Association between APOE4 and biomarkers in cerebral adrenoleukodystrophy

Cerebral adrenoleukodystrophy (cALD) is an inflammatory neurodegenerative disease associated with mutation of the ABCD1 gene. Proteomic analysis of cerebral spinal fluid (CSF) from young males with active cALD revealed markers of inflammation including APOE4. APOE4 genotype has been associated with an inferior prognosis following acute and chronic neurologic injury. We assessed APOE4 inheritance among 83 consecutive young males with cALD prior to hematopoietic cell transplant and its association with markers of cerebral disease. The allele frequency of APOE4 was not significantly different from that of the general population at 17%. Young males with cALD that were APOE4 carriers had similar CSF protein and chitotriosidase activity to that of non-carriers. In contrast, APOE4 carriers had an increased burden of cerebral disease involvement as determined by MRI severity score (10.5 vs 7.0 points, p = 0.01), higher gadolinium intensity score (2.0 vs 1.3 points, p = 0.007), inferior neurologic function (neurologic function score 2.4 vs 1.0, p = 0.001), and elevated CSF MMP2 levels compared to that of non-carriers (13168 vs 9472 pg/mL, p = 0.01). These are the first data showing that APOE4 is associated with increased severity of cerebral disease in cALD and suggest it may be a modifier of disease.


Results
Proteomic analysis of CSF samples. We have previously published several individual CSF biomarkers that are associated with disease burden in cALD [11][12][13][14] . In this study, we sought to undertake a comprehensive analysis of the CSF from young males with cALD using a proteomics-based approach. To investigate the proteomic composition of CSF in young males with cALD, we pooled the CSF from 18 young males with cALD prior to bone marrow transplant. A reference sample derived from non-ALD CSF was used for comparison as we have previously published 12,15 . The pooled samples were subjected to trypsin digestion followed by fragment labeling using iTRAQ isobaric tags, separation by liquid chromatography, and peptide identification by tandem mass spectrometry.
With a 1-peptide threshold and 90% confidence setting to identify proteins, we found 450 proteins in the CSF (5316 spectra); the false discovery rate (FDR) was 1%. The median amino acid length for the CSF proteome was 463 (range 45 to 4753) (Fig. 1A). The median molecular weight was 51.8 kDa (range 4.7 to 805 kDa) (Fig. 1B). The number of unique peptides assigned to each protein ranged from 2 to 97 (median of 4). Interrogation of Uniprot subcellular location ontogeny terms indicated that 53.1% of identified proteins were secreted, 24% cytoplasmic, 21% associated with the cell membrane, and 1.1% associated with the cell junction (Fig. 1C).
To understand the ontogeny of the proteins expressed at a higher level in cALD CSF, we performed Ingenuity Pathway Analyses (IPA) from the top quartile of proteins expressed in cALD ( Fig. 1D-F). Top quartile proteins in cALD CSF in the Diseases and Disorders category were found to be involved in "Inflammatory Response" and "Cardiovascular Disease" (47 and 35 proteins, respectively). The top Molecular and Cellular Functions were "Cell-to-cell Signaling" and "Cellular Movement" (35 and 31 proteins, respectively). For Physiological System Development and Function, the top ontogenies were the "Hematological System Development and Function" and "Humoral Immune Response" (42 and 15 proteins, respectively).
The protein sequence coverage from MSMS-identified proteins was at an average of 13.7% with a range of 0.2% to 85% (Fig. 1G). We noted that one of the proteins with high sequence coverage from our analysis was ApoE at 67% (Fig. 1H). ApoE exists as 3 isoforms: ApoE2, ApoE3, ApoE4. The single amino acid differences that separate these isoforms were not delineated in the mass spectrum analyses. The most well-studied of these isoforms is ApoE4, a significant risk factor for the development of early Alzheimer's disease and may play a direct role in blood-brain-barrier (BBB) breakdown 5,7,10,[16][17][18] . Therefore, we sought to determine if ApoE4 played a role in the manifestation of cerebral disease in cALD.

APOE4 increases endothelial cell oxidative stress and alters mitochondrial function.
The presence ApoE4 has been shown to be associated with increased oxidative stress in the brains of affected animals 19 . We evaluated the total ApoE concentration in the CSF of young males with cALD and found a mean of 17 µg/ mL, consistent within the range of previously published results [20][21][22] (Fig. 2A). ApoE4 has recently been shown to play a role in the maintenance of the endothelial BBB 7 , we next determined if endothelial cells exposed to ApoE4 had elevated levels of ROS. We exposed human endothelial cells (HDMECs) to ApoE4 for 72 hours and measured ROS production using H 2 DCFDA, a dye with fluorescent activity in an increased ROS state (Fig. 2B,C). We found a significant increase in ROS in endothelial cells exposed to ApoE4 compared to ApoE3 with elevated levels of ROS increasing with increasing ApoE concentration (Fig. 2C).
ApoE4 has been previously found to dysregulate energy metabolism and mitochondrial function 23 . We evaluated the effects of ApoE4 on the mitochondrial oxidative phosphorylation (OXPHOS) of human endothelial cells and found that the baseline oxygen consumption rate (OCR) was significantly reduced with the addition of ApoE4 to the cell culture media (30.6 versus 22.7 pmol/min, p = 0.03). Measurement of peak OCR and mitochondrial reserve were also significantly reduced in the presence of ApoE4 (43.5 versus 30.7 pmol/min, p = 0.02 and 13.5 and 7.6 pmol/min, p = 0.008, respectively) ( Fig. 3D-H). These data agree with prior reports indicating that ApoE4 can have detrimental effects on endothelial cell physiology and ApoE4 can directly affect ROS and mitochondrial metabolism in vitro 23 . Apoe genotype distribution in cALD. Given that cALD is manifest by blood brain barrier breakdown and cerebral inflammation, we next evaluated young males with cALD who carried the APOE4 allele to compare the extent of cerebral disease versus non-carriers of APOE4. We assessed 83 young males with cALD during pre-treatment evaluations. We found the distribution of APOE2, APOE3, and APOE4 alleles to be 0.060, 0.765, 0.175, respectively, which is very similar to that of the general population (Table 1). We observed that only three young males were APOE4 homozygotes. We performed a Chi-square test for Hardy-Weinberg equilibrium and found p = 0.51 indicating that our allele frequencies were in equilibrium. The mean age of APOE4 carriers at evaluation was 8.8 years (range 3.8-14.2 years), which was not statistically different from non-APOE4 carriers at 8.4 years (range 4.5-16.7 years, p = 0.59).

Correlation of Apoe genotype with clinical biomarkers of cALD. Earlier work demonstrated a role
for ROS and oxidative stress in ALD [24][25][26] . Exposure of fibroblasts from patients with ALD to excess levels of fatty acid gives rise to mitochondrial ROS production and reduced mitochondrial respiration as demonstrated by Lopez-Erauskin et al. 26 . Given that young males with cALD present with neuro-inflammation and BBB (i.e. endothelial) disruption, we sought to further investigate any modifying effect that inheriting APOE4 could have on patients with cALD. We evaluated CSF total protein concentration, which we previously have described as a biomarker of disease burden 12 , and found that carriers of APOE4 had a mean CSF total protein concentration of 40.5 mg/dL. This was not different from APOE4 non-carriers (mean 38.5 mg/dL, p = 0.70) (Fig. 3A). Monocyte/ macrophage-produced plasma chitotriosidase has been shown to be biomarker associated with Gaucher disease 27 , and we have previously shown that CSF and plasma chitotriosidase levels are associated with the amount of cerebral disease burden in cALD 14,27 . In the current study, we found CSF chitotriosidase activity in APOE4 carriers to www.nature.com/scientificreports www.nature.com/scientificreports/ be 4021 nmol/mL/hr which was higher than that of APOE4 non-carriers at 2347 nmol/mL/hr, but did not reach statistical significance (p = 0.056) (Fig. 3B).
MRI is the gold standard to determine the extent of cerebral disease in young males with cALD which can be quantified via a 35-point scale based on areas of white matter involvement, aka "Loes Score" 28 . We determined Loes scores of the young males in this cohort and found those carrying APOE4 had a significantly more MRI white matter involvement with a mean Loes score of 10.5 compared to APOE4 non-carriers with a mean Loes score of 7 (p = 0.01) suggesting that more demyelination was present in young males carrying APOE4 (Fig. 3C).
The presence of gadolinium enhancement defines "active" cALD, as it provides evidence of ongoing neuroinflammation. We have recently developed a semi-quantitative method of assessing gadolinium enhancement using a 0-3 gadolinium intensity score (GIS) 11 . A score of 0 indicates no gadolinium enhancement, while scores of 1, 2 or 3 correspond to increasing levels of gadolinium enhancement on MRI. We have shown the GIS to correlates to www.nature.com/scientificreports www.nature.com/scientificreports/ CSF chitotriosidase, Loes score, and neurologic outcomes after hematopoietic cell transplant (HCT) 11 . In the current study, young males with APOE4 had a significantly higher mean GIS (mean score 2.0) compared to APOE4 non-carriers (mean score 1.3, p = 0.007) (Fig. 3D).
Neurologic function is an important indicator of quality of life and ability to participate in everyday activities. We have used a 25-point scoring system developed by Moser and Raymond to assess the neurologic function where points are assessed in several domains when dysfunction is present and a neurologic function score (NFS) is derived 29 (Table 2). We have previously shown the NFS can be correlated to CSF cytokines, chitotriosidase, Loes score, GIS, and that a higher NFS predicts a more severe outcome after HCT 3,11-14 . As shown in Fig. 3E, APOE4 carriers had a significantly worse NFS (mean NFS 2.4) as compared to non-carriers (mean NFS 1.0, p = 0.001) indicating more neurologic dysfunction at the time of evaluation.  show results from testing of CSF pre-BMT. Scores shown in (C,D) were obtained from brain MRI within 45 days prior to BMT. Shown are means and standard error of the mean. P-values were generated from a Student's t-test. (G) shows a receiver operator characteristic plot from multivariate analysis of APOE4 that included the variables: Loes Score, Chitotriosidase, gadolinium intensity, NFS, and MMP2 concentration. www.nature.com/scientificreports www.nature.com/scientificreports/ non-carriers (not shown). A receiver operating characteristic curve (ROC) from nominal logistic fit analysis indicated reasonable association between the above biomarkers and APOE4 status (included all biomarkers except CSF Total Protein) (Fig. 3G).

Discussion
In this large cALD cohort, we report a potential genetic modifier of cALD disease related to the inheritance of the APOE4 allele. Frequency of APOE4 inheritance among patients with cALD (17.5%) was slightly higher than the 12% rate reported in the general population 6 , but this difference was not significantly different on Chi-square analysis. Our data indicate that young males with cALD carrying APOE4 had a higher cerebral disease burden at the time of evaluation for HCT. Our data show a 50% higher average Loes score in the APOE4 carrier group (i.e., 10.5 vs 7.0). Elevated Loes score in cALD has been shown to be associated with inferior outcomes after HCT 3,30 . Similarly, GIS was also increased in the presence of APOE4 among the young males with cALD. GIS has also been associated with a higher risk of poor neurologic function after HCT 11 . More intense gadolinium enhancement suggests a more permissive BBB in APOE4 carriers. Bell et al. have shown that a mechanism through which ApoE3 (the most common allele in the general population) may maintain the BBB is via the inhibition of cyclophilin A, an activator of MMP9. It was further demonstrated that carriage of ApoE4 allowed cyclophilin A to activate MMP9 and led to increased BBB leakiness. Unlike the prior work that showed MMP9 was the key mediator of BBB disruption, we found MMP2 is higher in the CSF of APOE4 carriers suggesting there may be a difference between murine and human biologic mechanisms. Furthermore, work recently reported by Paik et al. demonstrated attenuation of endothelial cell permeability with reduction in Aß induction of MMP-2 expression, suggesting that MMP-2 could be involved in BBB permeability, but it may vary depending on the pathophysiological process (in this case ß-amyloid toxicity). Interestingly, while CSF total protein may be a sign of increased BBB disruption, we did not find an association between APOE4 and CSF protein. In contrast, we did find an increase in CSF chitotriosidase activity, suggesting that inflammation and BBB disruption is higher in APOE4 carriers.
It is well described that HCT is the only established treatment modality to stabilize cALD once neuroinflammation has started 3,31-34 . Interestingly, one of the common immunosuppressive drugs given to cALD patients post-HCT to suppress graft versus host disease is cyclosporine, which binds intracellular cyclophilins in T-cells and inhibits their activity 35 . Cyclosporine can also bind and inhibit extracellular cyclophilins such as CypA 36 . Furthermore, Bell et al. has shown the APOE4 transgenic mice treated with cyclosporine had restored BBB integrity 7 . It is interesting to speculate that perhaps one of the mechanisms by which HCT stabilizes cerebral disease in young males with cALD is through the action of cyclosporine on the BBB apart from the allogeneic donor cell engraftment.
Our study was limited in that it was a cross-sectional analysis of young males with a new diagnosis of ALD, and all evaluated patients already manifested cerebral disease. Ideally, future studies should focus on longitudinal evaluation of young males with ALD who do not yet manifest cerebral disease to learn if APOE4 status truly modulates the onset of cerebral disease as well as learn if there is any change in the velocity of cerebral disease progression observed by serial MRIs and neurologic function.
There are numerous examples in the literature in which APOE4 is linked to poorer neurologic function following onset of acute or chronic neurologic injury [37][38][39] . In Alzheimer's disease, studies have shown a link

Neurologic Function Score Points
Hearing/auditory processing problems 1  www.nature.com/scientificreports www.nature.com/scientificreports/ between APOE4 inheritance and increased pericyte destruction leading to increased BBB permeability as contributors to the disease process 8 . Persons with neuropsychiatric disease have an increased prevalence of both anti-N-methyl-D-aspartate receptor (NMDA) antibodies and APOE4, demonstrating the a between BBB permeability allowing the penetrance of antibody producing immune cells with a neurologic illness 40,41 . Finally, APOE genotype can somewhat modify disease onset in metachromatic leukodystrophy, another leukodystrophy often beginning in childhood 42 .
Our data suggest that young males who inherit APOE4 have more advanced than those who do not carry APOE4. We speculate that they may have an accelerated course of disease, but without a true longitudinal study, this is unknown. As newborn screening for ALD becomes more prevalent, longitudinal studies looking at cALD disease modifiers will become increasingly important. Such studies will help inform us which patients may be at greatest risk of developing the cerebral form of the disease, perhaps allowing for increased surveillance with MRI and other biomarker testing. In-solution digestion. The 20-µg aliquot of each sample was diluted four fold with ultra-pure water. A pooled control sample of 125 µg was made from equivalent amounts of all the samples in the experiment (6.9 µg per sample) and diluted four-fold with ultra-pure water. Trypsin digestion was performed by adding 1:35 ratio of trypsin (Promega, Madison, WI) to total protein. Samples were incubated (16 hrs) at 37 C, frozen at −80 C (30 minutes), and dehydrated in vacuo. Each sample was then cleaned with a 4 mL Extract Clean ™ C18 SPE cartridge (Grace-Davidson, Deerfield, IL). Eluents were vacuum-dried and suspended in dissolution buffer (0.5 M triethylammonium bicarbonate) from the iTRAQ kit to a final 2 µg/µl concentration.

Methods
iTRAQ labeling and offline fractionation. 20 µg of a research sample or pooled reference was labeled for each iTRAQ channel in the set. A single iTRAQ assay consisted of iTRAQ isobaric tags 113-118 that were used to label CSF proteins from ALD patients 1-6. iTRAQ isobaric tags 119 and 121 were used to label a reference CSF protein extraction from children without adrenoleukodystrophy which we have previously published. The CSF protein from a total of 18 young males with cALD were analyzed in three iTRAQ experiments. The iTRAQ labeling and offline 1 st dimension fractionation was done as previously described followed by second dimension separation using liquid chromatography 43 .
Liquid chromatography and mass spectrometry. We analyzed the first dimension LC fractions by capillary LC-MS on an Orbitrap Velos MS system as previously described 44 . We made minor modifications to the LC and MS parameters: HCD activation time was 20 msec; lock mass was not employed; dynamic exclusion settings were: repeat count = 1, exclusion list size = 500, exclusion duration = 30 seconds, exclusion mass width (high and low) was 15 ppm and early expiration was disabled. www.nature.com/scientificreports www.nature.com/scientificreports/ Criteria for protein Identification. Scaffold (version Scaffold_4.7.3, Proteome Software Inc., Portland, OR) was used to validate MS/MS based peptide and protein identifications. Peptide identifications were accepted if they could be established at greater than 97.0% probability to achieve an FDR less than 1.0% by the Scaffold Local FDR algorithm. Protein identifications were accepted if they could be established at greater than 99.0% probability to achieve an FDR less than 1.0% and contained at least 2 identified peptides. Protein probabilities were assigned by the Protein Prophet algorithm 45,46 . APOE genotyping. HCT candidates' APOE genotyping was performed on prepared DNA samples by assessing APOE single nucleotide polymorphism (SNP) variants rs7412 and rs429358 using the SNP Genotyping Assay (Life Technologies, Grand Island, NY, USA) 44 .
Oxidative stress measurement. Human dermal microvascular endothelial cells (HDMECs) were plated in 24-well plate at 10,000 cells/well/mL media +/− 500 nM ApoE4, and incubated for 24 hours at 37C, 5% CO 2 . After incubation, wells were trypsinized with 200 µL of 1x trypsin and washed three times with PBS/0.2% BSA. Cells were next stained with 100 µL CM-H2DCFDA solution according to the manufacturer's protocol (Thermo Fisher) at room temperature for 5 minutes. Samples were washed three times with PBS/0.2% BSA. 5 µL of 7-AAD (Invitrogen) was added tube for live-dead discrimination. CM-H2DCFDA was read in the FL1 channel; 7AAD was read in the FL3 channel on a BD Accuri C6 analyzer.
Oxygen consumption rate. HDMECs were plated at 15,000 cells/well +/− 500 nM ApoE4. Plate was incubated at 5% CO 2 , 37 C for 72 hours prior to analysis. The plate was gently washed and each well received 500 µL Assay Medium (non-buffered DMEM without phenol red, 2 mM sodium pyruvate, 25 mM glucose, 1x GlutaMax, pH 7.43) that was pre-warmed to 37 C. The plate was incubated at 37 C for a minimum of 1 hour without CO 2 as is standard for Seahorse biochemical analysis experiments.
MMp determination. Samples were analyzed as a multiplex for MMP-1, 2, 9, and 10 using the Luminex platform performed as a multiplex. The magnetic bead set (cat # HMMP2MAG-55K-05, lot 2723323) was purchased from EMD Millipore Corporation (Billerica, MA). Samples were assayed according to manufacturer's instructions. Fluorescent color-coded beads coated with a specific capture antibody were added to each sample. After incubation, and washing, biotinylated detection antibody was added followed by phycoerythrin-conjugated streptavidin. The beads were read on a Luminex dual-laser fluidics based instrument (Bioplex 200). Samples were run in duplicate and values were interpolated from 5-parameter fitted standard curves.
Magnetic resonance imaging. Young males with cALD are defined by finding T2-weighted white matter signal abnormalities as well as gadolinium enhancement on MRI. The extent of cerebral involvement (i.e. T2 signal abnormality) can be quantified using a 35-point scoring system develop by Loes et al. 28 . Qualitative enhancement was assessed using the gadolinium intensity score (GIS) developed by Miller et al. 11 Loes and GIS scores were determined by a single neuroradiologist (D.R.N.).
Neurologic function score. Clinical cerebral disease severity was scored at the time of initial patient assessment in clinic by T.C.L., W.M., or G.V.R. using the ALD neurologic function score (NFS) 47 .

Data Availability
The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.