0.001	F(2,49) = 0.304, P = 0.739

Abbreviations: Geno, genotype; Trt, treatment (control, restraint, restraint+recovery).

There were statistically significant main effects of treatment and sex, but no significant effect of genotype or interactions between factors on plasma B concentration (Table 1). Mice of both sexes and genotypes showed statistically significant (P < 0.05) and similar increases in plasma B concentration after restraint stress (Fig. 1C).

Restraint stress increased liver Fkbp5 mRNA level in WT and Klf9^-/- mice

We investigated changes in the mRNA level of a direct GR target gene that does not have KLF9 binding sites, Fkbp5 (19, 28). Fkbp5 mRNA levels showed similar, 2- to 2.5-fold increases in the restraint+recovery group compared with the control group in both sexes and genotypes; the mean mRNA level during this period was lower in male WT compared with Klf9^-/- animals (Fig. 2). Three-factor ANOVA revealed statistically significant main effects of treatment and sex, but no effect of genotype, and a significant genotype × treatment × sex interaction (Table 1). The Fkbp5 mRNA level did not change in male or female WT mice after one hour of restraint (Fig. 2), but there were statistically significant increases (P < 0.05) in both male and female Klf9^-/- mice during this time. Activation of Fkbp5 by restraint in both WT and Klf9^-/- mice is consistent with normal GR function at genes not directly regulated by KLF9.

Figure 2.

Restraint stress induced Fkbp5 mRNA in wild-type (WT) and Klf9^-/- female and male mice. We analyzed the Fkbp5 mRNA level by RT-qPCR, and normalized it to the the reference gene Ppia. Bars represent the mean ± SEM and sample sizes are given in parentheses. *Significantly different from control group within a genotype, P < 0.05, Fisher’s least significant difference post hoc test. Abbreviation: RT-qPCR, real-time qPCR.

Injection of B increased liver Per1 mRNA level in WT and Klf9^-/- mice

To test if the failure of Per1 mRNA to respond to restraint in Klf9^-/- mice was due to an impaired response to B, we injected WT and Klf9^-/- mice with B. The Per1 mRNA level was increased by B injection ~2- to 3-fold in all animals, and this response was unaffected by genotype (Fig. 3). Three-factor ANOVA revealed a statistically significant main effect of treatment, and a significant treatment × sex interaction (Table 2). Injection of B caused a 5- to 10-fold increase in plasma B concentration in all animals (Fig. 3); there were significant main effects of treatment and sex, and significant genotype × treatment and treatment × sex interactions (Table 2).

Figure 3.

Intraperitoneal injection of corticosterone (B) increased Per1 mRNA level in liver of wild-type (WT) and Klf9^-/- mice. Sexually mature mice (8-12 weeks of age) were divided into 3 treatments (n = 4-6/treatment): unhandled (left in their home cage until killed), vehicle-injected (corn oil) and B-injected (14 mg/kg body weight in corn oil). They were given IP injections (or left unhandled) and killed 2 hours later. We analyzed liver Per1 mRNA by RT-qPCR, and normalized it to the reference gene Ppia. We analyzed plasma B by ELISA. Bars represent the mean ± SEM and sample sizes are given in parentheses. *Significantly different from the vehicle control group within a genotype; #significantly different from the unhandled group, P < 0.05, Fisher’s least significant difference post hoc test. Abbreviation: RT-qPCR, real-time qPCR.

Table 2.

Three-factor ANOVA results for corticosterone injection experiment

	Per1	Plasma corticosterone
Geno	F(1,40) = 0.150, P = 0.701	F(1,42) = 0.761, P = 0.388
Trt	F(2,40) = 42.059, P < 0.0001	F(1,42) = 183.195, P < 0.0001
Sex	F(1,40) = 1.421, P = 0.241	F(2,42) = 10.062, P = 0.003
Geno × Trt	F(2,40) = 0.514, P = 0.602	F(2,42) = 4.266, P = 0.045
Geno × Sex	F(1,40) = 0.123, P = 0.728	F(1,42) = 0.159, P = 0.854
Trt × Sex	F(2,40) = 3.312, P = 0.047	F(2,42) = 5.519, P = 0.007
Geno × Trt × Sex	F(2,40) = 0.633, P = 0.536	F(2,42) = 0.686, P = 0.509

Abbreviations: Geno, genotype; Trt, treatment (control, restraint, restraint+recovery).

Discussion

Here, we show that an intact Klf9 gene is required for liver Per1 transcriptional response to restraint stress in both male and female mice. The lack of a Per1 response to stress in Klf9^-/- animals was not due to an impaired hypothalamic-pituitary-adrenal axis response, as both WT and Klf9^-/- mice increased circulating B to similar levels after restraint. A role for KLF9 in Per1 transcription is consistent with findings that KLF9 associates in chromatin at the Per1 5′ upstream region and first intron in HT22 (17, 19). Surprisingly, injection of B increased liver Per1 mRNA similarly in WT and Klf9^-/- animals, suggesting that the impaired Per1 mRNA response to stress in Klf9^-/- mice was not due to a deficient response to B (but see below).

Several Klf genes including Klf9 are clock-output genes, and their protein products regulate transcription of cellular circadian clock genes (12-18). KLF9 associates in chromatin at circadian clock genes (17), and it is one of a handful of transcription factors required for rhythmic activation of gene enhancers in hepatocytes (18). It can activate or repress transcription (19, 29-31), and in HT22 it acts predominantly as a transcriptional repressor (19) (including repression of Per1; A. Subramani and R.J. Denver, unpublished data). We therefore hypothesized that restraint stress would amplify or lengthen induction of Per1 in mice lacking Klf9. However, our data show the opposite: no change in Per1 mRNA after restraint in Klf9^-/- mice despite a normal hypothalamic-pituitary-adrenal axis response. Our results support that KLF9 maintains transcriptional competency for liver stressor-dependent Per1 activation (29).

KLF9 has emerged as a key player in neuroendocrine stress responses in vertebrates. Klf9 transcription in mammals is induced by GCs via 2 GREs located upstream of the transcription start site (20). In zebrafish, KLF9 is a major mediator of secondary transcriptional responses to cortisol (32). Cortisol drives circadian variation in Klf9 mRNA levels in human keratinocytes, and KLF9 acts downstream of the core circadian transcription-translation feedback loop to modulate circadian variation in keratinocyte proliferation (12). KLF9 mediates chronic stress-induced enlargement of dendritic spines in mouse hippocampus and fearful/depressive behaviors (33).

Restraint stress or B injection induced Per1 mRNA in different mouse tissues (7, 34), and Per1 mRNA pulsing in rat hippocampus depends on the ultradian rhythm in plasma B (35). The actions of B on Per1 and other cellular circadian clock genes may facilitate SCN synchronization with peripheral tissues (7, 10, 11, 36-38). The lack of a Per1 response to restraint stress in Klf9^-/- mice led us to hypothesize that the response to elevated circulating B was impaired. Stressor-dependent induction of Per1 depends on GCs in many but not all tissues (11). In liver, stressor induction of Per1 is mediated by a GRE in the Per1 promoter (7). However, there is also evidence that increases in Per1 mRNA after stress may be independent of GCs, or GCs have different actions (activation vs. repression) in different brain regions (39, 40). In a pilot experiment using HT22 cells in which we inactivated Klf9 (19), the Per1 mRNA level was significantly lower during the recovery period after 1 hour of B treatment compared with WT, supporting that Klf9 deficiency impairs the Per1 response to B (A. Subramani and R.J. Denver, unpublished data). However, here we found no effect of Klf9 deficiency on the liver Per1 mRNA response to injected B, which may indicate that the actions of other factors induced by stress (e.g., catecholamines) are impaired by Klf9 deficiency. There is evidence that acute, stress-induced changes in gene expression depend on β-adrenergic signaling. Injection of propranolol blocked the increase in several stress-induced genes in hippocampus, although there was no effect of propanolol on Per1 (41). However, it is also possible that our findings were caused by a supraphysiological increase in plasma B concentration after B injection (~10- to 20-fold above baseline, ~3-fold greater than that achieved after exercise) (42), which could have led to artifactual overactivation of GR-dependent signaling, with any effect of Klf9 deficiency masked by the supraphysiological increase in plasma B.

KLF9 associates in chromatin at 2 sites within Per1, overlapping 2 GREs (7, 17, 19, 27) where it could act as an accessory transcription factor for GR transcriptional activity under physiological B concentration. Some KLFs, in addition to being regulated by nuclear hormone receptors, function as accessory transcription factors for nuclear hormone receptor activity (29, 43). For example, KLF15 acts cooperatively with GR on select genes with binding sites for the 2 transcription factors (23, 44). KLF9 is an accessory transcription factor for autoinduction of thyroid hormone receptor beta in Xenopus tadpole brain (45), and it acts cooperatively with progesterone receptor to maintain uterine function and embryo implantation (46-49). Additional experiments are needed to elucidate the molecular mechanism by which Klf9 deficiency impairs stressor activation of liver Per1.

Glossary

Abbreviations

B

corticosterone

BMAL

brain and muscle ARNT-like

ChSP-seq

chromatin-streptavidin sequencing

CLOCK

Circadian Locomotor Output Cycles Kaput

Cry

Cryptochrome

ELISA

enzyme-linked immunosorbant assay

GC

glucocorticoid

GR

glucocorticoid receptor; GRE, glucocorticoid response element

HT22

mouse hippocampus-derived cell line

KLF

Krüppel-like factor

Per

Period

SCN

suprachiasmatic nucleus

WT

wild-type

ZT

zeitgeber time

Additional Information

Disclosures: The authors declare that they have no conflict of interest with regard to the published work.

Data Availability

The datasets generated during and/or analyzed during the current study are not publicly available but are available from the corresponding author on reasonable request.