Tumour Necrosis Factor-α Inhibition Improves Stroke Outcome in a Mouse Model of Rheumatoid Arthritis

Rheumatoid Arthritis (RA) is a chronic inflammatory disorder where incidence and severity of myocardial infarction are increased. Data on the incidence and outcome of stroke are conflicting. Thus, we investigated outcome after Ischemia/Reperfusion (I/R) brain injury in a mouse model of RA and assessed for the role of the tumour necrosis factor-α (TNF-α) inhibitor Infliximab herein. We used a TNF-α reliant mouse model of RA. RA and wildtype (WT) animals were treated with vehicle (RA/WT) or Infliximab (RA Infliximab) for 4 weeks, before undergoing I/R brain injury. RA-animals displayed larger strokes and poorer neurological performance. Immunohistochemistry on brain sections revealed increased numbers of resident and peripheral innate immune cells (microglia and macrophages); increased Blood-Brain-Barrier (BBB)-disruption; decreased levels of the tight junction proteins (TJPs) claudin-5 and occludin; increased expression of matrix-metalloproteinases (MMP)-3 and -9 and enhanced lipid peroxidation. Treatment with Infliximab corrected these alterations. We show that RA associates to worse stroke-outcome via exacerbated BBB degradation by decrease of the TJPs claudin-5 and occludin. We identified MMPs-3 and -9 and increased oxidative stress as potential mediators thereof. Increased numbers of resident and peripheral innate immune cells (microglia and macrophages) may in turn contribute to all these effects. Infliximab-treatment restored the phenotype of RA-mice to baseline. Our data provide evidence clearly linking RA to adverse stroke-outcome in mice and indicate an approved TNF-α inhibitor as a potential strategy to reduce stroke-burden in this setting.

Rheumatoid arthritis (RA) is an immune-mediated, chronic inflammatory disorder. With a prevalence of ~1%, it ranks among the top 15% of diseases causing disability worldwide 6 . Apart from debilitating articular effects, associated systemic complications reduce median survival by 17 years 7 . Cardiovascular mortality is hereby increased by about 50% [8][9][10][11] . Particularly, the risk for myocardial infarction (MI) is increased by at least 2-fold and acute coronary syndromes in RA patients are clinically more severe and associate to higher fatality rates 12,13 . While the epidemiology of MI in RA is well characterized, the one of stroke is less defined with some studies reporting an increased risk 11,[14][15][16] and others finding no association [17][18][19][20] . Also, data on stroke outcome are conflicting, with some studies showing increased mortality rates and others not 7,10,[21][22][23][24][25][26] . Meanwhile, data on clinical stroke severity and presentation are sparse.
TNF-α inhibitors, such as the monoclonal TNF-α antibody Infliximab, are clinically approved for the treatment of RA which remains active despite therapy with disease modifying anti-rheumatic drugs. TNF-α can play a dual role in stroke, promoting inflammatory stroke progression on one hand and mediating cerebral tolerance to hypoxia and ischemia on the other. Therefore, the potential effect of TNF inhibitors in RA patients with stroke is far from obvious 27 .
We hereby investigated outcome after I/R induced brain injury in a mouse model of RA and assessed for the role of Infliximab in this setting.

Methods
Animals. Sixteen weeks old male and female TNF-α transgene over-expressing mice on a CBA x C57BL/6 hybrid background were used as a murine model for RA 28 . Briefly, a 2.8 kb fragment of the human TNF-α genomic sequence comprising 0.6 kb of 5′ regulatory sequences, introns and exons up to the stop codon was used for the transgene. The 3′ region of the human TNF-α gene was replaced with that of the human β-globin gene, resulting in TNF-α dysregulation and pathology development 28 .
The RA mouse model exists in two severity degrees, depending on the copy number of the transgene. The more severely affected TG197 line expresses five copies, while the milder TG3647 line expresses only one 29,30 . At 6-8 weeks of age, the TG3647 line develops an arthritic phenotype with 100% penetrance. Symptoms chronically progress over a possible lifespan of about 1 year. Therefore, the line well-reflects chronic, adult-onset RA and allows for the study of advanced stages of RA. Anti-TNF-α treatment was shown to correct the arthritic phenotype of TG3647 mice 31,32 . To reduce post-surgical mortality and suffering, the milder TG3647 mouse line was used in thus study. The TG3647 mice will be referred to as RA-mice throughout the manuscript.
Breeding was maintained by crossing transgenic males with wild type (WT) females. Animals were maintained at 24 °C under a 12 h light/dark cycle and were fed a normal chow diet with ad libitum access to food and water. Study design and experimental protocols were approved by the institutional animal care committee (License N° TVA 077_2016; Kommission für Tierversuche des Kantons Zürich, Switzerland). All protocols were carried out in accordance with the relevant guidelines and regulations.
Infliximab treatment. RA-mice of both sexes were randomly assigned to either weekly intraperitoneal injections of Infliximab (Remicade ® , MSD AG, NJ, USA) at a clinically relevant dose (10 mg/kg bodyweight) for 4 weeks between weeks 12 to 16 (RA Infliximab), or an equivalent regimen of distilled water (RA). Wildtype animals received the same vehicle treatment (WT). Infliximab was acquired from the Zurich cantonal pharmacy.
Clinical scoring. During the treatment period from week 12 to 16, animals were scored weekly using a clinical score for arthritic symptoms, social behaviour and bodyweight-loss.
The arthritic joint assessment was based on a previously published scoring system 33 : 0 (normal); 1 (edema or reddening of paw and/or ankle joints); 2 (distortion of paw and/or ankle joints); 3 (subluxation or ankyloses of paw and/or ankle joints).
Scores from each of the three categories were summed up to obtain the final score.

Transient Middle Cerebral Artery Occlusion. Transient Middle Cerebral Artery Occlusion (tMCAO)
was performed to induce I/R brain injury, as previously described 34,35 . Briefly, mice were anaesthetized using isoflurane 4% and 1.5% for induction and maintenance respectively. Body temperature was maintained at 37 °C. For analgesia, buprenorphine-HCL (0.1 mg/kg) was infiltrated subcutaneously at the incision site 30 minutes prior to surgery. The common, internal and external carotid arteries were dissected and a 6-0 silicone-coated filament (Doccol Corporation, Sharon, MA, USA) inserted into the common carotid artery and advanced to the origin of the left MCA to induce tMCAO for 45 min. After ischemia, middle cerebral artery (MCA) reperfusion was allowed for 24 h. During this time, animals were carefully observed and received analgesia with buprenorphine-HCL at a dose of 0.1 mg/kg s.c. every 6 hours. After 24 h of reperfusion, mice were sacrificed using an overdose of CO 2 .
stroke Volume. After euthanasia, mice were perfused with 10 ml of phosphate buffered saline (PBS) and relevant organs were excised. Murine brains were cut into 6 equally spaced (1 mm) coronal sections and immersed in a 2% solution of 2,3,5-triphenyltetrazolium chloride (TTC) (Sigma-Aldrich, Chemie GmbH, Buchs, Switzerland) at 37 °C for 20 min. TTC is a redox indicator, which changes its colour from white to red, as it gets reduced to 1,3,5-triphenylformazan in viable tissues. In metabolically inactive tissues lacking cellular respiration, this reduction does not take place, leaving TTC in its unreacted state. Therefore, the stroke area can be distinguished from viable brain tissue by its white appearance 36 . Areas of infarction, ipsilateral and contralateral hemispheres were quantified using ImageJ software (Image J, NIH, MD, USA). To correct infarct size measurement for cerebral edema and consequent overestimation, we applied the following formula as previously described 37 : Corrected infarct volume = contralateral hemisphere volume -(ipsilateral hemisphere volume -infarct volume). Infarct size was expressed as volume in mm 3 .
Neurological Deficiency Assessment. Baseline and post-infarction neurological status were assessed by a four-point scale neurological score according to Bederson et al. 38 and the RotaRod test as previously described 34 . Both tests are composite sensory-motor tests, which take into account motor functions, proprioception, spatial orientation and balance 39 . The neurological score test according to Bederson was performed at baseline, 2 h and 24 h after reperfusion and consists of the following scores: grade 0, normal neurological function; grade 1, forelimb and torso flexion on and towards the contralateral side upon lifting of the animal by the tail to 1 m above the work surface; grade 2, circling to the contralateral side; grade 3, leaning to the contralateral side at rest; grade 4, no spontaneous motor activity. The RotaRod test was performed at baseline and 24 h after reperfusion. To this purpose, mice were placed on a rotating rod at increasing speed and the time to tendency to fall was measured in seconds. Per animal, three consecutive measurements were performed at each time-point and the best score was used.
Histology. Histological procedures were performed as previously described 40,41 . Briefly, following euthanasia, mice were exsanguinated by cardiac puncture and transcardially perfused with PBS and 4.0% paraformaldehyde (PFA) (Sigma-Aldrich, Chemie GmbH, Buchs, Switzerland) in PBS at room temperature. The brains were removed and consecutively incubated overnight in 4.0% PFA at 4 °C and afterwards transferred to 30% sucrose in PBS for 36 h. Cryoprotected brains were cut into 100-µm thick free-floating sections using a microtome (Leica Jung HN40), pre-treated with proteinase K or 1 M HCL for antigen retrieval and immune-blocked with 10% donkey serum. After these steps, they were incubated with primary antibodies at the following dilutions (1:500; Abcam, Cambridge, United Kingdom) at 4 °C overnight, respectively. Secondary antibodies were added at a dilution of 1:750 (Jackson Immunoresearch, West Grove, PA, USA) for 24 h at 4 °C. Images were acquired using a confocal microscope (Leica SP8; Leica, Wetzlar, Germany). Cells positively stained for the microglial and activated macrophage marker Iba-1, were counted in a specified ipsilateral area (the CA1 hippocampal area) using ImageJ software. Stained areas of claudin-5, occludin, VE-cadherin, MMP-3 and MMP-9 were measured using ImageJ software (Image J, NIH, MD, USA) and normalized to the total endothelial cell surface area assessed by CD31 staining. The area positively stained for 4-HNE on the ipsilateral hemisphere was measured using ImageJ software (Image J, NIH, MD, USA) and expressed as a percentage of the contralateral hemisphere.
BBB permeability was assessed by quantifying endogenous immunoglobulin G (IgG) extravasation. Sections were incubated with Alexa647-conjugated donkey anti-mouse IgG for 24 h (1:600; Jackson Immunoresearch, West Grove, USA). IgG-stained area was expressed as a percentage of the contralateral hemisphere.
Statistical Analysis. Data were expressed as mean ± standard error of the mean (SEM). All statistical analyses were performed using GraphPad Prism 7 software (GraphPad Software, Inc, La Jolla, CA, USA). Results were confirmed to follow a normal distribution with the Kolmogorov-Smirnov test of normality. Data that passed the normality assumption were analyzed with two-tailed unpaired Student's t-test and data that failed the normality assumption were analyzed with the nonparametric Mann-Whitney U test. For repeated measurements, two-way ANOVA with Sidak post hoc test was applied. A probability value (P) below 0.05 was considered statistically significant. For the graphical depiction of p-values in the figures, the following applies: *P < 0.05, **P < 0.01, ***P < 0.001.

Results
RA mice develop a progressive arthritic phenotype which can be averted by Infliximab. RA mice displayed a progressive increase in the composite clinical score over time. At the age of 12 weeks, RA animals of both the treatment and the control groups showed relatively mild arthritic symptoms. Thereafter, symptom severity steadily increased in the vehicle-injected animals, while Infliximab treatment could blunt the increase (week 14: RA 2.44 ± 0.53, n = 9 vs. RA Infliximab 1.44 ± 0.52, n = 9; p < 0.0001; week 15: RA 3 ± 0.70, n = 9 vs. RA Infliximab 1.33 ± 0.50, n = 9; p < 0.0001; week 16: RA 3.44 ± 0.72, n = 9 vs. RA Infliximab 1.11 ± 0.33, n = 9; p < 0.0001; Fig. 1A). At all time-points, treated and untreated RA animals scored significantly higher than WT animals, where the clinical score remained 0 for the whole duration of the study.

RA mice display increased numbers of microglia and activated macrophages.
Microglia as the resident immune cells of the brain, are the first to respond to exposure of damage associated molecular patterns (DAMPs) following ischemia. Upon activation, they can secrete pro-inflammatory cytokines such as TNF-α and IL-1β and thereby exacerbate neuronal injury and para-cellular permeability. Ionized calcium binding adaptor molecule 1 (Iba-1) is a marker for both microglia and activated macrophages. In order to assess the contribution of microglia and invading macrophages (M/M) to the observed stroke phenotype, we have measured Iba-1 as a surrogate for these cells in the ipsilateral hemisphere.

RA mice display increased BBB disruption and tight junction protein (TJP) degradation. I/R-induced
BBB breakdown is characterized by extravasation of large circulating molecules, such as IgG into the affected parenchyma 42 . A disrupted BBB allows for the formation of vasogenic edema and the invasion of systemic inflammatory cells, which both drive stroke progression. It is thus an important mediator of unfavourable stroke outcome. As to be expected, we observed some IgG extravasation into ischemic hemispheres in all experimental groups, whereas the contralateral hemispheres remained unaffected.
Integrity of the endothelial component of the BBB is regulated by tight and adherens-junctional proteins (TJP). During stroke, paracellular permeability increases due to the degradation of these proteins. Here, we assessed the integrity of the 2 TJP occludin and claudin-5 as well as the adherens junction protein VE-cadherin by immunostaining.

Discussion
RA is a chronic systemic, immune-mediated inflammatory disorder. Apart from debilitating articular effects, RA associates to a decreased median survival 7 , mainly driven by cardiovascular complications. In particular, the frequency, clinical severity and mortality of acute coronary syndromes are increased 12,13 . Data on incidence and mortality of stroke in RA patients are more conflicting 16,19 , while data on clinical severity are sparse. Moreover, the influence of biologicals like Infliximab on stroke outcome in RA patients remains unclear 27 . Filling this knowledge gap is crucial for designing personalised therapeutic approaches to improve prognosis in these patients. To this end, we aimed to investigate the morphological and functional outcome after I/R brain injury in a RA mouse model.
RA-mice displayed an increased stroke volume by approximately 60% compared to WTs. The physiological relevance of these findings was supported by a greater post-ischemic neurological impairment in RA animals. Infliximab-treatment of RA mice not only improved arthritic symptoms, but also reduced stroke volumes and neurological dysfunction to WT levels, thereby confirming the relevance of inflammation and in particular, of the cytokine TNF-α.
Microglia are the resident brain immune cells and as such, the first line of response to damage associated molecular patterns exposed upon ischemia 3 . In previous studies, they have been shown to be important secretors of pro-inflammatory cytokines such as TNF-α and IL-1β in the brain 44 . Furthermore, microglia-derived TNF-α was shown to induce apoptosis in neuronal precursor cells 45 , attract peripheral immune cells 46 and facilitate extravasation of immune cells, platelets, erythrocytes and fibrin by increasing endothelial adhesion molecule expression 47,48 . Moreover, in an attempt at neo-vascularization upon ischemia, microglia along with endothelial cells and astrocytes increase the generation of matrix metalloproteinases, which in turn contribute to BBB dysfunction by degrading extracellular matrix and tight junction proteins 49 .
We found increased numbers of activated microglia and macrophages in the stroke hemispheres of RA animals, when compared to WT. Treatment with Infliximab meanwhile significantly decreased the post-ischemic counts of these cells, both compared to untreated RA animals and WT mice. These findings suggest not only a pertinent role of resident and peripheral immune-cells in mediating the observed stroke phenotype, but also a role for TNF-α in activating and recruiting these cells to the site of injury.
BBB disruption is a hallmark of stroke and a mediator of brain injury and stroke progression 50 . This notion is supported by data showing BBB-breakdown as early as 30 minutes after ischemia 51 . Inflammatory cytokines play a crucial role in this respect. Indeed, TNF-α and IL-6 were shown to dose-dependently decrease tightand adherens-junction expression in vitro 52 and to be increased in ischemic hemispheres of rats after tMCAO 53 . Furthermore, TNF-α serum-and cerebrospinal fluid-levels are increased in stroke patients and positively correlate to stroke severity 54 .
We found that RA mice display an increased BBB disruption after stroke as compared to WTs. As possible mediators thereof, we describe reduced protein levels of the TJPs claudin-5 and occludin. These transmembrane proteins are the backbone of cerebral endothelial TJs, which uphold three key functions: blocking the paracellular diffusion of blood-borne polar substances; maintenance of cell polarization by preventing lateral diffusion of lipids and membrane proteins and the provision of an intracellular signalling platform. Generally, distancing of these proteins from the cellular borders or decreases in their expression at the TJ cleft mediate a loss of junctional integrity and increased paracellular permeability, which in turn directly contributes to vasogenic edema, haemorrhagic transformation and increased mortality 55,56 . Infliximab-treatment of RA mice restored TJP expression to WT levels and in doing so, also conserved BBB integrity as demonstrated by diminished IgG extravasation.
MMPs are proteolytic enzymes degrading all components of the extracellular matrix. In cerebral ischemia, they can mediate early degradation of TJPs 50,57 . The main inducible MMPs in the brain are MMP-3 and MMP-9 58 . Thus, we investigated their putative involvement in the observed BBB-breakdown and found that expression levels of the active enzymes were significantly increased in ischemic hemispheres of RA animals and that these alterations could be reversed by Infliximab, supporting the crucial role of inflammation in post-ischemic MMP expression.
Both MMPs-3 and -9 are secreted in a latent form to undergo activation by nitrosylation or oxidation 59 , whereas active MMP-3 can also directly activate MMP-9. Additionally, oxidative stress has permeabilizing actions of its own and is subject to inflammatory stimuli. Thus, we investigated oxidative stress levels and showed that Scientific RepoRts | (2019) 9:2173 | https://doi.org/10.1038/s41598-019-38670-z ipsilateral hemispheres of RA animals present significantly more lipid peroxidation -a marker of oxidative stressas compared to WT, while Infliximab-treatment blunted this increase. This indicates that oxidative stress partially links inflammation to BBB disruption in our model. The overall-beneficial effects of Infliximab not only underline the pertinence of inflammation to the observed changes in RA mice, but also implicate this drug as a potential modulator of stroke in RA-patients.
One limitation of the herein reported study is represented by the use of vehicle as a control to Infliximab as opposed to using an isotype control antibody. IgG are capable of regulating the immune response through Fc binding alone 60 thus, the optimal control would have to be an isotype antibody. However, the key role of TNF-α in the pathogenesis of RA is understood [61][62][63] ; accordingly, anti-TNF-α treatment is well-established and beneficial in RA patients 64 . Taken together, this should confer additional confidence that the observed effects in our RA mouse model are specific to the antibody used.
In conclusion, results from this study contribute additional mechanistic evidence to epidemiological data concerning the link between RA and stroke. They call for further clinical studies in this direction to allow for the design of personalised treatment strategies, which could involve tailored use of biologicals such as Infliximab.