Involvement of trabecular meshwork phagocytic suppression by sympathetic norepinephrine in nocturnal intraocular pressure rise

Intraocular pressure (IOP) is important in glaucoma development and depends on aqueous humor (AH) dynamics, involving inow from the ciliary body and outow through the trabecular meshwork (TM). IOP has a circadian rhythm entrained by sympathetic noradrenaline (NE) or adrenal glucocorticoids (GCs). Here, we investigated the involvement of GC and NE in AH outow. Pharmacological prevention of inow/outow in mice indicated an AH outow increase during day. Although TM phagocytosis can determine AH drainage, only NE showed a non-self-sustained inhibitory effect in phagocytosis of immortalized human TM cells. Pharmacological approach and RNA interference identied β1-adrenergic receptor (AR)-mediated cAMP-EPAC-SHIP1 signal activation by ablation of phosphatidylinositol triphosphate regulating phagocytic cup formation. Furthermore, pharmacological instillation in mice revealed the role of β1-AR-EPAC-SHIP1 pathway in nocturnal IOP rise. These suggest that IOP rhythm is partially regulated by this pathway. This rst demonstration of TM phagocytosis suppression by NE could be useful in glaucoma management. via and PKA. is a hub molecule to induce cup


Introduction
In most organisms, multiple physiological and behavioral processes, including sleep-wake cycles, endocrine systems, and metabolism, are controlled by circadian rhythms lasting approximately 24 h. The suprachiasmatic nucleus (SCN) in the anterior hypothalamus acts as a circadian pacemaker in mammals, sensing optical information from the retina to manage the daily body rhythms of physiological and behavioral processes 1 . The SCN synchronizes most peripheral tissues and cells through various complex pathways, involving mainly the autonomic nervous system and endocrine signals 2 . Norepinephrine (noradrenaline, NE) released from the superior cervical ganglion (SCG), a part of the sympathetic nervous system, transmits circadian timing signals to the ciliary body of the eye to regulate pupil size, and to the pineal gland to regulate nocturnal melatonin synthesis 3 . Furthermore, for most peripheral tissues, glucocorticoids secreted from the adrenal glands via the hypothalamus-pituitaryadrenal axis-mediated SCN act as strong endocrine timing signals because glucocorticoid receptors (GRs) are expressed in most peripheral cell types 4 .
Glaucoma is a leading cause of blindness in elderly people; however, no effective cure exists. Abnormal intraocular pressure (IOP) inside the eye, e.g. high IOP, contributes to glaucoma development and progression characterized by vision loss 5 . IOP is balanced by aqueous humor (AH) and has a circadian rhythm. In humans, nocturnal IOP increases irrespective of posture 6 . IOP is also elevated at night in nocturnal rats, mice, rabbits, and diurnal humans 7 , and is controlled by the SCN in mice 8 . Nocturnal IOP is also elevated in patients with glaucoma 6 ,9 , and the IOP rhythm undergoes phase shifts in patients with primary open-angle glaucoma and normal-tension glaucoma 9 . Normal tension glaucoma is also reported to involve an abnormal IOP rhythm 10 . Furthermore, aging desynchronizes the IOP rhythm from the sleep/wake rhythms, causing a delayed IOP rhythm in older healthy subjects 11 . Interestingly, the IOP rhythm is disrupted in night-shift workers 12 . A recent study reported that a disrupted circadian IOP rhythm causes optic nerve damage and increases the risk of glaucoma 11 . Thus, regulation of nocturnal IOP is central to glaucoma management, and the circadian mechanism in AH dynamics is important for glaucoma therapy. However, the regulatory molecular mechanisms of this axis remain unknown. IOP is reportedly mediated by sympathetic nerve-released NE 13,14 . However, individuals with Horner syndrome, who exhibit unilateral reduced or absent sympathetic innervation, show normal circadian aqueous-ow patterns 15 . Although adrenergic β1/2 receptors mainly mediate sympathetic nerve regulation of IOP, mice knocked out for these receptors still maintain the IOP rhythm 7 . Furthermore, many peripheral clocks in rodents are controlled by CORT 16 . Although adrenalectomy (ADX) reportedly dampens the IOP rhythms in mice under a light/dark cycle 17 , the circadian aqueous-ow pattern is normal in human patients after surgical ADX 18 . These contradictory effects of NE and glucocorticoids (GCs) may explain the recent discovery demonstrating the dual pathway by which both NE and GCs transmit timing information to the eye to form the IOP rhythm in mice 19 . However, the detailed molecular mechanism remains unclear.
In AH dynamics, the non-pigmented epithelial cell (NPE) of the ciliary body participates in AH production 20 . In contrast, the trabecular out ow pathway is responsible for homeostatically regulating IOP and is regulated by the coordinated generation of AH out ow resistance mediated by the constituent cells of the trabecular meshwork (TM) and Schlemm's canal (SC) 21,22 , and partially uveoscleral out ow is known to be involved in human IOP rhythm 23 . In particular, 58-70% of AH passes through the trabecular pathway in several young mouse strains 24 . Most of this resistance is believed to be generated in the inner wall region, comprising the juxtacanalicular tissue and the inner wall endothelium of the SC and its pores 25 . The mechanisms that regulate aqueous out ow resistance in normal and glaucomatous eyes remain unclear, and only a few newly approved medications target this site of resistance. To traverse through the out ow pathway, AH passes through the SC endothelial cells in transient, pressure-driven cellular outpouchings, termed "giant vacuoles" (GVs) 25 . Pores (small openings in the GVs) allow AH to enter the SC from the endothelium. Previous studies have demonstrated a reduction in the number of GVs and pores in glaucoma. However, the β1β2-adrenergic receptor (AR) antagonist timolol reduces IOP during the night only in rodents 26 but does not alter the number and size of GVs 27 . On the other hand, TM phagocytosis and phagocytosis can decrease particulate material and debris from AH, attenuating out ow resistance and contributing to IOP reduction 21 . Phagocytosis is thought to play an important role in the normal functioning of the out ow pathway by keeping the drainage channels free.
Long-term dexamethasone (Dex) treatment is known to decrease TM phagocytosis in the human eye, as well as primary TM cells 28,29 , leading to increased AH out ow resistance, and ultimately, glaucoma development. In addition, NE suppresses wound macrophage phagocytic e ciency through α-and β-ARdependent pathways 30 . However, the detailed effect of NE on TM phagocytosis has not been elucidated. Hence, this study aimed to uncover the effects of GC/NE on trabecular phagocytosis and its molecular mechanism. We addressed the effect of AH out ow on diurnal IOP changes in mice. In addition, the e cacy of phagocytic activity by NE/GC exposure was monitored in real time using immortalized human TM cells. Pharmacological approaches and RNA interference identi ed the pathways involved in phagocytosis regulation. Furthermore, the role of the identi ed regulatory pathway in IOP rhythm regulation was assessed using pharmacological instillation in mice.

AH out ow increased in the daytime in mice
To address the effects of ciliary AH-production on nocturnal increase in IOP, we injected the Na + /K + ATPase antagonist, ouabain, into the posterior chamber of the mouse eye at zeitgeber time (ZT) 10 ( Fig. 1A). Ouabain allowed a nocturnal IOP increase (Fig. 1B), but prevented IOP increase in individual data (p < 0.01) (Fig. 1C), indicating the effect of AH out ow. Since AH drainage suppression from the TM by microbeads i.o. injection elevates IOP in rats 31 , to identify the role of AH out ow on mouse IOP rhythm, microbeads were injected into the anterior chamber of the eye to mildly suppress the AH out ow ( Fig. 1D), as previously reported 31 . After 2 weeks, IOP in bead-injected mice at ZT6 increased up to the nocturnal level (Fig. 1E), consistent with previous reports 31,32 . Day-night differences in individual IOP were arrested by bead injection (Fig. 1F), indicating a daytime decrease in IOP due to AH out ow. To con rm diurnal changes in AH drainage in the TM, we administered small uorescent particles to the anterior chamber at ZT0 and ZT12, and subsequently observed the anterior eye extracted at ZT6 and ZT18, respectively, and measured the uorescence intensity; here, we assumed that the particles were taken into the SC or passed through the TM (Fig. 1G). Thus, it seems plausible that AH out ow through TM/SC may be involved in the daytime decrease in IOP in mice.
Driven-output stimuli of NE may suppress phagocytosis in the TM To investigate the effect of GC and NE on phagocytosis in the TM, we used pH-sensitive pHrodo particles to evaluate phagocytosis in real time in immortalized human TM cells (iHTMCs) ( Fig. 2A). The validity of this assay was con rmed by using a control with the phagocytosis inhibitor, cytochalasin D (p < 0.001) (Fig. 2B). After medium changes, including NE or Dex, phagocytosis was signi cantly suppressed by NE ( Fig. 2C, D). Continuous NE treatments dose-dependently prevented phagocytic activity in iHTMCs during 2 days of culture (Fig. 2D). Although Dex alone had no effect (Fig. 2E, F, Fig. S1A), Dex suppressed phagocytosis in a dose-dependent manner (Fig. S1B). These results indicate the importance of NE in phagocytosis.
To investigate the short-term effect of NE on phagocytic circadian rhythm and activity, we exposed iHTMC to NE for 30 min, and subsequently observed the phagocytic activity in iHTMCs in real-time for 3 days (Fig. 2G). Although we could not observe the circadian phagocytosis rhythm in the normalized uorescence signal by control, high NE pulse stimulation immediately began to suppress phagocytosis, reaching the minimum after 9 h and showed an inhibitory effect over 24 h, while Dex stimulation did not modulate phagocytosis rhythm (Fig. 2H). These ndings suggest the possibility that diurnal changes in phagocytosis in TM may occur as a result of a driven output, but is not self-sustainable after a single NE stimulation (Fig. 2I).

β1-AR mainly attenuated phagocytosis in iHTMCs
As NE suppresses wound macrophage phagocytic e ciency through α-and β-AR dependent pathways 30 , we rst con rmed the gene expression of nine AR subtypes in iHTMC. Strong expression of ADRA2A, ADRA2B, ADRA2C, ADRB1, and ADRB2 was observed, while very weak expression of ADRA1s and ADRB3 (Fig. S2A) was seen, consistent with the gene expression patterns analyzed from previous singlecell RNA-seq data in human TM macrophages (Fig. S2B) (Fig. 3A), even though some showed slight changes in cell viability occurred after 72 h of treatment (Fig. S3). Further detailed analysis revealed the signi cant suppression of phagocytosis by L-NE, dobutamine, and isoproterenol, as well as NE, but not by salbutamol (Fig. 3B), indicating the involvement of β1-AR in phagocytosis suppression. Furthermore, AR antagonists timolol (β1β2), betaxolol (β1), and ICI-118.551 (β2) tended to rescue NE-suppressed phagocytosis but not phentolamine (α-AR antagonist) in iHTMC, while antagonist alone showed no effect (Fig. 3C), indicating the involvement of β1β2-AR in phagocytosis suppression. To clarify this, we analyzed the effect of RNA interference on β1β2-AR and iHTMC phagocytosis, using siRNA against ADRB1 and ADRB2 (Fig. 3D), after veri cation of the inhibitory effect on gene expression (Fig. S4). Although siRNA exposure to control iHTMC did not show any changes, ADRB1 siRNA signi cantly increased NE-reduced phagocytic activity, but not ADRB2 siRNA (Fig. 3E). Taken together, these results present clear evidence that β1-AR mainly mediates NE effects in phagocytosis.
β1-AR mainly attenuates phagocytosis through the cAMP-EPAC pathway β1-AR is a prototypical G protein-coupled receptor (GPCR) that preferentially couples with the stimulatory G protein G s to induce cyclic adenosine monophosphate (cAMP) production. After dobutamine stimulation in iHTMC, we detected the downstream phosphorylation of cAMP response element binding protein (CREB), but not other Gq downstream Ca 2+ /calmodulin-dependent protein kinase II (CaMKII), and Gi/Gq downstream protein kinase C ( Fig. 4A-C), indicating the activation of the Gs-coupled GPCR. In fact, prostaglandin E2 (PGE2), which activates Gs-coupled GPCR (EP4 receptor), prevents iHTMC phagocytosis in a dose dependent manner (Fig. 4D). In addition, dose-dependent intracellular cAMP accumulation by cAMP inducers forskolin (FSK; Fig. 4E) and β1-AR agonist (L-NE and dobutamine; Fig. 4F) were observed, while betaxolol suppressed dobutamine-induced cAMP accumulation (Fig. 4G). These data demonstrate the importance of functional Gs-coupled β1-AR in HTMC phagocytosis.
cAMP binds to activate protein kinase A (PKA) or exchange proteins directly activated by cAMP (EPACs) 34 . In microglia cells and in peritoneal macrophages, myelin phagocytosis occurs with the involvement of both EPAC1 and PKA 35 . iHTMC phagocytic activity was slightly but signi cantly reduced by cAMP inducers (FSK and IBMX) and cAMP analogs, such as the PKA activator Sp-cAMP and the Epac/PKA activator 8-CPT-cAMP (Fig. 4H). To clarify the importance of PKA and EPAC in phagocytosis, RNA interference against PRKACA, RAPGEF3, and RAPGEF4, encoding PKA, EPAC1, and EPAC2, respectively, insu ciently but signi cantly rescued β1-AR -mediated suppression of iHTMC phagocytosis (Fig. 4I). Furthermore, blocking of PKA and EPAC1/2 by antagonists (KT5720 and ESI09, respectively) dosedependently rescued this effect (Fig. 4J), indicating the involvement of both pathways. NE and isoproterenol suppress the phagocytosis of microglia cells via EPAC activation 36 , while these ndings clearly suggest that Gs-coupled β1-AR modulates iHTMC phagocytosis via PKA and EPAC. Class I PI3K has four isoforms: α, β, γ, and δ. PI3Kγ, but not α, β, and δ, modulates the phagocytosis of microglia, which is suppressed by cAMP-mediated EPAC activation 39 . When we exposed LY294002 (broad-spectrum inhibitor of PI3Kαβδ) and CAY10505 (selective PI3Kγ inhibitor) to iHTMC, only CAY10505 slightly but signi cantly restrained phagocytosis (Fig. 5C). Furthermore, we detected a decrease in PIP3 levels by dobutamine (Fig. 5D). These results clari ed that β1-AR-mediated PIP3 modulates phagocytosis in iHTMCs.
In macrophages, PKA does not inhibit phagocytosis, whereas Epac1 exerts an inhibitory effect mainly through activation of tyrosine phosphatase SHIP1 40 (Fig. 5A). SHIP1 converts PIP3 to PI(3, 5)P2 38 (Fig.   5A). In addition, activated SHIP1 has been reported to dephosphorylate PTEN, catalyzing PIP3 in macrophages 40 , and single-cell RNA-seq analysis has suggested that both SHIP1 and PTEN are expressed in human TM macrophages 33 . In fact, exposure of iHTMC with PKA and EPAC inhibitors improved the PIP3 suppression e cacy of dobutamine (Fig. 5D). In addition, the PTEN inhibitor bisperoxovanadium (pyridine-2-carboxyl) [bpV(pic)] and SHIP1 inhibitor 3-a-aminocholestane (3AC) also improved this e ciency (Fig. 5D), indicating the involvement of the β1-AR-signaling pathway in PIP3 reduction. Based on the results presented above, we next performed western blot analysis of SHIP1 phosphorylation to determine the effects of β1-AR-mediated PKA and EPAC signaling in iHTMCs. Dobutamine induced the phosphorylation of SHIP1, which was prevented by EPAC inhibition, but not PKA inhibition (Fig. 5E), indicating the importance of β1-AR-mediated SHIP1 activation through EPAC.
As PIP3 stimulates AKT and ERK1/2 signaling to modulate phagocytosis in macrophages 40 , we next veri ed the effect of the signaling pathway of NE on AKT/ERK1/2 phosphorylation (Fig. 5A, F). Dobutamine cleanly inhibited their phosphorylation, which recovered signi cantly with EPAC inhibitors, but not with PKA inhibitors (Fig. 5F). AKT/ERK1/2 phosphorylation suppressed by dobutamine was upregulated by PTEN and SHIP1 inhibitors (Fig. 5F). These results indicate the involvement of SHIP/PTEN-reduced PIP3 in the EPAC-mediated suppression of AKT/ERK activation. Furthermore, when we monitored the effect of PTEN and SHIP1 inhibitors on the phagocytosis suppression effect of dobutamine, only SHIP1 inhibitors dose-dependently rescued it up to the near control level, but no effect was seen for the PTEN inhibitor or the antagonist alone (Fig. 5G). These results suggest that, at least in vitro, β1-AR-mediated SHIP1 activation through EPAC reduces PIP3 to suppress phagocytic cup formation.
Nocturnal activation of the β1-AR-EPAC-SHIP1 pathway enhanced IOP in mice Based on the results presented above, we next performed immunohistochemical analysis to determine the localization of Adrb1 and Ship1 in the mouse eye (Fig. S6). This demonstrated the colocalization of Adrb1 and Ship1 within the enthothelial cells in the SC and the non-pigmented epithelial cells in the ciliary body ( Fig. S6), providing evidence of the existence of the β1-AR-Epac-Ship1 system in mice.
Since NE released from the SCG has a circadian rhythm with nocturnal increase in rodents 41 , which seems to also be true in humans, a nocturnal increase of NE in TM may cause IOP increase by inhibiting phagocytosis in the TM. To validate this hypothesis, we gave mice inhibitor instillation at ZT10 and measured IOP in the night (ZT15) (Fig. 6A). First, the β1-AR antagonist slightly blocked the nocturnal IOP increase (Fig. 6B). Furthermore, instillation with EPAC and SHIP1 inhibitors signi cantly suppressed the nocturnal IOP increase (Fig. 6B) and also in individual level (p < 0.01); however, the PKA inhibitor did not show this effect (Fig. 6C), suggesting the role of the β1-AR-EPAC-SHIP1 pathway in nocturnal IOP increase. As we demonstrated the possibility of a driven-output phagocytic activity (Fig. 2), we next measured day-night changes in IOP 1 day before and after instillation to verify whether SHIP1 inhibition suppresses IOP rhythm in a self-sustainable manner (Fig. 6D). Interestingly, SHIP1 inhibition by single instillation did not reveal any inhibitory effect on nocturnal IOP increase at next day after instillation (Fig.   6D), providing evidence of driven-output IOP enhancement by nocturnal NE (Fig. 6E).
To verify the IOP inhibitory effects of drugs on β1-AR-mediated IOP regulation in vivo, we then gave mice dobutamine instillation at ZT4 to increase the IOP (Fig. 6F). Five hours after administration, the IOP was signi cantly and clearly enhanced (Fig. 6G, H). Importantly, preinstillation with a β1-AR antagonist prevented this increase (Fig. 6G, H), con rming that IOP enhanced the effect of β1-AR. To determine the role of EPAC and SHIP1 in this effect, we blocked dobutamine-activated PKA, EPAC, or SHIP1 (Fig. 6G, H). Preinstillation of EPAC and SHIP1 inhibitors signi cantly suppressed IOP increase (Fig. 6G, H) and prevented IOP increase in individual data (p < 0.01); however, the PKA inhibitor did not show this effect ( Fig. 6H), indicating the importance of the β1-AR-EPAC-SHIP1 pathway in IOP regulation. Taken together, these ndings revealed that nocturnal NE suppresses phagocytosis-mediated AF out ow through β1-AR-EPAC-SHIP1 activation, leading to an increase in the IOP at night (Fig. 6I).

Discussion
Previous studies have indicated that sympathetic NE and adrenal GC transmit circadian timing signals to the eye to generate IOP 19 . However, the involvement of NE and GC in TM phagocytosis contributing to IOP regulation remains unknown 42,43 . In this study, we found that suppression of AH drainage in mice may partially contribute to diurnal IOP reduction. In addition, we rst demonstrated that driven-out suppressed TM phagocytosis by NE in vitro, which is regulated by β1-AR-cAMP-EPAC-SHIP1 activation.
cAMP binds to PKA and EPAC1/2 34 , which are both involved in myelin phagocytosis in microglia cells 35 and macrophages 40,44 . Long-term Dex exposure decreased TM phagocytosis in the human eye and primary TM cells 28,29 , and GC alone showed no effect up to 3 days after exposure (Fig. 2). We further identi ed β1-AR -suppressed TM phagocytosis by silencing PIP3-AKT or -ERK signaling. Although it is known that NE suppresses macrophage phagocytosis 30 , and PIP3 stimulates AKT and ERK1/2 signaling to modulate phagocytosis 37,40 , the connection between NE and PIP3-triggered phagocytosis remains unclear. We demonstrate success, for the rst time, in connecting the signaling pathway: PIP3 reduction through the cAMP-EPAC-SHIP1 pathway. SHIP1 activation by cAMP-EPAC suppresses phagocytosis in macrophages 40 by decreasing PIP3 38 . These results support our hypothesis that cAMP-EPAC-SHIP1 activation prevents phagocytic cup formation by decreasing PIP3 (Fig. 6I).
We further found a slight but signi cant involvement of PKA in β1-AR-mediated phagocytosis and PIP3 reduction (Fig. 5). Although we demonstrated TM phagocytosis through PI3Kγ, PI3Kγ may modulate phagocytosis not only by EPAC activation 39 , but also by PKA activation. PKA activation can cause the conversion of Gαs to Gαi in β2-AR in HEK293 cells 45 . It is possible that β1-AR-mediated TM phagocytosis inhibition may be mediated by PKA-suppressed Ras homolog gene family A (RHOA), which contributes to phagocytosis. RHO is a member of the small GTPases CDC42 and RAC1 37 associated with the cytoskeleton. RHO-kinase inhibitors enhance the drainage of AH into the SC to reduce IOP, and RHOA blocking prevents nocturnal IOP elevation 46 . In fact, RHOA activation decreases TM phagocytosis 47 . The involvement of these small GTPases in NE-mediated TM phagocytosis remains to be elucidated.
In general, β-AR blockers, including betaxolol, effectively reduce IOP by decreasing AH in ow in patients 48 . Several studies have demonstrated the opposite effect of the sympathetic role in AH out ow. β-AR may be involved in the increase of AH out ow by reducing the size of cells in the TM 49 . Continuous electrical stimulation of the cervical sympathetic nerve decreases IOP only during 1 h 50 . However, since these ndings suggest the importance of β2-AR, β1-AR may contribute to nocturnal AH out ow resistance. β1-AR stimulates cAMP production only through Gαs, while β2-AR may also couple to the Gαi protein, which enhances the PIP3/Akt pathway in cardiomyocytes 51 and HEK293 cells through PKA activation 45 . Although the differences in the human TM or eye remain unclear, the strong inhibitory effect of β1-AR and weak inhibition of β2-AR in the present study may be the cause of the differences.
Furthermore, primary open-angle glaucoma (POAG) involves a greater increase in IOP at night than during the day 10 . Interestingly, a recent study revealed that the β1-AR antagonist betaxolol, a clinically used medication for the treatment of ocular hypertension and chronic open-angle glaucoma, is the top compound most opposed to POAG signatures calculated by microarray database analysis 52 .
Furthermore, in the TM of the POAG donors, the cAMP signaling pathway and CREB were activated 53 , while ERK phosphatase activity was downregulated 54 , consistent with the NE-activated cAMP signaling pathway and suppressed ERK activation in the present study. It is certain that further understanding of this pathway is necessary to fully explain the complex cellular mechanisms by which occupancy of speci c ARs regulates AH dynamics, which may contribute to the establishment of chronotherapy.
Betaxolol decreased the β1-AR-mediated IOP increase, while slightly prevent nocturnal IOP (Fig. 6). This slight effect may be caused by the maintenance of β2-AR-or GC-mediated AH production in the NPE of the ciliary body 19 . β-AR1/2 double knockout mice maintain this IOP rhythm 7 . It can be seen that the contribution of β1-AR-mediated circadian rhythm of AH resistance for IOP rhythm formation is not very high. Ouabain instillation did not completely arrest the IOP rhythm (Fig. 1), indicating the involvement of not only the out ow rhythm in IOP rhythm generation, but also the ciliary carbonic anhydrase (CA) in HCO 3 − production 55 . DEX stimulation increases CAII expression in the mouse brain 56 . Soluble adenylyl cyclase has been reported to be activated to produce cAMP by CA-converted HCO 3 − 57 . Ciliary CA may produce a GC/NE-mediated AH rhythm. In contrast to the slight effect of betaxolol on nocturnal IOP increase, inhibitors of EPAC or SHIP1 dramatically arrested nocturnal IOP increase. This could be explained by two hypotheses: the inhibitors enhance the AH out ow by acting on pathways other than TM phagocytosis, such as GV in the endothelial cells of the SC, or AH in ow, that is, Na + /K + ATPases and CA in the NPE of the ciliary body. In fact, single-cell RNA-seq analysis has suggested that RAPGEF3, RAPGEF4, and SHIP1 are expressed in human SC endothelial cells 33 . We further detected SHIP1 expression in NPE. Since instillation of these inhibitors did not suppress the dobutamine-induced IOP increase to below the normal level, the latter possibility is more likely, although it may not be detected well because the IOP level is usually quite low.
TM phagocytosis rhythm seems to be directly driven by NE, but not by an autonomous rhythm (Fig. 2, 6).
This regulatory pathway is consistent with β1-AR-mediated circadian regulation of the pineal gland 58 . In our previous report, NE instillation generated a transient IOP increase within 1 day, but not the day after instillation 19 . It may be possible to partly explain not only the AH production rhythm, but also the NEmediated driven-out rhythm of AH out ow resistance. The physiological importance of the driven-output rhythm and circadian clock in the TM remains unclear. Considering the role of IOP in maintaining the camera function of the eye, the strength of being able to respond exibly to acute stress and uctuations may be important in adjusting IOP.
In the present study, a pharmacologically high dose of Dex triggered the inhibition of TM phagocytosis only with NE, but Dex alone had no effect even under high doses, indicating the existence of the dual regulatory pathway, suggesting that NE is a necessary condition for GC function in the TM. Although some previous studies have demonstrated that high doses of GCs (10 −6 to 10 −5 M) suppress phagocytosis of neutrophils 59 and TM (10 −7 to 10 −6 M) 28,29 , GCs combined with β2-AR have no effect on neutrophil phagocytosis 59 . Since the physiological dose of Dex (10 −9 M) did not alter phagocytosis in iHTMCs (Fig. S1), the physiological implications of these dual regulatory pathways remain unknown.
However, both β-adrenergic signaling and glucocorticoids are mediators of SCN timing signals in osteoblasts 60 . Interactions between the sympathetic nervous system and GCs have also previously been reported. In particular, GC transcriptionally modulates β2-AR expression by modulating GC-response elements (GREs) on the promoter 61 . Interestingly, GCs rapidly activate cAMP production via Gαs to initiate non-genomic signaling, which contributes to one-third of their canonical genomic effects 62 . Thus, in TM, Gs-bound GR may enhance the β1-AR-Gs signal to suppress phagocytosis. GC also binds to the TM and endothelial cells in the SC in humans 63 , and its receptor localizes in mouse TM 19 . β2-AR acts on human TM and SC 49 , and β1-AR is expressed in mouse TM, SC, and human TM cells (Fig. 3, Fig. S2, Fig.  S6). Thus, the interaction between GCs and NE can generate an appropriate AH drainage rhythm.
Our study has several inherent limitations. First, although we provide a model of IOP induction in nocturnal NE, the circadian rhythm of NE released from SCG in humans remains unclear. Although GC secretion peaks at the light offset 64 , and NE is released with a nocturnal peak from the SCG in rodents 65 , GC rhythms are anti-phasic but not SCG-NE in diurnal animals 65,66 . Nocturnal NE release from SCG generally stimulates melatonin synthesis in humans, as well as in other mammals 67 . In human, βblockers inhibit nocturnal melatonin levels 68 , and suppress IOP increase during late night to morning 69 .
Furthermore, in nocturnal rabbits, β-blockers suppress IOP increases only at night 26 . These results indicate the involvement of nocturnal NE release from the SCG in regulating the IOP rhythm. Second, we cannot fully explain the differences in IOP rhythms in diurnal and nocturnal animals. The IOP rhythm peaks early at night in nocturnal animals 7 , while in healthy humans, it appears to be elevated during the night and peaks from late night to early morning 6,9 . In phasic SCG-NE and anti-phasic GC, the action of both factors in IOP increase may be able to explain such differences. Third, we explain AH out ow from the AH resistance rhythm by NE. However, the involvement of other determinants of out ow in circadian AH out ow remains unclear. IOP-independent uveoscleral out ow to the ciliary muscle may explain this paradox (Fig. 6I). In mice, 30-42% of AH passes through the uveoscleral pathway 24 . To understand the AH dynamic rhythm completely, these determinants need to be elucidated.
Taken together, these results suggest a potential circadian role for NE in the modulation of phagocytosis and AH out ow resistance by TM contributing to the IOP rhythm. Currently available drugs lower IOP mainly by diminishing the in ow of AH or enhancing drainage via the uveoscleral pathway. Although the TM accounts for most out ow and is the major site of AH out ow resistance 24,70 , until the approval of the ROCK inhibitor, no clinically administered drug had a direct effect on the TM. The major reason for this is a poor understanding of the mechanisms underlying its function. In the present study, we provide the possibility of therapeutic drug development targeting TM using NE-mediated resistance to AH out ow.
In addition, although we demonstrated the GC/NE interaction in TM phagocytosis, the synergistic regulatory mechanism of GC in the presence of NE in AH dynamics remains unknown. Since nongenomic signaling may also contribute to this mechanism 62 , further understanding of time-dependent e cacy and the duration of GC and NE on AH in ow and out ow will lead to a complete elucidation of the regulatory mechanisms of IOP rhythm. Although new therapeutics with new mechanisms, such as chronotherapy, are urgently needed for glaucoma treatment, the development of multiple types of drugs using this interaction could be very useful for glaucoma treatment in the future. supplemented with 1% penicillin/streptomycin and growth factors (6591; Sciencell). To measure phagocytosis, pHrodo Green Zymosan Bioparticles (P35365; ThermoFisher) were resuspended in PBS and vortexed to disperse. After 90% con uence, the medium was removed by aspiration, and 100 μL of serum-free TMCM was immediately added. After 24 h, the medium was replaced with serum-free TMCM containing pHrodo Zymosan (2.5 µg / well) in the presence of several kinds of drugs, and the plate was placed in the Incucyte ZOOM instrument (Essen Bioscience, Ann Arbor, MI, USA), installed in a 5% CO 2 incubator at 37 °C. Each well was imaged at 3 points, every 0.5 h or 1 h for more than 72 h using the phase and green uorescence channels and the 10× objective. No pHrodo Zymosan was used for uorescent control using background uorescent intensity because of auto uorescence in TMCM, and vehicle control included 0.1% DMSO. At the end of the experiment, the uorescence intensity at each time point in each well was measured using the IncuCyte ZOOM 2015A software (Essen Bioscience). To perform the detailed analysis of the pulse stimulation, we calculated the difference from the control. We calculated the average daily uorescence intensity normalized to that of the control DMSO treatment group to show statistical changes. The phagocytosis assay was independently repeated three times using four biological replicates. Data were combined and averaged, and the standard error was calculated.
To con rm the phagocytic activity in iHTMC, we treated to iHTMCs with the phagocytosis inhibitor RNA was extracted and puri ed as described above for knockdown con rmation by qPCR.

Drug instillation
Drug instillation was performed as described in our previous report 19   Ouabain individually prevents a nocturnal IOP increase (t-test, **p < 0.01). Data are presented as scatter plots with mean ± sem (n = 11). (D-F) Diurnal changes of IOP (ZT6 and ZT15) were measured 2 weeks after intraocular injection of beads. (E) IOP at ZT6 was increased up to nocturnal level (paired t-test, **p < 0.01). Data are presented as box-and-whisker plots (n = 7). (F) Day-night differences in individual IOP were arrested by bead injection. Data are presented as scatter dot plots with mean ± sem (n = 7). (G-I) Fluorescent particles were injected into the anterior chamber of the eye at ZT0 or ZT12, and after 6 h, the uorescence of extracted cornea was observed with a microscope (Ex: 480 nm, Em: >510 nm) and quanti ed by microplate reader (Em: 525 nm). (H) Diurnal change of the particles at the edge of cornea (speculating the TM) was detected (arrowhead). (I) Fluorescent intensity at ZT6 was signi cantly higher than that at ZT18 (t-test, *p<0.05). Data are presented as scatter plots with mean ± sem (n = 8). ZT, Zeitgeber time.   Instillation with KT5720 and ESI09 signi cantly suppressed the nocturnal IOP increase (paired t-test, **p <