Structural, functional and behavioral impact of allergic rhinitis on olfactory pathway and prefrontal cortex

Background: Allergic rhinitis (AR) has been identified as a cause of olfactory dysfunction. Beyond the classic symptoms, AR has been associated with altered sleep patterns, a decline in cognitive performance and higher likelihood of depression and anxiety. The olfactory pathway has been postulated to be a possible link between nasal inflammation and central nervous system (CNS) modifications. Thus, we aimed to investigate the structural, functional and behavioral changes in the olfactory pathway and related areas in an animal model of AR. Methods: AR was induced in adult Wistar rats by ovalbumin sensitization and challenge. Following olfactory and behavioral tests we investigated the synaptic structure of the olfactory bulb (OB), anterior olfactory nuclei (AON), piriform cortex and prefrontal cortex (PFC), by immunofluorescence detection of synaptophysin (Syn) and glutamatergic, GABAergic and dopaminergic neuronal markers. Results: We detected a significant decrease in Syn in the glomerular layer (GL) of OB and in the PFC of the AR group. Additionally, the optical density of GAD67 and VGLUT2 was reduced in the OB, AON and PFC, compared to controls. The behavioral tests demonstrated olfactory dysfunction and reduced male aggressiveness in AR rats, but we did not find any difference in the cognition and anxiety-like behavior. Conclusions: We confirmed olfactory dysfunction in a rat model of AR and we identified modifications in synaptic activity by reduction of Syn optical density in the GL of the OB and in the PFC. This was accompanied by structural changes in glutamatergic and GABAergic activity in essential components of the olfactory pathway and PFC.


Introduction
Allergic rhinitis (AR) is a chronic inflammatory disorder of the nasal mucosa, mediated by Immunoglobulin E (IgE) and characterized by nasal congestion, sneezing, itching and rhinorrhea [1]. Hyposmia is also a common symptom in AR and other mechanisms have been recently proposed to justify the decrease in smell, in addition to nasal obstruction [2]. Beyond the classic symptoms, AR has been associated with altered sleep patterns and a decline in cognitive performance, leading to an important impact on the quality of life and work/school productivity [3][4][5]. Previous studies have described an association between allergic disease and mental health disorders and a meta-analysis recently emphasized the higher likelihood of anxiety and depression in AR patients [6]. Despite the underlying mechanisms remaining unclear, in-vivo studies supported the notion that molecular and cellular mediators released in AR inflammatory process can activate common neuroimmune mechanisms involved in certain circuits of the central nervous system (CNS) related to emotions [7][8][9].
The olfactory system could be an important link between external environmental stimuli and the CNS. The axons projecting from the olfactory sensory neurons (OSN) terminate within the olfactory bulb (OB) [10]. The mitral cells (MC) in the OB can reach the piriform cortex (PirC), the entorhinal cortex and the amygdala. Anxiety and mood disorders result from disruption in the balance of limbic system activity, where amygdala and PirC have an important role. In turn, the olfactory cortex has important connections with the prefrontal cortex (PFC), a cognitive center responsible for decision making, planning and regulating social behavior [11,12]. AR seems to induce neuroinflammation in PFC [9,13] and can be related to an increased level of anxiety in AR rats [7]. In an experimental model of AR, Salimi and collaborators proposed that the allergic response might disrupt the OB-medial PFC circuit, inducing anxiety-like behavior [14]. Thus, the olfactory pathway, by its particular anatomy, organization and connectivity, may have a potential contribution in the transmission of the allergic inflammatory process of the nasal mucosa to the OB and to the brain, In addition, neuroinflammation has been related to synapse loss in the brain [15]. Induction of systemic inflammation results in increased release of pro-inflammatory cytokines, activation of microglia and loss of synaptic proteins [15]. Synaptophysin (Syn) is a pre-synaptic membrane protein associated with recycling vesicles shown to be crucial for neurotransmission. Due to it being restricted to neuronal synapses [16] it has been shown to be an indicator of synaptic density [17] and thus a useful tool to understand the synaptic structure underlying some functional circuits of the brain. Furthermore, the additional use of specific antibodies could determine the kind of neurons (glutamatergic, GABAergic or dopaminergic) affected in the process [18]. Presently, it is not known the effect of AR on synaptic structure, strength or plasticity, which can lead to brain function impairment and AR comorbidities.
In the current study, we investigated the synaptic function and structure of the olfactory pathway and PFC in an animal model of AR, recurring to immunohistochemical methods. Furthermore, we also evaluated the influence of AR on olfaction, anxiety and depression-like behaviors, learning and memory.

Animals
The experiments were conducted in male 6-8 weeks old Wistar rats (Charles River, France). Rats were housed in groups of two, in appropriate environmental conditions (light/dark cycles: 12 h, ambient temperature: 21 ± 1 • C, relative humidity: 45 ± 5%) and with free access to water and food. All animal procedures were approved by ORBEA, the internal committee of the Faculty of Medicine, University of Porto (Portugal) and performed in accordance with the guidelines of the European Communities Council Directives of 22 September 2010 (2010/ 63/EU) and Portuguese Act n • 113/13.

Ovalbumin-driven induction of allergic rhinitis
An adapted version of a previously published rhinitis induction protocol was used [19]. Male rats were sensitized on alternate days for two weeks with ovalbumin (OVA, grade V, Sigma, 50 μg per rat, 500 μl, intraperitoneal) or vehicle (saline) co-administered with Alum™ (Fisher Scientific Porto Salvo, Portugal) as an adjuvant. Following this sensitization phase, topical OVA (intranasal (i.n.) 2% OVA in saline, 25 µl/nostril) was administered daily, for 14 consecutive days, under light sevoflurane anesthesia (SevoFlo, Abbott Laboratories Ltd, Maidenhead, UK), as previously described [19,20]. The i.n. instillation was dropped on the external nares and sniffed by the rats. Control animals received i. n. saline in the same manner. After the instillation phase, the animals were submitted to several behavioral studies. Rats were weighed every morning. Animals were anaesthetized and sacrificed 24 h after the last i. n. challenge (Fig. 1).

Fig. 1.
Experimental timeline of the AR induction protocol and sequence of behavioral tests. The beginning of synthetization was represented as D1 of the protocol. The challenge phase started on D14 and rats were divided in two independent cohorts: one cohort was subjected to the Morris water-maze test; the second cohort performed the open-field, elevated plus maze, aggressive behavior and forced swim test. On day 26 all rats were submitted to the buried food test. Allergic symptoms were measured on D14 and D27. Blood samples were collected on D27 immediately before the perfusion.

Symptom scoring
On day 14, before initiation of the topical OVA or saline administration and on day 27, after the last i.n. OVA administration, the rats were placed into the observation cage (10 min) for acclimatization (Fig. 1). The numbers of sneezes and bouts of nasal rubbing were noted by 3 investigators for 30 min, as described by Narita and collaborators [21]. Hygiene maintenance stereotypic behaviors were also evaluated.

Behavioral studies
All behavioral tests were executed during the standard light phase and after a period of habituation (30 min) to the testing room. On day 14, each of the two main groups (AR model and controls) were divided in two independent cohorts. One cohort (n = 10/group) was subjected to the Morris water-maze between day 19 and day 25. In the second cohort of rats (n = 10/group) tests were executed in the following order: openfield, elevated plus maze, aggressive behavior and forced swim test, between day 21 and day 25 (1 test/day). On day 26 all rats were submitted to the buried food test. A detailed overview can be found in Fig. 1.
The buried food test was performed, in accordance with previous descriptions, to evaluate the ability of the rats to smell volatile odors and [22,23]. A cookie was buried in the bedding of a test chamber, at a random location approximately 2 cm under the surface. After a period of food deprivation (18 h), the rat was placed into the chamber and the time taken to find the buried cookie was counted up to a maximum of 15 min. A surface cookie test was executed two hours later.
The Morris water-maze test was performed to evaluate spatial learning and memory, as detailed described by Vaz and collaborators [22]. Rats were placed in a black circular pool with an escape platform 2 cm below the water surface. In acquisition period rats were trained to locate the submerged platform during 4 trials per day (60 s/trial), for 7 consecutive days. After the acquisition period, rats performed a probe trial (60 s) without platform and the number of times the rats swam through the former location of the platform and the time they spent swimming on the target quadrant were recorded by a computerized video-tracking system (EthoVision XT 8.5, Noldus).
To assess general exploratory and anxiety-like behaviors, we performed the open-field test and the elevated plus maze test. These tests were executed as described in detail by Vaz and collaborators [22]. The time spent and the distances travelled by the rat in each zone of the open-field and elevated plus maze platforms were recorded using a computerized video-tracking system (EthoVision XT 8.5, Noldus, The Netherlands). The urine and the fecal boli left were noted after each test.
Male aggression was evaluated by the intruder test, as described in literature [22,24]. An unfamiliar and higher adult rat was placed into the home cage of the test animal for 15 min and offensive and defensive behaviors were videorecorded and analyzed.
To assess the depressive state of the rats, the forced swim test was performed as reported in detail by Vaz and collaborators [22,25]. The rat was introduced into a transparent glass cylinder filled with tap water during 5 min and forced to swim. The behavior of the animal was videorecorded and the time of immobility was analyzed.

Tissue collection
After behavioral studies, animals were deeply anesthetized with sevoflurane (SevoFlo, Abbott Laboratories Ltd, Maidenhead, UK) and blood samples were reaped from the heart and analyzed. Then, rats were euthanized by transcardiac perfusion of 150 mL of 0.1 M phosphate buffer (PB), pH 7.6, for vascular rinse, followed by 250 mL of a fixative solution containing 4% paraformaldehyde in PB. The adrenal glands and lungs were quickly removed after euthanasia and weighted. The brains were weighed, divided by a mid-sagittal cut and placed in a solution of 10% sucrose in PB overnight, at 4 • C. On the following day, brains were transferred into de Olmos cryoprotectant solution and stored at − 20 • C until further processing. For histochemical analysis cryoprotected brains were rinsed in phosphate-buffered saline (PBS), to wash out de Olmos solution. The obtained hemispheres were transected rostrally through the anterior border of the mammillary bodies and embedded in agar. The blocks of agar embedded tissue were serially sectioned in the coronal plane at 40 μm through the OB, AON, PirC and PFC on a vibratome.
Sections were serially collected in PBS, transferred into de Olmos solution and stored at − 20 • C until immunohistochemical analysis. After brain removal, the nasal cavity was dissected, excess tissue removed and transferred to 4% paraformaldehyde in PB for 48 h post-fixation. Following post-fixation, the nasal cavities were removed and stored in 70% ethanol until analysis.

Total IgE
Blood samples collected immediately before perfusion were centrifuged at 2000 RPM. Then, serum was extracted, diluted 1/1000 in dilution buffer and processed in duplicate for total IgE analysis via ELISA (Abcam, ab157736), following the manufacturers instruction.

Histology
For histological assessment, the nasal cavity was decalcified using nitric acid 5% for 7 days. Following decalcification, the tissue block containing the rostral nasal cavity was embedded in paraffin and coronally sectioned (5 µm thick) at a level rostral to the incisive papilla of the hard palate. Then, sections were deparaffinized and stained with hematoxylin and eosin. At the level of the vomeronasal organ, the number of eosinophils was counted over both sides of the nasal septum mucosa, using an oil immersion objective lens (magnification: x1000; Carl Zeiss Axio Imager 2.0 microscope coupled with a colored camera and a computer with the software Carl Zeiss AxioVision Rel. 4.8 (New York, USA)). Two consecutive sections were counted per animal and data was presented as mean ± SD.

Immunofluorescence
Immunofluorescence detection of Syn, vesicular glutamate transporter 1 (VGLUT1), vesicular glutamate transporter 2 (VGLUT2), glutamic acid decarboxylase 67 (GAD67) and tyrosine hydroxylase (TH) was performed in four sets of adjacent sections containing the OB, AON, PirC and PFC sampled at regular intervals of 480 µm (1 out of 12). Previously sliced sections were washed four times with PBS (15 min) and incubated in 5% normal horse serum in PBS with 0.25% Triton X-100, for 1 h at room temperature, to block nonspecific binding sites. Sections were incubated in two separate steps. Each group of sections was incubated with one of the primary antibodies used: anti-Syn, anti-VGLUT1, anti-VGLUT2, anti-GAD67 or anti-TH antibodies, for 72 h at 4 • C (Supplementary Table 1). Antibody solutions and washes used PBS with 0.25% Triton X-100. After washing, sections were incubated in the specific secondary antibodies: anti-rabbit IgG, anti-mouse IgG and antiguinea pig IgG antibody, for 1 h at room temperature (Supplementary Table 1). Sections incubated with the biotinylated antibody were washed and then incubated with Streptavidin, for 1 h at room temperature. All sections were then mounted on gelatin-coated slides and coverslipped with Fluorsafe™ mixed with DAPI at a 1:100 dilution.

Image processing
Photographs of immunostained sections were acquired with equal gain, offsets and exposure times for image acquisition, in a microscope (Carl Zeiss Axio Imager 2.0 microscope) coupled with a colored video camera and a computer with the software Carl Zeiss AxioVision Rel. 4.8 (New York, USA). Images were obtained by manually scanning the OB, AON, PirC and PFC at a 40 × magnification. For determination of the optical intensity per area of each protein in each brain area, the software ImageJ 1.52a was used (Figs. [2][3][4][5]. Sections from animals belonging to AR groups and controls were processed in parallel to prevent variability in staining.

Statistics
Behavioral results are presented as mean ± SEM and morphological data as mean ± SD. Statistical analyses were executed using GraphPad Prism version 7.02 for Windows (GraphPad Software, La Jolla, CA, USA). Repeated measures analysis of variance (ANOVA) was used to analyze data from escape latency in the Morris water-maze test. Twoway ANOVA was used for analysis of data obtained from buried food test, open-field test, elevated plus maze test and aggressive behavior test. A one-way ANOVA was used to analyze the effect of AR on the symptoms scoring, weight of urine and number of fecal boli, forced swim test data, body and organ weights and histological data. ANOVAs were followed by Tukey highest signification difference (HSD) post-hoc comparisons, when applicable. The t-test was used in eosinophils count, total IgE level and immunofluorescence analyses, to compare the differences between AR and control groups. Differences were considered to be statistically significant when P<0.05.

Body and organs weights
On D1, mean body weights were similar between AR and control groups. Body weights increased about 1.2 times during the experiment, without differences between groups. On experimental final day (D28), there were no significant differences in body, brain, lung and adrenal weights between groups (Table 1).

Fig. 2. (i).
Immunohistochemical detection of Syn and VGLUT1 in different layers of the OB of control (A) and AR groups (B). Immunofluorescence labelling with DAPI for nucleus (blue), Alexa Fluor 568 for Syn (red), Alexa Fluor 488 for VGLUT1 (green) and merge. Scale bar = 40 µm. (ii). Immunohistochemical detection of VGLUT2 and GAD67 in different layers of the OB of control (A) and AR groups (B). Immunofluorescence labelling with DAPI for nucleus (blue), Alexa Fluor 488 for VGLUT2 (green) and Alexa Fluor 568 for GAD67 (red) and merge. Scale bar = 40 µm. (iii). Immunohistochemical detection of TH in different layers of the OB of control (A) and AR groups (B). Immunofluorescence labeling with DAPI for nucleus (blue), Alexa Fluor 488 for TH (green) and merge. Scale bar = 40 µm.

AR model confirmation
After the sensitization phase (D14), we evaluated allergic symptoms and verified that the number of sneezes and nasal itching was similar between AR and control groups. On D27 we repeated this evaluation and identified similar results for all symptoms in the control group. However, AR rats exhibited a significant increase in the number of sneezes (P<0.001) and nasal itching (P<0.01) compared to controls. Hygiene maintenance stereotypic behaviors had also increased significantly in the AR group on D27 compared to controls (P<0.05). On D27, the number of sneezes (P<0.001) and nasal itching (P<0.01) had increased significantly compared to the data obtained on D14 (Fig. 6).
On the experimental final day (D28) blood samples were collected and total IgE measured in each group. Total serum IgE levels were 39 ng/ml in the control group and 228 ng/ml in the AR group, confirming a significantly higher level of total IgE in AR rats than in controls (P<0.01) ( Supplementary Fig. 1).
In turn, the number of eosinophils counted in nasal septum mucosa also revealed a significant increase in AR group (AR group: 37.2/field vs control group: 4.1/field; P<0.01).

Olfactory assessment
Olfactory assessment was performed using the buried food test. As shown in Fig. 7, the average latency to locate the hidden cookie was 3.1 times higher in rats with AR compared to controls (P<0.05), showing reduced olfactory function of rats in the AR group. There were no differences between the groups in the time taken to locate the cookie on the surface.

Locomotor activity
Locomotor activity was evaluated in the open-field and elevated plus maze tests. There was a significant effect of AR induction on the total distance travelled in the open-field test, since AR rats traveled significantly more distance (P<0.05). However, there were no significant alterations in the elevated plus maze test (Fig. 8A).

Anxiety
Anxiety-like behavior was evaluated in the open-field and elevated plus maze tests. Animals of both groups travelled longer distances in the outer zone than in the inner zone of the open-field (P<0.001) (Fig. 8D). There was also a significant AR induction × zone interaction (P<0.05) and post-hoc analysis showed that AR rats travelled significantly more distance in the outer zone than controls (P<0.01). However, when we analyzed the time spent in the inner and outer zones, we verified that there was a significant effect of zone (P<0.05), but there was no significant effect of AR induction and no significant AR induction × zone interaction (Fig. 8C). In the plus maze test, animals traveled more distance (Fig. 8H) and spent more time (Fig. 8G) in closed arms than in open arms and central area (P<0.001), however this was independent of AR induction. There were also no significant alterations in the number of entries in the different zone of the apparatus and also in percentage of entries on the open arms ( Fig. 8E-F). There were no significant differences between groups on the urination and defecation scores obtained in the open-field or the elevated plus maze test (Fig. 8B).

Spatial learning and memory
The mean distances travelled by rats to find the submersed platform in the Morris water-maze test are show in Fig. 9. During the period of acquisition, AR and control rats gradually improved their ability to find the hidden platform (P<0.001) (Fig. 9A). However, there was no significant effect of AR group × trial blocks interaction. Behavioral analyses of the probe trial are presented in Fig. 9C. Two-way ANOVA identified that animals of both groups spent more time in the target quadrant than in opposite quadrant (P<0.001), but the time spent in the target quadrant was similar between the groups. The number of times that rats crossed the previous position of the platform was similar between groups (Fig. 9B).

Behavioral adaptation and survival
Concerning to behavioral adaptation and survival, there was no difference between AR and control groups in the immobility time evaluated in the forced swim test, which indicates that AR does not influence the cognitive functions underlying behaviors of adaptation and survival ( Supplementary Fig. 2).

Aggressive behavior
As shown in Fig. 10A, we found that AR rats spent significantly less cumulative time in offensive behaviors (P<0.05) and more time in defensive behaviors (P<0.05) than controls. Indeed, it was observed that AR rats spent significantly less time in all offensive behaviors, although this difference was significant only at keep down behavior (P<0.05) (Fig. 10B). Moreover, AR rats spent more time in all defensive behaviors than control rats, this difference being only significant at defensive upright behavior (P<0.05) (Fig. 10C).

Effects of AR on synaptic structure in the OB
The different layers that form the OB were plotted and analyzed together, revealing a decrease in the optical density of VGLUT2 (46%) in AR group when compared with controls (P<0.05). We also detected a decrease in the optical density of GAD67 (16%) in AR rats (P<0.05). When we analyzed OB layers separately, we saw a reduction in Syn optical density (44%) in the glomerular layer (GL) of the AR group (P<0.001). No significant differences were measured in Syn of others OB layers. We also detected a decrease in GAD67 (23%), in VGLUT1 (36%) (P<0.01) and in VGLUT2 (49%) (P<0.05) optical density in the GL, compared to controls. The AR group also presented a reduction of GAD67 (14%) and VGLUT2 (31%) optical density in the external plexiform layer (EPL) of the OB and a reduction in the optical density of VGLUT2 (39%) in the mitral layer (ML) (P<0.01) (Fig. 11).

Effects of AR on synaptic structure in the AON
AR induction significantly decreased the optical density of GAD67 (30%) (P<0.01) and of VGLUT2 (70%) (P<0.001) in the AON, compared to the control group (Fig. 12). When we separately analyzed the different subdivisions of the AON, we did not find differences in the distribution of the studied proteins (Data not shown).

Effects of AR on synaptic structure in the PirC
AR induction decreased the optical density of VGLUT1 (29%) in PirC, comparing with control group (P<0.001). Furthermore, we also saw a reduction in the optical density of TH (17%) in AR group (P<0.05). Other analyzed proteins involved in the synaptic function showed no differences between groups (Fig. 13).

Discussion
Our study is the first report suggesting synaptic changes in the structure of olfactory pathway in a rat model of AR. Here we have shown that the OVA sensitization and challenge protocol to induce AR results in increased eosinophil infiltration into the nasal epithelium, symptomatology representing human disease and olfactory dysfunction associated with altered synaptic function and structure in OB, PirC and PFC. Our results show reduced Syn optical density in the GL of AR rats, a reduction in the optical density of glutamatergic (VGLUT1 and VGLUT2) and GABAergic (GAD67) neuronal markers in the same layer of OB accompanied by a decline in the optical density of VGLUT2 and GAD67 in the EPL and VGLUT2 in the ML. The PirC and PFC also showed altered neuronal marker optical density. Behavioral analyses identified a reduction of male aggression. Behaviors associated with anxiety and depression, spatial learning and memory, however, were unaffected.
We assessed clinical symptoms associated with rhinitis to confirm AR in our allergen sensitized and challenged animals and show that our AR induction protocol results in increased number of sneezes, itching and increased facial grooming. These results were accompanied by increased numbers of eosinophils within the nasal epithelium and higher level of IgE. Although higher levels of IgE can be used as an indicator of allergic disease [26,27], research has shown that these levels not necessarily correlate with olfactory scores [28].
Thus, although local inflammation resulting in the obstruction of the nasal passages and subsequent hyposmia or anosmia could explain the symptomatology and olfactory dysfunction observed in our study, previous studies have shown that the reversion of inflammatory processes does not return the entirety of olfactory function, suggesting that olfactory dysfunction in AR is probably caused by additional mechanisms beyond nasal obstruction [2,29]. Indeed, some studies have reported on the role of microglia in modulation of synapse activity and regulation of synaptic turnover following inflammatory insult [15]. Additionally, the allergic inflammatory process with persistent release of cyto-and neurotoxic eosinophilic proteins such as major basic protein and eosinophilic cationic protein in the nasal mucosa could indeed compromise OSN survival and regeneration [30]. OSN are directly exposed to inflammation in the olfactory epithelium and their axons synapse exclusively in the GL of OB [31]. Therefore, as suggested by various authors, olfactory epithelium inflammation can contribute to an impairment in stimulation of OSN and a consequent decline in synaptic activity of the OB [29,32]. As such, local nasal epithelial inflammation could affect OSN function and consequently affect GL neuronal structure. In the present study, we confirmed a decline in Syn of the GL, which indicates that AR may be affecting OSN-GL synapses and, as suggested above, could indeed be due to the loss of OSN synapses in the OB of AR rats. In the GL, this decline in Syn was accompanied with a reduced optical density of VGLUT 2 and VGLUT 1. Despite being present in other OB layers, VGLUT2 is highly expressed in axon terminals of OSN, which synapse with second-order neurons in the GL [33], while VGLUT1 is identified at dendrodendritic synapses between the output neurons (TC and MC) and the interneurons in the glomeruli of the GL [33]. Our results show a reduced optical density of VGLUT2 not only in the GL, but also in the EPL and the ML, where TC and MC reside respectively. Interestingly, VGLUT1 optical density was only altered within the GL. Thus, we propose that the altered synaptic function of OSN in the GL, as measured by a reduced Syn optical density, results in compromised downstream synapses with TC of the EPL and the MC of  Data are presented as mean ± SD. One-way ANOVA: *** P<0.001, compared with D1.
the ML as evidenced by our observed reduction of VGLUT2 in the GL, EPL and ML. Consequently, loss of TC and MC activity, due to their GL connections, would explain the reduction of VGLUT1 optical density in the GL, without repercussion in other layers.
In addition, almost all periglomerular cells of GL and interneurons of EPL are GABAergic neurons and regulate the olfactory stimulus by inhibition [31]. These neurons share dendrodendritic synapses with output neurons. Therefore a compromised activity of TCs and MCs could be reflected in these GABAergic cells decreasing their optical density in GAD67. Our results seem to support this hypothesis as GAD67 optical density was reduced in both the GL and the EPL, but not in the other OB layers.
The OB catecholaminergic systems are thought to play a role in the processing of olfactory information [34][35][36]. TH is an important enzyme in the synthesis of dopamine, but within the OB only 10% of all justaglomerular cells expresses TH [31]. These cells mostly release both dopamine and GABA, and play a role in modulating signal processing in the OB and thus the modulation of information relayed from the OB to the cortex [37,38]. The inhibitory dopaminergic neurons are distinct from other GL inhibitory neurons because of their neurotransmitter-synthesizing enzyme expression plasticity. Sensory input deprivation in experimental settings, generally including naris occlusion, are known to produce alterations in TH expression [39,40]. These protocols rely mostly on chronic nasal occlusion and long-term manipulations, though some studies have shown similar results following short term protocols [41]. However, in our analyses, we did not find any difference in TH measured in the OB, probably because the changes induced by AR were not strong enough to be detectable by our methods in this small neuronal population in OB. This is most likely coupled with the fact that our protocol results in local nasal inflammation and associated olfactory deficits. Mechanistically this is likely to differ from the complete blockage of naris employed in other studies.
All these changes in OB circuitry are in line with the olfactory dysfunction verified in the buried food test of AR rats, compared to controls. Thus, we affirm that the inflammatory allergic process in the nasal mucosa could be responsible for a decline in synaptic drive in the OB, with important changes in synaptic function of different neuronal populations in the distinct layers of the OB. The chronic impairment in olfactory function could also explain the volume loss of the OB previously described in patients with sinonasal disease [42].
We then extended our analysis to the AON and the olfactory cortex, we observed that AR was associated with a decline in the optical density of VGLUT2 and GAD67 in the AON and VGLUT1 and TH in the PirC. This is of particular interest as the AON receives direct input from the OB and sends associative projections to PirC, contributing to integrate convergent inputs from multiple olfactory bulb glomeruli [43]. These AON glutamatergic excitatory inputs to the PirC provide a potentially important influence over the cortical odor responses [43]. The loss of this excitatory input could explain the reduced glutamatergic and dopaminergic activity in the PirC. Dopamine, due to its targeted affect in the PirC, has been shown to alter the balance between inhibition and excitation of neuronal circuits associated with odor perception [44]. We propose that the synaptic changes observed in the AON and PirC are likely due to its functional dependency of OB projections and contribute   to the olfactory deficits in our AR group.
Current studies warn of the effect of allergic diseases on the CNS [45,46] and the PFC has received increasing interest mainly due to its role in cognition, social behavior and anxiety. Indeed, experimentally induced AR in rats precipitates anxiety-like behaviors and an increase in Th2 cytokines in the PFC, suggesting that AR is a disorder able to model behavioral responses [7]. Yang and colleagues identified neuroinflammation not only in the OB, but also in the PFC and hippocampus of animals with AR [9]. However, the repercussion of this inflammation on the activity of the CNS is not yet established. To the best of our knowledge, there are no reports about the impact of AR on structure, type and strength of the synapses in the PFC. Our results show a reduction in Syn optical density in the PFC of animals with AR which was accompanied by a decrease in VGLUT1, VGLUT2, GAD67 and TH optical density. These changes in synaptic structure in the PFC suggests a certain degree of synaptic pathology and corroborates the results described by Salimi and colleagues, who proposed an impairment of the OB-medial PFC circuit as a result of the allergic reaction [14]. Furthermore, it could be surmised that compromised neuronal projections to the PFC can lead to widespread neuronal impairment in this area, as we have demonstrated for glutamatergic, GABAergic and dopaminergic neurons. A long-term decrease in synaptic activity in some regions of the PFC could justify the deterioration in cognitive performance and anxiety detected in human studies [47]. In our study however, we did not find changes in memory and learning, though, it could be argued that the length of time that the animals lived with AR was likely not sufficient to induce these kinds of alterations. Human studies that support behavioral and psychological changes use adult participants, with a longer history of allergic disease [6,48].
Behavioural analysis aimed to evaluate olfaction, aggression, anxiety and depression in rats with AR revealed interesting findings. The AR rats displayed severe olfactory dysfunction, since these animals needed significantly more time to discover the buried cookie than controls. However, there was no difference between groups with the cookie visible on surface, proving the absence of motivational, motor or other sensory deficits.
Because the olfactory dysfunction observed in AR rats might be related to an increase of anxiety-like behaviours, we evaluated exploratory and locomotor activities in two tests, the elevated plus maze and open-field tests, since assessing anxiety levels in rodent models is limited  when using only one test [49]. Interestingly, we found that AR animals had increased locomotor activity when compared to control animals in the outer zone in the Open-Field test. However, as there was no significant increase of the distance travelled in the unprotected inner zone, which is particularly sensitive to anxiolytic effects, we think that these alterations are more likely to be related to increased exploratory activity rather than to a variation of anxiety levels [50]. The plus maze test confirmed these results, suggesting once more that the AR in our model did not induce changes in anxiety levels. We also did not find Similarly, there were no differences in the urination and defecation scores. This is in contrast to previous reports that OVA-immunised mice spent less time in the open arm than controls in the elevated plus maze test [51], though the difference in protocol and species could explain these differing results. Taken together our results shows that AR, notwithstanding changes in the locomotor and exploratory activities, did not impact anxiety levels. We postulate that the increase in the exploratory activity of the AR rats in the open-field test may be due to the olfactory deficits of these animals. Lacking, at least partially, their Fig. 11. Graphical representation of Syn, GAD67, VGLUT1 and VGLUT2 optical density on the entire OB and its layers (GL, ML and EPL) of control and AR rats. Columns represent means ± SD (N = 10/group). T-test: * P<0.05; ** P<0.01; *** P<0.001, compared to controls. main sensory system, AR animals need to compensate, increasing the distance travelled during the exploratory activity, to establish a correct mapping of the environment. Indeed, these results are not surprising since previous studies have reported that olfactory bulbectomy induced rat ambulation in the open-field test [52,53].
Behavioural changes in the exploratory activity in AR rats and the results observed in the forced swim test allow to exclude a state of anhedonia [54]. Likewise, AR did not affect cognition underlie stress coping, survival and adaptation, also assessed in the forced swim test. In the same way, cognitive functions assessed in the Morris water-maze test remain unaltered, suggesting that the learning and memory processes in the hippocampus were not altered by our model of AR, corroborating previous studies that reported that there is no direct association between olfactory dysfunction and learning and memory impairment [47,55]. Conversely, several other studies have found that ZnSO 4 olfactory dysfunction models induced impairment of learning and memory [56,57]. Furthermore, it was also found an association between AR and cognition deficits [58]. In the present work, we did not find cognitive alterations associated with the AR. Indeed, these results were very consistent since we did not find alterations in acquisition process nor in the probe trial of Morris water maze test. Furthermore, when we analysed the percentage of time spent in target quadrant versus opposite quadrant there was also no alterations. All this together show clearly that our AR model did not induces changes of spatial learning and memory. We hypothesize that in our AR model the time that the animals lived with AR was not sufficient to affect cognition significantly and highlights the importance to take into account AR model induction protocols when evaluating and extrapolating results. To the best of our knowledge, there are only few studies that have focused on the effects of AR on aggressive behaviors. Tonelli and collaborators reported that AR did not change the aggressive behaviors, although social interaction of these rodents was changed [7]. In the present work, using the resident intruder test we found that AR rats were less aggressive when compared to control rats. These discrepancies between both studies could be due to different methodology or animal housing. In the present test, the animals stayed in their home cage during several days without changing the bedding in order to develop a sense of ownership and to clearly identify that cage as their home. Naturally, this process is most effective with all sensory organs working correctly and efficiently. In the AR animals, there is a loss of olfactory abilities and, therefore, it is natural that these animals might have more difficulty in developing the sense of possession of its cage and, consequently, will develop less aggressiveness towards   As reviewed by Hasegawa et al., behavioral changes caused by olfactory dysfunction have been described in various rodent models using for instance viral infection, nasal irritants, or air pollutants [59]. What links all these models, irrespective of the external trigger, is olfactory epithelial inflammation. As shown in these and our model, it is conceivable that this local nasal inflammation impacts not only central olfactory neural structures, but also neural circuitry associated with aggravated mood disturbances. However, the inflammation and its underlying pathophysiological processes in these various diseases and their symptomatology are very different. For instance, while intranasal lipopolysaccharides in rodents results in a Th1 cytokine profile in the OB (e. g. IL-1b, TNF-a and IL-6) [60], in models of allergic rhinitis a Th2 type cytokine profile (e.g. IL-4, IL-5, IL-13) is characteristic [7,61]. An elegant study by Hasegawa-Ishii et al. clearly show a big impact of nasal lipopolysaccharides on the synaptic structure in the EPL of the OB [32]. In our model however, while some changes were observed in the EPL and the ML, the GL is where the expression of all measured synaptic markers was significantly reduced. Mitral and tufted cells residing in the ML and EPL establish synapses within the GL and project to other cortical regions such as AON, PirC and PFC. How different inflammatory conditions impact on mitral and tufted cell neuronal circuitry and how this impacts other cortical regions is of particular interest. Our results indicate that the PFC synaptic structure is particularly affected, though no behavioral changes were detected other than those relying on olfaction.
Taken together, our immunofluorescent and behavioral data suggests that OB synaptic pathology associated with the allergic nasal inflammation results in downstream modulation of the synaptic structure of the AON, PirC and PFC ultimately leading to olfactory deficit which precipitates modulation of olfaction dependent behavior. Although we cannot rule out completely that this olfactory deficit and OB synaptic changes are retrograde changes following systemic allergic inflammation induced modulation of brain egions including the PFC and PirC. These are interesting data, however, when investigating human disease in animal models and interpreting data, one needs to be mindful and consider important differences between the species: rats are macrosmatic animals and the loss of olfactory function can be perceived as more threatening than in humans; it is more difficult to perform an evaluation of anxiety and humor disorders in animals; rats have a reduced lifespan and the relatively short time lived with AR might not be sufficient to develop the same co-morbidities presented in humans. Similar to other studies, we only used male rats, however we think that our conclusions can be extended to both genders because the investigated areas do not present sexual dimorphism that could bias our results [58,62]. We reproduced a sub-chronic AR model while human AR constitute a chronic disease and, as such, alterations arising from cumulative exposure could have been undervalued.
In conclusion, considering these caveats, in this study we evaluated the OB providing important new perspectives to understand other mechanisms, beyond nasal obstruction, that justify olfactory loss. This is the first study, to our knowledge, that evaluated the synaptic structure of OB and related projections to PirC and PFC, reinforcing the hypothesis that AR induces structural changes in olfactory-related neuronal circuits which affect behaviours that rely heavily on olfaction such as aggression towards intruders. Since many other CNS regions, such as the amygdala and hippocampus are related to olfactory function, we feel the possible repercussions of AR on these and other brain regions warrant further investigation. In turn, we also suggest to further analyze the olfactory function in AR rats under adequate treatment either topically or systematically, to fully understand the reversibility of these changes in the olfactory pathway and their dependence on inflammation and/or olfactory dysfunction.

Funding
This article was supported by National Funds through FCT -Fundação para a Ciência e a Tecnologia,I.P., within CINTESIS, R&D Unit (reference UIDB/4255/2020) and within the scope of the project RISE, Associated Laboratory (reference LA/P/0053/2020).

Declarations of Competing Interest
None.

Data availability
Data will be made available on request.