Nuclear import of cellular retinoic acid-binding protein type I in mouse embryonic cells

ELSEVIER

Mechanisms of Development 58 (1996) 27-38

Nuclear import of cellular retinoic acid-binding protein type I in mouse embryonic cells Anne-Lee

GustafsonaJ, Mark Donovan bJv2,Eva Annerwalla, Lennart Denckera, Ulf Eriksso&*

aDepartment ~fPharmaceutica1 Biosciences, Division qf Toxicology, Biomedical Centre, Uppsala University. Box 594, S-751 24 Uppsala, Sweden bLudwig Institute,for Cancer Research, Stockholm Branch, Box 240, S-I 71 77 Stockholm, Sweden

Received 17 October 1995; revision received 15 April 1995; accepted 7 May 1996

Abstract Using confocal microscopy we show that cellular retinoic acid-binding protein type I (CRABP I), expressed in several embryonic cell types, displays a compartmentalized subcellular distribution. The protein was excluded from the nucleus in some cells, while in others it accumulated in the nucleus. In the rat cerebellar cell line STISA, which expresses CRABP I, the protein was found in the cytoplasm with a prominent nuclear exclusion. Addition of retinoic acid to embryos in vivo and to ST15 A cells in vitro did not affect the localization of the protein. Localization of CRABP I and CRABP I fused to a nuclear localization signal, expressed in transfected cells, suggested that cell-specific factors may regulate nuclear import of CRABP I. The potential role of a CRABP I-controlled nuclear import of retinoic acid is discussed. Keywords:

Retinoic acid; Nucleus; Import; Mouse; Gene regulation

1. Introduction Retinoic acid (RA) plays an important role during embryonic development of higher organisms (Conlon, 1995). It is well established that RA acts as a hormone by interacting with two groups of nuclear ligand-controlled transcription factors, the RA receptors (RARs) and the retinoid X receptors (RXRs) (Giguere et al., 1987; Petkovich et al., 1987; Benbrook et al., 1988; Brand et al., 1988; Zelent et al., 1989; Mangelsdorf et al., 1990, 1992). The RAR and RXR subtypes form heterodimers and interact with specific response elements in different genes (reviewed in Leid et al., 1992; Mangelsdorf et al., 1994). During murine development, the expression of the FURS and RXRs are spatially and temporally controlled, but in general, RARs are expressed in most tissues (Doll6 et al., 1989, 1990, 1994; Ruberte et al., 1991, 1993; Mangelsdorf et al., 1992).

* Corresponding author. Tel.: +46 8 7287109; fax: +46 8 332812; e-mail: [email protected] ’ Present address: Unilever Research, Colworth Laboratory, Coiworth House, Shambrook, Bedford, MK44 ILQ, UK. * A.-L.G. and M.D. contributed equally to this work.

The structure, expression and mode of action of nuclear RARs have gained much attention, but less is known regarding the synthesis of RA from retinol (vitamin A) and the possible regulatory steps which might control the action of RA in cells. Apart from interacting with the nuclear receptors, some of the different congeriers of RA bind to two types of cellular RA-binding proteins (CRABP I and CRABP II) (Giguere et al., 1990; Ong et al., 1994). However, the physiological roles of these proteins in metabolism and action of RA are not clear. So far, all-trans RA is the only endogenous retinoid identified in association with a CRABP (CRABP I; Saari et al., 1982) while the endogenous ligand for CRABP II has not been identified to our knowledge. The dissociation constant (Kd) of RA is significantly higher for CRABP II than for CRABP I (Bailey and Siu, 1988; Ong et al., 1994) but the functional consequences of this in the regulation of intracellular RA-metabolism is not known. The initial studies of the RARs have clearly demonstrated that CRABPs are not necessary for the action of RA in transcriptional activation (Giguere et al., 1987; Petkovich et al., 1987). However, these studies do not rule out a role of the CRABPs as modulators of this signalling pathway, since the transactivation assays were performed

09254773/96/$15.00 0 1996 Elsevier Science Ireland Ltd. AI1rights reserved PII SO925-4773(96)00554-O

28

A.-L. Gustqfson et al. /Mechanisms qf Development 58 (1996) 27-38

under rather non-physiological conditions using transfected cells exposed to high extracellular concentrations of RA. Some evidence suggests that, at least, CRABP I may act as a cytoplasmic sink and prevent RA from reaching the nuclear RARs (Boylan and Gudas, 1991). Furthermore, it has been proposed that RA complexed with CRABP I is a substrate in RA catabolism (Fiorella and Napoli, 1991; Boylan and Gudas, 1992) CRABP I is a major RA-binding protein in the early mouse embryo with abundant expression in a limited number of cell types including neural crest cells, subpopulations of cells in the early CNS, and limb bud mesenchymal cells (Dollt et al., 1989, 1990; Perez-Castro et al., 1989; Dencker et al., 1990; Maden et al., 1991a,b; Ruberte et al., 1991). In this paper we show that CRABP I displays a differential subcellular compartmentalization (nuclear and/or cytoplasmic) in a number of embryonic cell types. Studies using different cell lines suggested that cell type-specific accessory factors may regulate the subcellular localization of CRABP I in a ligandindependent fashion. We propose that a controlled subcellular compartmentalization of CRABP I, and thus of RA, may be important in regulation of RA-controlled transcription. 2. Results 2.1. Differential subcellular localization of CRABP I in mouse embryos Tissue sections obtained from day 9 to day 17 post coitum (p.c.) mouse embryos were stained with Ig to CRABP I, and the subcellular distribution of CRABP I was investigated by laser scanning confocal microscopy. In this work we have focused on three different types of cells which are known to abundantly express CRABP I, namely limb bud mesenchymal cells of the progress zone, neural crest-derived facial mesenchymal cells, and two types of neuronal cells, the commissural neurons of the developing neural tube and the olfactory receptor cells of the olfactory mucosa (Vaessen et al., 1989, 1990; Dencker et al., 1990; Maden et al., 1991a,b; Ruberte, 1991, 1993; Gustafsson et al., 1993). To delineate the nucleus of individual cells, the tissue sections were counterstained with propidium iodide (PI). In the images, nuclei are represented in red and CRABP I in green (Figs. 1 and 2). Nuclear CRABP I, consequently, appears in yellow (mix of red and green). Densitometric measurements along the lines in Figs. 1B and 2B,C, were used for further illustration and semiquantification of the intracellular distribution of CRABP I in individual cells (Fig. 3). In the developing limb bud, CRABP I was expressed in the progress zone with no detectable expression in the surface ectoderm (Fig. 1A-C). The subcellular distribution of the protein varied in individual cells. Thus, in some cells CRABP I was localized in the cytoplasm,

while in other cells, CRABP I was preferentially localized in the nuclei. Significant nuclear staining could be seen in at least 60-70% of the CRABP I-expressing cells in each projection. In many cells excluding CRABP I from the nucleus, an accumulation of the protein close to the nuclear envelope could be seen (yellow rims around the red nuclei). This was confirmed by the densitometric measurements showing that cells excluding CRABP I from the nuclei often displayed a marked accumulation of CRABP I in close proximity to the nuclear envelope while the central areas of the nuclei were essentially devoid of detectable CRABP I (Fig. 3A,B). Conversely, cells with a marked nuclear accumulation of CRABP I displayed lower levels of detectable CRABP I in the cytoplasm. In the CRABP I-positive nuclei, the protein was not evenly localized but instead displayed a speckled distribution (Fig. 1C). In the frontonasal mesenchyme, CRABP I is abundantly expressed in neural crest-derived mesenchymal cells and the subcellular distribution of the protein was similar to that previously described for the limb bud mesenchymal cells. Thus, CRABP I was excluded from the nuclei of some cells while in other cells, CRABP I accumulated in the nuclei (Fig. 2A,B). Calculations revealed that about 40-50% of the CRABP I-expressing cells displayed significant amounts of CRABP I in their nuclei. Densitometric measurements of selected individual cells verified that the protein often accumulated in close proximity to the nuclear envelope of cells with a nuclear exclusion of the protein (Fig. 3C,D). The nuclear CRABP I displayed a speckled distribution similar to that of the limb bud mesenchymal cells (Fig. 2B). Within the two mesenchymal tissues (limb bud and frontonasal) investigated here, there appeared not to be any association between the proximity to the surface epithelium and the subcellular distribution of CRABP I. The commissural neurons of the neural tube (Fig. 2C,D), being post-mitotic neurons, and the olfactory receptor cells of the olfactory epithelium (Fig. 2E) displayed a differential subcellular localization of CRABP I in a manner similar to that previously described for the mesenchymal cells of the limb bud and the frontonasal mesenchyme. The densitometric measurements of selected individual cells verified these observations (Fig. 3E,F). Although nuclear CRABP I was not evenly distributed within the nuclei of these cell types, a marked speckled distribution as seen in the mesenchymal cells could not be observed. The PI intensity was measured in limb bud mesenchymal cells excluding CRABP I from the nucleus and in cells with a significant nuclear accumulation of CRABP I. Although there was an expected variation in PI staining, we found no significant difference in PI staining between cells with or without nuclear CRABP I. This suggests that there is no correlation between the subcellular distribution of CRABP I and the amount of DNA in the two types of

A.-L.

Gusrafsonet al. /Mechanisms ef Developmenr 58 (1996) 27-38

29

Fig. 1. Differential subcellular localization of CRABP I in limb mesenchymal cells. Tissue sections from early mouse embryos (day 10.5 p.c.) were stained for CRABP I and the subcellular localization of the protein and the location of PI-stained nuclei were determined by confocal microscopy. Cytoplasmic CRABP I is represented in green, the PI-stained nuclei in red and nuclear CRABP I in yellow. (A) Projection showing an overview picture of a horizontal section of a limb bud. (B) Magnification of a subectodermal area of (A). The subcellular distribution of CRABP I was quantified in two cells (a and b) along the two lines, s denoting the startingpoint for the measurements (see Fig. 3 for the results). (C) A high contrast projection of a subectodermal area of a different limb bud. Cells with preferential cytoplasmic localization of CRABP I and with a speckled nuclear distribution of CRABP 1 are seen. aer, apical ectodermal ridge; ec, ectoderm. Bars, 10pm.

cells (correlation factor 0.97-0.98, n = 25) (data not shown). Furthermore, since we also observe the differential compartmentalization of CRABP I in post-mitotic neurons, nuclear uptake of the protein is not obviously related to the cell cycle. To evaluate if the ligand saturation affected the subcellular distribution of CRABP I, dams were administered teratogenic doses of RA, and tissue sections from exposed embryos were analyzed by confocal microscopy as above. For the tissues examined, we found more differences between embryos in the same treatment group than be-

tween control and RA-exposed embryos, suggesting that the subcellular localization of CRABP I in embryonic cells is ligand-independent (data not shown). 2.2. Quantitative estimation of retinoid-binding CRABP l-expressing embryonic tissues

sites in

The differential compartmentalization of CRABP I would drastically affect the subcellular location of RA, provided that CRABP I is expressed in a molar excess compared to endogenous RA. To roughly estimate the

30

A.-L. Gustafson et al. /Mechanisms of Development 58 (I 996) 27-38

Fig. 2. Differential subcellular localization of CRABP I in neural crest-derived frontonasal mesenchymal cells and in some neuronal cells revealed by confocal microscopy. The experimental procedures and the color codes are the same as in Fig. 1. (A) Optical section showing an overview of a coronally cut medial nasal process with CRABP I-expressing neural crest-derived mesenchymal cells (day 10.5 pc.). (B) Projection showing a part of a coronally cut lateral nasal process (day IO.5 p.c.). The subcellular distribution of CRABP I was quantified in two cells (a and b) along the two lines, s denoting the starting point for the measurements (see Fig. 3 for the results). (C) Projection from a transverse section of the neural tube (day 10 p.c.). The subcellular distribution of CRABP I was quantified in two cells (a and b) along the two lines, s denoting the starting point for the measurements (see Fig. 3 for the results). (D) Optical section showing two CRABP l-expressing commissural neurons of the neural tube (day 10.5 pc.). (E) CRABP I-expressing olfactory receptor cells of the olfactory epithelium (day 17.5 p.c.). Cells with a preferential cytoplasmic localization of CRABP I and cells with a nuclear localization of CRABP I are shown. e, ectoderm of the olfactory placode; nc, neural crest cell. Bars, 1Opm (A,C) and 5,~rn (B,D,E).

A.-L.. Gustgfson et ul. /Mechanisms of Development

A

Limb, cell a 260

I

58 (1996)

B

27-38

31

Limb, cell b

-,

200-

r-

’ P’ ‘3,’

s

2i

160-

0

2

4

6

-*

6

, #.- I b' ’\

10

,

12

14

16

0

2

4

6

b

LNP, cell a

0

2

4

6

6

10

NT, cell

0

2

4

12

14

16

12

14

0

2

4

6

6

10

12

14

16

Distance (pm)

F

a

6 6 10 Distance (pm)

10

LNP, cell b

Distance (pm)

E

6

Distance (pm)

Distance (pm)

12

14

16

NT, cell b

0

2

4

6

6

1012

14

16

16

Distance (pm)

Fig. 3. Semiquantitative measurements of the subcellular localization of CRABP I and PI by densitometric scanning. The intensities of the CRABP I expression and of the corresponding PI staining in the projections were quantified along the lines shown in Figs. IC and 2B,C (the starting points of the measurements are indicated by s in the figures). (A,C,E) Histograms showing the intensity of CRABP I expression (solid line) and PI staining (dotted line) in cells with a preferential cytoplasmic localization (cells labelled a in the figures). (B,D,F) Cells with a strong nuclear accumulation of CRABP 1. LNP, lateral nasal process; NT, neural tube.

ratio between available RA-binding sites and previously measured concentrations of endogenous RA, a radiolabelled analogue of RA ([14C]TIXPB) was administered to pregnant mice, and the concentration of TTNPB was measured in maternal blood and embryonic tissues after autoradiographic processing. We have previously published a qualitative study using this procedure showing that CRABP I is the major binding protein for TINPB in the embryo (Dencker et al., 1990) (see also Section 3).

Fig. 4 presents data collected at 4 h after an intravenousinjection into pregnant mice at day 10 p.c. (approximate peak concentration in the embryo; shorter or longer postinjection intervals are not shown). At this semi-steady state situation between maternal and embryonic compartments, the average concentration of TTNPB in loose mesodermally-derived mesenchyme was found to be only slightly lower than that in maternal blood. This indicates that there is no active transport of the compound by the

A.-L. Gustafson et al. /Mechanisms

32

1200 L

;-:,

:,

of Development 58 (1996) 27-38

sues concentrations of RA (40 nM, Thaller and Eichele 1987; 50-100 nM, Horton and Maden, 1995), suggests that the ligand saturation of CRABP I is low and probably under 5%. 2.3. Localization of CRABP I endogenously rat cerebellar

_

01 1 0

15

30

45

60

Mmutes

Fig. 4. Quantitation of RA-binding sites in early mouse embryonic tissues using autoradiography. The diagram presents data collected 4 h TTNPB into pregnant after an intravenous injection of t4C-labelled mice on day 10 pc. The average concentration of TTNPB in nonCRABP I-expressing cells was determined in maternal blood (MB) and in mesodermally derived mesenchyme (MM). Tissues strongly expressing CRABP I, like the limb bud (LB), neural tube (NT), frontonasal mesenchyme (FNM) and hindbrain floor (HBF) exhibited strong TTNPB accumulation (up to 25x that of MM). The estimated values in the different tissues disregard the fact that CRABP I may not be fully saturated, indicating that the real number of RA-binding sites is higher than presented here. The values represent the means + SEM for n = 1530 individual measurements. For calculations see Section 4.

placenta into the embryonic tissues. The loose mesodermally-derived mesenchyme is an approximation of the extracellular fluid, due to the low cell density and the absence of detectable CRABP I expression. On the other hand, the strongly CRABP I-expressing tissues such as frontonasal mesenchyme, hindbrain floor, neural tube and limb buds accumulated radioactivity up to 25X the embryonic background levels (loose mesodermally-derived mesenchyme; see above). The highest actual concentration recorded was approximately 9 nmol/g wet weight tissue which corresponds to approximately 9,uM. This value is probably an underestimation of the number of potential RA-binding sites in the cells since oral predosing of dams with 50 mg/kg body weight of unlabelled RA 2 h prior the intravenous injection of i4C-labelled TINPB reduced its accumulation by more than 50% in CRABP Iexpressing tissues, but not in loose mesodermally derived mesenchyme (data not shown). In addition, the results from the densitometric measurements of an area probably underestimates the concentration of RA-binding sites in individual cells since not all cells in that area express CRABP I (most obvious in neural tissues), and the extracellular space is also included in the area of measurement. In summary, our measurements suggest that the concentration of available specific RA-binding sites, mainly represented as CRABP I in the developing embryo (Dencker et al., I990), may exceed 10,~M in some cells. Comparison of available RA-binding sites and the measured tis-

expressed

in

STISA cells

To delineate the possible cellular mechanisms involved in regulation of subcellular compartmentalization of CRABP I, we have explored the rat cell line STlSA, a CRABP I-expressing cell line obtained from rat primary cerebellar cells infected with a recombinant retrovirus carrying a temperature-sensitive allele of the SV40 Tantigen (Fredriksen et al., 1988). Localization by immunofluorescense showed that fixation in cold methanol or in p-formaldehyde gave identical results and that CRABP I was largely confined to the cytoplasm in these cells (Fig. 5A). Cultivation of the cells in the presence of 1 ,uM RA for up to 72 h did not markedly affect the localization of CRABP I compared to cells grown in the presence of vehicle alone (Fig. 5C,D). We conclude that the ligand saturation of the protein is not regulating its subcellular compartmentalization in these cells. Likewise, culturing STlSA cells for up to 8 h in the presence of 50pM cycloheximide, an inhibitor of protein synthesis, did not markedly affect the subcellular localization of CRABP I, although the intensity of the immunofluorestense signal faded away with time (data not shown). This suggests that CRABP I is not retained in the cytoplasm by a rapidly turning over cytosolic protein. 2.4. Localization of CRABP I and CRABP I fused to a nuclear localization signal in transfected cell lines

To explore in more detail the mechanisms underlying the cytoplasmic localization of CRABP I, expression vectors encoding wildtype CRABP I (wtCRABP I) and CRABP I fused, at the N-terminus, to the nuclear localization signal (NLS-CRABP I) from SV40 T-antigen (Kalderon et al., 1984) were constructed. In transfected Cos- 1 cells both proteins were efficiently synthesized and in SDS-PAGE, the NLS-mutant migrated slightly slower than the wildtype protein (Fig. 6). Immunocytochemical localization of the two forms of CRABP I in transfected Cos- 1 and chorioncarcinoma JEG-3 cells, two cell lines devoid of detectable endogenous expression of the protein, and in transfected STlSA cells, with endogenous expression of CRABP I, were carried out. In Cos- 1 and JEG-3 cells expressing wtCRABP I, the protein was found both in the cytoplasm and in the nucleus with no apparent sign of cytoplasmic or nuclear accumulation (Fig. 7A,B). In contrast, expression of NLS-CRABP I in Cos- 1 and JEG-3 cells resulted in a clear nuclear accumulation of the protein (Fig. 7C,D; data not shown for JEG-3 cells). From these experiments we conclude that

A.-L. Gusfq&x~ et al. /Mechanisms

ofDevelopment

58 (1996) 27-38

33

8). Using confocal microscopy and semiquantitative measurements of the subcellular distribution, we concluded that NLS-CRABP I in STlSA cells was efficiently retained in the cytoplasm with up to 10X stronger signal in the cytoplasm compared to that of the nucleus (Fig. 8A,B). This result is remarkable as it contrasts the subcellular localization of the mutant protein when overexpressed in heterologous cells without endogenous synthesis of CRABP I. We conclude that in STlSA cells NLS-CRABP I remains in the cytoplasm despite the presence of a functional nuclear localization signal in the protein. We have also studied the properties of Cterminally epitope-tagged versions of the two forms of CRABP I in STlSA cells with essentially the same results as outlined above (unpublished observations). 3. Discussion The mechanisms underlying inter- and intracellular transport of retinoids are only partly understood. Yet, in light of the involvement of RA in growth, differentiation and pattern formation during embryonic development, it seems likely that formation of RA from its precursor reti-

Fig. 5. Localization of CRABP I in STISA cells. (A) Immunolocalization of CRABP I in methanol-fixed STISA cells using specific Ig to CRABP I revealed a predominant cytoplasmic localization. (B) Immunostaining using preimmune Ig gave a weak non-specific staining of the cells. (C) Immunolocalization of CRABP I in methanol-fixed STlSA cells cultivated in the presence of 1 PM RA (obtained from a stock solution in ethanol) for 72 h. (D) Immunolccalization of CRABP I in methanol-fixed STlSA cells cultivated in the presence of ethanol (at a concentration identical to that in (C)) for 72 h. Bar, 10,um.

the primary structure of CRABP I does not contain information regulating its subcellular localization in heterologous cells. Furthermore, the mutant NLS-CRABP I is efficiently targeted to the nucleus, suggesting that the artificial nuclear localization signal is functional when fused to CRABP I. Analysis of transfected Cos-1 and JEG-3 cells, grown in the presence of 1 ,uM RA for 1224 h, showed that ligand saturation did not affect the subcellular localization of CRABP I (data not shown). Both CRABP I and the mutant NLS-CRABP I remained cytoplasmic when overexpressed in transfected STISA cells, with or without the addition of 1 ,uM RA to the culture medium for 12-24 h (data not shown, and Fig.

NL SCRABP I WI CRABP I

Fig. 6. Expression of CRABP 1 and NLS-CRABP I in transfected Cos-I cells. Expression vectors encoding wtCRABP 1 and NLS-CRABP 1 were separately transfected into Cos-I cells. Protein extracts from the transfected cells were subjected to SDS-PAGE and transferred onto nitrocellulose filters. The two forms of CRABP I were. detected with specific Ig and visualized using the ECL technique.

34

A.-L. Gust~@on et al. /Mechanisms

of Development 58

(1996) 27-38

Fig. 7. Localization of CRABP I and NLS-CRABP I in transfected Cos-I and JEG-3 cells. (A) Cos-I cells and (B) JEG-3 cells transfected with an expression vector encoding CRABP I (wtCRABP I). Note the weak non-specific staining of the surrounding non-transfected cells. (C) Cos-I cells transfected with an expression vector encoding NLS-CRABP 1. The cells were fixed 48 h posttransfection and the two forms of CRABP I were detected using specific Ig to CRABP I. (D) Visualization of cell nuclei by staining with Hoechst dye of the cells in the field shown in (C). A transfected Cos-I cell expressing NLS-CRABP 1 with a nuclear localization is marked (arrows in (C,D)). Bar, 10pm.

nol, as well as transport of RA between cells and in target cells, are strictly regulated events. Our present results suggest that the subcellular distribution of CRABP I, the major RA-binding protein during early embryonic development (Dencker et al., 1990), is regulated. These results provide the first in vivo evidence for a role of CRABP I in transport of RA to the nucleus, as was previously proposed by others (Takase et al., 1986). Differential nuclear import of CRABP I was found in mitotically active and inactive embryonic cell types expressing the protein. Nuclear import of CRABP I appeared not to be obviously related to either the position of the expressing cell in the embryo, or to the ligand saturation. However, we could observe that a larger fraction of the limb bud mesenchymal cells, compared to the frontonasal mesenchymal cells, accumulated CRABP I in the nuclei at day 10.5.

Nuclear import and export of proteins are generally well regulated processes involving specific carrier proteins (Gerace, 1995; Melchior and Gerace, 1995). It is believed that the nuclear pores in the nuclear envelope allow cytoplasmic proteins with a molecular mass below 50-60 kDa, to passively diffuse into the nucleus and to equilibrate with the cytoplasm (reviewed in Silver, 1991). The low molecular mass of CRABP I (15.6 kDa) suggests that this protein would freely enter the nucleus. The results from the overexpression studies of CRABP I in heterologous cells are consistent with this notion, as similar concentrations of CRABP I were found in the cytoplasm and in the nucleus. The data presented further show that the primary structure of CRABP I (Sundelin et al., 1985) lacks a functional nuclear localization signal, and that cells without endogenous expression of CRABP I apparently cannot control the compartmentalization of the pro-

A.-L. Gust&on

ofDevelopment 58

et al. / Mechunisms

3.5

(1996) 27-38

NLS-CRABP I

0

5

10

15

20

25

30

35

Distance (pm)

Fig. 8. Immunolocalization of NLS-CRABP I in transfected STISA cells by confocal microscopy. (A) Projection of a STISA cell transfected with an expression vector encoding NLS-CRABP 1. The cells were stained with Ig to CRABP I and a predominant cytoplasmic localization of NLS-CRABP I is seen (green). The PI-Iabelled nuclei are in red. Weak expression of endogenous CRABP I is seen in the surrounding non-transfected cells. Bar, Sprn. (B) Histogram showing the relative optical density of the CRABP I signal (solid line) and of the PI signal (dashed line) measured along the line in (A).

tein, which occurs in various cell types in vivo and in STlSA cells in vitro. The absence of effects of RA on the subcellular localization of CRABP I in embryos in vivo, in ST 15A cells and in the transfected Cos-1 and JEG-3 cells in vitro, as well as the calculated low ligand saturation of CRABP I in embryos, furthermore suggested that nuclear import of CRABP I is ligand-independent. Given the low levels of expression of the nuclear RARs, it seems likely that CRABP I is responsible for the high level of RA binding recorded by autoradiography The low concentration of endogenous RA in embryonic tissues compared with the high concentration of CRABPs reported here, suggests that the concentration of free, non-protein bound RA in the embryo is low and probably below the &s of the nuclear RARs (Allenby et al., 1993). Previous studies suggested that CRABP I prevents nuclear uptake of RA by acting as a cytoplasmic sink, and that the role of the protein is to protect vulnerable cells from the teratogenic action of excess RA (Doll6 et al., 1990; Boylan and Gudas, 1991). However, the overexpression study was interpreted without considering the possibility of a regulated nuclear uptake pathway for RA via CRABP I (Boylan and Gudas, 1991). Based on our data, we propose that CRABP I instead may act as a reservoir, and a buffering protein for RA in embryonic cells with the ability to control nuclear import of RA. Whether CRABP II might have a similar role remains to be established, but it should be noted that CRABP II binds RA less efficiently than CRABP I. A regulated nuclear import of RA via CRABP I may be a means of regulating the availability of RA for the nuclear RARs which may indirectly affect RA-controlled transcription. Furthermore, storage of presynthesized RA bound to CRABP I in the cytoplasm may uncouple de novo synthesis of RA in the

embryo from the actual need of RA during embryogenesis. Nuclear import/export of CRABP I is unlikely to be directly related to the proposed role of CRABP I in catabolism of RA. Using gene targeting techniques it has been demonstrated that mice deficient in CRABP I developed normally (de Bruijn et al., 1994; Gorry et al., 1994). However, considering the strong selection pressure on the protein, which has almost totally conserved its structure during mammalian evolution (Donovan et al., 1995), a physiological role is likely. Since several genes encoding proteins involved in the RA-signalling pathway contain RA-responsive elements (see Gudas et al., 1994; Mangelsdorf et al., 1994) the RA-signalling pathway may be autoregulated and obviously able to compensate for the loss of CRABP I. We observed cytoplasmic localization of NLS-CRABP I in STlSA cells but nuclear accumulation of the same mutant protein in cells not constitutively expressing CRABP I. It thus seems likely that accessory factors, expressed in a cell-specific fashion, may regulate subcellular compartmentalization of the protein. In addition, the speckled nuclear distribution of CRABP I seen especially in the mesenchymal cells suggests that CRABP I binds to some, yet unidentified, subnuclear structures. Likewise, the accumulation of cytosolic CRABP I in close proximity to the nuclear envelope, in embryonic cells excluding the protein from the nucleus, demonstrates that localization of CRABP I is tightly controlled. That CRABP I may be involved in protein-protein interactions is well in line with the fact that it is highly conserved (Donovan et al., 1995). It is now well established that a strictly controlled subcellular localization of several transcription factors or

36

A.-L. Gustqkm

et al. /Mechanisms

components in signal transduction pathways are important in gene regulation during differentiation and embryonic development (Vandromme et al., 1996) Examples are the glucocorticoid receptor (Picard and Yamamoto, 1989) and the family of Rel/NF/c-related proteins, which includes the mammalian transcription factor NFK (reviewed in Bauerle, 1991) and the dorsal gene product in Drosophila (Roth et al., 1989; Rushlow et al., 1989; Steward, 1989). One property of these and other signalling pathways is that latent forms of essential components are stored in the cytoplasm and can be rapidly activated by nuclear import. The data presented in this work imply that RA-controlled gene expression may be regulated in a similar manner and that RA, a potent transcriptional activator, is stored as a latent compound complexed with CRABP I in the cytoplasm of some cells. A protein-mediated system for regulation of subcellular compartmentalization and nuclear import of RA may allow several levels of regulation of the RA-signalling pathway. It remains a challenge for the future to characterize this machinery and to identify the possible external or internal cues which may be responsible for its activity. 4. Experimental

procedures

4.1. Immunohistochemical confocal microscopy

localization

of CRABP I using

C57B1/6 mice were maintained, and the embryos were isolated and processed as previously described (Dencker et al., 1990; Gustafson et al., 1993). For laser scanning confocal microscopy, paraffin sections from formaldehyde-fixed mouse embryos were incubated overnight with rabbit Ig to CRABP I (.Spg/ml) as previously described (Gustafson et al., 1993). The used antibodies were raised against a synthetic peptide, corresponding to residue 68-81 of bovine CRABP I (Eriksson et al., 1987; Busch et al., 1990). Following extensive washing in phosphate-buffered saline (PBS) and 50 mM Tris-HC1 buffer (pH 7.6), the sections were incubated with RNAse A (250,uglml) for 30 min to remove RNA and then washed with PBS. Bound Ig was visualized by fluorescein (FITC)-conjugated donkey anti-rabbit Ig (Jackson Inc.) diluted l/50 in PBS. The sections were subsequently washed in PBS and Tris-buffer and the nuclei were stained with PI (2pg/ml in PBS). Following washings in PBS and in Tris-buffer, the sections were mounted in Vectashield (Vector) to reduce fading. The sections were viewed with a Nikon inverted microscope and dual scannings were performed using a Molecular Dynamics Multiprobe 2001 instrument. Excitation wavelengths of 488 nm (for fluorescein) and 568 nm (for PI) were used. Using the ImageSpace 3.10 Software, projections (look through type) corresponding to 0.6-0.9pm thickness were created from the optical

ef Development 58 (1996) 27-38 section series (Carlsson et al., 1985, Mossberg et al., 1990). Noisy projections were subjected to mean filtration. The optical densities were recorded along the lines indicated in Figs. 1, 2 and 8. 4.2. Immunohistochemical localization embryos exposed in utero to RA

of CRABP I in

Pregnant dams (day 10 p.c.) were given a single peroral dose of all-trans RA (100 mg/kg; a kind gift from F. Hoffmann-La Roche AG, Basel). The substance was dissolved in DMSO and further diluted in soybean oil, giving a final DMSO concentration of 8%. The dams were killed 3 h after dosing (approximately peak concentration of RA in the embryo; Creech Kraft et al., 1989) and the embryos were processed for confocal microscopy analysis as described above. In the neural tube, the number of cells with exclusively cytoplasmic or nuclear accumulation of CRABP I were counted in optical section series from four tissue sections from each of two control and two RAexposed embryos. The more speckled distribution of CRABP I in frontonasal and limb bud mesenchyme called for a different quantification in these cells, Optical section series were scanned from two tissue sections from each treatment group at a confocal microscope setting giving a similar intensity in all sections. Thereafter, by using the threshold function in the image space software, cutting off low intensity levels, the intensity level at which 50% of the cells, originally showing highest concentration of nuclear CRABP I, still showed nuclear CRABP I were determined. These levels were then mutually compared between the control and RA-exposed groups. 4.3. Quantitation of RA-binding using autoradiography

sites in embryonic

tissues

Pregnant dams (C57B1/6; day 10 p.c.) were given a single intravenous injection of a i4C-labelled RA analogue, the arotinoid Ro 13-7410 (TTNPB; specific activity 1.85 GBq/mmol) (a kind gift from F. Hoffmann-La Roche AG, Basel) dissolved in ethanol-mouse serum, using a dose of 1 mg/kg. Twenty min. 1, 4 and 8 h later, the dams were killed, the uteri removed and quickly frozen. Tape-attached tissue sections (20pm) were freezedried and apposed against X-ray film (Industrex C, Kodak) together with calibrated i4C-standards (Amersham) (Dencker et al., 1987; Ullberg, 1977). The autoradiograms were developed 7 weeks later. The optical density of the autoradiograms corresponding to embryonic tissues (at five locations in each of 3-5 embryos) and the i4Cstandards, respectively, were digitized by computerassisted image analysis using the MicroScale TM/TC image analysis system (Dighurst Ltd., UK) (d’Argy et al., 1990). The densities corresponding to the steps in the 14Cstandard staircase were plotted against the volume-based radioactivity they represent to produce a standard curve.

A.-L. Gustufson et at. /Mechanisms of Development 58 11996) 27-38

Then, the tissue-derived into volume-based tissue could be recalculated to TTNPB, since the specific was known.

densities could be translated radioactivity, which in turn real tissue concentration of activity of the labelled TI’NPB

4.4. Cell culture and immunocytochemical localization of CRABP I in STISA cells The rat postnatal cerebellar cell line STlSA (Fredriksen et al., 1988) and Cos-1 cells were grown at 37°C in Dulbecco’s modified Eagle’s medium containing 10% fetal bovine serum, 2 mM glutamine and appropriate antibiotics (Gibco BRL). Chorion carcinoma JEG-3 cells were grown in minimal essential medium containing the same supplements as outlined above. For immunocytochemical studies of endogenous CRABP I expression in ST15A cells, 1 X lo5 cells were seeded onto glass coverslips precoated with polylysine (2 mg/ml) in 60 mm Petri dishes and cultured overnight. Some cultures of ST15A cells were incubated with 1 PM RA (stock solution of RA was dissolved in ethanol) or in a corresponding concentration of ethanol for 24, 48 and 72 h. Other cultures were grown in the presence of cycloheximide (5Opg/ml medium) for 1, 2, 4 and 8 h. For analyses, the glass coverslips were rinsed twice in ice-cold PBS, the cells were fixed with -20°C methanol for 2 min and subsequently washed with PBS. Some cells on coverslips were fixed in 3% p-formaldehyde in PBS for 30 min on ice and then permeabilized in 0.1% Triton X-100 in PBS. The coverslips were incubated with 5% BSA/PBS for 1 h at room temperature and then incubated with either antiCRABP I Ig (S,~g/ml) or control Ig (5pg/ml) in PBS containing 5% bovine serum albumin overnight at 4°C in a humidified chamber. Bound Ig was visualized by FITClabelled secondary antibodies (Sigma Chemicals). The coverslips were viewed in a Zeiss microscope equipped for fluorescence microscopy. Kodak Tri X-400 film was used for recording. 4.5. Expression of wildtype and mutant CRABP I in STISA cells, Cos-I cells and JEG-3 cells The 5’ and the 3’ ends of a bovine CRABP I cDNA clone (Nilsson et al., 1988) were modified using polymerase chain reactions (PCR) at standard conditions employing Taq DNA polymerase (Perkin Elmer) (denaturation at 95°C for 1 min, annealing at 53’C for 1 min and elongation for 2 min at 72°C). The oligonucleotides ON1 and ON2 were used to engineer wildtype CRABP I. To generate an amino terminal region containing the prototype nuclear localization from SV-40 T-antigen (Pro-LysLys-Lys-Arg-Lys-Val-, NLS-CRABP I) (Kalderon et al., 1984) the oligonucleotide ON3 together with ON2 were used as above.

Oligonucleotides ON1 ON2 ON3

37

used were:

5’-ACGTGAAITCGCCACCATGCCCAACTIC-3’ Y-ACGTGGATCCCCGCCTICA’ITCCCGAACA-3’ S-ACGTGAA’ITCGCCACCATGCCCAAAAAGAAA CGCAAAGTCCCCAACTI-C-3’

The oligonucleotides contained EcoRl (ONl, 0N3) and a BamHl restriction site (ON2) (underlined) for subsequent cloning into expression vectors. The generated PCR fragments were digested with EcoRl/BamHl, gel purified and cloned into the EcoRl/BamHl-digested expression vector pSG5 (Green et al., 1988). The inserts of the generated expression plasmids were sequenced and no introduced mutations were observed. To monitor expression of CRABP I and NLS-CRABP I by immunoblotting, Cos-1 cells were seeded in 60 mm Petri dishes (1 X lo6 per dish) and transfected with 3 pg of expression plasmid per dish using DEAE-dextran for 3-4 h, and then shocked with 10% DMSO/PBS for 2 min, washed with PBS and cultured in fresh media for an additional 48 h. Total protein extracts were prepared, analyzed by sodium dodecylsulfate polyacrylamide gel electrophoresis using a 12.5% gel and electrophoretically transferred onto a nitrocellulose filter. Blocking of non-specific binding sites and anincubations were as previously described tibody (Gustafson et al., 1993). Bound Ig was visualized using the ECL system as recommended by the supplier (Amersham Inc.). For immunofluorescence localization of CRABP I in transfected ST15A cells, Cos-1 cells or JEG-3 cells, respectively, 1 x lo5 cells were seeded onto glass coverslips in 60 mm Petri dishes and the cells were transfected with 3 ,ug expression plasmid per dish. ST15A cells were transfected using Lipofectamine (Gibco) (2 mg/ml), using a ratio of Lipofectamine to DNA of 5:l (w/w). The mixture was added to cells for 3-4 h, after which the cells were washed with PBS and fresh media applied. Cos-1 cells were transfected by using DEAEdextran as described above. JEG-3 cells were transfected using standard calcium phosphate precipitation. The transfected cells were cultured for 48 h to allow expression and then prepared for immunofluoresence localization of CRABP I and NLS-CRABP I as described above. To visualize the nuclei, some specimens were treated with Hoechst dye (2 pg/ml). Analyses by confocal microscopy of methanol-fixed ST15A cells were carried out essentially as described above. Acknowledgements We thank Raili Engdahl for skillful technical assistance, Robert England, David Hanzel and Borje Bjelke for helpful suggestions regarding the confocal study and Ralf Petterson for valuable comments on the manuscript. This study was supported in part by grants from the Swedish Medical Research Council and the Alice and Knut Wallenberg Foundation.

38

A.-L. Gustqfson et al. /Mechanisms

References Allenby, G., Bocquel, M.-T., Saunders, M., Kazmer, S., Speck, J., Rosenberger. M., Lovey, A., Grippo, J.F., Chambon, P. and Levin, A.A. (1992) Proc. Natl. Acad. Sci. USA 90,30-34. Baeuerlen, P.A. (1991) Biochem. Biophys. Acta 1072.63-80. Bailey, J. and Siu, C.-H. (1988) J. Biol. Chem. 263,9326-9332. Benbrook, D., Lemhardt, E. and Pfahl, M. (1988) Nature 333,669472. Boylan, J.F. and Gudas, L.J. (1991) J. Cell Biol. 112.965-979. Boylan, J.F. and Gudas, L.J. (1992) J. Biol. Chem. 267, 2148621491. Brand, N., Petkovich, M., Krust, A., Chambon, P., de The, H., Marchio, A., Tiollais, P. and Dejean, A. (1988) Nature 332, 850-853. Busch, C., Saksenam, P., Funa, K., Nordlinder, H. and Eriksson, U. (1990) Methods Enzymol. 189, 315-324. Carlsson, L., Danielsson, P.E., Lenz, R., Liljeborg, A. and Aslund, N. (1985) Optics Letter, 10, 5.5-63. Cordon, R.A (1995) TIGS 11,314-319. Crech Kraft, J., tifberg, B., Chahoud, I., Bochert, G. and Nau, H. (1989) Toxicol. Appl. Pharmacol. 100, 162-176. d' Argy, R., Sperber, G.O., Larsson, B.S. and Ullberg, S. (1990) J. Pharmacol. Methods 24, 165-18 1. de Bruijn. D.R.H., Oerlemans, F., Hendriks, W., Baats, E., Ploemacher, R. and Wieringa, B. (1994) Differentiation 58, 141-148. Dencker, L., d’Argy, R., Danielsson, B.R.G., Ghantous, H. and Sperber, G.O. (1987) Dev. Pharmacol. Ther. 10.212-223. Dencker, L.. AnnerwaII, E., Busch, C. and Eriksson, U. (1990) Development 110,343-352. Dollt, P., Ruberte, E, Kastner, P., Petkovich, M., Stoner, CM., Gudas, L.J. and Chambon, P. (1989) Nature 342.702-705. Doll&, P., Ruberte, E., Leroy, P., Morriss-Kay, G. and Chambon, P. (1990) Development 110, 1133-1151. Doll& P., Fraulob, V., Kastner, P. and Chambon, P. (1994) Mech. Dev. 45,91-104. Donovan, M., Olofsson, B., Gustafsson, A.-L., Dencker, L. and Eriksson, U. (1995) J. Steroid Biochem. Mol. Biol. 53,459-465. Eriksson. U., Hansson, E., Nordlinder, H., Busch, C., Sundelin, J. and Peterson, P.A. (1987) J. Cell. Physiol. 133,482-490. Fiorella, P.D. and Napoli, J.L. (1991) J. Biol. Chem. 266, 1657216579. Fredriksen, K., Jat., J.S.. VaItz, N., Levy, D. and McKay, R. (1988) Neuron 1.439-448. Gerace, L. (1995) Cell 82, 341-344. Giguere, V., Lyn, S., Yip, P.. Siu, C.-H. and Amin, S. (1990) Proc. Natl. Acad. Sci. USA 7,6233-6237. Giguere, V., Ong, ES., Segui, P. and Evans, R.M. (1987) Nature 330, 624-629. Gerry, P., Lufkin, T., Dierich, A., Rochette-Egly, C., Decimo, D., Doll& P., Mark, M., Durand, B. and Chambon, P. (1994) Proc. Natl. Acad. Sci. USA 91.9032-9036. Green, S., lsserman, 1. and Sheer, E. (1988) Nucleic Acids Res. 16,369. Gudas, L., Spom, M.B. and Roberts, A.B. (1994) In Spom, M.B., Roberts, A.B. and Goodman, D.S. (eds.), Retinoids: Biology, Chemistry and Medicine, Raven Press, New York, pp. 443-520. Gustafson, A.-L., Dcncker, L. and Eriksson, U. (1993) Development 117.343-351. Horton, C. and Maden, M. (1995) Dev. Dynam. 202.312-323.

qf Development

58 (1996) 27-38

Kalderon, D., Roberts, B., Richardsson. W. and Smith, A. (1984) Cell 39.499-509. Leid, M., Kastner, P. and Chambon, P. (1992) TIBS 17.427433. Maden, M., Horton, C., Graham, A., Leonard, L., Pizzey, J., Siegenthaler, G. and Eriksson, U. (1991a) Mech. Dev. 37, 13-23. Maden, M., Hunt, P.. Eriksson, U., Kuriowa, A., Krumlauf, R. and Summerbell, D. (1991 b) Development I I 1.3544. Mangelsdorf. D.J., Ong, E.S., Dyck, J.A. and Evans, R.M. (1990) Nature 345.224-229. Mangelsdorf, D.J., Borgmeyer, U., Heyman, R.A., Zhou, J.Y., Ong, E.S., Oro, A.E., Kakizuka, A. and Evans, R.M. (1992) Genes Dev. 6,329-344. Mangelsdorf, D.J., Unesomo, K. and Evans, R.M. (1994) In Spom, M.B., Roberts, A.B. and Goodman, D.S. (eds.), Retinoids: Biology, Chemistry and Medicine, Raven Press, New York, pp. 283-3 18. Melchior, F. and Gerace, L. (1995) Curr. Opin. Cell Biol. 7, 3 1O-3 18. Mossberg, K., Arvidsson, U. and Ulfhake, B. (1990) J. Histochem. Cytochem. 38, 179-190. Nilsson, M., Spurr, N., Sakena, P., Busch, C., Nordlinder, H., Peterson, P.A and Sundelinz, J. (1988) Eur. J. Biochem. 173,45-51. Ong, D.E., Newcomer, M.E. and Chytil, F. (1994) In Spom, M.B., Roberts, A.B. and Goodman, D.S. (eds.), Retinoids: Biology, Chemistry and Medicine, Raven Press, New York, pp. 283-3 18. Perez-Castro, A.V., Toth-Rogler, L.E., Wei, L.-N. and Nguyen-Hou, M.C. (1989) Proc. Natl. Acad. Sci. USA 86,8813-8817. Petkovich, M., Brand, N.J., Krust, A. and Chambon, P. (1987) Nature 330.444-450. Picard, D. and Yamamoto, K.R. (1989) EMBO J. 6,3333-3340. Roth, S., Stein, D. and Nusslein-Volhard, C. (1989) Cell 59, 1 l891202. Ruberte, E., Dollt, P., Chambon, P. and Morris+Kay, G. (199 I) Development 111,45-60. Ruberte, E., Friederich, V.. Chambon, P. and Morriss-Kay, G. (1993) Development 118, 267-282. Rushlow, CA., Han, K., Manley, J.L. and Levine, M. (1989) Cell 59, 1165-1177. Saari, J., Bredbcrg, D. and Garwin, G. (1982) J. Biol. Chem. 257, 13329-13333. Silver, P.A. (1991) Cell 64,489-497. Steward, R. (1989) Cell 59, 1179-I 188. Sundelin, J. Das, S.R., Eriksson, I_.. Rask, L. and. Petersson, P.A. (1985) J. Biol. Chem. 260,6494-6499. Takase, S., Ong, D.E. and Chytil, F. (1986) Arch. Biochem. Biophys. 247,328-334. Thaller, C. and Eichele, G. (1987) Nature 327, 625-628. Ullberg, S. (1977) Science Tools (The LKB Journal), LKB, Bromma, Sweden, pp. l-29. Vaessen, M.-J., Kootwijk, E., Mummery, C., Hilkens, J., Bootsma, D. and van Kessel, A.G. (1989) Differentiation 40,99-105. Vaessen, M.-.J., Meijers, J.H.C., Bootsma, D. and van Kessel, A.G. (1990) Development 110. 371-378. Vandromme, M., Gauthier-Rouvikre, C., Lamb, N. and Fernandez, A. (1996) TIBS 21.59-64. Zelent. A., Krust, A., Petkovich, M., Kastner, P. and Chambon, P. (1989) Nature 339.714-717.