2.1. Animals

A total of 137 cynomolgus monkeys (comprising 127 males and 10 females) aged between 8 and 23 years (with an average of 13.08 years ± 4.42 SD) and weighing between 3.4 and 15.1 kg (with an average of 8.14 kg ± 1.94 SD) were enrolled in the study. The experimental protocol was approved by the Institutional Animal Care and Use Committee at the Guangdong Institute of Zoology, Guangdong Academy of Sciences. All monkeys originated from Guangdong Landau Biotech Inc., an institution accredited by AAALAC and CNSA for its experimental nonhuman primate breeding and research environment. Immediately prior to their introduction into the environment, the monkeys’ health was assessed based on their appearance, movement, mood, and general laboratory test results, which indicated that they were in a healthy condition. Monkey groups were subsequently allocated into rooms based on their prior social relationships in a randomized manner. People who allocated the monkey were unaware of the experimental procedure, including the night brightness in the rooms. None of the monkeys’ characteristics, such as age, gender, weight, or glucose metabolism status, were predetermined. Throughout the duration of the experiment, the research team and veterinary staff conducted daily checks on the rooms twice to ensure that all monkeys were in good health.

2.2. The Living Environment

Nine rooms were utilized, each with dimensions of 15 × 5 × 3.5 m (depth, width, and height, respectively). Each room features a south-facing back window measuring 0.5 × 2.5 m, a top window spanning 3 × 5 m, and a metal net door (5 × 3.5 m) that leads into a 2-m-wide corridor on the northern side. Additionally, there is an ad libitum access food tray (1 × 1 m) on the floor, a push-drink distilled water fountain, six-level shelves for sitting and playing on the wall, and recreational equipment including a swing, tumbling barrel, log, and balls.

2.3. Lighting Condition, Background Noise, and Temperature

The experiment was conducted at the Tropic of Cancer (north latitude 23°26 ^′) with the exception of nocturnal brightness. The brightness during the day was determined by direct sunlight filtering through the windows and scattered light. Considering that white light from a lamp and blue light from a screen are the most common pollution types in the dark hours of a modern living environment, the light spectrums and brightness in each room were accordingly provided by light tubes on the ceiling of the corridor, which shone through the net door (Supporting Information 1). Regardless of the time of day, every room had shadowy areas. The controlled lighting environment commenced once baseline data (Month 0) for all monkeys had been gathered. Throughout the duration of the experiment, the lights were illuminated daily from 5 PM to 8 AM.

Background noise was not intentionally restricted. Typically, the living environment was peaceful, and the room temperature was maintained between 18°C and 30°C.

2.4. Adjust of Living Condition for Individual Monkeys

During the experiment, the monkeys were typically housed in groups. However, any monkey that was bullied by others (1/137) or exhibited any diabetic complications would be promptly moved to a normal circadian environment. It would be treated by a veterinarian and raised separately in a cage (1.2 × 1.5 × 1.5 m). For statistical purposes, any data already collected from these monkeys remains valid. Diabetic complications included, but were not limited to, coma, skin ulcers (1/137), cataracts (1/137), and other related conditions.

2.5. Food, Fruit, and Cookie

Food was not designated as an experimental factor. A standard food pellet was provided twice daily at 9 AM and 3 PM, each lasting for 2 h. Fruits, including apples, bananas, sliced cucumbers, sweet potatoes, carrots, tomatoes, pitaya, and peaches (one kind each time), were provided every other day at 12 PM with unlimited access for a duration of 1 h. Cookies, peanuts, and popcorn were provided irregularly. On days of test and blood sampling (fasting 13–17 h), the meal at 9 AM would be avoided, but the monkeys had unrestricted access to drinking water, fruit, or their next meal once they woke up from anesthesia.

2.6. FBG, HbA1c, Insulin, C-Peptide, and Melatonin Concentration Testing

Monkeys were anesthetized monthly by receiving an intravenous injection of 1% sodium pentobarbital (50 mg/kg) to minimize unexpected fluctuations in blood glucose levels, and blood samples were drawn 30 min later. The FBG was measured on-site using the Ascensia Breeze Blood Glucose Monitoring System (Newbury, Berks, United Kingdom). The acceptable range for FBG levels was set at 1.1–33.3 mmol/L. Any concentration found to be below 1.1 was disregarded, while any concentration exceeding 33.3 was recorded as 33.3 for statistical analysis.

The plasma concentrations of HbA1c, insulin, C-peptide, and melatonin were separately evaluated using enzyme-linked immunosorbent assay (ELISA) kits (R&D System, Beijing, China) and analyzed by Huanya Biomedicine Technology Co. Ltd. (Beijing, China). The results were read using a microplate reader (Multiskan MK3, Thermo Scientific, Beijing, China) at a wavelength of 450 nm. Any uncertain HbA1c results, whether they are too high or too low, such as those that cannot be determined due to hemolysis, and results below 2.6%, will be discarded.

2.7. Verify of the Diabetes Condition

After an intravenous injection of glucose solution, the diabetes was confirmed in 12 randomly selected monkeys from the light-induced diabetes (LID) group using the 2hGTT (5 g/kg weight). This test was adapted from the IGT test in humans, which measures blood glucose levels after consuming a standardized amount (75 g) of glucose. All 12 monkeys tested positive for diabetes.

2.8. Retrospective Grouping of Monkeys According to Glucose Metabolic Status

Hyperglycemia, typically indicated by elevated blood glucose levels as reflected by FBG, is the primary and most frequently utilized marker of diabetes. Depending on the severity of the hyperglycemic condition, patients can be categorized into two groups: those with diabetes and those with impaired fasting glucose tolerance (IFG). The latter is a prediabetic condition characterized by elevated FBG levels but below the diagnostic threshold for diabetes [21]. For the retrospective grouping of monkeys, we utilized the clinical diagnostic criteria for diabetes that are currently employed in humans [21, 22]. However, we also took into account the concentration of HbA1c, a glycosylated hemoglobin that serves as an indicator of the average blood glucose level over a period ranging from weeks to months. Additionally, weight loss was another diagnostic criterion in our study, but it was evaluated in conjunction with FBG.

Based on the collected data, the monkeys were categorized into four groups: (i) SDM (spontaneous diabetes mellitus, consisting of monkeys diagnosed with diabetes at the onset of the formal experimental session), (ii) LID (consisting of monkeys that developed diabetes during the experimental session), (iii) IFG (consisting of monkeys with elevated FBG and/or HbA1c level but below diabetes diagnosis standard), and (iv) NGT (monkeys with normal glucose tolerance, all the remaining monkeys). The diagnosis of the monkeys that died during the experimental session (eight out of 137, three from SDM, two from LID, and three from the NGT group) was based on the data collected prior to their deaths. Notably, a female monkey that was enrolled in the 6-month preliminary session at a normal glucose level developed diabetes at Month 2. She survived for an additional 4 months in the environment and another 6 months of normal circadian period before being reenrolled in the formal experimental session with a diabetic status. This monkey was counted into the LID group in animal number, but all other data collected from it during the formal session was statistically used in the SDM group. Following this adjustment, the diagnostic rates are comprehensively listed in Table 1.

2.9. Statistical Analysis

Raw data were collected for gender, age, weight, concentrations of FBG, HbA1c, insulin, C-peptide, and melatonin. All statistical tests in this report were two-tailed. To produce statistical results, Microsoft Excel, GraphPad InStat 3.10, and SigmaPlot 11.0 for Windows were utilized.

Using GraphPad InStat 3.10, an ordinary one-way ANOVA was executed to compare raw data across the entire batch or within a subgroup of monkeys. This was done to determine if there were any significant differences among various testing time points. Subsequently, the Tukey–Kramer multiple comparison test was applied to compare data from each time point against the baseline of the same group. To determine if there were any differences between the two subgroups, raw data were compared using GraphPad InStat 3.10 ordinary one-way ANOVA, followed by the Tukey–Kramer multiple comparison test to compare data from each time point with the other group at the same point, thereby identifying the source of the difference and the significance of its magnitude. The data were presented as mean ± standard error (se). Differences were found to be statistically significant at the α = 0.05 level. The p value, F value, animal numbers (N), and degrees of freedom (DF) for the ANOVA results are listed in Supporting Information 2, 3, 4, and 5. The sources of the difference and the significance of the magnitude were displayed in the corresponding figures.

The correlation between the night brightness and concentrations of FBG, HbA1c, insulin, and C-peptide in overall or subgroups of monkeys was calculated in SigmaPlot 11.0 for Windows by running the linear regression, using the median of nocturnal light intensities of 13, 35, and 75 Lm as independent, while concentrations of FBG, HbA1c, insulin, or C-peptide of each monkey lived in the environments or in subgroups of LID, IFG, and NGT were used as dependents. The time course of the correlation coefficient between predicted concentrations of FBG and light, age, weight, light plus age, light plus weight, or light plus age plus weight was calculated by running linear regression or multiple linear regression in SigmaPlot. Median nocturnal light intensity was used as the dependent variable, while baseline age and the average weight of a monkey during the experiment session were used as independent variables.

The concentrations of plasma melatonin in the fifth month were compared using one-way repeated AVOVA in SigmaPlot among different light intensities.

3.1. Light Pollution Intensity-Dependently Inhibited Melatonin Production

Testing from the Experimental Month 5 sample found that compared to the plasma melatonin in monkeys exposed to the nocturnal brightness of 35 Lm light blue, both that exposed to 75 Lm white and 13 Lm flash neon were statistically lower (Figure 1).

3.2. Exposure to Light Pollution Deteriorated Glucose Metabolism in SDM Monkeys

The SDM rate was 5.84% (8/137). In this group, three monkeys died prematurely (37.5%, 3/8). Of these, two monkeys were from the 75 Lm room, including a 17-year-old female who died in the second experimental month with multiple skin ulcers and an 18-year-old male who passed away in the sixth month without any apparent diabetes complications. Another was an 18-year-old male from the 13 Lm flash neon room who died in the sixth month with cataracts in both eyes.

In the remaining SDM monkeys, exposure to light pollution led to a decline in glucose metabolism ability, evidenced by elevated average FBG concentrations at later time points (Figure 2(a), N = 6, p = 0.001 excluding data from the prematurely dead; N = 9, p = 0.0053 including the dead). Despite the high HbA1c concentrations observed in most SDM monkeys at the onset of the experiment, there was a slight increase in HbA1c levels at one of the later time points for every monkey. However, one-way ANOVA did not reveal statistically significant differences, likely due to the significant interindividual variance at each time point (Figure 2(b)).

3.3. Routine and Long-Term Exposure to Light Pollution Intensity-Dependently Induced Diabetes and IFG in Previously Healthy Monkeys

Although monkeys in the LID, IFG, and NGT groups entered the experiment with normal glucose metabolism status, their averaged FBG levels gradually increased during the first 6 months and remained high in the following months (Figures 3(a) and 3(b)). Exposure to nocturnal light elevated the FBG concentration significantly by the third month of the experiment for the entire group, with noticeable differences appearing by the fifth month in the LID group and by the sixth month in both the IFG and NGT groups. Furthermore, the concentration of FBG in monkeys exposed to 75 Lm of nocturnal light remained higher throughout the entire session. There were statistically significant differences in the FBG concentrations between monkeys exposed to 75 and 35 Lm of light at months 1, 3, and 6 (Figure 3(a)). However, the FBG concentrations in the IFG and NGT groups showed no significant difference at any time point under different light intensities. Nevertheless, a similar upward trend was observed in both groups, similar to that seen in the overall and LID groups. Although the study was terminated at month 10, the FBG level in the NGT group appeared to continue increasing up to that month (Figures 3(a), 3(b), 3(c), and 3(d)).

After calculating the morbidities of LID and LID+IFG in monkeys subjected to varying degrees of light pollution, a linear regression analysis revealed a significant positive correlation between the light intensities and the morbidities (Table 2). However, due to the limited number of dependent and independent variables, as well as a p value exceeding 0.05, it is possible that this correlation may not be conclusive based on the current sample size.

The point-to-point t-test results revealed that the average FBG concentrations of monkeys exposed to 75 Lm were consistently and significantly higher (results not shown). Subsequent linear regression analysis, which utilized median light intensities as independent variables and the average FBG concentration of each monkey living in the corresponding environment as the dependent variable, demonstrated that FBG levels were dependent on light intensity in both the LID and IFG groups (Table 3). However, a month-by-month analysis revealed that the impairment of the light to FBG was not consistent, with the initial month being the most susceptible period (Table 4 and Supporting Information 6).

In each group, there were general similarities between the line arrays representing HbA1c concentrations and those representing FBG. For instance, in the overall and LID groups, the 75 Lm HbA1c lines were above the other lines, albeit with varying peak times. Conversely, in the IFG and NGT groups, the lines were intertwined. The position of the line indicating 75 Lm brightness potentially reinforces the notion of light pollution’s intensity-dependent impairing effect; however, linear regression analysis did not reveal any notable disparities in the average HbA1c concentrations or light intensities among the monkey groups (Figures 3(e), 3(f), 3(g), and 3(h)). However, unlike the consistently increasing concentrations of FBG during the initial 6 months, the elevation in HbA1c levels reached its peak at Month 4 and subsequently returned to a level comparable to the baseline. Additionally, when compared to the baseline, the concentrations of HbA1c in the overall, IFG, and NGT groups were significantly elevated starting at Month 2, which was significantly earlier than the Month 6 time point observed for FBG. Considering that HbA1c reflects the average blood glucose levels rather than real-time concentrations, this phenomenon might indicate that, upon exposure to light pollution, the blood glucose concentrations were significantly elevated in the initial months following meals. After further calculation, we found a moderate correlation between the average HbA1c concentration during the Experimental Months 2–5 and light intensity. The correlation coefficients were R = 0.275 (p = 0.002, N = 127) for the overall group and R = 0.346 (P = 0.033, N = 38) for the LID group. These tampered coefficients suggest a reliable correlation between FBG and light pollution in both groups during Months 2–5, but not in the IFG or NGT groups (Figures 3(e), 3(f), 3(g), and 3(h)).

3.4. Light Pollution–Induced Diabetes as Reflected by Insulin and C-Peptide Production

Insulin is the hormone produced in the pancreas that enables glucose to enter cells [23]. A characteristic of T2D is insulin resistance, in which the body cells cannot functionally use insulin despite a normal or even higher than normal circulating concentration. Studies have shown that insulin sensitivity decreases in situations such as misaligned circadian rhythms or reduced sleep duration [24, 25]. We have now discovered that the production of insulin increased in monkeys that were exposed to the experimental environments. On one hand, the one-way ANOVA and the time course of fasting insulin concentrations revealed the following: (1) There were no significant statistical differences observed between different light intensities (Figure 4(a)) or among various monkey groups (Figure 4(b)). (2) The insulin concentration in the 35 Lm group of monkeys was statistically elevated at Month 8 compared to the baseline (Figure 4(a)). (3) While the insulin production appeared to be suppressed in the 13 Lm group of monkeys, this suppression might have been attributed to an exceptionally high insulin concentration in an individual at the baseline. On the other hand, during the experimental months of 1–8, a Mann–Whitney rank sum test revealed that, when comparing 75–35 Lm or 13 Lm environments, exposure to 75 Lm led to (1) elevated average insulin concentrations in monkey groups across most time points (Figure 4(a)), and (2) the line depicting the average insulin levels for LID monkeys was higher than that of the IFG and NGT groups (Figure 4(b)). Taking into account the rapid clearance rate of insulin (with a half-life of only 5 min), our findings revealed the following: (1) insulin production was not substantially diminished in any of the monkey groups, (2) exposure to brighter nocturnal lighting led to increased insulin resistance, and (3) there was a varying degree of susceptibility to glucose metabolic disorders among the monkeys, with those in the LID group being more susceptible.

The estimation of insulin production can be more precisely determined by measuring plasma C-peptide levels [26]. According to the results obtained from the ELISA test on the plasma of the entire batch of monkeys, the C-peptide levels were statistically elevated at Months 6 and 8 compared to Months 1–5 (Figures 4(c) and 4(d)). This elevation indicates heightened insulin production by the end of Experimental Month 6.

Based on the lines depicting insulin and C-peptide concentrations, it was observed that monkeys exposed to 75 Lm nocturnal light or those in the LID group exhibited higher insulin production compared to monkeys exposed to 35 and 13 Lm, or those in the IFG and NGT groups (Figures 4(c) and 4(d)). When considering light intensity as a standalone factor, a positive correlation was noted between the concentrations of insulin/C-peptide and the light intensities in all monkey groups, including the LID and NGT groups (Table 5). Despite experiencing a higher insulin production following exposure to a light-polluted environment, the FBG levels in monkeys of LID, IFG, and NGT groups increased over the course of the experiment. These findings suggest that, after prolonged exposure to the light-polluted environment, monkeys developed insulin resistance. Consequently, the insulin production was unable to meet the glucose metabolism demands, leading to a disruption in homeostasis and subsequently resulting in IFG and T2D.

3.5. The Involvement of Age, Weight, and Sex in the Development of Glucose Metabolic Disorders

Although monkeys were not specifically allocated to nocturnal light intensity by age or weight, when all four groups were compared together, the SDM group had statistically older average age, while there was no significant difference among the other three groups (Figure 5(a), repeated one-way ANOVA, F = 6.507, p = 0.0004, N = 137, DF = 136). However, even when comparing group data from previously healthy monkeys with age as an independent variable, linear regression again demonstrated a weak correlation between age and the average FBG concentration (Table 6).

3.6. The Synergistic Effect of Light Pollution, Age, and Weight in the Development of Glucose Metabolic Disorders

As previously mentioned, there was a weak or nonexistent correlation between weight and age with the average FBG level. Interestingly, in the presence of light pollution, multiple linear regressions indicated that both age and weight were significant factors that negatively impacted glucose metabolism in previously healthy monkeys (Table 6). A comparison clearly demonstrated this synergistic effect when examining the time course of the correlation coefficient between the monthly averaged FBG concentrations and factors such as light, age, weight, or any combination in previously healthy animals (Figure 6(a)). In parallelism with Table 6, the coefficients revealed that (1) the impact power of the light-polluted environment on glucose metabolism was uneven throughout the experimental session, with the initial month experiencing the most significant impact. (2) As a single determining factor in glucose metabolism, weight had the least impact, while age was the most powerful, with light in the middle. (3) When combined, age–weight–light had the most powerful impact on glucose metabolism disorders.

The monthly counting revealed that during the first 6 months, half of the LID monkeys (20/40) met the diabetes criteria, while the other half achieved it during Months 7–10. Furthermore, an additional 27 monkeys met the IFG criteria during Months 6–10, as shown in Figure 6(b). These trends suggested that, as regular exposure to the experimental environment was prolonged, the incidence of diabetes and IFG in monkeys increased. However, due to the limited number of animals in the study, the total number of monkeys with glucose metabolism issues was limited, and consequently, the number of healthy monkeys decreased.