Báo cáo khoa học: Comprehensive interaction of dicalcin with annexins in frog olfactory and respiratory cilia pdf

S100 proteins show no enzymatic activities by themselves and, instead, modulate the function of other proteins through direct binding to Keywords annexin; dicalcin; olfactory cilia; resp

in frog olfactory and respiratory cilia

Tatsuya Uebi1, Naofumi Miwa1,2,* and Satoru Kawamura1,2

1 Department of Biology, Graduate School of Science, Osaka University, Japan

2 Graduate School of Frontier Biosciences, Osaka University, Japan

Calcium ions are known to modulate signal

transduc-tion in various cells This effect is usually mediated

by Ca2+-binding proteins For example, in olfactory

receptor cells, odorant stimuli induce Ca2+ influx

through a cyclic nucleotide gated channel [1] The

increase in the Ca2+ concentration is detected by

calmodulin, a well-known Ca2+-binding protein The

Ca2+-bound form of calmodulin has essential roles in

olfactory adaptation [2,3] In photoreceptor cells,

sev-eral Ca2+-binding proteins are known to be present

and to modulate phototransduction signals [4]

We previously found a Ca2+-binding protein,

dical-cin (renamed from p26olf [5]), in frog olfactory

epithe-lium, and reported that dicalcin is expressed in the

olfactory epithelium, lung, and spleen [6,7] In the olfactory epithelium and lung, dicalcin localizes in the cilia Dicalcin has partial homology to S100 proteins, a family of EF-hand Ca2+-binding proteins, and consists

of two S100A11-like regions aligned in sequence The amino acid sequences in the N-terminal and the C-ter-minal halves show 58% and 45% identity, respectively,

to chick S100A11 [7] The predicted structure of dical-cin is similar to that of an S100 dimer [8]

S100 proteins are known to be involved in various cellular functions, such as cell cycle progression and cell survival [9–11] S100 proteins show no enzymatic activities by themselves and, instead, modulate the function of other proteins through direct binding to

Keywords

annexin; dicalcin; olfactory cilia; respiratory

cilia; S100

Correspondence

S Kawamura, Graduate School of Frontier

Biosciences, Osaka University, Yamada-oka

1–3, Suita, Osaka 565-0871, Japan

Fax: +81 6 6879 4614

Tel: +81 6 6879 4610

E-mail: kawamura@fbs.osaka-u.ac.jp

*Present address

Department of Physiology, School of

Medicine, Toho University, Tokyo, Japan

Database

Amino acid sequences have been submitted

to DDBJ under the following accession

numbers: frog annexin A1, AB286845; frog

annexin A2, AB286846; frog annexin A4,

AB286848; frog annexin A5, AB286847

(Received 8 May 2007, revised 20 July

2007, accepted 24 July 2007)

doi:10.1111/j.1742-4658.2007.06007.x

Dicalcin (renamed from p26olf) is a dimer form of S100 proteins found

in frog olfactory epithelium S100 proteins form a group of EF-hand

Ca2+-binding proteins, and are known to interact with many kinds of tar-get protein to modify their activities To determine the role of dicalcin in the olfactory epithelium, we identified its binding proteins Several proteins

in frog olfactory epithelium were found to bind to dicalcin in a

Ca2+-dependent manner Among them, 38 kDa and 35 kDa proteins were most abundant Our analysis showed that these were a mixture of

annex-in A1, annexannex-in A2 and annexannex-in A5 Immunohistochemical analysis showed that dicalcin and all of these three subtypes of annexin colocalize in the olfactory cilia Dicalcin was found to be present in a quantity almost suffi-cient to bind all of these annexins Colocalization of dicalcin and the three subtypes of annexin was also observed in the frog respiratory cilia Dicalcin facilitated Ca2+-dependent liposome aggregation caused by annexin A1 or annexin A2, and this facilitation was additive when both annexin A1 and annexin A2 were present In this facilitation effect, the effective Ca2+ con-centrations were different between annexin A1 and annexin A2, and there-fore the dicalcin–annexin system in frog olfactory and respiratory cilia can cover a wide range of Ca2+ concentrations These results suggested that this system is associated with abnormal increases in the Ca2+concentration

in the olfactory and other motile cilia

these proteins p53, RAGE and annexins are known to

be binding proteins of S100 proteins S100 proteins are

known to form dimers, and the dimer form binds to

the binding protein to exert the effect Because dicalcin

consists of two S100-like domains aligned in sequence,

the function of dicalcin is probably similar to that of

an S100 dimer

Although the Ca2+-binding property has been

inves-tigated in detail in dicalcin [12], little is known about

its physiologic function To investigate this, in the

present study we first tried to determine the binding

proteins of dicalcin We found that several of the

pro-teins in frog olfactory epithelium bind to dicalcin in a

Ca2+-dependent manner Among them, 38 kDa and

35 kDa proteins were the major proteins We identified

them as annexin A1, annexin A2 and annexin A5 We

further examined their localizations and the effect of

dicalcin on the activities of these annexins by

measur-ing liposome aggregation

Results

Purification of binding proteins of dicalcin

Binding proteins of dicalcin were searched for among

the soluble and membrane-associated proteins of frog

olfactory cilia Because dicalcin is an S100-like

EF-hand Ca2+-binding protein, we expected that the

binding proteins would bind to dicalcin in a Ca2+

-dependent manner The Chaps-solubilized fraction

(see Experimental procedures) containing the

mem-brane-associated proteins in frog olfactory cilia

(Fig 1, cilia) was loaded onto a dicalcin-Sepharose

column at 1 mm Ca2+ Most of the proteins were

found in the pass-through fraction (Fig 1, elution

peak A and lane A), but some of the proteins were

retained, and eluted by reducing the Ca2+

concentra-tion (Fig 1, eluconcentra-tion peak B and lane B) Several

pro-teins were found in lane B, but the major propro-teins

were 38 kDa and 35 kDa proteins The latter could

be one of the binding proteins detected in our

previ-ous dicalcin-overlay analysis [13] In control studies,

we did not see the binding of these proteins when

dicalcin was not attached to the Sepharose beads

(Fig 1C) Although the amount of each of the eluted

proteins varied among preparations, 38 kDa and

35 kDa proteins were always the major constituents

We therefore focused on these proteins in the

follow-ing study Essentially similar bindfollow-ing proteins were

detected when we used the soluble protein fraction,

but the amounts of the proteins were greater in the

Chaps-solubilized fraction For this reason, we used

this fraction in the following studies

Amino acid sequence analysis of 38 kDa and

35 kDa proteins During the course of this study, we realized that

35 kDa proteins contained proteolytic fragments of

38 kDa proteins: in the presence of protease inhibi-tors, the amount of 38 kDa proteins was larger than that in the absence of the inhibitors However, we could not inhibit the proteolysis completely: even in the presence of a cocktail of inhibitors, our immuno-logic study detected signals of 38 kDa proteins at the

35 kDa position (see Fig 3A below) In addition, the degree of inhibition was variable, depending on each preparation Nevertheless, the binding proteins of di-calcin, mainly the 38 kDa and the 35 kDa proteins, were fragmented by a protease The resultant proteo-lytic fragments were isolated by RP-HPLC, and their amino acid sequences were determined The result suggested that the 38 kDa and the 35 kDa proteins are the annexin family proteins The result, however, was complex: the amino acid sequences of the frag-ments did not match the sequence of a single annexin family protein Instead, the sequence of a fragment showed some similarity to the sequence of

annex-in A1, annexannex-in A2, annexannex-in A4 or annexannex-in A5 of

Fig 1 Purification of binding proteins of dicalcin by affinity column chromatography The Chaps-solubilized protein fraction of the cilia

of frog olfactory epithelium was loaded to a dicalcin-Sepharose col-umn at 1 m M Ca2+ Most of the proteins passed through the col-umn (A in the elution profile) at a high (1 m M ) Ca 2+ concentration, but some proteins remained in the column and came out only after addition of 5 m M EGTA (B in the profile) Inset: SDS ⁄ PAGE patterns

of the Chaps-solubilized cilia protein fraction (cilia), the pass-through fraction (A) and the eluate in the presence of 5 m M EGTA (B) As a control, an eluate was obtained similarly as in (B), but with the use

of Sepharose beads without dicalcin conjugated (C) Proteins were stained with silver.

other animal species, which suggested that the

38 kDa and the 35 kDa proteins were a mixture of

these annexins We therefore tried to isolate cDNAs

of annexin A1, annexin A2, annexin A4 and

annex-in A5 to identify which annexannex-ins were annex-in the fraction

of the 38 kDa and the 35 kDa proteins

Cloning of annexin cDNAs

On the basis of the partial amino acid sequences of the

proteolytic fragments as determined above, we

synthe-sized oligonucleotide degenerate primers and used

them to search for the cDNA fragments of the

corres-ponding annexins Partial cDNA fragments of

annex-in A1, annexannex-in A2, annexannex-in A4 and annexannex-in A5 were

amplified, and the frog olfactory cDNA library

was screened with these fragments The full-length

sequences of frog annexin cDNAs were obtained, and

the amino acid sequences were deduced

(supplemen-tary Fig S1) The amino acid sequences detected in

the proteolytic fragments were found in the deduced

amino acid sequences of frog annexin A1, annexin A2,

and annexin A5, but not in the sequence of frog

an-nexin A4 This result indicated that anan-nexin A4 was

not present, or the content of annexin A4 was small in

the fraction of the 38 kDa and 35 kDa proteins

Among our recombinant annexins (see below), the

apparent molecular mass of annexin A4 was slightly

lower than 35 kDa on our SDS⁄ PAGE gel Because

the density of the corresponding position on the

SDS⁄ PAGE gel of the binding proteins of dicalcin was

faint, this result also suggested that the content of

annexin A4 in the 35 kDa proteins was small even if

it was present For these reasons, we did not study

annexin A4 further

Identification of annexin A1, annexin A2 and

annexin A5 as the binding proteins of dicalcin

Our results were so far consistent with the notion that

the 38 kDa and the 35 kDa proteins are annexin A1,

annexin A2, and annexin A5 However, we were not

totally sure of this at this stage Therefore, we first

tried to confirm that annexin A1, annexin A2 and

an-nexin A5 show Ca2+-dependent binding to dicalcin, as

the 38 kDa and the 35 kDa proteins do For this, we

obtained recombinant annexin A1, annexin A2 and

annexin A5 expressed in Escherichia coli The apparent

molecular masses of recombinant annexin A1 and

ann-exin A2 were both 38 kDa, and that of annann-exin A5

was 35 kDa (Fig 2), and all of them bound to the

di-calcin-Sepharose beads in a Ca2+-dependent manner

(Fig 2), as the native 38 kDa and 35 kDa proteins do

Second, we identified the 38 kDa and the 35 kDa pro-teins as annexin A1, annexin A2 and annexin A5 immu-nologically We raised specific antiserum against annexin A1, annexin A2 or annexin A5 in mouse and rabbit using recombinant annexins (supplementary Fig S2) Antiserum against annexin A1 recognized both the 38 kDa and the 35 kDa proteins (Fig 3A, low Ca2+ eluate), and antiserum against annexin A2 also detected the 38 kDa and the 35 kDa proteins Antiserum against annexin A5 detected only the 35 kDa proteins

From the above results, it became evident that the

38 kDa proteins contained both full-length annexin A1 and annexin A2, and the 35 kDa proteins contained full-length annexin A5 together with proteolytic fragments of annexin A1 and annexin A2 Our two-dimensional electrophoresis confirmed this (Fig 3B) This two-dimensional analysis also indicated that pro-teins other than annexin A1, annexin A2 and

annex-in A5 were not present annex-in significant amounts annex-in the

38 kDa and 35 kDa proteins (Fig 3B) In Fig 3B, there are weak signals of annexin A1 at around

pH 5.1 They are probably the signals of annexin A1 that was not focused in our two-dimensional electro-phoresis

Fig 2 Ca 2+ -dependent binding of recombinant annexins to dicalcin The cell lysate of E coli (lysate) expressing recombinant

annex-in A1, annexannex-in A2 or annexannex-in A5 was mixed with dicalcannex-in-Sepha- dicalcin-Sepha-rose beads at 1 m M Ca 2+ The beads were washed 10 times by centrifugation with K-gluc buffer supplemented with 1 m M Ca 2+ , and the 1st and the 10th extracts were subjected to SDS ⁄ PAGE (high-Ca 2+ wash 1 and high-Ca 2+ wash 10) The beads were finally washed with K-gluc buffer supplemented with 5 m M EGTA, and the extract was subjected to SDS ⁄ PAGE (low-Ca 2+

wash).

Colocalization of annexins and dicalcin in frog

olfactory and respiratory epithelium

Dicalcin has been reported to localize in the cilia of

frog olfactory and respiratory epithelium [7] To

understand the possible association of annexin A1,

annexin A2 and annexin A5 with the function of

dical-cin, we examined the colocalization of each annexin

with dicalcin, using specific antisera (supplementary

Fig S2) In addition, we also examined whether

differ-ent subtypes of the annexins colocalize in the same

cilia Figures 4 and 5 show the immunohistochemical

studies of dicalcin and annexin A1, annexin A2 and

annexin A5 In Fig 4, the olfactory cilia, which were

identified immunohistochemically with olfactory

cilia-specific Golf antibody (Fig 4M), were found to be

reactive to antiserum against dicalcin (Fig 4A,D,G)

The cilia were also positively stained with antiserum

against annexin A1 (Fig 4B), annexin A2 (Fig 4E),

and annexin A5 (Fig 4H) The merged image clearly

showed colocalization of dicalcin with each of the

an-nexins (Fig 4C,F,I) In this study, the conditions for

obtaining immunofluorescence were kept constant in

each of the observations with rabbit antiserum (Texas

Red) or mouse antiserum (fluorescein isothiocyanate),

and therefore the color in the merged picture was

dependent on the relative intensities of red and green

fluorescence, namely, the titers of antisera against di-calcin and annexins Preabsorption of the specific anti-bodies by recombinant proteins significantly reduced the signals (Miwa et al [13] for anti-dicalcin serum and Fig 4N for anti-annexin A2 serum)

Because all the annexins examined in this study co-localized with dicalcin, we then examined whether ann-exin A1, annann-exin A2 and annann-exin A5 all colocalize in the same cell Figure 5 shows the immunohistochemi-cal study of coloimmunohistochemi-calization of annexin A1, annexin A2, and annexin A5 For any combination of these three subtypes of annexin, colocalization was demonstrated (Fig 5) Therefore, it was evident that all three sub-types of annexin are present in the same olfactory cilium From the results in Figs 4 and 5, it became evident that dicalcin, annexin A1, annexin A2 and annexin A5 all colocalize in the same olfactory cilium

In the respiratory epithelium, similar colocalization was observed (supplementary Fig S3), although the signal of Golf, a marker protein of olfactory cells, was not seen

Estimation of the relative molecular abundance

of dicalcin and annexins in frog olfactory cilia The above immunohistochemical study showed that all subtypes of the annexins studied here colocalize with

Fig 3 Identification of annexin A1, annexin A2 and annexin A5 by western blot analysis (A) Determination that the 38 kDa proteins are a mixture of annexin A1 and A2 and that the 35 kDa proteins are a mixture of annexin A5 and proteolytic fragments of annexin A1 and

annex-in A2 Purified recombannex-inant annexannex-in A1, annexannex-in A2 and annexannex-in A5 (A1, A2 and A5), together with the bannex-indannex-ing proteannex-ins of dicalcannex-in (low-Ca 2+ eluate), were electrophoresed on an SDS ⁄ PAGE gel, and the proteins were stained with silver (silver stain) The proteins were probed with specific antisera against annexins (anti-A1, anti-A2 and anti-A5) by western blot The 38 kDa proteins contained both annexin A1 and

annex-in A2, and the 35 kDa proteannex-ins contaannex-ined annexannex-in A5 together with annexannex-in A1 and annexannex-in A2, possibly fragmented by proteolysis durannex-ing preparation (B) Two-dimensional electrophoretic identification of the 38 kDa and the 35 kDa proteins as annexin A1, annexin A2, and

annex-in A5 A similar analysis as annex-in (A) was performed by two-dimensional electrophoresis Annexannex-ins were identified at the apparent molecular mass of 38 kDa with pI values of 6.2–7.1 (annexin A1), and of c 8 (annexin A2), and a single spot at 35 kDa with pI ¼ 5.6 (annexin A5) Each annexin subtype is indicated by a circle.

dicalcin in frog olfactory cilia To understand the

sig-nificance of this colocalization, we tried to estimate the

relative molecular abundance of dicalcin and annexins

In this quantification, we used both the soluble and

the membrane fraction after detachment of the cilia

(see Experimental procedures) They were solubilized

with the SDS⁄ PAGE sample buffer, and were directly

electrophoresed with known amounts of recombinant

dicalcin and annexins The contents of annexins and

dicalcin in the cilia were estimated by western blot,

and their ratio determined in three frogs was

annex-in A1⁄ annexin A2 ⁄ annexin A5 ⁄ dicalcin ¼ 1.0 : 0.42 ±

0.09 : 0.54 ± 0.15 : 1.9 ± 0.6 Dicalcin is a soluble

protein, and annexins were mostly present in the

Chaps-solubilized fraction Dicalcin might have been

lost during isolation of the olfactory epithelium, and therefore the content of dicalcin could be higher than the value determined above Because the number and the volume of the cilia in the sample were not known,

it was not possible to determine the actual concentra-tions of these proteins

Effect of dicalcin on the activity of annexins

As has been reported previously, annexins are known

to induce membrane aggregation in a Ca2+-dependent manner [14], and it is also known that this activity of annexins is enhanced by binding of S100 proteins [15]

We therefore examined the effect of dicalcin on the membrane aggregation activity of annexins The

Fig 4 Colocalization of dicalcin with annexin A1, annexin A2 or annexin A5 in frog olfactory epithelium (A–I) Immunofluorescence double-staining of dicalcin and annexins A section was treated with rabbit anti-dicalcin serum (red; A, D and G) and mouse antiserum raised against one subtype of annexin (green: B, annexin A1; E, annexin A2; H, annexin A5) The corresponding images were merged (merged; C, F and I) (J–L) Controls A section for controls was treated with normal serum of rabbit (J) and mouse (K), and the images were merged (L) (M) A representative section treated with antibody to G olf All positive signals against dicalcin, annexins and G olf were observed in the cilia layer (arrowheads) (N) A control Antiserum against annexin A2 was preabsorbed with recombinant annexin A2 (O) Frog olfactory epithelium stained with toluidine blue Bars indicate 20 lm in (L) (applicable to A–N) and 50 lm in (O).

activity was measured as the increase in the absorbance

due to aggregation of phosphatidylserine liposomes

(see inset in Fig 6E, for example) The dose effect of

each of the annexins in the presence or absence of

di-calcin was examined (Fig 6A) Annexin A1 and

annex-in A2 alone annex-increased liposome aggregation similarly annex-in

a dose-dependent manner (filled rectangles and filled

circles, respectively) Dicalcin increased their activities,

and the effect was higher on annexin A2 (open circles)

than on annexin A1 (open rectangles) Annexin A5 did

not show liposome aggregation activity (open and filled

triangles) Although the effect of dicalcin was obvious

at annexin concentrations above 40 nm, the increase in

the absorbance was often too rapid for reliable data to

be obtained For this reason, we used annexins at low

concentrations The concentrations of annexins were

kept at 12.5 nm (annexin A1), 5 nm (annexin A2) and

7.5 nm (annexin A5) throughout the measurement,

based on the relative molecular abundance of ins in the cilia, i.e annexin A1 : annexin A2 :

annex-in A5 ¼ 1.0 : 0.42 : 0.54 (see above) Dicalcin was added in excess

The effect of dicalcin on liposome aggregation induced by annexins was measured at various Ca2+ concentrations, and the initial rate of increase was plotted as a function of Ca2+ concentrations As shown in Fig 6B, no significant aggregation was observed in the absence of annexins (filled triangles) or dicalcin (open triangles) In the absence of liposomes,

no significant increase in absorbance was detected (not shown) In the presence of annexins alone, slight aggregation was observed, but the effect was not so large (filled circles in Fig 6B–E) at the annexin con-centrations used (see above) When dicalcin was present (open circles), the liposome aggregation activi-ties of annexin A1 or annexin A2 were facilitated

Fig 5 Colocalization of annexin A1, annexin A2 and annexin A5 in frog olfactory epithelium (A–I) Immunofluorescence double-staining of one subtype of annexin with the other subtype of annexin A section was treated with rabbit antiserum raised against one subtype of

in (red: A, annexin A1; D, annexin A5; G, annexin A5) and mouse antiserum raised against the other subtype of annexin (green: B,

annex-in A2; E, annexannex-in A1; H, annexannex-in A2) The correspondannex-ing images were merged (C, F, I) (J–L) Controls A section for controls was treated with normal serum of rabbit (J) and mouse (K), and the images were merged (L) Positive signals were observed only in the cilia layer (arrowhead).

greatly when the Ca2+ concentration was increased

(Fig 6B,C) Essentially, the effect of dicalcin was not

seen with annexin A5 (Fig 6D)

The effective Ca2+ concentrations depended on the

subtype of annexin: annexin A2 was more sensitive to

Ca2+ than annexin A1 The half-maximal dicalcin

effect was observed at < 5 lm Ca2+ with annexin A2,

but at about 30 lm with annexin A1 Although the

ini-tial rate of aggregation increased to a similar level for

both annexin A1 and annexin A2 at high Ca2+

con-centrations (Fig 6B,C), this was partly because of the

difference in the concentrations used (12.5 nm

annex-in A1 vs 5 nm annexannex-in A2; see above) When the

con-centration of annexin A2 was increased to the same

level as that of annexin A1, the effect of dicalcin was

at least two times larger for annexin A2 than for

annexin A1 (Fig 6A)

To simulate the effect of dicalcin in a cell, dicalcin

was added to the mixture of annexin A1, annexin A2

and annexin A5 according to their ratios of the

con-centrations in the cilia (see above) The observed

acti-vity (Fig 6E, filled lines) was equal to the calculated

sum of each of the activities of annexin A1,

annex-in A2 and annexannex-in A5 (Fig 6E, thick dotted lannex-ines)

Binding of truncated forms of annexins

to dicalcin

In the present study, we found that dicalcin binds to

annexin A1, annexin A2 and annexin A5, and that it

facilitates the membrane aggregation activities of

ann-exin A1 and annann-exin A2 In mammal S100 proteins

and annexins, an S100–annexin complex is formed in

a subtype-specific manner: S100A10 binds to

annex-in A2 [16], and S100A11 bannex-inds to annexannex-in A1 [17] In the case of mammal annexin A1 and annexin A2, the specificity has been reported to arise in part at their N-terminal 1–13 amino acids [18,19] Because dicalcin binds to both annexin A1 and annexin A2, in addi-tion to annexin A5, as shown in this study, the bind-ing sites of dicalcin and those of frog annexins could

be different from those known previously To test this possibility, we examined the binding to dicalcin of

Fig 6 Effect of dicalcin on liposome aggregation induced by

an-nexins Time courses of annexin-induced liposome aggregation

were measured as the increase in the absorbance at 350 nm [see

inset in (E)] In (A), the time course was measured at various

con-centrations of annexin in the presence (open symbols) and absence

(filled symbols) of 200 n M dicalcin at 100 l M Ca 2+ The initial rate of

the absorbance increase was plotted against the annexin

concen-tration In (B)–(E), liposome aggregation was measured at various

Ca 2+ concentrations in the presence (open circles) and absence

(filled circles) of dicalcin (DC) The initial rate of the absorbance

increase was plotted against the Ca 2+ concentration [annexin A1 in

(B), annexin A2 in (C), annexin A5 in (D), annexin A1 +

annex-in A2 + annexannex-in A5 annex-in (E)] Data poannex-ints represent mean ± standard

error determined in two different preparations (n ¼ 3 in each

prepa-ration) For controls, the result with dicalcin but no annexins

pres-ent (open triangles) and that with neither dicalcin nor annexins

(filled triangles) are shown in (B) These two controls are shown as

thin dotted lines in (C) and (D) The result obtained in the presence

of dicalcin and all of the annexins (E) was compared with the

calcu-lated sum of each of the initial rates obtained in (B)–(D) (thick

dot-ted lines).

N-terminal-truncated forms of frog annexin A1 and

annexin A2 The result showed that, indeed, dicalcin

binds to these truncated forms (Fig 7A), which

indi-cated that the N-terminal region is not essential for

the interaction of frog annexin A1 and annexin A2

with dicalcin Consistently, we observed that the

35 kDa forms of annexin A1 and annexin A2 found

in the fraction of the binding proteins of dicalcin

(Fig 1) were the N-terminal-truncated annexins

(Fig 7B)

Discussion

In the present study, we showed that the major

bind-ing proteins of dicalcin in frog olfactory epithelium are

annexin A1, annexin A2 and annexin A5 (Figs 1–3

and supplementary Fig S1) The binding does not

require the N-terminal region of annexins (Fig 7)

Di-calcin and all these annexins colocalize in the olfactory

and respiratory cilia (Figs 4 and 5 and supplementary

Fig S3) Dicalcin was found to increase the rate of liposome aggregation caused by annexins (Fig 6)

Specificity of the binding between dicalcin and annexins

In the present study, we identified the 38 kDa and the

35 kDa proteins as annexin A1, annexin A2 and ann-exin A5 Annann-exins are known to bind to a dimer form of S100 proteins In mammals, the binding between annexins and S100 dimer proteins has been shown to be subtype-specific S100A11 binds to

annex-in A1 [17] (but see [20] also), and S100A10 bannex-inds to annexin A2 [16] Because dicalcin in frogs shows the highest amino acid sequence homology to S100A11 (45–58%), the binding of dicalcin to annexin A1 is not surprising However, binding to all of annexin A1, annexin A2 and annexin A5 is a rather unique charac-teristic of dicalcin, although similar comprehensive binding has been suggested for some of the S100 pro-teins [11] The comprehensive binding of dicalcin to various subtypes of annexin could be due to the char-acteristics of frog annexins and⁄ or dicalcin (see below) Annexin consists of two domains, the N-terminal region and the C-terminal protein core Although the N-terminal region has been suggested to be responsible for the binding to S100 proteins [21], the N-terminal truncated forms of annexin A1 and annexin A2 bind

to dicalcin (Fig 7) The binding of these forms sug-gests that these annexins bind to dicalcin not with the N-terminal regions but with the sites that have not yet been identified in their core domains

In S100A10 and S100A11, the amino acid residues contacting the corresponding annexins are known [22– 24] In dicalcin, several of them are conserved (supple-mentary Fig S4) The amino acids thought to give the subtype-specificity of S100 binding to annexin are also known [25] However, these residues in dicalcin are dif-ferent from those in S100A10 or S100A11 (supplemen-tary Fig S4), which suggests that the specificity of binding of dicalcin to annexins is not so strict

From the above considerations, we speculate that the binding between annexins and dicalcin occurs via the interaction between the conserved amino acids in dicalcin and the still unknown site in the core domain

of annexin Because annexin A5 lacks the correspond-ing N-terminal region of annexin A1 or annexin A2 (supplementary Fig S1), it would not be surprising if frog annexin A5 bound to dicalcin Recombinant frog annexin A4, which also lacks the corresponding N-ter-minal region, also showed Ca2+-dependent binding to dicalcin (data not shown) Similarly, as in the present study, it was reported recently that the N-terminus of

A

B

Fig 7 Ca2+-dependent binding to dicalcin of N-terminal

region-trun-cated annexin A1 and annexin A2 (A) Recombinant annexin A1 and

annexin A2 were truncated at their N-termini with elastase and

chy-motrypsin, respectively, and mixed with dicalcin-Sepharose beads.

The truncated annexin A1 and annexin A2 bound to the beads at a

high Ca 2+ concentration, but they were eluted by reducing the Ca 2+

concentration (low-Ca2+ wash) (B) Amino acid sequence analysis

showed that the proteolytic fragments used in (A) lacked the

N-ter-minal regions Arrowheads show the sites cleaved and the

mole-cular masses of the rest of the cleaved peptides Arrows show the

N-termini of the 35 kDa forms of annexin A1 and annexin A2.

annexin 6 is not required for the interaction of annexin

6 with S100A11 [26]

Colocalization of dicalcin and annexins

in the cilia

We previously reported that dicalcin is present in the

olfactory and the respiratory cilia [7] Expression of

S100 proteins has been reported in the olfactory

epi-thelium in teleosts and rodents [27,28], and in the cilia

of human bronchial epithelial cells [29] Annexins have

been detected in the tissues containing ciliated cells:

the respiratory epithelium [30,31] and bronchial

epithe-lial cells [29] So far, however, localization of annexins

in the olfactory cilia has not been reported, and

there-fore, this is the first report that annexin A1,

annex-in A2 and annexannex-in A5 are expressed annex-in the cilia of

olfactory cells In the present study, we showed that

dicalcin, annexin A1, annexin A2 and annexin A5

co-localize in the olfactory cilia Because ciliated cells

seem to express both S100 proteins and annexins, our

result could apply to cells that contain motile cilia in

general

Annexin A1, annexin A2, annexin A5 and dicalcin

are present in the olfactory cilia at a ratio of

1 : 0.42 ± 0.09 : 0.54 ± 0.15 : 1.9 ± 0.6, and dicalcin

may be present in greater amounts (see Results) A

molecular modeling study showed that the structure of

dicalcin is similar to that of an S100 dimer [8] Because

a dimer form of S100 protein binds two annexin

mole-cules [21], one dicalcin molecule would bind to two

molecules of annexins If it is the case, the amount of

dicalcin is stoichiometrically sufficient to form

com-plexes with annexin A1, annexin A2 and annexin A5

Facilitation by dicalcin of membrane aggregation

induced by annexins

The half-maximal dicalcin effects were observed at

< 5 lm Ca2+ with annexin A2 and at about 30 lm

with annexin A1 (Fig 6) These Ca2+ concentrations

are the effective ranges of annexin A2 and annexin A1

of other animal species [32] The dissociation constant

of Ca2+ binding to dicalcin has been reported to be

10–20 lm [12] A simple expectation, therefore, was

that the Ca2+ concentration effective for liposome

aggregation in the presence of annexin A2 and dicalcin

would be determined by dicalcin, which shows lower

affinity for Ca2+than does annexin A2 Similarly, one

could expect that the effective Ca2+ concentration in

the presence of annexin A1 and dicalcin would be

determined by annexin A1 However, the results were

different from what we expected The effective Ca2+

concentrations did not change significantly in the presence or absence of dicalcin The results indicated that the Ca2+ dependency of liposome aggregation in the presence of dicalcin is determined by annexins, not

by dicalcin The result therefore suggested that there is cooperative regulation of Ca2+ binding to dicalcin by annexins The increase in the degree of Ca2+ binding

in the presence of binding proteins is known for S100A4 [33] and has been suggested for S100A11 [34]

We measured liposome aggregation in a mixture of dicalcin, annexin A1, annexin A2 and annexin A5 (Fig 6D) The observed liposome aggregation profile could be explained by the sum of each of the constitu-ents in the mixture In this study, we mixed all of these proteins at once If, as we assumed, dicalcin binds to two molecules of annexin, a dicalcin molecule would

be able to bind two annexin molecules of different sub-types, such as annexin A1 plus annexin A2, and ann-exin A1 plus annann-exin A5 However, the aggregation profile obtained in the mixture could be explained by the sum of the results obtained independently using single species of annexin This result suggests that even when all of the annexins are present in the mixture, annexins of a homomeric pair, not a heteromeric one, tend to bind to dicalcin to form a complex

Possible physiologic functions of dicalcin and annexins in the cilia

It has been estimated that the intracellular Ca2+ con-centration in the olfactory cilia is about 40 nm at the resting level, and increases to higher levels after odorant stimulation [35] In respiratory cilia, the intra-cellular Ca2+ concentration increases up to a sub-micromolar level at the maximum [36] The range of

Ca2+ concentration where the dicalcin–annexin com-plex has an effect seems to be higher than these ‘physi-ologic’ Ca2+concentrations Therefore, we believe that the dicalcin–annexin complex exerts its effect when the

Ca2+ concentration is abnormally increased The cell membranes of motile cilia are subject to mechanical stress and are often disrupted [37] In addition to this, the olfactory cilia are exposed to environmental chemi-cals, microorganisms and viruses, etc., so that the cil-ium membrane is likely to be damaged In these cases, the cytoplasmic Ca2+ concentration at the disrupted site could possibly be quite high Because (a) the effec-tive Ca2+ concentrations are different between

annex-in A1 and annexannex-in A2 (Fig 6), (b) dicalcannex-in is present

in a quantity sufficient to bind all of the annexins (see Results), and (c) all these molecules colocalize in the same cilia (Figs 4 and 5), it is possible that the dical-cin–annexin system could cover a wide range of Ca2+

concentrations inside the cell to reseal the disrupted

membranes It has been reported that annexin A1 [38]

and annexin A1 and annexin A2 [39] have important

roles in membrane repair

Annexin A5 did not show liposome aggregation

activity, in agreement with the findings of a previous

study [14], even in the presence of dicalcin (Fig 6D)

Because antibody against annexin A5 has been reported

to inhibit the survival of oxidation-damaged cells [40],

the dicalcin–annexin A5 complex may possibly

contri-bute to a recovery process after chemical damage

Dicalcin in other species

So far, we have found dicalcin in Rana catesbeiana [6]

and Xenopus laevis [5] In addition, the sequence of

dicalcin mRNA of X tropicalis has been registered in

a database (NM_001016706) Thus, dicalcin has been

found only in the three species of frogs The Mexican

salamander, Ambystoma mexicanum, has an

S100A11-like protein with an insertion of four amino acid

resi-dues in its C-terminal half EF hand (supplementary

Fig S3), and this insertion is characteristically observed

in dicalcin Nevertheless, this S100A11-like protein is a

monomer form of an S100 protein and is not like

dical-cin Therefore, dicalcin might be derived from a unique

S100 protein of ancestral amphibia, and could be a

frog-specific protein Members of the Caudata,

includ-ing the Mexican salamander, have a tendency to stay

either in an aquatic or a terrestrial environment In

contrast, most frogs are more biphasic, and actively

move between land and water Because the olfactory

motile cilia in these frogs could be exposed to vigorous

mechanical stress very often, they might have needed to

have a very effective membrane repair system Dicalcin,

a homodimer form of S100 proteins, could be the form

of S100 protein that exerts this effect most efficiently

Experimental procedures

Solutions

The standard buffer solution contained 115 mm potassium

gluconate, 2.5 mm KCl, and 10 mm Hepes (pH 7.5) (K-gluc

buffer) Low-salt K-gluc buffer (LS-K-gluc buffer)

con-tained 50 mm potassium gluconate and 20 mm Hepes

(pH 7.5) Either 1 mm CaCl2or 5 mm EGTA was added to

the LS-K-gluc buffer Ringer’s solution contained 115 mm

NaCl, 3 mm KCl, 2 mm MgCl2, 2 mm CaCl2, 10 mm

glu-cose, and 5 mm Tris⁄ HCl (pH 7.5) Tris-buffered saline

(NaCl⁄ Tris) contained 0.9% NaCl and 100 mm Tris ⁄ HCl

(pH 7.5) NaCl⁄ Pi contained 137 mm NaCl, 2.7 mm KCl,

8.1 mm Na2HPO4, and 1.5 mm NaH2PO4(pH 7.4)

Preparation of Chaps-solubilized proteins of the olfactory cilia

Animal care was carried out in accordance with the institu-tional guidelines of Osaka University

Partially purified cilia from frog olfactory epithelium were obtained as described previously [13] Briefly, olfactory cilia were detached from the epithelia by abruptly raising the Ca2+ concentration to 10 mm The deciliated epithelia were removed by brief centrifugation (1500 g, 5 min; TOMY MRX-150, TMA-11 rotor, TOMY SEIKO, Tokyo, Japan), and the supernatant containing the cilia was further centrifuged at 12 000 g for 15 min (TOMY MRX-150, TMA-11 rotor) The supernatant was removed and used as the soluble protein fraction of frog olfactory epithelium The resulting pellet containing the isolated cilia was washed twice with K-gluc buffer, resuspended in LS-K-gluc buffer containing 4% Chaps, and kept at 4C overnight to solubi-lize the membrane-associated proteins of the isolated cilia The Chaps-solubilized proteins were then obtained in the supernatant after centrifugation at 440 000 g for 5 min (Hitachi CS100, RP100AT4 rotor, Hitachi Koki, Tokyo, Japan) The supernatant was diluted with LS-K-gluc buffer containing 1 mm Ca2+so that the concentration of Chaps was reduced to 0.05% The diluted fraction was centrifuged

at 12 000 g for 30 min (TOMY MRX-150, TMA-11 rotor)

to remove any precipitates before subjecting it to affinity column chromatography as described below A cocktail

of protease inhibitors (leupeptin, 5 lgÆmL)1; phenyl-methanesulfonyl fluoride, 5 lgÆmL)1; aprotinin, 5 lgÆmL)1; bestatin, 40 lgÆmL)1) was present at the indicated final con-centrations during the preparation of the above fraction

Affinity purification of binding proteins of dicalcin

A dicalcin-Sepharose column was prepared as previously described [13] Chaps-solubilized proteins of the isolated cilia were loaded on the dicalcin-Sepharose column pre-equilibrated with LS-K-gluc buffer containing 1 mm CaCl2

and 0.05% Chaps After elution of unbound proteins, pro-teins that were bound to the column at 1 mm Ca2+were eluted by reducing the Ca2+concentration with LS-K-gluc buffer containing 5 mm EGTA and 0.05% Chaps In some studies, K-gluc buffer was used instead of LS-K-gluc buffer

to isolate the binding proteins, but no significant differences were observed in the detected proteins

Determination of partial amino acid sequences

of binding proteins of dicalcin Purified binding proteins of dicalcin were digested with lysyl endopeptidase (Wako, Osaka, Japan) at an enzyme⁄ substrate ratio of 1 : 100 in 1 mL of a Tris buffer solution (100 mm Tris, pH 9.2) overnight at 37C The