Systemic treatment with a selective TNFR2 agonist alters the central and peripheral immune responses and transiently improves functional outcome after experimental ischemic stroke

C T Ischemic stroke often leaves survivors with permanent disabilities and therapies aimed at limiting detrimental inflammation and improving functional outcome are still needed. Tumor necrosis factor (TNF) levels increase rapidly after ischemic stroke, and while signaling through TNF receptor 1 (TNFR1) is primarily detrimental, TNFR2 signaling mainly has protective functions. We therefore investigated how systemic stimulation of TNFR2 with the TNFR2 agonist NewSTAR2 affects ischemic stroke in mice. We found that NewSTAR2 treatment induced changes in peripheral immune cell numbers and transiently affected microglial numbers and neuroinflammation. However, this was not sufficient to improve long-term functional outcome after stroke in mice.

* Corresponding author at: Neurobiology Research, Department of Molecular Medicine, University of Southern Denmark, J. B. Winsløwsvej 21 st, 5000 Odense C, Denmark.
The tumor necrosis factor (TNF) signaling system is an interesting treatment target as increases in both TNF and its two receptors, TNF receptor 1 (TNFR1) and TNF receptor 2 (TNFR2), are seen after ischemic stroke in patients and preclinical animal models (Lambertsen et al., 2019;Zaremba et al., 2001).TNF is synthesized as a transmembrane protein (tmTNF), which can be cleaved by the enzyme TNF-α converting enzyme (TACE), to release soluble TNF (solTNF).solTNF, despite highaffinity binding to both TNF receptors, dominantly signals via TNFR1, initiating signaling cascades that can lead to apoptosis, necroptosis, cellular survival and pro-inflammatory signaling (Dong et al., 2015;Probert, 2015).tmTNF stimulates signaling by TNFR1 and TNFR2.Contrary to TNFR1, the latter is not directly linked to cytotoxic programs and has limited proinflammatory activity.In contrast, TNFR2 has pronounced anti-inflammatory effects and promotes cellular survival and regeneration (Dong et al., 2015;Probert, 2015).The currently approved TNF therapies are inhibitors of both tmTNF and solTNF, and although they are used to treat a range of chronic inflammatory diseases, none has been approved for use in stroke (Lambertsen et al., 2019;Leone et al., 2023).Nevertheless, studies have found functional improvements and reduced pain in stroke patients treated off-label with the non-selective TNF-inhibitor etanercept (Joseph et al., 2023;Ralph et al., 2020;Tobinick et al., 2012).However, inhibiting both solTNF and tmTNF can result in severe side effects as the protective and detrimental effects of TNF signaling are inhibited simultaneously (Kristensen et al., 2021;Li et al., 2021;Scheinfeld, 2004).A protective function of TNF after stroke is supported by animal models in which both conventional and conditional TNF ablation leads to bigger infarcts (Clausen et al., 2016;Lambertsen et al., 2009).Therefore, more specialized treatments aimed at the TNF system that promote protective TNFR2 signaling (TNFR2 agonists) or inhibit detrimental solTNF-TNFR1 (solTNF antagonists) or sol/ tmTNF-TNFR1 (TNFR1 antagonists) signaling axes, could therefore prove more beneficial (Siegmund and Wajant, 2023).Previous studies from our lab have shown that maintaining only tmTNF by either genetic deletion or pharmacological inhibition of solTNF is protective in an experimental stroke model (Clausen et al., 2014;Madsen et al., 2016;Yli-Karjanmaa et al., 2019).Additionally, promoting TNFR2 signaling is beneficial in a model of Alzheimer's disease (Ortí-Casañ et al., 2023;Ortí-Casañ et al., 2022).
To our knowledge, it has not been investigated how selective activation of TNFR2 affects the outcome of stroke.In this study, we therefore studied how systemic treatment with the TNFR2 agonist NewSTAR2 affects mice subjected to experimental stroke.NewSTAR2 is a ligandbased TNFR2 agonist with high serum retention (Vargas et al., 2022).We found that systemic TNFR2 activation induces significant changes in peripheral circulating immune cells and transient changes in the brains of pMCAO mice.However, these effects failed to translate into an improvement of experimental stroke in mice neither with respect to functional nor the local inflammatory response at the chronic phase after stroke.

Animals
Experiments were performed using 2-month-old, male C57Bl/6 mice purchased from Janvier (Janvier Labs).The mice were housed in a temperature-and humidity-controlled environment with diurnal lighting and food and water available ad libitum.All animal experiments were performed in accordance with approved permits (J.no.2019-15-0201-01620).

Permanent middle cerebral artery occlusion (pMCAO)
The mice were anesthetized using a mix of Hypnorm (fentanyl citrate (0.135 mg/ml, VetaPharma) and fluanisone (10 mg/ml VetaPharma)), Midazolam (5 mg/ml, Hameln), and sterile water in a 1:1:2 ratio before being subjected to experimental stroke, as previously described (Madsen et al., 2016).Briefly, the left middle cerebral artery (MCA) was exposed by making an incision from the eye to the ear, dissecting the temporal muscle, and drilling a small hole in the skull.Bipolar forceps were used to permanently coagulate the MCA at the level of the inferior cerebral vein.Following surgery, the mice were supplemented with isotonic saline and kept in a 28 • C controlled heating cabinet for 24 h.Subcutaneous injections with Temgesic (0.001 mg/20 mg body weight buprenorphium; Indivior) were given every 8 h for the first 24 h, starting at the time of surgery.To prevent dehydration of the eyes, these were covered with Viscotears ointment (2 mg/g, Bausch and Lomb).

Pharmacological treatment
Sixty (60) μg of NewSTAR2 (Vargas et al., 2022) or a corresponding isotype control antibody (Recombinant Human IgG1 N297A Isotype Control Antibody, Syd Labs Inc., #PA007133) were intraperitoneally injected every second day, starting at the time of surgery.Dosing and treatment schedule are based on previous published papers on the effect of NewSTAR2 in animal models of Alzheimer's disease (Ortí-Casañ et al., 2023;Ortí-Casañ et al., 2022).

Study design
The mice were euthanized 5 or 14 days post-pMCAO.A total of mice underwent surgery and treatment for the 5-day timepoint (isotype, n = 26; NewSTAR2, n = 25), while 24 mice underwent surgery and treatment for the 14-day timepoint (isotype, n = 12; NewSTAR2, n = 12).One isotype-treated mouse died 1 day post-pMCAO.Mice were excluded from all analyses if no infarct could be verified, if they had striatal damage, or if they had suffered MCA bleeding.A total of isotype-treated and 2 NewSTAR2-treated mice were excluded.

Behavioral assessment
The functional outcome of the mice was tested using behavioral tests of grip strength and Y-maze.The grip strength test assessed neuromuscular function and asymmetry of the mice following ipsilateral pMCAO compared to baseline.The Y-maze tested short-term spatial memory and turning preference of the mice.

Grip strength
Grip strength was measured before pMCAO (baseline) and on day post-pMCAO (n = 9-12/group) using a grip strength meter (BIO-GT-3, BIOSEB), as previously described (Clausen et al., 2016;Madsen et al., 2016).The highest peak grip strength (grams) measured from consecutive trials was recorded for the right and left front paw individually, and the asymmetry in strength was calculated as the percentage from baseline.Two isotype-treated mice were excluded from the analysis due to lack of grip during testing.

Y-maze
Y-maze was performed before pMCAO (baseline) and on day 5 (n = 12-14/group) or 14 (n = 11-12/group) post-pMCAO.The test was performed by placing the mouse in the center of a Y-shaped maze, followed by recording of the sequence of arms the mouse entered over min.An entry was recorded when the mouse had all four paws placed inside the arm, and the same arm could not be recorded multiple times in a row.The first two entries were excluded from the analysis as they depended on the orientation of the mouse when placed in the maze.For each mouse, we calculated the number of entries, the percentage of right turns, and the spontaneous alternation behavior (SAB) as described in (Raffaele et al., 2021).An alternation is when the mouse visits all three arms consecutively.One mouse was excluded from the analysis as it had no arm entries during the entire trial.Data were tested for outliers using the ROUT method in GraphPad Prism (Q = 1%), and significant outliers were excluded from the analysis.

Tissue processing
For collection of fresh-frozen tissue, animals were sacrificed by cervical dislocation, and their brains were dissected and snap-frozen using gaseous CO 2 snow, before being kept at − 80 • C. The brains were cryosectioned into 30 μm coronal sections in six parallel series.Two series were collected on gelatin-coated glass slides, and four series were collected in Eppendorf tubes.We collected 17-18 mice/group from 5 days post-pMCAO and 6 mice/group from 14 days post-pMCAO as freshfrozen tissue.For infarct volume analysis, only the mice collected at 5 days post-injury (dpi) were used (n = 17-18/group).For autoradiography, gene expression, and protein analyses, representative mice from both 5 dpi (n = 10/group) and 14 dpi (n = 5/group) were used.
For collection of brain tissue for flow cytometric analysis, mice were euthanized using an intraperitoneal injection of 0.2 ml pentobarbital (200 mg/ml) containing lidocaine (10 mg/ml), followed by transcardial perfusion with 20 ml phosphate-buffered saline (PBS) and dissection of the ipsilateral cortex.Six (6) mice/group were collected for flow cytometric analysis from both timepoints.

Infarct volume and rostro-caudal distribution
One series of coronal brain sections on glass slides from mice with 5 days survival post-pMCAO (n = 17-18/group) was used for volume estimation of the ischemic infarct.The brain sections were stained with toluidine blue, and the total infarct volume was estimated using the stereological CAST 2000 system (Olympus) and applying Cavalieri's principle, as previously described (Gregersen et al., 2000).The distribution of the infarct was estimated from 3600 μm rostral to 3600 μm caudal of the anterior commissure, by calculating the infarct areas from equidistant sections and using the anterior commissure as an anatomical landmark (Madsen et al., 2016).The analyses were performed by an investigator blinded to the treatment groups.Three mice were excluded from the analysis due to technical issues.

Autoradiography
Two glass slides per mouse with coronal brain sections from one series were used for autoradiography analysis of translocator protein (TSPO) binding (8 coronal brain sections in total).TSPO binding was assessed at both 5 dpi (n = 10/group) and 14 dpi (n = 5/group).One slide had sections from the striatal level (Bregma 0.74 mm), and the other had sections from the hippocampal level (Bregma − 1.70 mm).Brain sections were excluded from the analysis when no lesion could be identified, if the lesion was too small, if the high-intensity infarct area was absent or destroyed, or if the section was damaged or blurred.The radioligand used in this study was [ 3 H]PBR28 with a concentration of 1 mCi/ml.The specific activity at the time of the experiment was 78.44 Ci/mmol.The glass slides were rinsed twice in pre-incubation buffer containing 50 mM Tris-HCl (pH 7.4) and 0.5% bovine serum albumin (BSA) diluted in distilled water, for 10 min at room temperature, followed by 60-min incubation in 800 μl 50 mM Tris-HCl buffer (pH 7.4) containing 5 mM MgCl2, 140 mM NaCl, 2 mM EGTA, 0.5% BSA, and 3 nM of [ 3 H]PBR28 per slide at room temperature on a Rotator machine.The slides were rinsed 3 × 5 min in 4 • C pre-incubation buffer and dipped once in 4 • C cold de-ionized water for 10 s, using a washing chamber.The slides were thoroughly dried in the fume hood for 45 min and subsequently exposed to paraformaldehyde (PFA) overnight.The slides were taken out of the PFA chamber and left to dry on the exicator for 30 min.The dry slides with brain sections bound by the radioligand were exposed to a Fuji imaging plate (BAS-IP TR 2040, FujiFilm) along with tritium standard slides (American Radiolabeled Chemicals, #ART-123) for 24 h at 4 • C. Using ImageJ software (NIH), the images were calibrated to the tritium standard, making each pixel's value correlate with the radioligand binding intensity.The images were changed to colors, and brightness and contrast were adjusted to make the highintensity binding areas red and the low-intensity binding areas yellow and green.The rest of the section had darker blue colors.A threshold was defined (12.581-41.479) to visualize the high intensity, low intensity, and no infarct areas.The total infarct and peri-lesion area was measured using the wand tool, and the mean pixel intensity was used to describe the TSPO binding.Total TSPO binding was calculated for each mouse at both the striatal and hippocampal levels.The image analysis was performed by an observer blind to the treatment of the animals.

Gene expression analysis
Gene expression was analyzed at 5 days post-pMCAO (n = 10/group) and 14 days post-pMCAO (n = 5/group).Total RNA was isolated from series of brain sections from each mouse, by first homogenizing the tissue in TRIzol (Thermo Fisher, #15596018) and then centrifuging at 12,000g for 10 min.The supernatant was mixed with chloroform (Sigma Aldrich, #C2432), and the samples were spun again to create a phase separation.The upper phase containing the RNA was mixed with isopropyl alcohol (Sigma Aldrich, #I9030) to precipitate the RNA, which could then be pelleted.The pellet was washed three times in 75% ethanol, after which it was air-dried and resuspended in nuclease-free water and incubated for 10 min at 58 • C. The RNA concentration and purity were determined using a Nanodrop Spectrophotometer (Thermo Fisher), and all samples were diluted to contain 250 ng/μl.RNA was transcribed to cDNA using the High Capacity cDNA Reverse Transcription Kit (Applied Biosystems, #4368814) according to the manufacturer's protocol, followed by dilution to 50 ng/μl.qPCR was performed using Maxima SYBR Green/ROX qPCR Master Mix (2X) (Thermo Fisher Scientific, #K0223) and primers for the genes of interest (Table 1).All genes were normalized to hypoxanthine phosphoribosyltransferase (Hprt1) expression, which we previously demonstrated remains stable over time after pMCAO (Meldgaard et al., 2006) and the gene expression was calculated as fold change of the isotype-treated group using the DeltaDeltaCt method.

Flow cytometry
Glial and immune cell populations were analyzed by flow cytometry at 5 days (n = 6/group) and 14 days (n = 6/group) post-pMCAO.The ipsilateral cortices from the PBS-perfused mice were manually dissociated by passing the tissue through a 70 μm strainer (Falcon, #352350).
Blood samples from NewSTAR2-and isotype-treated mice with 5 days survival (n = 4-5/group) were collected for flow cytometry analysis.Red blood cells were lysed by mixing the blood with distilled water for 30 s, followed by addition of 10X PBS to create a 1X PBS solution.The samples were stained for live cells and blocked as described above, followed by incubation with CD45-PerCP-Cy™5.5 (Clone: 30-F11; BD Biosciences, #561869), CD11b-BB515 (Clone: M1/70; BD Biosciences, #564455), and Ly-6G-BV786 (Clone: 1A8; BD Biosciences, #740953) for 30 min.The samples were fixed and counting beads were added, as described above.Results are presented as number of cells per ml blood volume.
All samples were run on a FACSymphony A1 flow cytometer (BD Biosciences), and data were analyzed using FlowLogic software (Inivai Technologies).Positive populations were gated based on relevant FMO controls.

Primary microglia cultures and analyses
The effect of NewSTAR2 on microglial phagocytosis, morphology, and gene expression was examined using adult primary microglial cultures.Two-month-old C57Bl/6 mice (bred in the animal facility of The Miami Project to Cure Paralysis, Miller School of Medicine, University of Miami) were deeply anesthetized with a 400 mg/kg intraperitoneal dose of Avertin (Sigma Aldrich, #T48202) and transcardially perfused with PBS, and then the brains were dissected.A single cell suspension was obtained from the whole brain by enzymatic dissociation (Neural tissue dissociation kit (P), Miltenyi Biotec, #130-092-628), followed by magnetic-activated cell sorting of microglia using CD11b (Microglia) MicroBeads (Miltenyi Biotec, #130-093-634) and LS columns (Miltenyi Biotec, #130-042-401), according to the manufacturer's protocol.Microglia were plated in Poly-D-Lysine (Sigma-Aldrich, #P7280) coated 96-well plates at 5000 cells/well for the phagocytosis and morphology assay, and at 10,000 cells/well for gene expression analysis, in media consisting of RPMI 1640 Medium (Gibco, A1049101) supplemented with 10% heat-inactivated fetal bovine serum, 20% L929 fibroblast conditioned media, 50 μM beta-mercaptoethanol, and 1% antibioticantimycotic (Gibco, #15240-062).After 5 days in culture, the cells were switched to serum-free media for 24 h, followed by stimulation with 4.5 μg/ml NewSTAR2 or 4.5 μg/ml of the corresponding isotype control (Recombinant Human IgG1 N297A Isotype Control Antibody, Syd Labs Inc., #PA007133).
For the phagocytosis and morphology assay, 1.5 μl of a 1:100 dilution FluoSpheres (Invitrogen, #F8823) was added after 22 h of NewS-TAR2 stimulation.After 24 h of stimulation, the microglia were fixed using 4% PFA and stained with CellMask Deep Red (Invitrogen, #C10046, diluted 1:5000) and 4′,6-diamidino-2-phenylindol (DAPI; Invitrogen, #D1306, diluted 1:2000) for 20 min at room temperature.The microglia were imaged with the Opera Phenix Plus High Content Screening System (Perkin Elmer), and images were analyzed for microglial morphology and phagocytosis with the Harmony software package.Filters were applied to eliminate artifacts using empirically determined cutoffs.Multiple fields of view within each well were sampled in an automated fashion, and the average per well was calculated.The culture was repeated twice, and each datapoint represents one well.From the analysis, we extracted the number of phagocytosed beads, the cellular area and roundness, and the percentage of elongated cells (with roundness below 0.6).Outliers were excluded from the analysis.For the gene expression analysis, RNA was extracted after 3 h of stimulation with NewSTAR2 and was transcribed to cDNA using the SYBR Green Fast Advanced Cells-to-CT Kit (Invitrogen, #A35379), according to the manufacturer's protocol.Technical replicates were pooled together two and two to create 5 samples/group, and the samples were analyzed for genes of interest by qPCR using PowerUp SYBR Green Master Mix (Applied Biosystems, #A25742) and specific primers (Table 1).The expression of each gene was normalized to Hprt1 expression and calculated as fold change of the isotype-treated group using the DeltaDeltaCt method.

Statistical analysis
Data are shown as mean ± SEM.All statistical analyses were performed in GraphPad Prism 10.Normal data distribution was tested using the D'Agostino & Pearson test (n ≥ 8) or the Shapiro-Wilk test (n < 8).Statistical comparisons between the NewSTAR2-and isotype-treated groups used unpaired two-tailed t-test for normally distributed data and Mann-Whitney test for non-normally distributed data.The rostrocaudal distribution of the infarct was analyzed using multiple unpaired t-tests.For multiple comparisons of grip strength data, we used a repeated-measures two-way ANOVA followed by Šídák's post hoc analysis.Results were considered significant at p ≤ 0.05.

Systemic TNFR2 activation improves short-term, but not long-term, functional outcome after experimental stroke
Recent studies revealed improved functional outcome in neurological disease models after TNFR2 agonist treatment (Fischer et al., 2019;Ortí-Casañ et al., 2022).Here, we tested whether systemic treatment with the TNFR2 agonist NewSTAR2 improves the functional outcome after pMCAO by behavioral assessment.A grip strength test to assess asymmetry and neuromuscular function was performed at 1 day post-pMCAO.We found that isotype control treated mice displayed paw asymmetry in their grip strength after pMCAO compared to baseline (Fig. 1A).This asymmetry was not observed in the mice treated with NewSTAR2 (Fig. 1A), suggesting that NewSTAR2 treatment acutely improves the neuromuscular function after pMCAO.No effect of treatment on the weight of the mice was observed.
Y-maze testing was performed on separate mice at 5 and 14 days post-pMCAO to assess turning preference and spatial learning and memory.No differences were seen between isotype control-and NewSTAR2-treated mice at day 5 (Fig. 1B-D) or day 14 (Fig. 1E-G) in the number of entries, SAB, or percentage of right turns in the Y-maze.
Taken together, these data show that systemic activation of TNFR2 can transiently improve functional outcome after experimental stroke, but this improvement does not persist in the later phases.

Systemic TNFR2 activation does not alter infarct volume or poststroke inflammation as assessed by TSPO binding after experimental stroke
TNF is important in the development of the ischemic infarct as mice lacking myeloid-derived TNF developed bigger infarcts (Clausen et al., 2016), and mice with only solTNF removed developed smaller infarcts (Madsen et al., 2016;Yli-Karjanmaa et al., 2019).This suggests a protective role of the tmTNF-TNFR2 signaling axis in infarct development.We therefore investigated whether exogenous TNFR2 activation by systemic treatment with NewSTAR2 affects infarct volume post-pMCAO.The infarct volume was estimated at 5 dpi on coronal, toluidine bluestained sections (Fig. 2A).Unbiased analysis revealed no difference in the total infarct volume between the NewSTAR2-and isotype-treated mice (Fig. 2B), and the rostro-caudal distribution of the infarct was similar in the two groups (Fig. 2C).Thus, activation of TNFR2 using the present administration route and dose did not alter infarct development in pMCAO mice.
One way to visualize and measure glial-mediated inflammation is through imaging of TSPO binding as TSPO expression is strongly increased by activated microglia and astrocytes as well as by infiltrating macrophages (Boutin and Pinborg, 2015;Van Camp et al., 2021).We analyzed TSPO binding in the total infarct and peri-lesion area in coronal brain sections from two different bregma levels: striatal and hippocampal.At both levels, the TSPO binding was comparable between the isotype-and NewSTAR2-treated pMCAO mice at both 5 dpi (Fig. 2D-E) and 14 dpi (Fig. 2F-G).These data indicate that systemic activation of TNFR2 has no effect on TSPO expression after experimental stroke.

Systemic activation of TNFR2 causes a transient upregulation of TNF after pMCAO
Since using a TNFR2 agonist could shift the balance in the TNF signaling system, we wanted to establish how treating mice with NewSTAR2 might affect the gene and protein expression of both TNF, TNFR1 and TNFR2 in the brain after stroke.No significant change in gene expression of Tnf, Tnfrsf1a or Tnfrsf1b was found 5 days post pMCAO (Fig. 3A-C).However, an increase in protein expression of TNF in the NewSTAR2-treated mice compared to isotype-treated mice was seen at day 5 (Fig. 3D), while again no differences were found for TNFR1 and TNFR2 protein (Fig. 3E-F).At day 14, no differences between the two groups were seen in neither gene nor protein expression (Fig. 3G-L).This indicates that while systemic NewSTAR2 treatment can cause a transient upregulation of TNF protein in the brain after pMCAO, it does not affect TNF receptor levels.

Systemic TNFR2 activation alters circulating immune cell numbers and transiently reduces the number of microglia after pMCAO
Stroke activates hematopoiesis in the bone marrow, causing an increased output of immune cells to the blood, peaking within the first week (Courties et al., 2015).We therefore assessed the number of circulating immune cells 5 days post pMCAO by flow cytometric analysis (Supplementary Fig. 1A).NewSTAR2-treated mice had a decreased number of CD45 + Cd11b − cells and increased numbers of monocytes and neutrophils in the blood (Fig. 5A) compared to isotype-treated mice, also reflected in the percentages (Fig. 5B).The mean fluorescence intensity (MFI) of the different cellular markers used in flow cytometry can provide information on the activation state of the cells as they will typically increase their expression with activation (Clausen et al., 2014).Minor changes in CD45 MFI was seen in the CD45 + Cd11b − cells and monocytes in the blood, between isotype-and NewSTAR2 treated mice 5 days post pMCAO (Fig. 5C).
We performed flow cytometric analysis on cell preparations of the ipsilateral cortex as well (Supplementary Fig. 1B) to determine whether systemic NewSTAR2 treatment affected microglia and infiltrating immune cell numbers (Fig. 6) and populations (Supplementary Fig. 2) after pMCAO.Within the microglia population, we analyzed the number of CD68 + and MHCII + microglia (Fig. 6A).CD68 is a scavenger receptor involved in phagocytosis, while MHCII is important for antigen presentation (Lier et al., 2021).We saw a decrease in the numbers of microglia and CD68 + microglia in the ipsilateral cortex of NewSTAR2treated mice compared to isotype-treated mice at 5 days post-pMCAO (Fig. 6A).The decrease in CD68 + microglia numbers was also reflected as a decrease in the percentage of CD68 + microglia out of the total microglial population (Supplementary Fig. 2A).No difference in MHCII + microglia was found (Fig. 6A).Despite the pronounced changes in the number of circulating immune cells in mice 5 days post pMCAO (Fig. 5), the number of infiltrating immune cells in the injured tissue was not changed in any of the analyzed cell populations (Fig. 6B and Supplementary Fig. 2B).At 5 days post-pMCAO, the CD45, CD11b, CD68, and MHCII expression levels on microglia were comparable between treatment groups (Fig. 6C).In the infiltrating immune cells, a difference was seen only in MHCII expression on macrophages, with the NewSTAR2-treated mice having a lower MFI than the isotype-treated mice (Fig. 6D).qPCR on brain tissue for genes typically expressed by microglia and infiltrating macrophages showed no difference between isotype-and NewSTAR2-treated mice 5 days post-pMCAO (Fig. 6E).
At 14 days post-pMCAO, the changes seen in microglial numbers were no longer present (Fig. 6F), and no differences were observed in the number of infiltrating immune cells between treatment groups (Fig. 6G).Looking at the MFI, NewSTAR2 treatment increased CD68 expression levels on infiltrating CD68 + macrophages and Ly-6G expression levels on infiltrating neutrophils, with no other changes observed between treatment groups (Fig. 6H-I).Again, no differences were seen in the expression of microglial/macrophage genes in the brain (Fig. 6J).
Taken together, these results indicate that systemic NewSTAR2 treatment can induce changes in circulating immune cell populations, transient changes in the number of microglia and changes in the activation of infiltrating immune cells.

Activation of microglial TNFR2 in vitro increases phagocytosis and alters gene expression
Microglia become rapidly activated following ischemic stroke and are then important in the TNF signaling response as the main producers of TNF (Gregersen et al., 2000;Lambertsen et al., 2005) and TNFR2 (Lambertsen et al., 2007).As we saw a decrease in microglia with NewSTAR2 treatment, we set out to assess the effect of NewSTAR2 on microglia in vitro.After stimulation with NewSTAR2, naïve microglia increased their phagocytosis of fluorescent beads compared to isotypetreated microglia (Fig. 7A-B).The increase in phagocytosis was followed by a morphological change where the phagocytosing microglia became more rounded and less elongated, but there was no change in cell size (Fig. 7C-E).
TNFR2-induced changes in the expression of various genes of interest in naïve microglia stimulated with NewSTAR2 or an isotype control antibody were analyzed by qPCR.NewSTAR2-treated microglia showed increased expression of Tnf (Fig. 7F) and a moderately decreased expression of Tnfrsf1a while Tnfrsf1b expression was not significantly affected (Fig. 7G, H).No significant change was seen in Trem2 expression with NewSTAR2 treatment (Fig. 7I).The NewSTAR2-treated microglia furthermore downregulated expression of Cd68 and Cx3cr1 compared to isotype-treated microglia (Fig. 7J-K).Similarly to Tnf, the cytokine Il6 was also upregulated in the NewSTAR2-treated microglia (Fig. 7L).Lastly, no change was seen Igf1 (Fig. 7M).

Systemic TNFR2 activation does not change astrocyte and oligodendrocyte populations after pMCAO
Astrocyte and oligodendrocyte cell populations were also assessed with flow cytometry (Supplementary Fig. 1C) with respect to absolute numbers and MFI for ACSA-2 on astrocytes and PDGFRα and O4 on oligodendrocytes (Fig. 8, Supplementary Fig. 2).No change in the number of astrocytes was observed between treatment groups on day post-pMCAO (Fig. 8A).Within the oligodendrocyte lineage, oligodendrocyte progenitor cells (OPCs) can differentiate into pre-myelinating oligodendrocytes, and hereafter into mature, myelinating oligodendrocytes.Through their development, they express different markers, with PDGFRα being expressed mostly by the more immature stages and O4 being expressed mostly at the more mature stages (Long et al., 2021).We observed no differences in cell numbers between treatment groups in any of the oligodendrocyte populations at day 5 (Fig. 8B).Expression levels of ACSA-2 on astrocytes was also not different between NewS-TAR2-and isotype-treated mice at day 5 (Fig. 8C).However, PDGFRα expression levels on immature PDGFRα + oligodendrocytes were higher in NewSTAR2-treated mice compared to isotype-treated mice at day (Fig. 8D), while none of the other oligodendrocyte populations displayed any differences in PDGFRα or O4 expression levels between the two treatment groups (Fig. 8D).We further investigated the gene expression of oligodendrocyte and myelin genes in the brain of NewS-TAR2-and isotype-treated mice.At 5 days post-pMCAO, a decrease in the expression of Bcas1 was seen with NewSTAR2 treatment (Fig. 8E), but no other differences were seen in the investigated genes (Fig. 8E).
At day 14, no differences in the number of astrocytes or oligodendrocytes were identified (Fig. 8F, G), and no changes in any of the MFIs were seen between the two treatment groups (Fig. 8H, I).The gene expression of oligodendrocyte and myelin genes was also similar between isotype-and NewSTAR2-treated mice (Fig. 8J).

Discussion
The neuroinflammatory response following ischemic stroke can have detrimental effects, hereby exacerbating the ischemic damage.Modulation of the inflammatory response to promote the more protective effects is thus a promising treatment strategy to improve the outcome of stroke.As TNF can promote such beneficial effects preferentially through TNFR2, we investigated how systemic treatment with a TNFR2 agonist, NewSTAR2, affected the outcome of experimental stroke.We found that although treatment with NewSTAR2 was able to transiently improve the neuromuscular asymmetry in the mice after pMCAO, it was not able to improve the functional outcome at the later phases.We observed significant changes in circulating immune cell numbers with NewSTAR2 treatment, but within the brain only transient changes in terms of protein expression of TNF, CXCL1, and CCL2 as well as a transient decrease in total microglia and CD68 + microglia numbers was found in NewSTAR2-treated mice compared to the isotype-treated counterparts.These effects did not translate into an effect on infarct development or the overall neuroinflammatory response, measured as TSPO binding.Taken together, no significant long-term improvements with systemic NewSTAR2 treatment were seen in the pMCAO mice.
These findings are in contrast to what we expected as improvements of functional outcome and neuropathology have been observed in animal models of Alzheimer's disease, multiple sclerosis, and spinal cord injury with TNFR2 agonist treatment (Fischer et al., 2019;Gerald et al., 2019;Ortí-Casañ et al., 2023;Ortí-Casañ et al., 2022).One possible explanation is the differences between acute and chronic Fig. 2. NewSTAR2 treatment has no effect on infarct volume and TSPO binding after stroke.(A) Representative images of toluidine blue-stained sections from isotype control-and NewSTAR2-treated mice at 5 days post-pMCAO with the ischemic infarct delineated.Scale bar = 2 mm.(B-C) Infarct volume (B) and rostrocaudal distribution (C) of isotype control-and NewSTAR2-treated mice at 5 dpi, n = 15-17/group.(D) [ 3 H]PBR28 (fmol/mg TE(tissue equivalent)) as a measure of translocator protein (TSPO) binding at the striatal and hippocampal level in isotype control-and NewSTAR2-treated mice on day 5 post-pMCAO, n = 7-9/group.(E) Representative images of TSPO binding intensity at the hippocampal level on day 5 in isotype control-and NewSTAR2-treated mice including a tritium standard slide illustrating binding intensity (red = high intensity, blue = low intensity).(F) [ 3 H]PBR28 (fmol/mg TE) as a measure of TSPO binding at the striatal and hippocampal level in isotype control-and NewSTAR2-treated mice on day 14 post-pMCAO, n = 3-5/group.(G) Representative images of TSPO binding intensity at the hippocampal level on day 14 in isotype control-and NewSTAR2-treated mice including a tritium standard slide illustrating binding intensity (red = high intensity, blue = low intensity).Data are shown as mean ± SEM. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)E. Thougaard et al.  neuroinflammation.Peripheral immune cells do not act in the same way in acute and chronic neurological disease (Gao et al., 2023;Giles et al., 2018), and since the TNFR2 agonist was administered systemically, it will mainly be the peripheral immune cells that are affected.Alzheimer's disease and multiple sclerosis are chronic neurodegenerative diseases in which neuroinflammation is more chronic (Dendrou et al., 2015;Kinney et al., 2018).In contrast, ischemic stroke is followed by a significant acute neuroinflammatory response where the levels of various chemokines, cytokines, and other signaling molecules increase rapidly (Jayaraj et al., 2019).It is likely that the systemic activation of TNFR2 in this study was not sufficient to counteract this strong local neuroinflammatory response, contrary to what is seen in Alzheimer's disease and multiple sclerosis models.Spinal cord injury is also associated with a strong acute neuroinflammatory response (Hellenbrand et al., 2021).Gerald et al. administered the TNFR2 agonist using osmotic pumps that centralized the effect of the agonist at the injury site (Gerald et al., 2019), possibly explaining why they were able to show a beneficial effect.It would be interesting to investigate the effect of systemic treatment with NewSTAR2 in a transient MCAO model, where reperfusion is reestablihsed following occlusion.As the peripheral immune cells would be able to reach the ischemic area more sufficiently, the treatment might have an increased effect.Another important aspect of why systemic treatment with a TNFR2 agonist may not be sufficient in stroke is the expression of TNFR2.While TNFR1 is expressed on almost all cell types, TNFR2 is only expressed on endothelial cells, immune cells, neurons, and glial cells (Probert, 2015), thus limiting the number of cells affected by the treatment.
We saw transient changes in microglia numbers in vivo, where NewSTAR2 treatment caused decreased numbers of microglia and CD68 + microglia 5 days post-pMCAO compared to isotype treatment.CD68 is expressed by activated microglia and is involved in phagocytosis (Lier et al., 2021).We have previously shown that TNFR2 is mainly expressed on microglia within the CNS and peaks at day 5 post-pMCAO (Lambertsen et al., 2007).This explains why the TNFR2 agonist only affected microglial numbers and not the number of astrocytes and oligodendrocytes.As activation of TNFR2 will promote cell survival, the decreased number of microglia and CD68 + microglia could potentially be due to a smaller need for microglial proliferation and phagocytosis.This is supported by the finding that ablating TNFR2 in microglia causes increased microglial proliferation and thus increased numbers, in a mouse model of multiple sclerosis (Gao et al., 2017).We also saw an effect of activating TNFR2 on microglia in vitro.Naïve microglia treated with NewSTAR2 phagocytosed more beads than the microglia treated with the isotype control.Gao et al. found that ablating TNFR2 from microglia decreased their phagocytosis in vitro (Gao et al., 2017), supporting an important role of TNFR2 in regulating microglial phagocytosis.Surprisingly, we saw downregulation of the scavenger receptor CD68 in the microglia treated with NewSTAR2 in vitro, supporting the finding of decreased CD68 + microglia in vivo.Many different receptors are involved in microglial phagocytosis (Fu et al., 2014), however, so it is possible that other, uninvestigated phagocytotic markers were responsible for the increased phagocytosis in vitro.
NewSTAR2 treatment upregulated Tnf gene expression in microglia in vitro, and upregulation of TNF protein was seen in vivo 5 days post-pMCAO in the NewSTAR2-treated mice.No other changes in TNF or TNFR1/2 gene or protein expression were observed.The increase in TNF points to a possible feedback mechanism where activation of TNFR2 using the agonist, in combination with the increased TNF signaling following ischemic stroke, promotes production of TNF (Kuno et al., 2005).This may happen through induction of the Nuclear Factor (NF)-κB and p38 Mitogen-Activated Protein Kinase (MAPK) signaling pathways, which in turn may regulate TNF transcription and translation (Gais et al., 2010;Pe ¸kalski et al., 2013).
It is possible that NewSTAR2 caused a transient improvement early after experimental stroke as we saw improved grip strength 1 day post-pMCAO.Many cytokines and chemokines peak within the first few days, so we cannot rule out the possibility of a beneficial effect of the treatment on the acute neuroinflammatory response, reflected in the improved functional outcome.Even if the NewSTAR2 treatment did cause an acute improvement, however, this was not reflected at the two later timepoints investigated in this study.We saw no improvements in the Y-maze test, infarct volume, or TSPO binding and only few changes to gene and protein expression and cell numbers.
Because NewSTAR2 was given systemically in the present study, we expected most of the effects to occur through modulations of the peripheral immune cells as only around 3% of the compound crosses the blood-brain-barrier in a mouse model of Alzheimer's disease (Ortí-Casañ et al., 2022).We did see large changes in the numbers of neutrophils, monocytes, and CD45 + Cd11b − cells in the blood of NewSTAR2-treated pMCAO mice, showing a clear effect of the treatment on immune cells.However, these changes were not reflected in the number of infiltrating cells in the ipsilateral cortex.The small changes observed in the activation states of some of the infiltrating cells, measured as changes in their MFI, may have contributed to the changes observed in the brains of the NewSTAR2-treated mice.
We have previously shown a beneficial effect of both systemic and central administration of a solTNF inhibitor after pMCAO (Clausen et al., 2014;Yli-Karjanmaa et al., 2019), illustrating how modulation of the TNF signaling system can be effective in ischemic stroke models.Inhibition of solTNF might be more effective in counteracting the strong inflammatory response as it dampens the increase in solTNF following ischemic stroke.It would be interesting to investigate whether central administration of NewSTAR2, e.g., via mini-osmotic pumps, could have a larger effect on the outcome of experimental stroke.Central administration is less therapeutically relevant in stroke, however.Another way to boost the effect of systemic treatment with a TNFR2 agonist could be to combine it with either a TNFR1 antagonist or a solTNF inhibitor.As we saw improved functional outcome after pMCAO with systemic treatment using a solTNF inhibitor (Clausen et al., 2014), it is easy to hypothesize that combining this treatment with a TNFR2 agonist might have an additive effect.Indeed, a protective effect of a TNFR2 agonist in combination with a TNFR1 antagonist has been shown in a mouse model of multiple sclerosis (Fiedler et al., 2023;Pegoretti et al., 2023) and should be investigated further in the context of acute ischemic stroke.
In conclusion, we showed that systemic activation of TNFR2 in vivo caused a significant change in the circulating immune cells and could induce transient changes in the brains of pMCAO mice.However, the treatment was not sufficient to significantly improve the outcome after experimental stroke in mice.Fig. 6.NewSTAR2 treatment transiently decreases the number of microglia after pMCAO.(A) Flow cytometric quantification of absolute numbers of microglia, CD68 + microglia, and MHCII + microglia in the ipsilateral cortex of isotype-and NewSTAR2-treated mice 5 days post-pMCAO.Microglia are defined as CD11b + CD45 dim cells.(B) Flow cytometric quantification of absolute numbers of infiltrating leukocytes in the ipsilateral cortex of isotype-and NewSTAR2-treated mice 5 days post-pMCAO.Leukocytes are defined as CD11b + CD45 high cells, infiltrating macrophages as Ly6G − and subpopulations as MHCII + and CD68 + , neutrophils as Ly6G + , and infiltrating lymphocytes as CD11b − CD45 + .(C, D) Flow cytometric analysis of mean fluorescence intensity (MFI) for CD45, CD11b, CD68, and MHCII expression levels on microglia (C) as well as Ly6G on leukocytes (D) in the ipsilateral cortex of isotype-and NewSTAR2-treated mice 5 days post-pMCAO.n = 6/group.(E) Quantification of whole brain qPCR for microglial/macrophage genes Itgam1, Trem2, Cd68, Arg1, Igf1, and Nos2 at 5 days post-pMCAO, n = 10/group.(F) Flow cytometric quantification of absolute numbers of microglia, CD68 + microglia, and MHCII + microglia in the ipsilateral cortex of isotype-and NewSTAR2treated mice 14 days post-pMCAO.(G) Flow cytometric quantification of absolute numbers of infiltrating leukocytes in the ipsilateral cortex of isotype-and NewSTAR2-treated mice 14 days post-pMCAO.(H, I) MFI for CD45, CD11b, CD68, and MHCII expression levels on microglia (H) as well as Ly6G on leukocytes (I) in the ipsilateral cortex of isotype-and NewSTAR2-treated mice 14 days post-pMCAO, n = 6/group.(J) Quantification of whole brain qPCR for microglial/macrophage genes Itgam1, Trem2, Cd68, Arg1, Igf1, and Nos2 at 14 days post-pMCAO, n = 5/group.Gene expression was normalized to Hprt1 and calculated as fold change of isotype.Data are shown as mean ± SEM. *p ≤ 0.05, **p ≤ 0.01 unpaired two-tailed t-test.Flow cytometric quantification of absolute numbers of oligodendrocyte progenitor cells, immature and mature oligodendrocytes in the ipsilateral cortex of isotypeand NewSTAR2-treated mice 5 days post-pMCAO.Oligodendrocyte progenitor cells are defined as ACSA-2 − CD45 − PDGFRα + , immature oligodendrocytes as ACSA-2 − CD45 − PDGFRα + O4 + , and mature oligodendrocytes as ACSA-2 − CD45 − O4 + , n = 6/group.(C) MFI for ACSA-2 expression levels on astrocytes in the ipsilateral cortex of isotype-and NewSTAR2-treated mice 5 days post-pMCAO, n = 6/group.(D) MFI for PDGFRα and O4 expression levels on oligodendrocyte progenitor cells, immature and mature oligodendrocytes in the ipsilateral cortex of isotype-and NewSTAR2-treated mice 5 days post-pMCAO, n = 6/group.(E) Quantification of whole brain qPCR for oligodendrocyte and myelin genes; Mog, Mbp, Cspg4, Gpr17, and Bcas1 at 5 days post-pMCAO, n = 10/group.(F) Flow cytometric quantification of absolute numbers of astrocytes in the ipsilateral cortex of isotype-and NewSTAR2-treated mice 14 days post-pMCAO, n = 6/group.(G) Flow cytometric quantification of absolute numbers of oligodendrocyte progenitor cells, immature and mature oligodendrocytes in the ipsilateral cortex of isotype-and NewSTAR2-treated mice 14 days post-pMCAO, n = 6/group.(H) MFI for ACSA-2 expression levels on astrocytes in the ipsilateral cortex of isotype-and NewSTAR2-treated mice 14 days post-pMCAO, n = 6/group.(I) MFI for PDGFRα and O4 expression levels on oligodendrocyte progenitor cells, immature and mature oligodendrocytes in the ipsilateral cortex of isotype-and NewSTAR2-treated mice 14 days post-pMCAO, n = 6/group.(J) Quantification of whole brain qPCR for oligodendrocyte and myelin genes; Mog, Mbp, Cspg4, Gpr17, and Bcas1 at 14 days post-pMCAO, n = 5/group.Gene expression is normalized to Hprt1 expression and calculated as fold change of isotype.*p ≤ 0.05 unpaired two-tailed t-test.Data are shown as mean ± SEM.

E
. Thougaard et al. (caption on next page) E. Thougaard et al.

Fig. 7 .
Fig. 7. NewSTAR2-stimulated microglia increase phagocytosis and change their expression of selected genes in vitro.(A) Representative images of naïve primary microglia (red) treated with either NewSTAR2 or the corresponding isotype control, with added fluorescent beads (green) to assess phagocytosis.Scale bar = 100 μm.

Table 1
Primer sequences used for qPCR.