(Received for publication, September 20, 1996, and in revised form, January 21, 1997)
From Clinique de Dermatologie, Hôpital Cantonal Universitaire, CH-1211 Genève 14, Switzerland and the § Ludwig Institute for Cancer Research, S-75124 Uppsala, Sweden
We show that unsaturated fatty acids (FAs) bind
reversibly and with high affinity to a heterocomplex of 34 kDa (FA-p34)
formed by the non-covalent association of two calcium-binding proteins of the S100 family: MRP8 (S100A8) and MRP14 (S100A9). Fatty
acid-competition studies on the [3H]oleic
acid·FA-p34-complex show that oleic, -linoleic,
-linolenic, and
arachidonic acids have IC50 values of about 1 µM, whereas palmitic and stearic acids are poor
competitors. The binding of arachidonic acid is saturable with a single
class of binding site per FA-p34, and a dissociation constant
(Kd) of 0.13 µM is calculated. The
individual subunits MRP8 and MRP14 show no binding properties for fatty
acids, whereas a p34 complex reconstituted in vitro by the
recombinant molecules exhibits binding properties, suggesting that the
fatty acid-binding site of FA-p34 is created through heterocomplex
formation. Furthermore, we demonstrate that lowering free
Ca2+ levels to 16 nM results in a loss of the
fatty acid-binding capacity of purified FA-p34. In calcium-induced
differentiating keratinocytes, the amounts of FA-p34 are increased in
the particulate (2.0 ± 0.5 pmol of [3H]oleic
acid/mg protein) and in the cytosolic (4.5 ± 0.6 pmol of
[3H]oleic acid/mg protein) fractions, whereas no FA-p34
can be detected in non-differentiated cultured keratinocytes.
In abnormally differentiated keratinocytes (psoriasis) and in human polymorphonuclear leukocytes, FA-p34 is highly expressed (31.35 ± 1.6 and 349.8 ± 17.9 pmol of [3H]oleic acid/mg protein, respectively), pointing toward a role for this heteromer in mediating effects of unsaturated fatty acids in a calcium-dependent way during cell differentiation and/or inflammation.
Fatty acids (FAs)1 are implicated in
energy delivery, membrane synthesis, in the lipid barrier function of
epidermis, in inflammation, and they modulate gene expression (1-3).
FAs are hydrophobic, labile molecules forming poor-soluble complexes
with intracellular calcium ions, which are required for keratinocyte
differentiation. Therefore, FAs need to be solubilized, stabilized, and
translocated by specific carrier proteins (4). Three distinct families
of lipid-binding proteins, i.e. extracellular albumin, the
cytoplasmic fatty acid-binding proteins (FABPs), and the peroxisome
proliferator-activated nuclear receptors (2, 3) are thought to mediate
the biological activities of FAs. Skin represents a very active
lipid-synthesizing tissue in mammals. Two recent reports describe that
human keratinocytes express the epidermal FABP, which is highly
up-regulated in the hyperproliferative and inflammatory skin disease
psoriasis (5-7). The epidermis lacks the 5- and
6-desaturases
(8), and, therefore, essential FAs like linoleic and arachidonic acid
must be acquired from circulation. E-FABP binds stearic, oleic, and
linoleic acid but has no affinity for arachidonic acid and a very low
affinity for the nearby precursor, linolenic acid (6). Since no other FABPs have been detected in keratinocytes so far, we investigated whether a carrier protein, capable to bind unsaturated fatty acids like
linolenic and arachidonic acid, might exist in psoriatic skin and
cultured human keratinocytes.
We characterized a fatty acid-binding heteromer of 34 kDa (FA-p34) isolated from human keratinocytes, which is composed of MRP8 (also referred to as cystic fibrosis antigen, the L1 light chain, p8, calgranulin A, S100A8) and MRP14 (L1 heavy chain, p14, calgranulin B, S100A9) (for a review, see Ref. 9). These proteins belong to the S100 family of Ca2+-binding proteins (reviewed in Ref. 10), and both molecules are highly up-regulated in psoriatic skin (5, 11-14). Furthermore, for both molecules several post-translational modifications and the formation of high molecular weight complexes in vivo have been reported (15-18). In this report we show that FA-p34 represents a novel class of FA-binding proteins and discuss its possible role in mediating the biological activities of unsaturated fatty acids.
[9,10-3H(N)]Oleic acid
(specific activity, 9.2 Ci/mmol),
[5,6,8,9,11,12,14,15-3H(N)]arachidonic acid
(100 Ci/mmol), and 45CaCl (21.9 mCi/mg) were purchased from
DuPont NEN (Regensdorf, Switzerland). Retinoic acid, -linoleic acid,
- and
-linolenic acid, palmitic acid, and stearic acid were all
obtained from Sigma.
Normal human
keratinocytes from foreskin were cultured in Dulbecco's modified
Eagle's medium (Life Technologies, Inc., Switzerland) containing 1.3 mM Ca2+ and 10% fetal calf serum (19); after
stratification, differentiating keratinocytes were separated from the
non-differentiated cells by the low Ca2+ switch method as
described (20). Human polymorphonuclear leukocytes (PMNL) from healthy
volonteers were isolated as described (21). Psoriatic scales were
obtained by gentle scraping of lesional skin from volunteer psoriatic
patients. Normal human skin was obtained using a keratome set at 180 µm on skin biopsies from patients that have undergone plastic
surgery. All cells and tissue samples were kept frozen at 20 °C
until use.
Keratinocytes (about 300 mg of lyophilized cells) were homogenized using a Polytron tissue
homogenizer in 1.5 ml of Tris buffer (50 mM Tris/HCl, 25 mM NaCl, 2.5 mM EDTA, 1 mM
dithiothreitol, pH 7.5) and centrifuged for 30 min at 100,000 × g. This supernatant is referred to as the cytosolic
fraction. The corresponding pellet was washed in 3 ml Tris buffer to
remove residual cytosolic proteins; the suspended pellet was
centrifuged at 5,000 × g for 10 min, and the
supernatant was discarded. This washing procedure was repeated twice.
The pellet constituted of cellular debris is called particulate
fraction. This particulate fraction was then treated in 1.5 ml of high
salt KCl buffer (10 mM Tris/HCl, 0.8 M KCl, 10 mM monothioglycerol, 10% glycerol, 1 mM
phenylmethylsulfonyl fluoride, 10 units/ml aprotinin, 10 µg/ml
leupeptin, pH 8.0) at 4 °C for 1 h before centrifugation at
20,000 × g. This procedure was repeated once, and the
supernatants were collected and concentrated using centrifugal
microconcentrators (Amicon) with a molecular mass cut-off of 3000 Da.
This preparation contains soluble proteins dissociated from the
particulate fraction by the high salt KCl buffer and is named high salt
extractable protein fraction, HSEPF. Cytosolic proteins and HSEPF were
aliquoted and frozen at 20 °C for storage.
About 5 g of psoriatic scales or normal skin were homogenized as above in 15 ml of Tris buffer and subsequently centrifuged at 100,000 × g for 1 h at 4 °C to obtain the cytosolic fraction. The corresponding pellet was washed in 5 ml Tris buffer to remove residual cytosolic proteins; the suspended pellet was centrifuged at 10,000 × g for 15 min, and the supernatant was discarded. This washing procedure was repeated twice. The pellet constituted of cellular debris is called particulate fraction. The pellet was then treated with 15 ml of high salt KCl buffer for 90 min and centrifuged at 10,000 × g for 15 min. This procedure was repeated once. The supernatants referred to as HSEPF were collected and subsequently concentrated and stored as described above.
About 2 × 109 PMNL were disrupted in 10 ml of Tris
buffer using a Polytron homogenizer. The cell debris were centrifuged
for 1 h at 100,000 × g and 4 °C, and the
supernatant, corresponding to the cytosolic protein fraction, was
aliquoted and stored at 20 °C. The pellet was washed in 2 ml of
phosphate-buffered saline (PBS) and centrifuged at 5,000 × g for 10 min; then, the supernatant was removed. The washing
procedure was repeated twice. The pellet was then resuspended in 5 ml
of high salt KCl buffer and incubated at 4 °C overnight. The
supernatant referred to as HSEPF was separated from insoluble material
by centrifugation at 12,000 × g for 10 min. The
procedure was repeated once, and the HSEPFs were pooled, aliquoted, and
stored at
20 °C. Protein concentrations were estimated by the
colorimetric method described in Ref. 22 using human serum albumin as
standard.
For keratinocyte proteins, [3H]oleic acid at a final concentration of 1.2 µM was deposited in glass microtubes, and 220 µg (in 100 µl PBS) of either cytosolic proteins or HSEPF were added. In the case of PMNL, 50 µg (in 100 µl of PBS) of cytosolic proteins or HSEPF were incubated with 7 µM [3H]oleic acid. In all analyses, tracer levels were used at saturating concentrations. These mixtures were incubated for 2 h at 22 °C before analysis. Competition studies were performed with a 200-fold molar excess of unlabeled oleic or retinoic acid. The samples were then subjected to gel filtration on Superose 12 (Pharmacia) connected to a high pressure liquid chromatography using PBS containing 0.2 M NaCl at pH 7.4 as eluant with a flow rate of 0.8 ml/min. After a delay of 8 min, fractions of 0.4 ml were collected for determination of bound radioactivity.
Similarly, [3H]oleic acid binding experiments were performed with 40 µg of rMRP8 and rMRP14 or with a mixture of both in a molar ratio of 2:1 in 200 µl of PBS.
Purification of FA-p34 from Psoriatic ScalesHSEPF from
psoriatic scales, dialyzed overnight against an imidazole buffer (20 mM imidazole/HCl, pH 6.0), was loaded (3 × 23.3 mg)
on a Resource S column (Pharmacia), which was equilibrated with the
same buffer. Elution was performed in a linear gradient using the
imidazole buffer containing 0.5 M NaCl. The column was first standardized with a small protein sample that was incubated with
[3H]oleic acid before being subjected to the column. One
major radioactive peak was eluted. For the preparative procedure,
protein fractions that co-eluted with the major radioactive peak,
obtained from standardization, were collected, concentrated, and
dialyzed against 20 mM Tris/HCl buffer, pH 8.0, before
subjected on a Resource Q column (Pharmacia) equilibrated with the same
buffer. Elution was performed in a linear gradient using the Tris
buffer containing 0.5 M NaCl. Again, the column was
standardized as above, and only the protein peak co-eluting with the
major radioactive peak was collected, dialyzed against PBS,
concentrated, and stored at 20 °C. FA-p34 was purified to
homogeneity on a Superose 12 column using similar conditions as
described for the gel filtration analysis. Purified protein, eluted at
34 kDa (FA-p34), was stored at
20 °C until further analysis.
About 30 µg of purified FA-p34 was subjected to SDS-PAGE (15%) under non-reducing conditions and stained with Coomassie Blue. Only two protein bands of approximately 8 and 14 kDa were detected. They were excised from the gel, transferred to Eppendorf tubes, and subjected to in-gel digestion according to Hellman et al. (23). Generated peptides were isolated by reversed phase liquid chromatography using the SMART System (Pharmacia Biotech, Uppsala, Sweden). Peptides were sequenced on an Applied Biosystems model 470A following the manufacturer's instruction.
Saturation KineticsSaturation assays were performed in 0.5% gelatin-Tris buffer without EDTA by adding [3H]arachidonic acid in increasing concentrations (0-5.5 µM) to 0.5 µg of FA-p34 in the presence (nonspecific binding) or absence (total binding) of a 200-fold molar excess of unlabeled arachidonic acid. Bound radioligand was separated from free ligand using the charcoal-dextran technique (24). Calculations for the saturation kinetics and Scatchard-plot analysis were performed, and the apparent Kd was calculated as described previously (25).
Competitive Binding Studies (IC50)Aliquots of 0.5 µg of FA-p34 were incubated with 500 nM [3H]oleic acid in gelatin-Tris buffer; then, increasing amounts of unlabeled FAs (concentrations ranged from 0.01 to 100 µM) were added to compete with [3H]oleic acid. Bound radioligand was separated from free ligand by the charcoal-dextran technique (24). Calcium dependence of [3H]oleic acid binding to FA-p34 was analyzed using increasing concentrations of EDTA (0-5 mM) in the same conditions as described above. The Ca2+ levels in the gelatin-Tris buffer were measured by atomic absorption spectrophotometry.
ImmunoblottingCytosolic and HSEPF proteins from psoriatic
scales, differentiating and non-differentiated cultured human
keratinocytes, and PMNL were separated by SDS-PAGE (15%) and
subsequently blotted onto a nitrocellulose membrane (Electran, BDH
Laboratory Supplies, United Kingdom). Membranes were incubated in PBS
containing 0.5% skimmed dry milk, 0.2% Tween 20. mAbs directed
against MRP8 and MRP14 (8-5C2 and S 36.48, respectively, purchased
from Biomedicals AG (Augst, Switzerland)) were used at a 1:50 dilution.
The immunoreactive bands were visualized using a peroxidase-labeled
goat anti-mouse IgG (Sigma) and 3,3-diaminobenzidine dihydrochloride
(Sigma) and H2O2 as substrates.
Aliquots of 9 µg of FA-p34 and 4 µg of recombinant MRP8 and MRP14 were separated by SDS-PAGE (15%), blotted onto a polyvinylidene difluoride membrane (Millipore) before probing with 2 µCi/ml of 45CaCl for 20 min in 10 mM imidazole, 5 mM MgCl, 50 mM KCl at pH 7.8 as described previously (26).
Protein extracts from various samples were analyzed for
proteins with fatty acid-binding capacity using gel filtration-high pressure liquid chromatography and [3H]oleic acid as a
ligand. The radioactive elution profile of the cytosolic fraction from
differentiating keratinocytes showed one large radioactive peak at 34 kDa (referred to as FA-p34), which virtually abolished with the
addition of an excess of unlabeled ligand (Fig.
1A), demonstrating high binding specificity.
The excess of unlabeled oleic acid revealed another radioactive peak at
15 kDa (Fig. 1A). This peak corresponded to
[3H]oleic acid bound to E-FABP, since it co-eluted with
purified human E-FABP. The appearance of the E-FABP peak upon gel
filtration can be explained by the finding that E-FABP has a higher
Kd value of 2.5 µM (measured with
dextran-coated charcoal) than the value described earlier using another
technique (20). Thus, a higher concentration of oleic acid was needed
to saturate E-FABP binding sites. The other eluted radioactive peaks
were either excluded (Vo) or included (free
excess [3H]oleic acid) from the gel matrix.
Analysis of HSEPF of differentiating keratinocytes showed, besides minor radioactive peaks, a large radioactive peak co-eluting with cytosolic FA-p34 (Fig. 1B). No E-FABP could be detected in the HSEPF, confirming that FABPs are essentially cytosolic. The binding was specific since a molar excess of unlabeled oleic acid almost abolished the radioactive peak, whereas a molar excess of retinoic acid had no effect on [3H] oleic acid binding to FA-p34. A UV trace experiment at 280 nm from the eluted material of cytosolic proteins and HSEPF from differentiating keratinocytes is shown Fig. 1E. As lesional psoriatic skin contains high levels of E-FABP, revealing high FA-traffic (6), the presence of FA-p34 was investigated in this tissue. HSEPF of scales showed a radioactive peak at 34 kDa, which was specific since it was almost abolished by a molar excess of unlabeled oleic acid (Fig. 1C). The radioactive elution profile of normal skin (cytosol and HSEPF) and psoriatic skin (cytosolic fraction) showed no and weak levels, respectively, of FA-p34 (data not shown). All elution times and ligand specificities observed for the radioactive peak at 34 kDa were identical for all samples investigated, suggesting that these [3H]oleic acid-binding proteins were identical FA-p34 species. To quantitate the binding capacity of FA-p34 from various samples, the radioactive peak of the [3H]OA·FA-p34 complex was integrated. The amounts of FA-p34 from various samples are summarized in Table I.
|
As HSEPF of psoriatic scales contains about 5 times more FA-p34 than the cytosolic fraction of differentiating keratinocytes, scales were used for the purification of FA-p34. Three purification steps were necessary to obtain a homogenous protein peak of FA-p34 when analyzed by gel filtration chromatography (Fig. 1D). About 389 µg of FA-p34 was obtained representing a yield of 0.0078% of starting material. Purified FA-p34 conserved its binding property and specificity during the different purification steps, since the FA-p34 peak co-eluted on Superose with the radioactive peak of [3H]OA·FA-p34 and was abolished by the addition of a molar excess of oleic acid.
Analysis of FA-p34 by SDS-PAGE and Partial Amino Acid SequencingAnalysis of FA-p34 by SDS-PAGE (15%) under
non-reducing conditions revealed the presence of two
Coomassie-stained protein bands of about 14 and 8 kDa (Fig.
2A, lane 1), suggesting that FA-p34 is a heterocomplex consisting of two non-covalently associated proteins. To unravel the identity of the two subunits, both proteins were digested with trypsine. Proteolysis yielded 7 peptides for the
8-kDa subunit and 15 for the 14-kDa subunit. One peptide of each
digested subunit was randomly selected and sequenced. The sequences
were GNFHAVYRD for the peptide obtained from the 8-kDa subunit and
LTWASHEK for the peptide from the 14-kDa subunit. By sequence
comparison with the published sequences, the 8-kDa subunit was
identified as MRP8 and the 14-kDa subunit as MRP14 (27). In addition,
SDS-PAGE analysis of rMRP8 and rMRP14 revealed an identical mobility
with the subunits of FA-p34 (Fig. 2A, lanes 2 and
3), suggesting that the rMRPs are very similar to the MRPs composing FA-p34.
Ca2+ Binding Studies of the FA-p34 Subunits
Since MRP8 and MRP14 are two calcium-binding proteins, we investigated whether the individual components of FA-p34 are able to bind Ca2+. Using the overlay technique (26), the direct autoradiography showed that 45Ca2+ bound to MRP8 and MRP14 separated from the purified FA-p34 complex by SDS-PAGE (Fig. 2B, lane 1) as did the recombinant proteins (lanes 2 and 3). The strong radioactive bands of native and recombinant MRP14 suggest higher calcium-binding capacity of MRP14 compared with MRP8.
Expression Studies of MRP8 and MRP14By protein-blot analysis
using mAbs directed against MRP8 (Fig. 3A)
and MRP14 (Fig. 3B), we studied the expression of these proteins in cytosolic fractions and HSEPF from the various samples. In
normal human skin (lanes 1 and 2) MRPs were not
detectable. In non-differentiated keratinocytes (lanes 3 and
4), only MRP8 was detectable. In contrast, high amounts of
MRPs were found in the cytosol and in HSEPF of differentiating
keratinocytes (lanes 5 and 6), psoriatic scales
(lanes 7 and 8), partially purified FA-p34 from
PMNL (fraction 16 from Fig. 8) (lane 9), and
purified FA-p34 from psoriatic scales (lane 10). The
immunoreactive bands of the rMRP8 and rMRP14 used as standards
(lanes 11 and 12) showed identical
electrophoretic mobilities as the MRPs from the samples.
Ligand Binding Studies of Purified FA-p34
A saturation curve
at the equilibrium was obtained for purified FA-p34 using increasing
amounts of [3H]arachidonic acid (Fig.
4A). The straight line of the Scatchard plot
indicates a single class of binding site for arachidonic acid with a
Kd of 0.13 µM (representative value of
two independent experiments) (Fig. 4B). The calculated
number of binding sites per FA-p34 was about 0.3. This low value might
be explained by (i) an overestimation of the protein concentration
measured by the colorimetric method used, compared with the intrinsic
concentration of FA-p34, and (ii) the presence of FA-p34 isoforms
without fatty acid-binding properties (see below). Competition binding
assays on [3H]OA·FA-p34 showed that palmitic acid and
stearic acid had poor competitive binding affinity versus
[3H]oleic acid bound to FA-p34 (Fig. 5),
whereas -linoleic acid,
-linolenic acid, and arachidonic acid
were good competitors with an IC50 of about 1 µM.
To study the role of free Ca2+ concentrations in
[3H]oleic acid binding capacity,
[3H]OA·FA-p34 levels were analyzed in the presence of
increasing amounts of EDTA. For each EDTA concentration, the
corresponding free Ca2+ concentration was calculated (using
a total and constant Ca2+ level of 150 µM)
and plotted versus the amounts of bound [3H]OA
(Fig. 6). The [3H] oleic acid binding
capacity of FA-p34 showed a plateau for values greater than 100 nM of free Ca2+, and about 70% of binding
capacity was lost at a free Ca2+ concentration of 10 nM. An IC50 value of about 18 nM
free Ca2+ was calculated.
Reconstitution of FA-p34 from rMRP8 and rMRP14
rMRP8 and
rMRP14 were tested for [3H]oleic acid binding capacity on
Superose column using the same elution conditions as for FA-p34
analysis. As seen in Fig. 7A, the rMRP8
protein appears as two protein peaks eluting near 34 kDa, probably
representing the dimer and tetramer. rMRP14 appears as a single peak at
about 34 kDa, which might correspond to the dimer. The absence of
monomers eluting at their respective molecular mass is explained by the fact that S100 proteins easily form homomers (13, 15, 28). Due to the
low resolution capacity of the gel filtration technique and the
physical properties (lipophilicity) of S100 molecules, the indicated
molecular weights are only indicative. When rMRPs were incubated with
labeled oleic acid, no radioactive protein peak was observed for rMRP8
and a weak peak for rMRP14 (Fig. 7B). The protein elution
profile of a mixture containing a 2 molar excess of rMRP8 over rMRP14
showed a broad and high radioactive peak with shoulders. The broadness
of this peak and the shoulders compared with the thin and symmetric
radioactive peak of FA-p34 suggest that several heterocomplexes,
including FA-p34, were formed. Ligand binding was specific since a
molar excess of unlabeled oleic acid almost abolished the peak.
However, the oleic acid-binding capacity of this peak (50 µg) was
lower compared with the 35 µg of purified FA-p34, analyzed under the
same conditions (Fig. 1D), suggesting that reconstitution of
the complex was only partial.
Measurement of FA-p34 from Polymorphonuclear Leukocytes
When cytosolic proteins of isolated human PMNL were analyzed by gel filtration using labeled oleic acid, a large radioactive peak of 34 kDa was detected. This peak was specific since it was almost abolished by an excess of unlabeled tracer (Fig. 8). This peak, composed essentially of MRPs, as analyzed by protein blotting (Fig. 3, A and B, lane 9), corresponded to FA-p34 from keratinocytes.
In this report, we describe the
purification and characterization of a heterocomplex of 34 kDa (FA-p34)
from human keratinocytes, formed by the non-covalent association of the
two well known calcium-binding proteins MRP8 and MRP14. FA-p34 is
capable of specifically binding unsaturated fatty acids with high
affinity and differs structurally from common FABPs of 15 kDa (reviewed
in Ref. 4). Arachidonic acid binds to FA-p34 in a saturating and
reversible manner to a single class of binding sites with a calculated
Kd of 0.13 µM, a value in the range
described for FABPs (4). Fatty acid-competition studies on the
[3H]OA·FA-p34 complex show that oleic, -linoleic,
-linolenic, and arachidonic acids have similar IC50
values of 1 µM, whereas palmitic and stearic acid, both
saturated fatty acids implicated mainly in energy delivery and
structural functions, or retinoic acid are poor competitors.
Interestingly, the binding capacity of FA-p34 was found to be dependent
on the free Ca2+ concentration. About 100 nM of
free Ca2+ were necessary to obtain full fatty acid-binding
capacity, whereas at 18 nM the binding property diminished
about 50%. Such a range in Ca2+ concentrations is
currently thought to be physiological. Taking in account that
keratinocytes do not express detectable amounts of other FABPs than the
epidermal type (29, 30), which has no or low affinity for arachidonic
and
-linolenic acid (6), our data suggest that these fatty acids
might be transported in differentiating keratinocytes in a
calcium-dependent way by the highly expressed FA-p34.
Whether more specific, yet unknown carriers for unsaturated fatty acids
might co-exist in keratinocytes remains to be determined.
Partial amino acid sequencing and protein blotting with specific mAbs revealed that FA-p34 is composed of the subunits MRP8 and MRP14. However, the stoichiometry of the subunits in FA-p34 remains to be determined. Recently, the migration inhibitory factor-related proteins (reviewed in Ref. 9) MRP8 and MRP14 have been isolated and molecularly cloned from human neutrophils (27). Both proteins are members of the S100 family of proteins that contain two calcium-binding domains (reviewed in Refs. 10 and 28). MRP8 and MRP14 are found predominantly as a non-covalently associated hetero- or homodimer, and higher molecular weight forms have been detected (15, 18). Whether FA-p34 is identical to a heterodimer MRP8/MRP14 of 35 kDa described earlier (15) remains to be determined. Preliminary data indicate that several isoforms of p34 exist but only some of which are able to bind fatty acids. In this context it should be mentioned that due to alternative translation initiation sites, two forms of MRP14 co-exist in cells, both of which can be phosphorylated (15, 31). Recently, it has been shown that MRP8 can also be phosphorylated (17). Several other post-translational modifications of the murine MRP14 have also been described (16). How these modifications influence complex formation of FA-p34 and its FA-binding properties is currently under investigation. MRP8 and MRP14 were previously studied by immunohistochemistry in human normal and pathological skin (11-14); however, these investigations did not concern FA-p34.
Fatty Acid Binding Site of FA-p34The fact that fatty acids form poor soluble salts with calcium ions (soap) suggests that the binding of fatty acid is coordinated by one of the calcium ions of the MRPs. This is unlikely since (i) retinoic, palmitic, and stearic acid do not exhibit similar binding capacities as arachidonic acid and its precursors and (ii) unsaturated fatty acids bind in a reversible manner to FA-p34. We hypothesize that the fatty acid-binding site of FA-p34 is formed by MRP complex formation. This idea is supported by the observations that individual recombinant MRPs showed no significant binding affinity for [3H]oleic acid, whereas mixed together they form complexes, including FA-p34, with fatty acid-binding properties. S100 proteins contain two hydrophobic and two ionic domains, which by complementary affinity domain association allow the formation of homo- or heteromers (10). We hypothesize that the specific juxtaposition of the hydrophobic domains of MRP8 and MRP14 in a yet undefined stoichiometry allows the formation of a fatty acid-binding site.
Postulated Functions of FA-p34FA-p34 levels are increased in keratinocytes that were induced to differentiate by extracellular calcium and in psoriatic skin, which displays higher than normal calcium concentrations and an alteration of the normal calcium gradient that programs keratinocytes' terminal differentiation (32). These findings reinforce our in vitro observations that increased Ca2+ concentrations preserve fatty acid-binding properties of FA-p34. Moreover, the high FA-p34 levels found in psoriasic skin might also correlate with high metabolism of unsaturated fatty acids in this disease compared with normal skin (33).
FA-p34 is not solely a cytosolic complex, and its presence in particulate fractions, from where it can be released by high salt buffers, suggests that FA-p34 is associated with membrane components. This is especially the case in psoriatic scales, were almost 90% of total FA-p34 is membrane associated.
Although no definite function has been assigned to MRP8 and MRP14, their expression in myeloid cells (up to 45% of total cytosolic protein of neutrophils) (34-36) as well as in epithelial cells of inflammatory skin (11-14) has suggested a role for MRPs in inflammation and differentiation (37). Preliminary data show that high levels of FA-p34 are also found in PMNL, suggesting that this complex is not specific for keratinocytes. Since unsaturated fatty acids play an important role in keratinocyte differentiation and represent precursors of inflammation (38), our findings make FA-p34 a good candidate for mediating effects of unsaturated fatty acids in a calcium-dependent way.
Recombinant MRP8 and MRP14 were a generous gift from Dr. A. Suter (Novartis, Basel). Drs. J. A. Cox and M. Rossier are gratefully acknowledged for valuable discussions and help in Ca2+ level measurements and calculations. Dr. R. W. James is thanked for reading the manuscript. We also thank Dr. R. Schmidt for helpful discussions.