Prostaglandin D2/J2 signaling pathway in a rat model of neuroinflammation displaying progressive parkinsonian-like pathology: potential novel therapeutic targets

PGJ2 (cat. # 18500) and ibuprofen (cat. # 70280) from Cayman Chemical (Ann Arbor, MI). PGJ2 (33.4 μg/injection) was diluted in DMSO and then further diluted in PBS to a final DMSO concentration of 17% for PGJ2 microinfusions. The PGJ2 solutions were freshly prepared and stored for a maximum of 2 h at 4 °C and in the dark. A 17% DMSO/PBS was microinfused as the vehicle control. Primary antibodies were: dopaminergic neurons [TH, tyrosine hydroxylase, 1:1000, cat.# MAB318 (mouse, Millipore, Billerica, MA) and cat.# Ab6211 (rabbit, Abcam, Cambridge, MA)]; microglia [Iba1, 1:500, cat.# 019-19741 (rabbit, Wako, Richmond, VA) and 1:1000, cat.# ab139590 (chicken, Abcam)]; astrocytes [GFAP, 1:1000, cat.# AB5541 (chicken, Millipore)]; oligodendrocytes [GST-pi, 1:200, cat.# ab53943 (goat, Abcam)]; polyubiquitinated proteins [Ub, 1:200, cat.# BML-PW8805 (mouse, Enzo Life Sciences, Farmingdale, NY)]; phosphoS129 α-synuclein [pS129 α-syn, 1:200, cat.# p1571-129 (rabbit, PhosphoSolutions, Aurora, CO)]; cyclooxygenase-2 [COX-2, 1:250, cat.# 160106 (mouse, Cayman Chemical)]; prostaglandin D synthase [L-PGDS, 1:200, cat.# ab182141 (rabbit, Abcam)]; prostaglandin D2 receptor [DP2, 1:1000, cat.# PA5-20332 (rabbit, ThermoFisher, Waltham, MA)]; 15-Hydroxyprostaglandin dehydrogenase [15-PGDH, 1:500, cat.# NB200-179, (rabbit, Novus, Littleton, CO)]. Secondary antibodies—Alexa Fluor 568 (1:250, cat.# A11036, rabbit and cat.# A11031, mouse), Alexa Fluor 488 (1:250, cat.# A11039, chicken), and Alexa Fluor 350 (1:250, cat.# A10039, rabbit and cat.# A10035, mouse)—were from Life Technologies (Carlsbad, CA). Vectashield mounting medium (cat# H-1000, Vector Laboratories, Burlingame, CA).

We followed the same procedures for the rats as described in our previous study with mice [11]. Rats received unilateral (right side) injections of vehicle (DMSO) or PGJ2 into the SN. Briefly, at 18 weeks of age, rats were anesthetized by isoflurane inhalation (induction 2.5–3.5%, maintenance 2.5%) administered in 100% oxygen and placed into a digital stereotaxic instrument (Model 51730D, Stoelting Co., Wood Dale, IL) fitted with a rat anesthesia mask (Model 906, David Kopf Instruments). A burr hole was drilled in the skull at coordinates for the SNpc [31]: anterior-posterior (AP) = − 5.6 mm; medial-lateral (ML) = + 2.0 mm; and dorsal-ventral (DV) = − 8.0 mm relative to the bregma. All injections were administered to the right SN, while the contralateral (left) side served as an internal control. A 2-μl microinjection Hamilton syringe (7002 KH) with a 25-gauge needle was slowly inserted into the brain and left in place for 2 min. Thereafter, 2 μl of solution was infused at an injection rate of 0.2 μl/min (Quintessential stereotaxic injector, Model 53311, Stoelting Co.). The needle was left in place an additional 3 min to ensure total diffusion of the solution. Following injection, the needle was slowly removed and the incision was closed with monofilament absorbable sutures (cat. # 038729; Henry Schein, Melville, NY). After surgery, rats were removed from the stereotaxic frame and administered a subcutaneous injection of 1 cc Lactated Ringer’s solution, given wet palatable rodent chow, and kept in a warm place to recover. Subsequent injections to the SN were administered via the same drill hole established during the first surgical procedure.

Ibuprofen can be administered to rats twice daily for up to 80 days without adverse secondary side effects [32]. In our experiments rats were fed ibuprofen fortified chow (ibuprofen: ~ 40 mg / kg body weight) chronically starting the day after the first PGJ2 treatment. Ibuprofen was added to Purina 5001 Rodent Chow (800 ppm, Research Diets, Inc., New Brunswick, NJ). Individual intake of ibuprofen was measured daily by weighing the given and leftover chow per animal.

Rats in each group received two or four unilateral DMSO or PGJ2 injections (once / week for two or four consecutive weeks), starting at the age of 18 weeks. Rats were randomly assigned to eight groups receiving the following microinfusions (Fig. 1): (1) four × DMSO alone (n = 4), sacrificed 4 weeks after the last injection; (2) four × DMSO/PGJ2 (n = 5), sacrificed 4 weeks after the last injection; (3) two × DMSO alone (n = 8), sacrificed 4 weeks after the last injection; (4) two × DMSO/PGJ2 (n = 6), sacrificed 4 weeks after the last injection; (5) two × DMSO alone (n = 4), sacrificed 8 weeks after the last injection; (6) two × DMSO/PGJ2 (n = 4), sacrificed 8 weeks after the last injection; (7) two x DMSO fed ibuprofen (n = 9), sacrificed 4 weeks after the last injection; (8) two × DMSO/PGJ2 fed ibuprofen (n = 9), sacrificed 4 weeks after the last injection.

Rats were tested for parkinsonian-like behavior 4 or 8 weeks after their last injection (Fig. 1). We used the cylinder behavioral test that assesses spontaneous activity and limb use asymmetry in a novel environment. This is considered a highly reliable and sensitive behavioral test used to detect unilateral damage to the nigrostriatal pathway [34]. An additional advantage of the cylinder test is that inter-rater reliability is very high (r > 0.95) even with relatively inexperienced raters [34, 35].

Rats were placed in a transparent cylinder (11 in. in diameter and 18 in. in height) and video-recorded for 5–10 min depending on the activity level. A mirror was placed at an angle underneath the raised cylinder to allow the view of the back wall and the bottom of the cylinder. To determine the extent of asymmetry, the use of the left and right forelimb was counted during the wall movements after a rear, and during the landings as described in [34].

Forelimb use asymmetry is determined by the preference toward the non-impaired forelimb for weight shifting during spontaneous vertical exploration and landing. These movements along the wall and landings after a rear, were separately recorded and then averaged for total asymmetry, to correct for variability as described in [34, 35]. Each cylinder video was scored by two raters blind to the experimental group to further reduce the variability. Their scores were averaged and statistically analyzed.

Rodents display innate limb preference correlated to endogenous nigrostriatal DA imbalance [36]. Thus, the effect of the unilateral surgery on the behavior was measured as percent (%) change of the forelimb use asymmetry post-surgery, compared to the baseline behavior pre-surgery.

After the behavior analyses, rats were anesthetized (i.p.) with ketamine (100 mg/kg) and xylazine (5 mg/kg), and transcardially perfused with 4% paraformaldehyde in PBS. The rat brains were removed, post-fixed overnight at 4 °C, followed by cryoprotection (30% sucrose/PBS at 4 °C). Brains were sectioned in the coronal plane using a freezing microtome at a thickness of 30 μm, and sections were collected serially along the anterioposterior axis of the SNpc [between − 4.56 mm and − 6.36 mm for the SNpc; coordinates relative to bregma, as in [31]]. Tissue series (eight series for the SNpc) were stored at 4 °C in cryoprotectant (30% glycerol and ethylene glycol in PBS) until use. Sections were processed free floating for immunohistochemical (IHC) analyses as described in [10]. At the end, sections were mounted on gelatin-subbed glass slides with Vectashield. Sections were viewed under a wide-field fluorescence microscope (Zeiss AxioImager) using a Zeiss AxioCam MRm Rev. 3 camera connected to a motorized stage; the software AxioVision 4 module MosaiX was used to capture whole SN region mosaics (× 10 magnification).

Exposure time for each channel was kept consistent between sections. For each captured image, ZVI files were loaded onto Image J (NIH, Bethesda, MD) and converted to .tif files for use in optical density and co-localization analyses. Each channel was analyzed to an antibody specific threshold [37]. Pixel areas meeting threshold intensity criteria were measured in delineated SN regions.

The SNpc was delineated based on tyrosine hydroxylase (TH+) staining at × 10 magnification while carefully excluding cells within the ventral tegmental area (VTA).

For optical density (O.D.) analysis, three tissue sections per treatment group were used for quantification. Nonspecific background density was corrected using ImageJ rolling-ball method [38]. Data was expressed as O. D. ratio from the ipsilateral SNpc over the contralateral.

Activated microglia exhibit a remarkable variety of morphologies that can be associated with their particular functions [39]. Using binary images of individual microglia silhouettes, we distributed activated microglia into three different groups according to their form factor (FF) value (Fig. 4c), which is defined as 4π X area/perimeter² [40]. Each of the three microglia groups is defined as follows [39]: Ramified, FF: 0 to 0.5; actively engaged in neuronal maintenance providing neurotrophic factors. Reactive, FF: > 0.5 to 0.7; responsive to CNS injury. Amoeboid, FF > 0.7 to 1; cell body is amorphous with pseudopodia.

Unbiased stereology analysis was conducted by scorers blind to the experimental treatments as described in [11]. Total number of TH+ and Iba1+ cells were obtained with the Zeiss AxioVision Release 4.8.2 Software unbiased stereology programming (Carl Zeiss Group, Jena, Germany). The outline of the SNpc delineated by TH+ staining [41] was obtained at low (× 10) magnification. Within the delineated SNpc area, depending on the anterioposterior position of the coronal section, 5–11 sites were sampled with the optical fractionator probe. For each brain, six SNpc sections spaced 240 μm apart were used for the analyses. Cells were counted at × 40 magnification along the AP axis, and at predetermined intervals [x-step = 230 μm; y-step = 170 μm; 39,100 μm²; frame associated area (grid size): horizontal = 226.057 μm, vertical = 168.380 μm; 38,063.478 μm²]. The z-depth was set at 20 μm to allow a 5-μm guard on the top and bottom surface of each section to avoid error in case of tissue damage. Estimated cell numbers in the whole SNpc were used for group comparison. Dopaminergic (DA) cell counts from the intact SNpc were similar to a previous report [42]. Care was taken to ensure that cells within the VTA were excluded from SNpc quantification.

All data are expressed as the mean ± SEM. Statistical analyses were performed with GraphPad Prism 6 (GraphPad Software, San Diego, CA). A p value < 0.05 was considered statistically significant. For group comparisons, we performed one-way analysis of variance (ANOVA) followed by post hoc Tukey’s test for multiple comparison. One-tailed Student’s T test was used to compare means between two groups. Correlations between two variables were evaluated by linear regression calculating the Pearson correlation coefficients.

Following our recently established PGJ2-induced mouse model of PD-like pathology [10, 11], we investigated in vivo the progressive effects of subchronic inflammation in rats. For this purpose, rats were administered unilateral (right side) injections of PGJ2 to the SNpc for 2 or 4 weeks (once per week) as depicted in Fig. 1. Following behavioral analysis, the rats were sacrificed 4 or 8 weeks post the last injection, and their brains analyzed by immunohistochemistry.

To assess DA-specific neuronal damage, we performed IHC staining for TH+. We used the unbiased optical fractionator stereological method and compared the ratio between the numbers of TH+ cells in the ipsilateral SNpc over that in the contralateral side for each rat.

Rats, treated with four PGJ2-microinfusions and sacrificed 4 weeks following the last injection, displayed intense (74%) DA neuronal loss in the ipsilateral SNpc (Fig. 2a, second panel, and 2b, right panel), compared to DMSO-treated rats. We thus decided to reduce the number of injections to two, with the hope of detecting progressive DA neuronal loss. Indeed, DA neuronal loss was significantly less upon two than four PGJ2 injections (Fig. 2a, third and fourth panels, and 2b, right panel). Moreover, DA neuronal loss was progressive, as it significantly [t(4) = 4.65; p = 0.005] intensified 8 weeks compared to 4 weeks post-injection. Compared to DMSO-treatment, DA neuronal loss was 32% and 20% respectively, 8 and 4 weeks following PGJ2-treatment. The contralateral SNpc served as the uninjected control for each rat, and no significant differences were detected among groups (Fig. 2a, b, left panel).

The DA neurons in the VTA were less affected than those in the SNpc (Fig. 2c). Immunostaining for neuron-specific nuclear protein (NeuN) decreased significantly in the SNpc, following PGJ2 administration [t(4) = 5.76; p = 0.002], thus corroborating that the loss in TH immunostaining was due to the degeneration of DA neurons and not to TH depletion (Fig. 2d, left panel). The decrease in NeuN+ immunoreactivity in the VTA and SN pars reticulata (SNpr) was either non-significant or less than in the SNpc (Fig. 2d, right panel).

Clinically, PD is characterized by the development of motor deficits caused by the loss of dopamine in the SNpc [43‐45]. To determine the effect of PGJ2 treatment on motor performance, we measured the forelimb use asymmetry with the cylinder test which is considered a sensitive measure of the degree of unilateral nigrostriatal damage [46]. Unilateral injections into the right SN should cause forelimb impairment in the contralateral (left) side and encourage the use of the ipsilateral, right-forelimb, thus reflecting a change in nigrostriatal DA. For each animal, the percent (%) change in asymmetry between pre (baseline) and post-injection was calculated. The mean ± SEM for each treatment group was used for statistical analysis.

Weekly PGJ2-microinfusions induced a gradual and significant bias toward use of the ipsilateral over the contralateral forelimb when compared to the DMSO controls (Fig. 3a). The Pearson correlation coefficient indicates a significant negative correlation between ipsilateral forelimb use and the number of ipsilateral TH+ cells in the SNpc in the PGJ2-treated rats (r² = 0.6, p = 0.007; Fig. 3b, right panel). Such correlation was absent for the DMSO-treated rats (r² = 0.047, p = 0.287; Fig. 3b, left panel).

These behavioral assessments together with the IHC data for the TH+ DA neurons (Fig. 2) strongly support that PGJ2 induces the development of PD-like pathology in a progressive manner over the period of 8 weeks post-treatment.

We analyzed the number of microglia in DMSO and PGJ2-treated rats. The intact contralateral SN served as the control for each rat. We thus evaluated and compared the activated microglia numbers for each rat, as a ratio between ipsilateral over contralateral sides.

Due to the surgical procedure, SN injections of DMSO alone increased microglia activity as assessed by Iba1 immunostaining. Compared to DMSO-treatment, PGJ2 did not significantly [t(4) < 0.449); p > 0.33] change the overall numbers of activated microglia, (Fig. 4a, b).

However, based on the FF values, we observed a significant rise in reactive [t(4) = 2.74; p = 0.026] and amoeboid [t(4) = 4.74; p = 0.005] microglia in PGJ2-treated rats compared to the DMSO-treated ones (Fig. 4d). The rise in reactive and amoeboid microglia was only detected in rats treated with two PGJ2 injections and sacrificed 4 weeks post-injection (Fig. 4d). The microglia rise was not apparent in the PGJ2-treated rats 8 weeks post-injection. The plotted values represent the ratio between ipsilateral over contralateral sides for each rat, normalized to the ramified microglia, which are the ones associated with normal physiological function.

Astrocytes in the brain become reactive under pathological conditions, including in neurodegenerative diseases [47]. Under these conditions, reactive astrocytes overexpress GFAP and undergo morphological changes (hypertrophy and process remodeling) [47].

Our studies demonstrate that, compared to DMSO, rats that received two PGJ2 microinjections exhibited in the SNpc, a significant [t(4) = 2.14; p = 0.049] increase in astrocyte reactivity 8 weeks, post-injections (Fig. 5a, b). The astrocyte reactivity in the other groups of PGJ2-treated rats [2X, 4 weeks, t(4) = 1.18; p = 0.152; 4X, 4 weeks, t(4) = 1.04; p = 0.179] was non-significantly different from controls.

The accumulation of proteinaceous intraneuronal inclusions known as Lewy bodies in SNpc DA neurons is one of the hallmarks of PD [48]. These inclusions contain ubiquitinated proteins [49] as well as α-synuclein phosphorylated at S129 (pS129 α-syn) [50]. We addressed this aspect of PD (Fig. 6a for ubiquitinated proteins, and 6b for pS129 α-syn, both quantified in 6c) by focusing on the TH+ DA neurons of the SNpc. Since our preceding studies already demonstrated the progressive nature of the PGJ2-induced rat model, for this analysis and all subsequent studies, we focused only on four experimental paradigms: rats treated with two injections or four injections of DMSO or PGJ2, and sacrificed 4 weeks post-injections.

Regardless of the number of PGJ2 injections, there was a shift from a diffuse distribution of ubiquitinated proteins and pS129 α-syn detected in TH+ neurons of DMSO-treated rats, to more of an aggregated appearance observed in the PGJ2-treated rats (white arrows, Fig. 6a, b). Moreover, we established that only treatment with four (not two) PGJ2 injections induced a significant [t(4) = 3.22; p = 0.016] elevation in the levels of ubiquitinated proteins in the few spared TH+ neurons (Fig. 6c, left panel). The levels of pS129 α-syn also increased under these conditions, almost reaching statistical significance [t(4) = 1.63; p = 0.089; Fig. 6c, right panel].

Up to now, our findings corroborated that PGJ2-treated rats develop PD-like pathology in a progressive temporal manner. We wanted to take our in vivo investigation one step forward and address the impact of PGJ2 on the cell-type distribution and levels of key factors of the prostaglandin D2/J2 pathway observed during the 4-week window, including COX-2, the PGD synthase L-PGDS, the PGD2/J2 receptor DP2, and the prostaglandin dehydrogenase 15-PGDH (Fig. 7). We would like to highlight that the cell-type distribution and levels of these four factors have never been investigated under control or stress conditions in the SNpc in relation to PD, at least to our knowledge. Like for the ubiquitinated proteins and α-syn studies, we focused only on four experimental groups: rats treated with two injections or four injections of DMSO or PGJ2, and sacrificed 4 weeks post-injections.

We examined by immunocytochemistry, the distribution and levels of these factors in DA neurons and microglia in the SNpc by assessing the optical density ratio or co-localization of each of these factors with TH+ or Iba1+, respectively. There was no significant difference in the levels of these four factors in DA neurons (Fig. 7a–d, left) in rats that received 2× PGJ2 injections (red circles) compared to controls (DMSO-treated, black circles). In contrast, COX-2, L-PGDS, and 15-PGDH levels increased significantly in the DA neurons of PGJ2-treated rats that received four injections, when compared to controls (Fig. 7a, b, d, left). The DP2 receptor levels were stable in control and PGJ2-treated (4×) rats (Fig. 7c).

In regard to microglia, there was more variation in their responses to PGJ2-treatment. Starting with COX-2, we observed a rise in its levels in rats treated twice with PGJ2, but not in those treated four times with PGJ2, compared to controls (Fig. 7a, right). For L-PGDS, significant increases were detected in both groups of PGJ2-treated rats compared to controls (Fig. 7b, right).

Notably, DP2 receptors in microglia were almost non-existent or expressed at very low levels (Fig. 7c, right, and Table 1). When considering ipsilateral and contralateral SNpc separately, values representing the number of microglia exhibiting co-localization of the other three factors (COX-2, L-PGDS, 15-PGDH) with the microglia marker Iba1+, varied between 117 and 250 (not shown). However, for the DP2 receptor, these values ranged from 17 to 50 (Table 1); thus, they were 2 to 15 times lower. We suspect that in the microglia case, co-localization of DP2+/Iba1+ immunostaining is a paradox and is most likely due to the proximity of a microglia to a DA neuron.

The microglia levels of the prostaglandin dehydrogenase 15-PGDH did not change in any of the control or PGJ2-treated groups of rats (Fig. 7d, right). We found that this enzyme was expressed in another cell type, i.e., oligodendrocytes. The latter were detected in the SNpc of all groups of rats with the marker for mature oligodendrocytes, the pi form of glutathione-S-transferase (GST-pi, Fig. 7e).

Ibuprofen (IBP) is an NSAID that inhibits COX-1 and COX-2, thus reducing the synthesis of all prostaglandins [1, 51]. We previously demonstrated that PGJ2 induces COX-2 upregulation in neuronal cells [25], potentially establishing a positive feedback loop between PGJ2 and COX-2. This positive feedback effect of PGJ2 on COX-2 could promote the transition from acute to chronic neuroinflammation and consequently promote progressive PD pathology.

To determine the contribution of the cyclooxygenase pathway to the PGJ2-induced PD-like pathology in vivo, we administered IBP to the PGJ2-treated rats. The four groups of rats (2X DMSO ± IBP and 2X PGJ2 ± IBP) were analyzed at 4 weeks post the last PGJ2 treatment, with the cylinder test and immunochemistry. Preventive IBP-treatment starting the day after the first surgery ameliorated all PD-like pathology induced by PGJ2, except for reactive astrocytes (Fig. 8).

It is clear that IBP reduced DA neuronal loss (Fig. 8a, b), as well as reactive and amoeboid microglia rise (Fig. 8a, c, d) in the SN, and prevented the bias toward use of the ipsilateral over the contralateral forelimb (Fig. 8f). There was no statistical difference between DMSO and PGJ2-treated rats that were fed IBP in relation to these four parameters: numbers of TH+ cells (Fig. 8b), reactive (Fig. 8c) and amoeboid (Fig. 8d) microglia, and bias toward use of the ipsilateral over the contralateral forelimb (Fig. 8f). These results support the preventive effects of IBP in terms of PD-like pathology induced by PGJ2, a mediator of inflammation.

In contrast, IBP administration significantly [t(4) = 2.14; p = 0.05] increased astrocyte reactivity upon PGJ2-treatment, compared to DMSO-treatment (Fig. 8a, e). Thus, we show that PGJ2-treatment causes sustained astrogliosis, which is not alleviated by the IBP treatment (Figs. 5 and 8a, e). We should note that the role of astrocytes in PD pathology is ambiguous as activated astrocytes can produce both pro-inflammatory and anti-inflammatory factors, as well as neurotoxic and neurotrophic factors [52]. The types of reactive astrocytes that are induced by PGJ2 ± IBP remain to be investigated.