TGF-beta is a pleiotropic, multifunctional cytokine with potent immunoregulatory properties. Nearly all cells of leukocyte lineage express the cytokine. Although not completely understood, the role of TGF-beta has been established as the primary inhibitory cytokine in many processes including inflammatory responses, carcinogenesis and oral tolerance. TGF-beta treatment leads to cell-cycle arrest at the late G1 phase through the expression of a cyclin-dependent kinase inhibitor p27 molecule. The TGF-beta receptor consists of two chains, type I and II; both are receptor serine/threonine kinases. The binding of TGF-beta to the type II receptor recruits and activates the type I receptor, which in turn activates the SMAD signaling pathway, leading to the regulation of gene expression. Due to the lack of structural work, the ligand-binding region of both the receptor and TGF-beta remains unclear. Our objective is to determine the crystal structures of the TGF- beta receptor alone and in complex with TGF-beta. We hope to gain functional insight into the receptor activation. Our first attempt was to crystallize a baculovirus expressed human type II TGF-beta receptor (TBRII) in complex with TGF- beta1. Due to extensive glycosylation on the receptor, the crystallization trials with this receptor did not yield any crystals. Subsequent deglycosylation procedures yielded a partially deglycosylated receptor that produced small crystals when complexed with TGF-beta1. However, these crystals diffracted poorly in the X-ray beam and were not suitable for structure determination. To overcome the carbohydrate heterogeneity and low expression yield of the baculovirus expression system, we have attempted to express the soluble type II receptor in several bacteria expression vector systems. Due to a large number of disulfide bonds (12 cysteines) present in this receptor, previous attempts by other groups to reconstitute the bacteria expressed receptor have all failed. The first bacteria construct we made was a fusion protein of the type II TGF-beta receptor and GST. This resulted in the production of a functional receptor in the soluble fraction of cell lysates, as evident from a TGF- beta binding ELISA assays. However, the yield of TBRII from the GST fusion construct remained low. This was primarily due to the degradation of the linker region between GST and the receptor, that resulted in the majority eluted sample from a glutathione affinity column being free GST rather than the GST-TBRII fusion protein. A typical 10 liter bacteria expression experiment yielded less than 0.5 mg of purified receptor protein. To overcome the problem of low expression yield, we began to investigate the purification schemes by refolding methods, despite previous attempts by others not being successful. As a result, we subcloned the ecto-TBRII (without any fusion partner) into two Novagen's pET vectors, pET14 and pET30, both of which are driven by T7 polymerase. The expression of this construct resulted in large quantities of inclusion bodies. Subsequent refolding experiments yielded milligram quantities of TBRII that appears active in a TGF-beta ELISA assay. The crystallization experiments using this receptor have produced crystals. More recently, we have completed the crystal structure of this receptor at 1.1 angstrom resolution.? Another issue hinders the project is the availability of large quantity of recombinant TGF-beta. Current recombinant expression systems yield less than 1-2 mg of the mature TGF-b1 per liter of cell culture medium. In an effort to produce large quantities of the recombinant cytokine for structural studies, we have constructed a mammalian expression system based on a modified pcDNA3.1 vector with a glutamine synthetase gene inserted for gene amplification. The leader peptide of TGF-b1 was replaced with that of rat serum albumin, and an eight-histidine tag was insertedimmediately after the leader sequence to facilitate protein purification. In addition, Cys 33 of TGF-b1, which forms a disulfide bond with LTBP, was replaced by a serine residue. The resulting expression construct produced a stable clone expressing 30 mg of mature TGF-b1 per liter of spent medium. Purified TGF-b1 bound with high affinity to its type II receptor with a solution dissociation constant of 70 nM, and was fully active in both a Mv1Lu cell growth inhibition assay and in a PAI-1 luciferase reporter assay. In the past year, we have constructed similar recombinant CHO expression systems for Tgf-b2 and Tgf-b3. Both recombinant systems yielded 10mg protein per liter of culture media. A patent for the expression system has been filed by NIAID under U.S. Provisional application Nos 60/534,379 and 60/575,839.? ? We have recently crystallized and determined the structure of TGF-beta 1 in complex with the extracellular domains of the type I and type II receptors at 3.0 Ǻ resolution. The overall assembly for TGF-b1 ternary complex is very similar to that of TGF-b3 complex with TBRII bound identically to both isoforms of TGF-b. The docking of TBRI on TGF-b1 is 10X different from that of TGF-b3. The structural analysis suggests that the lower TBRII binding affinity to TGF-b2 is the result of replacement of both Arg 25 and Arg 94 in TGF-b1 and -b3 with lysines in TGF-b2. Comparison between the TGF-b and BMP-2 ternary complexes revealed that predominantly hydrogen bonds mediate binding of the cytokines to their high affinity receptors (TBRII and BMPR-IA, respectively), whereas hydrophobic interactions dominate their contacts to the low affinity receptors (TBRI and ActRII, respectively). This suggests the importance of hydrogen bonding in determining the receptor preference for these cytokines. The solution binding studies unexpectedly revealed significant binding of TGF-b2 and -b3 to TBRI. While all three isoforms of TGF-bnassembled their ternary complexes equally efficient, the preference for the type II receptor varied from TGF-b1 being the most to TGF-b2 being the least TBRII dependent. The equal energetic contribution of TBRI and TBRII to the assembly of TGF-b2 signaling complex suggests a simultaneous rather than sequential receptor recruitment model for TGF-b2. The differences in the receptor recruitment provide a potential mechanism for a distinct function of three TGF-b isoforms that is probably coordinated with the cellular and tissue-dependent receptor expressions.
