Parkinson’s disease (PD) is characterized by progressive loss of dopaminergic neurons in the substantia nigra and Lewy bodies’ formation in the midbrain. These pathological changes result in motor symptoms. Braak’s staging theory [1] and the presence of multiple non-motor symptoms reflect the complex and multifaceted nature of PD [2]. Non-motor symptoms are often resistant to L-DOPA treatment and will inevitably become a significant factor affecting patients’ function and quality of life in advanced PD stages. Therefore, further exploration of other pathophysiological mechanisms involved in these symptoms holds promise for other effective therapeutic strategies. During the last decade, increasing attention has been paid to the link between circadian rhythms and PD. It has been recognized that the circadian rhythm is disrupted in patients with PD [3]. First, circadian function biomarkers, such as body temperature, blood pressure, and hormone secretion, are dysregulated in PD [4, 5]. Second, patients in advanced stages lose normal motor activity, and actigraphy studies in patients with PD reported flattened diurnal motor activity patterns [6]. Third, patients receiving repeated L-DOPA administration often report a lower response to L-DOPA in the evening [7]. Fourth, common non-motor symptoms of PD, such as sleep-wake disturbances, depression, and visual symptoms, are modulated by circadian rhythms [8, 9].

A large sample cohort study of community-dwelling older men without PD reported that reduced circadian rhythmicity at baseline is associated with an increased risk of developing PD during an 11-year follow-up, suggesting that circadian disruption rhythms might be prodromal to PD [10]. Evidence indicates that the electrical activity of neurons within the master pacemaker of the circadian system, the suprachiasmatic nuclei (SCN), is disrupted at the onset of motor symptoms [11]. Therefore, circadian dysfunction may be a key cause of non-motor symptoms, especially for pre-motor symptoms. Sleep disturbances are common non-motor symptoms reported by 64% of patients with PD [12], including insomnia, rapid eye movement sleep behavior disorder, restless legs syndrome, sleep-disordered breathing, and excessive daytime sleepiness. Patients often demonstrate sleep alterations during the prodromal phase of PD [13]. It has been hypothesized that circadian and sleep changes may trigger neurodegeneration in the earliest phases of the disease [14]. Therefore, it is important to understand and improve sleep disturbances and circadian rhythm in patients with PD.

Even though there are reciprocal interactions between the dopaminergic system and circadian rhythms, the underlying mechanisms of circadian dysfunction in PD remain elusive. Circadian rhythms are regulated by circadian gene networks consisting of a series of clock genes (such as CLOCK, BMAL1, BMAL2, PER1, PER2, PER3, CRY1, and CRY2) and clock-controlled genes (such as TEF, DEC1, DEC2, and TIMELESS) [15]. Several studies have explored desynchronization at the molecular level. A study examining the expression profile of several principal circadian genes in peripheral leukocytes in whole blood for a 12-hour nighttime period found that the mRNA expression of BMAL1 and BMAL2 was significantly reduced in patients with PD in the evening. Furthermore, the level of BMAL2 correlated with motor severity and sleep quality in the morning [16, 17]. Cai and his colleagues [18] discovered that genetic polymorphisms in ARNTL and PER1 were significantly associated with PD risk. These reports suggest that circadian rhythm disruption is a result and a cause of the degeneration process. In this case, it would be expected to occur early in the disease’s course (or even preceding disease onset) and exacerbate disease progression. There has been no longitudinal study focusing on the effect of circadian genes on the progression of PD.

The thyrotroph embryonic factor (TEF) gene is one of the downstream genes in the circadian gene network and expresses a broad range of cells and tissues. It encodes a thyrotroph embryonic factor, together with albumin D box-binding protein (DBP), human hepatic leukemia factor (HLF) are three members of the proline and acidic amino-acid-rich basic leucine zipper (PAR bZip) transcription factor family. Different members of the subfamily can readily form heterodimers and share DNA-binding and transcriptional regulatory properties. The thyrotroph embryonic factor can bind to proline and acidic amino-acid-rich response elements (PARRE) in promoters and elevate the transcription levels of light-induced circadian genes NR1D1, NR1D2, CRY1a, CRY2b, and PER2 [19]. The TEF rs738499 polymorphism has been suggested to be associated with depression and sleep disturbance in PD in previous cross-sectional studies [20, 21]. The present study was conducted longitudinally in a cohort of patients with PD to investigate whether TEF is associated with worsening PD patients’ clinical symptoms with an average follow-up period of 3.3 years.

The distribution of genotypes was fitted with the Hardy-Weinberg equilibrium in the PD sample. The mean age of patients with PD was 58.7 $\pm$ 11.3 years, 58.8% were men. The mean duration at baseline was 4.5 $\pm$ 3.5 years and the mean follow-up period was 3.3 $\pm$ 1.1 years. The demographics and clinical characteristics at baseline showed no significant difference between genotypes (TT vs. TG + GG). All clinical symptoms worsened significantly during the follow-up period (Table 1). A faster annual decline rate of the PDSS score (P = 0.019) and progression rate of the H-Y stage (P = 0.016) was found in carriers of the TT genotypes of TEF rs738499 (Table 2).

Spearman’s rank correlation showed that the annual decline rate of the PDSS score was associated with the PDSS score at baseline, the annual progression rates of HAMA and NMSQ during follow-up, and the TT genotype of TEF rs738499. The H-Y stage’s progression rate was associated with the H-Y stage, tremor score, rigidity score at baseline, and the annual progression rates of HAMD, HAMA, NMSQ, ADL, MMSE, tremor, rigidity, akinesia, axial symptoms, and the TT genotype of TEF rs738499. Table 3 displays all related factors with P 0.2.

Stepwise linear regression showed that the progression rate of anxiety, the initial severity of sleep disturbance and axial symptoms, and the TT genotype of TEF rs738499 were independent risk factors affecting the decline rate of PDSS score. These four predictors together could explain the 40.5% variation in the decline rate of PDSS. TEF rs738499 alone accounted for 3.1% of the variance (P = 0.030) (Table 4). Meanwhile, stepwise linear regression revealed that the progression rate of axial symptoms, the TT genotype of TEF rs738499, and the initial severity of anxiety were independent risk factors affecting the H-Y progression rate stage. These three predictors together could explain the 49.1% variation in the progression rate of the H-Y stage. TEF rs738499 alone accounted for 8.3% variance (P = 0.021) (Table 5).

The present research indicates that the TT genotype in TEF rs738499 correlated with a faster deterioration rate of sleep quality and H-Y stage of PD patients. Moreover, patients with more severe sleep disturbance and axial symptoms at baseline and faster worsening rate of anxiety during the follow-up period presented a faster deterioration rate of sleep quality, and patients with more severe anxiety at baseline and faster worsening rate of the axial symptoms during the follow-up period experienced a quicker progression rate of the H-Y stage.

The sleep-wake cycle is under the dual control of homeostatic and circadian processes. Homeostatic processes are responsible for the sleep rebound following sleep deprivation, and the circadian processes determine the timing of the sleep period via the circadian clock in the SCN. However, little is known about the circadian system’s molecular mechanism in patients’ sleep problems with PD. Several neurotransmitters with circadian rhythmicity, such as dopamine, melatonin, 5-hydroxytryptamine (5-HT), glutamate, and gamma-aminobutyric acid (GABA), have been reported to be implicated in the regulation of sleep-wake behavior [31, 32, 33]. David et al. [34] found that reduced circulating melatonin levels were associated with reduced slow-wave sleep and reduced rapid eye movement (REM) sleep.

Furthermore, a lower melatonin onset concentration was associated with increased sleep latency in patients with PD. TEF is involved in maintaining neurotransmitter homeostasis (such as 5-HT and dopamine) [35]. We speculate that the association of TEF with the rapid deterioration of sleep quality and H-Y stage might be related to its above-described function, but further studies are needed to provide further evidence.

Our results show that anxiety and axial symptoms were closely related to worse sleep quality and motor staging. The etiology of sleep disturbances in PD is multifactorial and previous studies have shown that anxiety symptoms are a risk factor for sleep quality and motor complications [36, 37]. Axial nocturnal hypokinesia in PD manifests fewer turning-over episodes, turns with smaller degrees, less velocity, and acceleration. Patients with severe nocturnal hypokinesia usually need to take a higher total and bedtime levodopa equivalent dose, and drinking more water at night could result in frequent nocturia that compels the patients to get out of bed, resulting in lower sleep quality [38, 39]. The onset of falls is a warning symptom of H-Y stage aggravation. Spanish researchers [40] employed an objective technology for rigidity assessment. They reported that axial rigidity was related to the risk of falls in patients with PD. Axial symptoms contribute to the postural instability and gait difficulty (PIGD) subtype of PD. The PROPARK study [41] reported that PIGD-dominant patients had more severe motor and non-motor symptoms, a more rapid disease progression, and a poorer prognosis PD compared to the tremor-dominant (TD) patients. Another circadian rhythm-related study [42] showed that agomelatine, acting as a melatonin receptor (MT1 and MT2) agonist and as a serotonin receptor (5-HT2b and 5-HT2c) antagonist, improves not only depressive symptoms in PD but also improves sleep problems and motor symptoms. This confirmed the intimate internal relations of these three clinical phenotypes. Therefore, much closer relationships between sleep and mood disorders and motor symptoms may have a specific impact on the genotypic association.

The longitudinal follow-up is the main strength of this study, and a comprehensive follow-up evaluation of multiple non-motor symptoms of PD was carried out. There are several limitations to this study. First, the sample size was small, and no control group to perform gene-disease association analysis was included. We genotyped the TEF rs738499 polymorphism in 480 PD patients and 500 controls in a previous study. We found that the T allele and TT genotype were significantly higher in patients with PD, but this result was not shown in that study [20]. Second, the participants were not assessed during the same time of the day or the same season for baseline, and follow-up visits as both factors will influence motor and non-motor scores. Therefore, case-control studies with a larger sample size and more strict control of various factors affecting circadian rhythm are needed to explore the role of circadian rhythm in different clinical stages of PD.

This longitudinal study aimed to determine the association between a circadian gene’s polymorphism and PD symptoms progression. It also revealed that TEF might be considered a predictive marker of the progression of sleep disturbances and motor staging in PD. This advances our understanding of the genetic associations in PD and suggests circadian dysfunction may be involved in Parkinson’s disease’s pathophysiology.

ADL, Activities of Daily Living; APR, annual progression rate; ARNTL, Aryl Hydrocarbon Receptor Nuclear Translocator Like (another name of BMAL1 gene); BMAL, brain and muscle ARNT-like protein; CLOCK, circadian locomotor output cycles kaput; CRY, cryptochrome; DEC, Differentiated embryonic chondrocyte; HAMA, 14-item Hamilton rating scale for Anxiety; HAMD, 24-item Hamilton rating scale for depression; H-Y, stage Hoehn-Yahr stage; LEDD, levodopa equivalent daily doses; MMSE, Mini-Mental State Examination; NMSQ, Parkinson’s Disease Non-Motor Symptom Quest; NPAS2, neuronal PAS domain protein 2; NR1D1, Nuclear Receptor Subfamily 1 Group D Member 1; PD, Parkinson’s disease; PDSS, Parkinson’s Disease Seep Scale; PER, period; TEF, Thyrotroph embryonic factor; TIMELESS, Timeless Circadian Regulator; UPDRS-III, Unified Parkinson’s Disease Rating Scale Part-III.

Study concept and design: Hua and Liu; Acquisition of data: Liu, Xu, Yu, Yao and Cui; Analysis and interpretation of data: Hua and Cui; Drafting of the manuscript: Hua; Critical revision of the manuscript for important intellectual content: Liu; Obtained funding: Liu; Administrative, technical, and material support: Chen.

The study was approved by the Ethics Committee of Nanjing Medical University. Informed consent was obtained from each participant.

The authors thank for Sheng-Han Kuo’s guidance in manuscript writing, and thank all the patients participated in this study.

The study was supported by the National Natural Science (81571348), Jiangsu Natural Science Foundation (BK20151077), National key research and development plan (2016YFC1306600, 2017YFC1310302).

The authors declare no competing interests.