The Impact of Cysteine Substitutions on TGF-β3 Expression, Purification, Folding, and Activity

Abstract

Transforming growth factor beta 3 (TGF-β3) is a homodimeric cytokine with potential therapeutic applications in wound healing, tissue engineering and regenerative medicine. Production of recombinant TGF-β3 in Escherichia coli faces significant challenges due to TGF-β3’s propensity for misfolding and aggregation, driven by a high disulfide bond content and low aqueous solubility. To address these limitations, the impacts of substituting non-conserved cysteine residues C7, C16 and C77 with serine on TGF-β3 folding, dimerization and activity were investigated. Whilst C7 and C16 form an intra-chain disulfide bond, C77 forms an inter-chain disulfide bond stabilizing dimer formation. Our results showed that the C7S, C16S double cysteine mutant protein exhibited reduced aggregation, increased dimer formation, and maintained wild-type biological activity in nano-luciferase reporter gene assay. In contrast, both C77S single and C7S, C16S, C77S triple mutants were purified predominantly in monomeric forms and displayed about 2.5-fold reduced activities. Our findings highlight the roles of the non-conserved C7, C16 and C77 cysteine residues in TGF-β3 folding and aggregation. The identification of the C7S, C16S mutant as a more soluble protein with wild-type TGF-β3 activity offers a promising strategy for improving recombinant TGF-β3 production to facilitate therapeutic applications. This study underscores the importance of targeted cysteine engineering to overcome the inherent challenges associated with the production of TGF-β3 and related complex disulfide-rich proteins.

1. Introduction

Transforming growth factor-beta 3 (TGF-β3) is a cytokine crucial for regulating cellular processes such as growth and differentiation, positioning it as a promising therapeutic agent for applications ranging from wound healing and scar prevention to tissue regeneration [1,2]. While Escherichia coli (E. coli) is a widely favored host for recombinant protein production due to its cost-effectiveness and rapid growth, the expression of complex proteins like TGF-β3 in this system often leads to the accumulation of misfolded protein within insoluble inclusion bodies (IBs), necessitating costly downstream refolding to obtain the biologically active form.

TGF-β3 is a member of TGF-βs subfamily, which encompasses TGF-β1, TGF-β2, and TGF-β3 itself. They are also part of the larger superfamily of signaling proteins known as the TGF-β superfamily [3]. Mature human TGF-β3 is a homodimeric protein consisting of two identical subunits of 112 aa, linked by a single inter-chain disulphide bond [4]. There are nine Cys residues in each subunit: six participate in the cysteine knot (CK) motif, two form an intrachain disulphide bond near the N-terminus, and the remaining cysteine forms the interchain disulphide bond that links the two subunits. Figure 1 provides a visual representation of this arrangement. Figure 1A depicts the primary amino acid sequence, highlighting the positions of the nine cysteine residues. Figure 1B shows a 3D model of a TGF-β3 monomer, illustrating the intricate folding pattern and the disulfide bonds that contribute to its stability, including the characteristic CK. Dimerization, as shown in Figure 1C, is mediated by an intermolecular disulfide bond between cysteine 77 residues in each monomer.

A significant hurdle in the production of functional TGF-β3 lies in its inherent biochemical properties: low aqueous solubility and a strong propensity to misfolding and thus protein aggregation [5]. These characteristics severely impede efficient oxidative folding and subsequent purification of the active homodimer. The folding of TGF-β3 homodimer is stabilized mainly by intramolecular and intermolecular disulfide bonds, including a characteristic CK motif [4]. The formation of mis-matched disulfide bonds is a primary driver of protein aggregation, resulting in significantly reduced yields of the desired active protein [6].

Inspection of the crystal structure of TGF-β3 revealed substantial exposed hydrophobic patches at the surface, which may promote the formation of protein aggregates [4,7]. In addition, TGF-β3 exhibits a strong tendency to adsorb to hydrophobic surfaces and form large aggregates. Furthermore, compared with TGF-β1 and TGF-β2, TGF-β3 exhibits greater conformational plasticity, adopting two conformations. Whereas TGF-β1 and TGF-β2 exhibit a single closed conformation, TGF-β3 exist in both closed and open state conformations. In the closed state, both monomers of the homodimer are rigidly packed against one another. While in the open state, both monomers are loosely packed against each other [8,9,10,11]. Thus, TGF-β3 displays increased confirmation flexibility and altered monomers arrangement relative to TGF-β1. A study showed that the difference in their activity arises from the differences in their conformations, rather than their receptor binding alone [10]. However, increasing conformational flexibility may enhance solvent accessibility of hydrophobic regions, thereby contributing to aggregation propensity under physiological conditions. Indeed, circular dichroism (CD) and ultraviolet (UV) spectroscopy, along with ultracentrifugation studies, revealed that TGF-β3 displays a bell-shaped pH-dependent aggregation. It is soluble in acidic (pH ≤ 3.8) and alkaline (pH ≥ 9.7) buffers, but has a high propensity for aggregation between pH 6.8 and 8.2, including the normal physiological pH of around 7 [5]. Collectively, these factors contribute to TGF-β3 low solubility at physiological pH. Consistently, most previous studies reported the production of TGF-β3 in E. coli as IBs, necessitating subsequent in vitro denaturation, solubilization and refolding to obtain the active proteins [5,6,10,12,13]. However, refolding is inefficient resulting in very low yields, often less than 1 mg per liter of bacterial culture [14]. Fusion or solubility-enhancing tags have been explored to improve the solubility and reduce aggregation during expression and purification of TGF-β family proteins. For example, superfolder GFP (sfGFP) has been used as a fusion partner to obtain soluble sfGFP–6His-TGF-β3 in E. coli at a high purity and yield [15]. Similarly, high-level expression of Trx-6His fusion tag with TGF-β1 C77S in E. coli was reported [16]. It simplified downstream purification, and produced an active monomeric TGF-β1 C77S after removing the tag [16].

To overcome these production challenges, strategies aimed at mitigating TGF-β3 aggregation are essential. Analysis of the mature TGF-β3 amino acid sequence reveals a high cysteine content (approximately 8%), with a spatial arrangement that increases the susceptibility to cysteine mispairing and subsequent aggregation (Figure 1A). The CK configuration is a conserved structural feature crucial for the folding and stability of TGF-β3 and other members of the TGF-β superfamily [17]. However, not all disulfide bonds within TGF-β3 are conserved across the entire TGF-β superfamily [6,8,18,19]. Notably, the intramolecular disulfide bond near the N-terminus (C7–C16) and the cysteine residue involved in the inter-monomer disulfide linkage for dimerization (C77) are not conserved. The dimerization cysteine, C77 in TGF-β3, is only conserved in some TGFβ superfamily proteins, including TGF-β subfamily, BMP-1, BMP-15, GDF-3, and GDF-9, and it is vicinal to C_IV of the knot configuration across all proteins where it is conserved. Similarly, the extra internal disulfide bond, the C7-C16 in TGF-β3, is only present in the TGF-β subfamily, GDF-8, GDF-11, and Inhibin proteins, but not in other members of the superfamily proteins. In this bond, one of the cysteine participants is located vicinal to C_I in the knot configuration in all the proteins where it is present (Figure 1A). This suggests targeted substitutions of these non-conserved cysteine residues may not disrupt the core structural integrity and function of the TGF-β3 protein but may facilitate its folding and production. Given the critical role of disulfide bonds in both the correct folding and aberrant aggregation of TGF-β3, understanding the contribution of non-Cys knot disulfide bonds to protein folding, stability and function is paramount. In this study, cysteine residues (C7, C16, and C77) were replaced with serine to evaluate the roles of these cysteine residues on the folding, and biological activities of TGF-β3. By systematically investigating the effects of these substitutions, we sought to identify potential modifications that could reduce aggregation and improve the recombinant production of functional TGF-β3.

2. Results

2.1. Cysteine Substitutions Do Not Impair TGF-β3 Expression

WT TGF-β3 and three cysteine mutant variants (C77S, C7S, C16S, and C7S, C16S, C77S) were successfully produced in E. coli and purified under denaturing conditions. All proteins contained an N-terminal 6xHis tag to facilitate purification. SDS-PAGE analysis showed that all constructs yielded a predominant ~15 kDa species consistent with monomeric TGF-β3 (Figure 2). While expression levels and recovery were broadly comparable across constructs, densitometric analysis revealed that most of the expressed TGF-β3 proteins accumulated as insoluble species in the form of IBs. The identities of TGF-β3 WT and mutants were confirmed by Western blotting (see below) and mass spectrometry analysis (Supplementary Figure S1).

Notably, although cysteine substitution did not significantly alter overall expression, it markedly influenced the aggregation profile of the protein. Non-reducing SDS-PAGE analysis (Figure 3, the last lanes) demonstrated that WT TGF-β3 formed prominent higher-order oligomers (e.g., ~37–50 kDa), whereas cysteine mutant proteins, particularly the C7S, C16S, C77S, exhibited reduced levels of high-molecular-weight species (Figure 3). These observations suggest that these cysteine residues contribute to misfolding and intermolecular disulfide-mediated aggregation during expression.

2.2. Cysteine Substitutions Reduce Aggregation and Influence Dimerization During Refolding

To assess whether these cysteine substitutions improve refolding of TGF-β3, oxidative refolding of the purified proteins was performed in vitro. First, 10 mM DTT was added to the urea-solubilized and denatured proteins to break disulfide bonds. Following reduction in disulphide bonds, the WT and mutant proteins migrated predominantly as monomers with apparent Mw of ~15 kDa on SDS-PAGE (Figure 4A) as expected. Next, oxidative refolding was initiated by 10-fold dilution of the reduced and denatured proteins in refolding buffer (see Figure 4 legend).

Following refolding, non-reducing SDS-PAGE analysis revealed marked differences in oligomeric behavior between the WT and cysteine mutant proteins (Figure 4B). The WT protein exhibited a heterogeneous banding pattern comprising monomeric, dimeric, and substantial higher-order oligomeric species, consistent with extensive aggregation during refolding. In contrast, all three mutants showed a pronounced reduction in high-molecular-weight aggregates, demonstrating an improved refolding outcome. Among the mutants, the C7S, C16S double substitution resulted in a defined and simplified profile dominated by discrete monomer (~15 kDa) and dimer (~29 kDa) species, with minimal higher-order oligomers. The C77S single mutant and the C7S, C16S, C77S triple mutant refolded predominantly as monomers, consistent with disruption of the intermolecular disulfide bond formed by Cys77; however, the triple mutant displayed significantly less higher order oligomers. Consistently, Western blot analysis under non-reducing conditions (Figure 4C) using mTGF-β3 antibody, which recognizes multiple forms of TGF-β3, revealed that the WT and C77S samples showed prominent bands for monomers, dimers, and diffuse high molecular weight bands indicative of aggregates. Notably, the C7S, C16S double mutant and C7S, C16S, C77S triple mutant proteins exhibited a marked reduction in these high molecular weight aggregates, predominantly showing monomer and dimer bands. The presence of the His x6-tag results in a marked reduction in electrophoretic mobility under non-reducing conditions, relative to the TGF-β3 dimer standard (12.7 kDa), which lacked a tag. These observations indicate that selective removal of non-conserved cysteine residues reduces aberrant disulfide pairing while preserving controlled dimer formation where structurally favorable, i.e., C7S, C16S variant.

Quantitative densitometric analysis of the refolded samples in Figure 4C substantiated these qualitative trends (

Table 1. TGF-β3 WT and mutant refolding yield.

TGF-β3 Protein	% Dimer	% Monomer	% Aggregates
WT	37.7 ± 1.09	14.5 ± 5.0	47 ± 6
C77S	30 ± 4.1	43 ± 2.2	25 ± 1.9
C7S, C16S	48 ± 4.8	34 ± 4.7	16 ± 0.1
C7S, C16S, C77S	16 ± 0.5	58 ± 2.7	25 ± 3.2

). Aggregated species accounted for approximately 52% of refolded WT protein, whereas aggregation was reduced to 27%, 17%, and 23% for the C77S, C7S, C16S, and C7S, C16S, C77S variants, respectively. Notably, the C7S, C16S mutant displayed the highest dimer fraction (45%) alongside the lowest aggregate content, indicating that reduction in aggregation enables productive dimer formation rather than nonspecific oligomerization. In contrast, the increased monomer proportions observed for the C77S-containing variants reflect the expected loss of the Cys77-mediated interchain disulfide bond.

Next, the refolded proteins were purified by size exclusion chromatography (SEC) to remove aggregates and to enable buffer exchange to a physiologically relevant buffer required for subsequent activity assays. The SEC elution profiles (Figure 5A) revealed three principal peaks for WT and C77S mutant, corresponding to high-molecular-weight oligomers (P_O), dimers (P_D), and monomers (P_M), respectively, whereas C7, 16S and C7, 16, 77S predominantly eluted as dimer and monomer with minimal higher-order species. The fractions corresponding to P_D (17–19 mL) were pooled together and analyzed using non-reducing and reducing SDS-PAGE (Figure 5B). Under non-reducing conditions, the proteins separated mainly as dimer (~25 kDa) and monomer (14.5kDa). Whilst the monomer bands were the dominant bands for all the proteins, including C77S mutant, very weak bands at ~37 kDa were also observed. As a control, under reducing conditions, almost all proteins are present as monomers. Taken together, these results suggest that the apparent 25 and 37 kDa bands contain proteins complexes (dimer or trimer) that stabilized by non-covalent protein–protein interactions with and without inter-chain disulphide. This observation is consistent with a previous TGF-β3 study [20], in which similar protein species were also detected under non reducing gel, whereas a single band corresponding to the monomer was observed under reducing conditions.

Next, these SEC-purified proteins of P_D were analyzed by Western blots using a conformation specific antibody, cTGF-β3, which specifically recognizes correctly folded forms of TGF-β3 (Figure 6). The cTGF-β3 antibody identified both monomeric and dimeric forms of the WT and all three cysteine mutants, confirming that the proteins were correctly folded. The presence of correctly folded monomer and dimer bands is consistent with the observation that these forms are interconvertible and in equilibrium.

Furthermore, to evaluate whether cysteine substitutions affected overall protein folding, far-UV circular dichroism (CD) spectra were recorded (Figure 7A). All mutants displayed highly similar spectra to the WT protein with a broad minimum at about 204 nm, consistent with β-sheet-rich TGF-β family structures. Secondary structure estimation using BESTSEL revealed comparable distributions across all proteins (Figure 7B). All samples displayed a similar amount of α-helix (3–4.6%), β-sheet (29–34%), and about 50% in random coil. Thus, there are no obvious effects from the cysteine substitutions on the overall secondary structure under the experimental conditions. Moreover, thermal stability screening of TGF-β3 WT and the cysteine mutants showed on clear thermal denaturation temperature (Tm) can be measured (Supplementary Figure S2), consistent with the facts that no Tm has been reported for TGF-β3 and the CD analysis showed that the proteins are largely in a flexible conformation with 50% in random coil. In addition, the thermal stability of the WT and cysteine mutants in various buffers over physiologically relevant pH 6.5–8 was investigated (Supplementary Figure S3). The results suggest that C7, 16S and C7, 16, 77S mutants (without C7-C16 disulphide bond) have a decreased aggregation propensity over pH 6.5–8.

2.3. Cysteine Substitutions Differentially Modulate TGF-β3 Signaling Activity

The biological activity of WT TGF-β3 and its cysteine mutants was assessed using two cell-based assays: a nano-luciferase reporter gene assay and Smad2 phosphorylation Western blot analysis.

The reporter assay is a quantitative assay measuring activation of canonical TGF-β/SMAD signaling in TC28a2 cells with a SMAD binding element nano-luciferase-pest (SBE nLUCp) construct [23]. This assay provides a sensitive and integrated readout of pathway activation and was therefore used as the primary measure of functional activity. Dose–response analysis revealed that the C7S, C16S double mutant retained signaling potency comparable to WT TGF-β3, with EC₅₀ values of 1.1 ± 0.3 ng/mL and 1.6 ± 0.2 ng/mL, respectively (Figure 8,

Table 2. Nano-luciferase reporter assay EC50 values of 6xHis-TGF-β3 WT and its cysteine mutant proteins in ng/mL.

TGF-β3 Protein	EC50
WT	1.6 ± 0.2 ng/ml
C77S	3.4 ± 0.6 ng/ml
C7S, C16S	1.1 ± 0.3 ng/ml
C7S, C16S, C77S	3.7 ± 0.7 ng/ml

). In contrast, variants containing the C77S substitution exhibited significantly reduced activity. The C77S single mutant and the C7S, C16S, C77S triple mutant displayed approximately 2–3-fold higher EC₅₀ values compared to WT (3.4 ± 0.6 ng/mL and 3.7 ± 0.7 ng/mL, respectively), indicating impaired signaling capacity. Furthermore, Smad2 phosphorylation Western blot analysis showed trends consistent with the nano-luciferase data, confirming preserved signaling for WT and C7S, C16S proteins and reduced potency for C77S-containing variants (Supplementary Figure S4).

Taken together, these results demonstrate that substitution of Cys7 and Cys16 preserves TGF-β3 activity, whereas mutation of Cys77 compromises functional signaling activity. The consistent reduction in activity observed for C77S-containing variants suggests that this residue plays an important role in maintaining the structural integrity required for efficient receptor activation.

3. Discussion

In this study, we systematically evaluated the impact of targeted cysteine-to-serine substitutions on the expression, aggregation behavior, oxidative refolding efficiency, and biological activity of TGF-β3. Our findings demonstrate that selective modification of non-conserved cysteine residues can substantially reduce aggregation during refolding without compromising functional signaling, thereby addressing a key bottleneck in the production of bioactive TGF-β3 proteins.

A central outcome of this work is that the aggregation during refolding can be selectively reduced without compromising biological activity. Although all cysteine variants were expressed and purified with comparable efficiency, the major distinction emerged during refolding from denatured IBs. WT TGF-β3 exhibited a strong tendency to form higher-order aggregates, consistent with its high cysteine content and complex disulfide-bonding requirements. In contrast, substitution of Cys7 and Cys16 markedly reduced aggregation, resulting in a greater proportion of soluble, correctly folded protein. These findings support the view that non-native disulfide bond formation during refolding is a primary driver of aggregation in TGF-β3 and that limiting the availability of reactive thiols can improve folding outcomes.

Mechanistically, the data indicate that Cys7 and Cys16 contribute disproportionately to non-productive disulfide interactions that promote aggregation rather than correct folding. Their substitution likely reduces kinetic traps associated with cysteine mispairing, thereby favoring productive folding pathways. Importantly, this reduction in aggregation did not compromise biological function: the C7S, C16S double mutant retained WT-like signaling potency in the nano-luciferase reporter assay, demonstrating that these residues are dispensable for receptor activation and downstream SMAD signaling. In addition, although TGF-β3 can adopt both open and closed dimer conformations, with a small α-helix disordered in the open form [10], our results suggest that the cysteine substitutions have not altered this conformational equilibrium significantly. It is because a similar far UV CD spectrum and α-helix content (3–4.6%) were observed (Figure 7), especially between the WT and C7S, C16S mutant. Our activity assay showed that the C7S, C16S mutant had WT (or slightly higher) activity, suggesting this mutation did not disrupt its binding to the receptors. On the other hand, reducing the number of available cysteine residues likely decreases the probability of off-pathway intermolecular disulfide mispairing during folding. While the open state is associated with increased solvent accessibility [6,10], limiting reactive thiols may reduce aberrant intermolecular crosslinking without redefining the intrinsic conformational states of the protein. Moreover, the more pronounced higher order oligomerization of C77S (Figure 4, Figure 5 and Figure 6), likely reflects retention of C7 and C16 which remain available for non-native intermolecular disulfide pairing.

In contrast, mutation of Cys77 had a pronounced negative effect on activity. Variants containing the C77S substitution consistently exhibited reduced signaling potency despite improved solubility and reduced aggregation. This highlights a critical distinction between cysteines involved in aberrant aggregation and those essential for native structure. Cys77 is known to participate in the intermolecular disulfide bond that stabilizes the active TGF-β3 dimer, and its substitution likely disrupts the formation or stability of this covalent linkage. Lack of the C77-mediated disulfide interchain bond between the monomers destabilized dimer formation which was reliant on less stable non-covalent interactions. The predominance of monomeric species observed for C77S-containing variants supports this interpretation and provides a mechanistic basis for the observed reduction in biological activity.

These findings underscore the importance of preserving the inter-molecular disulfide bond while selectively removing cysteines that contribute to misfolding. The enhanced dimer formation observed for the C7S, C16S mutant relative to WT is particularly notable, suggesting that reducing competing aggregation pathways can indirectly promote productive dimerisation. This effect likely explains why the double mutant not only folds more efficiently but also retains full signaling competence.

Our results are consistent with previous reports demonstrating reduced activity of C77-mutated TGF-β proteins, and the magnitude of activity loss observed here aligns well with earlier in vitro studies [24]. As is well-established, TGF-β superfamily members primarily exert their biological functions in their dimeric forms [25,26]. Dimerization enhances potency, likely by increasing the apparent affinity for TGF-β receptor binding through membrane-localization effects [24]. Together, these data reinforce the conclusion that Cys77 is indispensable for functional dimer formation, whereas Cys7 and Cys16 primarily modulate folding efficiency rather than receptor engagement.

More broadly, this study provides a conceptual framework for improving the production of disulfide-rich growth factors. By distinguishing cysteine residues that are structurally essential from non-essential cysteine residues that drive aggregation, it is possible to rationally engineer protein variants with enhanced solubility and refolding efficiency while preserving biological activity. This targeted reduction in disulfide complexity represents a generalizable strategy that may be applicable to other members of the TGF-β superfamily and to disulfide-rich therapeutic proteins that are prone to aggregation during recombinant expression.

4. Materials and Methods

All chemicals used in this study were analytical grade and were from Sigma-Aldrich or Fisher unless specified.

4.1. Construction of the Expression Vectors

The human mature TGF-β3 protein (GenBank: CAR70088) was back translated into DNA choosing frequent codons in the E. coli codon usage table by Genscript (Oxford, UK). The coding sequences were cloned into pET30 vector by the company between the Nde I and Hind III or Xho I site. The start codon was located at the ATG within the CATATG Nde I recognition site and 6× Histidine tag and TEV protease recognition site were inserted to the TGF-β3 N-terminus. Sequences were manipulated with Vector NTI v 11.5 (Invitrogen, Carlsbad, CA, USA).

4.2. Plasmid Transformation into Expression Host

Competent cells of (BL21 star (DE3)) E. coli (~50 μL) of each respective strain were thawed on ice. Then, 1–5 μL of corresponding plasmid (at least 50 ng) was added to the competent cells and incubated for 30 min. The cells were then heat-shocked for exactly 30 s in a 42 °C and placed on ice for 5 min. Afterwards, 950 μL of warm LB broth was added per tube and cells were allowed to recover at 37 °C for 1 h with gentle shaking (New Brunswick Innova 42 shaker, Eppendorf, Hamburg, Germany). Approximately up to 100 µL of each transformation mixture was spread onto warm selective plates with 50 μg/mL Kanamycin, then incubated overnight at 37 °C in a static incubator (Genlab Ltd., Widnes, UK). This only allows bacteria that contain the gene of interest to grow.

4.3. Protein Expression and Purification

The transformed E. coli cells were cultured in a primary culture of 5ml LB medium with selection antibiotic (kanamycin (50 µg/mL)), and grown over-night at 37 °C. Then, an optical density of 0.05 E. coli (from pre-culture) was used to inoculate one liter of LB medium containing selection antibiotic and incubated at 37 °C until OD600 of (0.6–0.8). Protein expression was induced by adding IPTG (1 mM) into LB medium after temperature reduction to 20 °C. The culture was then incubated with shaking at 200 rpm overnight. Afterwards, cells harvested by centrifugation (5500× g, 4 °C, 30 min) were resuspended in buffer containing 50 mM Tris-HCl (pH 8.0), 300 mM NaCl, protease inhibitor cocktail tablet, 10% (w/w) glycerol, 0.1 mM EDTA and 1 mM PMSF (lysis buffer). Cells were then disrupted by sonication with a tapered microtip 10 s on/off for 8 min at 25–30% amplitude. The resulting cell lysate was clarified by centrifugation (55,000× g, 4 °C, 30 min). Pellets (or IBs), isolated from cells lysate, were first resuspended in 50 mM Tris-HCl pH 8.0 with 0.5 mM EDTA, 1 M NaCl, 0.05% (v/v) TritonX-100, and 5% (v/v) glycerol using a tissue homogenizer and pelleted by centrifugation (55,000× g, 4 °C, 15 min) in between wash cycles. The resulting pellet was then dissolved in 50 mM Tris-HCl pH 8.0 containing 8M urea and incubated at room temperature for two hours to solubilize and unfold all protein content. The His-tagged target proteins were purified from the solubilized pellet by affinity purification using Ni-NTA beads (stated in next sections) before refolding.

Clarified solubilized IBs containing the protein of interest was filtered through 0.2 μm filter and applied onto a column with 2 ml bed volume of Ni²⁺ charged Ni-NTA resin equilibrated with wash buffer (50 mM Tris-HCl pH 8 congaing 150 mM NaCl, 6M Urea and 20 mM Imidazole). After washing step with 15 column volumes of wash buffer, the protein of interest was eluted with 3 column volumes of elution buffer (wash buffer + 250 mM Imidazole), all done at room temperature.

4.4. Protein Reduction and Oxidative Refolding

Oxidative refolding of the affinity-purified TGF-β3 was carried out using a dilution method according to Cerletti patent [12], wherein solubilized IBs were incubated with DTT (typically 10–20 mM) for 30–60 min to reduce disulfide bonds. The mixture was then subjected to centrifugation (55,000× g, 4 °C, 30 min). The resulting supernatant was collected and diluted by 10-fold with a refolding buffer containing 0.1M Tris-HCl (pH 9.5), 0.7M 2-(cyclohexylamino)ethanesulphonic acid (CHES), 1 M NaCl, 3 mM glutathione in 1:1 GSH/GSSG ratio, and 10% (v/v) DMSO. The final protein concentration was at 0.1–0.2 mg/mL. The samples were incubated with gentle stirring at 10 °C for 3–4 days to allow refolding and dimerization of TGF-β3.

4.5. Size Exclusion Chromatography (SEC)

Superdex 200 10/30 column (Cytiva, Uppsala, Sweden) connected with an AKTA Purifier FPLC system (Cytiva, Uppsala, Sweden) was used for SEC. Oxidative refolded proteins were concentrated using a spin concentrator (MWCO 3 kDa) and centrifuged for 5 min in a bench-top centrifuge to remove protein precipitates. Then, the samples were passed through 0.2 μm filter before injection into SEC Superdex 200 10/30 column. A running buffer at pH 8 containing 50 mM Tris-HCl and 150 mM NaCl was used at a flow rate of 0.3 mL/min. Protein elution was collected in 0.2 mL fractions.

4.6. Circular Dichroism (CD) Spectra Measurements

CD analysis was performed using Chirascan spectropolarimeter V100 (Applied Photophysics Ltd., Leatherhead, UK) with a 1 mm path length quartz cuvette. Far-UV CD spectra were measured using SEC buffer (20 mMTris-HCl pH 8, 150 mM NaCl) at 22 °C. Typically, 300 µL of 10 µM proteins were used. Each spectrum represents an average of four scans from 190 to 260 nm at 0.2 nm intervals with the spectra for buffer alone subtracted. CD spectra were normalized by protein molar concentration and number of residues.

4.7. TGF-β3 Activity Assays

Chondrogenic cell line (TC28a2) [23] transfected with luciferase reporter plasmids, SMAD Binding Element nano-luciferase-pest (SBE nLUCp), in which luciferase expression is under control of the SMAD binding element (SBE), was used to evaluate SMAD-induced signaling and transcription. TC28a2 cells cultured at 37 °C, 5% CO₂, in 75 cm² cell culture flasks with DMEM containing 10% (w/v) fetal bovine serum, 1% (w/v) L-glutamine and 1% (w/v) penicillin/streptomycin. For routine maintenance, cells were sub-cultured into flasks containing fresh warmed medium at a passage ratio of 1:10. Briefly, cells were washed twice with PBS before cell dissociation with 5 mL of Trypsin. Cells were then centrifuged at 600× g for 5 min before pellet resuspension in medium and continued culture.

For the nano-luciferase reporter gene assay: A total of 10,000 reporter cells (TC28a2) were seeded into each well of a black walled 96 well plate and incubated overnight, following which cells were serum starved for a further 24 h. Protein samples, diluted in serum free media, were then applied to cells and incubated for 1 h, after which Nanoglo reagent Nano-Glo^® Luciferase Assay System (Promega Corporation, Madison, WI, USA) was applied to cells and luminescence was read using the plate reader. Relative luminesce units (RLUs) are a measure of SMAD 2/3 binding (SBE nLUCp activity) reflecting canonical TGF-β signaling.

4.8. SDS-PAGE

Protein samples were re-suspended in 2× SDS-PAGE-sample buffer −/+ DTT, heated for 10 min at 70 °C, cooled down at RT, centrifuged for 3 min and the supernatant fractionated on 10% (w/v) Bis-Tris gels (NuPAGE™, Invitrogen, Carlsbad, CA, USA; cat.no. STM4004) with NuPAGE™ MES SDS Running Buffer (Invitrogen, Carlsbad, CA, USA). Protein samples taken from the refolding buffer were mixed with folding stopping buffer (FSB) containing 0.1 M iodoacetamide (IAM) in 2× SDS-PAGE sample buffer.

4.9. Western Blotting

Protein samples were separated using SDS-PAGE and electro-transferred to a nitrocellulose membrane, blocked with 5% (w/v) non-fat dry milk, incubated with (1:1000 dilution in 5% (w/v) BSA 1× PBS 0.1% (v/v) tween 20) the primary antibody overnight at 4 °C, followed by incubation with HRP-conjugated secondary antibody (1:2000 dilution) for 1 h at room temperature. The antigen–antibody complex was then visualized using enhanced chemiluminescence (ECL) system. The ECL detection system (Bio-Rad, Hercules, CA, USA) was utilized for chemiluminescence development. Between each step the membrane was washed three times for 5 min with 1× PBS 0.1% (v/v) tween 20.

The primary antibodies used were mouse anti-human TGF-β3 (Bio-Techne (Minneapolis, MN, USA), MAB643), referred to as cTGF-β3 antibody; goat anti-human TGF-β3 (Bio-Techne, BAF243), referred to as mTGF-β3 antibody; phospho-SMAD2 rabbit monoclonal antibody (Cell Signaling Technology (Danvers, MA, USA), 138D4); and SMAD2 rabbit antibody (Cell Signaling Technology, D43B4). The secondary antibodies used were HRP-conjugated goat anti-mouse antibody (Invitrogen, cat. 31430) and HRP-conjugated rabbit anti-goat antibody (Sigma-Aldrich (St. Louis, MO, USA), cat. AP106P), as well as HRP-conjugated goat anti-rabbit antibody (Invitrogen, cat. 31460).

4.10. Protein Quantifications

Concentrations of purified protein were determined using a NanoDrop 2000 spectrophotometer (Thermo Scientific, Waltham, MA, USA). An extinction coefficient of 23,482 M⁻¹cm⁻¹ at 280 nm measured in an aqueous solution at an unreduced state and a molecular weight of 14,526.32 Da calculated using ProtParam tool from the ExPASy server (https://web.expasy.org/protparam/, accessed on 10 March 2021) [27] were used. For protein quantification from gels, ImageJ software (version 1.48v, National Institutes of Health, Bethesda, MD, USA) was used for densitometry scanning. For activity assays, the total protein concentrations in the tissue lysate were determined using Micro Bicinchoninic acid (BCA) assay kit (cat. no. 23235, Thermofisher Scientific, Waltham, MA, USA).