P301L-Tau mutation alters bioenergetic functions of neurons and these

Autophagy has an important role in cellular respiration through quality control of mitochondria (Ding, Gao et al. 2015). Because bexarotene had a significant effect on autophagy flux, primary cortical neurons were exposed to a mitochondrial stress with simultaneous measurement of oxygen consumption rate. P301L-Tau transgenic cortical neurons had a significantly lower ATP and oxidative phosphorylation-linked oxygen consumption rates (Figure 8A). These genotype-related differences in respiratory performance were normalized with bexarotene-pretreatment (Figure 8B). Despite the beneficial effect on cellular bioenergetics in primary cortical neurons, bexarotene did not affect ROS production of N2a cells as measured by the CellRox flow cytometry assay (Figure S5).

5.2 COPPER BIS(THIOSEMICARBAZONE) COMPLEXES MODULATE EXPERIMENTAL NEUROINFLAMMATION IN VITRO AND IN VIVO BY INCREASING METALLOTHIONEIN 1 (II)

We performed a study (II) to assess the therapeutic effect of copper delivery in acute (initiated with peripheral LPS injection into mice) and chronic inflammatory conditions (mice with AD pathology). The inflammatory status of the animals was measured using VCAM-1 expression with contrast MRI in acute inflammation, and with RT-PCR for cytokines in AD mice.

Primary microglia and astrocyte cultures were used to detect treatment effects on individual cell types using PCR, cytokine measurements, NO measurement, mass spectrometry and microscopy.

5.2.1 Copper bis(thiosemicarbazone) complexes scavenge acute and chronic neuroinflammation

Peripheral LPS administration is known to induce inflammation also in brain which in turn increases VCAM-1 expression in endothelial cells reflecting inflammatory status of brain (Gamal, Moawad et al. 2015). More importantly, in a neuroinflammation model of peripheral LPS injection VCAM-1 expression correlates with behavioral deficits thus allowing the peripheral monitoring of both CNS inflammation and brain function (Gamal, Moawad et al.

2015). VCAM-1 expression was measured by conjugating a VCAM-1 antibody to microsized particles of iron oxide (MPIO), injecting this conjugate intravenously into animals, and measuring hypointense brain areas in MRI images. Peripheral LPS induced a clear increase in brain signal area, indicating elevated vascular VCAM-1 expression 24 h after induction of peripheral inflammation (Figure 1). The signal was significantly reduced by p.o.

administration of Cu^II(atsm) 2 h after LPS injection.

Another copper delivery complex, Cu^II(gtsm), was previously demonstrated to have a beneficial effect in AD mouse models (Crouch, Hung et al. 2009). The anti-inflammatory potential of this compound was tested in APdE9 mice modelling AD by using quantitative RT-PCR due to lack of MRI-detectable VCAM-pathology. Results show that transcription of anti-inflammatory proteins Arginase-1 and TGF-β was increased in Cu^II(gtsm)-treated APdE9-animals (Figure 2). Transcription of pro-inflammatory MCP-1, IL-1, NOS2 and TNFα remained unaltered (data not shown). Taken together, these results indicated that copper delivery has anti-inflammatory effects in both acute and chronic inflammation.

5.2.2 Both microglia and astrocytes gain a less pro-inflammatory phenotype in the presence of Cu^II(atsm) in vitro

Cell-specific anti-inflammatory effects were tested using primary microglia and astrocyte cultures. An inflammatory milieu was induced in these models by the addition of pro-inflammatory cytokines IFN-γ and TNFα, as well as LPS (for astrocytes). Minocycline was used as a positive control as it has anti-inflammatory effects on microglia (Tikka, Fiebich et al. 2001). Co-stimulation of microglia with IFN-γ and TNFα increased NO production, which was reduced with low-dose co-treatment with Cu^II(atsm) (Figure 3). Cu^II(atsm) co-treatment also reduced MCP-1 transcription and secretion into the media, and the compound was more potent than minocycline in this assay. In addition, TNFα expression was reduced by Cu^II(atsm) treatment.

Further studies with astrocytes revealed that Cu^II(atsm) also reduced astrocytic production of MCP-1 during IFN-γ and TNFα co-stimulation (Figure 4). This effect was similar during LPS stimulation, but in addition, Cu^II(atsm) also reduced IL-6 secretion. Cu^II(atsm) was toxic for neither microglia nor astrocytes based on MTT assays (Figures 3 and 5).

5.2.3 Cu^II(atsm) increases the copper content of cytokine-stimulated cells

Increased copper content by Cu^II(atsm) treatment in primary microglia and neonatal astrocytes during IFN-γ and TNFα co-stimulation was confirmed with ICP-MS (Figure 6), indicating that the complex delivers copper into cells. Furthermore, X-ray fluorescence was used to demonstrate an elevation in microglial copper content during combined Cu^II(atsm) and IFN-γ/TNFα exposure.

5.2.4 Anti-inflammatory effects of Cu^II(atsm) are mediated by metallothionein 1

As a metal-binding protein with anti-inflammatory and antioxidant properties, metallothionein 1 was a suspected mediator of the beneficial effects of Cu^II(atsm). Indeed, both cytokine stimulated and unstimulated primary microglia treated with Cu^II(atsm) had increased levels of MT1 mRNA (Figure 7). Moreover, the anti-inflammatory effect of Cu^II(atsm) was abolished in primary microglia treated with propargyl glycine, an inhibitor of metallotionein synthesis. In this case, both the reduction in MCP-1 levels and NO release were absent, indicating that metallothionein 1 is important in the observed anti-inflammatory effect of Cu^II(atsm).

5.3 CU^II(ATSM) IMPROVES OUTCOME IN PERMANENT AND TRANSIENT MODELS OF CEREBRAL ISCHEMIA AND MODIFIES MICROGLIAL ACTIVATION (III)

In study (III) we tested therapeutic potential of Cu^II(atsm) in transient and permanent models of ischemic stroke in mice. The extent of brain injury was measured with MRI, histochemical stainings and behavioral assessment. We also used primary cortical neurons to study the effect of Cu^II(atsm) on glutamate-induced excitotoxicity. The effect of Cu^II(atsm) on post-stroke neuroinflammation was further analyzed by various methods, including immunohistochemistry, FACS and CBA.

5.3.1 Cu^II(atsm) protects cortical neurons against excitotoxicity and N2a cells against OGD in vitro

The effect of Cu^II(atsm) on excitotoxicity was studied using primary cortical neurons. In this assay, exposure to 400 µM resulted in 40 % decrease in cell viability in 24 h (Figure 1A). Co- or 2-h post-treatment with Cu^II(atsm) protected neurons partially from glutamate-induced excitotoxicity based on MTT assay, bringing the cell viability up to 70 %. Similarly, viability of N2a cells exposed to 24h OGD was reduced by 50 %, and 0,1 µM Cu^II(atsm) pretreatment had a slight protective effect, increasing the viability to 55 % (Figure 1B).

5.3.2 Cu^II(atsm) is protective in a mouse model of transient ischemia

Cu^II(atsm) was first tested in the transient MCAO model of cerebral ischemia. We found that Cu^II(atsm) treatment (15 mg/kg) both prior to MCAO and at the start of reperfusion reduced infarct sizes 24 h later (Figure 2). MCAO-induced functional impairment (assessed with neuroscore with a 5-point scale) was also alleviated by pre-treatment with Cu^II(atsm).

5.3.3 Cu^II(atsm) reduces ischemic damage in a permanent model of stroke

The therapeutic potential of Cu^II(atsm) was also tested in a permanent ischemia model. It was found out that p.o. 2 h post-treatment with 60 mg/kg significantly reduced lesion size as assessed by MRI at 24 h after stroke (Figure 3). In addition, co-treatment with Cu^II(atsm) during ischemia improved neurological status of the animals by reducing time spent in the latency to move testing at 1 dpi. Cu^II(atsm)-mediated copper delivery to the brain was confirmed using ICP-MS - animals treated with the Cu^II(atsm) complex had a significantly higher concentration of copper in the peri-ischemic area than vehicle treated controls.

5.3.4 Cu^II(atsm) reduces CD45 expression in cells located in the ischemic core

Copper bis(thiosemicarbazone) complexes have been described to modulate inflammatory reactions in animal models of neurological diseases (Crouch, Hung et al. 2009). To study the anti-inflammatory potential of Cu^II(atsm) in ischemia we applied immunohistochemical techniques to brain sections collected from animals with permanent MCAO. Astrocytic reactivity was not affected based on GFAP staining (data not shown). Instead, Cu^II(atsm) treatment significantly reduced the amount of CD45 immunoreactivity in the ischemic core both at 1 and 3 dpi (Figure 4). CD45 is highly expressed by monocytes and a lower expression is typical for reactive microglia.

5.3.5 Iba1 expression is reduced and the morphology of Iba-1 positive cells is altered by Cu^II(atsm) treatment after ischemia

Next, a set of immunohistochemical stainings were performed to distinguish the inflammatory cell types affected by Cu^II(atsm) treatment in mice that underwent permanent MCAO. No differences between treatment groups were found based on cells expressing CD68 (marker used for phagocytotic cells) or Arginase-1 (alternatively activated cells) (data not shown). However, reduced Iba1 immunoreactivity was observed 3 dpi in the peri-ischemic area of Cu^II(atsm) treated animals (Figure 5). Because microglial morphology and function are linked, a computer algorithm was used to analyse the morphology of Iba1-positive cells in detail. Results indicate that Cu^II(atsm) did not affect cellular branches, but significantly increased the cellular area of Iba-1 positive cells.

5.3.6 The proportion of resident microglia is increased in Cu^II(atsm) treated ischemic brains

To gain further information about the alterations in proportions of different inflammatory cell populations in the ischemic brain following Cu^II(atsm) treatment, cells were isolated from the ischemic hemisphere and analysed by flow cytometry 3 days after permanent MCAO.

The proportions of multiple cell types, such as lymphocytes, neutrophils and myeloid cells, were affected by stroke. Cu^II(atsm) increased the proportion of resting microglia, characterized by CD45^low CD11b⁺ Ly6G^- indicating reduced microglial activation, increased viability or increased proliferaton (Figure 6).

5.3.7 Cu^II(atsm) has beneficial effects on further inflammatory markers after ischemia Finally, immunohistochemistry and CBA were used to detect the effects of Cu^II(atsm) on inflammatory cascades. Activated p38 MAPK, a major signalling molecule mediating inflammatory activation in microglia, was found to be reduced in Cu^II(atsm) treated animals 1 day after permanent MCAO (Figure 7). We also saw a reduction in IL-12 levels at the same time point and increased IL-10 levels 3 days after ischemia.

5.4 COLLAGEN XV DEFICIENCY AMELIORATES ISHCEMIC DAMAGE IN MICE (IV)

Study (IV) aimed to assess the role of collagen XV in thromboembolic stroke. Sham and stroke animals received either vehicle or rtPA and underwent MRI for lesion size quantification.

Western blotting, cytokine measurements and immunohistrochemistry were used to detect genotype and treatment-related differences in cytokines and proteins (e.g. collagen XV and VEGF-A).

5.4.1 Α1-Collagen XV deficiency protects against thromboembolic stroke

A mechanistically relevant model of thromboembolic stroke was used to test rtPA-treatment and genotype effect on MRI-measured outcome in mice 2 days after stroke. Thrombolysis with tPa 20 minutes after MCAO significantly reduced lesion size in wildtype animals (Figure 1). However, α1-ColXV KO mice had significantly smaller lesion sizes than wildtype animals and there was no therapeutic effect of rtPA on lesion size in these animals.

To exclude obvious genotype-mediated reasons for the observed protection, cerebral vasculature, blood parameters and cerebral edema of the mice were assessed. No differences were detected between treatment groups in either cerebral vasculature or blood parameters (such as partial pressures of blood gases, Figure 2). Also, brain swelling and aquaporin-4 immunoreactivity (water channel contributing to increased edema) following stroke were similar in all groups (Figure 3).

5.4.2 RtPA increases Collagen XV levels in the plasma of WT mice but not in α1-Collagen XV KO mice.

Next, we used ELISA to measure the protein levels of ColXV in the plasma of WT and α1-ColXV KO mice with or without rtPA treatment 3 days after stroke. Interestingly, we found

that rtPA tends to increase α1-ColXV protein levels after ischemia in the plasma of WT mice (Figure 4).

5.4.3 rtPA treatment does not reduce post-stroke cytokine production in α1-Collagen XV KO mice

Immunohistochemical examination of brains with markers for astrocyte and microglial/monocyte activation (GFAP and Iba1) at 3 days after stroke did not show any treatment or genotype effects. However, CBA from protein extracts derived from ipsilateral and contralateral cortex revealed an ischemia-induced elevation in IL-6 and CCL2 levels at 3 days after stroke (Figure 5). Levels of IL-6 and CCL2 were reduced by rtPA in wildtype animals. Of particular interest was the finding that rtPA failed to reduce increased cytokine levels in α1-Collagen XV KO mice.

5.4.4 Α1-Collagen XV deficiency and rtPa treatment results in a similar pattern of increased VEGF-A expression during ischemia

Type A vascular endothelial growth factor (VEGF-A) is a potent neuroprotective growth factor in the CNS. Western blotting results showed that rtPA treatment increases VEGF-A expression in wildtype animals (Figure 6). Importantly, α1-Collagen XV deficiency caused similar increase in VEGF-A expression and this increase was not connected to rtPA treatment.

6 Discussion

Ischemic stroke is one of the leading causes of death worldwide, and long-term disabilities and reduced quality of life are common among stroke survivors. The current treatment regimen of ischemic stroke relies mostly on pharmacological or mechanical recanalization of the occluded artery. As the primary treatment option, thrombolysis with rtPA still has limitations considering timeframe and increased risk for hemorrhages. Moreover, successful recanalization does not necessarily mean reperfusion due to the no-reflow phenomenon. This is one example indicating that the neural network undergoes a dramatic change during the ischemic period, and the damaged tissue may not respond to restored blood flow. Thus, future treatment options for ischemic stroke should aim to maintain normal function of all the pieces of the network and target cascades beyond those mediated by rtPA treatment, such as inflammation. In addition, models used for preclinical testing of potential compounds should include real life factors affecting the neuronal-glial-vascular unit function, e.g. high age and the presence of other diseases. In the present study we have attempted to address the above mentioned issues by (1) using more complex models for testing drugs and drug candidates that have previously shown preclinical efficacy, (2) inspecting the protective potential of compounds with mechanism of actions beyond neurons in addition to the direct neuroprotective effect, and (3) studying role of collagen XV in the therapeutic effect of rtPA.

We studied three treatments for novel indication in animal and cell culture models of ischemic stroke, inflammation and AD. First, we studied the efficacy and mechanism of action of an RXR agonist, bexarotene, in a mouse model combining high age, tau pathology and ischemic stroke. Next, we assessed the effects of copper delivery agents, the bis(thiosemicarbazone) complexes, in in vivo and in vitro models of inflammation and AD.

Third, we determined that a certain copper bis(thiosemicarbazone) complex, Cu^II(atsm), is protective in two in vivo models and in in vitro model of stroke. Finally, we provide new insight into the role of a basal membrane protein, collagen XV, in thromboembolic stroke.

6.1 MODULATION OF AUTOPHAGY BY BEXAROTENE PROTECTS OLD MICE EXPRESSING P301L-TAU AGAINST STROKE

Bexarotene has been demonstrated to have beneficial effects in a wide range of preclinical models of neurodegenerative diseases (Cramer, Cirrito et al. 2012, Bomben, Holth et al. 2014, McFarland, Spalding et al. 2013, Riancho, Ruiz-Soto et al. 2015). Bexarotene-mediated protection has also been shown in two stroke studies using young, healthy male animals (Certo, Endo et al. 2015, Xu, Cao et al. 2015). To fully understand the therapeutic potential of a medicine in stroke, the model should mimic aspects of human stroke: both genders, ageing, comorbidity and mechanistic relevance. Thus, our study utilized old male and female mice expressing P301L-Tau subjected to thromboembolic stroke.

Our results show that bexarotene reduces lesion size and improves motor coordination after ischemic stroke only in old P301L-Tau animals. While there was no indication of genotype-related differences in lesion size 3 days after stroke, gait analysis revealed more

stroke-induced deficits in P301L-Tau animals 7 days after stroke. Thus, it is possible that transgenic animals had more severe pathology, since lesion volume and neurological outcome in stroke patients are loosely connected (Pineiro, Pendlebury et al. 2000). Previous studies have shown that young P301L-Tau animals have reduced damage in hypoxia/ischemia injury (Liao, Zhou et al. 2009). The reason for divergent results might be explained by the thromboembolic model used and the aggravated pathology in aged P301L-Tau mice compared to wildtype animals. In fact, our results also demonstrate that in a simple in vitro model of embryonic cortical neurons, the cells from wildtype and P301L-Tau mice are equally protected from excitotoxic insult by bexarotene.

ApoE driven clearance of Aβ was shown to mediate the therapeutic effect of bexarotene in mouse models of AD (Cramer, Cirrito et al. 2012). Surprisingly, bexarotene did not affect total or phosphorylated tau or Aβ, even though ApoE levels of bexarotene treated animals were elevated. In our study, we did not observe treatment-dependent differences in MMP-9 levels contributing to BBB permeability even though this process can be manipulated by increasing ApoE with bexarotene treatment (Xu, Cao et al. 2015). In addition, both peripheral and neural inflammatory mediators that were measured in this study remained unaltered between treatment groups, even though bexarotene has been shown to affect neutrophil infiltration in a transient filament stroke model (Certo, Endo et al. 2015). It is likely that the model used in the present study does not cause a tremendous or long-lasting effect on the peripheral immune response. In fact, even though splenic responses are a common finding in animal models of stroke, they are evident in only approximately 40 % of stroke patients (Vahidy, Parsha et al. 2016). It is also evident that different administration routes (intraperitoneal in previous study, oral gavage in the present study) result in different drug kinetics and concentration in blood leading to different splenic and inflammatory reactions.

Altered autophagy is a common finding in neurodegenerative diseases (Nixon 2013). This is also true in studies, which have inspected the combined effect of neurodegeneration and ischemia. As an example, the 3xTg-AD mouse model (including P301L-Tau) that underwent transient hypoperfusion had activated autophagy in the brain (Koike, Green et al. 2010). In our study, P301L-Tau transgenic mice had a drop in a key autophagy marker, LC3b-II, and this drop was normalized with bexarotene treatment. The changes in LC3b-II were coincident with reduced lesion volume. At a later time point, LC3b-staining revealed accumulation of LC3b in neurons of bexarotene-treated P301L-Tau mice.

Because accumulation of LC3b-II may be a consequence of both altered autophagosome synthesis and turnover, we used an N2a cell line to investigate the effect of bexarotene on autophagy flux. Bexarotene treatment induced both LC3b-II and p62 accumulation in P301L-Tau transgenic cells suggesting reduced autophagosome degradation. Furthermore, the autophagy flux assay revealed that the process of autophagy was slowed down by bexarotene treatment in transgenic cells. Overall, we hypothesize that induced autophagy may be a compensatory mechanism to clear abnormal tau in our model, and that exposure to bexarotene might release this pressure to autophagy and slow down autophagic flux.

Autophagy is an important process in mitochondrial quality control (Chen, Sun et al.

2014). In our study, the mitochondrial stress test revealed deficits in ATP production and maximal respiration of primary cortical neurons with P301L-tau mutation without affecting ROS production. Similar findings have been reported previously by other groups (David, Hauptmann et al. 2005). We also demonstrated that alterations in bioenergetic functions were normalized by bexarotene treatment. This treatment effect might be independent of changes

seen in autophagy flux and result from alterations in the AMPK-pathway as described previously (Zhu, Ning et al. 2014).

Taken together, our study demonstrates that bexarotene protects mice expressing human mutant tau from cerebral ischemia by altering autophagy. This is the first demonstration that bexarotene is protective in a comorbidity model of stroke and the study demonstrates a novel autophagy-modulating function of bexarotene. Taken into account the fact that autophagy is a key phenomenon in various neurodegenerative diseases, this study gives insight into the use of medicines to manipulate autophagy for a favorable outcome. Further studies should assess whether similar effects on autophagy can be achieved with other medicines targeting RXRs with less impact on plasma triglyceride levels. Moreover, this study demonstrates the importance of aging in modelling ischemic stroke by revealing genotype-dependent effects of bexarotene treatment in old animals. Differences in stroke outcome due to aging arise from neurovascular remodeling and this affects especially the regenerative potential of brain after ischemic injury (Hermann, Buga et al. 2015).

6.2 COPPER DELIVERY IS PROTECTIVE IN STROKE AND ALLEVIATES NEUROINFLAMMATION

Despite the fact that copper bis(thiosemicarbazonato) complexes have been extensively applied in models of neurodegenerative diseases, their effect on the inflammatory

Despite the fact that copper bis(thiosemicarbazonato) complexes have been extensively applied in models of neurodegenerative diseases, their effect on the inflammatory