Calcium-dependent complex assembly of the myeloic differentiation proteins MRP-8 and MRP-14.

MRP-8 and MRP-14 are calcium-binding proteins belonging to the S-100 protein family which have been shown to be associated with specific stages of myeloic/monocytic cell differentiation. Members of this protein family are shown to form homo- and heterodimers. Complex formation has also been observed in preliminary experiments for MRP-8 and MRP-14. To evaluate the in vivo relevance of the MRP complex formation and the stoichiometric ratio of individual components complexes were isolated from granulocytes and monocytes by immunoaffinity chromatography using monospecific antibodies. The purified fraction of the MRPs was found to contain monomers and dimers as shown on sodium dodecyl sulfate-polyacrylamide gel electrophoresis by silver staining and immunoblotting. Similar results were obtained by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and immunoblotting of crude cell extracts. The existence of the MRP complexes in vivo was demonstrated by chemical cross-linking and subsequent isolation of complexes by immunoaffinity chromatography. Two new, highly abundant complexes were found in addition to the heterodimer, but neither monomers nor homodimers were detected. The two larger protein complexes (35.0 and 48.5 kDa) were identified as [MRP-8)2.(MRP-14] trimer and [MRP-8)2.(MRP-14)2) tetramer, respectively. All complexes could be shown to be noncovalently associated in vivo. Furthermore, the association of MRPs was shown to be Ca2+ dependent.


Calcium-dependent Complex Assembly of the Myeloic Differentiation Proteins MRP-8 and MRP-14"
(Received for publication, February 21,1991) Stefan Teigelkamp, Ranjit S . Bhardwaj MRP-8 and MRP-14 are calcium-binding proteins belonging to the S-100 protein family which have been shown to be associated with specific stages of myeloic/ monocytic cell differentiation. Members of this protein family are shown to form homo-and heterodimers. Complex formation has also been observed in preliminary experiments for MRP-8 and MRP-14. To evaluate the in vivo relevance of the MRP complex formation and the stoichiometric ratio of individual components complexes were isolated from granulocytes and monocytes by immunoaffinity chromatography using monospecific antibodies. The purified fraction of the MRPs was found to contain monomers and dimers as shown on sodium dodecyl sulfate-polyacrylamide gel electrophoresis by silver staining and immunoblotting. Similar results were obtained by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and immunoblotting of crude cell extracts. The existence of the MRP complexes in vivo was demonstrated by chemical cross-linking and subsequent isolation of complexes by immunoaffinity chromatography. Two new, highly abundant complexes were found in addition to the heterodimer, but neither monomers nor homodimers were detected. The two larger protein complexes (35.0 and 48.5 kDa) were identified as ((MRP-S)z*(MRP-14)) trimer and ((MRP-8)2*(MRP-14)2) tetramer, respectively. All complexes could be shown to be noncovalently associated in vivo. Furthermore, the association of MRPs was shown to be Ca2+ dependent.
The two proteins MRP-8 and MRP-14 have recently been isolated and molecularly cloned (1,2). It was shown that the MRP-14 was largely identical with the so-called cystic fibrosis antigen (3, 4), and both proteins were identical with the socalled light and heavy chain of the L1 antigen which has been found in granulocytes (5). The expression of the proteins is not confined to granulocytes. They are also expressed in monocytes but disappear in mature forms of macrophages (6). Although the proteins are generally expressed by infiltrating monocytes/macrophages in chronic inflammatory reactions such as rheumatoid arthritis or sarcoidosis, they seem to be expressed separately by infiltrating macrophages in acute inflammations (1,6,7). The expression of the two proteins, * This work was supported by grants from the Bundesministerium fur Forschung und Technologie. 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.
A computer-assisted analysis of the primary structure of the proteins revealed two EF-hands indicating their calciumbinding properties (1). These structural characteristics assign the molecules to the S-100 protein family, a group of low molecular weight calcium-binding proteins, which include S-100 (Y and fi proteins, calcyclin (2A9), intestinal calciumbinding proteins and p l l (9,10).
Although no definite function could be assigned for this protein family there is evidence that they are involved in cell cycle progression, cell differentiation, cytoskeletal membrane interaction, and phosphorylation events (9,lO). Recent studies have shown an inhibitory activity of a complexed form of MRP-8 and MRP-14 for casein kinases I and I1 (11).
From the cited work and our own data it seems obvious that the two molecules generally are coexpressed by the same cell, and a great part of the functions may be exerted by complexes of the two molecules. However, the composition of the complexes including participation of other molecules in complex formation has not been studied thoroughly. In addition it is unclear whether different MRP-8.MRP-14 complexes exist in vivo and what function they exert with respect to different phenotypic functions of e.g. the respective macrophages.
In the present paper we demonstrate three distinct complexes and their stochiometry as the predominant form of intracellular MRP-8 and MRP-14. No third component was detected. The complexes were noncovalently associated in dependence of the calcium concentration.
Purification of the MRP-8 and MRP-14 Protein Complexes-Purification of native MRPs was achieved by immuoaffinity chromatography, using monospecific, affinity chromatography-purified rabbit antisera raised against recombinant MRP-8 as well as recombinant MRP-14 proteins. Antisera were coupled to BrCN-activated Sepharose, according to the protocol of the manufacturer (Pharmacia).
Supernatants from crude cell extracts, prepared as described above, were loaded onto the immunoaffinity column, washed with TBS, and eluted with 0.1 M glycine HCl, pH 2.5. Fractions containing protein were desalted immediately using a Sephadex G-25 column (PD-10, Pharmacia) equilibrated with 50 mM ammonium bicarbonate and subsequent lyophilization. Samples were diluted in PBS or TBS and stored at -80 "C.
Cross-linking of MRPs-The cross-linking of MRP complexes was carried out by using BS' (bis-(sulfosuccimidmidy1)suberate) or DTSSP (3,3'-dithiobis(sulfosuccinimidylpropionate)) (Pierce Chemical Co.), according to Staros (14). Briefly, affinity-purified MRPs (see above) were diluted in PBS, pH 7.4, to a concentration of 0.5, 0.25, and 0.125 mg/ml. Crude extracts of granulocytes and monocytes were prepared in the same concentrations as described above. A 20 mM stock solution of cross-linkers was prepared shortly before use. Protein samples were treated with various concentrations of crosslinkers (0.05-4.5 mM) for 30 min a t room temperature. Subsequently, the reactions were quenched by addition of 0.166 volume of 50 mM ethanolamine, 20 mM N-ethylmaleimide, and 50 mM sodium phosphate, pH 7.4, and then stored a t -80 "C. Samples cross-linked with HSI were incubated with 1% B-mercaptoethanol before separation on SDS-gel electrophoresis.
Silver Staining of SDS-PAGE-Affinity-purified MRP complexes, separated on SDS-PAGE, were silver stained by the method of Ansorge (method A) (17).
A modified silver staining (method B) for subsequent densitometric quantitation of the protein bands was carried out as follows. After fixation and washing with H 2 0 gels were rinsed in a 0.15% solution of AgNOs for 30 min, washed in H20 for 25 s, and developed in 2.5% Na2CO:,, 0.01% formaldehyde. The reaction was stopped by incubation in 50 mM EDTA. The gels were slightly destained with 0.2% Farmer's reagent and then washed extensively with H20. Quantification of protein bands was carried out on a laser densitometer (Ultra Scan XL, Pharmacia).
Additional Methods-Mass spectroscopy of MRPs was carried out by matrix-assisted laser desorption/ionization (LDI) (18,19). Samples were diluted to a concentration of 0.1 mg/ml in H20 and then mixed with a solution of 50 mM nicotinic acid in a 1:l ratio, dried onto a metallic substrate, and transferred to the vacuum chamber of a time-of-flight mass spectrometer.
Enzyme-linked immunosorbent assay was performed according to Briiggen and Sorg (20).
For alkylation of free thiol groups supernatants of cell lysates were prepared as described above in PBS containing either 10 or 40 mM iodoacetamide (21). After a 45-min incubation a t room temperature, unbound iodoacetamide was removed by Sephadex G-25 gel filtration (PD-10, Pharmacia) and prepared for SDS-PAGE as usual.

RESULTS
Purification and Characterization of MRP-8 and MRP-14 from Monocytes and Granulocytes-As a source for MRP-8 and MRP-14 purification, monocytes and granulocytes were isolated from human buffy coats. Cells were disrupted by nitrogen cavitation and centrifuged. The supernatants thus obtained from granulocytes and monocytes contained 4-6 mg/ ml total protein (Bradford method). The content of MRPs differed as determined by enzyme-linked immunosorbent assay and was 5-10% of the total protein content of granulocyte lysate supernatants; levels of MRPs in monocyte lysate supernatants were found to be 20-50 times lower. This supernatant was loaded on an affinity column with monospecific anti-rMRP-14 antibodies. The eluate was analyzed for the presence of MRPs by mass spectroscopy and silver staining after SDS-PAGE was carried out under reducing and nonreducing conditions (Fig. 1, A and B). Under reducing conditions only two bands were found representing MRP-8 and MRP-14 monomers. Nonreducing conditions revealed dimeric forms, indicating that cysteine linkage is responsible for their existence (Fig. 1B). Western blot analysis with monospecific antisera showed that the band at 22.0 kDa is an MRP-8 homodimer; at 24.5 kDa, an MRP-8-MRP-14 heterodimer; and at 26.5 kDa, an MRP-14 homodimer (Fig. IC).
To determine the molecular weight of MRP-8 and MRP-14 more precisely, anti-rMRP-14 antibody affinity-purified MRPs were measured by matrix-assisted LDI mass spectroscopy. It could be shown that MRP-8 has a molecular mass of 11,007 k 33 Da f 55 Da (n = 3), which indicated the existence of two MRP-14 isoforms (see "Discussion"). In addition, a heterodimer of 24,236 & 84 Da (n = 3) was detected. No evidence for homodimers or larger protein complexes was found. Identical results were obtained by the isolation of MRPs by affinity chromatography with monospecific anti-rMRP-8 antibody.
Both of the monospecific antibodies used in these experiments demonstrated an identical MRP pattern in immunopurified eluates derived from granulocytes and monocytes. Therefore, for further investigation, as a source for MRP-8 and MRP-14 molecules lysates from human granulocytes, purified by an anti-rMRP-14 affinity column were used.
Chemical Cross-linking of MRP-8 and MRP-14"The antisera used were highly monospecific. This could be confirmed further by immunoblot experiments with recombinant (data not shown) and native MRPs (Fig. 1C). Thus, one would not expect to obtain MRP-8 monomers and homodimers from an anti-rMRP-14 affinity column, and vice versa. This, however, is different from our results obtained in immunoblot experiments (Fig. IC). A likely explanation could be that both proteins form noncovalently linked complexes that disintegrate during SDS-PAGE. Therefore, cross-linking experiments employing Bis(sulfosuccinimidy1) suberate (BS") were carried out. Upon cross-linking of affinity purified MRPs and subsequent separation on SDS-PAGE under reducing conditions, two new protein bands of 35.0 and 48.5 kDa (Fig. 2 4 ) in addition to the MRP-8.MRP-14 heterodimer (24.5 kDa) were detected by silver staining. To evaluate the specificity of the cross-linking procedure, experiments were performed with different MRP concentrations. In parallel, cross-linking was carried out in supernatants prepared from monocyte and granulocyte lysates, separated by SDS-PAGE, and electroblotted. Blots were stained with anti-rMRP-14 or anti-rMRP-8 antibodies. Each of these experiments exhibited the identical pattern of MRP complex formation (Fig. 2B) as was apparent in the affinity-purified fractions. This confirms the complex formation in vivo. Further, to show the specificity of the cross-linking procedure, the cross-linking reagent was employed at a concentration even 5-10 times higher than recommended by Staros (14). Again, previous results were reproduced, and no larger complexes were seen (Fig. 2B). Interestingly, monomers and homodimers were either absent or only very weakly stained (Fig. 2).
Composition of the MRP-8-MRP-14 Complexes-The further attempt was to evaluate the composition of the complexes described above. To evaluate the possibility of the existence of a third component in the large complexes, an affinitypurified MRP fraction was split into two aliquots. One of the aliquots was cross-linked. Both were separated under reducing conditions on SDS-PAGE and visualized by silver staining (method A). In the cross-linked aliquot the very prominent 35.0-and 48.5-kDa band and a weak 24.5-kDa band were detected whereas in the other aliquot only MRP-8 and MRP-14 monomers could be seen. These results indicate that MRP-8 and MRP-14 are the only molecules involved in complex formation. Because of the apparent molecular weight of the two new protein bands, the 35.0-and 48.5-kDa complex represent a trimer and a tetramer, respectively, whereas the 24.5-kDa complex is a heterodimer, as described above. To determine the stoichiometric ratio of the single components in the trimer and the tetramer, the following experiment was performed. The affinity-purified MRP complexes were crosslinked with DTSSP, a cross-linker containing a P-mercaptoethanol-cleavable disulfide bond. The complexes were separated on SDS-PAGE under nonreducing conditions. Each

14-
"mv B kD a e f g a e t g single protein band, corresponding to its molecular weight, was excised from the gel and boiled for 10 min in SDS-PAGE sample buffer containing 5% @-mercaptoethanol for cleavage of cross-linkers. Subsequently, each gel patch was placed onto a second SDS gel to separate the now existing monomers. Protein bands were detected by silver staining (method B), and the intensity of the stained bands was determined densitometrically with a laser scan densitometer. Integration of peak areas indicated that the 35.0-kDa protein is an ((MRP-8 ) 2 * (MRP-14)) trimer and the 48.5-kDa protein is an ((MRP-8)2. (MRP-14)2) tetramer (Fig. 3). However, the possibility that both MRP-14 isoforms mentioned above are involved in the complex formation cannot be ruled out (see "Discussion"). Noncovalent Association of MRP Complexes-The results shown above indicate the presence of a disulfide bridge in the protein complexes. To determine whether the disulfide bond found in the heterodimer is of in vivo relevance, a granulocyte lysate was prepared in PBS containing different amounts of the free thiol group blocker iodoacetamide. Samples were A indicates the percentage of the appropriate peak areas (peak 1 + peak 2 = 100%) used for the evaluation of the stoichiometric compositibn shown in B.

FIG
separated on SDS-PAGE performed under nonreducing conditions, electroblotted, and immunostained with anti-rMRP-8 and anti-rMRP-14 antibodies. Only monomers were seen in samples treated with iodoacetamide, indicating that no cysteine-linked dimers could be relevant for complex formation. After cross-linking the same samples the formation of the heterodimer, the trimer, and the tetramer could be shown (Fig. 4). To prove that no cleavage of already existing disulfide bonds resulted under the chosen experimental conditions, unspecific cysteine linkage was induced by boiling the MRP fractions (cell extracts) in nonreducing SDS-PAGE sample buffer followed by treatment with iodoacetamide. Again, the dimeric form was detected as shown in experiments carried out without iodoacetamide treatment. Thus, it is confirmed that in the above described experiments only free thiol groups were blocked by iodoacetamide. Hence, it is concluded that only noncovalently linked heterodimers, trimers, and tetramers are formed in vivo.
Ca2+-dependent Complex Formation of MRPs-Both MRP-8 and MRP-14 have been shown to bind Ca2+ via two EFhands. Thus it is of great interest whether the complex formation described above is Ca'+ dependent. For this purpose, granulocytes were prepared in PBS containing different concentrations of EDTA or Ca'+ and were immediately disrupted and then incubated for 1 h at 37 "C. The samples were cross-linked, separated on SDS-PAGE, electroblotted, and immunostained with anti-rMRP-8 and anti-rMRP-14 antibodies.
The results showed that high amounts of EDTA (0.1-1 mM) yielded a weak staining of the larger protein bands (35.0 and 48.5 kDa) whereas in samples with high amounts of Ca2+ more intensive staining of these protein bands could be detected (Fig. 5). To prove whether the weaker staining of the larger protein complexes in EDTA-containing samples is caused by the decrease of free Ca2+ from intracellular stores caused by EDTA or by a nonspecific EDTA-MRP interaction, the following control experiment was performed. Granulocyte lysate was incubated with 1 mM EDTA and 1 mM Ca" simultane- 14.
-.---"7"   ously as described above. Results obtained from these experiments were identical to those of untreated samples. This demonstrates that a decreased free Ca'+ concentration is responsible for the weaker staining of the larger protein bands in EDTA-containing samples.

FIG
Further results obtained from identical experiments with affinity-purified MRPs corroborate these observations.

DISCUSSION
Previously we have described two Ca2+-binding proteins associated with defined stages of myeloid cell differentiation. The amino acid sequence of MRP-8 and MRP-14 (1, 6), derived from cDNA studies, revealed that MRP-8 contains 93 amino acids and MRP-14 114 amino acids. Molecular weights of MRPs predicted from amino acid composition derived from cDNA studies ignores post-translational modifications. Applying SDS-PAGE, different investigators have assigned different molecular weights to the native molecules (22-24). Therefore we used matrix-assisted LDI mass spectroscopy, which has been shown to provide a far more exact molecular weight than the methods mentioned above (18,19). Here we found molecular masses of 11,007 f 33 Da for MRP-8 and a doublet of clearly distinguishable molecular masses of 12,878 f 61 Da and 13,288 f 55 Da corresponding to MRP-14 (calculated molecular mass for MRP-8 was 10,835 Da and for MRP-14,13,242 Da). The possibility that one of these doublet peaks arose from a contaminating protein could be excluded by two-dimensional gel electrophoresis and subsequent immunoblotting with anti-rMRP-14 antibodies.' These results indicate the existence of two forms of MRP-14. Similar results have also been reported by Berntzen and Fagerhol (23) and Edgeworth et al. (25). Furthermore, the latter group reported a post-translational phosphorylation of both of these MRP-14 proteins (25).  (1,2). Consequently, the ATG coding for the methionine at position 5 may also function as a start codon, as it is shown for some other proteins (27)(28)(29). Alternative splicing as a cause for the existence of the two isoforms may not be likely because only full-length cDNA clones have been found (l), which could merely be derived from one mRNA species.
In addition to the distinction of MRP-14 and MRP-14', we could also detect a peak at 24.2 kDa representing a heterodimer of MRP-8 and MRP-14. The heterodimer could also be visualized by immunoblotting with anti-rMRP-8 and anti-rMRP-14 antibodies. This shows that both molecules are able to form complexes. The question of whether larger complexes of these proteins exist arose from affinity purification of the MRPs, as we resolved the identical elution pattern with both of the applied monospecific antibodies. The methods employed in this study do not preserve complexes stabilized only by noncovalent interactions. Therefore chemical cross-linking was applied immediately after cell disruption to exclude decomposition of complexes. The results derived from these experiments revealed a 48. 5  Karas, and C. Sorg, unpublished data.
Oppositely, the related proteins S-100-a (cysteine 85), S-100p (cysteine 68 and 84), and p l l (cysteine 61 and 82) are shown to be disulfide bound (30)(31)(32). This difference, however, can easily be explained by the fact that MRP-8 and MRP-14 lack a cysteine residue in that part of the C-terminal half which has been proposed to participate in intermolecular connections (32). Evidently, the N-terminal cysteine residues in MRP-8 (cysteine 42) and MRP-14 (cysteine 3) do not take part in intermolecular disulfide bonds in the assembly of MRP complexes.
In further studies the influence of Ca2+ on complex formation was investigated. A decomposition of the MRP complexes in granulocyte/monocyte lysates as well as in affinity-purified eluate upon trapping Ca2+ by EDTA was shown. Evidently, the MRP complex assembly is a Caz+-regulated process. This observation allows us to speculate on a possible function of the MRPs as a modulator/regulator in the Ca2+ signal pathway, which could be executed by their complex formation/ degradation. One of the potential candidates influenced by this proposed mechanism might be casein kinase I and 11, which is reported to be inhibited by an MRP-8.MRP-14containing complex ( 11).
This study demonstrates the existence of three different MRP-8 and MRP-14-containing complexes, the formation of which is Ca2+ dependent. Several studies of our group demonstrated up-regulation of MRP-8 and MRP-14 in myeloid cells in the course of inflammatory events (1,6,33). Preliminary evidence suggests that the expression of MRP-8 and MRP-14, which are both coded for on chromosome l3 (4), are under separate control. Although both proteins are expressed by monocytes/macrophages in chronic inflammatory processes such as rheumatoid arthritis or sarcoidosis, a dissociation is observed in acute inflammation (1, 6, 7). Furthermore the assembly of the complexes seems to represent distinct steps in the maturation of monocytes in chronic inflammation although it is not seen in acute inflammation (6). Future studies therefore will be directed at the functional characterization of MRP-8-and MRP-14-expressing monocytes/macrophages. The clearcut differences to the differentiation patterns in chronic inflammation might provide clues as to the pathomechanisms of a variety of chronic inflammatory diseases.