Introduction
The blood–brain barrier (BBB) and other brain barriers such as the meningeal blood-cerebrospinal fluid barrier are comprised of highly specialized brain endothelial cells (BECs) that promote proper brain function by separating the circulation from the central nervous system (CNS) [
1‐
3]. BECs possess unique phenotypes that include the presence of specialized efflux transporters, complex tight junctions, and low rates of endocytosis [
1‐
3]. Together, these properties maintain brain homeostasis and help to prevent the entry of pathogens and toxins into the CNS. P-glycoprotein (P-gp) is a major efflux transporter expressed in BECs that is able to efflux a wide variety of lipophilic molecules back into the bloodstream and because of this, can regulate the CNS accumulation of drugs [
4‐
9]. Much work has been conducted to understand and modulate P-gp function at the BBB in order to enhance drugs access to the brain [
4‐
9]. Moreover, the effects of disease conditions on P-gp function are increasingly being explored [
10‐
13].
Bacterial meningitis is a serious infection of the CNS that occurs when blood-borne bacteria are able to breach BECs and cause inflammation [
14‐
17]. Certain pathogens possess virulence factors that promote BEC interaction, and previous studies have characterized the molecular interactions that result in penetration of brain endothelium [
14‐
17].
Streptococcus agalactiae, also known as Group B
Streptococcus (GBS), is a Gram-positive bacterium that is the leading cause of neonatal meningitis [
18]. GBS disruption of the BECs and other brain barriers during meningitis has been documented, and it has been shown that tight junctions are disrupted and endocytosis pathways altered [
19,
20]. However, little is known about the effect of bacterial infection on P-gp function. Here, we show that bacterial infection can alter P-gp function in BECs, suggesting another mechanism by which bacterial pathogens contribute to BEC dysfunction.
Discussion
Previous work has shown that GBS can interact with BECs resulting in tight junction disruption, an innate immune response, and alteration of other cellular processes [
14,
19,
20,
36]. Here we observe for the first time that GBS is also able to disrupt BEC P-gp efflux transport. Using iPSC-derived BECs, an
MDR1 (P-gp) overexpressing MDCK line, and a mouse model of GBS infection we have established that P-gp function appears to be diminished during GBS infection.
The impact of GBS on P-gp was independent of substrate given the increased accumulation of two different substrates, R123 and FLUO-3-AM (Fig.
1a, b). Previously, we and others have demonstrated that iPSC-BECs possess functional BCRP and MRP that can be inhibited by Ko143 or MK571 respectively [
21,
22,
24‐
29]. Inhibition of BCRP or MRP family proteins did not result in a significant increase in R123 accumulation suggesting that the CsA- induced increase in R123 accumulation is mediated by P-gp in the iPSC-BEC model (Fig.
1e). The more specific second-generation P-gp inhibitor PSC-833 (Valspodar) [
6], showed a similar impact on R123 accumulation (Fig.
1c). In addition, the effects appear to be dependent on what bacterium is interacting with the BECs. For instance, while GBS affected P-gp function, another Gram-positive bacterium,
L. lactis, that is non-pathogenic, did not impact P-gp function. More work is required to identify the specific bacterial factor in GBS responsible for the loss of P-gp function during infection, as several known GBS invasion mutants did not affect the observed P-gp inhibition (Fig.
2c). However, our data suggest that direct live bacterial challenge is required since neither conditioned medium nor nonviable bacteria resulted in the same functional inhibition of P-gp. Meanwhile, direct interaction of live GBS on a non-BEC cell line overexpressing P-gp was also sufficient for loss of function indicating that the effects may not be BEC-specific. Moreover in BECs, the decrease in P-gp function accompanied a decrease in P-gp expression.
P-gp function in BECs impacting neurological disorders such as Alzheimer’s disease has been examined, where decreased function is associated with more Aβ deposition in the brain [
10,
12]. Interestingly, while modeling Huntington’s disease in iPSC-derived BECs, it was observed that P-gp function was decreased but expression was increased [
11]. P-gp function is also altered in various inflammatory states. Isolation and treatment of primary BECs from guinea pigs with pro-inflammatory cytokines led to a decrease in P-gp function and expression similar to our observations [
37]. In vivo, inflammatory molecules such as lipopolysaccharide (LPS) and cytokines have also been shown to decrease P-gp function while increasing P-gp expression in a murine model [
38]. However in rats, LPS treatment resulted in both lower expression and function of P-gp in the brain, suggesting model and/or species dependent differences [
39]. This disjointed observation between expression levels and function is further confirmed in other P-gp expressing tissues that have been challenged with LPS or bacteria with varying expression levels after treatment [
40‐
45]. Further work will be required to determine mechanisms by which GBS infection diminishes P-gp function. We observed decreased expression of P-gp in BECs on both transcript and protein levels as a result of GBS infection (Fig.
3a–e). However, in the MDCK-MDR1 cells, human P-gp is overexpressed and is not under the control of endogenous P-gp gene regulatory systems [
30], and yet we still observed disruption of P-gp function (Fig.
2a). These data then suggest that in addition to the decrease in expression, a bacterial factor may also directly interact with and contribute to the inhibition of P-gp.
In barrier forming intestinal epithelial cells,
Listeria monocytogenes and
Salmonella enterica serovar
Typhimurium have been shown to disrupt P-gp function [
46,
47]. In the case of
Listeria monocytogenes, P-gp acted as protective factor to invasive infection, and Listerial proteins inhibited normal P-gp function by acting as a competitive substrate [
47]. Additionally,
S. typhimurium infection resulted in R123 accumulation into epithelial monolayers [
46]. GBS is carried in up to 30% of healthy individuals in the gastrointestinal tract and could potentially be interacting with similar P-gp expressing epithelial cells in a colonization state [
14,
48]. Invasive GBS late-onset disease in newborns manifesting in sepsis and meningitis up to 3 months postnatally coincides with increasing P-gp expression in the brain that reaches maximum levels between 3 and 6 months of life [
49,
50]. It is possible that functional P-gp may possess BEC barrier protecting functions, such as in the case with
L. monocytogenes in the gut [
47], and may contribute to the decrease in invasive GBS disease after 3 months of life. Here we demonstrate that GBS can disrupt P-gp function in BECs, however further investigation is required to understand how inhibition might impact BEC barrier function and bacterial invasion.
Our study suggests that a generalized response to bacteria, such as through Toll-like receptor agonists, may not be sufficient to decrease P-gp function since non-pathogenic live bacteria, and nonviable GBS do not decrease P-gp function (Fig.
2b). Future work will be required to determine the precise mechanism of P-gp dysfunction during infection, signaling pathways involved with P-gp downregulation, and impacts other transporters in BECs. Finally, these findings suggest that P-gp inhibition should be taken into consideration when exploring therapeutic strategies for bacterial infection of the CNS.
Materials and methods
Bacterial strains and cell lines used
Induced pluripotent stem cell (iPSC) line IMR-90-C4 (WiCell) was maintained per previous reports and grown on Matrigel (WiCell) coated plates (Corning) in mTeSR1 medium (WiCell) changed daily. IMR-90 iPSCs were passaged twice a week as needed [
21‐
23,
26,
36,
51]. MDCK-MDR1 [
52] (ATCC) cells were maintained on tissue culture treated 25 cm
2 flasks in DMEM (Life Technologies) + 10% Fetal Bovine Serum (WiCell). Group B
Streptococcus (GBS,
Streptococcus agalactiae) hypervirulent clinical isolate COH1 (serotype III, multilocus sequence type 17) strain was used [
53]. Previously described COH1 mutants
Δiag [
33],
Δsrr2 [
54,
55], and
ΔsfbA [
34] were employed for mutant analysis. All GBS strains were grown in static Todd-Hewitt broth (THB) at 37 °C.
Lactococcus lactis was grown in M17 medium at 30 °C in static culture as previously described [
56].
L. lactis was prepared exactly like GBS as mentioned above.
Brain endothelial cell differentiation
iPSC-derived brain microvascular endothelial cells (BECs) were differentiated as previously described [
21‐
23,
26,
36,
51]. To ensure the quality of the cultures, mycoplasma testing was conducted periodically through services from WiCell, and PCR based kit (PanReac AppliChem). Briefly, single cell suspension of iPSCs were seeded at a density of 10,000/cm
2 onto Matrigel (WiCell) cell culture plates or flasks (Corning) and expanded for 3 days in mTeSR1 changing media daily. Initiation of differentiation was conducted by exchanging to unconditioned media (UM; DMEM-F12 base medium [Life Technologies], 20% Knockout serum replacement [Life Technologies], 1% minimal essential medium-nonessential amino acids [Life Technologies], 0.5% Glutamax [Life Technologies], and 0.07% beta-mercaptoethanol [Sigma]), for 6 days changing media daily. After the 6 days, media was then changed to EC medium (human endothelial cell serum-free media [Life Technologies], 1% platelet-poor plasma derived serum [Fisher], 500 ng/ml basic fibroblast growth factor, and 10 μM all trans-retinoic acid [Sigma]) for 2 days. Finally, differentiated BECs were purified onto collagen IV (Sigma) and Fibronectin (Sigma) coated plates and transwells inserts (Corning), and the following day, media was changed to EC medium without basic fibroblast growth factor or retinoic acid. BECs were analyzed for TEER using an EVOM2 instrument (World Precision) to ensure high electrical resistance unique to BECs.
Bacterial infection
iPSC-derived BECs that are purified onto collagen-fibronectin 24 well plates at 500 k cells/well were grown until Day 10 of the differentiation [
23]. A multiplicity of infection (MOI) of 10 was used for all experiments unless specifically noted otherwise as previously described [
36]. Overnight cultures of GBS were grown in THB media at 37 °C + 5% CO
2. The following day bacteria are subcultured and grown to an optical density at 600 nm (OD
600) of 0.400–0.600. Bacteria are then spun down and washed in phosphate-buffered saline (PBS) prior to infecting BECs. BECs were infected for 5 h at 37 °C at 5% CO
2 followed by sample collection or assays. Heat killed and formalin-fixed GBS were prepared by growing WT GBS to an OD
600 of 0.400–0.600 in THB followed by centrifugation and resuspension of the pellet in PBS and treated by heating to 95 °C for 15 min, or treatment of GBS with 4% paraformaldehyde for 15 min. Bacteria were then washed and used to treat BECs at an estimated MOI of 10 for 5 h like live infections described above. Killed GBS preparations were confirmed by plating undiluted fractions onto THB plates where no growth was observed overnight at 37 °C + 5% CO
2. GBS conditioned media was generated by growing WT GBS in endothelial cell assay media for 5 h, then sterile filtered through a 0.2 μM filter as previously described [
19]. Conditioned media preparations were also confirmed sterile by plating undiluted fractions onto THB plates where no growth was observed overnight at 37 °C + 5% CO
2.
P-gp activity assay
P-gp activity is assessed by the accumulation of the fluorescent substrates Rhodamine 123 (R123) (Sigma) and FLUO-3-AM (Thermo) [
23,
57]. The specific P-gp inhibitor cyclosporine A (CsA) is used as a control to monitor P-gp activity [
23]. For other inhibitor experiments PSC-833 (Sigma), MK571 (Sigma), or Ko143 (Sigma) were used to inhibit P-gp, MRPs, and BCRP respectively. Uninfected BECs are used as a control for comparison. After infection, cells are washed with Hank’s Balanced Salt Solution (HBSS) (Thermo) and pre-incubated with or without inhibitor (10 μM CsA, 10 μM PSC-833, 10 μM MK571, or 1 μM Ko143) for 1 h at 37 °C + 5% CO
2. Following pre-incubation, cells were incubated with 10 μM R123 or FLUO-3-AM with or without inhibitors mentioned for 2 h at 37 °C + 5% CO
2. After incubation, cells are washed twice in PBS and 200 μl of RIPA buffer (Thermo) is added and placed on a shaker for 10 min at room temperature protected from light. Fluorescence was measured on a plate reader (Tecan). Next, a BCA protein assay (Thermo) was conducted on each well and fluorescence values are normalized to BCA to account for relative cell number as previously described [
23].
Flow cytometry
To assay for the expression of P-gp, BECs were either infected with GBS or left uninfected as a control. After the infection, cells are washed 3× in PBS and 100 μl of Accutase to each well and incubate at 37 °C + 5% CO2 to remove cells from the plate. Cells are then fixed by adding 900 μl of 1% paraformaldehyde in PBS for 15 min at room temperature. Cells are then washed twice in wash buffer (5% bovine serum albumin (BSA), 0.1% Triton-X in PBS) to block and permeabilize for the staining of total P-gp. Samples were stained in the wash buffer with anti-P-gp clone F4 (Thermo) (1 μg/million cells) overnight at 4 °C. Following primary stain, samples were washed and stained with secondary anti-mouse 488 (Thermo) at 1:5000 in wash buffer for 1 h at room temperature. Cells were washed and data was collected on a MACSQuant Analyzer 10 (Miltenyi Biotec) and analyzed on FlowJo v10.
Western blotting
BECs were infected as described above for 5 h at an MOI of 10. After infection, cells were washed three times in PBS and lysates were taken using RIPA buffer (Thermo) plus HALT protease inhibitor cocktail (Thermo). Proteins were quantified using a standard BCA assay kit (Thermo), and equal amounts of protein were loaded onto Nu-PAGE gels (Thermo) and transferred onto nitrocellulose membranes. Membranes were blocked in 5% milk in tris-buffered saline with 0.1% Tween 20 (TBST) and anti-COX-IV antibody 1:1000 (Cell Signaling Technologies) was used as a protein loading control, and anti-P-gp (F4) (Thermo Fisher) was used to determine P-gp abundance. To visualize protein abundance horseradish peroxidase (HRP)-conjugated secondary antibodies (Jackson Laboratory) and a BioRad ChemiDoc XRS + instrument were used to image blots.
RNA isolation and quantitative PCR
Monolayers of BECs were either left uninfected, or infected with GBS for 5 h at an MOI of 10. After infection, cell lysates and RNA was purified using a NucleoSpin RNA kit (Machery Nagel). cDNA was generated using VILO first-strand synthesis kit (Thermo Fisher), or LunaScript RT (New England BioLabs). SYBR green qPCR was conducted for human
ABCB1 (P-gp, MDR1) forward primer 5′-GAAGAGATTGTGAGGGCAGC-3′, and reverse primer 5′-CCACCAGAGAGCTGAGTTCC-3′. 18S rRNA was used to normalize results and primers used, forward primer 5′-GTAACCCGTTGAACCCCATT-3′ and reverse primer 5′-CCATCCAATCGGTAGTAGCG-3′ were previously described [
58]. qPCR data was collected on an Applied Biosystems StepOnePlus and data are presented as fold change using the cycle threshold (ΔΔC
t) calculation.
Murine model and staining
Animal experiments were approved by the committee on the use and care of animals at the University of Colorado School of Medicine (protocol #00316) and performed using accepted veterinary standards. The University of Colorado School of Medicine is AAALAC accredited and the facilities meet and adhere to the standards in the “Guide for the Care and Use of Laboratory Animals”. We utilized a mouse GBS infection model as previously described [
19,
33,
35]. Briefly, 8 week old male CD-1 mice (Charles River) were injected intravenously with 10
8–10
9 of GBS. At the experimental endpoint mice were euthanized and brain tissue was collected. One half of the brain was frozen in OCT compound (Sakura) and sectioned using a CM1950 cryostat (Leica). Sections were fixed with ice cold methanol (Sigma) for 20 min, blocked with Mouse on Mouse blocking reagent (Vector Labs), and incubated with mouse anti P-glycoprotein antibody clone C219 (ThermoFisher) overnight at 4 °C followed by goat anti-mouse conjugated to Cy3 (Jackson Immunoresearch) and tomato lectin conjugated to DyLight488 (Vector Labs). Coverslips were mounted with Fluoroshield + DAPI (Abcam) and images were taken using a BZ-X710 microscope (Keyence). Pixel intensities of P-gp staining were estimated by using the FIJI image analysis program. Briefly, lectin positive vessels were traced manually and the selection was transposed onto the P-gp image and mean pixel intensity was measured.
Immunofluorescence
iPSC-BECs were fixed and stained for markers exactly as described previously [
23], with the exceptions that anti-CD-31 (Abcam, cat# ab32457), and anti-ZO1 (Proteintech, cat# 21773-1-AP) were utilized in place of referenced reagents. Briefly, BECs were fixed for 15 min in cold methanol and stained overnight at 4 °C. The following day, secondary antibodies anti-mouse 488 (Invitrogen, cat# A11001), and anti-rabbit 555 (Invitrogen, cat# A31572) were used at a dilution of 1:200. Samples were then visualized on a Nikon Eclipse T
i using Nikon NIS image acquisition software.
Statistics
GraphPad Prism version 5.0 (GraphPad Software Inc.) was used for all statistical analysis. For pairwise comparison, 2-tailed Student’s t test was used where appropriate. For multiple comparisons, analysis of variance (ANOVA) followed by Dunnett’s multiple comparisons test was used where appropriate. Data are represented as mean, ± standard deviation (SD) since all raw values are presented. Statistical significance was accepted at a P value of less than 0.05.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.