Solid-state NMR analysis of ligand--receptor interactions reveals an induced misfit in the binding site of isorhodopsin

Rhodopsin is the photosensitive protein of the rod photoreceptor in the vertebrate retina and is a paradigm for the superfamily of G-protein-coupled receptors (GPCRs). Natural rhodopsin contains an 11-cis-retinylidene chromophore. We have prepared the 9-cis analogue isorhodopsin in a natural membrane environment using uniformly (13)C-enriched 9-cis retinal. Subsequently, we have determined the complete (1)H and (13)C assignments with ultra-high field solid-state magic angle spinning NMR. The 9-cis substrate conforms to the opsin binding pocket in isorhodopsin in a manner very similar to that of the 11-cis form in rhodopsin, but the NMR data reveal an improper fit of the 9-cis chromophore in this binding site. We introduce the term "induced misfit" to describe this event. Downfield proton NMR ligation shifts (Deltasigma(lig)(H) > 1 ppm) are observed for the 16,17,19-H and nearby protons of the ionone ring and for the 9-methyl protons. They provide converging evidence for global, nonspecific steric interactions between the chromophore and protein, and contrast with the specific interactions over the entire ionone ring and its substituents detected for rhodopsin. The Deltasigma(lig)(C) pattern of the polyene chain confirms the positive charge delocalization in the polyene associated with the protonation of the Schiff base nitrogen. In line with the misalignment of the ionone ring, an additional and anomalous perturbation of the (13)C response is detected in the region of the 9-cis bond. This provides evidence for strain in the isomerization region of the polyene and supports the hypothesis that perturbation of the conjugation around the cis bond induced by the protein environment assists the selective photoisomerization.     

Fluoroquinacrine binding to nucleic acids: Investigation of the 19F-nmr,optical, and fluorescence properties in the presence of DNA,poly(A), and tRNA

The interaction of fluoroquinacrine, 3-fluoro-7-chloro-9-(diethylamino-1-methylbutyl-amino)acridine, with poly(A), DNA, and tRNA has been investigated by monitoring changes in the ¹⁹F-nmr properties, the fluorescence, and the optical absorbance of the drug. The changes in the properties of fluoroquinacrine in the presence of nucleic acids are similar to those observed for quinacrine and suggest that the drugs bind in a similar fashion. The molecular dynamics of fluoroquinacrine bound to nucleic acids were determined by interpreting the data from a number of different nmr relaxation experiments with a two-correlation-time model. The two motions are the long-range bending motion of the drug-nucleic acid complex and the sliding of the drug between the base pairs. Both dipolar and chemical shift anisotropy contributions to the nmr relaxation parameters were taken into consideration. The binding of fluoroquinacrine to tRNA appears to be different from that observed for binding to DNA. Optical absorbance and ¹⁹F-nmr were also used to examine the helix-to-coil transitions of the drug–nucleic acid complexes. In the DNA complex, the ¹⁹F chemical shift changes parallel the absorption changes that occur during the transition. ¹⁹F-nmr and absorption show that the drug–tRNA complexes undergo a cooperative helix-to-coil transition, with the drug binding sites melting when the tRNA is 70% denatured.     

19 F-NMR studies of oxygen binding to hemoglobin

The binding of oxygen to hemoglobin has been investigated by ¹⁹F-nuclear magnetic resonance spectroscopy. The ¹⁹F-nmr spectrum of hemoglobin trifluoroacetonylated at cysteine β 93 exhibits chemical shift changes on binding of ligands, which differ depending on which chains are undergoing complexation. Comparison of these changes to the fractional ligation of all chains, determined concurrently from the fractional change in the visible spectrum, shows that initial oxygen molecules bind preferentially to α-chains. The ¹⁹F-nmr spectrum of partially oxygenated hemoglobin contains resonances at the normal chemical shift positions of the oxygenated and deoxy species, in addition to two small resonances at intermediate positions. Analysis of the relativ magnitudes of these four peaks as functions of oxygen pressure permits identification of the intermediate species     

Purple membrane: color, crystallinity, and the effect of dimethyl sulfoxide

In an effort to understand the nature of chromophore-protein interactions in bacteriorhodopsin (bR), we have reinvestigated dimethyl sulfoxide (DMSO)-induced changes in bR [Oesterhelt et al. (1973) Eur. J. Biochem. 40, 453-463]. We observe that dark-adapted bR (bR560) in aqueous DMSO undergoes reversible transformation to a species absorbing maximally at 480 nm (bR480). Beginning at 40% DMSO, this change results in complete conversion to bR480 at 60% DMSO. The kinetics of the reaction reveal that this transformation takes place predominantly through the all-trans isomeric form of the pigment. Thermal isomerization of the 13-cis chromophore to the all-trans form is, therefore, the rate-limiting step in the formation of bR480 from the dark-adapted bR. As in native bR, the chromophore in bR480 is linked to the protein via a protonated Schiff base, and its isomeric composition is predominantly all-trans. The formation of bR480 is associated with minor changes in the protein secondary structure, and the membrane retains crystallinity. These changes in the protein structure result in a diminished chromophore-protein interaction near the Schiff base region in bR480. Thus, we attribute the observed spectroscopic changes in bR in DMSO to structural alteration of the protein. The 13-cis chromophoric pigment appears to be resistant to this solvent-induced change. The changes in the protein structure need not be very large; displacement of the protein counterion(s) to the Schiff base, resulting from minor changes in the protein structure, can produce the observed spectral shift.     

Photochemistry in dried polymer films incorporating the deionized blue membrane form of bacteriorhodopsin.

The preparation and photochemical properties of dried deionized blue membrane (dIbR600; lambdamax approximately 600 nm, epsilon approximately 54, 760 cm-1 M-1, f approximately 1.1) in polyvinyl alcohol films are studied. Reversible photoconversion from dIbR600 to the pink membrane (dIbR485; lambdamax approximately 485 nm) is shown to occur in these films under conditions of strong 647-nm laser irradiation. The pink membrane analog, dIbR485, has a molar extinction coefficient of approximately 39,000 cm-1 M-1 (f approximately 1.2). The ratio of pink --> blue and blue --> pink quantum efficiencies is 33 +/- 5. We observe an additional blue-shifted species (dIbR455, lambdamax approximately 455 nm) with a very low oscillator strength (f approximately 0.6, epsilon approximately 26,000 cm-1 M-1). This species is the product of fast thermal decay of dIbR485. Molecular modeling indicates that charge/charge and charge/dipole interactions introduced by the protonation of ASP85 are responsible for lowering the excited-state all-trans --> 9-cis barrier to approximately 6 kcal mol-1 while increasing the corresponding all-trans --> 13-cis barrier to approximately 4 kcal mol-1. Photochemical formation of both 9-cis and 13-cis photoproducts are now competitive, as is observed experimentally. We suggest that dIbR455 may be a 9-cis, 10-s-distorted species that partially divides the chromophore into two localized conjugated segments with a concomitant blue shift and decreased oscillator strength of the lambdamax absorption band.     

Structure of the retinal chromophore in the hRL intermediate of halorhodopsin from resonance raman spectroscopy

Time-resolved resonance Raman spectra of the hRL intermediate of halorhodopsin have been obtained. The structurally sensitive fingerprint region of the hRL spectrum is very similar to that of bacteriorhodopsin's L550 intermediate, which is known to have a 13-cis configuration. This indicates that hRL contains a 13-cis chromophore and that an all-trans----13-cis isomerization occurs in the halorhodopsin photocycle. hRL exhibits a Schiff base stretching mode at 1644 cm-1, which shifts to 1620 cm-1 in D2O. This demonstrates that the Schiff base linkage to the protein is protonated. The insensitivity of the C-C stretching mode frequencies to N-deuteriation suggests that the Schiff base configuration is anti. The 24 cm-1 shift of the Schiff base mode in D2O indicates that the Schiff base proton in hRL has a stronger hydrogen-bonding interaction with the protein than does hR578.     

Protein and ligand adaptation in a retinoic acid binding protein.

A retinoic acid binding protein isolated from the lumen of the rat epididymis (ERABP) is a member of the lipocalin superfamily. ERABP binds both the all-trans and 9-cis isomers of retinoic acid, as well as the synthetic retinoid (E)-4-[2-(5,6,7,8)-tetrahydro-5,5,8,8-tetramethyl-2 napthalenyl-1 propenyl]-benzoic acid (TTNPB), a structural analog of all-trans retinoic acid. The structure of ERABP with a mixture of all-trans and 9-cis retinoic acid has previously been reported. To elucidate any structural differences in the protein when bound to the all-trans and 9-cis isomers, the structures of all-trans retinoic acid-ERABP and 9-cis retinoic acid ERABP were determined. Our results indicate that the all-trans isomer of retinoic acid adopts an 8-cis structure in the binding cavity with no concomitant conformational change in the protein. The structure of TTNPB-ERABP is also reported herein. To accommodate this all-trans analog, which cannot readily adopt a cis-like structure, alternative positioning of critical binding site side chains is required. Consequently, both protein and ligand adaption are observed in the formation of the various holo-proteins.     

The effect of protein conformation change from alpha(II) to alpha(I) on the bacteriorhodopsin photocycle

The bacteriorhodopsin (bR) photocycle was followed by use of time-resolved Fourier-transform infrared (FTIR) spectroscopy as a function of temperature (15-85 degrees C) as the alpha(II) --> alpha(I) conformational transition occurs. The photocycle rate increases with increasing temperature, but its efficiency is found to be drastically reduced as the transition takes place. A large shift is observed in the all-trans left arrow over right arrow 13-cis equilibrium due to the increased stability of the 13-cis isomer in alpha(I) form. This, together with the increase in the rate of dark adaptation as the temperature increases, leads to a large increase in the 13-cis isomer concentration in bR in the alpha(I) form. The fact that 13-cis retinal has a much-reduced absorption cross-section and its inability to pump protons leads to an observed large reduction in the concentration of the observed photocycle intermediates, as well as the proton gradient at a given light intensity. These results suggest that nature might have selected the alpha(II) rather than the alpha(I) form as the helical conformation in bR to stabilize the all-trans retinal isomer that is a better light absorber and is capable of pumping protons.     

The nature of the primary photochemical events in rhodopsin and isorhodopsin.

The nature of the primary photochemical events in rhodopsin and isorhodopsin is studied by using low temperature actinometry, low temperature absorption spectroscopy, and intermediate neglect of differential overlap including partial single and double configuration interaction (INDO-PSDCI) molecular orbital theory. The principal goal is a better understanding of how the protein binding site influences the energetic, photochemical, and spectroscopic properties of the bound chromophore. Absolute quantum yields for the isorhodopsin (I) to bathorhodopsin (B) phototransformation are assigned at 77 K by using the rhodopsin (R) to bathorhodopsin phototransformation as an internal standard (phi R----B = 0.67). In contrast to rhodopsin photochemistry, isorhodopsin displays a wavelength dependent quantum yield for photochemical generation of bathorhodopsin at 77 K. Measurements at seven wavelengths yielded values ranging from a low of 0.089 +/- 0.021 at 565 nm to a high of 0.168 +/- 0.012 at 440 nm. An analysis of these data based on a variety of kinetic models suggests that the I----B phototransformation encounters a small activation barrier (approximately 0.2 kcal mol-1) associated with the 9-cis----9-trans excited-state torsional-potential surface. The 9-cis retinal chromophore in solution (EPA, 77 K) has the smallest oscillator strength relative to the other isomers: 1.17 (all-trans), 0.98 (9-cis), 1.04 (11-cis), and 1.06 (13-cis). The effect of conformation is quite different for the opsin-bound chromophores. The oscillator strength of the lambda max absorption band of I is observed to be anomalously large (1.11) relative to the lambda max absorption bands of R (0.98) and B (1.07). The wavelength-dependent photoisomerization quantum yields and the anomalous oscillator strength associated with isorhodopsin provide important information on the nature of the opsin binding site. Various models of the binding site were tested by using INDO-PSDCI molecular orbital theory to predict the oscillator strengths of R, B, and I and to calculate the barriers and energy storage associated with the photochemistry of R and I for each model. Our experimental and theoretical investigation leads to the following conclusions: (a) The counterion (abbreviated as CTN) is not intimately associated with the imine proton in R, B, or I. The counterion lies underneath the plane of the chromophore in R and I, and the primary chromophore-counterion electrostatic interactions involve C15-CTN and C13-CTN. These interactions are responsible for the anomalous oscillator strength of I relative to R and B.(ABSTRACT TRUNCATED AT 400 WORDS)     

Interaction of aromatic retinal analogues with apopurple membranes of Halobacterium halobium

Absorption spectral properties of aromatic analogues of retinal with apopurple membrane of Halobacterium halobium were studied. The spectra of the all-trans forms were composed of two or more absorption bands. During incubation at 20 degrees C, an absorption band above 500 nm increased in intensity gradually at the expense of an absorption band in the shorter wavelength region with no isomerization of the chromophore. The longer wavelength species was shown to be the protonated form of the shorter wavelength species by changing the pH of the medium. Upon irradiation with blue light, the bandwidth of the spectrum became smaller with isomerization of the chromophore to its 13-cis form. Irreversible binding of protons on the membrane occurred during this process. The rate of the increase in the longer wavelength absorption band was especially low in the reaction with the all-trans form of retinal analogues having a bulky substituent at the para or meta positions of the phenyl ring. In contrast, the 13-cis isomer of aromatic retinal analogues gave a single absorption peak. The extent of the spectral shift upon binding to apopurple membranes was compared over a series of aromatic retinals, and the results were explained in terms of steric interactions of the chromophore with the protein.     

Interaction of the antimalarial drug fluoroquine with DNA,tRNA, and poly(A): 19F-NMR chemical-shift and relaxation,optical absorption,and fluorescence studies

The interaction of the fluorinated antimalarial drug fluoroquine [7-fluoro-4-(diethyl-amino-1-methylbutylamino)quinoline] with DNA, tRNA, and poly(A) has been investigated by optical absorption, fluorescence, and ¹⁹F-nmr chemical-shift and relaxation methods. Optical absorption and fluorescence experiments indicate that fluoroquine binds to nucleic acids in a similar manner to that of its well-known analog chloroquine. At low drug-to-base pair ratios, binding of both drugs appears to be random. Fluoroquine and chloroquine also elevate the melting temperature (T_m) of DNA to a comparable extent. Binding of fluoroquine to DNA, tRNA, or poly(A) results in a downfield shift of about 1.5 ppm for the ¹⁹F-nmr resonance. The chemical shift of free fluoroquine depends on the isotopic composition of the solvent (D₂O vs H₂O). The solvent isotope shift is virtually eliminated by fluoroquine binding to any one of the nucleic acids. ¹⁹F-nmr relaxation experiments were carried out to measure the spin-lattice relaxation time (T₁), ¹⁹F{¹H} nuclear Overhauser effect (NOE), off-resonance intensity ratio (R), off-resonance rotating-frame spin-lattice relaxation time (T), and linewidth for fluoroquine in the nucleic acid complexes. By accounting for intramolecular proton-fluorine dipolar and chemical-shift anisotropy contributions to the fluorine relaxation, all of the relaxation parameters for the fluoroquine–DNA complex can be well described by a motional model incorporating long-range DNA bending on the order of a microsecond and an internal motion of the drug on the order of a nanosecond. Selective NOE experiments indicate that the fluorine in the drug is near the ribose protons in the RNA complexes, but not in the DNA complex. Details of the binding evidently differ for the two types of nucleic acids. This study provides the foundation for an investigation of fluoroquine in intact cells.     

Retinyl esters are the substrate for isomerohydrolase

Regeneration of 11-cis retinal from all-trans retinol in the retinal pigment epithelium (RPE) is a critical step in the visual cycle. The enzyme(s) involved in this isomerization process has not been identified and both all-trans retinol and all-trans retinyl esters have been proposed as the substrate. This study is to determine the substrate of the isomerase enzyme or enzymatic complex. Incubation of bovine RPE microsomes with all-trans [(3)H]-retinol generated both retinyl esters and 11-cis retinol. Inhibition of lecithin retinol acyltransferase (LRAT) with 10-N-acetamidodecyl chloromethyl ketone (AcDCMK) or cellular retinol-binding protein I (CRBP) diminished the generation of both retinyl esters and 11-cis retinol from all-trans retinol. The 11-cis retinol production correlated with the retinyl ester levels, but not with the all-trans retinol levels in the reaction mixture. When retinyl esters were allowed to form prior to the addition of the LRAT inhibitors, a significant amount of isomerization product was generated. Incubation of all-trans [(3)H]-retinyl palmitate with RPE microsomes generated 11-cis retinol without any detectable production of all-trans retinol. The RPE65 knockout (Rpe65(-/-)) mouse eyecup lacks the isomerase activity, but LRAT activity remains the same as that in the wild-type (WT) mice. Retinyl esters in WT mice plateau at 8 weeks-of-age, but Rpe65(-/-) mice continue to accumulate retinyl esters with age (e.g., at 36 weeks, the levels are 20x that of WT). Our data indicate that the retinyl esters are the substrate of the isomerization reaction.     

Retinoid signalling and gene expression in neuroblastoma cells: RXR agonist and antagonist effects on CRABP-II and RARbeta expression

9-cis Retinoic acid (RA) induces gene expression in neuroblastoma cells more effectively and with different kinetics than other RA isomers, and could be acting in part through Retinoid X Receptors (RXRs). The aim of this study was to characterise the effects of an RXR agonist and RXR homodimer antagonist on the induction of cellular RA binding protein II (CRABP-II) and RA receptor-beta (RARbeta) in neuroblastoma cells in response to different retinoids. The RXR agonist, LDG1069, was as effective as all-trans RA in inducing gene expression, but less effective than 9-cis RA. The RXR-homodimer antagonist, LG100754, inhibited the induction of CRABP-II mRNA in SH-SY5Y neuroblastoma cells by 9-cis RA or the RXR-specific agonist LGD1069, but had no effect when used with all-trans RA. Conversely, LG100754 did not inhibit induction of RARbeta mRNA by 9-cis or all-trans RA, or by LGD1069. RAR- and RXR-specific ligands used together induced CRABP-II and RARbeta as effectively as 9-cis RA. These results demonstrate the value of combining RXR- and RAR-specific ligands to regulate RA-inducible gene expression. The possibility that RXR-homodimers mediate, in part, the induction of CRABP-II by 9-cis RA and RXR-specific ligands is discussed.     

Resonance Raman spectroscopy of carotenoids in Photosystem I particles

Low-temperature resonance Raman (RR) spectroscopy was used for the first time to study the spectral properties, binding sites and composition of major carotenoids in spinach Photosystem I (PSI) particles. Excitation was provided by an argon ion laser at 457.9, 476.5, 488, 496.5, 502 and 514.5 nm. Raman spectra contained the four known groups of bands characteristic for carotenoids (called from nu(1) to nu4). Upon 514.5, 496.5 and 476.5 nm excitations, the nu(1)-nu(3) frequencies coincided with those established for lutein. Spectrum upon 502-nm excitation could be assigned to originate from violaxanthin, at 488 nm to 9-cis neoxanthin, and at 457.9 nm to beta-carotene and 9-cis neoxanthin. The overall configuration and composition of these bound carotenoid molecules in Photosystem I particles were compared with the composition of pigment extracts from the same PSI particles dissolved in pyridine, as well as to configuration in the main chlorophyll a/b light-harvesting protein complex of photosystem II. The absorption transitions for lutein, violaxanthin and 9-cis neoxanthin in spinach photosystem I particles are characterized, and the binding sites of lutein and neoxanthin are discussed. Resonance Raman data suggest that beta-carotene molecules are also present in all-trans and, probably, in 9-cis configurations.     

Early steps of the intramolecular signal transduction in rhodopsin explored by molecular dynamics simulations

We present molecular dynamics simulations of bovine rhodopsin in a membrane mimetic environment based on the recently refined X-ray structure of the pigment. The interactions between the protonated Schiff base and the protein moiety are explored both with the chromophore in the dark-adapted 11-cis and in the photoisomerized all-trans form. Comparison of simulations with Glu181 in different protonation states strongly suggests that this loop residue located close to the 11-cis bond bears a negative charge. Restrained molecular dynamics simulations also provide evidence that the protein tightly confines the absolute conformation of the retinal around the C12-C13 bond to a positive helicity. 11-cis to all-trans isomerization leads to an internally strained chromophore, which relaxes after a few nanoseconds by a switching of the ionone ring to an essentially planar all-trans conformation. This structural transition of the retinal induces in turn significant conformational changes of the protein backbone, especially in helix VI. Our results suggest a possible molecular mechanism for the early steps of intramolecular signal transduction in a prototypical G-protein-coupled receptor.     

9-cis-retinoic acid decreases the level of its cognate receptor, retinoid X receptor, through acceleration of the turnover.

In this study, we investigated the ligand-mediated regulation of retinoid X receptor (RXR) in two human cell lines (HepG2 and JEG-3 cells), which have been reported to express RXRalpha predominantly. Western blot analysis revealed that a treatment with 1 microM 9-cis-retinoic acid (9-cis RA) for 24 h decreased RXRalpha protein level to 72 +/- 9 and 75 +/- 7% in HepG2 and JEG-3 cells, respectively, when the levels in the non-treated cells were expressed as 100%. The decrease was not due to the changes in steady-state level of RXRalpha mRNA or its stability as revealed by Northern blot analysis. However, the 9-cis RA treatment decreased the half-life of RXRalpha protein as determined by pulse-chase analysis. It was thus demonstrated that 9-cis RA downregulates RXRalpha by increasing the turnover of the protein in the two cell lines. The ligand-dependent downregulation of RXRalpha protein may be important for several hormonal signalings, in which the receptors heterodimerize with RXR.     

Synergistic effect of vitamin D derivatives and retinoids on C2C12 skeletal muscle cells

The response of C2C12 myoblasts to 1 nM 1,25-dihydroxyvitamin D3 [1,25(OH)2D3], two vitamin D analogues (KH 1060 and EB 1089, which are 20-epi-22-oxa and 22,24-diene-analogues, respectively), 100 nM retinoids (9-cis retinoic acid, all-trans retinoic acid) and to combination treatments, after 72 h incubation, was studied. The incubation with 1,25(OH)2D3 was ineffective on either cell proliferation or [3H]thymidine incorporation (expressed as DPM per cell) or protein content per cell. On the contrary, all the other treatments inhibited cell proliferation, this inhibition being synergistic when the vitamin D derivatives were combined with 9-cis or all-trans retinoic acid, and increased [3H]thymidine incorporation and protein content per cell. The levels of the VDR protein remarkably increased in comparison with control cells, except for the incubation with 9-cis retinoic acid. This increase was particularly accentuated in C2C12 cells treated with KH 1060 and 9-cis retinoic acid in combination. These results, taken together, suggest a role for vitamin D derivatives and retinoids on C2C12 cells.     

Expression of adipogenesis markers in a murine stromal cell line treated with 15-deoxy Delta(12,14)-prostaglandin J2, interleukin-11, 9-cis retinoic acid and vitamin K2.

Recent studies have demonstrated that bone marrow stromal cells can undergo adipogenesis or osteoblastogenesis in vivo, and in vitro, and that peroxisome proliferator-activated receptor gamma (PPAR gamma) plays a central role in the control of adipocyte differentiation. In the present study, we treated a murine stromal cell line (TMS-14) with a cocktail of dexamethasone, insulin and glucose (DIG cocktail), which caused the cells to convert to fat-laden cells with adipocyte-like morphology. We also exposed TMS-14 cells to DIG cocktail followed by 15-deoxy Delta(12,14)-prostaglandin J2 (15d-PGJ2), a ligand of PPAR gamma, interleukin- 11 (IL-11), 9-cis retinoic acid (9-cis RA) and vitamin K2. 15d-PGJ2 enhanced DIG cocktail-induced adipogenesis, whereas IL-11, 9-cis RA and vitamin K2 each inhibited adipogenesis induced by DIG cocktail. The gene expressions of four adipogenesis markers, PPAR gamma 2, adipocyte P2 (aP2), adipocyte determination and differentiation factor 1 (ADD1), and fatty acid synthase (FAS) were enhanced by DIG cocktail and these expressions were more enhanced by 15d-PGJ2, in contrast they were attenuated by 9-cis RA. IL-11 also attenuated the adipogenesis markers except ADD1. Western blotting showed that 15d-PGJ2 enhanced the levels of PPAR gamma, C/EBP alpha and RXR alpha proteins, while IL-11 and 9-cis RA decreased the level of PPAR gamma protein, but not C/EBP alpha protein and vitamin K2 decreased the level of C/EBP alpha protein. We also tested the effect of 15d-PGJ2 on osteoblastogenesis, using TMS-12 cells, another stromal cell clone from the same mouse, which differentiate into osteoblasts spontaneously. 15d-PGJ2 did not affect osteoblastogenesis, as detected by von Kossa staining and Cbfa-1 gene expression. These data indicate that 15d-PGJ2 enhances the expression of both PPAR gamma and C/EBP alpha and as a result it stimulates adipogenesis in murine bone marrow cells.