Sleep and Circadian Disruption and the Gut Microbiome-Possible Links to Dysregulated Metabolism

Abstract

Insufficient sleep and circadian misalignment are associated with adverse metabolic health outcomes. Alterations in gut microbial diversity occur with insufficient sleep and circadian misalignment, which can lead to modifications in microbial structure and function. Changes in microbially produced and modified metabolites such as short chain fatty acids and secondary bile acids may contribute to chronic inflammation, positive energy balance and endocrine changes, and represent potential mechanisms linking insufficient sleep and circadian misalignment with metabolic dysregulation. Literature primarily from the last two years is reviewed here, examining the impact of sleep and circadian rhythms and their disruption on the gut microbiome in human and non-human models, with an emphasis on the hypothesis that the altered gut microbiome may be one pathway by which insufficient sleep and circadian misalignment dysregulate metabolism.

Globally, overweight and obesity has increased over the past four decades leading to an obesity pandemic and increased cardiometabolic disease [1–4]. Sleep and circadian rhythms impact most aspects of human physiology and behavior, and insufficient sleep and circadian misalignment contribute to cardiometabolic dysregulation including positive energy balance, weight gain, impaired insulin sensitivity, and inflammation [5–11]. Insufficient sleep is common, with over 35% of adults in the United States reporting less than the recommended 7h sleep per night [12–14]. Circadian misalignment is associated with shift work and social jetlag, which are also prevalent with ~20% of the workforce being shift workers [15] and ~ 33% of the population experiencing two or more hours of weekly social jetlag [16–17] (i.e., later sleep and circadian time on weekends than weekdays). Circadian misalignment often ensues when behaviors occur at an incorrect internal biological time [18]. Examples include sleeping during the biological day (i.e., when levels of the circadian hormone melatonin are low) or eating during the biological night (i.e., when levels of melatonin are high) [5–6]. Collectively, findings suggest that insufficient sleep and circadian misalignment are risk factors for weight gain and related cardiometabolic disorders [9] [19]. However, our understanding of the pathways by which insufficient sleep and circadian misalignment influence cardiometabolic disease risk remains incompletely understood.

Emerging findings on the microbiome afford a novel insight into potential mechanisms linking insufficient sleep, circadian misalignment, and metabolic dysregulation. The gut microbiome consists of trillions of microbes with rich genetic diversity [20]. When alterations of the composition of the gut microbes result in an imbalance in harmful and protective microbes, this results in gut dysbiosis [21]. Many pathological states that arise from insufficient sleep and circadian misalignment also occur in states of gut dysbiosis, such as cardiometabolic disorders (obesity, type II diabetes, heart disease) [22–26], cognitive impairment [27–28], and other proinflammatory and neuro-behavioral disorders such as multiple sclerosis [29], Alzheimer’s disease [30], depression [31] and anxiety [27] [32]. Altered gut microbiome profiles and microbial modified metabolites during insufficient sleep and circadian misalignment may be mechanisms linking insufficient sleep, circadian misalignment, and metabolic dysregulation [33–34]. Here we review the most recent literature regarding this possible mechanism outlined in Figure 1.

Figure 1. Proposed model linking insufficient sleep, circadian misalignment and gut dysbiosis to cardiometabolic disease.

The gut-brain-axis bi-directionally links gut dysbiosis to sleep and circadian misalignment. Gut dysbiosis can alter microbially modified metabolites like short chain fatty acids and secondary bile acids which may contribute to chronic inflammation, changes in energy absorption and endocrine dysregulation. Insufficient sleep and circadian misalignment lead to alterations in behavior such as the timing and amount of food intake as well as endocrine dysregulation and chronic inflammation. Positive energy balance, endocrine dysregulation and chronic inflammation may contribute to weight gain, insulin resistance and metabolic syndrome which are risk factors for cardiometabolic disease.

Metabolism and the Gut Microbiome

Findings from human and animal studies suggest a role for the gut microbiome in host metabolism as alterations in specific microbes have been observed in many metabolic disorders [35–37]. The gut microbial community is quantified using a variety of approaches, including alpha and beta diversity, abundances of microbes, and the ratio of Firmicutes:Bacteriodetes (F:B), two of the most highly prevalent phyla in the gut microbiome. Alpha diversity quantifies the richness and evenness of microbes within a single sample, whereas beta diversity quantifies the differences between samples [38]. Typically, findings in humans show decreases in microbial alpha and beta diversity are associated with obesity [37] [39]. The F:B ratio has been shown to be positively associated with Body Mass Index (BMI) [40], whereas diet induced weight loss has been shown to decrease the F:B ratio [41]. Other findings have associated a higher F:B ratio in obese individuals with metabolic syndrome versus obese people without metabolic syndrome [42].

More recently, links between functional outputs of specific gut microbes and metabolic health has been demonstrated. For example, short chain fatty acids (SCFA) produced by gut microbes in the large intestine [43] are known to mediate changes in inflammatory processes and improve gut integrity, energy homeostasis and glucose metabolism [44–48]. The primary short chain fatty acids in the intestinal tract include acetate, butyrate and propionate and these are produced by a variety of bacteria [43] [45]. Acetate and butyrate are involved in lipid biosynthesis and propionate is involved in glucose metabolism via gluconeogenesis in the liver [44][49]. In general, ~5% of SCFAs are excreted in feces [50], whereas higher fecal concentrations of SCFAs have been associated with systemic inflammation, gut permeability, obesity and hypertension [39]. Interestingly, fecal SCFA concentrations in both humans and mice models have been found to demonstrate time of day variation and are influenced by behavioral inputs such as food intake [51–53]. Findings from studies in mice have shown that high fat diet feeding eliminates the diurnal oscillations observed in fecal butyrate and propionate concentrations [54]. Additionally, in antibiotic induced microbially depleted mice, oral gavage of SCFAs at ZT5 has been shown to advance the peak of peripheral PER2::LUC phase in the kidney, liver and submandibular gland [55]. Because findings from Tahara and colleagues [55] suggest the effect of SCFAs on the circadian clock are mediated through an indirect mechanism, further work is required to characterize the underlying mechanism/s by which SCFAs may influence the circadian clock.

Another class of microbially derived molecules that influence host metabolic physiology are bile acids. Primary bile acids (BA) are produced from cholesterol in the liver and secreted into the gut in response to food intake by the host to aid in lipid absorption. Conversion to secondary bile acids is completed by bacteria in the gut [56]. Bile acids can impact metabolic physiology by acting as ligands for the farnesoid x receptor (FXR) and G protein coupled receptor TGR5 which influence lipogenesis, glucose homeostasis and energy metabolism [56–57]. Findings from mice with ablated circadian rhythm genes Per1 and Per2 showed increased serum and hepatic bile acid levels [58], and oral administration of unconjugated bile acids has been shown to alter expression of clock genes in the colon, ileum, and liver [59]. It is possible that dysregulation of these bile acids may contribute to host metabolic dysregulation. This review focuses on the above identified microbially modified SCFAs and BAs in line with the recently published literature in the field. Note that other non-microbial modified metabolites implicated in the complex relationship between sleep, circadian rhythms and alterations in metabolism are beyond the scope of this short review.

Sleep and the Gut Microbiome

Findings from initial controlled laboratory studies of insufficient sleep and changes in the human gut microbiome community structure have been inconsistent. Specifically, two days with 4.25h sleep opportunities [60] and five days with 4h sleep opportunities [61] did not alter beta diversity; whereas two days with 4.25h sleep opportunities increased the F:B ratio and increased relative abundance of families Coriobacteriaceae and Erysipelotrichaceae and decreased Tenericutes in one of the studies [60]. These initial findings from humans generally differ from initial findings using rodent models. For example, seven days with 4h sleep opportunities in rats decreased beta diversity [61] and four weeks with daily sleep fragmentation in mice altered beta diversity and increased the relative abundance of families Lachnospiraceae and Ruminococcaceae, and decreased Lactobacillaceae [62]. Additionally, findings show transfer of fecal material from sleep-fragmented mice to germ free mice led to increased food intake and higher inflammatory markers, as well as reduced whole-body and adipose insulin sensitivity [62]. These initial findings in non-humans and more recently published findings (Table 1) support that insufficient sleep alters the microbiota with potential implications for dysregulated metabolism.

Table 1.

Summary of recent (2018–2020) findings from studies examining insufficient sleep and gut microbiome.

Study (first author, year)	Host	Age, BMI (human studies)	N, Sex Information	Location (Country)	Protocol Type	Protocol Summaries	Primary Results Related to Gut Microbiome
Smith, 2019 [63]	Human	~22 years; BMI 25	26 Males	USA	Cross-sectional field based	Baseline appointment between 2–4pm; fecal sample collected within 12 hours; 30 days of actigraphy for behavioral sleep assessment after sampling; food intake not reported.	Sleep efficiency (SE) positively associated with alpha diversity. SE negatively associated with taxa from the Lachnospiraceae family, Blautia, Coprococcus and Oribacterium
Grosicki, 2020 [64]	Human	~30 years; BMI 24.7	17 Males, 11 Females	USA	Cross-sectional field based	Pittsburg Sleep Quality Index (PSQI) for self-reported sleep assessment over the past month (PSQI >5 poor sleep); fecal sample collected at home; food intake not reported.	Lower PSQI score (better sleep quality) positively associated with alpha diversity, and F:B ratio, Blautia and Ruminococcus; negative association with Prevotella
Bowers, 2020 [65]	C57BL/6N mice	7 weeks	Experiment 1, N = 7 Males; Experiment 2, N = 20 Males	USA	Laboratory	5 days of 20h EEG verified sleep disruption using a slowly rotating bar; fecal samples collected between ZT8 and ZT12; ad-libitum food.	Sleep disruption altered beta diversity compared to controls at day 2 and 4 post-sleep disruption; sleep disruption did not alter alpha diversity, but increased F:B ratio on day 2 of recovery sleep and increased bacteria in the class Clostridia. Sleep disruption altered the fecal metabolome, including microbially modified metabolites such as bile acids
El Aidy, 2019 [66]	C57BL/6J mice	9 weeks	N = 13 Males, 7 controls, 6 sleep deprivation	Netherlands	Laboratory	One day of 5h sleep deprivation starting at lights on using shaking/tapping of cage (ZT0-ZT5); fecal samples collected immediately after the 5h; Sleep outcomes not reported; ad-libitum food.	Sleep deprivation did not alter alpha or beta diversity
Triplett, 2020 [67]	Specific Pathogen Free Sprague Dawley rats	7 weeks	Acute sleep fragmentation (SF), n = 9, controls n =10. Chronic SF, n = 10, controls n = 12	USA	Laboratory	6 days (acute) SF using a rotating bar every 3 mins for 24h/day and 6 weeks (chronic) SF with same protocol but a 3h rest period/day; Microbiome sample collected from different intestinal regions.; sleep outcomes not reported; ad-libitum food.	6 days and 6 weeks of SF increased alpha and beta diversity compared to controls. Alpha diversity in distal ileum was significantly distinct from species in the cecum and proximal colon in both chronic and acute SF conditions. Generally, SF increased Ruminococcacea, decreased Lactobacillaceae. Lachnospiraceae; Alterations after SF were predominant in the distal ileum.
Maki, 2020 [68]	Wistar-Kyoto rats	8–10 weeks	N = 15 Males, n = 8 SF, n= 7 controls	USA	Laboratory	28 days of 8h sleep fragmentation in the light phase per day; fecal samples collected at baseline and days 3, 6, 9, 13, 20 and 27; EEG verified SF but 24h sleep amount not reported; ad-libitum food.	Days 6–13 of SF decreased alpha diversity, decreased the F:B ratio, reduced butyrate producing bacteria including Oscillospira, and increased proteobacteria compared to control rats. Recovery sleep increased putative acetate and butyrate -producing bacteria
Ma, 2019 [69]	Wistar rats	Age not reported	N = 20 Males, n = 10 REM sleep deprivation, n = 10 controls	China	Laboratory	7 days of REM sleep deprivation using modified multiple platform over water tank; fecal samples collected after 7th day; sleep outcomes not reported; ad-libitum food.	“REM sleep deprivation” decreased alpha diversity, altered multiple taxa including increased proteobacteria, Oscillospira and Ruminococcus compared to controls. REM sleep deprivation increased systemic inflammatory markers as well as increased urinary metabolites and related biochemical pathways linked to metabolic disease compared to controls.

PSQI = Pittsburg Sleep Quality Index, SE = Sleep efficiency, SF = Sleep fragmentation.

Opposite to initial findings from studies on insufficient sleep [60–61], findings from recent observational studies on sleep health in humans suggest that better sleep health is associated with increased gut microbial diversity [63–64]. Interestingly, better self-reported sleep quality was positively associated with Blautia and Ruminococcus [64], whereas actigraphy measured sleep efficiency, a behavioral measure related to sleep quality, was negatively associated with Blautia, Coprococcus and Oribacterium [63]. Individuals who reported better subjective sleep quality showed a higher F:B ratio [64], which is in contrast to findings from a controlled-laboratory study that showed experimentally induced insufficient sleep increased the F:B ratio [60]. Thus, beyond findings for diversity measures, recent findings are inconsistent with regards to associations between specific microbes and different measures of sleep health. Findings from recent studies of insufficient sleep in non-human animal models have shown either increased [67], decreased [68–69] or no change in alpha diversity [65–66]. Five days of 20h sleep disruption with two days of recovery sleep increased the F:B ratio [65], similar to findings from Poroyko and colleagues [62], whereas twenty-eight days of 8h sleep fragmentation decreased the F:B ratio [68]. While alterations in metabolic outcomes were not directly investigated in most of the studies in Table 1, seven days of REM sleep deprivation in rats was reported to alter pathways related to metabolic disease [69]. Microbes reported to change in studies of non-human animals are consistent with some of the reported findings in studies of humans; including changes in taxa within the class Clostridia including microbes from the family Lachnospiraceae and Ruminococcus that include known producers of SCFAs [45], (e.g.,butyrate production [70]). Oral gavage of the SCFA butyrate decreased time awake by increasing NREM sleep and decreasing REM sleep in mice. The mechanism of action of these changes in sleep in response to SCFA administration may be related to concurrent changes in body temperature seen in the mice [71]. Alterations in gut microbiota SCFA production during insufficient sleep may represent a possible link between insufficient sleep and dysregulated metabolism.

Circadian Misalignment and the Gut Microbiome

Findings from initial studies have shown daily rhythms in the gut microbiome community structure that appear to be associated with the timing of food intake and fasting [72–73]. Environmental [73–75] and genetic [73, 76–77] disruption of circadian rhythms, which also lead to alterations in food intake in rodents, can change the gut microbial community structure. Furthermore, transplantation of fecal material from jet lagged humans into germ free mice resulted in increased adiposity and reduced glucose tolerance compared to germ free mice receiving transplants from non-jet lagged humans [73]. Thus, there is evidence that circadian disruption may change the microbiome and contribute to metabolic dysregulation (Table 2).

Table 2.

Summary of recent (2018–2020) studies examining circadian misalignment and gut microbiome findings.

Study (first author, year)	Host	Age BMI (Human Studies)	N, Sex Information	Location (Country)	Protocol Type	Protocol Summaries	Primary Results Related to Gut Microbiome
Liu, 2020 [78]	Humans	~25 years BMI not reported	N = 22, 14 Males, 8 Females	China	Field based	One night of delayed bedtime 2–4 hours, and no scheduled wake time the next morning, one night of recovery sleep; fecal sample at baseline, after delayed bedtime and after two days of return typical schedule; no sleep or circadian assessments; food intake not reported.	Delay in bedtime did not alter alpha or beta diversity compared to baseline. F:B ratio and phyla Tenericutes were increased on the day after the delayed bedtime. Delayed bedtime altered multiple pathways including one which increased the acetyl-CoA fermentation to butanoate II pathway
Mortaş, 2020 [79]	Humans	25–40 years BMI >18.5 and <29.5	N = 10 Males	Turkey	Field based	Shift workers working days for 4 weeks then nights for 2 weeks; fecal sample collected ~same time of day after day and night shift schedule; no sleep or circadian assessments; food intake restricted to 0700–2300h.	Night shift work did not alter alpha or beta diversity, relative abundance of Firmicutes increased and Bacteroidetes decreased. Day work was associated with increased family Faecalibacterium and Prausnitzii whereas night shift work was associated with increased family Lachnospiraceae, genera Coprococcus, Gauvreauii,Dorea and [Ruminococcus] torques.
Kim, 2019 [80]	C57BL/6J mice	11 weeks	N = 106 Males	USA	Laboratory	3 cohorts: 12 weeks continuous dark (DD), light (LL) or alternating 12h:12h LD cycle, fecal samples collected every 4 weeks; no circadian assessments; ad-libitum food.	Continuous DD or LL did not alter microbial diversity between cohorts but altered specific phyla;, 12 weeks DD increased members of Clostridia including Christensenellaceae and Lachnospiraceae and 12 weeks of LL and DD decreased plasma metabolites associated with glucose metabolism compared to LD controls.
Deaver, 2018 [81]	C57BL/6J mice	16 weeks	N = 8 Males, 4 constant light (LL), 4 controls (LD)	USA	Laboratory	LD or LL for 4 weeks, fecal samples taken at baseline then end of 4 weeks; no circadian assessments; ad-libitum food.	4 weeks of LL decreased alpha diversity compared to LD controls, however no difference in alpha or beta diversity after 4 weeks of LL compared to baseline. LL decreased an enzyme involved in butyrate synthesis-3-hydroxybutyrl-CoA Dehydrogenase; LL induced intestinal barrier dysfunction (higher permeability) compared to controls.

LD = Light:Dark cycle, LL = continuous light. DD = continuous dark

Findings from recent studies in humans show that circadian misalignment did not alter alpha or beta diversity. Two weeks of night shift work increased the relative abundance of genus Coprococcus compared to day shift work [79]; night shift work is associated with disturbed sleep [15] and as noted, levels of Coprococcus of the family Lachnospiraceae has been negatively associated with behaviorally assessed sleep quality in healthy adults [63]. Whether altered levels of Coprococcus are causally related to changes in sleep during circadian misalignment remains to be established. Findings from Mortaş and colleagues [79] showed night shift work increased the abundance of Firmicutes and decreased Bacteroidetes compared to day shift work, and findings from another study showed one night of a 2–4 hour delay in bedtime with an unscheduled waketime increased the F:B ratio, which decreased when individuals returned to their habitual bedtime [78]; sleep and circadian data were not reported in the latter study.

In recent studies in rodents, twelve weeks of constant darkness in mice increased relative abundance of bacteria in family Lachnospiraceae [80], consistent with findings from other studies that induced physiological stress [82], and sleep fragmentation [62]. Additionally, four weeks of constant light in mice resulted in intestinal barrier dysfunction [81], which may contribute to metabolic dysregulation. Findings from non-human animal studies showed constant light or darkness conditions increased microbes associated with SCFA production, but that plasma levels of metabolites associated with glucose metabolism were decreased compared to controls [80].

Sleep and Microbiome Based Countermeasures

Manipulations of the sleep and circadian systems and of the gut microbiome with the goal of improving health have also been examined. Two weeks of at-home sleep extension in eight participants improved fasting insulin resistance compared to within-subject baseline [83], but beta or alpha diversity were not found to be altered [84]. Regardless of condition, sleep efficiency was positively correlated with relative abundance of phyla Tenericutes. Of note, the sleep extension condition increased actigraphy monitored sleep duration from 5.6h to 6.6h, which is still under the 7h per night recommended sleep duration for adults [14]. Thus, whether increasing sleep duration to adequate levels alters the gut microbiome is unknown.

Findings from intervention studies supplementing with probiotics and prebiotics have showed beneficial effects on sleep and associated alterations in fecal metabolites. Twenty-four weeks of a once daily probiotic supplement, Lactobacillus gasseri CP2305, in healthy young individuals while preparing for a stressful examination, reduced anxiety levels and improved subjective sleep quality on the PSQI and objective sleep quality assessed by a portable single channel EEG device compared to placebo [85]. Further, probiotic supplementation prevented a decrease in Bifidobacterium that was found in the placebo group in response to the stressful examination. Probiotic supplementation also increased levels of the SCFA valeric acid compared to placebo. In non-human animal models, prebiotic supplementation in rats has been reported to increase NREM sleep in early life, increase REM rebound in response to a stressor, and prevent stress induced decreases in alpha diversity [86]. Links between the gut microbiome and fecal metabolome were also described, associating alpha diversity, hyodeoxycholic acid (a secondary bile acid) and allopregnalone. Prebiotic supplementation also decreased levels of a fatty acid derivative that was increased in response to unpredictable and inescapable tail shock stress [87]. In a db/db mouse model of metabolic dysfunction, prebiotic supplementation of oligofructose increased abundance of Bifidobacterium spp. and decreased food intake [88]. Additionally, fasted and fed glucose levels were improved in the db/db mice given the prebiotic. Modifying the gut microbiome composition via prebiotic supplementation has thus been shown to be associated with improved sleep and metabolic outcomes in animal models and therefore represents a countermeasure of interest to be tested in humans.

Finally, manipulation of the timing of food intake has recently been demonstrated to impact the microbiome. This is relevant to insufficient sleep and circadian misalignment as both are associated with alterations in the timing of food intake, which could contribute to gut dysbiosis in humans. Findings from a recent study on time restricted feeding in healthy young males [89] showed 25 days of eating all calories within an 8h window per day increased alpha diversity, reduced serum triglyceride levels, and upregulated mRNA levels of circadian genes Bmal1 and Clock in serum (time of day not reported) compared to controls. The impact that these changes may have on sleep or circadian clocks in metabolic tissues are unknown as such outcomes were not measured. Regulation of the circadian timing of food intake could be a countermeasure strategy to modify the gut microbiome in a way that promotes metabolic health.

Conclusions and Future Directions

Although recent evidence contributes to our current understanding of the impact the gut microbiome in the development of metabolic phenotypes, carefully controlled laboratory and field studies utilizing state-of-the-art methodologies in the metabolic and sleep and circadian fields are needed to better elucidate underlying microbiome related mechanisms involved in physiological and behavioral responses to insufficient sleep and circadian misalignment. Existing evidence suggests that specific taxa and their related microbial derived or modified metabolites may play a role in metabolic dysregulation associated with insufficient sleep and circadian misalignment. Furthermore, early findings suggest that increased microbial diversity is associated with better sleep health. Similarly, there is an emerging body of evidence that prebiotic supplementation can benefit the microbiome and improve sleep health outcomes.

Although this short review focused on findings and methodologies from papers published in the past two years on this topic, it is important to highlight that community composition was the major focus in these studies. Community composition is only one component of microbiome analysis, as functional changes in the microbiome are also important. There is functional redundancy across microbes and the functional capacity of a given microbe does not necessarily capture functional output, especially at a specific time and under physiological challenges such as sleep and circadian disruption. Thus, to advance the field, sleep and circadian rhythm studies need to examine the impact of sleep duration and circadian timing on the functional output of microbes and ultimately link changes in functional output with changes in metabolic physiology. Combining other omics-based analyses with microbiome analyses, such as proteomics and metabolomics, represents one approach to quantify and link functional output of the microbiome to metabolic dysregulation associated with sleep and circadian disruption.

It is possible that some inconsistent findings to date are due, in part, to methodological differences in study populations, analyzing the gut microbial community, and assessing sleep and circadian outcomes. Transdisciplinary microbiome, sleep, and circadian science requires expertise from sleep and circadian and microbiology fields be incorporated to optimize the rigor of the research as expected from each field. Careful screening of research participants for documentation of disorders, conditions, and medications and reporting these in manuscripts is essential for reproducibility and comparison of findings across studies. Some important considerations when analyzing the microbiome include variability in sequencing methodology, batch and location effects [90], body location sampling site and protocols [91] [67], and other factors [92] which make inter-study replicability difficult and highlights the importance of having publicly available data, including relevant metadata, for meta-analyses [93–95]. Additionally, current microbiome sampling techniques in humans are limited in that microbiome samples cannot easily be taken from different regions throughout the intestine. In terms of sleep and circadian protocols, careful consideration of the timing of samples and the research design used to differentiate behavioral (e.g., sleep-wake, fasting-food intake) versus circadian regulated alterations in the microbiome composition and functionality outside of the experimental perturbations are needed. Further, measurement of sleep and circadian outcomes and the timing of behaviors, such as the timing of food intake are necessary for research rigor; especially when considering metabolic outcomes. Variability of gut microbial composition between individuals [96] combined with sample sizes that may be insufficient to detect alterations in gut microbes in response to insufficient sleep and circadian misalignment may also contribute to the heterogeneity in findings among protocols.

Future interventions aimed at improving sleep and circadian health while observing changes in the microbiome and the direct impact these alterations have on metabolic health are necessary to understand this proposed link. While recent studies add to the existing literature supporting that insufficient sleep and circadian misalignment alter the gut microbiome in ways that may affect cardiometabolic health, mechanistic questions remain largely unanswered and represent fruitful areas for future research.