Poldip2 controls leukocyte infiltration into the ischemic brain by regulating focal adhesion kinase-mediated VCAM-1 induction

Stroke is a multiphasic process involving a direct ischemic brain injury which is then exacerbated by the influx of immune cells into the brain tissue. Activation of brain endothelial cells leads to the expression of adhesion molecules such vascular cell adhesion molecule 1 (VCAM-1) on endothelial cells, further increasing leukocyte recruitment. Polymerase δ-interacting protein 2 (Poldip2) promotes brain vascular inflammation and leukocyte recruitment via unknown mechanisms. This study aimed to define the role of Poldip2 in mediating vascular inflammation and leukocyte recruitment following cerebral ischemia. Cerebral ischemia was induced in Poldip2+/+ and Poldip2+/− mice and brains were isolated and processed for flow cytometry or RT-PCR. Cultured rat brain microvascular endothelial cells were used to investigate the effect of Poldip2 depletion on focal adhesion kinase (FAK)-mediated VCAM-1 induction. Poldip2 depletion in vivo attenuated the infiltration of myeloid cells, inflammatory monocytes/macrophages and decreased the induction of adhesion molecules. Focusing on VCAM-1, we demonstrated mechanistically that FAK activation was a critical intermediary in Poldip2-mediated VCAM-1 induction. In conclusion, Poldip2 is an important mediator of endothelial dysfunction and leukocyte recruitment. Thus, Poldip2 could be a therapeutic target to improve morbidity following ischemic stroke.

Polymerase δ-interacting protein 2 (Poldip2, PDIP38, Mitogenin-1) is a ubiquitously expressed multifunctional protein first identified as a binding partner of polymerase δ and proliferating cell nuclear antigen 1 . While originally described to be involved in DNA replication and damage repair, subsequent studies have implicated Poldip2 as a critical regulator of focal adhesion turnover, mitochondrial function and morphology, cell migration, proliferation and differentiation, sensory nerve activation, autophagy and, more recently, metabolic adaptation 1,2 . Homozygous gene trap mice suffer from perinatal lethality 3 and Poldip2 heterozygous mice have a clear vascular phenotype characterized by arterial stiffness, increased collagen deposition, disordered elastic laminae and impaired collateral formation 4,5 . Studies from our group have shown that Poldip2 heterozygous mice are strongly protected against blood-brain barrier (BBB) permeability and the expression of inflammatory cytokines that accompanies cerebral ischemia 6 and sepsis-associated encephalopathy 7 . These studies, in addition to our report on the protective effect of Poldip2 depletion in a lipopolysaccharide (LPS) model of acute respiratory distress syndrome 8 , suggest that Poldip2 may play a critical role in regulating endothelial barrier function. However, whether or not a reduction in Poldip2 also influences leukocyte recruitment and transmigration across the BBB is unknown.
Adhesion molecules, such as ICAM-1, VCAM-1 and selectins, mediate the recruitment and firm adhesion of circulating leukocytes from the bloodstream to the postischemic cerebral microvasculature, facilitating their transmigration across the vascular endothelial barrier 9 . Leukocyte-endothelial cell adhesion, which is largely absent in the normal healthy cerebral microcirculation, becomes substantial and persists for hours and days following the onset of brain ischemia, compromising cerebral microvascular perfusion and contributing to the severity of stroke outcomes 9,10 .
Cytokines, reactive oxygen species (ROS), and prostaglandin E2 (PGE2) are some of a large number of inflammatory mediators released into ischemic brain tissue and plasma, activating transcription-dependent and -independent pathways, resulting in upregulation of cell adhesion molecules 9,11 . Patterns of adhesion molecule expression are very dynamic and are closely associated with the infiltration of different subsets of leukocytes. Evaluation of plasma levels of adhesion molecules in patients with acute ischemic stroke and histopathological examination of brain tissue in experimental stroke models revealed that expression of E-selectin, which is associated with neutrophil extravasation, peaks early (6-24 h) after acute cerebral ischemia and rapidly decreases thereafter, corresponding to the onset of neutrophil recruitment into the brain tissue 12,13 . VCAM-1, in turn, which predominantly mediates adhesion of monocytes, progressively increases and peaks 5-10 days after cerebral ischemia, corresponding to the onset of monocyte recruitment [12][13][14][15] . Because VCAM-1 mediates sustained and prolonged leukocyte adhesion and transmigration, there is an increasing appreciation for the role of VCAM-1 in coordinating the inflammatory response and leukocyte recruitment in cerebral ischemia.
In endothelial cells, transcriptional regulation of VCAM-1 involves AP-1, NF-κB and GATA-4 activation [16][17][18] . In addition, nuclear focal adhesion kinase (FAK) activity has been shown to be involved in promoting tumor necrosis factor (TNF)-induced VCAM-1 expression in human umbilical vein endothelial cells 19 . However, specific factors regulating FAK-induced VCAM-1 expression remain unclear and comparatively much less is known about the mechanisms that regulate inflammation-induced FAK activation. Of interest, Poldip2 has been shown to regulate FAK activity 20 .
Based on these considerations and our previous work implicating Poldip2 in endothelial barrier function, we hypothesized that heterozygous deletion of Poldip2 will diminish infiltration of leukocytes into the brain. We examined the effect on several different subtypes of leukocytes and investigated the mechanistic basis of the role of Poldip2 by testing the hypothesis that Poldip2 mediates TNF-induced VCAM-1 expression via regulation of FAK activity. Herein we report that Poldip2 +/− mice exhibit reduced infiltration of leukocytes into the brain, as well as lower VCAM-1 expression, following cerebral ischemia. Our data further suggest that Poldip2 depletion in cultured rat brain microvascular endothelial cells (RBMECs) prevents VCAM-1 induction, leukocyte adhesion and TNF-induced FAK activation. Overexpressing FAK partially rescues VCAM-1 expression induced by TNF in spite of Poldip2 depletion. Thus, we substantiated Poldip2 as an important mediator of leukocyte infiltration into the brain and identified FAK activation as a critical intermediary in Poldip2-mediated VCAM-1 induction.

Methods
Animals. Poldip2 gene trap mice on the C57BL/6 background were produced by the Texas A&M Institute for Genomic Medicine (College Station, TX). Mice were genotyped using a 4-primer qPCR method. Homozygous gene trap mice suffer from perinatal lethality; therefore, Poldip2 +/− mice were used for this study. Characterization of these mice has previously been described 3 . Studies involving animals were carried out in compliance with the ARRIVE guidelines 21,22 . All animal experiments were conducted with the approval of the Institutional Animal Care and Use Committee at Emory University (protocol reference number: 201700864). Ethical approval was received before conducting the study.
Animal model of cerebral ischemia. Hypoxia-induced permanent stroke was induced in 10-to 14-weekold male mice, as previously described 23,24 . This stroke model leads to reperfusion deficits, fibrin and platelet deposition, and significant infarct in the middle cerebral artery-supplied territory 24 . Mice were anesthetized with isoflurane (medical air (21% oxygen) delivered at 0.5 l/min with 3% isoflurane for induction and 1.5% for maintenance). The right common carotid artery (RCCA) was exposed and transiently occluded, using 5-0 silk sutures to interrupt blood flow to the ipsilateral side of the brain. The wound was temporarily closed and animals were turned to a prone position. To induce transient hypoxia, medical air in the inhalation mixture was replaced with 7.5% oxygen balanced by nitrogen (92.5% N 2 ) for 30 min. A rectal probe was connected to a temperature controller system with heating lamps and core body temperature was maintained at 37.5 ± 0.5 °C during the entire procedure. At the end of the hypoxic stimulation, mice were returned to their original supine position, the inhalation mixture was switched back to medical air and the RCCA sutures were released. Sham-operated control mice simply underwent exposure of the RCCA and were kept under anesthesia for the same duration as the others.
Lesion volume. Twenty-four hours after cerebral ischemia induction, mice were deeply anesthetized and sacrificed by cervical dislocation. Brains were removed, sliced into 2-mm coronal sections, and the volume of the ischemic lesion was measured by 2,3,5-triphenyltetrazolium chloride (TTC) staining as described previously 6,25 .
The lesion volume was calculated using ImageJ.
Brain dissociation for immune cell isolation. Forty-eight hours after cerebral ischemia induction mice were anesthetized with isoflurane and sacrificed by cervical dislocation. Whole brains were immediately removed, finely minced in 1 × Hank's balanced salt solution (HBSS, without calcium, magnesium, or phenol red; Invitrogen, Carlsbad, CA), transferred to an enzymatic solution in DMEM/F12 media containing 1 mg/ml papain (Sigma Aldrich, St. Louis, MO), 1.2 U/ml dispase II (Roche diagnostics, Risch-Rotkreuz, Switzerland), and 220 U/ml DNase I (Invitrogen) and incubated at 37 °C for 20 min with agitation every 5 min. Warm fetal bovine serum (10%; heat inactivated, Atlanta Biologicals) was added to neutralize the enzymatic reaction, and samples were spun at 300×g for 5 min. The supernatant was removed and the tissue pellet was homogenized in ice cold 1 × HBSS by pipetting up and down 30 times using a fine-tip (Thermo Fisher Scientific, Waltham, MA), fire-polished glass pipette. The homogenate was passed through a sterile 70 μM cell strainer and resuspended in 37% Percoll (Percoll pH 8.5-9.5; Sigma Aldrich). Four ml of 70% Percoll were layered below the tissue suspension, and 4 ml of 30% Percoll were overlain, followed by 4 ml of 1 × HBSS and spun at 500×g for 30 min at room temperature with no brake. The immune cell cloud was identified and collected between the 70% and 37% layers and washed with 4 volumes of 1 × HBSS. Cells were plated on a 96 well plate for immunocytochemistry. Single cell lymphocytes were gated on forward scatter area (FSA; size) and side scatter area (SSA; granularity) and then by forward scatter height (FSH) and FSA. Live cells were selected as the negative Fixable Aqua population. CD45 + cells were selected and gated by CD3 (T cells) versus CD11b (myeloid cells). The CD3 + population was then gated on CD4 (helper T cells) and CD8 (cytotoxic T cells). CD11b + cells were further gated on CD11c and Ly6G for the following populations: CD11c − Ly6G − (microglia and macrophages), CD11c-Ly6G + (neutrophils), and CD11c + Ly6G − (dendritic cells). CD11c − Ly6G − cells were gated for CD45 and low and high populations were further gated for MHCII and Ly6C to determine their activation state (Suppl. Fig. S1).  Adenoviruses. Adenoviruses expressing FAK (AdFAK), LacZ control (AdLacZ, control for AdFAK) and

RNA extraction and RT-qPCR. Total
coexpressing human Poldip2-myc (AdPoldip2) and green fluorescent protein (GFP) or GFP alone (AdGFP) were prepared in HEK293T cells as described previously and precipitated with polyethylene glycol 6000 (PEG-6000) 29 . AdLacZ was a kind gift of Dr. Aviv Hassid (University of Tennessee) and the AdFAK was a kind gift from Dr. Joseph C. Loftus (Mayo Clinic, Phoenix, AZ). RBMECs were transduced with recombinant adenoviruses for 3 h at 37 °C in complete endothelial cell medium. Following 3 h, fresh media was added to RBMECs and cells were allowed to recover for 24 h. Cells were then washed with HBSS and fresh media was added for an additional 24 h before use in experiments. Overexpression of FAK and Poldip2 was confirmed by immunoblotting.
Western blotting. Whole cell lysate was prepared from cultured RBMECs using a buffer containing 50 mM HEPES, 50 mM NaCl, 5 mM EDTA, 10 µg/ml aprotinin, 10 µg/ml leupeptin, 1 mM PMSF, and Halt phosphatase inhibitor cocktail (Thermo Fisher Scientific). Homogenates were centrifuged at 17,700×g at 4 °C for 2 min and then sonicated for 10 s. Homogenates were aliquoted, and samples were brought to equal volume in Laemmli buffer and water before boiling for 10 min at 100 °C. Samples were stored at − 20 °C until gel loading.

Results
Poldip2 regulates the recruitment of immune cells into brain tissue after cerebral ischemia. Quantitative flow cytometric analysis of CD45 expression, a leukocyte common antigen, performed 48 h after cerebral ischemia induction, demonstrated that the number of CD45 + cells was dramatically increased in the brains of Poldip2 +/+ , but not in Poldip2 +/− mice (Fig. 1). Similar results were found for the CD45 high immune cell population. Additionally, the infiltration of CD45 + CD11b + myeloid cells, Ly6C high CD45 high and MHCII low CD45 high inflammatory monocytes/macrophages was significantly reduced in Poldip2 +/− mice after cerebral ischemia when compared to Poldip2 +/+ mice. CD45 low inflammatory cells were not significantly increased in the brains of either Poldip2 +/+ or Poldip2 +/− mice (Fig. 1). Further characterization of invading cell populations demonstrated that CD3 + , CD8 + and CD4 + T cells and CD11b + Ly6G − dendritic cell counts were not significantly increased in the brains of either Poldip2 +/+ or Poldip2 +/− mice 48 h after cerebral ischemia (Suppl. Fig. S2

Poldip2 does not affect lesion volume after cerebral ischemia induction. We then tested whether
Poldip2 depletion affects the lesion volume following cerebral ischemia induction. Brains obtained from Poldip2 +/+ and Poldip2 +/− mice were analyzed using TTC staining. Twenty-four hours after cerebral ischemia induction, no significant differences were observed between Poldip2 +/+ and Poldip2 +/− mice (Suppl. Fig. S3).

Heterozygous deletion of Poldip2 reduces leukocyte adhesion molecule expression in vivo. Following cerebral ischemia, increased expression of adhesion molecules in cerebral endothelial cells
promotes the recruitment of circulating leukocytes into brain tissue. To determine whether Poldip2 regulates the expression of leukocyte adhesion molecules, brains from Poldip2 +/+ and Poldip2 +/− mice were harvested 24 h after cerebral ischemia induction and E-selectin, P-selectin, ICAM-1 and VCAM-1 mRNA levels were measured. P-selectin mRNA levels remained unchanged in the brain tissue upon cerebral ischemia induction in both Poldip2 +/+ and Poldip2 +/− mice. While E-selectin expression was not detectable in sham samples, E-selectin mRNA levels were increased 24 h after ischemia, but significantly less so in Poldip2 +/− , compared to Poldip2 +/+ mice. Similarly, ICAM-1 and VCAM-1 mRNA levels were increased in the brain tissue of Poldip2 +/+ mice after ischemia, but not in Poldip2 +/− mice (Fig. 2).

Poldip2 silencing attenuates TNF-induced VCAM-1 upregulation and leukocyte adhesion to brain endothelial cells in vitro.
Longitudinal studies evaluating the levels of soluble adhesion molecules in human peripheral blood after cerebral ischemia revealed that while ICAM-1 and E-selectin levels peaked within the first 6-24 h and rapidly decreased thereafter, VCAM-1 levels progressively increased and peaked after 5-10 days, suggesting that VCAM-1 plays an important role mediating prolonged and sustained leukocyte adhesion and migration in cerebral ischemia [13][14][15] . We have previously determined that Poldip2 is required for LPS-induced VCAM-1 expression in human primary pulmonary microvascular endothelial cells 8 . To examine the role of Poldip2 in the regulation of VCAM-1 induction in the brain, cultured RBMECs were transfected with siPoldip2 or control siRNA (siControl) and treated with TNF (10 ng/ml) for 6 h before evaluating VCAM-1 expression. TNF was used to stimulate RBMECs because it is one of the main proinflammatory cytokines implicated in the vascular permeability induced by stroke 30,31 and because it is a key factor regulating inflammatory VCAM-1 expression 19 . Consistent with our in vivo observation, Poldip2 silencing significantly inhibited TNF-induced VCAM-1 upregulation. Poldip2 silencing efficiency was evaluated by western blotting (Fig. 3A).
The physiological consequence of VCAM-1 reduction in RBMECs was assessed by evaluating THP-1 monocyte adherence. RBMECs were transfected with siControl or siPoldip2, stimulated with TNF (10 ng/ml) for 6 h and then incubated with THP-1 monocytes. Poldip2 downregulation in RBMECs significantly decreased TNFinduced THP-1 adherence (Fig. 3B), suggesting that Poldip2 in endothelial cells may play a role in regulating leukocyte adhesion during cerebral ischemia. To determine if VCAM-1 is required for Poldip2-mediated effects on monocyte adhesion, we examined the ability of Poldip2 to mediate THP-1 adhesion upon VCAM-1 depletion. TNF (10 ng/ml, for 6 h) significantly enhanced THP-1 adhesion in RBMECs transfected with siControl compared with cells treated with PBS, regardless of whether Poldip2 was also overexpressed. However, in both control and Poldip2 overexpressing RBMECs, depletion of VCAM-1 using siRNA almost completely abolished THP-1 adhesion, indicating that VCAM-1 is required for monocyte adhesion in both control and Poldip2overexpressing situations (Fig. 3C). VCAM-1 gene silencing (Fig. 3D) and Poldip2 overexpression (Fig. 3E) were confirmed using immunoblotting.  20 . To first determine whether TNF induces FAK activation in RBMECs, cells were stimulated with TNF (10 ng/ml) for either 30 min, 1.5, 3 and 6 h and FAK phosphorylation on Y397 (FAKpY397) was assessed by western blotting assay. While total FAK protein levels remained unchanged, FAK phosphorylation was increased in response to 6 h of TNF treatment (Fig. 4A).

FAK is required for TNF-induced VCAM-1 expression in vitro.
We next sought to determine whether FAK is required for TNF induction of VCAM-1 in RBMECs. Monolayers of RBMECs were exposed to increasing concentrations (0.5-2.5 µM) of the FAK inhibitor, PF-562,271 for one hour before co-treatment with TNF for 6 h. VCAM-1, FAKpY397 and total FAK levels were assessed by western blotting. TNF-induced VCAM-1 expression was attenuated in response to FAK-I, reaching a maximum effect at 1 µM (Fig. 5). This result suggests FAK activity is required for TNF-mediated VCAM-1 expression.

Discussion
We previously reported that Poldip2 depletion in vivo protects against BBB permeability in the ischemic brain and improves survival 33 . In this study, we build upon our earlier work and report a novel mechanism by which Poldip2 regulates vascular inflammation and leukocyte recruitment into the ischemic brain. Notably, we observed that Poldip2 depletion in vivo blocked infiltration of CD45 + , CD45 high immune cells, CD11b + CD45 + myeloid cells, as well as Ly6C high CD45 high and MHCII low CD45 high inflammatory monocytes/macrophages. We also show that Poldip2 depletion attenuated the induction of adhesion molecules such as E-selectin, ICAM-1 and VCAM-1. Focusing on VCAM-1, we demonstrate mechanistically that Poldip2 is required for both TNF-induced FAK activation and downstream induction of VCAM-1.
Stroke is a multiphasic process involving direct ischemic brain injury exacerbated by a secondary insult associated with the subsequent inflammatory response. Rapid activation of resident microglia and the production of proinflammatory cytokines, including TNF and IL-6, occur within hours of cerebral ischemia onset 34 . and mRNAs were measured by RT-PCR in the brain tissue 24 h after sham surgery or hypoxia-induced permanent stroke in Poldip2 +/+ and +/− mice. Bar graphs represent means ± SEM from 5 to 9 mice per group normalized to GAPDH. Two-way ANOVA *P < 0.05, **P < 0.01, ***P < 0.001 vs. Poldip2 +/+ sham mice and # P < 0.05 vs. Poldip2 +/+ ischemic mice. www.nature.com/scientificreports/ Inflammatory cytokines, in turn, upregulate adhesion molecules in endothelial cells, promoting transmigration of leukocytes across the BBB and post-ischemic neuroinflammation. Inhibition of leukocyte adhesion and transmigration across the BBB following cerebral ischemia reduces neuronal damage and improves functional outcome 35,36 . Therefore, elucidation of the molecular mechanisms regulating infiltration of inflammatory cells into ischemic tissue remains critical for the identification of potential pharmacological targets aimed at improving stroke outcomes. Previous work from our group has demonstrated that Poldip2 depletion attenuates TNF and IL-6 levels following ischemic stroke, possibly via regulation of NFκB 33 . Both activation of brain resident immune cells and infiltration of blood-derived inflammatory cells contribute to the inflammatory reaction triggered by cerebral ischemia. Given the reduced levels of inflammatory markers in Poldip2 +/− mice 6 , in this report we investigated the contribution of Poldip2 to leukocyte recruitment into brain tissue after cerebral ischemia. Our results show an overall reduction in CD45 + , CD45 high lymphocyte, CD11b + CD45 + myeloid cell, as well as Ly6C high CD45 high and MHCII low CD45 high inflammatory monocyte/macrophage in brains of Poldip2 +/− mice 48 h after cerebral ischemia induction. To investigate how Poldip2 depletion attenuates leukocyte recruitment, the mRNA levels of adhesion molecules were measured. ICAM-1, VCAM-1, and E-selectin mRNA levels were significantly attenuated in Poldip2 +/− animals following ischemic stroke compared to sham controls. Likewise, our in vitro studies demonstrated that depletion of Poldip2 attenuates TNF-induced VCAM-1 expression in RBMECs. While future work will be necessary to determine how Poldip2 regulates ICAM-1 and E-selectin, we focused here on determining the mechanism by which Poldip2 regulates VCAM-1.
TNF induces VCAM-1 expression via activation of the MAPK/FAK signaling pathways, which ultimately results in transcription factors, including AP-1, GATA-binding proteins and NFκB, binding to the VCAM-1 promoter 19 . In vascular smooth muscle cells, depletion of Poldip2 reduces FAK Y397 phosphorylation 20 , which. Raised the question of whether Poldip2 regulates FAK activation in RBMECs. Consistent with earlier findings, depletion of Poldip2 in RBMECs blocks TNF activation of FAKpY397. Additionally, our in vitro studies using a FAK inhibitor, PF-562,271, suggest that FAK is required for TNF-induced VCAM-1 expression in RBMECs. Together, our results suggest Poldip2 regulates TNF-mediated VCAM-1 expression via FAK activation, thus providing a potential mechanism by which Poldip2 mediates leukocyte infiltration, as shown in our in vivo flow cytometry results. Since FAK has also been described to regulate ICAM-1 induction, it is possible that Poldip2 enhances FAK activity to upregulate ICAM-1 37 . Future studies should investigate this potential mechanism.
Our data demonstrate that overexpression of FAK in TNF-treated cells with Poldip2 depletion largely, but not completely, recovers VCAM-1 expression (reduction of 68% between siControl + TNF and siPoldip2 + TNF in the presence of AdLacZ control virus vs. reduction of only 20% between siControl + TNF and siPoldip2 + TNF in the presence of AdFAK virus), suggesting the regulation of TNF-mediated VCAM-1 expression by Poldip2 may involve additional pathways. These additional pathways likely include NFκB, as Poldip2 depletion has previously been reported to inhibit NFκB activation and induction of downstream inflammatory effectors 7,33 . Interestingly, VCAM-1 induction in human pulmonary microvascular ECs was partially attenuated when cells were pre-treated with the mitochondrial ROS scavenger MitoTEMPO, which is consistent with the idea that additional mechanisms, such as FAK and NFκB, may be involved 8 .
Moreover, the specific downstream consequences of FAK activation by Poldip2 in RBMECs remain unknown. As described above, FAK-mediated VCAM-1 induction is associated with the activation of at least two different Figure 5. FAK activity mediates TNF-induced VCAM-1 upregulation. Confluent RBMECs were treated with vehicle or PF-562,271 (0.5-2.5 μM) for 1 h before cells were exposed to TNF (10 ng/ml) for 6 h. VCAM-1, FAKpY397, total FAK and β-tubulin protein levels were evaluated by western blotting. One-way ANOVA *P < 0.05 vs. no TNF and no PF-562,271.

Conclusions
In the context of our previous reports 7,8,33 , our findings provide additional evidence that Poldip2 is a critical regulator of endothelial barrier function and the inflammatory response through its modulation of adhesion molecule expression. Further work is needed to explore the role of Poldip2 in leukocyte activation per se as well as the upstream regulators of Poldip2. Future investigation should also focus on further determining how Poldip2 regulates FAK activation. Ultimately, Poldip2 inhibition may provide therapeutic benefit in a wide spectrum of brain pathologies involving barrier dysfunction.

Data availability
All data generated or analyzed during this study are included in this article. The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.