Oleic Acid-Containing Phosphatidylinositol Is a Blood Biomarker Candidate for SPG28

Hereditary spastic paraplegia is a genetic neurological disorder characterized by spasticity of the lower limbs, and spastic paraplegia type 28 is one of its subtypes. Spastic paraplegia type 28 is a hereditary neurogenerative disorder with an autosomal recessive inheritance caused by loss of function of DDHD1. DDHD1 encodes phospholipase A1, which catalyzes phospholipids to lysophospholipids such as phosphatidic acids and phosphatidylinositols to lysophosphatidic acids and lysophoshatidylinositols. Quantitative changes in these phospholipids can be key to the pathogenesis of SPG28, even at subclinical levels. By lipidome analysis using plasma from mice, we globally examined phospholipids to identify molecules showing significant quantitative changes in Ddhd1 knockout mice. We then examined reproducibility of the quantitative changes in human sera including SPG28 patients. We identified nine kinds of phosphatidylinositols that show significant increases in Ddhd1 knockout mice. Of these, four kinds of phosphatidylinositols replicated the highest level in the SPG28 patient serum. All four kinds of phosphatidylinositols contained oleic acid. This observation suggests that the amount of oleic acid-containing PI was affected by loss of function of DDHD1. Our results also propose the possibility of using oleic acid-containing PI as a blood biomarker for SPG28.


Introduction
Hereditary spastic paraplegia (HSP) is a genetic neurological disorder with progressive spasticity of the lower limbs and muscle weakness caused by axonal damage in the pyramidal tract [1][2][3][4][5]. The prevalence of HSP is estimated to be 1.2 to 9.6 per 100,000 [6]. HSP is genetically heterogeneous and is classified into 83 types, at least by 2022, with different responsible genes [7]. HSP is thought to be caused by the obstacles of mitochondrial metabolism, lipid metabolism, membrane transport, axonal transport and myelin formation, and its pathogenesis is extremely diverse [4,6]. Although diagnoses of specific types of HSPs are generally difficult due to the subtle or no differences in their symptoms among types, the molecular mechanisms of pathogenesis can be at least partially different depending on the causative genes. Due to the late onset of HSP [8], it would be useful if we could identify their types with biomarkers before HSP symptoms develop. In this study, we  [17,23,24].

Strain Phenotypes
Strain 1 (Baba, 2014 [17]) · Shortened mitochondrial sheath in sperm · Impaired sperm motility Strain 2 (Inloes, 2018 [24]) · Accumulation of PI 18:1/20 :4 · Decrease of LPI 20:4 Strain 3 (Morikawa, 2021 [23]) · Gait disturbance-like symptpms · Decrease of LPI 20:4 (sn-2) in cerebra · Axonal decrease in the pyramidal tract HSP is clinically as well as genetically heterogeneous. Neurological examination of patients often fails to distinguish their types without DNA testing [1]. Whole exome sequencing (WES) is commonly used as DNA testing to conduct molecular diagnoses of specific types of HSP [3]. Therefore, if biomarkers specific to particular types of HSP and applicable to patients' peripheral blood are available, they lead to cheaper and faster molecular diagnosis without DNA testing. Furthermore, biomarkers are potentially useful for examining pathogenic mechanisms, disease progression and drug efficacy.  [17,24]. In strain 3, Ddhd1 is disrupted by inducing 5 bp deletion to induce premature termination [23]. The map is not drawn to scale.
Biomarkers have been established for some monogenetic diseases, such as phenylketonuria (PKU) and alkaptonuria (AKU). These diseases are caused by defects in genes encoding enzymes PHA and HGD as in the case of DDHD1 in SPG28 [26,27]. L-phenylalanine and homogentistic acid are substrates of PHA and HGA, respectively, and have been established as biomarkers for these diseases [28,29]. In monogenic diseases caused by defects in enzyme genes, it is most straightforward to examine the substrates of the enzymes as candidate biomarkers. As quantitative changes of specific phospholipids in Ddhd1 knockout mice have been reported [23,24], lipid molecules are good candidates for the biomarker of SPG28. Since there are many species of phospholipids with different hydrocarbon groups, it is essential to distinguish molecular species with different side chains by high-resolution lipidome analyses. To address the need for SPG28 biomarkers, here we searched for lipid molecules whose abundances are altered by DDHD1 dysfunction.

Animals
We previously established Ddhd1 knockout mouse strain (Ddhd1 (−/−)) by introducing a 5-base-pair deletion in the second exon of the Ddhd1 gene [23]. The deletion results in a premature termination within the second exon of the DDHD1 gene, resulting in the expression of a protein completely lacking the DDHD domain. They were fed a standard pellet diet (CLEA Japan, Inc., Tokyo, Japan) and filtered water. The animals were kept under condition of a 12:12 h light:dark cycle and housed in groups of two to five animals per cage. The cages were changed once per week. Mouse husbandry and all mouse experiments were carried out at The Animal Center at Nakamura Gakuen University. All mice were bred in SPF (specific pathogen free) area. Although we did not examine sperm motility abnormalities in our Ddhd1 (−/−), we maintained the strain by mating Ddhd1 (+/−) females and Ddhd1 (+/−) males because they were considered to be infertile based on a previous report [17]. The six individuals in each genotype were chosen completely at random and used for blood collection. Animals used in the experiments were observed by visual inspection to be in normal health status. The background of all mice used in the experiments was C57BL/6J. The animal experiments were conducted with the approval of the Animal Ethics Committee of Nakamura Gakuen University (protocol code: 2016-1; date of approval: 4 April 2016). The study is reported in accordance with the recommendation of the Animal Research: Reporting of In Vivo Experiments (ARRIVE) guidelines.

Genotyping
Tails were cut with lengths of 2 mm at 2 weeks of age for genotyping. DNA was extracted by heating the tail tissues in 50 mM sodium hydroxide solution at 95 • C for 10 min. After heating, 1M Tris-HCl (pH 8.0) was added to neutralize, and the supernatant obtained by centrifugation at 16,000× g for 10 min was used as a template for PCR. The second exon of Ddhd1 was amplified using primers 5 ′ -AAGGTACATCTGGCTTGAAG-3 ′ and 5 ′ -GAGCTGTGGGTATAGTTGTG-3 ′ . Genotypes were determined by Sanger sequencing of PCR products.

Sample Preparation of Mouse Plasma
Blood samples were collected from the heart of 6-month-old Ddhd1 (+/+), Ddhd1 (+/−) and Ddhd1 (−/−) (n = 6 each). Blood collection was performed at 6 months of age in all mice. Both male and female mice were used in a 1:1 ratio to experimental groups. The mice were anesthetized by injecting 10 mL per kg of body weight of mixed anesthesia. Mixed anesthesia was composed by adjusting medetomidine hydrochloride (0.3 mg/kg) (Kyoritsu Seiyaku, Tokyo, Japan), midazolam (4 mg/kg) (Teva Takeda Pharma Ltd., Nagoya, Japan) and butorphanol tartrate (5 mg/kg) (Meiji Seika Pharma Co, Ltd., Tokyo, Japan) with saline (Otsuka Pharmaceutical Co., Ltd., Tokyo, Japan). After complete disappearance of the reflection, the chest was opened, and approximately 1 mL of blood was collected from the left ventricle. The syringe used for blood collection was moistened beforehand with sodium heparin (Mochida Pharmaceutical Co., Ltd., Tokyo, Japan). The blood was collected in a tube containing 1 µL of sodium heparin and centrifuged at 3000× g for 20 min at 4 • C. The supernatant was transferred to a new tube and stored as plasma at −80 • C until use.

Human Sample Preparation
Peripheral blood was collected from the SPG28 patient carrying the homozygous 4 bp deletion in the DDHD1 gene, originally described in Miura et al., 2016 [22], and from unrelated five control individuals free of HSP. The SPG28 patient was a male aged 61 at the time of blood collection. All control individuals were males aged 30, 47, 60, 75 and 76 at the time of the blood collection. Serum was isolated by centrifugation at 3000× g for 10 min after blood collection. Peripheral blood was taken following informed consent. The study was approved by the ethics committees of Kyushu University (protocol code: 2020-444; data of approval: 9 October 2020), Faculty of Medicine and Kurume University School of Medicine (protocol code: 21057; the data of approval: 16 June 2021). All methods were carried out in accordance with relevant guidelines and regulations, including the Declaration of Helsinki.

Lipidome Analysis for Mouse Plasma and Human Sera
The levels of PC, PI, LPC, LPI and LPA in mouse plasma and human sera were quantified using a supercritical fluid supercritical fluid chromatography (SFC) system (Shimadzu Co., Kyoto, Japan) coupled to a triple-quadrupole mass spectrometer (QqQMS) equipped with an electrospray ionization ion source (Shimadzu Co., Kyoto, Japan) as described previously [30]. The SFC system was equipped with a binary pump (i.e., CO 2 pump and a pump for the modifier and make-up solvent), autosampler, temperature-controlled column oven, and back pressure regulator (BPR). The SFC/QqQMS systems and data acquisition were controlled using Shimadzu LabSolutions ver. 5.99 SP2 software.
All samples were resuspended in 200 µL methanol/chloroform (1:1, v/v), and 2 µL was injected onto an ACQUITY UPC2™ Torus diethylamine (DEA) column (3.0 mm i.d. × 100 mm, 1.7 µm particle size, Waters Co., Milford, MA, USA). The temperature of the column oven and autosampler was set at 50 and 4 • C, respectively. BPR was set at 10 MPa. The mobile phase was composed of supercritical carbon dioxide (A) and methanol/water (95/5, v/v) with 0.1% (w/v) ammonium acetate (B). The mobile phase B was also used for modifier and make-up solvent. The flow rate of the mobile phase and make-up pump was set at 1.0 and 0.1 mL/min, respectively. The chromatographic separation was performed using a gradient of increasing the mobile phase B concentration as follows: 0-1 min: gradient was held at 1% B; 1-24 min; linear gradient of 1% to 75% B; 24-26 min; gradient was held at 75% B; 26-26.1 min; gradient was returned to 1% B; 26.1-30 min; the initial conditions were restored and the column was allowed to equilibrate for 3.9 min.
SFC/QqQMS analyses under multiple-reaction monitoring (MRM) were carried out in the positive-ion mode and in the negative-ion mode (polarity switching mode). The QqQMS parameters were set as follows: electrospray voltage of 4.0 kV in the positive-ion mode and −3.5 kV in the negative-ion mode; heating gas flow rate, 10 L/min; drying gas flow rate, 10 L/min; nebulizing gas flow rate, 3 L/min; heat block temperature, 400 • C; desolvation temperature, 250 • C; detector voltage, 2.3 kV; dwell time, 2 ms; pause time, 2 ms, and polarity switching time of 15 ms. The information of optimized MRM parameters for the lipid molecules is shown in Tables 2-5 and Tables S1-S10. Identification and quantification of the lipid molecules were performed using Multi-ChromatoAnalysT (Beforce Co., Fukuoka, Japan). The quantitative content of identified lipid molecules was calculated using the ratio of the peak area of each analyte to that of the internal standard of its representative lipid class. Table 2. PIs identified from mice plasma.

PIs
Retention Time (min)  The unit of measurement is pmol/mg. The q value was calculated by Student's t test followed by correction by the Benjamini-Hochberg method. * q < 0.05.
When the objective to examine which molecules among several molecular species changed significantly between Ddhd1 (+/+) and Ddhd1 (−/−), the analysis was performed by Student's t test followed by correction using Benjamini-Hochberg method.

PC, PA and PI in Mouse Plasma
Phosphatidylcholine (PC), PA and PI in mouse plasma from Ddhd1 (+/+) and Ddhd1 (−/−) were analyzed by lipidome analysis. PI and PA are substrates of DDHD1. PC is a precursor of PA, of which the conversion is catalyzed by phospholipase D [31,32]. PCs, PAs and PIs are expected to accumulate due to the loss of DDHD1 enzymatic activity in Ddhd1 (−/−) mice. A total of 88 different species of PCs and 64 different species of PIs were identified in mouse plasma (Tables S1 and S2). PA was not detected in any of the molecular species. Although there was no significant change in the total amount of PC in Ddhd1 (−/−) (Figure 2A)

Focused lipidome Analysis of Human Sera
We performed lipidome analysis on human sera by focusing on nine molecule species significantly changed in the mouse lipidome analyses. Since simple statistical analyses are not applicable due to the unavailability of biological replication of SPG28 patients, we ranked the levels of the seven lipid molecules in the six human samples, including the SPG28 patient. The total amount of PIs observed in the SPG28 patient was within the distribution of the ones in the controls ( Figure 4A). Of the nine PIs that showed sig-  Figure 4B). The amounts of all four PIs in SPG28 consistently exceeded the 95% CI calculated from the control samples (Tables 4 and S8). This indicates that these four PIs were elevated in the SPG28 sera. Notably, all four of these PIs contain oleic acid (18:1) in their side chains. LPA 24:0 was not detected in human sera. The SPG28 serum showed the lowest value for total LPAs, in contrast to the elevation of LPAs observed in mice ( Figure 4C and Table S9).

PIs and LPIs Containing Oleic Acid
Since the four PIs showing the highest values in SPG28 commonly contain oleic acid, we examined the amounts of PIs containing oleic acid in mouse plasma and human sera. Oleic acid-containing PIs were significantly elevated in Ddhd1 (−/−) mice (p = 7.0 × 10 −4 ) and showed the highest values in the SPG28 patient ( Figure 5). We examined the ranks of all oleic acid-containing PIs from the results of the lipidome analysis for human sera. Twelve oleic acid-containing PIs were detected in human sera. Nine of them showed the highest values in SPG28. Furthermore, the levels of these nine oleic acid-containing PIs exceeded the 95% CI of the control samples (Tables 5 and S10). This indicates the tendency of an increase in oleic acid-containing PIs in SPG28 sera.

Discussion
We observed a significant increase in the total amount of PI in Ddhd1 (−/−) mice that is consistent with the function of PLA1 converting PI into LPI. The increase in total PI is considered to be due to the accumulation of the substrate PI by the failure of the metabolism of PI to LPI due to the loss of DDHD1 function, which is consistent with the prediction based on the metabolic pathway of DDHD1. PI is one of the essential phospholipid components of the plasma membrane. PI is biased toward the inner side of the plasma membrane [33]. Since DDHD1 is a type of intracellular PLA1 (iPLA), the intracellular amount of PI can be greatly affected by the function of DDHD1 [13]. We observed consistent trends of increase in four PIs, 16:0/18:1, 18:1/18:1, 18:1/20:3 and PI 18:1/22:5, both in Ddhd1 (−/−) plasma from mice and in the human SPG28 serum [23]. Although differences in the content of several phospholipids have been reported between plasma and sera, we assumed the correspondence of quantitative changes in phospholipids between plasma and sera [34]. Notably, all of the PIs that have the highest levels in SPG28 contain oleic acid (18:1) in the side chain, an unsaturated fatty acid with 18 carbons and one double bond. Significant increases in PI 18:1/20:4 have also been observed in the brains of a different strain of Ddhd1 knockout mice previously published [24]. There was no significant increase in oleic acid-containing PI in Ddhd1 (+/−) mice, suggesting that sufficient enzymatic activity of DDHD1 remained ( Figure 4A). This is consistent with the mode of inheritance of SPG28 being autosomal recessive. Since the increased levels of oleic acid-containing PIs in the peripheral blood are reasonably attributable to the deficiency of DDHD1, our results indicate the possibility of oleic acid-containing PIs as a blood biomarker for SPG28.
LPA is a phospholipid that is abundant in plasma. LPA is also known to be a potential biomarker for ovarian cancer since LPA in plasma is shown to be specifically increased in ovarian cancer patients [35][36][37]. Although LPA is the product of a reaction catalyzed by DDHD1, the total amount of LPA was increased in Ddhd1 (−/−) mice. Among the eight species of LPAs identified, only LPA 24:0 showed a significant increase, suggesting that the increase in total LPAs is mainly attributable to LPA 24:0. LPA is yielded not only by the PLA1-mediated pathway but also by other pathways involving other enzymes such as lysophospholipase D (LysoPLD) catalyzing lysophoshatidylcholine, lysophosphatidyllethanolamine (LPE) or lysophosphatidylserine (LPS) into LPA [38]. LysoPLD is a secreted enzyme that is a major enzyme-producing PLA in plasma [39]. The excess of LPAs observed in our current analyses is, therefore, attributable to the compensatory pathway such as the one mediated by LysoPLD. Increased activity of LysoPLD and accumulation of LPA are known to occur in female-specific cancers, hepatitis C virus and inflammatory diseases [40,41]. The activity change in LysoPLD is thought to be involved in these diseases in multiple aspects. Increased activity of LysoPLD has been shown to be triggered primarily by inflammatory signals [42]. We therefore speculate that the loss of motoneurons in the pyramidal tract in Ddhd1 knockout mice can act as inflammatory signals, triggering an increase in LysoPLD activity [23]. Since no significant decrease in LPC 24:0 was observed in the Ddhd1 (−/−) mice ( Figure S1), the observed increase in LPA 24:0 may be supplied by the metabolism of LPE or LPS, which are out of the scope of our focused lipidome analyses. LPA was at the lowest level in the SPG28 patient in contrast to the observation in mice. This is partly attributable to the differences in biological age between the human and mice at examination. The age of 6 months in the Ddhd1 (−/−) mice is prior to the onset of HSP-like phenotypes and is likely in the middle of the inflammatory process, which may trigger LysoPLD activation without neurological phenotypes. In contrast, since the human sera were collected from the SPG28 patient at the age of 61 after the progression of HSP symptoms, it is possible that the motoneuron degradation prior to the HSP symptoms had reached a plateau and the inflammatory signals might have been weakened.
The significant reduction in LPI 20:4 (sn-2) found in Ddhd1 (−/−) cerebra was not observed in Ddhd1 (−/−) plasma from mice (Table S6) [23]. This discrepancy can be partly attributed to the difference in tissues used for lipid extraction. Plasma is cell-free. DDHD1 is one of the iPLAs and is mainly involved in the intracellular metabolism of phospholipids [43]. Therefore, it is likely that the lipidome analysis using plasma, which is cell-free, does not directly reflect the effects of intracellular DDHD1 metabolism. Since DDHD1 seems to be functionally defective in blood cells, a decrease in LPI 20:4 (sn-2) as in the cerebra is likely to occur.
SPG39, SPG54 and SPG56 are also caused by defects in enzymes involved in PI metabolism, PNPLA6, DDHD2 and CYP2U1, respectively [10]. Notably, DDHD2 is responsible for SPG54 functions such as in iPLA1 as well as in DDHD1 [44,45]. Since abnormal PI metabolism is likely to be the common pathological mechanism for these types of HSPs, peripheral PI levels are good candidates for biomarkers of other HSPs sharing pathogenesis with SPG28.
The major problem with the current study is that statistical analyses were not applicable to human samples due to the unavailability of additional SPG28 samples, since the disease is extremely rare. Although further studies of specificity and sensitivity of oleic acid-containing PIs using additional SPG28 patients are required for establishment as a biomarker, oleic acid-containing PIs may still be useful to group SPGs sharing similar pathogenic pathways and may still help in clarifying their pathogenic mechanisms, potentially contributing to the treatment. Identification of the blood biomarker is useful not only for diagnosis of HSP types in clinical practice but also as a platform for therapeutic drug development. For example, we should be able to evaluate the effect of a drug by measuring oleic acid-containing PI after administering an SPG28 therapeutic agent to patients. To achieve these goals, it is essential to establish standard values for the amount of oleic acid-containing PI. Measurements of oleic acid-containing PI in larger numbers of SPG28 patients and establishment of the standard values are important tasks for this goal.

Conclusions
Lipidome analysis of SPG28 model mice and a SPG28 patient was performed to search for a biomarker in blood that is useful for the diagnosis of SPG28. Lipidome analysis of mouse plasma revealed a significant increase in total PI and LPA in Ddhd1 (−/−) mice. We examined the changes in each PI and identified nine PIs that were significantly elevated in Ddhd1 (−/−) mice. Next, we examined the level of these nine PIs and total PA in human serum. Seven of the nine PIs found to be elevated in Ddhd1 (−/−) mice were also detected in human sera. Four of these has the highest level in the SPG28 patient, and all of them were oleic acid-containing PIs (PI 18:1/18:1, PI 18:1/20:3, PI 18:1/20:4 and PI 18:1/22:5). LPA showed lower levels in the SPG28 patient, different from the results observed in Ddhd1 (−/−) mice. From our observation, we propose that oleic acid-containing PI is a promising blood biomarker for SPG28.

Institutional Review Board Statement:
The study was conducted in accordance with the Declaration of Helsinki and was approved by the Ethics Committee of Kyushu University (protocol code: 2020-444; date of approval: 9 October 2020) and Kurume University (protocol code: 21057; date of approval: 16 June 2021). The animal study protocol was approved by the Institutional Review Board (or Ethics Committee) of Nakamura Gakuen University (protocol code 2016-1; date of approval: 4 April 2016).

Informed Consent Statement:
Informed consent was obtained from all subjects involved in the study.

Data Availability Statement:
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request, while all the analyzed data are included within this article and its supplementary information files.

Conflicts of Interest:
The authors declare that they have no competing interests.