neo-inositol polyphosphates in the amoeba Entamoeba histolytica.

We have reexamined the structure of inositol phosphates present in trophozoites of the parasitic amoeba Entamoeba histolytica and show here that, rather than being myo-inositol derivatives (Martin, J.-B., Bakker-Grunwald, T., and Klein, G. (1993) Eur. J. Biochem. 214, 711-718), these compounds belong to a new class of inositol phosphates in which the cyclitol isomer is neo-inositol. The structures of neo-inositol hexakisphosphate, 2-diphospho-neo-inositol pentakisphosphate, and 2, 5-bisdiphospho-neo-inositol tetrakisphosphate, which are present in E. histolytica at concentrations of 0.08-0.36 mM, were solved by two-dimensional (31)P-(1)H NMR spectroscopy. No evidence for the co-existence of their myo-inositol counterparts has been found. These neo-inositol compounds were not substrates of 6-diphospho-inositol pentakisphosphate 5-kinase, an enzyme purified from Dictyostelium discoideum that phosphorylates 6-diphospho-myo-inositol pentakisphosphate and more slowly also myo-inositol hexakisphosphate, specifically on position 5. Because preliminary data indicate that large amounts of the same neo-inositol phosphate and diphosphate esters are also present in another primitive amoeba, Phreatamoeba balamuthi, the occurrence of high concentrations of neo-inositol polyphosphates may be much more general than previously thought.

Inositol phosphates are members of a large family of naturally occurring compounds that because of their complex role in cell signaling and homoeostasis have been intensively studied during the last two decades (1). Research has focused mainly on the commonly known myo-derivatives, although many other naturally occurring inositol stereoisomers are known. The most abundant and ubiquitous member of the family, myo-InsP 6 , has been shown to reach an intracellular concentration close to 1 mM in Dictyostelium discoideum (2,3). Recently, new members of this class of compounds have been identified in cellular slime molds as 5-PP-and 6-PP-myo-InsP 5 1 and 3,5-bis-PP-and 5,6-bis-PP-myo-InsP 4 (4 -6). The same or similar highly phosphorylated and diphosphorylated compounds have been detected in free living and parasitic amoebae and a number of mammalian cell types (7)(8)(9)(10)(11). The latter molecules may function in cell signaling (2,(12)(13)(14). Entamoeba histolytica is a human intestinal parasite that causes amoebiasis. Its trophozoites contain high amounts of inositol polyphosphates (8). myo-Inositol is the major cyclitol building natural inositol polyphosphates, and other isomers are only rarely found and in very low concentrations. Consequently, the two major inositol phosphates of E. histolytica have been preliminarily identified as myo-inositol 2,4,6triphosphate and 5-PP-myo-InsP 5 (8). However, we subsequently noticed during comparative studies by high resolution anion exchange chromatography that the inositol polyphosphates from E. histolytica were not eluted exactly with the same retention time as the myo-inositols used as reference. This casted some doubt on the previous identification. Therefore, we have reinvestigated the structures of the inositol phosphates present in E. histolytica by two-dimensional 31 P-1 H NMR analysis after HPLC purification, and we show in the present study that E. histolytica possesses high levels of several neo-inositol polyphosphates.

Culture of Cells and Preparation of Acellular Extracts-E. histolytica
HM-1:IMSS (ATCC 30459) were cultured and perchloric acid extracts prepared as detailed in Ref. 8.
Isolation and Analysis of Highly Phosphorylated Inositol Phosphates-The HPLC-MDD analysis for inositol phosphates and their further purification were performed as described (5,15).
NMR Analyses-NMR spectra were recorded at room temperature on a Bruker AMX400 WB spectrometer at 400-MHz frequency for 1 H, 162-MHz frequency for 31 P, and 100.6-MHz frequency for 13 C.
One-dimensional 1 H spectra of purified inositol phosphates from E. histolytica dissolved in 2 H 2 O were recorded in a 5-mm inverse probehead with an excitation pulse of 60°. Acquisition time was 2 s, and 32 scans were accumulated. The spectral width was set to 4032 Hz at a 8192 data size, which yields a digital resolution of 0.50 Hz/point. The 1 H H 2 O signal was suppressed by a selective 2-s presaturation pulse. 1 H chemical shifts are given relative to external tetramethylsilane at 0 ppm. For 31 P-decoupled spectra, a WALTZ-16 sequence is used.
Other NMR spectra were recorded in a 10-mm X-( 1 H) probehead. 1 H-decoupled 13 C spectra were recorded with an excitation pulse of 60°. * This work was supported by grants from the Commissariat à l'Energie Atomique (Département de Biologie Moléculaire et Structurale), the Centre National de la Recherche Scientifique (UMR 314), and the Université Joseph Fourier-Grenoble and by Deutsche Forschungsgemeinschaft Grants Vo 348/2-2 and SCHO315/7-2. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The spectral width was set to 25 kHz at a 32,768 data size, which yields a digital resolution of 0.76 Hz/point. Spectra were WALTZ-16 protondecoupled with two levels of decoupling: 1.5 W for 0.65 s (acquisition time) and 0.5 W for 0.35 s (relaxation delay). Free induction decays (68,400 scans for neo-InsP 6 and 25,200 scans for 2-PP-neo-InsP 5 ) were processed with 1.5 Hz exponential line broadening. 13 C chemical shifts are referenced to ␦ of external tetramethylsilane at 0 ppm.
One-dimensional 31 P spectra were recorded with an excitation pulse of 45°. The spectral width was set to 5880 Hz at a 16,384 data size, which yields a digital resolution of 0.35 Hz/point. For WALTZ-16 1 H decoupling, two levels of decoupling were used: 2.5 W during the acquisition time (1.4 s) and 0.5 W during the relaxation delay (0.6 s). Free induction decays were processed with 1 Hz exponential line broadening. 31 P chemical shifts are referenced to ␦ of external phosphoric acid at 0 ppm.
Two-dimensional 31 P-1 H heteronuclear shift correlation spectroscopy correlation spectra were recorded with the following pulse sequence (16): RD-90°( 1 H)-1/2t 1 -180°( 31 P)-1/2t 1 -⌬ 1 -90°( 1 H)90°( 31 P)-⌬ 2 -acquisition ( 31 P) with polarization transfer from 1 H to 31 P via J POCH . The time intervals ⌬ 1 and ⌬ 2 were set to 56 and 37 ms, respectively. The /2 pulse lengths were 23 s ( 31 P) and 24 s ( 1 H). Acquired spectra were WALTZ-16 1 H-decoupled, and 31 P acquisition time was 0.4 s with 0.6 s recycling delay. Spectral width was 2604 Hz in the F 2 domain (2048 points) and 400 Hz in the F 1 domain (32 points). Two-dimensional contour plot was calculated by Fourier transformation after zero filling to a 2048 ϫ 128 data matrix and applying a sine-bell function in the F 1 dimension and a Gaussian multiplication in the F 2 dimension. The one-dimensional 1 H and 31 P NMR spectra were the projections of the F 1 and F 2 dimensions.
Two-dimensional 31 P-1 H TOCSY-DEPT correlation spectra were acquired with the following pulse sequence (17) Quantification by NMR of E. histolytica Inositol Phosphate Content-Absolute amounts of inositol phosphates were calculated as described previously (8).
FIG. 2. Phosphorylation assay of inositol phosphates from E. histolytica. Compounds II and III isolated from E. histolytica were incubated for 4 h with 0.5 milliunit of 6-PP-InsP 5 5-kinase prepared from D. discoideum. A, compound II (2 nmol) was not phosphorylated by the kinase. B, 0.6 nmol myo-InsP 6 added to 2 nmol of compound II as a control for kinase activity was completely phosphorylated to 5-PP-myo-InsP 5 after 60 min. C, compound III was not phosphorylated by the kinase. The position of a minor hydrolysis product of compound III, with the same retention time as compound II, is labeled with an asterisk. D, 0.6 nmol 6-PP-myo-InsP 5 added to compound III as a control for kinase activity was completely phosphorylated to 5,6-bis-PP-myo-InsP 4 after 10 min.
A). Although the retention time of compound I was close to that of the various myo-inositol pentakisphosphates, it did not coincide with any of them. Compound II eluted with the same retention time as myo-InsP 6 , and compound III eluted slightly later than 6-PP-myo-InsP 5 purified from D. discoideum (Fig. 1,  trace B). This result suggested that the inositol phosphates present in E. histolytica might be different from the highly phosphorylated inositol phosphates presently known in D. discoideum. The intracellular amounts of the E. histolytica compounds I, II, and III were determined by 31 P NMR to be about 0.11, 0.18, and 0.04 mol/g wet cells, respectively. For further characterization, a perchloric acid extract was prepared from 18 g (wet weight) cells, and compounds I, II, and III were purified on an HPLC anion exchange column, with a 42-68% yield.
Inositol Phosphates Extracted from E. histolytica Are Not Phosphorylated by 6-PP-InsP 5 5-Kinase-The 6-PP-InsP 5 5-kinase extracted from D. discoideum specifically phosphorylates position 5 of 6-PP-myo-InsP 5 leading to 5,6-bis-PP-myo-InsP 4 (4). This enzyme also phosphorylates more slowly position 5 of myo-InsP 6 to form 5-PP-myo-InsP 5 . 2 Compound II, although co-eluting from the HPLC column with myo-InsP 6 , was not 2 T. Laussmann and G. Vogel, unpublished results. HPLC-purified compound I was adjusted to pH 6.0. Acquisition conditions are detailed under "Experimental Procedures." Spectral width was 320 Hz in the F 2 domain (256 points) and 1200 Hz in the F 1 domain (128 points), and the number of scans for an increment was 384. The twodimensional contour plot was calculated by Fourier transformation after zero filling to a 256 ϫ 256 data matrix and applying an exponential multiplication in F 1 and F 2 dimensions. The one-dimensional spectra correspond to the high resolution spectra and not to the projections of the two-dimensional map. The letters d and r refer to direct and relayed correlations, respectively.

neo-Inositol Phosphates in Entamoeba
phosphorylated by this enzyme (Fig. 2, trace A), whereas added myo-InsP 6 was completely converted into 5-PP-myo-InsP 5 (Fig.  2, trace B). Similarly, compound III, with an elution time from the HPLC column close to that of 6-PP-myo-InsP 5 , was not phosphorylated by the kinase (Fig. 2, trace C), whereas added 6-PP-myo-InsP 5 was converted into 5,6-bis-PP-myo-InsP 4 (Fig.  2, trace D). Compound I was no substrate for the kinase, either (data not shown). Taken together with the HPLC data, these results demonstrate that the highly phosphorylated inositol phosphates from E. histolytica were not identical to known myo-inositol phosphates. Compounds I, II, and III were further analyzed by NMR spectroscopy to solve their structures. Identification of Compound I as neo-InsP 6 -The one-dimensional 1 H NMR spectrum of compound I showed two resonance lines at about 4.4 and 4.9 ppm in an intensity ratio of 2:1, respectively. No other proton peak was detectable in this 1 H spectrum between 10 and 0 ppm, except for some residual water at 4.8 ppm (F 1 dimension, Fig. 3). Because the inositol ring has six ϪCH groups, this results indicates that compound I had just two groups of equivalent protons, one group consisting of four protons and the other group consisting of two protons. Similarly, in a 31 P NMR spectrum, compound I gave also two resonance lines at 0.8 and 0.6 ppm with the same intensity ratio of 2:1 or 4:2 (F 2 dimension, Fig. 3), suggesting an inositol molecule with either three or six phosphorus atoms. Only a 3-phosphate molecule is compatible with the symmetry of the myo-inositol ring. Assuming that myo-inositol phosphates were indeed present in E. histolytica, compound I was previously identified as myo-Ins (2,4,6)P 3 (8). The above data on its elution position from the HPLC column, the kinase assays, and the equivalence of the protons of the molecule suggested that this identification of compound I was wrong and that E. histolytica possessed an inositol isomer other than myo-inositol.
To determine the structure of compound I, we recorded a two-dimensional 31 P-1 H relayed, correlative NMR spectrum on the HPLC-purified compound I. The phosphate atoms at 0.8 ppm were correlated with the four protons at 4.42 ppm, whereas the phosphate atoms at 0.6 ppm were correlated with the two protons at 4.90 ppm. Relayed cross-peaks were also detected between the four phosphate atoms and the protons at 4.90 ppm and between the two phosphate atoms at 0.6 ppm and the protons at 4.42 ppm. Because all protons of the inositol ring are correlated with phosphates (F 1 dimension, Fig. 3), the six hydroxyl groups must be phosphorylated. The peculiar 31 P and 1 H NMR spectra of compound I with only two resonance groups in the ratio 4:2 were also found with 13 C NMR, with two lines (both of them broadened because of the multiple 13 C-31 P coupling constants) at 72.8 ppm (intensity 2) and 75.3 ppm (intensity 1) (data not shown). This unique order of symmetry is only compatible, among the nine isomeric inositols, with compound I being neo-InsP 6 . Chemical shifts and coupling constants are summarized in Table I Identification of Compound II as 2-PP-neo-InsP 5 -Compound II co-eluted from the HPLC column with myo-InsP 6 , after compound I identified just above as neo-InsP 6 (Fig. 3). Because the negative kinase assays eliminated the possibility of compound II being myo-InsP 6 (Fig. 2), we had a closer look at the structure of this component as well. Compound II had a 31 P-decoupled 1 H NMR spectrum with three singlets at 5.00, 4.93, and 4.44 ppm in the ratio 1:1:4 (Fig. 4A). All protons were coupled to a phosphorus atom with J POCH coupling constants around 10 -11 Hz for the two downfield protons and 7-8 Hz for the four upfield protons (Fig. 4B and Table I). Again, these values of chemical shifts and J POCH coupling constants are an indication for the presence of the two equatorial and four axial protons characteristic of the neo-inositol isomer. The 1 H-decoupled 31 P NMR spectrum of compound II exhibited three singlets at 0.85, 0.72, and 0.55 ppm and two doublets at Ϫ9.60 and Ϫ11.30 ppm in the ratio 2:2:1:1:1 (Fig. 4C and Table I), suggesting a molecule with seven phosphate atoms among which is a diphosphate group responsible for the peak splitting in the two upfield doublets (J POP ϭ 16.2 Hz). Its 1 H-coupled 31 P NMR spectrum showed a splitting of each line into two, except for the doublet corresponding to the ␤P of the diphosphate group (Fig. 4D). This is an indication that each of the phosphate groups, except the ␤P, is coupled with one proton only. The HPLC elution order, the symmetry of the protons, the number of phosphate atoms, and the two doublets at Ϫ9.60 and Ϫ11.30 ppm were suggestive of PP-neo-InsP 5 . The ␣-phosphate of the diphosphate group was correlated to an equatorial proton of the neo-inositol, which can be in position 2 or 5 (Fig. 5). Because these two positions are equivalent because of the symmetry of the molecule, compound II was named 2-PP-neo-InsP 5 , according to the "lowest locant rule" of the IUPAC-IUB (18, 19).

neo-Inositol Phosphates in Entamoeba
1 H-decoupled 31 P NMR spectrum (F 2 dimension, Fig. 6) of compound III exhibited a singlet at about 1 ppm and 2 doublets indicative of diphosphate bond(s) at Ϫ9 and Ϫ11 ppm in a 2:1:1 (or 4:2:2) ratio. This would be compatible with a total of either four or eight phosphate groups. The fact that compound III was eluted from the HPLC column after compound II identified above as a 2-PP-neo-InsP 5 indicates that it must contain eight phosphate groups, a number that can only be reached if two diphosphate groups are present in the molecule. In a 31 P-1 H correlation map, the singlet and the doublet corresponding to the ␣-phosphate of the diphosphate groups are correlated to two groups of protons at 4.4 and 5.0 ppm in the ratio 2:1 (4:2). No other protons were detectable in a one-dimensional 1 H NMR spectrum between 10 and 0 ppm, except for the signal of residual water at 4.8 ppm (F 1 dimension, Fig. 6). These results are an indication that the two positions carrying equatorial protons (positions 2 and 5) were bearing the diphosphate groups. Compound III was thus identified as 2,5-bis-PP-neo-InsP 4 .
Presence of neo-Inositol Polyphosphates in Phreatamoeba balamuthi-P. balamuthi is a free living amoeba that lacks mitochondria (20,21). It also contains unusual diphospho-inositol polyphosphates in large quantities (9). An HPLC-MDD analysis of the three inositol-containing compounds present in P. balamuthi indicated that they exactly co-eluted with neo-InsP 6 , 2-PP-neo-InsP 5 , and 2,5-bis-PP-neo-InsP 4 purified from E. histolytica (data not shown). We conclude from this that P. balamuthi probably contains the same set of neo-inositol metabolites.

DISCUSSION
Distribution of myo-versus neo-Inositols-Our data show that E. histolytica trophozoites contain phosphate and diphosphate esters of neo-inositol: neo-InsP 6 , 2-PP-neo-InsP 5 , and 2,5-bis-PP-neo-InsP 4 at cytosolic concentrations of 0.22, 0.36, and 0.08 mM, respectively. So far, the most abundant inositol isomer encountered in nature has been myo-inositol, whereas neo-inositol is only a relatively rare isomer. Neo-InsP 6 has first been recognized as a soil constituent (22). Mixtures of myo-, scyllo-, chiro-, and neo-inositol are indeed present in soil as their pentakisphosphate and hexakisphosphate esters (23). Lneo-Inositol 1-phosphate has been found in the brain, heart, testis, and spleen of rat in micromolar concentrations (24). Phosphorylated neo-inositol has been identified besides polyphosphates on the cell surface of the freshwater carnivorous Amoeba discoides (25).
In vitro experiments show that E. histolytica trophozoites are sensitive to micromolar concentrations of exogenously added myo-inositol phosphates. E. histolytica possesses specific and separate binding sites for myo-Ins (1,4,5)P 3 and myo-Ins (1,3,4,5)P 4 , and independent calcium stores are releasable by each of these myo-inositol phosphates (26,27). The presence of high concentrations (ϳ100 -400 M) of neo-inositol polyphosphate derivatives suggests that neo-inositol metabolism may be regulated independently from myo-inositol in this amoeba. Comparable cases are found in nature where the myo-isomer does not represent the major inositol in terms of concentration. High concentrations of scyllo-inositol have been found in human brain, breast tumors, and the skate Raja erinacea (28 -31), and chiro-inositol has been identified in bovine liver and mouse brain (32).
Biosynthesis of neo-Inositols-Two biosynthesis pathways by which neo-inositol might be formed have been proposed: 1) the cyclization of mannose 6-phosphate into neo-inositol 1-phosphate by L-myo-inositol 1-phosphate (D-myo-inositol 3-phosphate) synthase, the same enzyme that cyclizes glucose 6-phosphate into myo-inositol 1-phosphate (24) and 2) the epimerization of myo-inositol into neo-inositol at C-5 by a de- The NMR two-dimensional map of compound II, HPLC-purified as described in the legend to Fig. 1, was acquired as detailed under "Experimental Procedures." hydrogenation to the 5-keto compound (myo-inosose-5) followed by reduction of the carbonyl group (33). Both processes have been proven to proceed in bovine brain (24,34). In support of the first pathway, a gene encoding L-myo-inositol 1-phosphate synthase from E. histolytica has been cloned, and the native enzyme from trophozoites has been purified and characterized (35).
Function of Diphospho-neo-inositols-Among the neo-inositol phosphates found in E. histolytica, two bear diphosphate groups. Diphospho-myo-inositol phosphates have been found recently in Dictyostelids and a number of mammalian cell types (4 -7, 10, 11). The wide phylogenetic spectrum of distribution of these diphosphate cyclitols is taken as an indication for a fundamental physiological function. Diphosphoinositol phosphates have been suggested to be a new form of phosphate donor (2) or serve as intracellular second messengers (13,14). Alternatively, in vitro studies suggest a regulatory function of vesicle trafficking (36,37). A high turnover rate of diphosphomyo-inositol phosphates has been revealed, because treatment with fluoride, which inhibits the bisphosphatase involved in their breakdown, elevated their levels up to 10-fold at the expense of myo-inositol pentakisphosphate and myo-InsP 6 (7, 10). By contrast, treatment of E. histolytica with 10 mM NaF, although drastically reducing the levels of nucleotide diphophates and triphosphates, had no significant effect on the levels of neo-InsP 6 , 2-PP-neo-InsP 5 , and 2,5-bis-PP-neo-InsP 4 (data not shown). This suggests that the diphospho-neo-inositols are stable end products that are not involved in serving as energy stores or in regulating vesicle traffic. Whatever their function, it is clear that the neo-inositol phosphates found in E. histolytica and strongly suspected to be present in P. balamuthi represent yet another class of inositol metabolites and that nature exploits the full versatility of the inositol moiety.
FIG. 6. Two-dimensional 31 P-1 H TOCSY-DEPT spectrum of 2,5-bis-PPneo-InsP 4 . HPLC-purified compound III was adjusted to pH 6.0. Acquisition conditions are detailed under "Experimental Procedures." Spectral width was 2604 Hz in the F 2 domain (2048 points) and 400 Hz in the F 1 domain (64 points), and the number of scans for an increment was 1152. The two-dimensional contour plot was calculated by Fourier transformation after zero filling to a 2048 ϫ 128 data matrix and applying an exponential multiplication in F 1 and F 2 dimensions. The one-dimensional spectra correspond to the high resolution spectra. The letters d and r refer to direct and relayed correlations, respectively.