Unchaining the beast; insights from structural and evolutionary studies on TGFβ secretion, sequestration, and activation

Abstract

TGFβ is secreted in a latent state and must be "activated" by molecules that facilitate its release from a latent complex and allow binding to high affinity cell surface receptors. Numerous molecules have been implicated as potential mediators of this activation process, but only a limited number of these activators have been demonstrated to play a role in TGFβ mobilisation in vivo. Here we review the process of TGFβ secretion and activation using evolutionary data, sequence conservation and structural information to examine the molecular mechanisms by which TGFβ is secreted, sequestered and released. This allows the separation of more ancient TGFβ activators from those factors that emerged more recently, and helps to define a potential hierarchy of activation mechanisms.

Keywords: Activation; Evolution; Extracellular matrix; LTBP; TGFβ.

Figure 1. Simplified scheme of TGFβ secretion and signalling

A summary of key steps in TGFβ production and signalling. Starting at the bottom left; TGFβ associates with LTBPs inside the cell to form the large latent complex (LLC). After secretion the LLC is localized to ECM fibres via LTBP mediated interactions. Active TGFβ may be released by a number of factors reviewed here. TGFβ can induce down stream signalling by binding its type 1 and type 2 receptors. The canonical signalling pathway involves phosphorylation of R-SMADs by the type I TGFβ receptor. Phosphorylated R-SMADs (pSMAD2/3) interact with SMAD4 and are translocated to the nucleus, where they regulate the transcription of many genes through interactions with CAGA repeat elements. Various nuances of TGFβ signalling have been excluded here for the purposes of simplicity.

Figure 2. Phylogenetic tree of the TGFβ superfamily

A) Phylogenetic tree of 359 TGFβ sequences from a variety of organisms (for search and tree construction methods see supplementary information). To save space sub-trees are compressed and labelled according to the predominant superfamily members they contain. Sub-trees are coloured depending on the pattern of cysteines present in the mature growth factor domains (colour code at bottom). The vertical dimension of each sub-tree is dependent on the number of sequences it contains. Bootstrapping values are shown at each node, and the scale bar at the bottom gives the average number of amino acid substitutions per site. B) The expanded sub-tree of bona fide TGFβ homologues. Sub-trees corresponding to TGFβ1, 2 and 3 isoform groups are highlighted with coloured boxes. Despite the wide range of animal genomes searched, only sequences from the deuterostomes are found in the true TGFβ sub tree.

Figure 3. LAP cysteines

A) Structure of TGFβ1 and homology models of TGFβ2 and 3 with LAP cysteines highlighted as spheres. TGFβ1 LAP is coloured orange while the growth factor is coloured dark red, TGFβ2 LAP is coloured light green and the growth factor is coloured dark green, and TGFβ3 LAP is coloured light blue and the growth factor is coloured dark blue. B) Alignments of cysteine containing regions, residues are highlighted blue if >60% conserved or yellow if a cysteine residue. Cysteines not conserved in over 60% of sequences are highlighted in pale orange. Specific cysteines of interest are numbered above human TGFβ1, TGFβ2 and TGFβ3 sequences (as highlighted in A).

Figure 4. Conservation of TGFβ residues interacting with LTBP

A) Alignment of the N-terminus of the TGFβ propeptide showing absolute conservation of cysteine 33, highlighted in yellow. Residues highlighted blue are also conserved in >75% or more of the sequences. Orange arrows at the top of the alignment highlight residues that when mutated disrupt LTBP1 binding. The grey arrow highlights arginine 58, which when mutated to alanine did not affect LTBP binding B) Alignment with other TGFβ superfamily members. Some members do possess cysteines in this region of sequence, but they do not align well with the conserved features of bona fide TGFβ. C) Structure of TGFβ-LAP (red and orange) and LTBP1 TB2 (blue), TGFβ residues that when mutated inhibit the LAP-LTBP1 interaction are highlighted as orange spheres. Residues in LTBP1 8-Cys3/TB2 that have been shown to contribute to TGFβ binding are highlighted as dark blue spheres. (Carbon atoms coloured as described, while all oxygen atoms are coloured red, nitrogen atoms coloured blue, and sulphur atoms coloured yellow)

Figure 5. Emergence of TGFβ and regulatory factors

Representation of the proteins present in various organisms selected to give a comprehensive overview of bona fide TGFβ evolution. The tree (top) illustrates the evolutionary relationships between each organism (according to the tree of life web project). The column below each illustrated organism shows whether or not the gene of interest is present, as indicated by a + or -, or other details such as the names of specific isoforms. Blocks of colour highlight where each gene emerges. Details on the methodologies and data used to construct this figure are described in supplementary material.

Figure 6. Alignment of LTBP 8-cys3/TB2 domains

Black arrowheads highlight acidic residues implicated in binding the TGFβ propeptide (numbered according to Chen et al.), and the 2 amino acid insert shown to be important for TGFβ binding is surrounded by a black box. Residues present in >60% of all sequences are highlighted as blue, or yellow when cysteine. Sequences from a variety of organisms representing all LTBP isoforms were aligned with the MEGA 5 BLOSUM matrix (Gap opening penalty 5, gap extension penalty 0.5). Alignment divided into blocks by LTBP isoform.

Figure 7. Integrin mediated TGFβ activation mechanisms

A) Scheme for traction mediated activation by integrins. On the left the large latent complex is shown in a “resting” state with integrin binding LAP via its RGD motif. On the right the latent complex is shown under traction, which deforms LAP and releases active TGFβ. Components are labelled within the figure. B) Scheme for protease mediated integrin activation. On the left the integrin is shown binding to LAP at the cell surface. This integrin may then recruit proteases that degrade LAP and release active TGFβ, as shown on the right. C) Alignment of the RGD containing region of TGFβ. Residues conserved in >60% or more of sequences were highlighted blue, or yellow if cysteine, a black line above the alignment highlights the RGD motif.

Figure 8. Conservation of protease cleavage sites in TGFβ and LTBP1

Ai) Alignments of the “latency lasso” region of LAP. Some potential MMP2/9 cleavage sites are boxed, and pairs of light red or dark red asterisks above the human TGFβ2 sequence highlight two potential MMP cleavage PxxV motifs. Aii) Homology model of TGFβ2, with LAP shown in light green, and the mature growth factor in dark green. The side chains of the two putative PxxV cleavage motifs highlighted in A) are shown as red coloured spheres. B) Sequence alignments of LTBP regions where putative BMP1 sites were detected (103). Residues present in 60% or more sequences are highlighted. The aspartate residues that define the putative cleavage sites are highlighted by boxes, but do not appear to be well conserved.

Figure 9. Glycosylation of TGFβ

A) Alignments of N-glycosylation sites from all human TGFβ isoforms, residues conserved in 60% or more sequences are highlighted blue. Putative N-glycosylation sites are boxed. Only the first glycosylation site is well conserved. B) Glycosylation sites highlighted on the structure of TGFβ1. TGFβ and LAP are coloured as in previous figures and sugar moieties are shown coloured pink and represented as spheres. In the crystallised protein the third glycosylation site, “NNS” has been replaced with “QDS”, but the Q residue is coloured pink here and shown in stick representation labelled “N”176. The position of the RGD motif is also highlighted by a black line, although it is not resolved in the structure.