Phospholipid Profiles Are Selectively Altered in the Putamen and White Frontal Cortex of Huntington’s Disease

Huntington’s disease (HD) is a genetic, neurodegenerative illness that onsets in late adulthood as a series of progressive and terminal cognitive, motor, and psychiatric deficits. The disease is caused by a polyQ mutation in the Huntingtin gene (HTT), producing a polyglutamine expansion in the Huntingtin protein (HTT). HTT interacts with phospholipids in vitro; however, its interactions are changed when the protein is mutated in HD. Emerging evidence suggests that the susceptibility of brain regions to pathological stimuli is influenced by lipid composition. This study aimed to identify where and how phospholipids are changed in human HD brain tissue. Phospholipids were extracted using a modified MTBE method from the post-mortem brain of 13 advanced-stage HD patients and 13 age- and sex-matched controls. Targeted precursor ion scanning mass spectrometry was used to detect phospholipid species. In the white cortex of HD patients, there was a significantly lower abundance of phosphatidylcholine (PC) and phosphatidylserine (PS), but no difference in phosphatidylethanolamine (PE). In HD putamen, ester-linked 22:6 was lower in all phospholipid classes promoting a decrease in the relative abundance of ester polyunsaturated fatty acids in PE. No differences in phospholipid composition were identified in the caudate, grey cortex or cerebellum. Ether-linked PE fatty acids appear protected in the HD brain, as no changes were identified. The nature of phospholipid alterations in the HD brain is dependent on the lipid (subclass, species, and bond type) and the location.


Introduction
Huntington's disease (HD) is an autosomal, dominant, neurodegenerative illness resulting from a CAG repeat mutation on exon-1 of the Huntingtin gene (HTT). This mutation causes a polyglutamine expansion at the N-terminus of the huntingtin protein [1] (HTT), referred to as mutant huntingtin (mHTT). Although ubiquitously expressed in the human body, the presence of mHTT coincides with the targeted degeneration of several brain regions, notably the striatum and cerebral cortex. This degeneration corresponds with the progressive and terminal cognitive, psychiatric, and motor symptoms which onset in late adulthood in HD [2]. Because there is limited knowledge concerning how the mHTT causes the selective degeneration of these brain regions, no effective treatment options are cortex) and those which are mildly degenerated (cerebellum). The use of these brain regions was to aid in understanding if phospholipid disturbance is region-specific in advanced HD. It appears that in clinically advanced HD cases, phospholipid disturbance is specific to brain regions.

Human Brain Tissue
The Victorian Brain Bank supplied human post-mortem brain tissue from 13 advancedstage HD subjects and 13 age-and sex-matched controls. Subject demographics have been published previously [24]. The Victorian Clinical Genetics Service determined the CAG repeat length of HD subjects' disease causing HTT gene. Tissue was taken from the brain's left hemisphere from five regions of interest: caudate, putamen, cerebellum, and the grey and white matter of the dorsomedial prefrontal cortex. The mean age, post-mortem interval, and brain pH between HD and controls were not different as determined by Kreilaus et al., 2016 [25]. All HD subjects had a Vonsattel pathological grading of IV [26] and were of advanced clinical stage. Brain tissue was stored at −80 • C until use. Ethics approval was granted by the UOW Human Research Ethics Committee (10/327), and this research was carried out in accordance with the Declaration of Helsinki (2008).

Mass Spectrometry
Nanoelectrospray ionization mass spectrometry of lipid extracts was performed using a hybrid triple quadrupole linear ion trap mass spectrometer (Q-Trap 5500, Sciex, Vaughan, ON, Canada), equipped with an automated chip-based nanoelectrospray source (TriVersa Nanomate, Advion Biosciences, Ithaca, NY, USA) as described previously [29]. Samples were loaded into a 96-well plate and sealed before direct infusion. Spray parameters were set at a gas pressure of 0.4 psi and a voltage of 1.2 kV for positive and 1.1 kV for the negative ion mode. Lipid data were acquired using targeted precursor ion scans, as shown in Supplementary Tables S1 and S2. Target lists for each molecular species were generated following a manual review of spectra in Analyst (v1.6; Sciex, Framingham, MA, USA). The mass spectrometry output was then analyzed using LipidView (v1.2; Sciex, Framingham, MA, USA) and quantified by comparing peak areas to class-specific internal standards. Processing settings were set at a mass tolerance of 0.5 kDa and a minimum signal to noise of 20. Smoothing and de-isotoping were enabled.

Data Processing
LipidView output was exported to Microsoft Excel. Positive ion data for phospholipid head groups were used to quantify lipids by matching the detected rations of isobaric species from the paired negative ion data for the fatty acids. Quantification of ether-linked phospholipid species was corrected using a 3.45 isotope correction factor as described previously [28]. Any lipids detected in 'blank' samples were subtracted from patient samples. Lipid species that were not detected in at least 60% of samples were excluded from the analysis.

Statistical Analysis
Outliers were identified using the 2.2 interquartile range of each phospholipid class total. Lipids were assessed individually for normality using the D'Agostino-Pearson Omnibus test and then analyzed using either a two-tailed unpaired t-test with Welch's correction or the Mann-Whitney U test where appropriate. Correlation analyses were conducted using Pearson's correlation. A two-way ANOVA was used to assess differences in lipid class totals between regions in control and HD subjects. Data were adjusted for a False Discovery Rate of 1% using the two-stage Benjamini, Krieger, and Yekutieli method for multiple comparisons. Statistical test information is provided for each lipid species in Supplementary Tables S3-S68. Principal component analyses were used to compare phospholipid composition between regions in control and HD subjects. Volcano plots were created manually to visualize patterns of lipid changes between HD and controls. Data are expressed as the mean ± the standard error of the mean in nmol lipid per mg brain tissue.
Processed lipid values are available in Supplementary Excel File.

Principal Component Analyses of Control and HD Brain Regions
Principal component analyses were used to determine the phospholipid classes which contributed to the separation of brain regions according to their phospholipid chemistry in control and HD subjects ( Figure 1). Variances were supported using a two-way ANOVA. The controls served as a baseline for the regional variation in phospholipid content. In controls, principal component 1 (PC1) contributed to 56% of the regional variances, while component 2 (PC2) contributed 29%. The white matter of the cortex was separated from grey matter regions by significantly higher concentrations of ether PC (+60-80%), ether PE (+60-70%) and PS (+40-70%) ( Figure 1A,B; PC1). The putamen was separated from the other grey matter regions by significantly higher PC (+50% caudate; +63% cerebellum), LPC (+8% caudate; 84% cerebellum) and PE (+35% caudate; +53% cerebellum).
In comparison, the white cortex and putamen were not as clearly separated from the other regions in HD as compared to controls. HD subjects had more significant intersubject variability in phospholipid content than controls, and therefore regions were not as well defined. In HD, PC1 contributed to 46% of the variance between regions and PC2, contributed 20%. In HD, the white matter was still distinguishable from the other grey matter regions, although not as robustly, by significantly higher concentrations of ether PC (+40-60%), ether PE (+40-60%) and PS (+40-70%) ( Figure 1C,D). In HD, the putamen was separated from the other grey matter regions by higher PC (+50-60%) and PE (+30-50%), which aligned with the control separations previously mentioned. Phospholipid class totals are provided in Table 1.    Data are presented as the mean ± SEM (n) in pmol lipid per mg tissue. Data were assessed using a two-way ANOVA. Multiple comparisons were adjusted using a False Discovery Rate of 1% (Benjamini, Krieger and Yekutieli method). The percentage difference of HD compared to controls is provided. *** p < 0.001. ns , not significant when adjusted for multiple comparisons. Abbreviations: CON, control; HD, Huntington's disease; LPC, lysophosphatidylcholine; LPE, lysophophatidylethanolamine; PC, phosphatidylcholine; PD, percentage difference; PE, phosphatidylethanolamine; PS, phosphatidylserine; SEM, standard error of the mean.     ure 2B). Statistical test information for PC, ether PC and LPC can be fou tary Tables S3-S6 and S15 (caudate), Supplementary Tables S16-S19 an Supplementary Tables S30-S33 and S42 (white cortex), Supplementary T S55 (grey cortex) and Supplementary Tables S56-S59 and S68 (cerebellu

Phosphatidylethanolamines
Ether PE was not different in any HD brain region when compared to controls. PE was lower in the putamen of HD subjects compared to controls (p < 0.0001). Lower PE in HD putamen was due to significant alterations in ester PE species. These shifts also caused a shifted dominance of long-chain over very long fatty acyl chain species. PE species  Figure 3G). In addition, multiple LPC species were lower in the putamen of HD patients. The most significant differences in LPC species identified in the putamen were in LPC 18:0 (−49%, p < 0.0001) and LPC 22:6 (−59%, p = 0.0008) ( Figure 5E). The remaining species were LPC 16:0 (−24%, p = 0.0012), LPC 18:1 (−23%, p = 0.0022) and LPC 20:4 (−40%, p = 0.0010). No differences in PC, ether PC nor LPC were found between HD and control in the caudate ( Figure 2D), or cerebellum ( Figure 2E). Total PC was lower in the grey cortex of HD subjects (p < 0.0001); however, no changes in ether PC or LPC were found ( Figure 2B). Statistical test information for PC, ether PC and LPC can be found in Supplementary Tables S3-S6 and S15 (caudate), Supplementary Tables S16-S19 and S29 (putamen), Supplementary Tables S30-S33 and S42 (white cortex), Supplementary Tables S43-46 and S55 (grey cortex) and Supplementary Tables S56-S59 and S68 (cerebellum).
Only one ether PE species was found to be different in the HD brain. PE O-18:1_22:6 was 35% lower in HD putamen compared to controls (p = 0.0007). PE species were not different in the caudate ( Figure 6D), cerebellum ( Figure 6E), or white or grey cortex ( Figure 6B) of HD patients. Statistical test information for PE, ether PE and LPE can be found in Supplementary Tables S7-S10 and S15 (caudate), Supplementary Tables S20-S23 and
No changes in PS species or-derived fatty acids were identified in the caudate ( Figures 3L and 7D), cerebellum (Figures 3O and 7E) and grey cortex (Figures 3F and  7B) between control and HD patients. Statistical test information for PS can be found in Supplementary Tables S11 and S12 (caudate), Supplementary Tables S24 and S25 (putamen),  Supplementary Tables S38 and S39 (white cortex), Supplementary Tables S51 and S52 (grey cortex) and Supplementary Tables S64 and S65 (cerebellum).

CAG Repeat Length Is Not Related to Neural Phospholipid Abundances in HD Patients
Pearson's correlation analysis was run to determine if a relationship existed between the CAG repeat length of HD patients and the concentration of phospholipids in each brain region. The p values were adjusted for multiple comparisons using the two-stage Benjamini, Krieger and Yekutieli method. No significant correlations were identified between CAG repeat length and any phospholipid class, species, or fatty acyl chains.

Discussion
The putamen and the white matter of the dorsomedial prefrontal cortex have a regionspecific vulnerability to phospholipid disturbance in cases of clinically advanced HD. The vulnerability is supported by the significant changes to the phospholipid profiles of these regions and the absence of change in the caudate, cerebellum, and grey matter of the dorsomedial prefrontal cortex. Previous reports on lipid metabolism in the cerebellum of HD post-mortem tissue have found little to no changes in this brain region [25,30]. However, the caudate is a region where multiple lipid classes are altered in HD [25,30,31]. The selective reduction in the PC and PS content of the white matter in the cortex and ester-linked 22:6 in the putamen suggests that phospholipid disturbance is both region and lipid specific in advanced HD. The absence of change in ether PE in the HD brain was perplexing, considering the immense importance of these lipids to neural function and the severe degeneration of the brain in HD. Whilst the extraction and mass spectrometric analysis used in this study cannot isolate the cellular location of the identified lipid changes (i.e., neural bodies, glia, synapses), phospholipids are most abundant in the cell membrane contributing to fluidity and function. Thus, this study revealed a critical feature of HD; not all brain regions experience the same disturbance in phospholipid metabolism. The nature of the changes to phospholipids between the caudate and putamen suggest that they are specific not only to whole regions, but also striatal subregions. In controls, the caudate and putamen were separated by PC and PE composition, as shown in the principal component analysis (Figure 1A,B). Considering the caudate and putamen share the same dominant cell type (~95% medium spiny neurons), their distinct phospholipid profiles were interesting. The putamen had almost twice the PC content compared with the caudate, and PC was the phospholipid which was reduced by 28% in the putamen of HD patients. The PE content of the putamen in controls was also significantly higher than the caudate, and again only the putamen was affected by changes in PE composition in HD. The difference in the phospholipid chemistry between the caudate and putamen in controls may simply be the result of changes in the numbers of neural cells (i.e., astrocytes, microglia), an infiltration of the biochemistry of neighboring brain regions (i.e., internal capsule), or it may reflect fundamental biochemical differences between striatal subregions. The difference in phospholipid disturbance between the caudate and putamen in HD may be explained by the differing rates and onset of degeneration, as well as their differing relationships to clinical indices in HD [32,33]. However, the previous investigation into cholesterol metabolism in HD has also indicated a difference in the vulnerability to lipid disturbance between the caudate and putamen [25,31]. The selective disturbance of PC and PE in the putamen of HD patients may reflect a difference in the molecular consequences or response to pathological triggers caused by HD compared to the caudate. The shift in the relative abundance of PE species towards an increased dominance of monounsaturated fatty acyl species in the putamen suggests alterations in cell membrane fluidity and permeability in HD, affecting cell function. The abundance of ester-linked 22:6 (also known as docosahexaenoic acid; DHA)-derived from PC, LPC, PE, and PS was lower in the putamen of HD patients. These reductions were specific to the putamen; no reductions in polyunsaturated fatty acid chains were identified in the caudate, which houses the same neuronal population as the putamen (medium spiny neurons). Mammals cannot synthesize 22:6, so the brain contributes to the abundance of 22:6 via the elongation and desaturation of dietary α-linolenic acid [34], the final step of which occurs exclusively in astrocytes [35]. Astrocytes can then mediate the release of 22:6 for incorporation into phospholipids [36] and inflammatory mediators maresins, resolvins and protectins [37,38]. Severe astrocytosis (an increase in the number of astrocytes) in the caudate and putamen indicates pathological grade IV classification of HD post-mortem tissue used in this study [26]. Neural inflammation is well documented in HD [37,39,40]. An increase in astrocytes would suggest an increase in the available 22:6, unless it is siphoned to inflammatory mediators, restricting the availability to phospholipids. Alternative explanations include a release of 22:6 by phospholipids for inflammatory mediators or changes in the transport of 22:6 into the brain by LPC. LPC is the favoured carrier of dietary 22:6 across the blood-brain barrier [41], and the reduction in LPC and PC would support this.
Although the exact sn-position of the fatty acyl chains was not determined, polyunsaturated chains (including 22:6) typically attach to phospholipids at the sn-2 position of the glycerol backbone. The release of fatty acyl chains at the sn-2 position is facilitated by phospholipase A 2 (PLA 2 ) [42]. PC, PE, and PS share synthesis and degradation pathways, making it difficult to discern if the reduction of 22:6 in one phospholipid class then had downstream effects on another. Fatty acyl chains are rapidly hydrolyzed and re-esterified to phospholipids allowing them to 'shuffle' between classes as needed by the cell [18]. Compared with other phospholipid classes, PS is enriched in 22:6 and thus acts as a storage facility for the fatty acid [43]. PS 18:0_22:6 is the most abundant PS species in the brain [36], and this species was the driving force of the reductions in PS-derived 22:6 in HD putamen. In neural cell cultures, 22:6 supplementation assists neurite growth and increases the number of dendritic spines, promoting more quality connections [36]. In addition, the esterification of 22:6 to PS allows the species to incur anti-apoptotic effects in vitro. These effects are thought to occur because of the ability of 22:6 PS species' ability to promote brain-derived neurotrophic factor induced translocation of Raf-1 kinase in the cell membrane, which is crucial to cell survival [36,43]. Other fatty acids, including 18:1 and 22:5, cannot rescue this feature in cells when 22:6 is absent. Therefore, the loss of 22:6 from PS species in the putamen of people with HD could have profound consequences for cell survival.
The white matter of HD patients was distinguishable from the other brain regions, by lower abundances of PC and PS, in the dorsomedial prefrontal cortex. If these lower abundances were a direct result of neuronal degeneration or atrophy, it is expected that PE (including ether) would be decreased by a similar fold. The principal component analysis of controls ( Figure 1A,B) indicated that white matter was distinguishable from the grey matter by ether PC, ether PE and PS content. The separation of matter type is likely due to the strong associations of these lipids with myelin, the dominant component of white matter. PC, PE, and PS represent 8.3%, 11.2% and 5.3% of the total lipid dry weight, respectively, in the aged brain [44]. Of PE, ether species are reported to contribute up to 70% [45]. Magnetic resonance imaging studies report decreases in the volume and structural abnormalities in major white matter tracts early in HD; however, there is little evidence of the molecular changes occurring [46][47][48]. Myelin breakdown in the white matter tracts surrounding the basal ganglia and those connecting it to the frontal cortex have been reported [47]. The high abundance of PE and ether PE in myelin and the absence of change in these lipids in the white matter of HD patients suggests that the selective reductions in PC and PS are either non-myelin related or selective targeting of these lipids within the myelin membrane.
Despite changes to multiple phospholipids in the putamen and cortical white matter of HD patients, ether PE species and ether-linked PE fatty acyl chains appear to be protected across multiple regions of HD brain. Albeit for one ether PE species in the putamen, these lipids and their ether-linked fatty acyl chains were unaltered in all five brain regions in HD. The reason for this is unclear, considering their high abundance in the brain and the enormous contribution that they have to neural function. Early reports on HD post-mortem brain tissue found no differences in the relative abundance of PE to PE plasmalogens in the caudate. However, these reports could not ascertain exact quantities [49]. The ether PE detected in this study include species with both alkyl ether bonds and vinyl ether bonds. The difference between these bonds is not easily detectable in phospholipids and the mass spectrometric techniques used could not distinguish them. Ether PE species with vinyl ether bonds are plasmalogens. There is evidence that PE plasmalogens are protective against oxidative damage. This protection is due to the ease of oxidation of their vinyl ether bond at the sn-1 position compared to the ester bond at the sn-1 position of diacyl species [50]. The evidence suggests that once vinyl ether bonds are oxidized, the propagation of reactive oxygen species, which typically occurs in ester-linked species, is stopped, effectively halting the oxidative event cascade [51]. Furthermore, incorporating highly oxidizable polyunsaturated fatty acids into ether PE is suggested to protect the fatty acids against oxidation [51].
It is established that ester-linked fatty acyl chains are cleaved from phospholipids by either phospholipase A 1 for those at the sn-1 position or PLA 2 at the sn-2 position. However, the ester-linked fatty acyl chains at the sn-2 position of ether PE are released via the selective phospholipase PlsEtn-PLA 2 [52]. The ether-linked fatty acyl chains at the sn-1 position of ether PE are cleaved via oxidation. The decrease in PE species' polyunsaturated fatty acid content in HD putamen was caused by a reduced abundance of PE species, not ether PE species. For example, the reduction in 22:6 from ester PE species was 42% (p = 0.001), whilst the reduction from ether PE species was 12% (p = 0.099). Therefore, reductions in the highly oxidative suspectable polyunsaturated fatty acids would likely be due to PLA 2 , releasing polyunsaturated fatty acyl chains from PE and not PlsEtn-PLA 2 from ether PE. PLA 2 is coupled to dopaminergic [53], serotonergic [54], and N-methyl D -aspartate [55] neuroreceptors, all affected in HD [56][57][58].

Conclusions
Changes to phospholipids are isolated to specific brain regions in advanced HD. In this study, those changes were in the white matter of the dorsomedial prefrontal cortex and the putamen. The absence of change to phospholipids in the caudate indicates that cell type may not dictate a region's susceptibility to phospholipid disturbance in HD; it may be that location in the brain or the cellular response of a specific region to pathological stimuli. It appears that ether bonds in PE may protect attached fatty acyl chains in HD, supporting evidence from in vitro studies that the molecular structure of phospholipids can influence its susceptibility to interference by mutant huntingtin. The selective changes in phospholipid classes in the white matter of the dorsomedial prefrontal cortex and the putamen prompt vital considerations for using lipid therapeutics to treat HD, as each brain region may need unique approaches.