Circadian misalignment by environmental light/dark shifting causes circadian disruption in colon

All animal studies in these experiments used a Rush University Medical Center (RUMC) IACUC approved protocol. At 21 days old, male C57BL/6J mice (hereafter BL6 mice) and homozygous Per2::Luc reporter BL6 mice [33] from Jackson Labs (Bar Harbor, ME) were bred at RUMC, weaned and group housed with 2–4 mice per cage. 6-8-week-old mice were then used for each experiment and placed in light-tight chambers as previously described [14]. Mice were all maintained on a 12h light/12h dark schedule. There were two groups per mouse type: non-shifted (NS) and shifted (S). The NS mice had a standard 12h LD cycle throughout the 22-week experiment. The shifted mice followed a pattern of adhering to a 6am-6pm light and 6pm-6am dark schedule during odd numbered weeks and were ‘shifted’ to the opposite LD schedule (6am-6pm dark, 6am-6pm light) on even numbered weeks for the 22 weeks of the experiment (S1 Fig). Both NS and S mice had free access to water and chow ad libitum throughout the study. All animal studies in these experiments used a Rush University Medical Center (RUMC) IACUC approved protocol. End of study euthanasia was performed at ZT0 using carbon dioxide (primary method) and cervical dislocation (secondary method).

For central circadian rhythm measurements of activity, both BL6 and Per2::Luc mice were provided food ad libitum and placed in singly housed cages with free rotating wheels at age 6–8 weeks [37]. To allow for acclimation to the wheel, mice were kept at the 12h LD cycle to which they were previously entrained for two weeks. Following two weeks of 12h LD, mice were kept in dark:dark (DD) ‘free-running’ conditions for two weeks. Under these conditions, mice had minimal disturbances, with brief randomly timed visual inspection with red light only. DD actigram data was analyzed following the completion of the two weeks using CLOCKLAB software (Actimetrics, Wilmette, IL).

The advantage of utilizing the Per2::Luc mice is that the circadian oscillations of luciferase-tagged Per2 in colonic tissue or colon epithelial cells can be captured using colon tissue or colon organoids in covered 35mm dishes with the eight-dish wheel Kronos Dio apparatus (AB-2500, Atto Co., Tokyo, Japan). The device quantified the bioluminescence expression levels of the Per2-luciferase protein in luciferin containing media measured for one minute of exposure every 10 minutes, with a controlled nominal temperature of 37°C/5%CO₂.

IRB approval was obtained from the RUMC IRB. All subjects completed and signed RUMC IRB-approved informed consent forms. Night shift nurses and day shift nurses were asked to have flexible sigmoidoscopies at 8 am. Biopsies of normal appearing colon mucosa were performed.

Participants were recruited through a combination of Rush University Medical Center GI clinic, advertisements, posted flyers, or self-referral. Day Shift Workers (DSW) were healthy subjects (primarily nurses) who and had been on a stable work schedule of 7 am– 7pm for at least 3 shifts a week for greater than 3 months. Night shift workers were healthy subjects on a stable work schedule of 7 pm– 7 am of at least 3 shifts a week for greater than 3 months. All subjects completed blood tests and questionnaires at their initial visit and were excluded if they met any of the following criteria: 1) Inability to sign an informed consent; 2) Any history or blood tests consistent with liver disease (LFTs > 1.5 normal), renal impairment (Creatinine > 1.5), cardiovascular disease, diabetes (Hgb-A1c>8%), or thrombocytopenia (<80k); 3) Major depression (score ≥ 15 or any endorsement of suicidal intent on the Beck Depression Inventory) [38]; 4) Sleep apnea (score high risk ≥ 2 or more categories on the Berlin Questionnaire) [39] or 5) Restless leg syndrome IRLSSG consensus criteria for Restless Leg Syndrome.

In vivo intestinal permeability was determined at the end of 22 weeks as we have previously described [14, 40]. The test was performed at Zeitgeber time (ZT) 0 (lights on), after mice had been fasted for eight hours. A 200 μL solution containing lactulose (3.2 mg), sucrose (0.45 mg), sucralose (0.45 mg) and mannitol (0.9 mg) was orally gavaged, after which 2.0 mL of lactated ringers was administered subcutaneously to promote urine output. Mice were placed in metabolic cages for five hours, after which urine was collected and the total volume recorded. Intestinal permeability was calculated by using gas chromatography to measure urinary sugar concentrations. Permeability is expressed as percent urinary excretion of the oral dose of each sugar [40].

The mouse intestinal epithelial organoid cultures from the colon were prepared using a method described by Sato et al. [30, 31, 41] as we have also published [42]. Briefly, the colon was washed with ice cold PBS without Ca⁺⁺ and Mg⁺⁺. The samples were cut longitudinally and gently scraped with a glass cover slip. The samples were washed with cold PBS twice and cut into small pieces, then transferred into a 50 mL conical tube. A further five washes were performed with PBS and then the samples were transferred into a separate 50 mL conical with 2.5 mM EDTA chelating buffer and placed on a shaker at 4°C for one hour. After the removal of the supernatant, the tissue pieces were washed in PBS and the first fraction was discarded. Three mL of PBS was added to the tubes and pipetted gently multiple times to loosen the crypts. Next, the crypts were filtered through a 70 μm cell strainer and saved in a fresh conical tube. This washing and filtering was repeated 3 times with each fraction added to the fresh conical tube. 10% FBS was added to the crypt suspension and the samples centrifuged for 5 min at 300g. The supernatant was discarded and the cell pellet resuspended in 15 mL of organoid media minus growth factors. As before, the cell suspension was centrifuged at 150g for 3 min and the supernatant was discarded. This organoid media washing and centrifugation step was repeated twice to get a clean crypt suspension. After the final centrifugation the crypts were counted and resuspended in cold Matrigel (60 μL per well) and plated on a warm 24 well culture plate. The plate was then incubated at 37°C for 30 min for the Matrigel to polymerize and then 500 μL of organoid medium with the growth factors was added to each well. The organoids were then incubated at 37°C/5% CO₂ and allowed to grow for 21 days (time required for colonic organoids to grow).

The organoid immunofluorescent staining was performed on 8 well chamber slides (Nunc, Rochester, NY). Immunofluorescent staining occurred following at least 21 days (3 weeks) following colon isolation and organoid generation, to allow optimal colonic organoid growth. The medium was removed and the organoids in Matrigel were washed three times with cold PBS. Corning Cell Recovery Solution (500 μL) was added to the well and the Matrigel suspension transferred to a 15 mL conical tube. An additional 500 μL of Cell Recovery Solution was added to wash the well. The samples were kept on ice for 1 hour and tube inverted regularly to completely dissolve the Matrigel. The cells were then centrifuged at 300g for 5 min. Upon removal of the supernatant the cells were washed twice further with ice cold PBS. The organoids were fixed with warm 4% paraformaldehyde (PFA) on chamber slides for one hour at room temperature (RT) and then permeabilized with 1% Triton-X100. The samples were blocked with a solution of 3% goat serum/1% bovine serum albumin (BSA)/0.2% Triton X-100 for 1 hour. Primary staining for the organoids was done overnight at RT in a humidified chamber. The next day, organoids were washed and incubated for two hours with the appropriate secondary antibodies conjugated to Alexa Fluor 488. After washing, they were further stained for DAPI and mounted using Flouromount aqueous mounting medium (Sigma Aldrich). A Zeiss LSM 700 confocal microscope (Zeiss, Oberkochen, Germany) was used to image the organoids. Specific antibodies for staining are detailed in S1 Table. All organoid staining data is from n = 4 mice with images from at least 10 stained organoids or tissue section areas used to determine relative expression of each marker (using ImageJ) and to pick the representative images. All staining was evaluated by two blinded independent observers.

For Western blot, either total proximal colon tissue or organoid protein was determined (Bio-Rad, Hercules, CA). Tissue samples were collected and immediately frozen in liquid nitrogen and stored at −80°C until use. Organoid protein was collected at least 21 days (3 weeks) following colon isolation and organoid generation to allow optimal colonic organoid growth. Laemmli sample buffer with 2-ME (Bio-Rad) was used to prepare the samples. Thirty micrograms of protein per lane were loaded into a 4%/10% stacking acrylamide Tris gel and electrophoresed at 100 V for two hours [43]. Membrane transfer, blocking, and primary and secondary antibody incubation were performed as previously described [44]. Chemiluminescent substrate (ECL, GE Healthcare) was then applied to the membrane for protein visualization with autoradiography film (HyBlot CL, Denville Scientific, Metuchen, NJ). Optical density was determined via densitometric analysis using Image J Software (NIH, Bethesda, MD). Specific antibodies are detailed in S2 Table and original uncropped and unadjusted images of blots and gel results are included in the Supporting Information section.

Analysis of variance (ANOVA) was used to detect effects of shifting in permeability, apical junction complexes (AJC) and cell fate markers. Significance levels were set at p < 0.05 in all analyses. Statistics were performed using GraphPad, SPSS v 26 (IBM), and circular data was measured by Oriana v 4. Biodare 2 was used to analyze luciferase time series data and periodogram was performed to measure period, acrophase and amplitude by Fast Fourier Transformation Nonlinear Least Squares algorithm (FFT-NLLS) [45]. FFT-NLLS is a variation of the cosinor method and is reported to be less sensitive to missing or noisy data. The general equation for the cosinor fit is as follows [46, 47]: Y(t)=M+Acos(2πt/τ+ϕ) where M is the MESOR (Midline Statistic of Rhythm, a rhythm adjusted mean), A is the amplitude (a measure of half the extent of the variation within the cycle), ϕ is the acrophase (a measure of the time of overall highest value), and τ is the period.

Circular variance is calculated as follows: [V = 1-r]. Where r = length of the mean vector, and V = Circular variance. Circular variance has been found to an important marker in genetic models of circadian disruption [48]. All circular statistics were calculated using Oriana, v4.

To further establish that Per2::Luc mice exhibit normal circadian function, we sought to measure the circadian phenotype of the Per2::Luc mice compared to the WT BL6 mice (S2 Fig). Activity was recorded on running wheels for 14 days in LD followed by 14 days in dark:dark (DD) for Per2::Luc and BL6 mice. The LD environment involves 12h light exposure from 6:00 am to 6:00 pm. The DD environment is constant darkness; therefore, mice express an endogenous circadian rhythm (free running) and period [37]. Based on wheel data in LD and DD environments, we found that Per2::Luc mice central circadian period (24.10±0.16 hours) differed slightly from WT BL6 mice at baseline (23.81±0.067 hours, p = 0.013) however the Per2::Luc mice were found to exhibit a clear and similarly shaped stable circadian rhythm curve to WT BL6 mice (S2 Fig). These data support that the Per2::Luc mice exhibit a normal circadian rhythm.

Because the Per2 protein in the Per2:Luc reporter mice is expressed as a ‘knock-in’ Per2-Luciferase fusion protein [33], we sought to validate that Per2 protein functioning was the same in the Per2::Luc reporter mice as in WT BL6 mice. To address this question, we used antibody to the Per2 protein and immunoprecipitation (IP) followed by Western blotting for two key validated circadian protein binding partners for Per2: Cry2 and Bmal1 in both WT BL6 and Per2::Luc mice (n = 3–4 each) [2]. Colon tissue samples were collected and stored in appropriate storage conditions prior to analysis. These samples were not dependent on the timeframe used for optimal colonic organoid growth (3 weeks) in analysis. Fig 1 shows Western blotting data from these IP experiments in which no difference in Per2 protein binding to Bmal1 or Cry2 proteins was detected in the Per2::Luc mice colon tissue vs. WT BL6 colon tissue (p = 1.0 and 0.7, respectively). These data support that the Per2::Luc fusion protein exhibits similar circadian-related binding function to the WT BL6 Per2 protein.

A key aim of this study was to assess the disruption in circadian period and acrophase in mice colonic epithelium undergoing the circadian stress of chronic LD shifting and determine if these changes were stable and therefore likely to possibly promote pathology in vivo. Circadian oscillations of luciferase-tagged Per2 were measured using colon tissue or colon organoids with the Kronos Dio apparatus, and periodogram was performed by FFT-NLLS. An example of the Per2::Luc activity for organoids is shown in S3 Fig. NS Per2::Luc period in tissue versus organoids was consistent at 26.7 ± 0.2 vs 26.5 ± 0.5 (p = 0.67) as was shifted period at 26.0 ± 0.9 vs 26.2 ± 0.3 (p = 0.70). There was not a significant difference in NS vs shifted period or phase, but chronic shifting caused Per2::Luc mice to have more variability (shortening and lengthening) in both the period and phase of Per2::Luc rhythms in the colon, as is demonstrated by an increase in circular variance of period and phase (Fig 2). Increased variability, defined as circadian disruption from our chronic phase shift model, in both the period and phase of these mice support that the circadian disruption observed in colonic tissue after chronic LD shifting is maintained in colonic epithelial stem cell-derived organoids 3 weeks after isolation from the shifted mice colons (as 3 weeks is the required time period for colonic organoids to grow).

Another aim of this study was to investigate potential intestinal epithelial mechanisms through which disruption of central circadian rhythms by chronic LD shifting results in colonic leakiness as we have shown [14]. As expected from those studies, chronic LD shifting increased the intestinal permeability of both BL6 [14] and Per2::Luc mice. Fig 3A demonstrates that the mean percent excretion of oral dose of sucralose for shifted Per2::Luc mice (2.921±0.318) (p = 0.0130) Per2::Luc is significantly greater than for non-shifted Per2::Luc mice (1.769±0.271). This is a measure of increased colonic permeability when combined with our data that the lactulose/mannitol ratio was not significantly different [40].

Fig 3B shows that the excreted sucralose/lactulose ratio is also greater in shifted vs. non-shifted Per2::Luc mice. Although a ratio of sucralose to lactulose is prone to outliers, this ratio is relevant to gastrointestinal permeability due to the kinetics of the sugar probes. We have previously shown this is a more reliable measure of colonic leakiness than sucralose excretion alone [40]. Together these data strongly support that central circadian disruption by chronic LD shifting increased colonic leakiness in our shifted Per2::Luc mice as shown previously in WT BL6 mice [14].

Our data above (Fig 2) and in previous studies support that chronic LD shifting for 22 weeks results in colonic hyperpermeability in Per2::Luc mice [14]. The apical junctional complex (AJC) of intestinal epithelial cell proteins contains the tight junction protein Zonula occludens protein 1 (ZO-1) and the adherens junction protein E-cadherin that are key regulators of gut leakiness. We therefore sought to determine whether chronic 22-week LD shifting resulted in changes in expression of these two key AJC proteins that regulate permeability in colon tissue and organoids from S vs. NS mice. Our protein staining data shows both proteins were found to be clearly lower in the colon epithelial tissue and organoids from S Per2::Luc mice. Fig 4A shows a pattern with the E-cadherin AJP protein, with IF staining demonstrating lower amounts of E-cadherin protein in the colon tissues and colon organoids from shifted Per2::Luc mice but not in NS Per2:Luc mice colon tissue or organoids (p < .05 for each). Fig 4B shows Western blot data for E-cadherin with both shifted organoid and tissue samples staining show a trend towards lower E-cad protein expression, though only the shifted mice organoids demonstrate significance (p<0.05). This is possibly because organoids only include the colonic epithelial layer, while tissue samples include extraneous non-epithelial intestine that may dilute the effect of shifting on AJC proteins in tissue samples. Fig 4C demonstrates the IF staining of ZO-1 in colon epithelial tissue and organoids. Importantly, this figure demonstrates that the Per2:Luc mice colonic tissue exhibits a decreased ZO-1 protein phenotype that is maintained in the organoids created from shifted Per2:Luc mice colons compared to NS Per2:Luc mice colon tissue and organoids. Western blot data for ZO-1 protein was too variable possibly due to the high molecular weight of ZO-1. Finally, Fig 4D shows IF staining for the colon enterocyte cytoskeletal protein cytokeratin-20 (Krt-20). Changes in Krt-20 are also associated with increased colon permeability. IF staining for Krt-20 was significantly decreased in both colon tissue and organoids from shifted Per2::Luc mice (p < .05). However, once again the Western blot data for Krt-20 was only significantly lower in colon organoids (p < .05) but not colon tissue from Per2::Luc shifted mice (Fig 4E). The lower amounts of these three key AJC-related proteins in the colons of S Per2::Luc mice are consistent with the increased gut leakiness seen in the shifted Per2::Luc mice shown in Fig 1.

Finally, in Fig 5, we wished to see if there was evidence for similar changes in colonic barrier proteins in humans with chronic circadian disruption due to night shift work as we found in the shifted Per2::Luc mice. We have previously shown these nurse night shift workers are prone to increased gut leakiness in response to alcohol [13]. To do this we performed a Rush IRB-approved study (see Methods) in which we obtained colonic biopsies from Rush day and night shift nurses and performed IF staining analysis for the AJC-related proteins E-cadherin and Krt-20. As seen in Fig 5, both proteins were significantly decreased in colon biopsy tissue from the night shift workers compared to day workers (p < .05). While these are only two human subjects, these data parallel our rodent data that colonic barrier proteins are decreased with chronic phase shifting leading to decreased colonic barrier resiliency. While cohort sizes for this experiment will need to be increased for future studies to allow for a more robust comparison of variance and shift work effects, these data serve as a “proof of concept” for our animal data for relevance in human disease.

Our Supplemental Data depict S4–S7 Figs in which we show data for staining Per2::Luc mice colon tissue and organoids for cell fate markers: Carbonic Anhydrase 2 (CA2), an enterocyte maker; Chromogranin A (ChgA) marker for enteroendocrine (EE) cells; Regenerating Islet-derived Family Member 4 (Reg4) a Paneth-like cell marker; and Alcian blue mucin staining, a marker for Goblet cells. Only Reg4 was significantly changed in shifted mice with an increase in shifted mice colon organoid Reg4 protein staining and Western blot (p < .05) but not significant in colon tissue. S8 Fig shows staining of colon tissue from the two nurse shift workers shown in Fig 5 was not different for ChgA, Reg4 or Alcian blue cell fate markers.