Pflügers Archiv - European Journal of Physiology

, Volume 456, Issue 6, pp 1199–1206

Discovery of protein-palmitoylating enzymes


  • Ryouhei Tsutsumi
    • Division of Membrane Physiology, Department of Cell PhysiologyNational Institute for Physiological Sciences
  • Yuko Fukata
    • Division of Membrane Physiology, Department of Cell PhysiologyNational Institute for Physiological Sciences
    • PRESTOJapan Science and Technology Agency
    • Division of Membrane Physiology, Department of Cell PhysiologyNational Institute for Physiological Sciences
    • PRESTOJapan Science and Technology Agency
Signaling and Cell Physiology

DOI: 10.1007/s00424-008-0465-x

Cite this article as:
Tsutsumi, R., Fukata, Y. & Fukata, M. Pflugers Arch - Eur J Physiol (2008) 456: 1199. doi:10.1007/s00424-008-0465-x


Posttranslational modification provides proteins with additional function and regulatory control beyond genomic information, allowing cells to maintain homeostasis and respond to extracellular signals. Protein palmitoylation, the common posttranslational modification with the lipid palmitate, plays a pivotal role in protein trafficking and function. Palmitoylation is unique in that it is reversible and dynamically regulated by specific extracellular signals. The reversible nature of protein palmitoylation enables proteins to shuttle between intracellular compartments upon extracellular signals. However, the molecular mechanisms of protein palmitoylation have long been elusive, mostly because the enzymes responsible for protein palmitoylation were unknown. Recently, genetically conserved DHHC family proteins have emerged as palmitoyl-acyl transferases. With the identification of specific enzymes for palmitoylated proteins, including H-Ras, PSD-95, and eNOS, the specificity and regulatory mechanism of DHHC enzymes are beginning to be clarified.


Protein palmitoylationDHHC proteinLipid modificationProtein targetingPalmitoyl-acyl transferase


Posttranslational processing events, including phosphorylation, glycosylation, and lipid modification, represent central molecular mechanisms for cells to respond to extracellular signals. Protein palmitoylation—frequent lipid modification with the lipid palmitate—regulates the membrane targeting, subcellular trafficking, and function of proteins [5, 18]. Palmitoylation occurs either through labile thioester linkage (S-palmitoylation) at cysteine residue or stable amide linkage (N-palmitoylation) at glycine/cysteine residue. Irreversible N-palmitoylation is observed with limited proteins, such as the secreted morphogen Sonic hedgehog [25]. S-Palmitoylation is found more commonly in most palmitoylated proteins. In this review, the term protein palmitoylation means “S-palmitoylation.”

Palmitoylation modifies many important proteins, including guanosine triphosphate (GTP)-binding proteins, enzymes, ion channels, neurotransmitter receptors, and synaptic scaffolding proteins. Examples include classical trimeric G protein α subunits (Gαs, Gαi, Gαq; Fig. 1), small GTPases (H-Ras, N-Ras, RhoB), nitric oxide synthases (NOSs; endothelial NOS [eNOS], inducible NOS), and PSD-95, a representative scaffolding protein at the neuronal postsynapse. Unlike other irreversible lipid modifications, such as myristoylation and prenylation, palmitoylation is relatively labile and palmitates on proteins turn over rapidly. This reversible nature of palmitoylation allows proteins to shuttle between the cytoplasm and intracellular organelles/plasma membranes upon extracellular signals. For example, PSD-95 is targeted to the postsynaptic membrane through palmitoylation [36] and anchors α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA)-type glutamate receptors at the excitatory postsynapse. Upon glutamate stimulation, depalmitoylation of PSD-95 is accelerated, and this process dissociates PSD-95 and AMPA receptors from the postsynaptic sites, contributing to downregulation of AMPA receptors [6]. Thus, dynamic protein palmitoylation is involved in various physiological phenomena, such as synaptic transmission. However, the enzymes that add palmitate to proteins (palmitoyl-acyl transferases [PATs]) and those that cleave the thioester bond (palmitoyl-protein thioesterases [PPTs]) remained unidentified for more 30 years.
Fig. 1

Palmitoylation is a dynamic reversible lipid modification. a The targeting and function of various signaling proteins, including Gα and H-Ras, are dynamically regulated by both palmitoyl-acyl transferases (PATs) and palmitoyl-protein thioesterases (PPTs). The activity of PATs and PPTs is considered to be regulated by extracellular signals. Although all PATs (DHHC proteins) are membrane-spanning proteins at the intracellular organelles, such as Golgi and endoplasmic reticulum, organelles are omitted in this figure. GPCR, G protein-coupled receptor. b Palmitoylation of Gαq is essential for its plasma membrane targeting. Wild-type (WT) Gαq-GFP clearly targets to the plasma membrane in Hela cells (top panel). In contrast, the palmitoylation-deficient mutant of Gαq-GFP (C9,10S) diffusely localizes in cytoplasm (bottom)

Yeast genetic analysis identified DHHC proteins as authentic PATs

Numerous biochemists have faced difficulties in developing a specific in vitro palmitoylation assay and purifying authentic PATs. Some results have suggested that palmitoylation in cells might occur nonenzymatically [4] because the formation of palmitoyl thioester linkage on proteins could occur spontaneously in vitro in the presence of palmitoyl-coenzyme A. Thus, identification of PATs has been challenging.

Elegant genetic studies in yeast have recently shed light on this fundamental question. Forward genetic screens identified Erf2/Erf4 [1, 19] and Akr1 [29] as the palmitoyl transferase for yeast Ras2 and yeast casein kinase 2 (Yck2), respectively. Deletion of Erf2/Erf4 or Akr1 prevents palmitoylation of Ras2 or Yck2, respectively. Also, recombinant Erf2/Erf4 and Akr1 palmitoylate Ras2 and Yck2, respectively, in vitro. Interestingly, Erf2 and Akr1 have four- to six-pass transmembrane domains and share a common domain referred to as a DHHC domain, a cysteine-rich domain with a conserved aspartate–histidine–histidine–cysteine signature motif. This DHHC domain is necessary for the PAT activity of Erf2 and Akr1, suggesting that DHHC-domain-containing proteins (DHHC proteins) may function as general PATs. In fact, other DHHC proteins (Pfa3, Pfa4, and Swf1) in yeast have PAT activity. Pfa3 [13, 33] and Pfa4 [17, 30] mediate Vac8 and Chs3 palmitoylation, respectively, and regulate their subcellular distributions. Swf1 palmitoylates yeast soluble N-ethylmaleimide-sensitive factor attachment receptor proteins [39].

One may ask if another class of PATs besides DHHC proteins exists in cells or if a nonenzymatic mechanism is still supported. Systematic proteomic analyses by Roth et al. [30] addressed this question. The group established a method (acyl–biotinyl exchange [ABE]) for purifying palmitoylated proteins and subsequently identifying proteins by mass spectrometry [3, 30]. This approach allowed the investigators to identify 35 new palmitoyl proteins, as well as 12 known palmitoyl proteins in yeast. They applied this proteomic method to mutant yeast strains either singly or multiply deficient for the seven yeast DHHC proteins. When six of seven DHHC genes in yeast were deleted, 29 of the 30 surveyed palmitoyl proteins were not detected by the ABE method [30]. Thus, DHHC family proteins are authentic PATs that catalyze most of the protein palmitoylation in yeast. Also, this loss-of-function study suggests that the DHHC protein family has discrete and partially overlapping PAT specificities. Besides yeast (seven genes), the DHHC family is conserved from the nematodes Caenorhabditis elegans (15 genes, predicted), fruit fly Drosophila melanogaster (23 genes, predicted), and mammals (23 genes). DHHC proteins are also found in the plant Arabidopsis thaliana and regulate plant cell growth [12]. Thus, the DHHC protein family is expected to represent palmitoylating enzymes conserved from various species.

Mammalian DHHC proteins

In mammalian genomes (human and mouse), 23 kinds of DHHC proteins are predicted (Fig. 2). So far, several DHHC proteins have been reported to be associated with human diseases (Table 1) or binding partners with neuronal receptors. Mutations of DHHC2 have been found in several colorectal cancers [24]. DHHC3, also called GODZ (Golgi-apparatus-specific protein with the DHHC zinc finger domain), was identified as a GluR1 (an AMPA receptor subunit)-interacting protein by yeast two-hybrid analysis [38]. DHHC8 was proposed as a candidate gene for schizophrenia [2, 22], and DHHC9 and DHHC15 have been reported to be associated with X-linked mental retardation [21, 27]. DHHC17, a counterpart of yeast Akr1, was identified as a huntingtin-interacting protein, HIP14 [32]. However, the biological function of mammalian DHHC proteins has been unclear. Although genetic studies in yeast inspired investigators to identify PATs toward a mammalian palmitoyl protein, the large size of the DHHC protein family has led investigators to wonder how to identify specific enzymes for specific substrates.
Fig. 2

DHHC protein family in mammals. a Domain structures of representative DHHC proteins, DHHC3 and DHHC6. These proteins have four transmembrane domains and a conserved cysteine-rich domain containing DHHC motif in the cytoplasmic loop. The DHHC sequence is essential for palmitoylating activity. DHHC3 and DHHC6 have a PDZ-binding motif and an SH3 domain, respectively. b Phylogenetic tree of the mouse DHHC protein family and a summary of screening for PAT candidates. The tree is based on alignment of the DHHC core domains. DHHC proteins apparently have substrate specificity. DHHC3 and DHHC7 palmitoylate Gα, SNAP-25, GABAA, receptor γ2 subunit (GABAARγ2), PSD-95, and GAP-43, suggesting that DHHC3 and DHHC7 show broad substrate specificity. DHHC3 is also known as GODZ, DHHC17 as HIP14, DHHC11 as NIDD, and DHHC2 as ream. GenBank accession numbers for each DHHC clone are listed in Fukata et al. [9]. Sc, Saccharomyces cerevisiae

Table 1

Pathophysiological functions of DHHC proteins

DHHC protein

Pathophysiological function(s)

Associated substrates



Colorectal cancer (Hs)




Schizophrenia (Hs, Mm)


[2, 22]


X-linked mental retardation (Hs)



Colorectal cancer (Hs)




Bladder cancer (Hs)




X-linked mental retardation (Hs)




Huntington disease (Mm)



Synaptic transmission (Dm)


[23, 34]

Hs, Homo sapiens; Mm, Mus musculus; Dm, Drosophila melanogaster

In 2004, three groups, including our own, revealed that mammalian DHHC proteins function as PATs [9, 14, 16]. Keller et al. [16] identified DHHC3/GODZ as a molecule that interacts with the γ-aminobutyric acid (GABA)A receptor γ2 subunit and showed that DHHC3/GODZ palmitoylates the GABAA receptor γ2 subunit in heterologous cells, whose palmitoylation is a major determinant of proper trafficking and subcellular targeting of GABAA receptors [7, 26]. This study raises the possibility that some DHHC enzymes tightly interact with their substrates, and we may identify the PATs by exploring the binding proteins with a substrate. In fact, DHHC17/HIP14, identified as a huntingtin-interacting protein [32], palmitoylates huntingtin [42]. However, most of the enzyme reaction is mediated by the transient interaction with its substrate and does not necessarily require such a tight interaction. In other words, this approach cannot be applied to every substrate. Huang et al. [14] noticed the predominant neuronal expression of DHHC17/HIP14 and found that DHHC17/HIP14-enhanced palmitoylation of several neuronal proteins, including SNAP-25, PSD-95, GAD65, synaptotagmin I, and huntingtin. However, most DHHC proteins express ubiquitously, judging from the several databases of messenger RNA expression. Therefore, it is difficult to determine physiological PATs based on expression profiles of DHHC proteins. A more systematic screening method is required to determine the specific PATs in mammalian cells.

Systematic method to identify specific PATs

To evaluate PAT activity quantitatively and relatively, we isolated the complementary DNAs (cDNAs) of all mouse DHHC proteins and established a simple, systematic, and straightforward screening method to identify the candidate PAT for specific substrates [9, 10]. We first selected PSD-95, a well-characterized palmitoyl protein, as a model substrate. The procedure includes three steps: (1) transfection of PSD-95 cDNA together with an individual DHHC clone into HEK293 cells, (2) metabolic labeling with [3H]palmitic acid, and (3) sodium dodecyl sulfate polyacrylamide gel electrophoresis and fluorography. We usually do not purify the substrate by immunoprecipitation for detection of DHHC-clone-induced palmitoylation. Immunoprecipitation does not necessarily reveal all palmitoylated proteins because usual extraction using Triton X-100 or radio-immunoprecipitation assay buffer cannot extract all palmitoylated proteins, which are strongly associated with membrane and go to the insoluble fraction. Taking advantage of our systematic method, we found that a subset of DHHC proteins (DHHC2, 3, 7, and 15: PSD-95-PAT [P-PAT]) quantitatively enhance palmitoylation of PSD-95. After this screening, we examined whether the candidate DHHC clone expresses in brain where PSD-95 expresses. Also, we examined whether palmitoylation of the endogenous substrate is affected when the candidate DHHC clone is inhibited by the dominant-negative mutant or small interfering RNA (siRNA) knockdown. This screening method can be applied to various substrates as well as to PSD-95, and we identified various enzyme-substrate pairs. Examples include eNOS [8], the GABAA receptor γ2 subunit [7], SNAP-25 [9], Gαs (Tsutsumi et al., unpublished data), Lck tyrosine kinase, GAP-43, and H-Ras (Fig. 2). The palmitoylation levels of tested substrates were all enhanced by some DHHC proteins, strongly suggesting that the DHHC protein family functions as the main PATs in mammals, as shown in yeast [30].

Substrate specificity

Using our DHHC clone library, we have already screened for PATs for more than 20 kinds of palmitoyl proteins. Our screening results apparently indicate that DHHC proteins have substrate specificity (Fig. 2), a finding that accords with the loss-of-function analysis in yeast as described above [30]. Among DHHC proteins, DHHC3 and the closely related DHHC7 have broad substrate specificity; in other words, DHHC3 and DHHC7 enhance the palmitoylation of various substrates, including PSD-95, eNOS, the GABAA receptor γ2 subunit, SNAP-25, Gαs, and GAP-43. In contrast, DHHC2 and DHHC15 are more specific to PSD-95 and GAP-43. DHHC9 and DHHC18 are also specific to H-Ras. These screenings will prompt us to classify 23 DHHC proteins into several subfamilies. There may be some structural correlations to enzyme specificity (Fig. 2).

Besides the catalytically critical DHHC domain, an individual DHHC protein has some regulatory regions such as the SH3 domain, ankyrin repeat, and PDZ-binding motif at the C-terminal and N-terminal regions (similarly, protein kinases also share a core catalytic region and differ in regulatory domains that afford differential control systems). It is conceivable that these regions may recruit specific substrates or regulators to DHHC proteins. It is worthwhile to identify the binding partners of the DHHC protein family.

Erf2, a yeast DHHC protein, functions as a PAT for yeast Ras2 in a complex with Erf4. Swarthout’s group [35] identified GCP16 as a mammalian functional ortholog of Erf4. They found that DHHC9 requires GCP16 for its PAT activity toward H/N-Ras and protein stability. Because several GCP16 family proteins are predicted in the mammalian genome, some other DHHC proteins may complex with GCP16 family proteins. The GCP16 family may determine the substrate specificity, enzymatic kinetics, subcellular localization, and stability of some DHHC proteins as an auxiliary subunit.

Consensus sequence of protein palmitoylation

Whereas the consensus sequences for myristoylation (N-terminal glycine in the sequence MGXXXS/T [M: Met; G: Gly; S: Ser; T: Thr; X: a variety of amino acids]) and isoprenylation (cysteine in the C-terminal CAAX motif [C: Cys; A: aliphatic amino acids]) are well known, the consensus sequence of protein palmitoylation has been unknown. In fact, palmitoylation occurs at the various cysteine residues located at (1) the amino terminal region in PSD-95 and Gα, (2) the internal region in SNAP-25 and GAD65, (3) the carboxyl terminal region in H-Ras and RhoB, and (4) the juxtamembrane region of various transmembrane proteins. Given our screening results, we may expect the consensus sequences of palmitoylation, which is DHHC subfamily specific. Our preliminary results suggest that DHHC21 favors the substrate with cysteines near myristoylated glycine, such as Lck, Fyn, and eNOS (Fukata et al., unpublished data). Mutagenesis analyses of residues surrounding palmitoylation sites and structural analyses will determine the exact consensus sequence of protein palmitoylation.

Regulatory mechanism of DHHC proteins

Protein palmitoylation is a reversible modification and dynamically regulated by the extracellular stimulus akin to protein phosphorylation. The finely tuned balance between palmitoylation by PATs and depalmitoylation by PPTs determines the palmitoylation levels of substrates. The best-studied substrates whose palmitoylation levels are regulated by extracellular signals are Gαs, neuronal scaffolding protein PSD-95, and eNOS. More than 10 years ago, it was shown that palmitate turnover on Gαs is accelerated by its upstream β-adrenergic receptor [40]. Recently, dynamic palmitate cycling on PSD-95 was reported. By the pulse-chase labeling method, the half-life of palmitate on PSD-95 is calculated to be about 2 h, whereas the half-life of PSD-95 protein itself is about 36 h [6]. The half-life of palmitate on PSD-95 increased (~4 h) when the glutamatergic synaptic activity was inhibited by kynureic acid, an inhibitor of glutamate receptors [6]. Conversely, glutamate-induced synaptic activity reduced the half-life of palmitate on PSD-95. These studies strongly suggest that receptor activation stimulates depalmitoylation, and unidentified PPTs seem to be the primary target regulated downstream of agonist stimulation. In contrast, we recently found that P-PAT activity is apparently elevated by kynureic acid (unpublished data). These observations indicate that when synaptic activity is reduced, PAT activity increases and PPT activity decreases, leading to incremental increases in palmitoylated PSD-95. Because increased palmitoylated PSD-95 recruits more AMPA-type glutamate receptors, regulated palmitate cycling on PSD-95 may be involved in homeostasis of glutamate receptors such as synaptic scaling [37].

Recently, Fernández-Hernando’s group [8] showed that DHHC21, one of the eNOS PATs, is involved in the release of nitric oxide stimulated by both ionomycin and adenosine triphosphate (ATP), suggesting that DHHC21 is regulated downstream of calcium and ATP. In contrast to PSD-95 and eNOS, palmitates on SNAP-25 are not dynamically cycled. Neuronal depolarization does not affect palmitate turnover on SNAP-25 [15]. This result implies that the individual substrate is differently regulated by both specific PATs and PPTs.

The most intriguing question is how PAT activity is regulated by extracellular signals. Given that the palmitate cycling on PSD-95 and that on SNAP-25 are apparently different and PATs responsible for palmitoylation are partly different (PSD-95 PATs are DHHC2, 3, 7, and 15; SNAP-25 PATs are DHHC3, 7 and 17), it is conceivable that some DHHC proteins (DHHC2 and 15) may be regulated downstream of synaptic activity. Several regulatory mechanisms are considered: (1) The activity of PATs may be modified by posttranslational modifications, such as phosphorylation and S-nitrosylation. Because S-nitrosylation occurs at cysteine residues and the cysteine residue of the DHHC motif is essential for its PAT activity, it is worthwhile to examine whether the DHHC protein itself is modified by S-nitrosylation. (2) Some second messengers, such as phospholipisds, cyclic AMP, and metal ions, may bind the DHHC protein and modulate its PAT activity. In fact, in yeast, the DHHC protein Akr1p palmitoylates casein kinase, and this activity is increased by ATP in vitro [29]. Further intensive studies are required on this important issue.

One may ask whether the subcellular localization of DHHC proteins is involved in the regulatory mechanism of PATs and in determining substrate specificity. Keller et al. [16] have reported that endogenous DHHC3 is localized in Golgi in primary cultured neurons. Recently, our group also confirmed this specific localization by the specific antibody validated by siRNA experiments (unpublished data). Given that DHHC3 localizes Golgi and mediates much of protein palmitoylation, it is possible that DHHC3 is mainly involved in the protein palmitoylation during or just after protein synthesis rather than the activity-dependent re-palmitoylation that may occur near the plasma membrane. Huang et al. [14] showed that endogenous DHHC17/HIP14 may be associated with several vesicular structures, including the Golgi as well as sorting/recycling and late endosomal structures. We have also preliminarily observed that DHHC2 is localized in dendrites and dynamically moves around synapses (unpublished data). It is also possible that DHHC distribution is dynamically regulated by extracellular stimulation.

Physiological significance of DHHC proteins

The physiological processes in which DHHC proteins are involved are beginning to be elucidated (Table 1). Very recently, two independent groups identified Drosophila DHHC17/HIP14 (dHIP14) as an essential gene for synaptic transmission [23, 34]. Stowers and Isacoff [34] showed that the synaptic components of electroretinograms are completely absent in flies with dHIP14 mutant photoreceptors. Furthermore, loss of dHIP14 leads to mislocalization of palmitoylated presynaptic proteins SNAP-25 and CSP in the ventral nerve cord and in the optic lobe, suggesting that dHIP14 palmitoylates these presynaptic proteins in vivo and regulates their presynaptic targeting and functions. Expression of human HIP14 (48% identity with dHIP14) restored the phenotype of the dHIP14 mutant. Ohyama et al. [23] isolated dHIP14 during forward genetic screening for novel proteins that affect synaptic transmission. The group found that evoked neurotransmitter release at the neuromuscular junction is impaired in the dHIP14 mutant and that CSP is severely mislocalized because of loss of palmitoylation in the mutant. Expression of wild-type dHIP14 in neurons could rescue the mutant phenotypes, indicating that dHIP14 controls neurotransmitter release, possibly through regulating trafficking of CSP to the presynaptic terminus. These works using genetic approaches not only addressed the physiological importance of dHIP14 in synaptic transmission but also established dHIP14 protein as a presynaptic PAT. Given that many postsynaptic proteins, such as PSD-95, are also palmitoylated and trafficked to the postsynapse, another subset of DHHC proteins should function at the postsynapse.

Human genetic evidence based on the single-nucleotide polymorphism rs175174 strongly suggests that ZDHHC8/DHHC8 is associated with schizophrenia [2, 22], although several reports do not support this linkage [11, 31]. By generating knockout mice, Mukai et al. [22] examined whether DHHC8 contributes to the risk of schizophrenia. DHHC8 knockout mice had a deficiency in prepulse inhibition and a decrease in exploratory activity in a new environment, supporting the hypothesis that DHHC8 may be a risk factor for schizophrenia. Interestingly, only homozygous mutant female mice but not male mice displayed a deficiency in prepulse inhibition and exploratory behavior. Identification of in vivo substrates of DHHC8 and further human genetic linkage studies will be required to clarify the physiological function of DHHC8.

Concluding remarks

Why have DHHC proteins not been identified for a long time? Recently, we estimated by its specific antibody that the amount of DHHC3 is only 0.001% of the total protein in brain tissue (unpublished data). This is one reason the numerous biochemical attempts to purify PATs have been unsuccessful. Several important questions remain unanswered, however. First, what are the physiological PPTs? Although one candidate, cytosolic APT1, has been proposed as a PPT for Gα, H-Ras, and eNOS [18], it remains largely unknown that APT1 and its family molecules are general PPTs as DHHC proteins for PATs. Because palmitoyl proteins are anchored to the plasma membrane, we wonder whether some PPTs may be membrane bound or transmembrane proteins. As in the case of PATs, genetic approaches, including genome-wide knockdown, may be useful for identifying the PPT candidate. Second, how can we visualize the dynamic cycling of palmitoylated/depalmitoylated states of proteins? Recently, the fluorescence recovery after photobleaching method combined with an inhibitor of protein palmitoylation (2-bromopalmitate) was applied to the analysis of H-Ras/N-Ras trafficking and showed that H-Ras/N-Ras rapidly cycles between the plasma membrane and Golgi apparatus in a depalmitoylation/repalmitoylation-dependent manner [28]. Photoconversion is also useful for analyzing dynamic trafficking of palmitoyl proteins. Just as the appearance of the phospho-specific antibody greatly contributed to the field of protein phosphorylation, so too will specific visualization of the palmitoylated protein be immensely helpful.

Finally, mutations of DHHC proteins in humans are associated with various diseases, including cancers and neurological disorders (Table 1). Studies of DHHC protein knockout mice will reveal unexpected physiological and pathological functions of DHHC proteins. Because palmitoylation modifies a variety of key proteins involved in cell growth, cell signaling, and synaptic transmission, the DHHC enzyme family may become an ideal therapeutic target.


We thank Dr. Catherine H. Berlot of the Weis Center for Research for kind gifts of Gαq-GFP plasmids, Dr. David S. Bredt of Eli Lilly and company for helpful suggestions and Dr. Jun Noritake for reading the manuscript. R.T. is supported by the Japan Society for the Promotion of Science. Y.F. is supported by the Human Frontier Science Program (HFSP) and Ministry of Education, Culture, Sports, Science and Technology (MEXT) of Japan (grant no. 18700376). M.F. is supported by grants from the HFSP, MEXT (nos. 18022054, 18057032, and 18687008) and the Ministry of Health, Labour and Welfare of Japan (no. 17C-2).

Copyright information

© Springer-Verlag 2008