ANALYTICAL BIOCHEMISTRY l&$220-227 (1985)

Separation of lnositol Phosphates and Glycerophosphoinositol

Phosphates by High-Performance Liquid Chromatography’

HELGE BINDER, PETER C. WEBER, AND WOLFGANG SIESS~

Medizinische Klinik Innenstadt der Universitiit, Ziemssenstrasse I, 8000 Munich 2, Federal Republic of Gemany

Received December 12, 1984

We developed a HPLC method which separates the following nine inositol-containing

compounds of biological interest: inositol, inositol I-monophosphate, inositol 2- or 4-mono-
phosphate, inositol 1,2-cyclic phosphate, inositol 1 ,Cbisphosphate, inositol 1,4,5-trisphosphate,
glycerophosphoinositol, glycerophosphoinositol Cmonophosphate, and glycerophosphoinositol
4,5-bisphosphate. The method shows good resolution and sufficient recovery (70-80%) for each
compound. By applying this method to human platelets prelabeled with [‘Hlinositol and
stimulated with thrombin, we found an early increase of inositol 1,Cbisphosphate and inositol
1,4,5trisphosphate. Accumulation of glycerophosphoinositol, inositol I-monophosphate, and
an inositol monophosphate which cochromatographs with inositol 2- and inositol 4-mono-
phosphate occurs later. The method is simple, and-after removal of salts from the incubation
buffer-can be directly applied to the measurement of aqueous soluble [‘Hlinositol-labeled
compounds in biological samples. 0 1985 Academic Press, Inc.
KEY WORDS: inositol phosphates; HPLC; phospholipase C; platelet activation; thrombin.

The hydrolysis of membrane inositol phos- not sufficiently resolve certain inositol phos-
pholipids is an early biochemical response to phates from glycerophosphoinositol phos-
cell stimulation induced by a wide variety of phates. To achieve a better resolution of the
neurotransmitters, hormones, eicosanoids, various inositol-containing compounds of
and growth factors (l-8). The measurement biological interest, we developed a new
of inositol phosphates provides direct evi- method using HPLC which can easily be
dence for a phospholipase C-catalyzed reac- applied to the separation of aqueous soluble
tion and is the only tool for determining [3H]inositol-labeled substances in biological
which type of phosphoinositide is degraded samples.
by phospholipase C (9-l 1). The determina-
tion of glycerophosphoinositol and glycero- MATERIALS AND METHODS
phosphoinositol phosphates could provide Materials. L-myo[2-3H]inositol (IO-20 Ci/
information concerning if and which type of mmol) was obtained from NEN, and L-3-
phosphoinositide is deacylated by phospho- phosphatidyl[U-14C]inositol (225 mCi/mmol)
lipases and lysophospholipascs AI/A2 present and L-myo[U-14C]inositol I-monophosphate
in many cells. The methods currently used (55 mCi/mmol) were from Amersham. L-cu-
to separate inositol phosphates comprise an- phosphatidylinositol, L-a-phosphatidylinositol
ion-exchange chromatography on Dowex 4-monophosphate, L-cr-phosphatidylinositol
columns and paper chromatography ( 1 l- 15). 4,5-bisphosphate, phosphoinositides, Dowex
Both methods are able to separate inositol IX8 (chloride form, 100-200 mesh), Dowex
mono-, bis-, and trisphosphates, but they do HCR-W2, myoinositol 2-monophosphate,
phospholipase C from Bacillus cereus, and
’ This study was supported by the Deutsche For-
cyclohexylamine were all obtained from
schungsgemeinschah (DFG). Sigma. ITLC-SA sheets were from Gelman
’ To whom correspondence should be addressed. Sciences, Ann Arbor, Michigan, and chro-

0003-2697185 $3.00 220

Copyright 0 1985 by Academic Press, Inc.
All rights of reproduction in any form reserved.
 CHROMATOGRAPHIC SEPARATION
matography paper No. 1 was from Whatman.
Other chemicals were from Merck, Darm-
stadt, FRG.
Preparations of inositol phosphate and
glycerophosphoinositol phosphate standards.
14C-Labeled or unlabeled Insl,2-cyclic P,3
Insl-P, and Ins2-P were prepared by incu-
bating [i4C]PtdIns or PtdIns with phospho-
lipase C from platelet cytosol or from B.
cereus ( 16,17). The aqueous soluble Insl-P
and Insl,Zcyclic P were extracted, separated
on ITLC-SA sheets, and eluted with ethanol/
H20, 1:l (16,18). Insl-P and InQ-P were
obtained by heating Insl,2-cyclic P at 100°C
for 15 min in 0.1 N HCl or by hydrolysis of
PtdIns for 40 min at 100°C in 0.1 N HCl
( 19). Insl-P and Ins2-P were separated by
descending paper chromatography developed
in ethanol/l3.5 N NH40H [3:2, v/v; 18 h;
see Ref. (20)] or by HPLC. To remove the
salts of the HPLC eluates, the fractions con-
taining Insl-P and InQ-P were subsequently
passed through Dowex HCR-W2 columns.
InsCP, Ins 1,4-P2, and Ins 1,4,5-P3 were
prepared by the alkaline hydrolysis of phos-
phoinositides as described by Grado and
Ballou (15). Phosphoinositides (25-30 mg)
were dissolved in 5 ml of 2 N KOH, refluxed
for 30 min, and passed through a Dowex
HCR-W2 column. The eluate was extracted
three times with diethyl ether to remove free
fatty acids. Then cyclohexylamine was added,
and the eluate was rotoevaporated at 50°C
and purified by descending paper chromatog-
raphy for 30 h ( 14,15). This procedure leads
to a mixture of two isomers of each glycero-
phosphate, inositol monophosphate, -bis-
3 Abbreviations used (recommended by IUPAC-IUB):
Ins, inositol; PtdIns, phosphatidylinositol; PtdIn&P,
phosphatidyhnositol 4-monophosphate; PtdIns4,5-PZ,
phosphatidylinositol 4,5bisphosphate; Insl,2-cyclicP,
inositol 1,2-cyclic phosphate; Insl-P, inositol I-mono-
phosphate; Ins2-P, inositol 2-monophosphate; Ins4-
P, inositol 4-monophosphate; Insl,4-P2, inositol 1,4-
bisphosphate; Insl,4,5-P, , inositol 1,4,5-trisphosphate;
GroPIns, giycerophosphoinositol; GroPInsrl-P, glycero-
phosphoinositol 4-monophosphate; GroPIns4,5-P2, glyc-
erophosphoinositol 4,5-bisphosphate; Hepes, 4-(2-hy-
droxyethyl)-I-piperazineethanesulfonic acid; EGTA, eth-
ylene glycol bis(&aminoethyl ether) N,Wtetmacetic acid.
OF INOSITOL PHOSPHATES 221
phosphate, and -trisphosphate, which were
visualized by an ammonium molybdate re-
agent ( 16). The slow-running compounds of
the two isomers of inositol bisphosphate and
inositol trisphosphate are identified as Ins 1,4-
PZ and Ins1,4,5-P3 ( 15,2 1). The inositol
monophosphate band contains mainly InsC
P and less Insl-P (15). These compounds
were eluted with water which was then ro-
toevaporated at 50°C. Ins1 ,4-PI and Ins1,4,5-
P3 were also obtained by alkaline hydrolysis
of pure PtdIns4-P and PtdIns4,5-P2, respec-
tively. GroPIns was prepared by mild alkaline
hydrolysis of PtdIns (22). GroPInsCP and
GroPIns4,5-P2 were prepared from phos-
phoinositides as described (13) and purified
by descending paper chromatography (14).
High-performance liquid chromatography.
The HPLC equipment (Waters Assoc., Mil-
ford, Mass.) included a precolumn (Bondapak
AX/Corasil, 37-50 pm), a separation column
(PBondapak NH2, 30 X 0.39 cm), two sol-
vent-delivery pumps, and a gradient pro-
grammer. The new columns were conditioned
as follows:
They were washed with 10 ml of chloro-
form followed by 30 ml of methanol, 30 ml
of H20, and 30 ml of the initial HPLC
solvent at a flow rate of 1 ml/min. The
columns were tested by injection of a mixture
of radiolabeled GroPIns, Insl-P, and Ins2-P.
Some of the columns had to be equilibrated
longer (2-4 h) with the elution buffer, until
the inositol phosphate standards appeared.
The dry sample was dissolved in 200-300 ~1
of the initial HPLC solvent. Separation was
carried out utilizing a 20-min isocratic elu-
tion with a 60 or 75 mrvr (depending on
the column) ammonium acetate/acetic acid
buffer, pH 4.0, followed by a 120-min linear
gradient to 2 M ammonium acetate/acetic
acid, pH 4.0, at a flow rate of 1 ml/min.
Fractions were collected every 0.5 min (from
1 to 40 min), every 1 min (from 41 to 80
min), or every 2 min (from 8 1 to 120 min),
and they were measured for radioactivity
by liquid scintillation counting or for phos-
phorus (24).
222 BINDER, WEBER, AND SIESS
Measurement of inositol phosphates in
platelets prelabeled with [‘Hlinositol. The
labeling of platelets with [3H]inositol has
recently been described (24). After washing
the platelets to remove the free [3H]inositol,
the platelets were resuspended in a Tyrode-
Hepes-glucose buffer containing 1mM EGTA.
Samples (0.5 ml) of platelet suspensions (6
X 108/ml) were placed in aggregometer tubes,
LiCl ( 10 mM final concn) was added, and
platelets were stirred ( 1100 rpm) for 2 min
at 37°C before exposure to thrombin (2 U/
ml tinal concn). Shape change and aggrega-
tion were recorded. To separate the platelets
from the incubation buffer, the suspension
was filtered rapidly (6-8 s). We used Millipore
prefilters (AP 40), reduced them to a diameter
of 9 mm, and placed them into the bottom
of 2-ml plastic syringes which were fixed to
the top of a vacuum bottle. The syringe tip
containing the filter with the platelets was
immediately transferred into 2.5 ml of chlo-
roform/methanol (1:2) and 0.5 ml of H20.
The solution was vortexed, the syringe tip
was removed, and 0.83 ml of chloroform
and 1.0 ml of Hz0 were added. After vortex-
ing and centrifugation, the upper (aqueous)
phase was removed.
RESULTS
This study describes a HPLC method
which separates nine inositol-containing
compounds of biological importance (Fig. 1).
It resolves Insl,Zcyclic-P from GroPIns and
separates positional isomers of inositol
monophosphate (Fig. 1) and inositol bis-
phosphate, but not of inositol &phosphate
(unpublished observations). The Ins1,4,5-P3
peak is smeared, perhaps due to the high
negative charge of its phosphate groups which
seem to interact strongly with the NH2 col-
umn. Insl-P is separated from InQ-P and
Ins4-P, but InQ-P and Ins4-P coelute. Al-
though Insl,Zcyclic-P is an acid-labile com-
pound, the HPLC solvent of pH 4.0 leaves
Ins 1,Zcyclic-P intact as demonstrated by the
HPLC profiles shown before (Fig. 2a) and
after (Fig. 2b) chemical hydrolysis.
The HPLC shows sufficient recovery: For
each of the compounds tested, the recovery
after HPLC was between 70 and 80% (see
Table 1). The recovery was independent of
the amount of substance applied to HPLC:
radioactive standards applied in picogram
quantities were recovered at similar rates as
unlabeled standards injected in amounts of
20 to 200 pg (Table 1).
A certain elution time was always needed
for a new HPLC column to become optimally
equilibrated. We observed that with a new
column the retention times of glycerophos-
phoinositol and the inositol phosphate stan-
dards were longer and became progressively
shorter until they were fairly constant after
5 to 10 elutions (Compare also Figs. 3 and
5). The only major problems which we en-
countered were (i) a decrease of resolution
with longer use of the column, and (ii) a
sensitivity of the chromatographic separation
to salts in the sample (unpublished observa-
tions). We could substantially increase the
number of elutions per column by using
ammonium acetate instead of ammonium
formate as HPLC solvent, and by acidifying
the HPLC solvent by means of acetic acid
instead of hydrochloric acid. Using this pro-
cedure, we have used this column up to now
for approximately 100 runs. The salt sensi-
tivity of the chromatographic separation
especially concerns compounds eluted early,
such as GroPIns, Ins 1,2-cyclic-P, Ins 1-P, and
InQ-P (unpublished observations), and can
cause a problem if the method is applied to
the separation of [3H]inositol-labeled com-
pounds in biological samples (see below).
The major goal of our efforts was to apply
this method to the separation of inositol
phosphates in stimulated cells. Human plate-
lets were prelabeled with [3H]inositol and
stimulated with thrombin. Since we had ob-
served that inositol phosphates-similar to
1,2diacylglycerol and phosphatidic acid (5)-
are not released into the extracellular me-
dium, but remain inside the platelets, the
platelets were filtered to remove the salts
from the incubation buffer and then ex-
tracted. The aqueous soluble [‘Hlinositol
CHROMATOGRAPHIC SEPARATION OF INOSITOL PHOSPHATES 223
224 BINDER, WEBER, AND SIESS

TABLE 1
RECOVERYOFSTANDARD SUBSTANCESOF~NOSITOL
PHOSPHATESANDI~LYCEROPHOSPHOINOSITOL
a PHOSPHATES AFTER HPLC

Recovery
Substance (%I

[‘HlInositoI 85 + 1 (3)
[‘4C]Glycerophosphoinositol 80 + IO (7)
Glycerophosphoinositol 78 + 8 (5)
[‘4C]Inositol 1,2-cyclic phosphate 74 + 9 (4)
[“‘C]Inositol I-monophosphate 78 + 11 (3)
[‘4C]Inositol 2-monophosphate 80 + 9 (3)
Inositol 2-monophosphate 86 iz 11 (7)
Inositol 4-monophosphate 76 + 12 (3)
Glycerophosphoinositol 4-phosphate 87 (1)
Inositol 1,Cbisphosphate 78 zk 6 (3)
Inositol 1,4,5-trisphosphate 76 + 6 (3)

Note. Values are means + SD. Number of determi-

nations in parentheses.

in vitro by phospholipase C-induced hydro-

I 4 lysis of PtdIns, was detectable in small
0 30 60
RfTENIION TIMt. ml” amounts in unstimulated platelets and
showed a small increase 5 min after addition
FIG. 2. Behavior of inositol I ,2-cycIic phosphate before
(a) and after (b) acidic hydrolysis on HPLC. of thrombin (Figs. 4, 5).

phosphates could then be directly separated

by the described HPLC method without fur-
ther sample purification. To identify the
[3H]inositol-labeled compounds in the bio-
logical samples they were cochromatographed
with unlabeled authentic standards (Fig. 3).
Ten seconds after addition of thrombin we
found a rapid increase of Insl,CP;! and
Ins1,4,5,-PS, the increase of the latter com-
pound being transient (Fig. 4). At later time
points an enhanced formation of 3H-labeled
GroPIns, Insl-P, and an inositol monophos-
phate coeluting with InsZ-P and InsCP was
observed (Figs. 4, 5). Ins1,4-Pz appeared as FIG. 3. Separation of [‘Hlinositol phosphates formed
two peaks (peaks I and II, Fig. 4), if no in platelets stimulated by thrombin. Samples (0.5 ml) of
unlabeled Ins1 ,4-Pz standard was added. If human platelets (6 X 10s per ml) prelabeled with
the samples were cochromatographed with [‘Hlinositol (24) were stimulated with thrombin (2 U/
ml) for 5 min. Platelets were rapidly filtered to remove
unlabeled Ins 1,4-P2, only one radioactive buffer salts, and the aqueous soluble [‘Hlinositol-labeled
and phosphorus-containing peak for Insl,4- compounds (0 - 0) were extracted. Unlabeled standards
P2 appeared. The reason for this observation (0 --- 0) of GroPIns, Insl-P, InQ-P, and Insl,4,5-Pj
is unknown. Insl,2-cyclic-P, which is formed were added before seoaration of the samde bv HPLC.
 Ins 1-P Ins 4-P
andI or
Ins 2-P

Ins 1.L -P2

peak II
i
I

Gro P Ins

Insl2-cyclic P

a b c 0 b c (I b c a b c 0 b c

FIG. 4. Aqueous soluble [3H]inositol-labeled compounds in control platelets and in platelets stimulated by thrombin (2 U/ml). (a) Control; (b) thrombin for IO-15 s; g
(c) thrombin for 5 min. Values are means f SD from 3 to 25 assaysfrom two to five experiments.
 BINDER.
IrIsI-P
d -
”
I I
0 28
REltNllON IlMt. min
FIG. 5. Formation of glycerophosphoinositol and ino-
sit01 monophosphates in thrombin-stimulated platelets.
To samples of human platelets (0.5 ml) prelabeled with
[3H]inositol (24) was added either thrombin (2 U/ml
final concn) for 5 mitt (O-O) or saline (0 --- 0).
Platelets were then filtered, and the aqueous soluble
inositol phosphates were extracted and separated on a
freshly conditioned pBondapak NH2 column.
Since the platelets were separated from the
incubation medium by filtration before the
biological reaction was stopped, they could
be stimulated by that procedure. We ob-
served, however, no platelet activation as
long as EGTA (1 mM) and no divalent
cations (Ca*+, Mg*+) were present in the
incubation medium. If extracellular divalent
cations are needed for platelet activation, we
transferred directly small aliquots (0.1-0.2
ml) of the platelet suspension into chloro-
form/methanol without previous filtering of
the platelets. We observed that salts contained
in 0.1-0.2 ml of the platelet buffer did not
WEBER, AND SIESS
significantly influence the HPLC separation,
except for the resolution of GroPIns from
Ins 1,2-cyclic-P (unpublished observations).
DISCUSSION
The investigation of the inositol phospho-
lipid metabolism in stimulated cells is of
increasing interest [for recent reviews see Ref.
(3,4)]. Two substances derived from break-
down of inositol phospholipids in stimulated
cells are viewed currently as second messen-
gers: 1,2-diacylglycerol activates protein ki-
nase C, and Ins1 ,4,5-P3 mobilises Ca++ from
the endoplasmic reticulum in different cell
types [for review see Ref. (3,4)]. In gen-
eral, the hydrolysis of PtdIns, PtdInsCP, or
PtdIns4,5-P2 by phospholipase C in vitro
leads to Ins1 -P, Insl,2-cyclic-P, Ins 1,4-Pz,
and Insl,4,5-P,, respectively. The exact se-
quence of hydrolysis of those phosphoinosi-
tides following receptor activation of cells is
not yet known, although at present the deg-
radation of PtdIns4,5-P2 is regarded as the
receptor-linked step (2-4,lO). In addition
55
to the phospholipase C-induced cleavage,
the phosphoinositides could also be deacyl-
ated by phospholipases and lysophospholip-
ases AI/A2 to the corresponding glycero-
phosphoinositol compounds, i.e., GroPIns,
GroPInsCP, and GroPIns4,5-P2 respectively.
Up to now, it was not possible to separate
inositol phosphates from glycerophosphoi-
nositol phosphates in one chromatographic
step. For example, on Dowex anion-exchange
chromatography inositol bisphosphate coe-
lutes with GroPIns4-P, and inositol trisphos-
phate coelutes with GroPIns4,5-P2 (unpub-
lished observations). On descending paper
chromatography, which separates even iso-
mers of inositol mono-, bis-, and trisphos-
phate ( 14,15), Ins 1,4-P* corn&rates with
GroPIns4,5-P2, Ins2-P with GroPInsCP, and
Ins 1,Zcyclic-P with GroPIns, respectively
(unpublished observations). Therefore, the
described HPLC method is superior to the
widely used Dowex anion-exchange chro-
matography and in some aspects also to
descending paper chromatography.
 CHROMATOGRAPHIC SEPARATION
The method was applied to measure
[3H]inositol phosphates in thrombin-stimu-
lated human platelets which were prelabeled
with [‘Hlinositol. The [3H]inositol-containing
peaks were identified by cochromatography
with unlabeled standard compounds which
were measured by phosphorus determination.
One has to be aware, however, that the
structural identification of any substance
which is based solely on cochromatography
with authentic compounds is not absolute.
We observed that the addition of thrombin
to human platelets stimulated the formation
of Gro-P-Ins, Insl-P, InQ-P and/or InsCP,
Ins1 ,4-P2, and Ins1 ,4,5-P3. In addition, a
small increase of Insl,2-cyclic-P could be
detected. GroPIns could derive from deacyl-
ation of PtdIns by phospholipases A and
lysophospholipases A which may be activated
in platelets stimulated by high concentrations
of thrombin (2 U/ml). The biological signif-
icance of the enhanced formation of inositol
mono-, bis-, and trisphosphates is discussed
elsewhere (24,25). The method can also be
used to measure 32P-labeled inositol phos-
phates, if 32P-labeled nucleotides are removed
before HPLC (25).
In conclusion, this new HPLC method for
the separation of inositol phosphates and
glycerophosphoinositol phosphates is simple
and reproducible, and can easily be applied
to the measurement of 32P- or 3H-labeled
inositol phosphates in biological samples if
care is taken to remove or reduce salts orig-
inating from the incubation buffer. In addi-
tion, the method could also be applied to
measure levels of the various phosphoinosi-
tides after their chemical deacylation to the
corresponding GroPIns, GroPInsbP,
GroPIns4,5-P2.
ACKNOWLEDGMENT
We are grateful to lngeborg Drijssel for the careful
_DreDaration
. of the manuxrid.
OF INOSITOL PHOSPHATES 227
REFERENCES
1. Michell, R. H. (1975) Biochim. Biophys. Actu 415,
81-147.
2. Downes, C. P., and Michell, R. H. (1982) Cell
Calcium 3, 461-502.
3. Benidge, M. J. (1984) B&hem. J. 220, 345-360.
4. Berridge, M. J., and Irvine, R. F. (1984) Nature
(London) 312, 315-321.
5. Siess, W., Siegel, F. L., and Lapetina, E. G. (1983)
J. Biol. Chem. 258, 11236-l 1242.
6. Siess, W., Siegel, F. L., and Lapetina, E. G. (1984)
Biochim. Biophys. Acta 801, 265-276.
7. Siess, W., Biihlig, B., Weber, P. C., and Lapetina,
E. G. (1985) Blood, in press.
8. Habenicht, A. J. R., Glomset, J. A., King, W. C.,
Nist, C., Mitchell, C. D., and Ross, R. (1981) J.
Biol. Chem. 256, 12329-12335.
9. Akhtar, R. H., and Abdel-Latif, A. A. (1980)
Biochem. J. 192, 783-791.
10. Berridge, M. J. (1983) Biochem. J. 212, 849-858.
II. Berridge, M. J., Dawson, R. M. C., Downes, C. P.,
Heslop, S. P., and Irvine, R. F. (1983) Biochem.
J. 212, 473-482.
12. Richards, D. E., Irvine, R. F., and Dawson,
R. M. C. (1979) Biochem. J. 182, 599-606.
13. Downes, C. P., and Michell, R. H. (1981) Biochem.
J. 198, 133-140.
14. Desjobert, A., and Petek, F. (1956) Bull. SIC. Chim.
Biol. 38, 871-883.
15. Grado, C., and Ballou, C. E. (196 1) J. Biol. Chem.
236, 54-60.
16. Siess, W., and Lapetina, E. G. (1983) Biochim.
Biophys. Acta 752, 329-338.
17. Michell, R. H., and Allan, D. (1975) FEBS Lett.
53, 302-304.
18. Hokin-Neaverson, M., and Sadeghian, K. (1976) J.
Chromatogr. 120, 502-505.
19. Pizer, F. L., and Ballou, C. E. (1958) J. Amer.
Chem. Sot. 85,9 15-92 1.
20. Dawson, R. M. C., and Clarke, N. (1972) B&hem.
J. 127, 113-118.
21. Irvine, R. F., L&her, A. J., Lander, D. J., and
Downes, C. P. (1984) Biochem. J. 223, 237-243.
and 22. Billah, M. M., and Lapetina, E. G. (1982) J. Biol.
Chem. 257, 12705- 12708.
23. Bartlett, G. R. (1959) J. Biol. Chem. 243, 466-468.
24. Siess, W., and Binder, H. (1985) FEBS Lett. 180,
107-111.
25. Watson, S. P., McConnell, R. T., and Lapetina,
E. G. (1984) J. Biol. Chem. 259, 13199-13203.