Altered expression of clock and clock-controlled genes in a hSOD1-linked amyotrophic lateral sclerosis mouse model.

Abstract

Most physiological processes in mammals are subjected to daily oscillations that are governed by a circadian system. Circadian rhythm orchestrates metabolic pathways in a time-dependent manner and loss of circadian timekeeping has been associated with cellular and system-wide alterations in metabolism, redox homeostasis, and inflammation. Here, we investigated the expression of clock and clock-controlled genes in multiple tissues (suprachiasmatic nucleus, spinal cord, gastrocnemius muscle, and liver) from mutant hSOD1-linked amyotrophic lateral sclerosis (ALS) mouse models. We identified tissue-specific changes in the relative expression, as well as altered daily expression patterns, of clock genes, sirtuins (Sirt1, Sirt3 and Sirt6), metabolic enzymes (Pfkfb3, Cpt1 and Nampt) and redox regulators (Nrf2, G6pd, Pgd). In addition, astrocytes transdifferentiated from induced pluripotent stem cells from SOD1-linked and FUS RNA binding protein-linked ALS patients also displayed altered expression of clock genes. Overall, our results raise the possibility of disrupted cross-talk between the suprachiasmatic nucleus and peripheral tissues in hSOD1^G93A mice, preventing proper peripheral clock regulation and synchronization. Since these changes were observed in symptomatic mice, it remains unclear whether this dysregulation directly drives or it is a consequence of the degenerative process. However, because metabolism and redox homeostasis are intimately entangled with circadian rhythms, our data suggest that altered expression of clock genes may contribute to metabolic and redox impairment in ALS. Since circadian dyssynchrony can be rescued, these results provide the groundwork for potential disease-modifying interventions.

Introduction

Circadian rhythms are genetically encoded by a molecular clock which generates a self-sustained internal rhythm of approximately 24 hours. The cell-autonomous molecular circadian clock is composed of interlocked positive and negative transcriptional/translational feedback loops (TTFL), which induces and represses the transcription of target genes, respectively. The positive arm of the core clock consists of brain and muscle ARNT-like 1 (Bmal1) and circadian locomotor output cycles protein kaput (Clock), while the negative arm of the core clock consists of Period1/Period2 (Per1/2) and Cryptochrome1/Cryptochrome2 (Cry1/2) (1). In addition, in a second interlocking TTFL, the nuclear receptor subfamily 1 Group D member 1 (Nr1d1, also known as Rev-erbα) suppresses the positive arm of the core clock. Nr1d1 and RAR-related orphan receptor α (Rorα) compete for binding to the E-box motif located in the Bmal1 promoter to directly repress or stimulate transcription, respectively (2–4).

Amyotrophic lateral sclerosis (ALS) is the most common adult onset motor neuron disease (5). ALS can be inherited (familial, FALS), or sporadic (SALS). SALS is the most common form of the disease, accounting for roughly 90% of cases. Superoxide dismutase 1 (SOD1) was the first gene linked to ALS (6), and rodents over-expressing ALS-linked mutant hSOD1 develop an ALS-like phenotype (7, 8). Mice overexpressing hSOD1 with the G93A mutation (hSOD1^G93A) are widely used for preclinical studies as they develop an ALS-like pathology that closely resembles the human disease. The pathophysiological analysis of this model has contributed to a significant portion of our mechanistic understanding of ALS pathology.

In addition to motor neuron degeneration, ALS patients and mouse models show signs of systemic metabolic alterations (9). A hypermetabolic state, marked by increased resting energy expenditure, has been observed in hSOD1-linked ALS mouse models, in familial cases of the disease, and in a subset of SALS patients (10–13). Several endocrine factors that display circadian rhythmicity and could participate in metabolic synchronization of peripheral tissues are altered in ALS patients and animal models of the disease. Among the best-studied examples are those governed by the hypothalamic-pituitary axis, such as cortisol and growth hormone. Morning plasma cortisol levels are increased in ALS patients (14–16) and ALS mice display circadian changes in serum corticosterone levels that negatively correlate with survival (17). In addition, growth hormone secretion, which has specific and powerful effects on metabolism, is impaired in ALS mice and patients (18, 19).

The circadian rhythm orchestrates metabolic pathways in a time-dependent manner (20) and loss of circadian timekeeping has been associated with cellular and system-wide alterations in metabolism, redox homeostasis, and inflammation (21–26). A growing body of data obtained from both human and animal studies has documented disruption of circadian timekeeping at physiological, molecular, and behavioral levels in Alzheimer’s, Parkinson’s, and Huntington’s diseases (25, 27, 28). A potential involvement of disrupted molecular clock in ALS pathology has not been evaluated. We examined the expression of clock and clock-controlled genes in the central nervous system and peripheral tissues of early symptomatic hSOD1^G93A mice over a 24-hour period. Compared to age-matched non-transgenic littermates, hSOD1^G93A mice displayed alterations of daily expression patterns of several clock and clock-controlled genes, which may contribute to the metabolic impairment and oxidative stress observed in this ALS mouse model.

Methods and Materials

Animals-

B6.Cg-Tg(SOD1*G93A)1Gur/J mice (7) were obtained from The Jackson Laboratory (Bar Harbor, ME, USA) and maintained in hemizygosis in a C57BL/6J background. Early symptomatic mice (around 120 days-old) were used for these studies. hSOD1H^46R/H48Q mice were provided by Dr. David Borchelt (29) and have been backcrossed into C57BL/6J pure background for more than 10 generations. Early symptomatic hSOD1H^46R/H48Q mice (around 210 days-old) were used for this study. Mice were fed ad libitum and kept under a 12/12 hours light/dark cycle in a light-tight cabinet (Actimetrics). All animal procedures were carried out in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the NIH. The Animal Care and Use Committee of UW-Madison and MUSC approved the animal protocol pertinent to the experiments reported in this publication.

Real-time PCR-

RNA extraction, RNA retrotranscription and real-time PCR were performed as previously described (30). Primers were obtained from Integrated DNA Technologies (Table 1).

Table 1.

Sequence of primers used for real-time quantitative PCR.

Target gene	Mouse-specific primers (sequence 5’-3’)	Human-specific primers (sequence 5’-3’)
Clock-F	CTACTGTCAGCCTACCTACCA	CTCTACTCATCTGCTGGAAAGTG
Clock-R	GTGTGTGCTATTCCTCCCATAA	ATGGCTCCTTTGGGTCTATTG
Bmal1-F	CAACCCATACACAGAAGCAAAC	CGGAGTCGATGGTTCAGTTT
Bmal1-R	CATCTGCTGCCCTGAGAATTA	CTTCCAGGACGTTGGCTAAA
Per1-F	CCTGGAGGAATTGGAGCATATC	GCAGCCACACAAGCAAATAC
Per1-R	CCTGCCTGCTCCGAAATATAG	GATCCTGGAGCACGTACTTAAT
Per2-F	CAACAACCCACACACCAAAC	GACGTGGCAGAATGTGTTTAC
Per2-R	CTCGATCAGATCCTGAGGTAGA	CTGAGTCCCAGAGAAGGAATATC
Cry1-F	CTCAGTCCTTATCTCCGCTTTG	GGATGTGGCTGTCTTGTAGTT
Cry1-R	CCACAGGAGTTGCCCATAAA	GTCTCCATTGGGATCTGTTCTC
Cry2-F	CCTGAGCAAGTCTTCCTTCTAC	TCGGAACAGTGCCTCAAATC
Cry2-R	TCACACACACACACACTCTC	CACACCTTCCCAGCAATAGAA
Nr1d1-F	CTTCATCCTCCTCCTCCTTCTA	GTCATCCTCCTCCTCCTTCTAT
Nr1d1-R	GTAATGTTGCTTGTGCCCTTG	TGATGTTGCTGGTGCTCTT
Rplp0-F	CCTCCTTCTTCCAGGCTTTG	TCCAAGTTCATGGGTCACAATA
Rplp0-R	CCACCTTGTCTCCAGTCTTTATC	TGGCTGACTTTGGAGATGAC
Sirt1-F	AGCAACATCTCATGATTGGCACCG
Sirt1-R	TCTGCCACAGCGTCATATCATCCA
Sirt3-F	ATCCCGTACCCTGAAGCCATCTTT
Sirt3-R	TCAAGCCCGTCGATGTTCTGTGTA
Sirt6-F	CGTCTGGTCATTGTCAACCT
Sirt6-R	GAGTCTGCACATCACCTCATC
Nampt-F	GCCACCTTATCTTAGAGTCATTCA
Nampt-R	GAGACATTCTCGATACTCCACTTC
Nrf2-F	TTCTTTCAGCAGCATCCTCTCCAC
Nrf2-R	ACAGCCTTCAATAGTCCCGTCCAG
Pfkfb3-F	GAAGATAAGAGTGGCAGGGAAC
Pfkfb3-R	CTCCTAGGCCAAGATGCATAAA
Cpt1a-F	TCGAAACCCAGTGCCTTAAC
Cpt1a-R	AAGCAGCACCCTCACATATC
Cpt1b-F	ATTCTGTGCGGCCCTTATT
Cpt1b-R	TGACTTGAGCACCAGGTATTT
G6pd-F	GAAGCTGCCAATGGATACTTAGA
G6pd-R	CCACCGTTCATTCTCCACATAG
Pgd-F	AGGACATGGTCTCCAAACTAAAG
Pgd-R	CGTGTCCAACAAGGGTACTAAT

Experimental Design and Statistical Analysis-

At each time point, groups of 3–4 animals per genotype were sacrificed every 4 hours over a 24-hour period and used for RNA collection. For all figures presenting animal data, the results corresponding to zeitgeber time 0 (ZT0) were also used as ZT24 for analysis and graphical representation. Significance for mean expression changes at individual timepoints was determined by multiple Student’s t-test analysis and differences were declared statistically significant if p ≤ 0.05. All mRNA expression data are reported as mean ± SEM.

To evaluate the periodicity of diurnal rhythms, MetaCyle R package, which incorporates Jonckheere-Terpstra-Kendall (JTK_cycle) algorithm (31), was run in RStudio (Version 1.2.5033.0). A p value cutoff of 0.05 and relative amplitude threshold of 0.15 were used to identify transcripts exhibiting significant daily rhythms. Transcripts that exhibited these two combined criteria in the non-transgenic animals were considered for further differential rhythmicity analysis between genotypes. Cosine fits of those transcripts were compared for differences in amplitude and phase. To establish differences between genotypes, data was fitted by cosinor analysis in GraphPad Prism 6.0 (GraphPad Software) using the following equation:

$${\text{x}}_{\mathrm{i}}=\text{M}+\text{A cos}\left(\frac{2{\mathrm{\pi }\text{t}}_{\text{i}}}{\text{P}}+\mathrm{\phi }\right)$$

This nonlinear least squares regression model estimates the unknown waveform parameters MESOR (M, midline estimating statistic of rhythm), amplitude (A), and phase (φ); assuming a fixed period (P) of 24 hours. Here, x_i is the relative mRNA expression level at which data were collected at (t_i, or ZT) (32). Significance for amplitude and phase shift differences were determined by 95% confidence intervals of the difference of means.

Patient-derived iPSC-induced astrocytes (iAs)-

Control and ALS induced pluripotent stem cells (iPSC) lines were obtained from the NINDS Human Cell and Data Repository or commercial vendors. iPSC lines details are as follows: SOD1^G86R (iPSC ID# ND39036, ALS-linked SOD1^G86R), Control 1 (iPSC ID# FA0000010, control), FUS^G522A (iPSC ID# ND39034, ALS-linked FUS^G522A) and Control 2 (iPSC ID# ND38555, control). iPSCs were differentiated into induced NPCs using an embryoid body formation protocol in the presence of SMAD signaling inhibitors (STEMdiff SMADi Media, Stemcell). Induction was confirmed by an increase in MAP2, PAX6 and NESTIN gene expression and concurrent decrease in SOX2, OCT3 and NANOG expression. Induced NPCs were cultured for three weeks in astrocyte differentiation media (STEMdif Astrocyte Differentiation Media, Stemcell), following by three weeks in astrocyte maturation media (STEMdif Astrocyte Maturation Media, Stemcell). Astrocyte differentiation was confirmed by assessing GFAP, S100B, and ALDH1L1 gene expression. Following differentiation, iAs were cultured in DMEM-F12 supplemented with 10% FBS and 0.3% N2 supplement. Confluent iAs were treated with 100nM dexamethasone (Sigma) for 1 hour in media with reduced serum (2% FBS), after which the dexamethasone was removed and cells were returned to 10% FBS media. mRNA samples for the first timepoint, circadian time 0 (CT0), were harvested 24 hours later and subsequent mRNA samples were harvested every 4 hours over a 24-hour period.

Results

We first examined the expression of core clock genes in the SCN of early symptomatic hSOD1^G93A mice (120 days-old) compared to age-matched non-transgenic controls (Fig. 1A). Overall, we did not observe many significant changes in the relative expression of core clock genes in the SCN of hSOD1^G93A mice. However, Nr1d1 mRNA levels revealed significant differential expression in the SCN of hSOD1^G93A mice compared to non-transgenic mice at both ZT16 and ZT20 (Fig. 1A). Furthermore, Nr1d1 mRNA displays a loss of rhythmic gene expression in the SCN of hSOD1^G93A mice (Table 2).

Figure 1. Altered expression of core clock genes in early symptomatic hSOD1^G93A mice.

mRNA levels of core clock genes were determined by qRT-PCR in tissues from early symptomatic hSOD1^G93A (G93A) and age-matched non-transgenic (NonTG) mice every 4 hours over a 24-hour period. A) Suprachiasmatic nucleus (SCN), B) spinal cord (SC), C) gastrocnemius (Gastroc), and D) liver. Time of harvest is shown in zeitgeber time (ZT) and for graphical representation ZT0 data was also used as ZT24. Data was normalized by Rplp0 and is presented as percentage of non-transgenic mice at ZT0 (mean ± SEM, * p < 0.05, n = 3–4).

Table 2.

Waveform analysis and statistical comparison of the expression of clock and clock-controlled genes in different tissues form early symptomatic hSOD1^G93A and non-transgenic mice.

	SCN		Spinal Cord		Gastrocnemius		Liver
	Amplitude (% change)	Phase Shift (hrs)	Amplitude (% change)	Phase Shift (hrs)	Amplitude (% change)	Phase Shift (hrs)	Amplitude (% change)	Phase Shift (hrs)
Bmal1	–	–	–	–	✓ (−27.2)	✓ (-21.7)	–	–
Clock	–	–	–	–	#	#	–	✓ (−23.1)
Cry1	–	–	–	✓ (−3.0)	–	–	–	–
Cry2	–	–	–	–	✓ (−15.4)	–	–	–
Per1	–	–	#	#	✓ (−169.1)	–	–	–
Per2	–	–	✓ (−32.1)	–	✓ (−100.7)	–	✓ (114.4)	–
Nr1d1	#	#	–	–	✓ (−70.5)	–	–	–
Sirt1	–	–	–	✓ (−7.0)	–	–	–	–
Sirt3	#	#	–	–	–	–	–	–
Sirt6	#	#	–	✓ (15.0)	–	–	–	–
Nampt	–	–	–	✓ (14.5)	–	–	–	–
Cpt1a	–	–	–	–	╱	╱	–	–
Cpt1b	╱	╱	╱	╱	–	–	╱	╱
Pfkfb3	–	–	–	–	#	#	–	–
G6pd	–	–	–	✓ (−13.1)	–	–	–	–
Pgd	–	–	–	✓ (−12.3)	–	–	–	–
Nrf2	–	–	–	–	#	#	#	#

Cosine fits of the transcripts scored as rhythmic by JTK_cycle analysis were compared for differences in amplitude and phase. Significant changes in mean amplitude and phase shift differences between age-matched non-transgenic and symptomatic hSOD1^G93A mice were determined by non-overlapping 95% confidence intervals of the difference in means. The dash line (-) indicates no difference in the fitted data between non-transgenic and hSOD1^G93A mice or the lack of daily rhythmic expression in non-transgenic mice (e.g. p > 0.05 and rAMP < 0.15 as determined with MetaCyle R package). The pound sign (#) indicates loss of daily rhythmic gene expression in hSOD1^G93A mice (e.g. p > 0.05 and rAMP < 0.15 as determined with MetaCyle R package). The check mark (✓) indicates a statistically significant difference, followed by the difference in means in parenthesis. A negative difference in amplitude indicates a decrease in mean amplitude in hSOD1^G93A mice when compared to non-transgenic animals. A negative phase shift indicates that the phase in hSOD1^G93A mice occurs at a later ZT than in non-transgenic controls.

We also analyzed the expression of core clock genes in the spinal cord of hSOD1^G93A mice (Fig. 1B). Notably, all genes of the negative arm, except for Nr1d1, had decreased expression at ZT16. Furthermore, the daily rhythm of Cry1 mRNA displayed a significant phase shift (Fig. 1B and Supplemental Fig. 1). A significant decrease in Per2 amplitude was observed in the spinal cord of hSOD1^G93A mice, while Per1 showed a loss of daily rhythmic expression (Table 2). We also found significant changes in the expression of core clock genes at multiple times of the day in the gastrocnemius of early symptomatic hSOD1^G93A mice (Fig. 1C). In, the positive arm, these changes include both a decrease in amplitude and a phase shift in Bmal1 expression and the loss of daily rhythmicity in Clock expression. In the negative arm, significant changes in amplitude of Cry2, Per1, Per2, and Nr1d1 were identified (Fig. 1C and Table 2).

Although the liver is not a traditional target organ in ALS, hepatic ultrastructural changes accompanied by liver dysfunction is observed in a significant number of patients (33), and disruption of metabolic circadian rhythms may contribute to intrahepatic metabolic abnormalities. Our analysis showed that the expression of Clock and all members of the negative arm was significantly increased in the liver of hSOD1^G93A mice at various ZTs (Fig. 1D). Nr1d1 expression displayed a significant increase at all ZTs except for ZT4. Furthermore, waveform analysis revealed significant changes in the phase of Clock and the amplitude of Per2 in the liver of hSOD1^G93A mice (Table 2 and Supplemental Fig.1).

In order to extend these results to another mutant SOD1-linked ALS mouse model, we analyzed core clock gene expression at a single time point, ZT8, in 7-month-old early symptomatic hSOD1^H46R/H48Q mice (Fig. 2A). Consistent with the results in hSOD1^G93A mice, few changes were seen in the SCN of hSOD1^H46R/H48Q mice when compared to non-transgenic littermates. However, there was a significant downregulation in the expression of all but one core clock gene in the spinal cord of symptomatic hSOD1^H46R/H48Q mice. Conversely, in the gastrocnemius and liver, most core clock genes displayed a significant upregulation in symptomatic hSOD1^H46R/H48Q mice. Although with a single time point it is not possible to draw conclusions regarding differences in diurnal patterns of gene expression, the data supports the notion that changes in the expression of clock genes also occur in another mutant hSOD1-linked mouse model. Notably, no changes in the expression of clock genes were observed in mice over-expressing hSOD1^WT when compared with non-transgenic mice at ZT8 and ZT16 (Fig. 2B, C), indicating that the altered expression of clock genes observed in hSOD1^G93A and hSOD1^H46R/H48Q mice is likely linked to the expression of the mutant hSOD1.

Figure 2. Expression of core clock genes in hSOD1^H46R/H48Q and hSOD1^WT mice.

mRNA levels of core clock genes were determined by qRT-PCR in tissues harvested from age-matched non-transgenic (NonTG) and symptomatic hSOD1^H46R/H48Q (H46R/H48Q) at ZT8 (A) or NonTG and hSOD1^WT mice at ZT8 (B) and ZT16 (C). [Suprachiasmatic nucleus (SCN), spinal cord (SC), gastrocnemius (Gastroc)]. Data was normalized by Rplp0 and is presented as percentage of non-transgenic mice (mean ± SEM, * p < 0.05, n = 3–4).

Sirtuins (Sir2-like enzymes, Sirt) are NAD⁺-dependent deacylases that play a key role in transcription, DNA repair, metabolism, and oxidative stress resistance. Sirt1 and Sirt6 have been linked to circadian control of gene expression, while Sirt3 regulates circadian enzymatic activity in fuel-producing and fuel-utilizing tissues (34–37). Interestingly, in the SCN of hSOD1^G93A mice Sirt3 and Sirt6 loss diurnal rhythmicity and displayed opposite expression patterns during the dark period, when compared to non-transgenic controls (Fig. 3A). In the spinal cord of hSOD1^G93A mice, the expression patterns of Sirt1 and Sirt6 displayed phase shifts (Fig. 3B, Table 2 and Supplemental Fig.1). Furthermore, the expression levels of Sirt3 and Sirt6 was significantly different at ZT16 in all the tissues analyzed. The liver displayed significant relative expression changes in Sirt1, Sirt3 and Sirt6 expression in hSOD1^G93A mice, with an overall trend towards upregulation (Fig. 3D).

Figure 3. Altered expression of sirtuins in early symptomatic hSOD1^G93A mice.

Sirtuin mRNA levels were determined by qRT-PCR in tissues from early symptomatic hSOD1^G93A (G93A) and age-matched non-transgenic (NonTG) mice every 4 hours over a 24-hour period. A) Suprachiasmatic nucleus (SCN), B) spinal cord (SC), C) gastrocnemius (Gastroc), and D) liver. Time of harvest is shown in zeitgeber time (ZT) and for graphical representation ZT0 data was also used as ZT24. Data was normalized by Rplp0 and is presented as percentage of non-transgenic mice at ZT0 (mean ± SEM, * p < 0.05, n = 3–4).

The observed altered daily rhythm in gene expression patterns of sirtuins suggests that the expression of genes involved in metabolism and energy homeostasis could also be disrupted. Therefore, we examined the expression of Nicotinamide phophoribosyltransferase (Nampt), Carnitine palmitoyltransferase I (Cpt1), and 6-Phosphofructo-2-kinase/fructose-2,6-biphosphate 3 (Pfkfb3). In the SCN, spinal cord and liver of hSOD1^G93A mice Cpt1a and Pfkfb3 exhibited significant expression changes at certain ZTs (Fig. 4A, B, D and Table 2). No significant changes in Nampt expression were observed in the SCN or liver, but Nampt displayed a significant phase shift in the spinal cord of hSOD1^G93A mice (Fig. 4B, Table 2 and Supplemental Fig. 1). All three metabolic genes showed significant downregulation in the gastrocnemius of hSOD1^G93A mice at most ZTs (Fig. 4C). Moreover, Pfkfb3 expression loses diurnal rhythmicity in the gastrocnemius of hSOD1^G93A mice (Fig. 4C, Table 2 and Supplemental Fig. 1).

Figure 4. Altered expression of metabolic genes in early symptomatic hSOD1^G93A mice.

Nampt, Cpt1a or Cpt1b, and Pfkfb3 mRNA levels were determined by qRT-PCR in tissues from early symptomatic hSOD1^G93A (G93A) and age-matched non-transgenic (NonTG) mice every 4 hours over a 24-hour period. A) Suprachiasmatic nucleus (SCN), B) spinal cord (SC), C) gastrocnemius (Gastroc), and D) liver. Time of harvest is shown in zeitgeber time (ZT) and for graphical representation ZT0 data was also used as ZT24. Data was normalized by Rplp0 and is presented as percentage of non-transgenic mice at ZT0 (mean ± SEM, * p < 0.05, n = 3–4).

There is a strong link between circadian timekeeping and redox homeostasis (38, 39). For example, the circadian clock regulates rhythmic activation of Nrf2 (nuclear factor, erythroid 2 like 2), a master regulator of antioxidant defenses (40); while the pentose phosphate pathway regulates circadian NADPH levels (39). Thus, we analyzed the circadian expression pattern of Nrf2 and two enzymes from the pentose phosphate pathway, G6pd (glucose-6-phosphate dehydrogenase X-linked) and Pgd (phosphogluconate dehydrogenase). The most notable changes were observed in the spinal cord, where G6pd and Pgd displayed significant phase shifts in their diurnal expression patterns (Figure 5B, Table 2 and Supplemental Fig. 1). In addition, Nrf2 displayed loss of diurnal expression patterns in the both the gastrocnemius and liver of early symptomatic hSOD1^G93A mice (Fig. 5C–D). Pgd expression in the gastrocnemius does not display diurnal rhythmicity but it showed significant up-regulation in hSOD1^G93A mice at all ZTs.

Figure 5. Altered expression of genes involved in redox homeostasis in early symptomatic hSOD1^G93A mice.

G6pd, Pgd and Nrf2 mRNA levels were determined by qRT-PCR in tissues from early symptomatic hSOD1^G93A (G93A) and age-matched non-transgenic (NonTG) mice every 4 hours over a 24-hour period. A) Suprachiasmatic nucleus (SCN), B) spinal cord (SC), C) gastrocnemius (Gastroc), and D) liver. Time of harvest is shown in zeitgeber time (ZT) and for graphical representation ZT0 data was also used as ZT24. Data was normalized by Rplp0 and is presented as percentage of non-transgenic mice at ZT0 (mean ± SEM, * p < 0.05, n = 3–4).

Neuronal-astrocyte coupling plays a role in almost all physiological processes that ensure the well-being of neurons (41–43) and it has been recently shown that astrocytes play a crucial role in proper circadian timekeeping (44, 45). Moreover, astrocyte-specific deletion of clock genes is sufficient to alter energy balance, glucose homeostasis, and the level of neurotransmitters in the brain, thus affecting lifespan and metabolism in vivo (46). Since astrocytes play a key role determining motor neuron fate in ALS and to investigate potential changes in the expression of clock genes in another model of ALS, we analyzed the expression of clock genes in iPSC-derived astrocytes (iAs) from ALS patients with SOD1 or FUS mutations. (Fig. 6). Treatment of cell cultures with glucocorticoids induces synchronized oscillation of clock genes, independent from cell cycle progression (47). iAs from controls and ALS patients were synchronized with 100nM dexamethasone (1-hour treatment) and 24 hours later cultures were harvested every 4 hours over a 24-hour period. Each ALS line was differentiated, treated, and analyzed in parallel with a control line. Similar to hSOD1^G93A mice, both ALS lines displayed altered expression of core clock genes (in absolute value) at multiple times when compared to controls (Fig. 6). The most notable change in rhythmicity was observed in the expression of Bmal1 and Nr1d1 in both lines of patient-derived astrocytes (Table 3).

Figure 6. Altered expression of clock genes in synchronized iPSC-derived astrocytes (iAs) from ALS patients.

Astrocytes differentiated from induced pluripotent stem cells (iPSCs) from control and ALS patients were synchronized with 100nM of dexamethasone and after 24 hours, harvested every 4 hours over a 24-hour period. mRNA levels of core clock genes were determined by qRT-PCR. A) Expression of core clock genes in a non-ALS control and SOD1^G86R iAs. B) Expression of core clock genes in a non-ALS control and FUS^G522A iAs. Data was normalized by Rplp0 and is presented as percentage of their respective control line at CT0 (mean ± SEM, * p < 0.05, n = 3–4/ct).

Table 3.

Waveform analysis and statistical comparison of the expression of clock genes in patient-derived induced astrocytes.

	Control 1 & SOD1G86R		Control 2 & FUSG522A
	Amplitude (% change)	Phase Shift (hrs)	Amplitude (% change)	Phase Shift (hrs)
Bmal1	#	#	#	#
Clock	–	–	–	–
Cry1	–	–	–	–
Cry2	–	–	–	✓ (−6.0)
Per1	–	–	–	–
Per2	–	–	–	–
Nr1d1	–	✓ (−3.3)	#	#

Cosine fits of the transcripts scored as rhythmic by JTK_cycle analysis were compared for differences in amplitude and phase. Significant changes in mean amplitude and phase shift differences between control and ALS-lines were determined by non-overlapping 95% confidence intervals of the difference in means. The dash line (-) indicates no difference in the fitted data between the control and ALS-line or the lack of rhythmic expression in the non-ALS control line (e.g. p > 0.05 and rAMP < 0.15 as determined with MetaCyle R package). The pound sign (#) indicates loss of daily rhythmic gene expression in the ALS-line (e.g. p > 0.05 and rAMP < 0.15 as determined with MetaCyle R package). The check mark (✓) indicates a statistically significant difference, followed by the difference in means in parenthesis. A negative phase shift indicates that the phase in the ALS-line mice occurs at a later time than in non-ALS controls.

Discussion

The SCN is the dominant circadian pacemaker driving behavioral rhythms and synchronization of peripheral clocks (48). Overall, our data reflects fewer and less pronounced differences in the expression of clock and clock-controlled genes in the SCN of early symptomatic hSOD1^G93A mice when compared to age-matched non-transgenic controls. This result may in part explain why a previous study observed no changes in circadian activity patterns in this ALS mouse model (49). However, analysis of the spinal cord and peripheral tissues (gastrocnemius and liver), showed numerous significant alterations in the expression of clock and clock-controlled genes in early symptomatic hSOD1^G93A mice. Our data raises the possibility of disrupted cross-talk between the SCN and peripheral tissues in hSOD1^G93A mice, preventing proper peripheral clock regulation and synchronization. Since metabolism and redox homeostasis are intimately entangled with circadian rhythms, our data suggest that altered expression of clock genes may contribute to the metabolic and redox impairment observed in ALS patients and mouse models (12, 50–62).

Although we analyzed the expression of the same set of genes across four different tissues during a 24-hour period, not every gene displayed true circadian rhythmicity across all tissues. In mice, different tissues maintain different sets of circadian expressed genes, with a relatively low overlap of common genes in different tissues (63, 64). Circadian information encoded by the core clock genes is spread at a tissue-specific level through the expression of specific sub-sets of clock-controlled genes which regulate rhythmic functions. For those genes that did not display significant daily rhythms in the non-transgenic control tissue, no assumption of circadian dysregulation in the hSOD1^G93A tissue can be made (Table 2, dash line). Thus, it is important to note that depending on the gene and tissue considered, some of the changes observed are changes in daily expression rhythms while others are relative expression changes at specific time points. For example, Pgd mRNA expression displayed altered daily rhythms in the spinal cord while Pgd mRNA does not display rhythmicity in the gastrocnemius but it is significantly up-regulated at all time point in hSOD1^G93A mice (Fig.5B and C).

In addition to its role in the molecular clock, Nr1d1, is also a known repressor of metabolic and inflammatory genes (65). Therefore, Nr1d1 can be considered as the link that coordinates the communication between circadian rhythm, metabolism, and inflammation. Interestingly, Nr1d1 was the only clock gene that displayed significant expression changes in all four tissue-types analyzed and its changes could be linked to altered daily rhythms in the expression of metabolic genes.

The transcription of the rate-limiting enzyme in the NAD⁺ salvage pathway, Nampt, is directly under circadian control, causing NAD⁺ levels to display circadian oscillation (34, 35). Thus, the activity of the sirtuins is inherently circadian based on their dependence on NAD⁺ as co-substrate and they appear to play a key role in the interplay between cellular metabolism and circadian rhythms. For example, Sirt1 associates with Clock and is recruited to the Clock:Bmal1 chromatin complex at circadian promoters (34, 35), completing a feedback loop that may coordinate daily rhythms with cellular energy levels. All major mitochondrial processes, such as tricarboxylic acid cycle, fatty acid metabolism, oxidative phosphorylation, mitochondrial acetyl-CoA production and antioxidant response, are regulated by acetylation. Thus, circadian oscillation of NAD⁺ levels also controls mitochondrial function through Sirt3 activation (22). Sirt6 also interacts with Bmal1:Clock and appears to mediate their recruitment to a set of circadian promoters different from Sirt1 (36). In addition, Sirt6 activity regulates glycolysis and mitochondrial respiration (66, 67). Moreover, Sirt6 can regulate inflammatory and antioxidant responses by regulating the recruitment and activation of NF-κB and Nrf2 (30, 66, 68, 69). Thus, altered daily rhythms in gene expression of Nampt, Sirt1, Sirt3 and Sirt6 in the CNS of early symptomatic hSOD1^G93A mice could contribute to both circadian clock dysfunction as well as altered metabolism and elevated oxidative stress. Importantly, altered expression of SIRT6 and enzymes involved in NAD⁺ synthesis have been described in ALS patients, suggesting a possible role of these changes in the disease (70).

We also present direct evidence of dysregulated gene expression of two key metabolic enzymes, Pfkfb3 and Cpt1. Pfkfb3 controls both the synthesis and degradation of fructose-2,6-bisphosphate (F2,6BP), a regulatory molecule that controls glycolysis in eukaryotes (71). F2,6BP activates glycolysis through allosteric modulation of phosphofructokinase (Pfkm). Therefore, it is possible that altered Pfkfb3 expression could contribute to disrupted glycolysis and gluconeogenesis in the spinal cord and gastrocnemius muscle of hSOD1^G93A mice. Cpt1 is the rate-limiting enzyme in the utilization of long-chain fatty acids for beta-oxidation in the mitochondria (72). Decreased Cpt1 levels result in reduced energy production. Interestingly, we found significant down-regulation in the relative expression of Cpt1 in both tissues directly affected by mutant hSOD1 over-expression, i.e., spinal cord and gastrocnemius muscle (Fig. 3B, C), pointing towards the potential involvement of these changes in the metabolic dysregulation observed in this ALS model. Moreover, treatment with an NAD⁺ precursor is neuroprotective in hSOD1^G93A mice, and this treatment partially corrects the changes in expression of both these enzymes observed in the gastrocnemius muscle of hSOD1^G93A mice (70).

Activation of Nrf2 is significantly increased in the spinal cord and muscle of mutant hSOD1 animal models (51, 73). Recently it has been suggested that Nrf2 participates in an additional interlocking loop that integrates cellular redox signaling and inflammation into circadian timekeeping (74, 75). We observed increased expression of Nrf2 mRNA at several ZTs in the spinal cord and loss of diurnal expression patterns in the gastrocnemius and liver of early symptomatic hSOD1^G93A mice. Furthermore, we identified significant changes in the relative expression levels, as well as significant changes in the phases, of both G6pd and Pgd in the spinal cord of hSOD1^G93A mice. We also observed significant up-regulation of the relative expression of G6pd at several ZTs in the liver and of Pgd at all ZTs in the gastrocnemius. Thus, it is possible that the altered expression levels of these important redox regulators could impair an appropriate antioxidant and inflammatory response during the pathological process.

The relevance that circadian regulation appears to have on metabolism and antioxidant defenses, together with the major contribution of astrocytes towards the regulation of these components in the CNS, prompted us to investigate the response of iPSC-derived astrocytes from ALS patients to a phase-setting cue. With the exception of PER2 expression, the changes of core clock gene expression have similar trends in both ALS lines when compared to the controls. Notably, NR1D1 displays significant altered rhythmicity in both ALS lines. As mentioned earlier, this change could potentially have decisive consequences in the metabolism and inflammatory response of ALS astrocytes.

Although it is difficult to precisely extrapolate in vitro data to the whole CNS, our data does demonstrate significant relative expression changes of several clock genes in synchronized ALS astrocytes, which could suggest that ALS astrocytes are unable to effectively respond to phase-setting cues. Abnormal astrocytic synchronization could result in uncoordinated metabolic and antioxidant defenses which, subsequently, could impair optimal neuronal requirements during circadian oscillations.

Taken together our data demonstrates altered expression of core clock and clock-controlled genes in mutant hSOD1-linked and non-hSOD1 ALS models. Since these changes were observed in symptomatic mice (i.e., after ongoing motor neuron degeneration), they may be a consequence of the degenerative process. In this regard, it is worth noting that previous research has found that daily body temperature and corticosterone circadian rhythm becomes dysregulated after symptoms onset in hSOD1^G93A mice (17, 49). Due to the cyclical and intricate relationship between circadian rhythm, metabolism and inflammation, further research is needed to establish whether this dysregulation is a cause or a consequence of the degenerative process. Nevertheless, the altered expression of many genes observed across multiple ALS models highlights the importance of better understanding the role of circadian rhythm in the pathophysiology of ALS.

Abbreviations

ALS

amyotrophic lateral sclerosis

Bmal1

brain and muscle ARNT-like 1

Clock

circadian locomotor output cycles protein kaput

Cry

cryptochrome

FUS

FUS RNA binding protein

Nr1d1

nuclear receptor subfamily 1 Group D member 1

Per

period

Rorα

RAR-related orphan receptor α

SCN

suprachiasmatic nucleus

SOD1

superoxide dismutase 1