Activation of GPR35 protects against cerebral ischemia by recruiting monocyte-derived macrophages

Pamoic acid is a potent ligand for G protein Coupled Receptor 35 (GPR35) and exhibits antinociceptive property. GPR35 activation leads to increased energy utilization and the expression of anti-inflammatory genes. However, its role in brain disorders, especially in stroke, remains unexplored. Here we show in a mouse model of stroke that GPR35 activation by pamoic acid is neuroprotective. Pharmacological inhibition of GPR35 reveals that pamoic acid reduces infarcts size in a GPR35 dependent manner. The flowcytometric analysis shows the expression of GPR35 on the infiltrating monocytes/macrophages and neutrophils in the ischemic brain. Pamoic acid treatment results in a preferential increment of noninflammatory Ly-6CLo monocytes/macrophages in the ischemic brain along with the reduced neutrophil counts. The neuroprotective effect of GPR35 activation depends on protein kinase B (Akt) and p38 MAPK. Together we conclude that GPR35 activation by pamoic acid reprograms Ly-6CLo monocytes/macrophages to relay a neuroprotective signal into the ischemic brain.

Pamoic acid (PA) is a potent GPR35 agonist that exhibits an antinociceptive property mediated through GPR35 14 . In contrast, pamoic acid is considered to be inert 15 and currently in use to improve the dissolution of pharmaceutical formulations 16 . GPR35 activation by pamoic acid may increase the phosphorylation of ERK1/2, which in turn initiates an anti-inflammatory signal by suppressing NF-κB-dependent inflammatory genes 17 . Activation of AKT signaling by pamoic acid through GPR35 may critically involve survival signals, anti-apoptosis 18 , and synthesis of essential cellular proteins 19 . Numerous studies reported inflammation as an integral part of cerebral ischemia and therefore responsible for the poor outcome 20 . Modulation of immune cells in CNS disorder, especially in stroke, has been reported to be beneficial. Until now, the impact of GPR35 activation in cerebral ischemia is not known. Therefore, we investigated the role of GPR35 activation by pamoic acid in a mouse model of stroke. Our data reveal that activation of GPR35 by pamoic acid reprograms monocytes that results in improved stroke outcome.

Results
Pamoic acid mediates the neuroprotective effect of GPR35. GPR35 is activated by pamoic acid 14 , which is currently used in many pharmaceutical preparations to modify dissolution rate 16 , and release property 21 . In our study (Fig. 1), We noticed pamoic acid treatment reduced the infarct size significantly at 100 mg/kg as well as 50 mg/kg body weight at 24 h ( Fig. 2A,B) and 48 h (Fig. 2C) after the Middle Cerebral Artery Occlusion (MCAO). Since pamoic acid is a ligand for GPR35, we sought to investigate whether GPR35 mediates the neuroprotective effect of pamoic acid. Therefore, we repeated the experiment with pharmacological inhibition of the GPR35 using ML194 22,23 . We noticed that pamoic acid was only effective in the absence of ML194 (Fig. 2D), and the effect was lost in mice after MCAO that were treated with ML194. These data demonstrate that pamoic acid produces a neuroprotective effect by activating GPR35. Since the presentation of the stroke patient is often delayed in clinical settings, we sought to investigate whether pamoic acid could exert its neuroprotective effect when administered after the stroke incidence. We administered pamoic acid (100 mg/kg) one hour after the MCAO in a separate study and found that pamoic acid was still effective in reducing the infarct volume significantly (Fig. 2E).
Pamoic acid also improved the stroke-induced sensorimotor dysfunctions as evaluated by corner test, latency-to-move test, and rotarod test. In the corner test, bias towards the right side after MCAO was observed, which was normalized by pamoic acid treatment (Fig. 2F). After 24 h of MCAO, the latency-to-move one body length was increased significantly. However, mice that were treated with pamoic acid took significantly less time to perform the task (Fig. 2G). In the rotarod test, latency to fall from the rod was reduced after MCAO. However, pamoic acid treatment restored this parameter (Fig. 2H) Cellular expression of GPR35 in the ischemic brain. We performed flowcytometry on isolated brain cells and sorted out the singlets (Fig. 3A-D).
The expression of GPR35 in the ischemic brain was localized by staining the cells with the GPR35 antibody (Fig. 3E,F). We noticed that CD45, CD11b, and Ly-6G positive cells expressed GPR35 (Fig. 3G-J), indicating that GPR35 is expressed on monocytes/macrophages and neutrophils that might infiltrate the ischemic brain.
Further analysis revealed that a substantial number of the gated GPR35 + -Ly-6G − cells were also positive for CD45 Hi -CD11b Hi , indicating that these were monocyte-derived macrophages (MDMs) 24 (Fig. 4A-D). Interestingly, pamoic acid treatment significantly increased the number of GPR35 + MDMs (Fig. 4E) 48 h after the MCAO. These findings suggest that the neuroprotective effect of pamoic acid is associated with an increased number of GPR35 + MDMs in a mouse model of stroke.
Pamoic acid favors the infiltration of neuroprotective subsets of monocytes into the ischemic brain. Since the number of MDMs that express GPR35 were increased after pamoic acid treatment in stroke,

Pamoic acid treatment ameliorates oxidative stress and iron deposition after stroke in mice.
Oxidative stress is critically associated with ischemic brain damage. Nitric oxide (NO) is an inorganic gas and known to play a significant role in modulating neuronal activity 28 . However, under ischemic conditions, it is generated in excess quantities and intervenes in inflammatory and cytotoxic action leading to neuronal death and poor outcome in stroke 29 . In line with others, we noticed an increase in NO release as measured by nitrate concentrations after ischemia in the brain of mice, and pamoic acid normalized the NO concentration 24 h and 48 h after the MCAO ( Fig. 8A and Supplementary Fig. 2A). Superoxide dismutase (SOD), catalase, and reduced glutathione (GSH) are the natural antioxidants that neutralize free radicals. When we measured SOD, catalases and GSH in brain tissue, we found that the concentrations of SOD ( Fig. 8B and Supplementary Fig. 2B), catalase ( Fig. 8E and Supplementary Fig. 2D), and GSH ( Fig. 8F and Supplementary Fig. 2E) increased substantially after stroke with pamoic acid treatment. Neutrophils and monocytes are the known source of myeloperoxidase (MPO) 30 . Increased MPO activity was observed in the brain after MCAO, which was normalized when mice were treated with pamoic acid (Fig. 8C and Supplementary Fig. 2C). Malondialdehyde (MDA) is a marker of www.nature.com/scientificreports www.nature.com/scientificreports/ lipid peroxidation associated with oxidative stress. In our study, pamoic acid treatment significantly reduced the MDA concentration after 24 h of stroke compared to the saline-treated mice ( Fig. 8D and Supplementary Fig. 2F).
Iron homeostasis is crucially involved in normal brain function. Iron overload during ischemic injury may induce oxidative stress leading to apoptosis. Under ischemic conditions, superoxide and NO are known to facilitate the process of ferritin-bound iron release substantially 31,32 . In our current study, we noticed increased iron deposition after stroke compared to the pamoic acid treated mice ( Fig. 8G-K).

Discussion
Our study revealed the neuroprotective effects of pamoic acid in a mouse model of stroke. Pamoic acid reduced the infarct volume, most likely in a GPR35 dependent manner (Fig. 2). Pamoic acid treatment activated neuroprotective subsets of monocytes/macrophages that resulted in improved functional outcome after cerebral ischemia. Till to date, the only FDA approved pharmacotherapy for stroke is tissue plasminogen activator (tPA). However, this therapy is limited to a golden timeframe of 4 hours of stroke incidence. Pamoic acid in this context showed promising effect in our current study. It reduced the infarct volume substantially when administered in a delayed time point after the stroke incidence (Fig. 2E).
Previous studies showed that GPR35 is expressed in CA1 neurons apart from immune cells and involved in the control of neuronal activity in the hippocampus 6 . Activation of GPR35 by pamoic acid, kynurenine, and zaprinast is associated with antinociceptive activity 14,33 . Pamoic acid, on the other hand, is known to increase the phosphorylation of extracellular signal-regulated kinase 1/2 in a G i/o -linked GPR35 dependent manner 14 . A separate study reported that GPR35 expressed in adipose tissue increased energy expenditure and is involved in the regulation of inflammation by enhancing the expression of anti-inflammatory genes 34 . So far, the potential of GPR35 to improve CNS disorders especially stroke was unknown. Here we provide evidence that in an ischemic condition, neutrophils and monocyte/macrophages that infiltrated the ischemic brain through the leaky blood-brain barrier (BBB) expressed GPR35. Treatment with pamoic acid increased the number of GPR35 expressing MDMs in the ischemic brain and improved stroke outcomes. Pamoic acid treatment also reduced the number of neutrophils and associated MPO activity in the ischemic brain.
Neuroinflammation plays a critical role in the pathogenesis of ischemic stroke. It may lead to both beneficial and detrimental consequences 35,36 . Due to compromised BBB function, blood-borne immune cells infiltrate into the ischemic brain during a stroke. Neutrophils are the first cell type to appear at the ischemic brain, followed by www.nature.com/scientificreports www.nature.com/scientificreports/ monocytes and macrophages. Although depletion of neutrophil resulted in neuroprotection across several studies, monocytes/macrophage ablation turned out to be either detrimental or did not benefit stroke outcomes 3,37 . MDMs that infiltrate the ischemic brain in a CCR2-dependent manner exhibit a high degree of functional plasticity and contribute to the post-ischemic repair mechanisms 38,39 . In line with other elegant studies 24 , we noticed MDMs in the ischemic hemisphere, and the number of these cell types were augmented in the ischemic hemisphere when treated with the GPR35 agonist pamoic acid (Fig. 4). Interestingly, pamoic acid treatment reduced the number of infiltrating neutrophils in the ischemic hemisphere in contrast to its effect on MDMs (Fig. 6).
Circulating monocytes exhibit two distinct phenotypes with inimitable functional properties. In mice, monocytes that express a high level of Ly-6C (Ly-6C Hi ) are known as the classically activated pro-inflammatory subset, which is specifically recruited to injury sites. On the other hand, monocytes with a low level of Ly-6C (Ly-6C Lo ) expression are considered to be anti-inflammatory and involved in repair mechanisms after injury 40,41 . In contrast to this traditional concept of phenotypes, a growing body of evidence suggest that while recruited into the inflamed tissue, Ly-6C Hi monocytes may differentiate into both classically activated pro-inflammatory macrophages (M1) and alternatively activated (M2) anti-inflammatory macrophages. Unlike Ly-6C Hi monocytes, Ly-6C Lo expressing monocytes that are recruited during inflammation may only differentiate into M2 macrophages 42 . The current consensus suggests that during a stroke, peripheral monocytes/macrophages that are recruited early (Ly-6C Hi ) into the ischemic brain, becoming M1 tissue macrophages. Afterward, these cells lose their Ly-6C and CCR2 expression and become capable of releasing repair mediators 43,44 . In our current study, we found that the number of Ly-6C Lo expressing monocytes was increased significantly in the ischemic hemisphere upon pamoic acid treatment 24 h and 48 h after the MCAO (Fig. 5). After 48 h of MCAO, we noticed a www.nature.com/scientificreports www.nature.com/scientificreports/ slight increment in Ly-6C Hi expressing monocytes upon pamoic acid treatment, which may reflect the dynamics between the protective monocytes subset (Ly-6C Lo ) and the inflammatory Ly-6C Hi expressing monocytes 45 .
Oxidative stress plays a critical role in reconciling ischemic brain damage. Cells maintain a low concentration of reactive oxygen species (ROS) to perform various functions 46 . However, to titrate the excess ROS produced during a stroke, endogenous antioxidants such as SOD and catalase are crucial. It is especially critical for the brain not only because neurons express a low level of antioxidant enzymes and have a high basal oxygen consumption rate, but also because concentrations of oxidizable lipids and iron that can act as pro-oxidant are very high 47,48 . In this context, p38 MAPK is of special importance. Apart from playing a crucial role in the monocyte differentiation and chemotaxis 49 , it substantially reduces the ROS mediated brain damage by enhancing the expression of SOD and catalase 50 . It is also implicated in the survival of endothelial cells in cerebral ischemia 51 . Cardiac ischemia is known to be improved upon p38 MAPK activation by carbon monoxide 52 . In line with these observations, we noticed that pamoic acid treatment enhanced the activity of antioxidant enzymes SOD, catalase, and GSH (Fig. 8) and reduced iron overload (Fig. 8G-K). Accordingly, it also increased the phosphorylation of p38 MAPK (Fig. 7C,D).
The PI3K/Akt pathway is known to regulate the survival, migration, and proliferation of macrophages and coordinate their response to diverse metabolic and inflammatory stimuli 18 . Akt activation is deemed necessary to facilitate M2 polarization of macrophages since its inhibition results in the abrogation of M2 gene expression 53 . Furthermore, signals such as BMP-7 and TGFβ promote M2 polarization through PI3K/Akt signaling 54,55 . Inhibition of PI3K/Akt signaling in synovial macrophages from rheumatoid arthritis patients is associated with increased apoptotic cell death 56 . Akt activation in macrophages results in reduced severity in experimental autoimmune encephalomyelitis 57 . Akt signaling was implicated in neuroprotection after stroke 13 . A recent study demonstrated that in non-small-cell lung cancer, Akt inhibition resulted in suppression of GPR35 expression 58 . In our study, we found that pamoic acid mediated neuroprotection is associated with increased phosphorylation of Akt, and pharmacological inhibition of Akt phosphorylation resulted in the abrogation of neuroprotection in stroke (Fig. 7A,E,F). GSK3β is a downstream signaling molecule of the PI3K/Akt cascade. It is known to be negatively regulated by Akt and its inhibition is associated with improved cognitive function after stroke 59 . p38 MAPK on the other hand, may also inactivate GSK3β in the brain by direct phosphorylation at its C terminus leading to beta-catenin accumulation and thus providing a p38 MAPK-mediated survival signal 60 . Indeed, in our study, we noticed increased phosphorylation of GSK3β upon pamoic acid treatment at both the time point www.nature.com/scientificreports www.nature.com/scientificreports/ Mouse stroke model. In this model, mice were subjected to left middle cerebral artery occlusion (MCAO) as described previously 3 . The mice were anesthetized with 2.5% 2,2,2-tribromoethanol (15 µl/g BW, i.p., CAS 75-80-9, Sigma-Aldrich). The skin between the ear and the orbit on the left side was incised, and the temporal muscle was removed. The stem of the middle cerebral artery (MCA) was exposed by drilling a burr hole and occluded using microbipolar electrocoagulation. The surgery was performed under a stereomicroscope, and the body temperature was maintained using a heating pad. The skin incision was closed, and mice were placed under the heating lamp until full recovery. After 24 h or 48 h of MCAO, mice were reanesthetized, and intracardiac perfusion with saline was performed. After removal, brains were placed in a brain matrice to obtain 1 mm thick coronal sections and were stained with 2,3,5-triphenyl-tetrazolium chloride solution (CAS 298-96-4; Loba Chemie) 61 . The stained sections were digitalized, and the infarct volume was determined using ImageJ. The calculated infarct volume was corrected for brain edema as described previously 62,63 . Mice were randomized to treatment groups. Pamoic acid (PA, 100 mg/kg BW, 50 mg/kg BW, s.c.; CAS: 130-85-8, Sigma-Aldrich) 64

Behavioral analysis.
To evaluate the sensorimotor function, we used the corner test, latency-to-move test, cylinder test, and rotarod test. These tests have been described previously 62 . In the corner test, mice were placed to enter into a 30°corner before and 24 hours after MCAO. When the mice reached the corner, they turned either left or right on rearing. The number of rights and left turns was counted out of 12 trials.
The latency-to-move test was performed by placing the mice at the center of a plain board. The time to cross one body length was measured before and 24 hours after MCAO.
Motor coordination and balance alterations of the mice were evaluated using a rotarod test 65 . During a four day training before surgery, mice were placed on the rotarod for the 30 s with no rotation, then 1 min with a rotation of 8 rpm. Mice were placed on the rod until they were able to stay for 1 min. During the test session, the latency-to-fall was recorded for each mouse to compare motor coordination.
Protein estimation. Protein concentration was estimated from the brain samples as described previously 67 .
Briefly, a 20 µl sample was taken in a 1.5 ml tube. 20 µl of sodium hydroxide solution was added and heated at 100 °C in a water bath for 10 min. The samples were then allowed to cool in room temperature, and 200 µl complex reagent was added with the sample mixture and incubated for 10 min. 20 µl Folin reagent was added and incubated for 60 min. The samples were then read at 750 nm using a microplate reader.