Human circadian system causes a morning peak in prothrombotic plasminogen activator inhibitor-1 (PAI-1) independent of the sleep/wake cycle

The ability to clot blood can be life-saving after an injury, but thrombi within vessels can contribute to myocardial infarction, ischemic stroke, and sudden cardiac death.^1-3  Ongoing fibrinolytic activity helps break down thrombi and maintain vessel patency and is largely managed by circulating tissue plasminogen activator, which converts plasminogen to plasmin.^1-3  Fibrinolytic activity is significantly reduced in the morning due to an increase in plasminogen activator inhibitor-1 (PAI-1), the primary inhibitor of tissue plasminogen activator.^1-3  Thus, increased PAI-1 and decreased fibrinolysis in the morning may increase the risk for development of occlusive thrombi and could help explain the morning peak in adverse cardiovascular events.^1-4  To begin to understand potential underlying mechanisms in humans, we tested the degree to which the morning increase in PAI-1 is caused by a direct endogenous circadian rhythm in PAI-1^5-8  vs influences from the daily behavioral/environmental changes across the night and day (eg, rest/activity, fasting/feeding, dark/light, and ambient-temperature cycles).

We studied 12 healthy volunteers taking no medications (mean [range], 25.8 [20-42] years; body mass index, 23.6 [19.9-29.6] kg/m²; 6 women). Studies were approved by the Partners Human Research Committee, and participants gave written informed consent in accordance with the Declaration of Helsinki.

Participants were assessed throughout a 2-week laboratory protocol designed to desynchronize daily behavioral rhythms from internal circadian rhythms, while maintaining environmental factors constant. Participants had 2 baseline 24-hour days in normal room lighting conditions (∼90 lux wake episodes/0 lux sleep episodes), followed by a forced desynchrony protocol (FD) consisting of 12 standardized 20-hour “days” with controlled activity, posture, meals, sleep, room temperature, and light (<4 lux), as previously published.^9-12  On the second baseline day and each 20-hour day, participants underwent a standardized test battery,¹²  which included a 15-minute 60° passive head-up tilt test and a 15-minute steady-state cycle ergometer test at 60% of maximal predicted heart rate. Core body temperature derived from a rectal thermistor was continuously recorded and used to determine circadian period and phase, with minimum core body temperature assigned as 0° (equivalent to ∼4:30 am in these participants).⁹  Blood samples were collected by intravenous catheter in EDTA tubes on ice, immediately spun down in a precooled centrifuge (4°C) for 10 min at 2200 rpm, and frozen at −80°C until assayed for PAI-1 antigen by enzyme-linked immunosorbent assay (R&D Systems, Minneapolis, MN). We note that PAI-1 antigen and PAI-1 activity are highly correlated.¹³  The normal daily pattern of plasma PAI-1 antigen was assessed from samples taken hourly across 24 hours on the second baseline day and night. Thereafter, the effect on PAI-1 of the endogenous circadian system, standardized activities, and their interaction was assessed from 4 samples taken across a range of activities on each of the 12 test batteries (so that all circadian phases were represented). The 4 standardized activities each day throughout the FD were supine rest (after 15 minutes lying on the tilt table), head-up tilt (10 minutes into the tilt), seated at rest (after 15-minute sitting on the ergometer), and mild cycle exercise (10 minutes into exercise). Endogenous circadian rhythms (2 harmonics) and behavioral effects (4 activities) were tested by cosinor mixed-model analysis of variance.

From group analysis of the FD data, there was a large amplitude endogenous circadian rhythm in PAI-1 with a peak at a circadian phase of 30°, corresponding to ∼6:30 am (∼1 hour prior to habitual wake time) and a trough at a phase corresponding to ∼3:30 pm (peak-to-trough amplitude = 1.24 ng/mL; 82% of mean; P < .0001; Figure 1A,C) with an increase from trough to peak of 124%. Although there was an effect in anticipation of the stress battery on the baseline day (see Figure 2B), throughout the subsequent FD there were no significant changes in PAI-1 across the sequence of standardized behaviors while controlling for circadian effects (Figure 1B,D) and no interaction between the effects of behaviors and circadian phase.

Figure 2 enables comparison of the effects on PAI-1 of the endogenous circadian cycle versus the effects of normal behavioral changes combined with the effects of the endogenous circadian cycle across the baseline day. Overall, there was great similarity between the baseline and circadian profiles of PAI-1 in terms of both the timing (<1 hour difference in peak) and absolute amplitude (<10% difference). The main differences were an overall average increase in PAI-1 from 1.00 ng/mL on the baseline day to 1.54 ng/mL during the 20-hour days) and a rapid increase of PAI-1 ∼3 hours after scheduled awakening on the baseline day, ie, after breakfast and likely in anticipation of the start of the very first stress test battery. Because samples in the FD were collected during the test battery and not before, we likely missed any such anticipation-triggered PAI-1 changes in the circadian analyses.

Previous studies have shown a significant daily rhythm in PAI-1 in humans, but the role of the endogenous circadian system vs behaviors was not established. Our main finding is that the endogenous circadian system appears to be the major contributor to the daily changes in PAI-1 seen during a normal sleep/wake cycle in humans. Although behavioral or environmental factors such as activity level could also influence PAI-1, our 15-minute mild stress tests likely did not capture the full range of PAI-1 changes that can be induced by behavior.^14,15  This complexity is highlighted by a study that found a blunted PAI-1 rhythm in 2 blind individuals who were not synchronized to the 24-hour light/dark cycle.¹⁶  In animals, lesioning the central circadian pacemaker (suprachiasmatic nucleus) abolished the daily rhythm in PAI-1¹⁷  but also abolished behavioral rhythmicity so it could not be determined if the PAI-1 rhythm is driven directly by the circadian system or indirectly through behavior. Nonetheless, recent genetic studies in animals have demonstrated that the PAI-1 promoter is under direct control of core clock genes^5-8  and, in humans, a PAI-1 promoter polymorphism (4G5G) was found to increase morning PAI-1.^18,19  Furthermore, a genome-wide association study identified a common variant in the core clock gene BMAL-1 (ARNTL) to be robustly associated with elevated morning PAI-1 levels.²⁰  Building on this work, we have now discovered that there is a true endogenous circadian rhythm in circulating PAI-1 in humans, independent of the behavioral and environmental factors. Moreover, the circadian peak in PAI-1 at ∼6:30 am is at the start of the vulnerable window for adverse cardiovascular events (6 am to noon), as we hypothesized based on the possibility that the circadian system causes a prothrombotic state in the morning. One could speculate that evolutionary pressures shaped the circadian influences to favor blood clotting (ie, elevated PAI-1) at times of increased activity after sleep when risk of blood loss due to laceration injury was increased, eg, due to predator/prey/competitor encounters. On the other hand, this large endogenous circadian rhythm in prothrombotic PAI-1, if it persists in vulnerable individuals, could contribute to the increased morning peak in PAI-1 observed in these populations^21,22  and to the morning peak in adverse cardiovascular events.^1-3  We note that in a prospective study, individuals who later developed coronary heart disease had PAI-1 levels at baseline that were 31% higher than in case-control individuals after correcting for age, sex, and race,²³  suggesting that the endogenous circadian modulation observed in the current study (82%) may be clinically meaningful.

We previously found that other cardiovascular risk factors are under direct circadian control, including platelet activation, count, and aggregability, plasma epinephrine and norepinephrine, plasma cortisol, systolic and diastolic blood pressure, heart rate, and vagal cardiac modulation.^9-12  Of these factors, platelet activation and cortisol had endogenous circadian peaks, and plasma epinephrine the steepest circadian increase, during the vulnerable window for adverse cardiovascular events, around 9 am. Furthermore, in response to exercise, the largest circadian increase in epinephrine and largest circadian decrease in cardiac vagal modulation also occurred around 9 am. Together with these previous data, our current data suggest that the circadian system could contribute to the increased risk for cardiovascular events in the morning via a number of related cardiovascular mechanisms. Such mechanisms may have homeostatic advantage in most circumstances in anticipation of the expected behavioral changes in the morning, such as vagal withdrawal, sympathetic activation, and increased blood clotting. However, these same physiological adaptations could exacerbate existing risk factors in certain individuals, such as those with vulnerable atherosclerotic plaques in a coronary artery.⁴ 

The authors thank Joanna Garcia and Carolina Smales for help with data processing and the staff of the Medical Chronobiology Program and Brigham and Women’s Hospital’s Center for Clinical Investigation for their help running the study. The authors also thank the participants for their participation in the study.

This research was supported by National Institutes of Health (NIH), Heart, Lung and Blood Institute grant R01-HL76409 (S.A.S.), a grant from the Harvard Catalyst Clinical Research Center (F.A.J.L.S.), NIH, National Center for Research Resources grant UL1-RR025758 (Harvard Clinical and Translational Science Center), NIH, Heart, Lung and Blood Institute grant P30-HL101299 (F.A.J.L.S.), and NIH, Heart, Lung and Blood Institute grant K24-HL076446 (S.A.S.).

Contribution: F.A.J.L.S. and S.A.S. designed and performed research and wrote the paper; and F.A.J.L.S. analyzed data.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Frank A. J. L. Scheer, Division of Sleep Medicine, Brigham and Women’s Hospital, and Harvard Medical School, 221 Longwood Ave, Boston, MA 02115; e-mail: fscheer@rics.bwh.harvard.edu.