Expression of tight junction proteins and transporters for xenobiotic metabolism at the blood-CSF barrier during development in the nonhuman primate (P.hamadryas)

Abstract

The choroid plexus (CP) is rich in barrier mechanisms including transporters and enzymes which can influence drug disposition between blood and brain. We have limited knowledge of their state in fetus. We have studied barrier mechanisms along with metabolism and transporters influencing xenobiotics, using RNAseq and protein analysis, in the CP during the second-half of gestation in a nonhuman primate (P.hamadryas). There were no differences in the expression of the tight-junctions at the CP suggesting a well-formed fetal blood-CSF barrier during this period of gestation. Further, the fetal CP express many enzymes for phase I–III metabolisms as well as transporters suggesting that it can greatly influence drug disposition and has a significant machinery to deactivate reactive molecules with only minor gestational changes. In summary, the study suggests that from, at least, midgestation, the CP in the nonhuman primate is restrictive and express most known genes associated with barrier function and transport.

1. Introduction

The choroid plexus, which is situated in the ventricles of the brain, is the main blood-CSF interface. It contains a range of mechanisms to either prevent or facilitate transport of compounds across the choroid plexus epithelium. One important mechanism of the barrier is to protect the brain from potentially harmful exogenous or endogenous compounds. The plexus is well adapted for high transport of compounds from the blood to the cerebrospinal fluid. It contains a richly vascularised core with fenestrated vessels for promoting flux of fluid and compounds. Outside of the blood vessels is a continuous epithelium that is the main cellular barrier. These cells have highly complex tight-junctions insuring selective transfer by forcing flux of compounds across, rather than between, the epithelial cells that contain a battery of transporters and receptors. The tight-junctions have been shown to be restrictive already early in rodent development [1, 2]. Many of the transporters have roles in the production of the cerebrospinal fluid or are important for the influx of nutrients to the central nervous system as well as the homeostasis of the brain.

The epithelial cells contain a range of enzymes and metabolic activities that act as a defence for the central nervous system to reduce xenobiotic loading on the brain (Miller et al 2001; [3]. Metabolism in the plexus can also make lipid-soluble compounds more hydrophilic and increases the time that they are available for further metabolisms or exclusion by the plexus. Drug metabolising mechanisms have been shown to be present throughout the brain but are enriched at the blood-CNS interfaces, such as the blood-brain and blood-CSF barriers, where they can act as an outer defence against access to the brain [4]. Several drug metabolising mechanisms are present in the developing plexus, including the energy-driven ATP-binding cassette proteins such as breast cancer related protein (BCRP) and multi-drug resistance related proteins in the ABCC family [5, 6]. Transport by these pumps may be facilitated by phase II metabolism of compounds including conjugation of glutathione, sulfation and glucuronidation. A wide variety of other solute carriers are also present on choroidal epithelial cells belonging to the SLC- or SLCO-families that are indicated in drug transport. Previous studies have demonstrated that at least some of these mechanisms are working at a similar level in the fetus as in the adult [7]. A WHO sponsored investigation showed that women ingest an average of three prescription medications during pregnancy [8]. There is also a growing concern how environmental toxins affects the development of the brain [9]. An important question in regard to the potential harmful effects on the fetal brain in relation to xenobiotics such as drugs in the maternal circulation is therefore how well the defence mechanisms at the blood-CNS interfaces are developed in the fetus. Recently the gene changes of drug metabolising and antioxidant systems were reported for the developing rat choroid plexus [10]. In order to further our understanding of these defence mechanisms at the choroid plexus in a translational model we sourced choroid plexus from a fetal nonhuman primate, the baboon (P. hamadryas), which has very close development to the human. Our study was conducted during the second half of gestation when the majority of the cortical expansion is occurring in primates [11]. We firstly investigated the genes and proteins associated with cell-junctions at the blood-CSF barrier and secondly explored changes in genes related to drug transport, metabolism, antioxidant defence and CSF production.

2. Material and Methods

2.1 Animals

Female baboons (Papio hamadryas) from the Southwest National Primate Research Center (San Antonio, TX, USA) were used for this study. All procedures were approved by the Institutional Animal Care and Use Committee of the Texas Biomedical Research Institute and conducted in AAALAC-approved facilities. Feeding and management have previously been explained in detail [12]. Food was withdrawn for 16 hours before surgery. Caesarean sections were performed at 90 (50% gestation), 120 (67% gestation) and 165 days (90% gestation) of gestation (term 184 days) under general anaesthesia as previously reported [13]. Following caesarean section maternal analgesia was provided with buprenorphine hydrochloride 0.015 mg/kg per day (Hospital, Inc., Lake Forest, IL, USA) for three days post-surgery. When caesarean section was performed to obtain samples from multiple areas of the brain and other fetal tissues, fetal telencephalic choroid plexus (lateral ventricular plexus) was collected and frozen in liquid nitrogen at gestation day (GD) 90 (2 males/1 female), GD120 (2 males/2 females) and GD165 (3 males/3 females). Choroid plexus was also obtained from two non-pregnant adult females at 9.4 and 10.2 years of age and two pregnant baboons at 9.4 and 11.3 years of age. These adult animals were necropsied as part of a wider study of aging. Where euthanasia was performed on foetuses and adult animals it was conducted by exsanguination while the animal was under general anesthesia as approved by the American Veterinary Medical Association.

2.2 RNA Preparation and RNA sequencing

A detailed description of RNA sequencing has been published elsewhere [14]. Briefly, frozen tissues were homogenised and RNA extracted with an RNAeasy mini-kit (Qiagen) according to manufacturer’s specifictions. RNA quality was analysed by Biorad Experion electrophoresis. RNA-seq was performed at the Genomics Core Facility at University of Gothenburg. Only high quality RNA was used for library preparation (RIN >8). Libraries were created using the TruSeq^™ RNA Sample Preparation v2 kit and the library was subjected to 50bp single end read cycles of sequencing on an Illumina HiScan SQ.

2.3 Analysis of sequencing data

RNA-seq reads were mapped to the current baboon genome assembly (https://www.hgsc.bcm.edu/content/baboon-genome-project) using Tophat [15]. Cufflinks [16] was used for transcript assembly of individual samples. All assemblies were merged to create a reference transcript, which was used to calculate gene counts with HTSeq (http://www-huber.embl.de/users/anders/HTSeq/doc/overview.html). DESeq [17] was then used to find differentially expressed genes. Annotation of genes was done through blasting (blastx) (Altschul et al., 1990) each transcript against the Swiss-Prot database (Uniprot Consortium). Differential expression of transcripts was considered between GD90 and GD165 with p-value considered significant when less than 0.05. The expression levels for genes were divided into five levels for the Results section whereby normalised raw counts between 1–10 were considered very low expression, 10–100 low expression, 100–1000 medium expression and 1000–10000 high expression and >10000 very high expression. The cut-off detection level was set to 1 (i.e. all transcripts where the raw read count was less than 1 across all ages were removed). In addition, we have not further deliberated on significant changes in genes for which expression was limited to the very low expression range (<10 counts) across all developmental ages. Changes in genes were visualised in heatmaps generated in the statistical software R along with cluster analysis. We have used GO terms for human proteins annotated to tight-junction and cell-cell adhesion to make relevant heatmaps of genes associated to these terms, however, genes were also manually collated with duplicates removed.

2.4 Immunohistochemistry (IHC)

After fixation for 24 hours in 4% paraformaldehyde, plexuses from fetuses at GD120, GD165 and adults (no GD90 animals were available) were dehydrated in increasing concentrations of alcohol and embedded in paraffin. Serial sections were cut on a microtome and prepared for IHC. Paraffin was removed from sections in xylene, sections were rehydrated in decreasing concentrations of ethanol and gently boiled in 0.01 M citrate buffer or protease digestion (Streptomyces griseus, Sigma; 1 mg/ml at 37°C) for 10 min. Blocking steps included H₂O₂ treatment (3% in methanol for 10min) and appropriate serum (5%) or DAKO serum-free protein block for 1 hour. Sections were then incubated in the following primary antibodies towards three key plexus tight-junctional proteins: ZO-1 (1:100; Cat#61-7300, Invitrogen), occludin (1:100; Cat#71-1500, Invitrogen) and claudin-1 (1:200; Cat#71-1500, Invitrogen). They were then incubated in biotinylated secondary against the appropriate species followed by the ABC kit (Vector) or HRP-labelled secondary antibody (BrightVision poly HRP-anti-rabbit IgG from ImmunoLogic) and developed with DAB. In between all steps, sections were washed in PBS with 0.1% tween20. After 5 minute in DAB solution, sections were washed in water, dehydrated and cover-slipped. Blank sections were obtained when primary antibody was omitted.

2.5 Protein analysis

Choroid plexus tissues (from the same samples as for RNAseq) were homogenised by sonication in ice-cold PBS containing 5mM EDTA, protease inhibitor cocktail (PIC, P8340, Sigma), phosphatase inhibitor cocktail 1 (P2850, Sigma), phosphatase inhibitor cocktail 2 (P5726, Sigma). The protein concentration was determined using a spectrophotometer at 280-nm absorbance. Sample lysates were mixed with 5X Laemmeli sample buffer and heated (70 C) for 10 min. To detect occludin and claudin-1 proteins 10 μg of total protein extract was resolve on 4–12% Tris–glycine gels (Novex, San Diego, CA, USA); For ZO-1 20μg of total protein was resolve on 3–8% Tris–acetate gel (Novex, San Diego, CA, USA). After electrophoresis, proteins were transferred to reinforced nitrocellulose membranes (Schleicher & Schuell, Dassel, Germany) and membranes blocked in 30 mM Tris–HCl (pH 7.5), 100 mM NaCl and 0.1% Tween-20 (TBS-T) containing 5% fat-free milk powder for 1 h at room temperature. Then, the membranes were incubated overnight at 4 °C with the following primary antibodies in 5% bovine serum albumin (BSA)-TBST: rabbit anti–occludin, 1:2000, rabbit-anti claudin-1, 1:1000, rabbit anti-ZO-1, 1:400 (Invitrogen, CA, USA) and rabbit anti-β-actin, 1:10000 (Sigma-Aldrich, St. Louis, MO, USA) or mouse anti-GADPH, 1:500 (Abcam, Cambridge Science Park, Cambridge, UK). After washing, the membranes were incubated with peroxidase-labelled goat anti-rabbit or anti mouse (1: 5000) secondary antibody (Vector Laboratories, Burlingame, CA, USA) at room temperature. Immunoreactive species were visualised using the Super Signal Western Dura substrate (Pierce, Rockford, IL, USA) and a LAS 1000-cooled CCD camera (Fujifilm, Tokyo, Japan). Immunoreactive bands were quantified using the Image Gauge software (Fujifilm) and normalised with respect to β-actin or GADPH. Data were analysed across ages with one-way ANOVA using SPSS software (α<0.05).

3. RESULTS

3.1 Tight-junctional proteins

3.1.1 Western blotting and immunolocalisation of tight-junctional proteins

We used immunohistochemistry to investigate the localisation of occludin, ZO-1 and claudin-1 proteins in the fetus and adult. All three proteins showed almost identical immunolocalisation at the apical cell membrane of epithelial cells (Figure 1). There was no difference in the staining pattern between the fetal ages or adults. We also used western blotting to quantitatively investigate the expression level of the three tight-junctional proteins in the fetal plexus (Figure 2). Western blotting produced bands in blots at expected sizes for each tight junctional protein (occludin ~ 60KDa, claudin-1 ~ 20KDa and ZO-1 ~ 225KDa) although an additional band at around 70KDa was detected for ZO-1. Protein quantification using western blotting showed that all three proteins were present in the choroid plexus in the 90, 120 and 165 GD plexus. There were only minor changes in protein levels of all three proteins during fetal development and it was only ZO-1 that showed a somewhat higher level at the intermediate age (GD120) but this did not reach statistical significance (p=0.13).

Figure 1.

A) Immunoreactivity for claudin-1, ZO-1 and occludin in choroid plexus of gestational day (GD) 90, GD165 fetuses and in adults. Immunoreactivity in tissue was only visible at the apical cell membrane of the epithelial cells and not in other parts of the plexus. There was no apparent difference in the staining pattern in fetuses compared to adult tissue. Tissue was counterstained with hematoxylin. B) Higher power image of occludin staining showing that staining was present on the apical side of epithelial cells (arrows), where the tight-junctions are situated, whereas no staining was present on the basal side of cells (arrowheads) in fetuses. C) Control slides where primary antibodies were omitted resulted in no staining of plexus. Scale bar is 50μm in A and 20μm in C.

Figure 2.

Protein quantification using western blotting of three tight-junctional associated protein occludin, ZO-1 and claudin-1 in the fetal nonhuman primate plexus at gestational day (GD) 90, 120 and 165. Western blot analysis showed only minor differences in protein levels between the three gestational ages and with no statistically significant differences (one-way ANOVA). Data are mean ± SEM.

3.1.2 Tight-junction and cell-adhesion associated genes

To further investigate cell-junction mechanisms at the fetal blood-CSF barrier we collated all genes associated to tight-junction and cell-adhesion across the three gestational ages (Figure 3). For tight-junction associated genes (Figure 3A) the GD90 animals showed the highest transcription of these genes with few changes in the genes between GD120 and GD165. This was also signified in the cluster analysis, which showed that GD90 animals form a separate group whereas GD120 and GD165 animals form a mixed group. A more distinct pattern is seen for cell-adhesion genes (Figure 3B). In this group of genes there was a stepwise reduction of transcription during gestation with GD90 having the highest transcription and GD165 the lowest. The cluster analysis for cell-adhesion genes showed that all animals cluster consistently within their respective gestational age group.

Figure 3.

Heatmap generated from the read counts of tight-junction associated genes (A) and cell-cell adhesion genes (B) along with cluster analysis. Each column in the heatmap represents an individual animal and the cluster analysis shows which animals have the closest relationship based on these genes. Note that expression across the genes was highest at the earliest gestational age and cluster analysis shows separation of animals into their respective gestation age except the GD120 and 165 animals in A. GD - Gestational Day.

3.2 Xenobiotic Metabolism and Transport

Xenobiotic metabolism is commonly divided into three phases whereby in phase I polar groups are added to compounds and in phase II, charged species are added to compounds such as glutathione, sulphate or glucuronic acid. Phase I/II metabolism makes compounds less lipid soluble and less likely to cross into the brain. In phase III metabolisms compounds are excreted by transport proteins such as the ATP-binding cassette proteins (ABC-proteins). We divided changes in genes in these three stages of metabolism and further explored other solute carriers that are important for xenobiotic transport as well as oxidative stress. Age-dependent regulation of proteins refers to significant expression differences between GD90 and GD165. Results are presented in Tables 1–5.

Table 1.

Transcripts associated to Phase I Metabolism. Fold changes (FC) and adjusted p-values were detected between gestational day (GD) 90 and GD165.

GENE	GD90	GD120	GD165	FC	p-value
CYP1B1				1.49	0.857
CYP2A13*				2.08	0.464
CYP2C8*				−4.32	0.128
CYP2C18				−1.12	0.831
CYP2C19				2.21	0.235
CYP2J2				1.03	1.000
CYP2R1				1.43	0.081
CYP2S1*				3.02	0.026
CYP2W1				−25.37	0.000
CYP3A5				−1.23	0.849
CYP4B1				−1.12	0.892
CYP4F12				4.64	0.002
CYP4V12				−1.10	0.764
CYP4F11*				−1.25	0.812
CYP4X1*				1.18	0.924
CYP4Z1*				−4.11	0.407
CYP17A1				1.88	0.064
CYP20A1				1.25	0.314
CYP21A2				1.37	0.376
CYP26B1				−1.27	0.842
CYP27A1				−1.16	0.485
CYP27B1*				1.18	0.840
CYP27C1				−1.75	0.003
CYP39A1				1.37	0.103
CYP51A1				1.13	0.730
MAOA				−2.49	0.015
MAOB				1.60	0.005
EPHX1				1.81	0.000
EPHX2				1.60	0.004
EPHX3				1.23	0.349
EPHX4				−2.06	0.010

^*<10 counts.

Colour indexing for number of read counts at each gestational age:

1 1000 10000 (counts)

Table 5.

Transcripts for antioxidant associated genes. Fold changes (FC) and adjusted p-values were detected between gestational day (GD) 90 and GD165.

Gene	GD90	GD120	GD165	FC	p-value
SOD1				2.39	0.000
SOD2				1.39	0.138
SOD3				21.9	0.000
CCS				1.34	0.184
PGC1A				1.75	0.002
PGC1B*				−1.98	0.565
NRF1				−1.18	0.420
NRF2				1.35	0.135
KEAP1				−1.01	0.986
HMOX1				1.45	0.714
HMOX2				1.28	0.221

^*<10 counts.

Colour indexing for number of read counts:

1 1000 10000

3.2.1 Xenobiotic Metabolisms - Phase I

In total 25 different iso-enzymes of cytochrome P-450 monooxygenases (CYPs) were detected (Table 1; Suppl. Fig. 1). The highest number of transcripts was found in the CYP2 (CYPA6/C8/C18/C19/J2/R1/S1/W1) and CYP4 (B1/F11/F12/V2/X1/Z1) families representing 56% of all CYPs detected. Only CYP1B1 and CYP3A5 were detected within their respective family. Most were in the very low or low expression level with five in the medium level (CYP1B1, CYP2R1/W1, CYP4F12/V2). Three CYPs showed regulation during fetal development with CYP2S1 increasing 3.02 fold, CYP2W1 decreasing 25.4 times and CYP4F12 increasing 4.64 fold from GD90 to GD165. Transcripts for the epoxide hydrolases EPHX1/2 were transcribed in the high range, EPHX3 in medium range and EPHX4 in the low range. Up-regulation during development occurred in both EPHX1/2 with 1.81 and 1.60 fold higher expression at GD165 compared to GD90, whereas EPHX4 was down-regulated 2.06 times during gestation. Transcription for monoamine oxidase MAO1 was in the medium range and MAO2 in the high range. MAO1 was significantly down-regulated from GD90 to GD165 (−2.49 times) whereas MAO2 was significantly upregulated (1.53 times).

3.2.2 Xenobiotic Metabolism - Phase II

We detected in total fourteen Glutathione S-transfereases (GSTs) belonging to the α (GSTA1/3/4), (GSTM1/2/3/4/5), κ (GSTK1), ω (GSTO1), (GSTP1), ζ (GSTZ1) and θ (GSTT1/2) classes with the only class not detectable being σ (Table 2; Suppl. Fig 2). Nearly all GST transcripts were in the medium or high range of expression with highest expression levels found for GSTM3, GSTM5, GSTK1 and GSTP1. Five microsomal GSTs (MGST1-3, LTC4S, PTGES) were also detected. MGST1-3 and LTC4S were transcribed at the medium range with no developmental regulation whereas was PTGES expressed at low level. The only regulated soluble GSTs were GSTK1 which was increased 2.49 during gestation whereas the only regulated microsomal GSTs was PTGES, which was down-regulated 1.86 times. Cytosolic glutathione reductase (GSR), maintaining glutathione in the reduced form, showed high expression levels across all ages and was significantly upregulated 1.96 times during gestation. GCLM (glutamate-cysteine ligase regulatory subunit) and GCLC (glutamate-cysteine ligase catalytic subunit), the first rate limiting enzymes in glutathione production, showed high and medium expression levels with no fetal regulation. Glutathione synthetise (GSS), which catalyses the last step of glutathione production showed medium expression with no developmental regulation.

Table 2.

Transcripts associated to Phase II Metabolism. Fold changes (FC) and adjusted p-values were detected between gestational day (GD) 90 and GD165.

Gene	GD90	GD120	GD165	FC	p-value
SULT1A1				1.07	0.902
SULT1A3				1.13	0.679
SULT1C4				2.40	0.024
SULT1E1				−10.79	0.000
SULT4A1				4.37	0.490
GSTA1*				−2.75	0.264
GSTA3				−1.09	0.794
GSTA4				1.24	0.270
GSTK1				2.49	0.000
GSTM1				−1.29	0.266
GSTM2				−1.23	0.491
GSTM3				1.42	0.064
GSTM4				1.34	0.273
GSTM5				1.05	0.825
GSTO1				1.35	0.151
GSTP1				1.00	1.000
GSTT1*				−1.04	1.000
GSTT2				1.43	0.069
GSTT4				−1.03	1.000
GSTZ1				1.49	0.051
MGST1				−2.29	0.000
MGST2				1.94	0.003
MGST3				1.41	0.059
LTC4S				1.74	0.148
PTGES				−1.86	0.038
GSR				2.05	0.005
GCLM				1.04	0.908
GCLC				1.26	0.275
GSS				−1.02	0.959
UGT3A1*				−3.05	0.729
NAT1				−1.96	0.169
NAT2*				1.06	1.000
TPMT				1.42	0.090
COMT				1.24	0.377

^*<10 counts.

Colour indexing for number of read counts:

1 1000 10000

Five sulfotransferases (SULTs) were detected in the choroid plexus (SULT1A1/A3/C4/E1, SULT4E1), with the highest expression being SULT1A3 with no developmental regulation. The only age-regulated SULT gene was SULT1E1, which was down-regulated 11.6 fold from GD90 to GD165. Only one UDP-Glucuronosyltransferase, UGT3A1, was detected in the very low level of expression at GD90/120 but was under the detection range at GD165.

Some other transferases are also considered important for phase II metabolism of xenobiotics [18]. Expression of the aralamine N- arylamine N-acetyltransferases NAT1/2 were detected, but both transcribed at a very low level. We could detect both thiopurine S-methyltransferase (TSMT) and catechol O-methyltransferase (COMT) at a high expression level in the choroid plexus with no developmental regulation.

3.2.3. Xenobiotic Metabolisms - Phase III

It is mainly the ABCB- and ABCC-families as well as ABCG2 that are important for xenobiotic transport. Within the ABCC family, encoding for the multi-drug resistance related proteins (MRPs), we detected nine variants (ABCC1-6, ABCC8-10), all in very low to medium expression level with ABCC1/C4 having the highest expression across all ages (Table 3; Suppl. Fig. 3). ABCC1/MRP1 and ABCC4/MRP4 were the only regulated transcripts with ABCC1 down-regulated 2.37 times whereas ABCC4 was upregulated 2.44 times from GD90 to GD165. We could also specifically identify isoform 6 and 8 of ABCC1 protein, which both showed significant down-regulation with age (−3.26 and −4.88 fold, respectively). Within the ABCB family 6 forms were detected (ABCB1, ABCB6-10) with ABCB1/B8 in the medium range and ABCB6 in the high range. ABCB1/MDR1 was down-regulated 2.22 times during gestation with no other ABCB genes regulated. ABCG2/BCRP decreased 1.91 times from GD90 to GD165 but the change was not significant (p=0.09).

Table 3.

Transcripts for ATP-binding cassette proteins. Fold changes (FC) and adjusted p-values were detected between gestational day (GD) 90 and GD165.

Gene ID	GD90	GD120	GD165	FC	p-value
ABCB1				−2.22	0.007
ABCB6				1.01	0.998
ABCB7*				−1.41	0.549
ABCB8				1.81	0.112
ABCB9				−1.13	0.776
ABCB10				1.10	0.751
ABCC1				−2.37	0.000
ABCC2				−1.38	0.076
ABCC3*				−1.53	0.561
ABCC4				2.44	0.000
ABCC5*				−1.59	0.401
ABCC7				−1.99	0.142
ABCC8*				1.07	0.977
ABCC9				−1.49	0.259
ABCC10				−1.53	0.024
ABCG2				−1.91	0.091

^*<10 counts.

Colour indexing for number of read counts:

1 1000 10000

3.2.4. SLC transporters

Within the solute carriers it is mainly the SLCO, SLC22, SLC47 and SLC15 families that are considered important in relation to pharmacological toxicology (Nigam 2015). We could detect transcription of eight genes in the SLCO family (formerly known as SLC21 family) with two in the very low (SLCO1B7, SLCO4A1), one in the low (SLCO1B1), three in medium (SLCO1C1, SLCO2A1, SLCO2B1) and three (SLCO1A2, SLCO1B3, SLCO3A1) in the high range (Table 4; Suppl. Fig 4). All but one (SLCO4A1) had significantly higher transcription at GD165 than at GD90 (1.49–27.5 times higher). Ten members of the SLC22A family were detected in the choroid plexus. Seven were in the very low or low range (SLC22A2/A3/A15/A16/A18), three in the medium range (SLC22A6/A20/A23) and two in the high range (SLC22A5/A8). SLC22A8 was regulated during development with 2.31 fold higher expression at GD165 than at GD90 as well as SLC22A23, which was down-regulated 1.58 times during gestation. Within the SLC15 class of solute carriers we detected SLC15A2/A3/A4 expression. SLC15A4 showed expression in the high range, SLC15A3 in the medium range and SLC15A2 in the low range. Only SLC15A2 showed age-dependent regulation whereby expression was significantly lower at GD165 than at GD90 (−2.34 times). We could detect medium expression of SLC47A1 and high expression of SLC47A2 in choroid plexus across all gestational ages. Both genes were upregulated from GD90 to GD165 with SLC47A1 increasing 4.47 times and SLC47A2 7.62 times.

Table 4.

Transcripts for solute carriers in SLCO, SLC15/22/47 families. Fold changes (FC) and adjusted p-values were detected between gestational day (GD) 90 and GD165.

Gene	GD90	GD120	GD165	FC	p-value
SLCO1A2				5.72	0.000
SLCO1B1				3.27	0.018
SLCO1B3				1.90	0.000
SLCO1C1				27.5	0.000
SLCO2A1				2.52	0.011
SLCO2B1				1.64	0.046
SLCO3A1				1.49	0.025
SLCO4A1*				−5.97	0.094
SLC15A2				−2.34	0.031
SLC15A3				−1.17	0.518
SLC15A4				−1.07	0.795
SLC22A2				6.53	0.626
SLC22A3*				−1.20	0.982
SLC22A5				−1.09	0.779
SLC22A6				1.15	0.620
SLC22A8				2.31	0.003
SLC22A15				−1.09	0.887
SLC22A16*				−2.40	0.443
SLC22A18				1.26	0.522
SLC22A20				1.24	0.774
SLC22A23				−1.31	0.242
SLC47A1				4.47	0.000
SLC47A2				7.62	0.000

^*<10 counts.

Colour indexing for number of read counts:

1 1000 10000

3.3 Antioxidant systems

Superoxide dismutase (SOD) 1 (soluble from), SOD2 (mitochondrial form) and SOD3 (extracellular form) were all detected in the choroid plexus (Table 5; Suppl. Fig 5). SOD1 showed high expression, SOD2 medium expression and SOD3 medium expression at GD90/GD120 but high at GD165. Both SOD1 and SOD3 were significantly increased at GD165 compared to GD90 increasing 2.39 and 21.9 times, respectively. The copper chaperone CSS gene, important for copper delivery to SOD1, was transcribed at medium levels across all ages with no developmental regulation. Both heme oxyganse-1 (HMOX1) and HMOX2 were detected in the medium range, but with HMOX2 about 3 times higher than HMOX1. None were regulated with development of the choroid plexus.

Two factors are thought to have particular importance in turning on antioxidant response elements (ARE): peroxisome proliferator-activated receptor gamma coactivator-1 alpha (PGC1α) and nuclear factor-erythroid 2 (NF-E2) related factor 2 (NRF2). PGC1A was expressed at medium levels at GD90 but high levels at GD165 and this increase (+1.75 times) was significant. PGC1B was expressed at a very low level. The transcript of NFE2L2 (encoding for NRF2) was expressed at high levels at all ages studied with no developmental regulation. Also the NRF2 regulating factor Kelch-like ECH-associated protein 1 (KEAP1) was expressed at high levels at all ages with no developmental regulation. Likewise, we found high expression of NRF1 gene (NF2L1) with no development regulation.

3.4 CSF production

The molecular mechanisms for the production of CSF are still not completely understood but are thought to rely on a range of membrane transporters, water channels, ion channels and enzymes. We have presented transcripts related to CSF production as based on Brown et al [19] and Janssen et al [20] giving a list of 41 transcripts in total that were detectable in the baboon choroid plexus (Table 6; Suppl. Fig 6). We found medium expression level for five of these (AT1B1, CA4, SLC4A8, CLCN3, KCNF1), high expression for eleven of these (AT1A2, AT1B1-3, CA2/14, SLC4A5, SLC12A2/A7, CLCN7, KCNH2) and very high for two transcripts (AQP1 and CA2). Transcripts that were up-regulated between GD90 and GD165 were AQP1 (1.91 times), AT1A2 (3.63), AT1B1 (4.11), AT1B2 (1.83), CA13 (2.61), SLC4A8 (2.57), SLC12A2 (1.40), KCNH2 (1.94), KCNK1 (2.63) and down-regulated were CA9 (−1.94), CLCN3 (−2.74), KCNA4 (−4.65), SCN5A (−4.88).

Table 6.

Transcripts for genes associated with CSF production. Fold changes (FC) and adjusted p-values were detected between gestational day (GD) 90 and GD165.

Gene	GD90	GD120	GD165	FC	p-value
AQP1				1.92	0.000
AQP3				1.01	0.980
AQP5*				−1.01	1.000
AQP7*				2.40	0.200
AQP8*				1.46	0.561
AT1A2				3.63	0.019
AT1A3				−1.07	1.000
AT1B1				4.11	0.000
AT1B2				1.83	0.000
AT1B3				−1.27	0.222
CA1				8.39	0.425
CA2				1.11	0.734
CA4				1.14	0.824
CA9				−1.94	0.035
CA12				1.06	0.960
CA13				2.61	0.007
CA14				1.08	0.867
SLC4A4				2.18	0.451
SLC4A5				−1.17	0.431
SLC4A7*				−2.28	0.260
SLC4A8				2.57	0.000
SLC12A2				1.40	0.049
SLC12A5				3.18	0.418
SLC12A7				−1.20	0.356
CLCN3				−2.74	0.000
CLCN5*				1.18	0.992
CLCN7				1.00	1.000
KCNA4				−4.65	0.005
KCNA5				1.37	0.633
KCNE1				−1.32	0.470
KCNF1				−1.23	0.428
KCNG1				−1.68	0.675
KCNH2				1.94	0.000
KCNK1				2.63	0.022
SCN1B				1.59	0.093
SCN2A*				1.91	0.912
SCN2B				−1.60	0.507
SCN4A*				30 −2.19	0.470
SCN4B				−3.10	0.278
SCN5A				−4.88	0.000
SCN8A				1.18	0.618

^*<10 counts.

Colour indexing for number of read counts:

1 1,000 10,000 35,000

4. DISCUSSION

We have investigated the blood-CSF barrier during the second half of nonhuman primate gestation using protein analysis for tight-junctional proteins and whole genome profiling, exploring enzymes and transporters with importance for xenobiotic metabolism and transfer as well as CSF production. Overall the data strongly indicate a well-developed choroid plexus barrier, with respect to tight junctions and development of many enzyme and xenobiotic transport mechanisms by mid-gestation. However, we did detect some genes associated with antioxidant systems being lower at midgestation compared to near-term.

4.1 Tight junction proteins

Firstly, we were interested in the age-dependent state of cell-junctions that make up the physical barrier at the choroid plexus. We focused on three key tight-junctional proteins; the transmembrane proteins occludin and claudin-1 as well as the tight-junctional anchoring protein ZO-1. These proteins were expressed in the apical cell membrane of epithelial cells as expected based on previous ultrastructure studies of choroidal epithelial cells [1] and mouse and human choroid plexus epithelium [21, 22]. We then examined the protein level of the tight junctions during gestation and showed similar expression across all ages. In order to further analyse molecular changes of cell-junction we explored gene transcription of tight-junction and cell-adhesion associated genes. This analysis showed that the highest level of transcription for both these groups of genes occurred at the earliest gestational age studied (GD90). Taken together, these data substantiates the overall notion that the molecular basis for cell-junctions at the blood-CSF is well established at mid-gestation in the nonhuman primate although this needs to be tested and confirmed with functional studies. Additionally, the high expression of many solute carriers and ion transporters support a functioning diffusion barrier at the plexus. The data corroborate with previous functional studies in rats and marsupials which have shown that these tight-junctions are restrictive to even small molecules soon after choroid plexus differentiation [1, 2, 23] and with presence of tight-junctions proteins at the apical membrane in the fetal human plexus [22]. Restrictive cell junctions is of fundamental importance to the choroid plexus in order to operate as a barrier forming tissue since it insures flux of compounds through the epithelial cells (not between cells) and is a prerequisite for selective transfer of compounds in/out of CSF. In addition, this makes it achievable to maintain concentration gradients between blood and CSF. Such concentration gradients between CSF and blood have been shown to be established already by gestational day 50 (earliest age studied) in the rhesus monkey [24] and the authors conclude that this shows the presence of effective passive permeability barriers between blood and CSF in the fetus supporting our results.

4.2 Xenobiotic metabolism

As compounds pass through the cells in the choroid plexus they can be subjected to facilitated influx/efflux by transporters as well as exposed to metabolism. In addition, the choroid plexus is thought to maintain homeostasis in the brain by removing substances from CSF. We thus further explored changes in genes related to xenobiotic metabolism, efflux proteins, solute carriers as well as oxidative stress at the blood-CSF barrier.

4.2.1 Phase I: xenobiotic metabolism

The most important group of compounds for phase I metabolism is the CYP superfamily. We detected in total 25 different CYP transcripts, which amounts to around one third of all known human CYP monooxygenase genes. In contrast, only a small number of CYPs were found in the rat choroid plexus [10]. It is uncertain if the variance in numbers is due to different analysis platforms between studies or if it is a species difference between primates and rodents. We have data from neonatal mouse choroid plexus using the same Illumina platform as used in this study showing a similar number of CYPs in the baboon as in the mouse plexus (A. Motaheddin, unpublished). Although many CYPs were detected in the present study, most were expressed at low levels and their biological significance during development remains unclear. Similar to reports on the rat choroid plexus [10] we found much higher expression levels for the epoxide hydrolases and MAOs than CYPs. Overall there was also little regulation during gestation among the CYP genes (4/25 regulated). On the other hand 3/4 epoxy hydrolases were regulated, including the most abundant phase I transcripts, MOA-B and EPX1/2. The role of monoamine oxidases in xenobiotic metabolisms is not well studied however and our present data suggest that they may play a more important role for phase I metabolisms in the developing plexus than the CYPs.

4.2.2 Phase II: xenobiotic metabolism

Transferases play a central role in detoxification processes tagging xenobiotics or waste products for elimination. The transferases glutathione-s-transferases (GSTs), sulfotransferases (SULTs) and UDP-glucuronosyltransferases (UGTs) are important enzymes for phase II metabolism [18]. The primate choroid plexus was rich in GSTs with a total of fourteen transcripts detected. It is thought that it is mainly the α, μ and π GST classes that are involved in drug metabolism [25]. Ten out of the 14 detected GSTs belong to these three classes suggesting a prominent role for GSTs in xenobiotic metabolism in the non-human primate choroid plexus during gestation. In addition, we found five microsomal GSTs. Microsomal GSTs are membrane bound and localised to the mitochondria and endoplasmic reticulum and are thought to protect these organelles from oxidative stress [25]. However, the exact role of the microsomal GSTs in relation to xenobiotic metabolisms is unclear at the moment and it has been difficult to study their specific contribution since they are masked by the activity of cytosolic GSTs.

Five SULTs were detected in the choroid plexus including SULT1A1, which has previously been shown to be transcribed and active in the rat choroid plexus [26]. The only regulated SULT-associated gene during gestation was SULT1E1 (−10.8 times), which catalyses the sulphonation of estradiol and estrone so may play an important role for estrogen uptake into the brain during development. There was a striking lack of transcripts for UGTs suggesting a minor role for these in phase II metabolism at the blood-CSF barrier. In support, studies indicate that glucuronidation by UGTs mainly occurs in the liver [27].

From our data there is little to suggest that phase II metabolism enzymes changes during the second half of fetal development. Instead there is a considerable phase II metabolic capacity in the fetal choroid plexus mainly within the GST family. The data also suggest that the plexus is well equipped for production of glutathione. This is in agreement with previous studies showing high GST activity in the developing rat and human choroid plexus [7].

4.2.3 Phase III: xenobiotic metabolism

ABC-transporters can significantly alter drug distribution across cellular membranes such as the gut epithelium. Their prominent role in reducing entry of various drugs at the blood-CNS interfaces, including anti-cancer medication, antibiotics and immunosuppresants as well as environmental toxins, is well documented [28–30]. Although these transporters largely appear to work in conjugation with phase II metabolism some have also been shown to be able to directly efflux drugs that have not undergone previous metabolism [29]. In the present study we found a broad complement of ABC-transporters being transcribed at the blood-CSF barrier across all ages with very small changes during gestation. Since ABC-transporters are membrane bound their cellular localisation becomes important for the interpretation. MRP1 has been shown to specifically localise to the basolateral membrane of choroidal epithelial cells [5, 6] suggesting it does contribute to keeping xenobiotics out of the brain. Studies by Wijnold and colleagues have also suggested that MRP1/ABCC1 presence at the choroid plexus can significantly reduce drug disposition into the brain [31]. It is enriched at the blood-CSF barrier compared to the blood-brain barrier [6] and protein expression localise in the choroid plexus, but not in the blood-brain barrier [5, 6]. The down-regulation we detected in ABCC1 is in contrast to rats where an increase has been shown during development in the plexus [5, 10], however, agrees with data from mice [32]. ABCC4/MRP4 has also been shown to localise to basolateral membrane of choroidal epithelial cells and limits drug movement of topotecan from blood to CSF [33]. Cellular protein localisation of the many MRPs is uncertain across epithelia and endothelia due to the difficulty in reliable antibodies. In the ABCB family we could detect ABCB1 and ABCB6-10 with ABCB6 the only transcript that showed high expression. To our knowledge there is no previous reports on MRP6/ABCC6 importance for drug disposition at the blood-brain barriers but it is suspected to confer anticancer drug resistance in cell lines and in particular protect against arsenic cytotoxicity [34]. The best known transporter related to drug efflux in this family is ABCB1 also known as p-glycoprotein. P-gp/ABCB1 can efflux drugs directly when passing through the cell membrane. P-gp deficient mice have dramatically increased brain tissue distribution and toxicity to several drugs such as cyclosporin A, etoposide, digoxin and dexamethasone [35] but the present data and previous studies suggest minor role at the blood-CSF barrier. ABCB1 was expressed in the medium range and was significantly down-regulated from GD90 to GD165 (2.32 times), in agreement with studies of the rat choroid plexus with development [10] although it has been shown not to change in mice (Liddelow et al. 2012).

4.2.4 Solute carriers

The SLCO family (Solute Carrier Organic Anion), also known as organic anion transporting polypeptides (OATPs), is present on various epithelia across the body mediating primarily transport of large hydrophobic organic anions. Drug pharmacokinetics that are influenced by these transporters include B-blockers, statins and artans but numerous other drugs have been implemented in their transport and a very comprehensive list is given by [36]. This group of transporters showed the most consistent change during gestation with 7/8 transporters increasing during gestation with SLCO1C1 the most regulated transporter in this family (+27.5 times). Slco1c1 is also highly increased during development in the rat [10]. The SLCO1C1/OATP-F transporter has particular high affinity for hormones thyroxine and estrogen [37] but several of the other SLCO transporters also have affinity for these hormones [36]. The lower expression of these transporters at earlier gestation substantiates the question raised by [10] how sufficient levels of these hormones, which are important for early brain development, reach the brain. There is a particular high divergence of genes encoding for these transporters between human and rodents making comparisons and translation of functional studies difficult. For instance, SLCO1A2 in humans encoding for OATP1A2 has five orthologues in the rat, slco1a1 and slco1a3-6. Within the SLC22 family we found high expression of SLC22A5/A8 similar to the rat choroid plexus. [38] found immunolocalisation of SLC22A6/OAT1 and SLC22A8/OAT3 to the human choroid plexus but membrane localisation was uncertain. In rats it has been shown that SLC22A6/OAT1 is present on the apical membrane is believed to remove substances from the CSF [39]. To our knowledge the membrane localisation of SLC22A5/OCTN2 or SLC22A8/OAT3 is not known in either rodents or human but would be of interest to know. There is considerable overlap of substrate specificity for these transporters making it difficult to study and interpret complement and changes at the choroid plexus. However, in general solute carriers facilitate the entry of compounds into the plexus in contrast to ABC transporters which limit their entry. The high expression of many SLCs in the plexus suggest that it contributes to the supply of nutrients to the developing brain and that within the outlined SLC families there are minor changes in gestation regarding transport capacity.

We found a similar complement of SLC15 family of genes in the nonhuman primate as that previously reported in the rat choroid plexus [10]. In this family SLC15A2/PEPT2 has been shown to be important in relation to drug transport and has affinity for peptide-mimetic drugs (such as aclovir) and is present on the apical membrane on epithelial cells in the choroid plexus. Studies using choroid plexus epithelial cultures and PEPT2 knockout mice have suggested it removes peptides from the CSF [40, 41]. Specificity of PEPT3 and PEPT4 (encoded by SLC15A3/A4) does not seem to be known yet. Of note is also that we found transcription for SLC47A1/MATE1 and SLC47A2/MATE2 proteins in the choroid plexus which both increased during gestation, however, their function at this interface remains elusive.

4.2.5 Antioxidants

As has been suggested before [10] that antioxidant systems are likely to be important for choroid plexus epithelial cells since they are readily exposed to blood solutes through the rich vasculature of the plexus. In addition, the majority of reactive oxygen species are produced in mitochondria as a by-product of oxidative phosphorylation and the choroidal epithelial cells are abundant in mitochondria [1]. Although there was an increased level of expression of SOD genes at the latest fetal age, the data also shows that transcription is already high at mid-gestation. Antioxidant response elements (ARE) can be turned on in the event of oxidative stress and there are two factors that have emerged as master controllers for these responses [42]. PGC1α is a regulator of mitochondrial biogenesis genes that simultaneously upregulates many genes known to protect against oxidative stress and NRF2, which turns on a battery of antioxidant genes but also xenobiotic metabolising enzymes. Together, PGC1α and NRF2 activation drives glutathione synthesis, heme-oxygeneases, SOD production and are thus important for antioxidant activity as well as drug metabolising with downstream effects on ABC-proteins. Most of the genes related to these two systems were transcribed at high levels across all gestational ages and there was little developmental regulation. Overall, this suggests that the fetal choroid plexus encompass significant antioxidant activity mechanisms. We have previously shown that the NRF2 system is turned on in the choroid plexus in neonatal mice in response to hypoxia-ischemia [43].

4.2.6 CSF production

A main site for the production of CSF is thought to be the choroid plexus although extra choroidal sources are probably also present and the classical view of CSF circulation has repeatedly been challenged [44]. The continual production and reabsorption of CSF act like a dilution process, traditionally known as the CSF sink effect, for substances entering the CNS and is therefore an governing factor determining CSF and brain concentrations. A relevant question in relation to neuroprotection of the fetus is therefore whether there are changes in the CSF production by the choroid plexus during fetal development. The molecular mechanisms behind CSF production is still tentative but is thought to be driven by the flux of Na, K, Cl and HCO3⁻ across the plexus epithelial cells with water movement regulated by aquaporins [19]. We have listed in Table 6 the level of transcripts thought to be related to CSF production [19, 20]. Although most of these transcripts did not show regulation during gestation, of the twelve transcripts showing high or very high expression levels, half were up-regulated between GD90 and GD165 (AQP1, AT1A2, AT1B1/2, SLC12A2, KCNH2) with none down-regulated. If we believe that these twelve genes are the ones most pertinent for CSF production, then the data suggest that the CSF production is higher near-term for the nonhuman primate plexus cells. Of special note is the increase of both Na-K-Cl co-transporter SLC12A2 and AQP1. SLC12A2, situated on the CSF-facing membrane of the epithelial cells is thought a key transporter for the production of CSF at the plexus [45]. AQP1, situated also on the apical membrane on epithelial cells, has previously been suggested as an important contributor of water movement across the plexus during CSF secretion [46, 47], which is supported by AQP1 showing the highest expression of all genes outlined in our study. However, it is important to mention that to fully understand the effects of CSF production by plexus cells there is a need to take into account many other factors such as the growth of the plexus and changes in the ventricular system dynamics during development.

4.3. Limitations of the study

One limitation of this study is that we have not compared gene transcription changes in the fetus with adult animals. To better understand the stage of fetal development such a comparison would be of interest and should be included in future work. It is also important to note that our anatomical and molecular analyses of the localisation and levels of tight-junctional genes and proteins are only indicative of barrier function and functional tests of the blood-CSF barrier should be performed for confirmation.

CONCLUSIONS

This study suggests that proteins that make up the fundamental structures of the blood-CSF barrier are in their adult location already at mid-gestation in the nonhuman primate fetus and that there is little regulation of these proteins during the second half of gestation. This indicates that the basis for the blood-CSF barrier is well established in the mid-gestation fetus although functional studies are needed for confirmation. There is variable expression of genes encoding for xenobiotic metabolism. Phase I metabolism transcripts such as monoamine oxidases and epoxy hydrolases are found in significant amount, while CYPs are less abundant. In contrast, expression of genes associated with phase II metabolism appears robust and especially dependent on a rich variety of GSTs. However, there was little regulation of either phase I/II genes during development. On the other hand, some solute carriers considered being involved in drug transport showed increased expression with gestation. In particular, transcription of OATPs increased at late gestation and ought to be further examined. Transcription of antioxidants, including the inducing antioxidant systems, was high in the choroid plexus across all gestational ages suggesting that the fetal choroid plexus possess many mechanisms to deactivate reactive molecules. We have only explored the transcription of genes and further functional data is needed to better interpret the observed changes. This is particularly difficult for many of the transporters as the direction of transport is uncertain. The nonhuman primate data validates most of the previous results from rodents in relation to xenobiotic metabolisms in the choroid plexus although there are some distinct differences on the complement of transporters and CYP enzymes that should be taken into consideration when translating data to the human.

Supplementary Material

1. Suppl. Fig 1.

Heatmap generated from read counts of genes associated to Phase I metabolism for each animal included in the study. GD – Gestational Day.

2. Suppl. Fig 2.

Heatmap generated from read counts of genes associated to Phase II metabolism for each animal included in the study. GD – Gestational Day.

3. Suppl. Fig 3.

Heatmap generated from read counts of genes associated to ATP-binding cassette (ABC) proteins for each animal included in the study. GD – Gestational Day.

4. Suppl. Fig 4.

Heatmap generated from read counts of genes associated to solute carriers thought to interact with xenobiotics for each animal included in the study. GD – Gestational Day.

5. Suppl. Fig 5.

Heatmap generated from read counts of genes associated to antioxidants for each animal included in the study. GD – Gestational Day.

6. Suppl. Fig 6.

Heatmap generated from read counts of genes associated to the production of cerebrospinal fluid (CSF) for each animal included in the study. GD – Gestational Day.

Highlights 4620ETS.

• Study of the fetal nonhuman primate choroid plexus in relation to neuroprotection

• Combination of RNAseq and protein analysis

• Barrier mechanisms of choroid plexus are developed in fetus

• Phase I–III metabolisms and antioxidant genes are highly expressed in plexus

• Fetal choroid plexus can influence xenobiotics and nutrient supply into brain