Phosphoinositide deficiency due to inositol depletion is not a mechanism of lithium action in brain

Molecular Genetics and Metabolism 82 (2004) 87–92 www.elsevier.com/locate/ymgme

Phosphoinositide deWciency due to inositol depletion is not a mechanism of lithium action in brain Gerard T. Berry,a,¤ Roberto Buccafusca,a John J. Greer,b and Eric Ecclestona a Children’s Research Institute, Children’s National Medical Center, Department of Pediatrics, George Washington University School of Medicine and Health Sciences, Washington, D.C., USA b Department of Physiology, Division of Neuroscience, University of Alberta, Edmonton, Alberta, Canada

Received 2 February 2004; accepted 4 February 2004

Abstract The “inositol depletion hypothesis” has been widely held to be the explanation for both the eVect of lithium on brain function, apropos of its use in mood disorders, and on the impairment of development and induction of embryonic malformations in diverse organisms. The essence of the hypothesis is that a deWciency in cellular myo-inositol (Ins), secondary to lithium inhibition of inositol monophosphatase and/or multiple inositol polyphosphate phosphatase activities with trapping of Ins as inositol phosphates, leads to a depression of phosphatidylinositol (PtdIns) and a secondary impairment in inositide signaling. However, the ability of relatively low micromolar levels of Ins to reduce mammalian PtdIns synthetase activity in vivo has never been adequately tested. We have generated a lethal murine brain Ins deWciency model and measured PtdIns content using a novel MALDI-TOF MS method. Our results show that in the most severe Ins deWciency ever recorded in a mammal, the brain PtdIns levels do not decrease. We conclude that PtdIns deWciency due to “inositol depletion” is not a mechanism of lithium action in brain, and that Ins plays another unidentiWed role in the mammalian brain.  2004 Elsevier Inc. All rights reserved.

Introduction A depression in neuronal myo-inositol (Ins) concentration leading to deWcient phosphoinositides has been hypothesized to be the biochemical basis for the eVect of lithium treatment on brain function [1–3]. It is also considered to be the etiology of malformations following exposure of a developing organism to lithium [4], as concomitant treatment with Ins can rescue the phenotype [5]. However, there have been no accurate mass measurements of the only inositolphospholipid, phosphatidylinositol (PtdIns), that is directly synthesized from Ins to support this hypothesis [6]. In order to gain more insight into the role of Ins in brain metabolism and, speciWcally, to determine whether millimolar concentrations of Ins ¤ Corresponding author. Present address: JeVerson Medical College, OYce of Vice Dean for Research, Suite 102, College Building, 1025 Walnut Street, Philadelphia, PA 19107-5083, USA. Fax: 1-215955-2868. E-mail address: [email protected] (G.T. Berry).

1096-7192/$ - see front matter  2004 Elsevier Inc. All rights reserved. doi:10.1016/j.ymgme.2004.02.002

are necessary for PtdIns production, we generated a murine brain Ins deWciency model [7], and measured PtdIns content in cellular membranes using a novel matrix-assisted laser desorption and ionization (MALDI) time of Xight (TOF) mass spectrometry (MS) method. Our results show for the Wrst time that the most severe deWciency of Ins ever recorded in mammalian brain does not cause PtdIns to decrease, yet, is lethal for the newborn. This work has major implications for lithium action. We conclude that Ins plays another unidentiWed role in the mammalian brain, that is distinct from its role as an osmolyte and the precursor of PtdIns. Properly stated, the “inositol depletion hypothesis” to explain lithium action is as follows: cellular Ins deWciency leads to decreased PtdIns which in turn leads to reduced polyphosphoinositides that are important in cell signaling. Based on enzyme kinetic principles, we thought that the PtdIns deWciency hypothesis was not a likely mechanism. The synthesis of all phosphoinositides are ultimately dependent on the activity of the enzyme, phosphatidylinositol synthase (CDP–diacylglycerol:myo-inositol

88

G.T. Berry et al. / Molecular Genetics and Metabolism 82 (2004) 87–92

3-phosphatidyltransferase, EC 2.7.8.11). This integral membrane protein catalyzes the reversible conversion of Ins and CDP-diacylglycerol to PtdIns and CMP [8,9], as well as PtdIns/Ins exchange [10,11]. Early studies of this enzyme employed detergents to solubilize membranes and yielded a Km for Ins that was in the millimolar range [8,9]. Utilizing the enzyme in the exchange mode and in the absence of detergent, we uncovered a Km for Ins of 38
M in a rat brain cortical synaptosome preparation [10]. Unlike brain with millimolar levels of Ins, adult hepatocytes contain approximately 100
M Ins [12], yet this extremely active cell has diverse needs for phosphoinositide synthesis. Taken together, the data suggested that only micromolar levels would suYce for adequate synthesis of PtdIns. Furthermore, since CDP-diacylglycerol is present in concentrations much lower than Ins, it was more likely the rate-limiting substrate in the PtdIns synthase reaction [11]. To properly address this fundamentally important biochemical problem, it was necessary to identify the key active Ins transporter in brain and signiWcantly curtail its activity. The brain tissue Ins concentration gradient was essentially ablated by generation of newborn mice with a homozygous targeted deletion of the Na+/myoinositol cotransporter (SMIT1 or SLC5A3) gene [7,13]. Newborn SMIT1 (¡/¡) fetuses at embryonic day (E) 18.5 have no evidence of SMIT1 mRNA, a 92% reduction in the level of brain Ins, an 84% reduction in whole body Ins, and expire shortly after birth because of central apnea [7]. This work showed that the SMIT1 transporter is responsible for the Ins concentration gradient in the fetal-placental unit and that the most severe Ins deWciency state generated to date in a mammal was uniformly lethal but not associated with any pathologic lesions in the fetus and placenta. Thus, this Ins deWciency model proved to be ideal for studying a lethal biochemical abnormality that is not inconsistent with a defect in neuronal signaling. Our goal was to measure PtdIns levels in the brain of these mice with Ins deWciency as any deWcit in PtdIns synthase activity and inositolphospholipid synthesis should Wrst be reXected in the membrane content of PtdIns.

Materials and methods Three diVerent methods were utilized to measure PtdIns mass: (1) MALDI-TOF mass spectrometry (MS); (2) HPLC and; (3) HPLC—online electrospray ionization (ESI) MS/MS. All of the brain PtdIns measurements were performed in a double-blinded manner. For the Wrst 2 techniques, the E 18.5 fetuses were removed from the uterine sacs of anesthetized pregnant mice, and euthanized by decapitation. The whole brains were removed within 1 min and immediately frozen by exposure to liquid nitrogen. The wet weight of each brain was

measured and they were subjected to an ice cold 7% perchloric acid extraction followed by a Folch extraction of the subsequent pellet with chloroform:methanol:HCI (2:1:0.01). For the MALDI-TOF MS analyses, an internal standard containing soybean PtdIns was employed. The major constituent in the soybean lipid mixture was PtdIns (16:0–18:2) with m/z 833.5 (minus one proton) containing palmitoyl and linoleoyl as the molecular species. Similar to the method of Schiller et al. [14], the MALDI-TOF MS was performed on an ABI 4700 instrument in the negative ion mode. The Folch extract of brain tissue containing the internal standard was evaporated to dryness with N2, reconstituted with chloroform and the matrix agent, 2,5-dihydroxybenzoic acid, and 0.1
l was applied to the target. In the second method, the dried Folch extracts were sent to Avanti Polar Lipids, in Alabaster, AL for analysis by gradient HPLC with evaporative light scattering detection. It was performed on an Agilent 1100 Quatenary system with autosampler and Sedex 55 detector and the peak identity of PtdIns was veriWed using a Hewlett–Packard 59987A/5989A electrospray mass spectrometer. An Advanced Separation Technologies Diol column was employed in a gradient elution system as described by Hollan et al. [15]. The levels of PtdIns, phosphatidylcholine/phosphatidic acid, phosphatidylethanolamine, phosphatidylserine, lyso-phosphatidylcholine, lyso-phosphatidylethanolamine, phosphatidylglycerol, cholesterol, triglycerides, and free fatty acids were determined in the Folch extracts of whole brain. Because of the possibility that degradation of phosphoinositides during the less than 1-min time required for removal of the fetal brain could have aVected the accurate measurement of the in situ PtdIns content, we measured PtdIns in whole fetal heads in the third method. Immediately following the removal of the fetuses from the uterine sacs, they were immersed in liquid nitrogen. Following decapitation of the frozen fetuses, the entire head was subjected to a Folch extraction. The ratio of PtdIns to phosphatidylserine was determined by Drs. P. Vreken and F. Valianpour using a HPLC—online—ESI MS/MS method [16] in the Laboratory for Genetic Metabolic Diseases, Academic Medical Center, at the University of Amsterdam (Head: Professor Dr. R.J.A. Wanders).

Results and discussion As shown in Fig. 1, the major PtdIns (18:0–20:4) in the mammalian brain is one with stearoyl and arachidonoyl as the molecular species and minus one proton giving m/z 885.5. This peak is not present in the soybean standard that contains the 833.5 PtdIns that in turn is absent in the brain of the E18.5 fetus. The results of the analyses of two litters are shown in Fig. 2. There was no diVerence in

G.T. Berry et al. / Molecular Genetics and Metabolism 82 (2004) 87–92

89

Fig. 1. MALDI-TOF MS analyses of soybean PtdIns standard and murine brain Folch extracts with and without standard. The upper panel shows the murine brain Folch extract spectrum with the prominent peak at 885.5 m/z representing PtdIns (18:0–20:4) with stearoyl and arachidonoyl as the molecular species. The lower panel shows the soybean standard with a peak at 833.5 representing PtdIns (16:0–18:2) with palmitoyl and linoleoyl as the molecular species. Please note that the mammalian brain extract does not contain a PtdIns molecule with this molecular composition. The middle panel shows the murine brain extract with the soybean standard present.

Fig. 2. Relative PtdIns content in E18.5 fetal brain versus genotype. Analyses were performed on brains obtained from several litters and the genotypes are shown in the x axis. The MALDI-TOF MS results are expressed as the ratio of ion intensity of 885.5/833.5 per 100 mg of brain weight.

the ion intensity ratio of PtdIns (885.5/833.5) per 100 mg brain in SMIT1 (¡/¡) and SMIT1 (+/¡) brain tissue compared to controls. There were no diVerences in

PtdIns as well as the major lipids in the SMIT1 (¡/¡) samples compared to the normals. The percent of PtdIns per total brain lipids in the samples from knockout and

90

G.T. Berry et al. / Molecular Genetics and Metabolism 82 (2004) 87–92

wild types fetuses were 5.2 § 0.18% (n D 3) and 5.2 § 0.27% (n D 5), respectively, and are shown in Fig. 3. The ratios of PtdIns to phosphatidylserine in the SMIT1 (¡/¡), (+/¡), and (+/+) were 0.33 § 0.06 (n D 3), 0.32 § 0.05 (n D 4), and 0.25 § 0.10 (n D 4), respectively, indicating that there was no diVerence in PtdIns mass in the knockout samples of which a signiWcant fraction represented nervous system tissue. We are not surprised by these Wndings despite the fact that it has been very popular for decades to assume that a reduction of Ins as with lithium treatment would decrease PtdIns and subsequently polyphosphoinositides in cellular membranes [17,18]. It is time that this hypothesis be abandoned. As suggested by AgranoV and Fisher [19], it is possible that an elevation of one or more inositol phosphates due to enzyme inhibition by lithium is important in the eVects of lithium in the adult brain. But it is not plausible that these eVects are mediated by a perturbation in phosphoinositide synthesis due to Ins deWciency per se, especially as it is on average of only modest proportions, e.g., 25–30% reduction. Aside from discrete microdomains of membranes with reduced PtdIns content, the only other way that a reduction in Ins could perturb PtdIns metabolism is by interfering with Xux into PtdIns without aVecting the concentration of PtdIns. Work on the mechanism(s) of lithium action on the developing organism have provided insight into the proteins such as the multifunctional protein kinase, glycogen synthase kinase 3 (GSK3) [18], that exhibit alterations in functions. Finally, Klein and Melton [20] showed that lithium directly inhibits the GSK3- kinase activity and this leads to activation of Wnt/-catenin

Fig. 3. The HPLC analyses of PtdIns in brain extracts of wild-type (Wt) and knock-out (KO) E18.5 fetuses. The results are expressed as the percent of PtdIns per total brain lipids.

signaling pathway. A re-interpretation of these Wndings may now be required for two important reasons: severe Ins deWciency independent of PtdIns metabolism is potentially lethal and Ins can rescue lithium-induced teratogenic eVects. Hedgepeth et al. [21] have demonstrated that the phenotypic malformations induced by a dominant negative form of GSK3- (DN-GSK-3-) in a transgenic line can also be rescued with Ins, even through Ins metabolism has never been linked or seemingly has nothing to do with the GSK3-/-catenin/axin/ APC/protein phosphatase 2A complex [22] that is perturbed by lithium and when defective is suYcient to cause the developmental defects. It may be premature, however, to link the role of Ins in the mammalian nervous system with its role in a developing organism as while high millimolar concentrations of Ins are central to the enigma in brain, it is not at all clear that such relatively high levels are found in the target cells that are perturbed by lithium in the Xenopus embryo or slime mold, Dictyostelium. Based on the accepted two roles of Ins, the only reasonable explanation for high millimolar levels would be for the buVering of osmotic changes in the brain. Yet, even though Ins is the most important non-nitrogenous osmolyte in brain and the SMIT1 gene is under osmoregulatory control [23–26], there were no pathologic Wndings in the knockout fetus that might be compatible with a lethal alteration in regulation of osmotic balance in the absence of hypertonic or hypotonic stress. Nonetheless, we can not rule out the possibility that, as Ins levels decline, in the interest of preserving osmotic balance, other osmolyte molecules accumulate that are toxic to brain cells. The absence of this enigmatic function of Ins in the nervous system may be the cause of the lethal central apnea in the newborn SMIT1 (¡/¡) fetuses. This same function that we propose is independent of inositolphospholipid synthesis may be augmented by raising ambient Ins levels into the millimolar range in a lithiumexposed embryonic tissue and developing cells during the rescue of lithium toxicity in the embryo and in the isolated protein kinase D/N GSK3- model. While it is possible that this eVect is mediated by Ins itself, the observation of Williams and colleagues on lithium and Dictyostelium suggests another possibility involving inositol polyphosphates. They found that the developmental defects induced by lithium may be abrogated when a spontaneous mutant form of this organism lacks the prolyl oligopeptidase (PO) enzyme [27]. The PO protein appears to have a positive eVect on the activity of the multiple inositol polyphosphate phosphatase and when absent produces an increase in Ins-1,4,5-P3 levels that can apparently overcome the eVect of lithium. It is clear that in this system the Ins-1,4,5-P3 can actually be derived from Ins-1,3,4,5-6-P5 [27]. Since Ins deWciency per se does not appear to reduce PtdIns-4,5-P2, the apparent obligatory precursor of D-Ins-1,4,5-P3 in mammalian

G.T. Berry et al. / Molecular Genetics and Metabolism 82 (2004) 87–92

cells, it is possible that higher-order inositol polyphosphates can be directly synthesized from Ins as well as from Ins-1,4,5-P3 derived from PtdIns-4,5-P2. This is true for Dictyostelium [28], as well as in the plant kingdom [29], in which inositol polyphosphates such as phytate are directly synthesized from Ins utilizing Ins-3-phosphate as the initial phosphorylated precursor [30]. Further support for the importance of this enzyme pathway in this primitive organism is that no developmental defects occur in genetic mutants that lack PtdIns-speciWc phospholipase C activity [31]. While the other special role of Ins in mammalian metabolism may be for direct synthesis of inositol polyphosphates initially utilizing an enzyme that has a Km for Ins that is in the millimolar range, it has not been possible to demonstrate that D-Ins3-P (or L-Ins-1-P), which can be synthesized from D-glucose-6—P in only certain cells such as brain capillary pericytes [32], can be further phosphorylated [30,33]. Therefore, although higher levels and turnover of inositol polyphosphates may be necessary in the nervous system to competitively regulate polyphosphoinositide binding to proteins [34,35], current data suggest that species such as phytate can only be synthesized from D-Ins1,4,5-P3. Further insight into how Ins independent of phosphoinositide metabolism may be critical to brain homeostasis will necessitate a detailed examination of alternate Ins pathways and metabolites including inositol isomers [36]. In summary, a loss of SMIT1 protein leads to massive Ins deWciency and a pathophysiology that centers around SMIT1 itself and/or Ins as either may function in roles that have yet to be elucidated. Our genetic model may facilitate the identiWcation of a new role for Ins in metabolism.

Acknowledgments This work was supported by a grant from the March of Dimes (to G.T.B.). We thank Professor R.J.A. Wanders, Dr. F. Valianpour, and the late Dr. P. Vreken for performing the ESI MS/MS analyses.

References [1] W.R. Sherman, A.L. Leavitt, M.P. Honchar, L.M. Hallcher, B.E. Phillips, Evidence that lithium alters phosphoinositide metabolism, J. Neurochem. 36 (1981) 1947–1951. [2] M.J. Berridge, C.P. Downes, M.R. Hanley, Lithium ampliWes agonist-dependent phosphatidylinositol responses in brain and salivary glands, Biochem. J. 206 (1982) 587–595. [3] R.H. Belmaker, Y. Bersudsky, J. Benjamin, G. Agam, J. Levine, O. Kofman, Manipulation of inositol-linked second messenger systems as a therapeutic strategy in psychiatry, Adv. Biochem. Psychopharmacol. 49 (1995) 67–84. [4] Y. Maeda, InXuence of ionic conditions on cell diVerentiation and morphogenesis of the cellular slime molds, Dev. Growth DiVer. 12 (1970) 217–227.

91

[5] W.B. Busa, R.L. Gimlich, Lithium-induced teratogenesis in frog embryos prevented by a polyphosphoinositide cycle intermediate or a diacyglycerol analog, Dev. Biol. 132 (1989) 315–324. [6] B.J. Holub, Metabolism and function of myo-inositol and inositol phospholipids, Annu. Rev. Nutr. 6 (1986) 563–597. [7] G.T. Berry et al., Loss of murine Na+/myo-inositol cotransporter leads to brain myo-inositol depletion and central apnea, J. Biol. Chem. 278 (2003) 18297–18302. [8] H. Paulus, E.P. Kennedy, The enzymatic synthesis of inositol monophosphatide, J. Biol. Chem. 235 (1960) 1303–1311. [9] J.A. Benjamins, B.W. AgranoV, Distribution and properties of CDP-diglyceride:inositol transferase from brain, J. Neurochem. 16 (1969) 513–527. [10] G.T. Berry, J.R. Yandrasitz, V.M. Cipriano, S.M. Hwang, S. Segal, Phosphatidylinositol:myo-Inositol exchange activity in intact nerve endings: substrate and cofactor dependence, nucleotide speciWcity, and eVect on synaptosomal handling of myo-Inositol, J. Neurochem. 46 (1986) 1073–1080. [11] A. Lykidis, P.D. Jackson, C.O. Rock, S. Jackowski, The role of CDP-diacylglycerol synthetase and phosphatidylinositol synthase activity levels in the regulation of cellular phosphatidylinositol content, J. Biol. Chem. 272 (1997) 33402–33409. [12] S.H. Sigal, J.R. Yandrasitz, G.T. Berry, Kinetic evidence for compartmentalization of myo-inositol in hepatocytes, Metabolism 42 (1993) 395–401. [13] K.E. McVeigh et al., Murine chromosome 16 telomeric region, homologous with human chromosome 21q22, contains the osmoregulatory Na+/myo-inositol cotransporter (SLC5A3) gene, Cytogenet. Cell Genet. 88 (2000) 153–158. [14] J. Schiller et al., Lipid analysis by matrix-assisted laser desorption and ionization mass spectrometry: a methodological approach, Anal. Biochem. 267 (1999) 46–56. [15] W.L. Hollan, E.C. Stauter, B.J. Stith, QuantiWcation of phosphatidic acid and lysophosphatidic acid by HPLC with evaporative light-scattering detection, J. Lipid Res. 44 (4) (2003) 854–858. [16] F. Valianpour, R.J.A. Wanders, P.G. Barthm, H. Overmars, A.H. Van Gennip, Quantitative and compositional study of cardiolipin in platelets by electrospray ionization mass spectrometry: application for the identiWcation of barth syndrome patients, Clin. Chem. 48 (9) (2002) 1390–1397. [17] R.S. Williams, L. Chen, A.W. Mudge, A.J. Harwood, A common mechanism of action for three mood-stabilizing drugs, Nature 417 (2002) 292–295. [18] A.J. Harwood, G. Agam, Search for a common mechanism of mood stabilizers, Biochem. Pharm. 66 (2003) 179–189. [19] B.W. AgranoV, S.K. Fisher, Inositol, lithium, and the brain, Psychopharmacol. Bull. 35 (2001) 5–18. [20] P.S. Klein, D.A. Melton, A molecular mechanism for the eVect of lithium on development, Proc. Natl. Acad. Sci. USA 93 (1996) 8455–8459. [21] C.M. Hedgepeth et al., Activation of the Wnt signaling pathway: a molecular mechanism for lithium action, Dev. Biol. 185 (1997) 82– 91. [22] C.M. Hedgepeth, M.A. DeardorV, K. Rankin, P. Klein, Regulation of glycogen synthase kinase 3 and downstream Wnt signaling by axin, Mol. Cell. Biol. 19 (1999) 7147–7157. [23] J.H. Thurston, W.R. Sherman, R.E. Hauhart, R.F. Kloepper, Myo-Inositol: a newly identiWed nonnitrogenous osmoregulatory molecule in mammalian brain, Pediatr. Res. 26 (1989) 482– 485. [24] J.S. Handler, H.M. Kwon, Cell and molecular biology of organic osmolyte accumulation in hypertonic renal cells, Nephron 87 (2001) 106–110. [25] J.J. Mallee, M.G. Atta, V. Lorica, J.S. Rim, H.M. Kwon, A.D. Lucente, Y. Wang, G.T. Berry, The structural organization of the human Na+/myo-inositol cotransporter (SLC5A3) gene and characterization of the promoter, Genomics 46 (1997) 459–465.

92

G.T. Berry et al. / Molecular Genetics and Metabolism 82 (2004) 87–92

[26] J.S. Rim, M.G. Atta, S.C. Dahl, G.T. Berry, J.S. Handler, H.M. Kwon, Transcription of the sodium/myo-inositol cotransporter gene is regulated by multiple tonicity-responsive enhancers spread over 50 kb pairs in the 50-Xanking region, J. Biol. Chem. 273 (1998) 20615–20621. [27] R.S.B. Williams, M. Eames, W.J. Ryves, J. Viggars, A.J. Harwood, Loss of a prolyl oligopeptidase conders resistance to lithium by elevation of inositol (1,4,5) trisphosphate, EMBO J. 18 (1999) 2734–2745. [28] L.R. Stephens, R.F. Irvine, Stepwise phosphorylation of myo-inositol leading to myo-inositol hexakisphosphate in Dictyostelium, Nature 346 (1990) 580–583. [29] C.A. Brearly, D.E. Hanke, Metabolic evidence for the order of addition of individual phosphate esters to the myo-inositol moiety of inositol hexakisphosphate in the duckweed Spirodela polyrhiza L., Biochem. J. 314 (1996) 227–233. [30] S.B. Shears, How versatile are inositol phosphate kinases? Biochem. J. 377 (2004) 265–280. [31] A.L. Drayer, J. Van der Kaay, G.W. Mayr, P.J. Van Haastert, Role of phospholipase C in Dictyostelium: formation of inositol

[32]

[33]

[34]

[35]

[36]

1,4,5-trisphosphate and normal development in cells lacking phospholipase C activity, EMBO J. 13 (1994) 1601–1609. Y.H. Wong, S.J. Kalmbach, B.K. Hartman, W.R. Sherman, Immunohistochemical staining and enzyme activity measurements show myoinositol-I-phosphate synthase to be localized in the vasculature of the brain, J. Neurochem. 48 (1987) 1434– 1442. S.K. Fisher, J.E. Novak, B.W. AgranoV, Inositol and higher inositol phosphates in neural tissues: Homeostatis, metabolism, and functional signiWcance, J. Neurochem. 82 (2002) 736– 754. K.A. Beck, J.H. Keen, Interaction of phosphoinositide cycle intermediates with the plasma membrane-associated clathrin assembly protein AP-2, J. Biol. Chem. 266 (1991) 4442–4447. P. De Camilli, S.D. Emr, P.S. McPherson, P. Novick, Phosphoinositides as regulators in membrane traYc, Science 271 (1996) 1533–1539. A. Shalovbina, M.L. Greenberg, Epi-inositol regulates expression of the yeast INOl gene encoding inositol-I-P synthase, Mol. Psychiatry 7 (2002) 174–180.