Brain microvascular endothelial cells (bEnd.3, ATCC CRL-2299) were purchased from ATCC and cultured in Dulbecco’s Modified Eagle’s Medium (DMEM, ATCC) supplemented with 10% FCS, L-glutamine and antibiotics. Cell clones with increased Zfp580 expression or expression of an shRNA embedded miRNA targeting Zfp580 or LacZ were generated using lentiviral particles. For a detailed description of lentiviral particle production, see Dätwyler et al. (Datwyler et al., 2011). Transfer vectors additionally encoded enhanced green fluorescent protein (EGFP) and puromycin resistance proteins separated by T2A self-cleavage peptides. After transduction, the cells were cultivated in DMEM with puromycin until 100% of the cells expressed EGFP. For experiments, the cells were cultured in supplemented DMEM without puromycin.

Lentiviral transfer vectors were designed based on a modified flap-Ubiquitin promoter-driven EGFP-WRE lentiviral transfer plasmid (pFUGW) [the FUGW plasmid was a gift from David Baltimore (Addgene plasmid #14883; http://n2t.net/addgene:14883; RRID:Addgene_14883)] (Lois et al., 2002). Zfp580, derived as a product of gene synthesis (Eurofins, Ebersberg, Germany), was followed by a T2A self-cleaving peptide sequence that was inserted by unidirectional cloning such that it preceded the sequences for EGFP and puromycin resistance in pFUGW. Small hairpin RNA (shRNA)-embedded microRNA for the expression of LacZ or Zfp580 targeting shRNA were designed using BLOCK-iT RNAi Designer (Invitrogen, USA) with the accession number NM_026900, which encodes murine Zfp580, and blasted against Mus musculus to control off-target effects. Annealing and ligation of shRNA targeting Zfp580 at nucleotide position 930 in the 5′ untranslated region (UTR) (Eurofins, Ebersberg, Germany) were performed using sticky ends in a BbsI-predigested vector (pcDNA6.2-GW/EmGFP with spectinomycin as a selection marker) following the manufacturer’s instructions (BLOCK-iT Pol II miR RNAi expression vector kit, Invitrogen). The ligated shRNAs were subcloned via BlpI and XhoI (204 bp-long fragments) into a modified version of the pFUGW lentiviral transfer vector for equal coexpression of the shRNA and EGFP (reporter protein) alongside puromycin. A LacZ nontargeting shRNA served as a control.

All transfer vectors and detailed steps for cloning and sequence verification are available upon request from the corresponding authors or via Addgene.

RNA was extracted using TRIzol (Invitrogen) and stored at -80 °C prior to transcription. One microgram of RNA was transcribed with MLV reverse transcriptase (Promega) using random primers (Roche) and oligo-dT primers (Eurofins-MWG). Real-time PCR was performed with exon-spanning primers for tyrosine 3-monooxygenase/tryptophan 5-monooxygenase activation protein (Ywhaz) and intron-spanning primers for succinate dehydrogenase complex, subunit A, flavoprotein (Sdha) as internal controls (Gubern et al., 2009), and intron-spanning primers for Zfp580, Igf1 or Igfp3 using SYBR Green (Qiagen). For primer sequences, see Table 1. Comparable primer efficiency was confirmed. Melting curves were analyzed for all runs. Experiments in the absence of a template and those with untranscribed RNA served as negative controls. The ddCT method was used to figure out how many fold differences there were in mRNA levels compared to internal control levels.

ELISAs to determine Igf1 and Igfbp3 (R&D Systems) levels in mouse blood serum were performed according to the manufacturer’s instructions. The Igf1: Igfbp3 molar ratio was calculated as previously described (Soubry et al., 2012).

Endothelial cells were subjected to combined OGD as described previously (Harms et al., 2007). In brief, cultures were placed in an “IN VIVO ₂ 300” OGD chamber (Ruskinn, Pencoed, United Kingdom) with 5% CO₂/0.3% O₂ in buffer free of glucose for 4 h. Control cells were cultured for 4 h under normoxic conditions in the same buffer but supplemented with 5.5 mM glucose.

All experimental procedures were approved by the local authorities (Landesamt für Gesundheit und Soziales, LaGeSo, Berlin (Reg G0254/16)) and were conducted following the German animal protection law and local animal welfare guidelines. Zfp580tm1a^(EUCOMM)Wtsi mice (constitutive Zfp580-knockout mice, C57/BL6/N background) were purchased from EuMMCR (Helmholtz Zentrum München). Transient filamentous blockage of the middle cerebral artery was performed on mice for 30, 45 or 60 min using a documented standard operating procedure from our laboratory (Endres et al., 2003). In brief, mice were anaesthetized with 1.5–3.5% Isoflurane and maintained in 1.0–2.5% Isoflurane with approximately 75/25 N₂O/O₂ and for pain relief, bupivacain gel (1%) was topically applied to the wound. The common carotid artery (CCA) external carotid artery (ECA) were ligated. A 180 µm diameter Doccol^® filament was introduced into the CCA and advanced up the internal carotid artery (ICA) to occlude the middle cerebral artery (MCA) for the indicated durations. For sham procedure, the filament was inserted to occlude the MCA and withdrawn immediately. Successful induction of stroke was proven by MRI-scans 24 h after surgery. For time course experiments, we used 3 male C57BL6/N wild-type mice for sham surgery, 4 mice for 30 min MCAo and 5 mice for 60 min MCAO for each reperfusion time as indicated in the respective figure (Figure 4). According to preset exclusion criteria concerning humane endpoints, 1 animal was excluded in the group of 7 days of reperfusion after 30 min MCAo and 1 animal in the group of 48 h as well as 2 animals in the group of 72 h of reperfusion after 60 min MCAO. For 45 min MCAo experiments with Zfp580tm1a and wild-type littermates, we used 6 age- and gender-matched animals in each group. None of these animals had to be excluded.

Animals were randomized for MCAo operation, and experimenters were blinded to the genotype. Reporting conforms to the ARRIVE guidelines.

All data are presented as scatter dot plots with the mean ± standard deviation. The data were analyzed with GraphPad Prism version 8.2.0 and tested for normal distribution using the Shapiro-Wilk test. A detailed description of the corresponding statistical analysis is provided in the figure legends.

We used the cerebral microvascular endothelial cell line bEnd.3. By lentiviral transduction, we generated cell clones with increased expression of Zfp580 or cell clones expressing a Zfp580 specific shRNA embedded in a miRNA cassette. A cell clone expressing miRNA embedded shRNA against LacZ served as negative control for knock down experiments. We determined the mRNA expression levels using real time RT-PCR. We achieved a strong increase of Zfp580 expression (Figure 1A) and knocked down Zfp580 efficiently by approximately 85% using RNAi (Figure 1B). Increased expression of Zfp580 led to a strong suppression of Igf1 and Igfbp3 mRNA (Figures 1C,E), whereas knock down of Zfp580 induced Igf1 and Igfbp3 mRNA (Figures 1D,F). Therefore, Zfp580 is a suppressor of Igf1 and Igfbp3 expression in endothelial cells.

To simulate a stroke in vitro, we exposed the cells to a 4-h period of combined oxygen and glucose deprivation (OGD). Cells cultured in OGD-buffer with glucose under normal conditions served as negative controls. Immediately after OGD, mRNA expression levels were determined by real-time RT-PCR. We identified a strong suppression of Zfp580 mRNA after OGD (Figure 2A) and, consistently, Igfbp3 mRNA was more than doubled (Figure 2C, left panel with native cells). However, Igf1 was suppressed after OGD (Figure 2B, left panel with native cells). Next, we evaluated whether Zfp580 mediates the effects of OGD on Igf1 and Igfbp3 expression. Since Zfp580 mRNA is strongly suppressed after OGD, we used bEnd.3 cells with exogenously increased expression of Zfp580 in a rescue experiment to prevent the loss of Zfp580 after OGD. In cells cultured under control conditions, increased expression of Zfp580 suppressed Igf1 and Igfbp3 (Figures 2B,C). Increased Zfp580 expression blunted the effect of OGD on Igf1 and Igfbp3 expression. Therefore, preventing Zfp580 loss after OGD interfered with Igf1 and Igfbp3 regulations after OGD.

We determined the effects of Zfp580 genomic ablation on Igf1 and Igfbp3 in vivo using Zfp580 knockout mice (Zfp580tm1a^(EUCOMM)Wtsi, a constitutive knockout mouse model from the EUCOMM project). Wild-type littermates served as controls. To assess paracrine effects, we extracted RNA from brain lysates and determined Igf1 and Igfbp3 mRNA levels by real-time RT-PCR. We identified no obvious differences between Zfp580tm1a and wild-type littermates in paracrine cerebral Igf1 and Igfbp3 expression (Figures 4A,B). Additionally, we determined blood levels of endocrine circulating Igf1 and Igfbp3 using ELISA. There was no difference in Igf1 and Igfbp3 blood levels or in the Igf1/Igfbp3 molar ratio between the genotypes (Figures 3C–E). Thus, under normal settings, genomic ablation of Zfp580 has no effect on paracrine or endocrine Igf1 or Igfbp3.

Next, we evaluated the effects of stroke on Zfp580, Igf1 and Igfbp3 expression after 60 min of transient filamentous middle cerebral artery occlusion (MCAo) for short-term regulation up to 3 days in wild-type C56/BL6 mice. Additionally, we used a 30-min MCAo to evaluate long-term effects up to 28 days. Negative controls were sham operated animals that had the exact surgery process but were immediately removed of the filament. Zfp580 expression was induced acutely after stroke, starting 3 h after ischemia (Figure 4A). Expression was induced globally in both hemispheres and normalized 3 days after ischemia. After 30 min of MCAO, there was no significant difference in Zfp580 expression in the long-term (Figure 3B). Igf1 expression was selectively induced in the ipsilesional hemisphere starting 2 days after ischemia (Figures 4C,D). Expression normalized between day three and day 7 after ischemia. Similarly, Igfbp3 expression was selectively induced in the ischemic hemisphere, while induction occurred earlier starting 24 h after ischemia and persisted up to 7 days (Figures 4E,F). When we investigated the effects of MCAo duration on expression levels on day 2 following ischemia, we discovered that a 60-min MCAo induced Zfp580 to a greater extent than a 30-min MCAo (Figure 4G). Igf1 was induced only after a 30 min MCAO (Figure 4H) at this early point of time. Induction of Igfbp3 was less pronounced after a 60 min MCAO than after a 30 min MCAo (Figure 4I). Thus, early induction of Zfp580 might inhibit Igf1 and Igfbp3 expression in the acute phase following stroke, but lowering Zfp580 after 3 days may allow Igf1 and Igfbp3 expression through disinhibition. Moreover, higher levels of Zfp580 might cause a reduced induction of Igf1 and Igfbp3 following a prolonged duration of MCAo.

We determined the expression levels 2 days after 45 min MCAo in Zfp580 knockout mice to determine if Zfp580 has an effect on the paracrine cerebral Igf1 and Igfbp3 regulation following stroke. Wild-type littermates and sham-operated mice served as negative controls. In wild-type littermates, we again identified an induction of Igf1 and Igfbp3 in the ischemic hemisphere (Figures 5A,B). There was no observable induction of Igf1 or Igfbp3 in Zfp580 knockout mice (Figures 5A,B). We measured Igf1 and Igfp3 levels in the blood 2 days after ischemia using an ELISA to examine the effects of Zfp580 on post-ischemic endocrine regulation. After stroke, Igf1 blood levels were significantly increased in Zfp580 knockout mice (Figure 5C). However there was no difference in Igfbp3 blood levels between the two genotypes (Figure 5D). The Igf1/Igfbp3 molar ratio was greater in Zfp580 knockout mice than in wild-type littermates (Figure 5E). Thus, Zfp580 deletion resulted in reduced paracrine cerebral Igf1 and Igfbp3 expression, increased endocrine Igf1 blood levels, and consecutively increased Igf1/Igfbp3 molar ratios 2 days after stroke.