Irisin, a proteolytic cleavage product of a transmembrane protein, fibronectin type III domain-containing protein 5 (FNDC5), is a myokine primarily expressed in skeletal muscle (Böstrom et al. 2012). Böstrom et al. (2012) first reported that the overexpression of proliferator-activated receptor gamma coactivator 1-alpha (PGC1α), a major mediator in the beneficial effect of exercise in muscle (Handschin and Spiegelman 2008), led to an increase in the expression of FNDC5, which is then secreted into circulation as irisin during exercise (Böstrom et al. 2012). They also observed that irisin mediates the conversion of white adipose tissue into brown adipose tissue by increasing the expression of uncoupling 1 (UCP1), resulting in increased energy expenditure and amelioration of obesity-induced insulin resistance. The discovery of irisin has provided evidence for the importance of crosstalk between muscle and other tissues, including adipose tissue, liver, kidney, brain, and bone (Momenzadeh et al. 2022).

Subsequent studies have indicated that irisin not only plays a pivotal role in energy metabolism but also exerts other beneficial effects such as anti-oxidant, anti-inflammatory, anti-metastatic, and cytoprotective effects in various tissues (Rabiee et al. 2020). Irisin exerts its biological effects through several intracellular signaling pathways. Irisin-induced MAPK signaling pathways are involved in neural differentiation, adipocyte browning, and osteoblast proliferation and differentiation. Other pathways involved in irisin action include AMPK, PI3K/AKT, and STAT3/Snail pathways. Despite controversies on the circulating levels of irisin in physiological and pathophysiological conditions (Albrecht et al. 2015; Crujeiras et al. 2015; Polyzos et al. 2015), studies that tested the therapeutic potential of irisin by injecting recombinant irisin have shown promising results (Liu et al. 2022). Therefore, it is important to efficiently produce active irisin and further characterize its potential mechanism against diseases.

To date, many studies have employed the Pichia pastoris system to obtain recombinant irisin (Zhang et al. 2014; Duan et al. 2015), although some reports have suggested that the recombinant irisin from the Escherichia coli expression system also has activity (Reza et al. 2017). Studies have reported that the glycosylated form of irisin has a significant biological function, which will be lost without glycosylation modification (Zhang et al. 2014). The P. pastoris system has several advantages for exogenous gene expression, including correct processing, folding, proper glycosylation, and other post-translational modifications (Hamilton et al. 2006). However, there is a need to develop means to construct a high-yield recombinant strain with optimal human irisin coding sequence and proper glycosylation.

Plants are considered a promising platform for the production of various pharmaceutical proteins because of their safety, cost effectiveness, and easy scalability. Further, recombinant protein production in plants has the advantage of few concerns about animal pathogens such as viruses, bacteria, and prions (Yusibov et al. 2011). In addition, the plant platform is easy to scale-up and has low infrastructure costs compared to the fermenter-based production platform (Tschofen et al. 2016; Edgue et al. 2017; Islam et al. 2019). Therefore, the plant platform has been widely accepted and advanced for the production of antibodies, vaccines, and protein therapeutics (van Dolleweerd et al. 2014; Park et al. 2019; Kim et al. 2020; Kim et al. 2023). In this study, we cloned and expressed human irisin (r-irisin) in Nicotiana benthamiana leaves and tested its biological activity in terms of adipocyte browning and mitochondrial biogenesis. Upon treatment with plant-made r-irisin on adipocytes, increased levels of UCP-1, PGC1α, nuclear respiratory factor 1 (NRF1), and mtTFA were observed, implicating their therapeutic potential in the treatment of obesity-induced metabolic diseases.

Construction and expression of plant-made irisin

The human irisin gene was synthesized by Genscript (Piscataway, NJ, USA) using an amino acid sequence from genbank number NP_715637.2 (amino acid 32-143) with codon optimization for N. benthamiana. The synthesized irisin gene was fused with nucleotides encoding an endoplasmic reticulum (ER) signal peptide from Arabidopsis-binding protein (BiP), a 6-histidine tag, and a histidine-aspartate-glutamate-leucine (HDEL) tetrapeptide. Then, the DNA fragment encoding BiP:His:irisin was cloned into a pCAMBIA1300 vector harboring the Cauliflower mosaic virus (CaMV) 35S promoter and heat shock protein terminator (Kim et al. 2020).

The plant-made irisin was transiently expressed as previously described with minor modifications (Kim et al. 2023). Briefly, Agrobacterium tumefaciens LBA4404 was transformed with irisin expression vector and cultured in YEP medium. The harvested Agrobacterium was resuspended in an infiltration buffer (10 mM MES [pH 5.6], 10 mM MgCl₂, and 100 mM acetosyringone) and was vacuum-infiltrated into N. benthamiana leaves. Four days after infiltration, leaves were harvested. For the solubility test, the harvested leaves were ground using a mortar and pestle, and the leaf powder was incubated with an extraction buffer (50 mM sodium phosphate [pH 8.0], 300 mM NaCl, 10 mM imidazole, 0.5% Triton X-100, 10 mM ascorbic acid, 100 mM Na₂SO₃, 1.5% polyvinylpolypyrrolidone). Then suspension was filtered using Miracloth (EMD Millipore Corp., Billerica, MA, USA). The filtrate was centrifuged at 20,000 ×g for 15 min, and the supernatant and pellet were collected. Each fraction was analyzed using western blot analysis with anti-His antibody (BioLegend, San Diego, CA, USA).

Purification and de-glycosylation of plant-made irisin

The plant-made irisin was purified as previously described with minor modification (Kim et al. 2023). Briefly, harvested leaves were homogenized in a blender (32,000 rpm) in the presence of an extraction buffer. To remove debris, the extract was centrifuged for 30 min at 12,000 ×g, and the supernatant was filtered through Miracloth. The extract was stirred for 1 h with Ni-IDA agarose resin (Clontech, Kyoto, Japan). The resin was washed with washing buffer (50 mM sodium phosphate [pH 8.0], 300 mM NaCl, 10 mM imidazole, 0.5% Triton X-100). The resin was finally washed with 50 mM imidazole to remove non-specific protein, and the protein was eluted with 300 mM imidazole. The eluted protein was exchanged with a final buffer (50 mM Tris-Cl [pH 7.2], 300 mM NaCl, 0.5 mM EDTA). The plant-made irisin was stored at −80℃ until further use.

For the de-glycosylation of the plant-made irisin, purified irisin was treated with endoglycosidase H (endo H) following the manufacturer’s protocol (New England Biolabs, Ipswich, MA, USA). Then, endo H-treated plant-made irisin was analyzed using SDS-PAGE followed by Coomassie brilliant blue staining.

Cell culture and treatment

The 3T3-L1 cell line was obtained from the American Type Culture Collection (Manassas, VA, USA) and cultured in Dulbecco’s modified Eagle medium (HyClone, Logan, UT, USA) with 10% fetal bovine serum (FBS, HyClone) and 1% penicillin/streptomycin (Himedia, Maharashtra, India) and maintained in a 5% CO₂ incubator at 37℃. Two days after reaching a confluence of 100% (day 0), differentiation into adipocytes was induced with 1 µg/mL insulin (Sigma, St. Louis, MO, USA), 0.25 µM dexamethasone (Sigma), 0.5 mM 1-methyl-3-isobutylxanthine (Sigma), and 1 µM rosiglitazone (Sigma) in DMEM supplemented with 10% FBS and 1% penicillin/streptomycin. After 48 h of differentiation induction, the medium was changed with 1 µg/mL insulin and DMEM supplemented with 10% FBS and 1% penicillin/streptomycin. Then, the cell culture medium was changed with a medium containing 10% FBS every 2 days during 6 days of culture. On day 8 after differentiation, cells were treated with 100 ng/mL of plant-made recombinant irisin or commercially available recombinant irisin (cat#067-29; Phoenix Pharmaceuticals, Inc., Burlingame, CA, USA). After treatment, cells were harvested for gene or protein expression analysis. To examine the effect on adipogenesis, 100 ng/mL recombinant irisin was treated to post-confluent preadipocytes from the start of differentiation, every other day for 6 days.

Gene expression analysis

Total RNA was isolated using TRIzol (MRC, Cincinnati, OH, USA), and cDNA synthesis was performed using TOPscript™ RT DryMIX (Enzynomics, South Korea). Quantitative real-time polymerase chain reaction (PCR) was performed using the TOPreal SYBR Green PCR Kit (Enzynomics). mRNA levels were detected with real-time PCR using the Rotor-Gene Q PCR cycler (QIAGEN, Hilden, Germany). The mRNA levels were normalized to 18S ribosomal RNA.

Western blotting

3T3-L1 cells were homogenized in RIPA buffer (Thermo Scientific, Rockford, IL, USA). Protein concentration was determined using the Pierce BCA Protein Assay Kit (Thermo Fisher Scientific). Western blotting was performed as described previously (Ma et al. 2019) using anti-UCP1 (ab209483; Abcam, Cambridge, UK), anti-PPARγ (#2435; Cell Signaling, Danvers, MA, USA), anti-C/EBPα (#8178; Cell Signaling), anti-FABP4 (#2120; Cell Signaling) and anti-β-tubulin (#2146; Cell Signaling) antibodies.

Oil red O staining

The 3T3-L1 cells were differentiated in 24 well plates and treated with recombinant irisin for 24 and 48 h on day 8 of differentiation. Then, the cells were washed with phosphate buffer saline and fixed with 10% formaldehyde for 1 h. The fixed cells were stained with filtered Oil Red O solution (O0625-25G, Sigma) for 1 h at room temperature. Subsequently, the staining solution was removed and washed with distilled water. For quantitative analysis of stained cells, absorbance was measured at 490 nm using a Leica microplate reader (Leica, Wetzlar, Germany).

Statistical analysis

Data were analyzed using Statview v5.0 (SAS Institute Inc., Cary, NC, USA). All data are reported as mean ± standard error of the mean. The mean values obtained from each group were compared through one-way analysis of variance (ANOVA) followed by Fisher’s protected least significant difference post hoc test. Graphs were constructed using GraphPad Prism 8.0 (San Diego, CA, USA). p-values less than 0.05 were considered to be statistically significant.

Expression and purification of plant-made irisin

To express irisin in N. benthamiana, a plasmid was constructed as shown in Fig. 1A. The irisin gene was codon optimized for N. benthamiana expression and then fused with a 6-histidine tag at the N-terminus of irisin. To accumulate His-irisin in the ER, a signal peptide from BiP and an ER retention signal were fused at the N- and C-termini, respectively.

Figure 1. Expression and purification of plant-made irisin. (A) Schematic representation of the plant-made irisin-expressing vector. (B) Solubility test result of plant-made irisin. (C, D) SDS-PAGE/CBB staining of eluate (C) from Ni-IDA chromatography and purified irisin (D) after buffer change. Bovine serum albumin (BSA) was used as protein standard. (E) SDS-PAGE/CBB staining of endo H-treated plant-made irisin. Asterisks (*) indicate the N-glycosylated form, while arrow indicates de-glycosylated irisin. CaMV 35S Pro, Cauliflower Mosaic Virus 35S promoter; BiP, Binding protein signal peptide; 6xHis, 6-histidine tag; HDEL, histidine-aspartate-glutamate-leucine tetrapeptide; HSP Ter, heat shock protein terminator; M, protein size marker; T, total fraction; S, soluble fraction; P, pellet fraction.

Next, to evaluate the expression and solubility of the plant-made irisin, proteins were extracted from leaves infiltrated with Agrobacterium harboring the irisin expression vector and analyzed with western blot. Two protein bands were mainly detected between 17 and 28 kDa in the soluble fraction (Fig. 1B). Because a 6-histidine tag was fused to irisin for purification, the plant-made irisin was purified with Ni-IDA chromatography. Although the His-irisin protein was mainly eluted in the elution buffer containing both 100 and 300 mM imidazole, only 300 mM imidazole eluate was collected for higher purity (Fig. 1C). Finally, after changing the buffer, three protein bands were observed between 25 and 17 kDa in SDS-PAGE followed by Coomassie brilliant blue staining. It has been reported that irisin protein has two putative N-glycosylation sites (Zhang et al. 2014). Therefore, the upper two protein bands were suspected to be N-glycosylated forms of irisin. To test this possibility, purified irisin was treated with endo H. In the presence of endo H, the upper two protein bands disappeared and the intensity of the bottom protein band increased, confirming that the upper two protein bands were N-glycosylated (Fig. 1D).

Effect of plant-made human recombinant irisin on UCP1 expression in adipocytes

To evaluate the performance of plant-made r-irisin we first compared its effect on the expression on UCP1, the primary target of irisin in adipocytes, with that of commercially available recombinant irisin obtained from Phoenix Pharmaceuticals. Of note, recombinant irisin from Phoenix Pharmaceuticals is derived from the E. coli expression system. On day 8 of differentiation, 3T3-L1 cells were treated with 100 ng/mL recombinant irisin. Then, UCP1 expression was examined with real-time PCR or western blot. No change in UCP1 gene expression was observed 6 h after treatment with both the plant-made recombinant irisin and recombinant irisin from Phoenix Pharmaceuticals. After 48 h of treatment, the plant-made r-irisin significantly increased UCP1 gene expression, whereas the r-irisin from Phoenix Pharmaceuticals only showed an increasing trend (Fig. 2A). Next, cells were harvested after 24 and 48 h of treatment for protein expression analysis. Cells treated with both the plant-based r-irisin and r-irisin from Phoenix Pharmaceuticals showed a significantly upregulated UCP1 protein levels (Fig. 2B). These findings confirm that the plant-made r-irisin is effective in inducing UCP1 gene and protein expressions in adipocytes. To verify that recombinant irisin induces energy expenditure through upregulation of UCP1 expression, we used Oil Red O staining to examine lipid accumulation. As a result, 24 h treatment of both the plant-made r-irisin and r-irisin from Phoenix Pharmaceuticals downregulated lipid accumulation in mature adipocytes, which became more evident after 48 h treatment (Fig. 2C, 2D).

Figure 2. Effect of plant-made human recombinant irisin (r-irisin) on UCP1 expression in adipocytes. (A) UCP1 gene expression was quantified using real-time PCR and normalized to 18S. (B) UCP1 protein expression was measured using western blot and normalized to β-tubulin. (C, D) Representative images (×200) and quantification of Oil Red O staining in 3T3-L1 mature adipocytes. Cells were treated with plant-made r-irisin or Phoenix r-irisin on day 8. Original magnification ×20. *p < 0.05 vs control.

Effect of plant-made human recombinant irisin on mitochondrial biomarkers in adipocytes

Next, we further evaluated the effect of the plant-made human r-irisin on the expression of genes related to mitochondrial biogenesis, including PGC1α, NRF1, and mtTFA (Fig. 3). After 6 h of treatment, only PGC1α gene expression was significantly increased by the plant-made r-irisin, whereas after 48 h of treatment, the mRNA levels of PGC1α, NRF1, and mtTFA were significantly upregulated, implicating the biological activity of plant-made irisin in mitochondrial biogenesis.

Figure 3. Effect of plant-made human recombinant irisin (r-irisin) on genes related to mitochondrial biogenesis in adipocytes. (A-C) mRNA expression was quantified using real-time PCR and normalized to 18S. *p < 0.05 vs control, ^†p < 0.05 vs plant-made r-irisin at 6 h.

Effect of plant-made human recombinant irisin on adipogenesis

We have previously reported that irisin exerts inhibitory effect on adipogenesis (Ma et al. 2019). To test whether the plant-made r-irisin can also inhibit adipocyte differentiation, 3T3-L1 cells were treated with recombinant irisin during adipogenesis. At day 6 after the differentiation, protein expressions of FABP4, C/EBPα, and PPARγ significantly downregulated (Fig. 4A), which resulted in less accumulation of lipids by both the plant-made r-irisin and r-irisin from Phoenix Pharmaceuticals (Fig. 4B, 4C). These results imply that the plant-made r-irisin possesses the ability to inhibit adipocyte differentiation through regulation of transcription factors.

Figure 4. Effect of plant-made human recombinant irisin (r-irisin) on adipocyte differentiation. Post-confluent 3T3-L1 preadipocytes were treated with plant-made r-irisin or Phoenix r-irisin every two days during differentiation process. (A) Protein expression of adipogenic markers were measured at day 6 using western blot and normalized to β-tubulin. (B, C) Representative images (×200) and quantification of Oil Red O staining in 3T3-L1 adipocytes at day 8 of differentiation. *p < 0.05 vs control.

Irisin is considered a promising myokine that links the beneficial effects of exercise to improvement of health. To date, several studies have provided evidence on the positive effects of irisin on adipose tissue browning, bone homeostasis, and cognitive function (Böstrom et al. 2012; Lee et al. 2014; Kim et al. 2018; Lourenco et al. 2019). Despite these results on the physiological effects of irisin and its potential as a therapeutic target, studies on the efficient production of active irisin are lacking. In the present study, we successfully constructed, expressed, and purified the glycated form of human irisin using N. benthamiana. Subsequent evaluation of its biological activity has confirmed that the plant-made r-irisin induces thermogenic program via UCP1 induction as well as upregulates mitochondrial biogenesis-related genes.

Although the predicted size of irisin is 12.7 kDa, irisin in humans exists in different molecular weights (MWs) because of post-translational glycosylation. Previous studies have reported non-glycosylated irisin produced from bacteria to be around 13 kDa, whereas glycosylated irisin produced from HEK293 cells had an apparent MW of 20-25 kDa (Albrecht et al. 2015) . Other studies have found that recombinant irisin produced from P. pastoris has two N-glycosylation sites at asparagine 7 (Asn7) and asparagine 52 (Asn52) (Zhang et al. 2014). Glycosylation at both Asn7 and Asn52 sites produced the largest size irisin (25 kD), glycosylation at a single site produced mid-sized irisin (20 kD), and non-glycosylation produced the smallest irisin (17 kD) (Zhang et al. 2014; Duan et al. 2015). Indeed, in our results, the plant-made r-irisin exhibited three protein bands between 25 and 17 kDa. Treatment with endo H caused the upper two protein bands to disappear and increased the intensity of the bottom protein band, confirming that the upper two protein bands were N-glycosylated.

The majority of secretory proteins have been suggested to be N-linked glycoproteins (Scott and Panin 2014), and the presence of glycan may contribute to secretory protein folding in the ER and maturation in the Golgi (Aebi et al. 2010; Aebi 2013). Consistently, Nie and Liu (2017) showed that FNDC5 is N-glycosylated at Asn36 and Asn81 sites and that removal of glycosylation site significantly affects the stabilization and localization of FNDC5, which is mostly retained in the ER. Although this study provided evidence for the role of N-glycosylation in FNDC5 stabilization and irisin secretion, it is unclear whether irisin’s bioactivity differs between glycosylated and non-glycosylated forms. While the study by Zhang et al. has suggested that glycosylated irisin may present high bioactivity compared with low levels of glycosylated irisin or non-glycosylated irisin (Zhang et al. 2014), other studies have reported that non-glycosylated irisin also has bioactivity (Reza et al. 2017; Panati et al. 2018). In our study, the plant-made r-irisin and commercially available r-irisin produced from E. coli both effectively induced UCP1 expression, questioning the role of glycosylation in the biological function of irisin.

Most previous reports have used recombinant irisin from either E. coli, P. pastoris, or HEK293 cells. In particular, the P. pastoris expression system has been successfully used over the past decade because it can form proper glycosylation and has many other advantages including correct processing, folding, post-translational modification, keeping antigenic determinants yielding robust immune responses, and direct secretion into liquid medium facilitating purification (Duan et al. 2015). For recombinant protein production, the plant system is attractive because of its scalability, safety, post-translational modification, and low cost. In particular, in terms of post-translational modification, while protein N-glycosylation from P. pastoris is hyper-mannose type, mature N-glycosylation pattern from plants are similar to that from animal cells (Cereghino and Cregg 2000; Nagashima et al. 2018). Plant-derived glycoproteins especially contain non-mammalian epitopes such as β(1,2)-xylose and α(1,3)-fucose residues, which may play a role as putative allergens or immunogen motifs. However, recent results from phase I/II clinical trials have revealed that allergy/hypersensitivity was not detected in participants receiving two doses of vaccine including plant-derived SARS-CoV-2 spike glycoprotein and adjuvants (Ward et al. 2021; Charland et al. 2022), suggesting that the plant system is a powerful and safe platform to produce recombinant proteins.

The biological activity of the plant-made r-irisin has been examined in 3T3-L1 adipocytes. The results showed that the plant-made r-irisin is effective in inducing adipocyte browning, as shown by increased UCP1 gene and protein expressions and decreased lipid accumulation, which were comparable to that of the most frequently used commercially available r-irisin. The results underline the active participation of purified r-irisin in the induction of thermogenic program, consistent with previous studies performed with r-irisin expressed in P. pastoris (Zhang et al. 2014). Although the UCP1 mRNA expression was not significantly elevated by r-irisin from Phoenix Pharmaceuticals, it significantly upregulated UCP1 protein expression, implying that Phoenix r-irisin may have induced UCP1 gene expression in a different timeframe. In addition, expressions of genes related to mitochondrial biogenesis were also significantly upregulated. Similar to our results, it was reported that in C2C12 myocytes, irisin treatment significantly increased oxidative metabolism, mitochondrial uncoupling, and the expression of PGC1α, NRF1, mtTFA, glucose transporter 4, and mitochondrial uncoupling protein 3 (Vaughan et al. 2014). Furthermore, plant-made r-irisin was able to inhibit adipocyte differentiation as evidenced by reduction in adipogenic markers, which could play additional role in treatment of obesity.

In addition to developing an optimal condition for the production of large-scale recombinant irisin that resembles its native form, modifications to improve and overcome the limitations of irisin are also needed. For example, the short half-life (1 h) of injected recombinant irisin is a major issue. In a recent study, integrins, primarily complexes involving alpha V integrin, were identified as long-sought receptors mediating the effects of irisin on bone and fat in mice (Kim et al. 2018), which opens up new possibilities for the synthesis of structure-based irisin receptor ligand. It is also important to note that biochemical studies that determined the structure of irisin by X-ray crystallography revealed that irisin forms a continuous intersubunit β-sheet dimer, which has important implications for receptor activation and signaling (Schumacher et al. 2013).

In conclusion, numerous studies have demonstrated the physiological properties of irisin, pointing to its beneficial health potential for various tissues and organs. With the promise of its health benefits, plants will provide safe, efficient, cost-effective means for the production of various pharmaceutical proteins, including irisin. Further studies to test the therapeutic applications of the plant-made r-irisin, including in vivo studies, are highly anticipated.

The authors Kyungmin Min, Minhee Park, Hyangju Kang, Bo-Hwa Choi, Eun-Ju Sohn are employees of BioApplications Inc., Pohang, Republic of Korea. Any other authors have no conflict of interest.

This study was supported by the National Research Foundation of Korea (NRF) grant funded by the Korean government (MSIT) (no. 2018R1C1B6003470 and 2022R1A2 C1002956).