Retinoic acid receptors and retinoid receptors ... · RXRs(a, (, and'y) bind9-cts-RAwithdissociation constants (Kd)of15.7, 18.3,and14.1nM,respectively. Incontrasttothe selectivity

Proc. Nati. Acad. Sci. USAVol. 90, pp. 30-34, January 1993Biochemistry

Retinoic acid receptors and retinoid X receptors: Interactions withendogenous retinoic acids

(afl-ftns-retinoic acid/9-cis-retinoic acid/3,4-didehydroretinoic add/binding affinity/transcriptional activation)

GARY ALLENBY*, MARIE-THERESE BOCQUELt, MICHAEL SAUNDERSt, SONJA KAZMER*, JEFFREY SPECK*,MICHAEL ROSENBERGERf, ALLEN LOVEY*, PHILIPPE KASTNERt, JOSEPH F. GRIPPO*, PIERRE CHAMBONt,AND ARTHUR A. LEVIN*§*Department of Toxicology and Pathology and tDepartment of Medicinal Chemistry, Hoffmann-La Roche, Nutley, NJ 07110; and Institut de ChimieBiologique, Unite 184 de Biologie Mol6culaire et de G6nie G6netique de l'Institut National de la Sante et de la Recherche M6dicale, Laboratoire de G6n6tiqueMol6culaire des Eucaryotes du Centre National de la Recherche Scientifique, Facult6 de M6decine, 11 rue Humann, 67085 Strasbourg Cedex, France

Contributed by Pierre Chambon, August 14, 1992

ABSTRACT The binding of endogenous retinoids andstereoisomers of retinoic add (RA) to the retinoid nuclearreceptors, RA receptor (RARs) and retinoid X receptors(RXRs), was characterized using nucleosol preparations fromtransiently transfected COS-1 cells. Among several stereoiso-mers ofRA tested, including 7-cis-, 9-cis-, 11-cis-, 13-cis-, andall-trns-RA, only 9-cis-RA effectively competes with 9-cis-[3HJRA binding to the RXRs. Additionally, the endogenousretinoid trans-didehydro-RA (t-ddRA) does not interact withRXRs, whereas the 9-cis form of ddRA competes effectively.RXRs (a, (, and 'y) bind 9-cts-RA with dissociation constants(Kd) of 15.7, 18.3, and 14.1 nM, respectively. In contrast to theselectivity of RXRs for 9-cis-RA, RARs bind both t-RA and9-cis-RA with high affinity, exhibitng Kd values in the 0.2-0.7nM range for both ligands. Unlike RARs, the cellular RAbindig proteins CRARPI or CRABPH bind t-RA but do notbind 9-cis-RA. Consistent with the binding data, 9-cis-RA and9-cis-ddRA tnscriptionally activate both GAL4-RXR andGAL4-RAR chimeric receptors with ECso values of 3-20 nMfor 9-cis-RA and 9-cis-ddRA, whereas t-RA and t-ddRAefficiently activate only GAIA-RAR chimeric receptors. Thus,9-cis forms of en s retinoids can contribute to thepleiotropic effects of retinoids by interacting with both theRARs and RXRs.

Retinoids have a broad spectrum of biological activities ingrowth and differentiation of epithelia (1), embryonic devel-opment (2), and spermatogenesis (3). These effects arethought to result from interactions of retinoids with nuclearreceptors (4, 5) that are members of the steroid-thyroidhormone superfamily ofreceptors and as such are consideredto be ligand-dependent transcription factors (6, 7). Oneexplanation for the diversity of retinoid action resides in themultiplicity of nuclear receptors (for a review, see ref. 8).Two families of nuclear retinoid receptors have been de-scribed, the retinoic acid receptors (RARs) (4, 5, 9, 10) andthe retinoid X receptors (RXRs) (11-13). These two retinoidreceptor families are divergent, sharing only 29%6 homologyin their ligand binding domains (8, 11), and may control in theform ofRAR-RXR heterodimers the expression ofRA targetgenes (12, 14-16 and references therein). They may alsoregulate separate gene pathways through distinct responseelements (17, 18). Furthermore, within the families of recep-tors there may be preferential activation of specific respon-sive promoters by the different receptor types (a, ,B, or y)(19). This multiplicity of receptors and gene pathways may inpart explain the pleiotropic effects of retinoids for the controlof a wide variety of cellular processes.

Another factor controlling the pleiotropic effects of retin-oids may be the existence of multiple ligand pathways for theretinoid receptors. The RARs directly bind all-trans-RA(t-RA) and as a consequence of this binding become tran-scriptionally active. The RXRs differ from the RARs in thatthey are incapable of binding t-RA, but they bind and areactivated by the 9-cis stereoisomer of RA (20, 21). The3,4-didehydro form of t-RA (t-ddRA) is also biologicallyactive and is found in the chicken limb (22) and in mammaliantissues (23, 24). It is not known whether t-ddRA has a uniquereceptor or whether its effects are mediated by interactionswith one of the RARs or RXRs.

In the present investigation, we examine the ligand spec-ificity of the receptors and the cellular RA binding proteins(CRABPs) for RA stereoisomers and other naturally occur-ring retinoids. The results demonstrate that the RXRs areselective for 9-cis-RA and 9-cis-ddRA but that the RARseffectively bind and are activated by 9-cis-RA as well ast-RA. These results suggest a role for 9-cis-RA in the induc-tion of RA target gene responses through RAR, RXR, orRAR-RXR heterodimer pathways.

MATERIALS AND METHODSMaterials. 9-cis-[10-3H]RA and the unlabeled RA isomers

were synthesized by the Department ofMedicinal Chemistry,Hoffmann-La Roche, and all-trans-[11,12-3H2JRA was pur-chased from NEN. HPLC analysis determined the purity ofthese reagents at >98%.

Transfections, Nucleosol Preparations, and Binding Assays.COS-1 cells were transfected by electroporation with pSG5expression vectors (25) containing cDNAs for mouse RARsa, f3, or y; RXRs a, ,B, or y; or CRABPI or CRABPII asdescribed (20). Nucleosol or cytosol fractions were prepared(26, 27) and stored at -80°C until use. Aliquots of nucleosolor cytosol were incubated in nuclei lysis buffer (26) withtritiated ligands for 4 h at 4°C. Retinoids were added inethanolic solutions that did not exceed 2% of the totalincubation volume. For competitive binding assays, theincubations were performed with increasing concentrationsof unlabeled competing ligand and a fixed concentration ofthe radioligand (10 nM 9-cis-[3H]RA). In saturation kineticsstudies, incubations were performed in the presence ofincreasing concentrations ofthe indicated radioligand. For allbinding assays, bound was separated from free radioactivityas described (20). Binding in the presence of a 100-fold molarexcess of unlabeled ligand was defined as nonspecific bind-

Abbreviations: RA, retinoic acid; dd, 3,4-didehydro; t, all-trans;RAR, RA receptor; RXR, retinoid X receptor; CRABP, cellular RAbinding protein; CAT, chloramphenicol acetyltransferase.§To whom reprint requests should be addressed.

30

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Dow

nloa

ded

by g

uest

on

Mar

ch 1

8, 2

020

Proc. Natl. Acad. Sci. USA 90 (1993) 31

ing. Specific binding was defined as total binding minusnonspecific binding.The concentration required to produce a 50% inhibition of

binding (IC50) was calculated from competition data (28).Saturation kinetics data were analyzed as described (20, 29,30).Chimeric Receptors and Transactivation Assays. GAL4-

RAR a, 13, and y chimeras were constructed by PCR ampli-fication of the D to F regions of mouse RAR a, 13, and y (31)to give DNA fragments with a 5' Xho I site and a 3' Kpn I sitethat were subcloned into the expression vector pGAL4(1-147)ER region F (32). The encoded chimeric GAL4-RARproteins, therefore, consist of GAL4 amino acids 1-147, ashort linker region encoding amino acids IGRPPRA, followedby the RAR DEF regions linked to region F of humanestrogen receptor amino acids 553-595 by 2 amino acids(GT), corresponding to the Kpn I site (see Fig. 4). TheGAL4-RXR a, f3, and y chimeras were constructed by PCRamplification of the DE regions of mouse RXR a, 13, and y(12), followed by subcloning into GAL4MpolyII (33). Thelinker region joining the GAL4 DNA binding region to RXRsequences encodes the amino acids IPRA (see Fig. 4). Thechimeric constructs were verified by DNA sequence analysisand Western blot analysis after expression in transientlytransfected COS-1 cells.HeLa cells were transfected as described (33) with 50 ng of

a given GAL4 chimera, 500 ng of the reporter gene 17M/ERE-G.CAT, 2 ,ug of pCH110, and 12.5 1ug of BSM13+ ascarrier. Twenty-four hours after transfection, cells weretreated for 12-15 h with various concentrations of t-RA,9-cis-RA, t-ddRA, or 9-cis-ddRA as indicated. The relativelyshort treatment with ligand was to minimize isomerization ofRA or ddRA (20). Chloramphenicol acetyltransferase (CAT)activity was determined as described after correction forvariations in transfection efficiency (33).

RESULTSWe have previously demonstrated (20) that when COS-1 cellsare cultured for 48 h with t-RA, the nuclei, enriched bytransfection with RXRs, selectively trap 9-cis-RA. This ste-reoisomer binds RXRs directly and activates these receptors(20, 21), suggesting a role for stereoisomers ofRA in definingligand pathways for retinoids. To determine whether otherstereoisomers of RA are ligands for the RXRs, competitivebinding assays were performed under equilibrium conditionsin which binding was proportional to nucleosol concentrationand remained constant from 5 min to 16 h ofincubation (G.A.,unpublished data).The 7-cis and 13-cis isomers of RA produce little or no

inhibition of 9-cis-[3H]RA binding to the RXRs so that IC50values for these compounds are greater than the maximumconcentrations of 50,000 nM tested. The li-cis isomer ofRAinhibits the binding of 9-cis-[3H]RA to the RXRs, but theinhibition is weak and the concentrations required to producea50% inhibition ofbinding are in the range of5000-12,000 nM

Table 1. IC50 values for competitive binding of 9-cis-[3H]RA tomouse RXRs by isomers of RA

IC50 value, nM

Ligand mRXRa mRXR/3 mRXRy7-cis-RA >50,000 >50,000 >50,0009-cis-RA 73 1 117 ± 21* 85 + 211-cis-RA 4940 ± 1540* 12,550 + 2120* 5430 ± 139013-cis-RA >50,000 >50,000 >50,000t-RA >50,000 >50,000 >50,000Values are mean + SEM for two experiments except where

indicated by an asterisk (n = 3). m, Mouse.

(Table 1). The IC50 values for 11-cis-RA were essentiallyunchanged with increasing durations of incubation from 1 to16 h. If 11-cis-RA were being isomerized over time to9-cis-RA, then the IC50 values would be expected to decreasewith increasing incubation times (data not shown). Althoughil-cis-RA is only a weak competitor for 9-cis-RA binding tothe RXRs, it competes more efficiently for binding to theRXRa and RXRy forms than to the RXR/3 form (Table 1),suggesting that there may be differences in the ligand bindingcharacteristics of the RXRs.Of the RA isomers investigated, 9-cis-RA most effectively

competes for 9-cis-[3H]RA binding to mRXRa, ,B, or y (Table1). Interestingly, the 9-cis form of ddRA (9-cis-ddRA) com-petes for 9-cis-[3H]RA binding to RXRs a, 1, and 'y with IC50values of 210, 172, and 142 nM, respectively (data notshown). While the 9-cis form of ddRA is an effective com-petitor, the all-trans form of ddRA, like t-RA, does notproduce significant inhibition of 9-cis-[3H]RA binding toRXRs a, 1, or y at concentrations up to 50,000 nM (data notshown). Thus, the RXRs demonstrate a striking selectivityfor 9-cis isomers of RA and ddRA.We characterized the binding affinities of 9-cis-RA to

RXRs directly by saturation kinetics and Scatchard analysis.The binding of 9-cis-RA to RXRs (Fig. 1 A-C) is specificand saturable. Scatchard analyses of the binding of 9-cis-RAto RXRs a, 1, and y yield Kd values of 15.7, 18.3, and 14.1nM, respectively (Fig. 1 D-F). These data indicate that9-cis-RA binds with relatively high affinity to all three mouseRXRs. Human RXRa binds 9-cis-RA with Kd values in the 10nM range (20, 21).Because 9-cis-RA had been shown to activate both RXRa

and RARa (21), we sought to determine whether 9-cis-RAdirectly bound to RARs. Interestingly, we found that9-cis-RA binds to RARs with high affinity. Scatchard anal-yses of the binding of 9-cis-[3H]RA to RARs yield Kd valuesin the range of 0.2-0.7 nM (Fig. 2), indicating that theaffinities for the binding of 9-cis-RA to RARs are at least20-fold higher than those for the binding of9-cis-RA to RXRs(Table 2). The human RARs bind 9-cis-RA with high affinitiessimilar to those reported here for the mouse receptors (G.A.,unpublished data). When the Kd values for the binding oft-RA and 9-cis-RA to mouse RARs are compared (Table 2 anddata not shown), they are remarkably similar, considering thestereochemistry of the two ligands. Only RARy seems to

600 RXRax A 40 RXRa D

°00 130 d15.7nM

3120A0AA10

400RXR ( B 12 RXR ( EFG1Strikitcf hbnno cs3] tE~~~~~~~~

U200o ~~ ~~~~~4AKd=18.3nM

500~~~~~~2300- 12.

100.

10 30 50 70 100 300 500Concentration (nM) Bound (fmnol)

FIG. 1. Saturation kinetics for the binding of 9-Cis-[3H]RA toRXRs a (A), (3 (B), and y (C). Specific binding (A) is defined as thetotal binding (o) minus the nonspecific binding (o). Scatchard anal-ysis of the saturation kinetic data for mouse RXRs a (D), , (E), andy (F). Dissociation constants (Kd) are displayed.

Biochemistry: Allenby et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 1

8, 2

020

32 Biochemistry: Allenby et al.

I_, G-Q - El N

2 4 6 8 10 200 400Concentration (nM) Bound (fmol)

FIG. 2. Saturation kinetics for the binding of 9-cis-[3H]RA toRARs a (A), P (B), and fy (C). Specific binding (v) is defined as thetotal binding (o) minus nonspecific binding (o). Scatchard analysis ofthe saturation kinetic data for mouse RARs a (D), ( (E), and y (F).Dissociation constants (Kd) are displayed.

show a slight difference in Kd values for the two ligands,exhibiting an 3-fold higher affinity for t-RA compared to-9-cis-RA.We also examined the binding of 9-cis-RA to other RA

binding proteins. CRABPs exist in two forms, CRABPI andCRABPII (ref. 34 and references therein). These proteinsmay play a role in modulating free concentrations of RAeither by sequestration or by participating in the metabolicdegradation of RA (35-37). We examined the binding oft-[3H]RA or 9-cis-[3H]RA to the CRABPs by incubatingcytosolic fractions from COS-1 cells transfected with thecDNAs for CRABPI, CRABPII, or parental expression vec-tor (mock). Transfection with CRABPI or CRABPII pro-duces a marked increase in t-[3H]RA binding compared tomock-transfected cells. However, there is little or no directbinding of 9-cis-[3H]RA to either CRABPI or CRABPII.Thus, in contrast to the RARs, which bind both 9-cis-RA andt-RA with approximately equal affinities, CRABPI andCRABPII bind t-RA but not 9-cis-RA (Fig. 3). These data donot exclude the possibility that CRABPs may bind 9-cis-RAwith very low affinity (B. P. Sani, J.F.G., and A.A.L.,unpublished results).To determine whether the binding specificity of the RXRs

for 9-cis-RA and the apparent high affinity of the RARs for9-cis-RA were correlated with transcriptional activation ofthe receptors, chimeric receptors were employed to limit theinterference from endogenous retinoid receptors. Chimericreceptors contained the DNA binding domain of the yeasttransactivator GAL4 [GAL4-(1-147)] fused to the C-terminal

Table 2. Dissociation constants for the binding of 9-cis-[3H]RAor t-[3H]RA to mouse RXRs and RARs

Kd, nM

Receptor type 9-cis-RA t-RA

mRXRa 15.7 NBmRXR(3 18.3 NBmRXRy 14.1 NBmRARa 0.24 0.20mRARI3 0.40 0.36mRARy 0.70 0.20

NB, no binding; m, mouse.

0300L0IiCD100~

CRABPI CRABPII' MOCK

FIG. 3. Binding oft-[3H]RA (hatched bars) or qcis-[3H]RA (solidbars) in cytosolic fractions isolated from COS-1 celis transfected withmouse CRABPI, mouse CRABPII, or parental expression plasmid.

regions ofRXR (regions D and E) or RAR (regions D, E, andF), which contain both a ligand binding and a transactivationdomain (ref. 19 and Fig. 4).These chimeras bind efficiently as homodimers to a 17-mer

GAL4 response element to which endogenous retinoid re-ceptors cannot bind (M.S., unpublished results). Cotransfec-tion of HeLa cells with a given GAL4-RXR or GAL4-RARchimera and the 17M/ERE-G.CAT reporter gene (33) resultsin a retinoid-concentration-dependent increase of CAT ac-tivity. For all three RXR chimeras, 9-cis-RA is much moreefficient than either t-RA or t-ddRA at activating transcrip-tion from the reporter gene (Fig. 5 A-C). The EC56 values foractivation by 9-cis-RA are 7 nM for GAL4-RXRa, 20 nM forGAL4-RXR(3, and 10 nM for GAL4-RXR'y. In all threecases, t-RA and t-ddRA are much less potent than 9-cis-RA,and no significant activation is seen at 10 nM. Of particularinterest, the 9-cis stereoisomer of ddRA (9-cis-ddRA) acti-vates the three GAL4-RXR chimeras with EC50 valuessimilar to those of 9-cis-RA (M.-T.B., unpublished data).Clearly the 9-cis forms ofRA and ddRA are active Zgands forthe RXRs exhibiting a good correlation between binding andtransactivation.The high affinity of 9-cis-RA for the RARs in binding

assays suggested that 9-cis-RA should be a potent activatorof these receptors. In fact, 9-cis-RA activates the RARs atleast as potently as t-RA. Additionally, t-RA, t-ddRA, 9-cis-RA, and 9-cis-ddRA have very similar effects (Fig. SD-Fanddata not shown). The 50%o maximal response values for9-cis-RA (or 9-cis-ddRA) and t-RA (or t-ddRA) are 10 nM and

1 147 206

I GAEm466

mRXRa (DE) GAL-RXRa

1 147 186 448

GAL4 M mRXRO (DE) GAL-RXRP

1 147 205N r

463

L UAL4 * mRXRy (DE) I GAL-RXRy

1 147 154L %

462 553 595'

| GAL4 6 mRARa (DEF) M hER(F)F

IGRPPRA GT

1 147 147 448 553 595

GAL4 i mRARP (DEF) hER(F)i

1 147 156 458 553 595

I GAL4 M mRARy (DEF) M hER(F)|

GAL-RARa

GAL-RARP

GAL-RARy

FIG. 4. Schematic representation ofthe GAL4-RXR and GAL4-RAR chimeric constructs. m, Mouse; hER, human estrogen recep-tor.

Proc. Natl. Acad. Sci. USA 90 (1993)

Dow

nloa

ded

by g

uest

on

Mar

ch 1

8, 2

020

Proc. Natl. Acad. Sci. USA 90 (1993) 33

11 9 7 5 11 9 7 5-Log concentration (M)

FIG. 5. Relative efficiency of t-RA (o), 9-cis-RA (*), and t-ddRA(o) at inducing transcriptional activation by GAL4-RXR and GAL4-RAR chimeric constructs. The chimeric constructs and the appro-priate reporter were transiently transfected into HeLa cells. For theGAL4-RXR chimeras, maximal activity in the presence of 9-cis-RAwas taken as 100%, whereas for the GAL4-RAR chimeras, 100%oCAT activity corresponded to the maximal activity produced byt-RA.

100 nM for GAL4-RARa, 3 nM and 5 nM for GAL4-RAR,8,and 20 nM and 6 nM for GAL4-RARy, respectively. GAL4-RARa and GAL4-RAR/B are more efficiently activated by9-cis-RA or 9-cis-ddRA than by t-RA or t-ddRA at all ligandconcentrations. In contrast, while 9-cis-RA (or 9-cis-ddRA) isless potent than t-RA in activating GAL4-RARy, it exhibitsa greater efficacy in that it induces a stronger maximalresponse (Fig. 5 D-F). These data confirm the binding data,which indicate that the affinity ofRARy for 9-cis-RA is lowerthan for t-RA.While there was a good correlation in the data for 9-cis-RA

binding and transactivation in the case ofRXRs, there was anotable lack of correlation between binding affinities andEC50 values for transactivation in the case of RARs (Table 2and Fig. 5). The binding affinities of RARs for 9-cis-RA andt-RA are in the range of 0.2-0.7 nM (Table 2), yet the EC50values for transcriptional activation are in the range of 3-100nM (Fig. 5). Previous studies on RA-induced transcriptionalactivation of wild-type RARs (21, 38) and other RAR chime-ras (39) have yielded similar EC50 values; thus the differencein Kd and EC50 values does not appear to be related to theGAL4 chimeras employed here. The magnitude of this dis-crepancy is most profound with RARa and corresponds to atwo order of magnitude difference between binding andtransactivation data. At this time, we do not understand themolecular basis for this difference. One interesting possibility(S. Nagpal and P.C., unpublished data) might be that, withinHeLa cells, GAL4-RARa heterodimerizes with RXRs. Con-ceivably, this heterodimer may bind and be activated betterby 9-cis-RA than t-RA. This would suggest that the lowpotency of t-RA activation of GAL4-RARa reported heremay depend on its conversion to 9-cis-RA. Such an expla-nation might account for the difficulty in reaching a maximumresponse with RARa by using t-RA (Fig. 5D, and see ref. 21).

DISCUSSION

We extended our previous observation that 9-cis-RA is an

endogenous ligand for RXRa (20) by demonstrating that 9-cis

forms of the endogenous retinoids RA and ddRA bind toRXRs a, /3, and y with relatively high affinities. RXRs areremarkably selective in that they do not bind the trans formsofRA and ddRA, nor do they bind other stereoisomers ofRAincluding 7-cis-, 11-cis-, and 13-cis-RA. These data clearlyindicate that the 9-cis conformation is essential for thebiological activity mediated by RXRs. An unexpected resultfrom these experiments is that 9-cis forms of RA and ddRAcan also strongly interact with the RARs. Scatchard analysesdemonstrate that 9-cis-RA and t-RA bind to RARs withsimilar high affinities. Additionally, we demonstrate that9-cis-RA activates all three RARs with high potency. In fact,9-cis-RA is more potent than t-RA for activating RARa andRAR/3 chimeric constructs. The RARy construct was some-what different in that t-RA yielded a slightly lower EC50value, but 9-cis-RA induced a greater maximum response.Thus the 9-cis conformation of the endogenous retinoidstested appears to be important for the activity of both familiesof retinoid receptors, RXRs and RARs. We propose that 9-cisforms of endogenous retinoids are the cognate ligands thatactivate RXR-mediated pathways. However, if we considerRAR-mediated activity, our data suggest that both the transand the 9-cis forms can function as active ligands.Recent data indicate that the RXRs form heterodimers with

RARs and other receptors in the steroid/thyroid superfamily.Additionally, the RXRs coregulate the activity of RARs ontheir response elements (see the Introduction for references).RAR-RXR heterodimers are interesting protein complexes inrelation to the observation that endogenous 9-cis-retinoidscan bind both RARs and RXRs. We can speculate that therewould exist a number of ligand-receptor interactions involv-ing both t-RA and 9-cis-RA. For example, the ligand bindingdomains within an RAR-RXR heterodimer could be occupiedby only 9-cis-RA or by both 9-cis-RA and t-RA. Bindingassays presented here were performed under conditions thatfavor monomers of RAR (26). Under these conditions, wefind that 9-cis-RA binds to the RARs with an =20-fold greateraffinity -than to the RXRs. If the RXR-RAR heterodimersbind 9-cis-RA in a similar fashion, then we would expect that9-cis-RA would compete with t-RA for binding to this com-plex. However, it is possible that heterodimerization canalter receptor binding affinities because the dimerizationdomain of these receptors overlaps with the ligand bindingdomain (12, 14).One interesting question that arises from these data con-

cerns the availability of 9-cis-RA at its cellular site of action.The concentration oft-RA in tissues and serum is in the rangeof 5-10 nM (2, 40, 41). The available data indicate that the9-cis-RA concentrations were 3-fold lower than availablet-RA concentrations in several tissues (21), which suggeststhat 9-cis-RA concentrations would be in the low nanomolarrange. According to the data reported here, these levels arelower than the Kd values for the binding of9-cis-RA to RXRs,but they are in the range of the Kd values for the binding of9-cis-RA to RARs. However, recent data indicate that nano-molar concentrations of 9-cis-RA promote RXR homodimerformation, leading to the activation of several responseelements (42). Conceivably, the binding properties of recep-tors in whole cells may differ from those of receptors in ournucleosol preparations yielding Kd values more in the rangeof the endogenous concentrations of 9-cis-RA.

Synthesis of t-RA from retinol has been implicated as onefactor in controlling t-RA concentration (43-45), but thesource of 9-cis-RA is less clear. Isomerization from t-RA isa possible route, or 9-cis-retinol precursors in tissues (46-48)may be oxidized. Another factor is the presence of CRABPsin the cell. The CRABPs are thought to control the levels offree t-RA by sequestration (35) and may play a role in thedegradation of t-RA by microsomal enzymes (36, 37). Wehave shown that 9-cis-RA does not bind to the CRABPs.

Biochemistry: Allenby et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 1

8, 2

020

Proc. Nadl. Acad. Sci. USA 90 (1993)

CRABP binding may directly modulate the concentration offree t-RA, but not that of 9-cis-RA. However, if 9-cis-RA isa product of the isomerization of t-RA, then CRABPs couldindirectly control the level of 9-cis-RA.We have demonstrated that there is an intricate pattern of

endogenous RA interactions with the nuclear retinoid recep-tors and CRABPs. This ligand pattern combined with themultiplicity of receptor types and isoforms (8, 49) and thedifferential transactivation of RA responsive promoters bythese receptors (19) yield a system with sufficient combina-torial factors to explain retinoid control ofcomplex biologicalprocesses (for further discussion, see ref. 8).

We thank the Isotope Synthesis Group, Hoffmann-La Roche,Nutley, NJ, for the synthesis of q-cis-(3H]RA and Debra Lucas forpreparation of receptor plasmids. This work was supported inStrasbourg by funds from the Institut National de la Sante et de laRecherche Mddicale, the Centre National de la Recherche Scienti-fique, the Centre Hospitalier Universitaire R6gional, the Associationpour la Recherche sur le Cancer, the Human Science FrontierProgram, and the Fondation pour la Recherche Midicale.

1. Roberts, A. B. & Sporn, M. B. (1984) in The Retinoids, eds.Sporn, M. B., Roberts, A. B. & Goodman, DeW. S. (Academ-ic, Orlando, FL), Vol. 2, pp. 209-286.

2. Thaller, C. & Eichele, G. (1987) Nature (London) 327,625-628.3. Thompson, J. N., Howell, J. M. & Pitt, G. A. J. (1964) Proc.

R. Soc. London Biol. Sci. 159, 510-535.4. Petkovich, M., Brand, N. J., Krust, A. & Chambon, P. (1987)

Nature (London) 330, 444-450.5. Giguere, V., Ong, E. S., Segui, P. & Evans, R. M. (1987)

Nature (London) 330, 624-629.6. Evans, R. M. (1988) Science 240, 889-895.7. Green, S. & Chambon, P. (1988) Trends Genet. 4, 309-314.8. Leid, M., Kastner, P. & Chambon, P. (1992) Trends Biochem.

Sci. 17, 427-433.9. Brand, N., Petkovich, M., Krust, A., Chambon, P., de The, H.,

Marchio, A., Tiollais, P. & Dejean, A. (1988) Nature (London)332, 850-853.

10. Krust, A., Kastner, P., Petkovich, M., Zelent, A. & Chambon,P. (1989) Proc. Nati. Acad. Sci. USA 86, 5310-5314.

11. Mangelsdorf, D. J., Ong, E. S., Dyck, J. A. & Evans, R. M.(1990) Nature (London) 345, 224-229.

12. Leid, M., Kastner, P., Lyons, R., Nakshatri, H., Saunders, M.,Zacharewski, T., Chen, J.-Y., Staub, A., Gamier, J.-M.,Mader, S. & Chambon, P. (1992) Cell 68, 377-395.

13. Mangelsdorf, D. J., Borgmeyer, U., Heyman, R. A., Zhou,J. Y., Ong, E. S., Oro, A. E., Kakizuka, A. & Evans, R. M.(1992) Genes Dev. 6, 329-344.

14. Yu, V. C., Delsert, C., Andersen, B., Holloway, J. M., De-vary, 0. V., Nair, A. M., Kim, S. Y., Boutin, J.-M., Glass,C. K. & Rosenfeld, M. G. (1991) Cell 67, 1251-1266.

15. Marks, M. S., Hallenbeck, P. L., Nagata, T., Segars, J. H.,Appella, E., Nikodem, V. M. & Ozato, K. (1992) EMBO J. 11,1419-1435.

16. Durand, B., Saunders, M., Leroy, P., Leid, M. & Chambon, P.(1992) Cell 71, 73-85.

17. Mangelsdorf, D. J., Umesono, K., Kliewer, S. A., Borgmeyer,U., Ong, E. S. & Evans, R. M. (1991) Cell 66, 555-561.

18. Rottman, J. N., Widom, R. L., Nadal-Ginard, B., Mahdavi, V.& Karathanasis, S. K. (1991) Mol. Cell. Biol. 11, 3814-3820.

19. Nagpal, S., Saunders, M., Kastner, P., Durand, B., Nakshatri,H. & Chambon, P. (1992) Cell 70, 1007-1019.

20. Levin, A. A., Sturzenbecker, L. J., Kazmer, S., Bosakowski,T., Huselton, C., Allenby, G., Speck, J., Kratzeisen, C.,Rosenberger, M., Lovey, A. & Grippo, J. F. (1992) Nature(London) 355, 359-361.

21. Heyman, R. A., Mangelsdorf, D. J., Dyck, J. A., Stein, R. B.,Eichele, G., Evans, R. M. & Thaller, C. (1992) Cell 68, 397-406.

22. Thaller, C. & Eichele, G. (1990) Nature (London) 345, 815-819.23. Tdrma, H. & Vahlquist, A. (1985) J. Invest. Dermatol. 85,

498-500.24. Tdrma, H. & Vahlquist, A. (1988) Biochim. Biophys. Acta 961,

177-182.25. Green, S., Isseman, I. & Scheer, E. (1988) Nucleic Acids Res.

16, 369.26. Nervi, C., Grippo, J. F., Sherman, M. I., George, M. D. &

Jetten, A. M. (1989) Proc. Natl. Acad. Sci. USA 86,5854-5858.27. Nervi, C., Poindexter, E. C., Grignani, F., Pandolfi, P. P.,

LoCoco, F., Avvisati, G., Pelicci, P. G. & Jetten, A. M. (1992)Cancer Res. 52, 3687-3692.

28. DeLean, A., Munson, P. J. & Robard, D. (1978) Am. J.Physiol. 235, E97-E102.

29. Grippo, J. F. & Gudas, L. J. (1987) J. Biol. Chem. 262,4492-4500.

30. Scatchard, G. (1949) Ann. N. Y. Acad. Sci. 51, 660-672.31. Zelent, A., Krust, A., Petkovich, M., Kastner, P. & Chambon,

P. (1989) Nature (London) 33, 714-717.32. Webster, N. J., Green, S., Tasset, D., Ponglikitmongkol, M. &

Chambon, P. (1989) EMBO J. 8, 1441-1446.33. Tora, L., White, J., Brou, C., Tasset, D., Webster, N., Scheer,

E. & Chambon, P. (1989) Cell 59, 477-487.34. Giguere, V., Lyn, S., Yip, P., Siu, C.-H. & Amin, S. (1990)

Proc. Natl. Acad. Sci. USA 87, 6233-6237.35. Boylan, J. F. & Gudas, L. J. (1991) J. Cell Biol. 112, 965-979.36. Fiorella, P. D. & Napoli, J. L. (1991) J. Biol. Chem. 266,

16572-16579.37. Napoli, J. L., Posch, K. P., Fiorella, P. D. & Boerman, M. H.

(1991) Biomed. Pharmacother. 45, 131-143.38. Astrom, A., Petterson, U., Krust, A., Chambon, P. &

Voorhees, J. J. (1990) Biochem. Biophys. Res. Commun. 173,339-345.

39. Graupner, G., Malle, G., Maignan, J., Lang, G., Prunitras, M.& Pfahl, M. (1991) Biochem. Biophys. Res. Commun. 179,1554-1561.

40. Napoli, J. L., Pramanik, B. C., Williams, J. B., Dawson, M. I.& Hobbs, P. D. (1985) J. Lipid Res. 26, 387-392.

41. Vahlquist, A. (1982) J. Invest. Dermatol. 79, 89-93.42. Zhang, T. K., Lehman, J., Hoffmann, B., Dawson, M. I.,

Cameron, J., Graupner, G., Hermann, T., Tran, P. & Pfahl, M.(1992) Nature (London) 358, 587-591.

43. Napoli, J. L. (1986) J. Biol. Chem. 261, 13592-13597.44. Connor, M. J. & Smit, M. H. (1987) Biochem. J. 244, 489-492.45. Posch, K. C., Boerman, M. H. E. M., Burns, R. D. & Napoli,

J. L. (1991) Biochemistry 30, 6224-6229.46. Ben-Amotz, A., Mokady, S. & Avron, M. (1988) Br. J. Nutr.

59, 443-449.47. Ben-Amotz, A., Mokady, S., Edelstein, S. & Avron, M. (1989)

J. Nutr. 119, 1013-1019.48. Stahl, W., Schwarz, W., Sundquist, A. R. & Sies, H. (1992)

Arch. Biochem. Biophys. 294, 173-177.49. Chambon, P., Zelent, A., Petkovich, M., Mendelsohn, C.,

Leroy, P., Krust, A., Kastner, P. & Brand, N. (1991) inRetinoids Ten Years On, ed. Saurat, J.-H. (Karger, Basel), pp.10-27.

34 Biochemistry: Aflenby et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 1

8, 2

020