Characterisation of a New Fungal Immunomodulatory Protein from Tiger Milk mushroom, Lignosus rhinocerotis

Lignosus rhinocerotis (Tiger milk mushroom) is an important folk medicine for indigenous peoples in Southeast Asia. We previously reported its de novo assembled 34.3 Mb genome encoding a repertoire of proteins including a putative bioactive fungal immunomodulatory protein. Here we report the cDNA of this new member (FIP-Lrh) with a homology range of 54–64% to FIPs from other mushroom species, the closest is with FIP-glu (LZ-8) (64%) from Ganoderma lucidum. The FIP-Lrh of 112 amino acids (12.59 kDa) has a relatively hydrophobic N-terminal. Its predicted 3-dimensional model has identical folding patterns to FIP-fve and contains a partially conserved and more positively charged carbohydrates binding pocket. Docking predictions of FIP-Lrh on 14 glycans commonly found on cellular surfaces showed the best binding energy of −3.98 kcal/mol to N-acetylgalactosamine and N-acetylglucosamine. Overexpression of a 14.9 kDa soluble 6xHisFIP-Lrh was achieved in pET-28a(+)/BL21 and the purified recombinant protein was sequence verified by LC-MS/MS (QTOF) analysis. The ability to haemagglutinate both mouse and human blood at concentration ≥0.34 μM, further demonstrated its lectin nature. In addition, the cytotoxic effect of 6xHisFIP-Lrh on MCF-7, HeLa and A549 cancer cell lines was detected at IC50 of 0.34 μM, 0.58 μM and 0.60 μM, respectively.

Computational studies can help to elucidate the structure and function of a new protein using simple homology modelling to predict the three dimensional (3-D) structure and docking techniques to determine its ligand binding sites with sufficient accuracy. The available 3-D structures of FIP-fve (1OSY.pdb), FIP-gmi (3KCW.pdb) from G. microsporum and FIP-glu (3F3H.pdb) show FIPs to occur constitutively as homo-dimers, arranged non-covalently in a dumb-bell-shaped similar to the variable region of Ig heavy chains 12,17,25 . The FIP-fve is folded into an Ig-like β -sandwich, each homodimer subunit consists of an N-terminal α -helix dimerization domain, followed by a fibronectin III (FNIII)-type fold. Dimerization is critical for the activity of the FIPs and occurs by swapping of the N-terminal helices and held strongly by hydrophobic interactions. Likewise, FIP-glu was found to resemble the structure of FIP-fve 17,25 (Fig. 1). On the contrary, FIP-gmi, occurs as a tetramer instead of dimer and the arrangement of loops and conformation are significantly different from FIP-fve.
FIPs contain a carbohydrate binding module (CBM) in the β -sandwich at the C-terminal 26 . It is similar to the noncatalytic CBM family 34 (CBM-34), a β -sandwich folding family that acts as a granule starch-binding domain 26,27 . Therefore, FIP can bind to complex sugars such as dextrin, cyclodextrin and N-acetyl neuraminic acid on cell surface, which explains why majority of the reported FIPs are lectin in nature with different ability to agglutinate red blood cells (RBCs) from rat, mouse, sheep or human. Due to their small size, FIPs can be easily modified, with potential use in wide-ranging industrial applications 16 .
Recent genome sequencing of L. rhinocerotis 5 revealed two putative FIPs (GME7566_g and GME10641_g) (FIP-Lrh), both exhibited 64% identity to FIP-glu. Here, we report the subsequent cloning and characterisation of the FIP-Lrh cDNA, including structure modelling for prediction of its putative carbohydrate binding module (CBM) and binding affinity to different cellular ligands. In addition, recombinant FIP-Lrh produced in E. coli was used to determine its haemagglutination ability and cytotoxicity on three cancer cell lines.

Results and Discussion
Isolation and cloning of FIP-Lrh cDNA. The extracted total RNA (yield of 0.153 μ g/mg of sclerotia) ( Fig. 2) when used in RT-PCR gave a PCR product of ~480 bp of the expected size which was subsequently cloned into pGEMT. Four recombinant clones were obtained, designated as pGEM_FIP_Lrh_1, pGEM_FIP_Lrh_2, pGEM_FIP_Lrh_3, and pGEM_FIP_Lrh_4. The pGEM_FIP_Lrh_3 gave a smaller sized PCR product (~450 bp) compared to the other three clones when subjected to PCR screening using the gene primer pairs (Fig. 2). The translated DNA sequence of inserts from the four clones showed that pGEM_FIP_Lrh_3 was 100% identical to the putative FIP-Lrh. The pGEM_FIP_Lrh_1, 2 and 4 contained an additional fragment of 18 amino acids (aa) followed by a stop codon which is probably part of an un-spliced intron of RNA processing due to the presence of a predicted acceptor splice site preceding the Met initiation codon (Fig. 3). Likewise, Zhou et al. 13 reported the presence of a 61 bp intron at the 5′ -flanking region preceding the Met initiation codon of FIP-gsi. Since the forward primer was designed to start at a methionine of the genomic putative FIP-Lrh ORF, untranslated sequences further upstream of the FIP gene was not available for further analysis.
Reported FIP protein sequences from other mushroom species showed the putative Methionine (Met) initiation codon starts preceding a serine. Correspondingly, multiple alignment of protein sequences, starting from this putative Met, for the four cloned FIP-Lrh cDNA and the reported FIPs from seven mushroom species are as shown in Fig. 4. The pGEM_FIP_Lrh_2 exhibit one conserved amino acid substitution at position 87 (isoleucine to valine) and a non-conserved substitution at position 98 (valine to aspartic acid) while pGEM_FIP_Lrh_4 showed two substitutions, at position 11 (valine to isoleucine) conserved and at position 80 (glutamic acid to glycine) non-conserved, respectively. Since the pGEM_FIP_Lrh_1 and pGEM_FIP_Lrh_3 cDNA is 100% identical to the putative FIP-Lrh, it was used for subsequent analysis. The 29 amino acids fragment preceding the Met initiation codon of pGEM_FIP_Lrh_3 did not match to any signal peptides indicating that it may possibly be the 5′ untranslated region of the cDNA.
Open reading frame (ORF) for the full-length FIP-Lrh cDNA encodes a protein of 112 aa (336 bp), with a calculated molecular mass of 12.59 kDa. Similar to mature FIPs from other mushrooms, FIP-Lrh contains a serine at the N-terminal that may be acetylated during post-translational modification. FIP-Lrh is predicted to be stable (instability index of − 0.44, aliphatic index of 73.75) and contain mainly hydrophilic regions, except for the relatively hydrophobic middle region of the protein (Fig. 5a), unlike other FIPs with relatively hydrophobic N-terminals. Similar to other FIP proteins, it contains no Cys, Met (except for the initiation codon) or His residues, but it contains numerous Asp (n = 9), Asn (n = 10), Lys (n = 10), Ser (n = 9), Gly (n = 9), Tyr (n = 9) and Val (n = 15) residues. The FIP-Lrh cDNA exhibited 54-64% identity to reported FIPs, highest identity is to FIP-glu (LZ-8) (63.96%), followed by FIP-tvc (59.46%), FIP-gap (58.56%), FIP-gsi (58.56%), FIP-gja (57.66%), FIP-fve (54.05%) and FIP-vvo (53.57%). Correspondingly, a phylogeny analysis using the protein sequences revealed that the aligned FIP from Lignosus (FIP-Lrh cDNA) arise from a closer common ancestor with FIPs from Flammulina, Ganoderma and Trametes compared to that for Volvariella (Fig. 4). As previously reported, the FIP-gsi and FIP-gja were clustered into one subgroup, suggesting a close relationship 13 . Structure modelling and ligand docking. Since the functional form of the FIP-Lrh is a homodimer, the three dimensional structure of FIP-Lrh was predicted as a homodimer (Fig. 5b). It had 53.6% sequence identity and root mean square deviation score of 0.38 Å with its template FIP-fve (1OSY.pdb). The discrete optimised energy (DOPE) score of the predicted structure was − 22852.3 kcal/mol which is an indication of the stability of the structure. The Ramachandran plot (Fig. 5d) of the predicted structure showed the bond length and angles were within limits with 87% of the amino acids in the core region. Four o-linked glycosylation were also predicted on each monomer. Ser1, Thr4, Thr6 and Ser58 of each monomer were predicted to be O-linked while none N-linked glycosylation sites were predicted due to lack of consensus Asn-Xxx-Ser/Thr sites 28,29 .
The N-terminal of FIP-Lrh monomer forms a single α -helix joined to seven β sheets sandwiched at the C-terminal via a loop, similar to structures of FIPs reported earlier 17 . The β -sheet residues fold in immunoglobulin-like (Ig) domain indicating the functional similarity of FIP-Lrh to cell adhesion proteins. However it lacks Cys, Met and His residues usually found in Ig structures 26 . As reported previously by Liu et al. 26 , the β -sandwich of FIP-Lrh contains the CBM-34 domain; the identified six key residues of W24, T28, D34, T90, I91, and W111 in this domain of FIP-fve monomer bind to glycans on the membrane of human peripheral blood mononuclear cells (hPBMCs), promoting production of IFN-γ by T-cells. Four of these key residues were also present in FIP-Lrh. However their relative position was slightly different such as W25, D35, I92 and W110, whereas two residues, T28 and T90 of FIP-fve, were substituted by a hydrophilic and polar N29 and a positively charged K91 respectively, at comparable positions in FIP-Lrh (Fig. 5c, green coloured chain). Drawing on the experimental results of FIP-fve by Liu et al. 26 subsequent docking analysis using glycans predicted that FIP-Lrh binds galactose, glucose, glucose-6-phosphate, maltose, mannose, N-acetylgalactosamine and N-acetylglucosamine with comparable binding energy as FIP-fve (Fig. 6). The CBM-34 binding pocket of FIP-fve comprised of hydrophobic and neutral amino acids and was experimentally demonstrated to bind strongly to N-acetylneuraminic acid, maltriose, cyclodextrin and dextrin, important for eliciting immune responses 26 . Comparatively, FIP-Lrh showed overall lower binding energies towards tested glycans especially for cyclodextrin, dextrin and N-acetylneuraminic acid. Among the 14 glycans, FIP-Lrh showed the best binding to N-acetylglucosamine and N-acetylgalactosamine with binding energy of − 3.98 kcal/mol. Presence of a more positively charged CBM-34 binding pocket in FIP-Lrh as predicted by in silico analysis explained for the decreased interaction with cyclodextrin, dextrin and maltriose, while the negatively charged N-acetylgalactosamine and N-acetylglucosamine can bind strongly. This is in accordance with the results of Liu et al. who experimentally proved the substitution of the residues in the CBM-34 domain led to decreased immunomodulatory and haemagglutination activity, especially T28 resulted in little or complete abolition of IFN-γ in hPBMCs and haemagglutination activity while substitution of T90 resulted in reduced activity as compared to wild type. In FIP-Lrh these two crucial residues were replaced with arginine (N29) and lysine (K90) resulting in overall low binding energy prediction (Fig. 6) 26  Expression and purification of 6xHisFIP-Lrh. Further functional studies of FIP-Lrh can be facilitated by the availability of an active recombinant FIP-Lrh, since direct purification of this protein from the mushroom is costly, time-consuming and low in yield. The FIP-Lrh cDNA was cloned in-framed into pET-28a(+ ) and preliminary time-point total lysate profile of p2/BL21 expression culture is as shown in Fig. 8. Overexpression of a soluble 6xHisFIP-Lrh protein of approximately 14.9 kDa (including the 22 aa of the vector sequence and 6xHis tag) was observed at 1 hr after 1 mM IPTG induction, which was absent in the uninduced sample (t = 0). The amount of protein obtained increased as indicated by an increase in the intensity of the recombinant protein band with the duration of expression from 1 to 4 hr after induction. Purification of the 6xHisFIP-Lrh was performed under native condition where protein binding was optimal at pH 8.0, as minimal protein was found in the flow through from the resin (Fig. 8). The presence of imidazole (20-50 mM) in the binding buffer and wash buffer prevented binding of unspecific E. coli host proteins, therefore increasing the efficiency of the purification. Yield of the purified 6xHisFIP-Lrh was 9 mg/50 mL of expression culture and was used for subsequent haemagglutination and anti-proliferative assays.  immunomodulatory protein (accession number 10641) from L. rhinocerotis genome database with mean peptide spectral intensity of 3.21E+ 06 (Table 1). Peptide fragments from tryptic digested 6xHisFIP-Lrh covered 71.4% of amino acid sequence of the recombinant protein, or 85.6% of the annotated FIP protein (FIP-Lrh), thus establishing the identity of the recombinant protein as FIP.

Verification
Haemagglutination and cytotoxicity of 6xHisFIP-Lrh. Results demonstrated that the 6xHisFIP-Lrh can agglutinate both human and mouse RBCs at concentration of ≥ 0.34 μ M (5 μ g/mL); greater degree of agglutination in mouse cells was observed compared to human cells (Fig. 9). Thus, similar to other FIPs, the FIP-Lrh is also part of a family of fungal lectins 30 with a functional CBM binding pocket confirming its glycans binding ability predicted by in silico analysis.
In addition, preliminary investigation also showed that 6xHisFIP-Lrh exhibited a dose dependent cytotoxic effect on HeLa and A549 cells with an IC 50 of approximately 0.58 μ M (8.64 μ g/mL) and 0.60 μ M (8.94 μ g/mL), respectively, whereas a lower IC 50 of 0.34 μ M (5.07 μ g/mL) was observed for MCF-7 cells (Fig. 10). The positive control with DMSO, likewise showed dosage dependent cell killing (IC 50 = 2.5-3.2%). This is in accordance to the cytotoxic effect on specific cancer cells lines reported for other recombinant FIPs. FIP from Nectria haematococca (FIP-nha) showed strong dose dependent antitumor effect on human gastric cancer MGC823 and liver cancer HepG2 cell lines with IC 50 values of 15.54 and 12.24 μ g/mL, respectively 31 . The rLZ-8 inhibited cancer cell proliferation by increasing G1 arrest 32-34 , rFIP-gts suppressed human alveolar A549 cells proliferation via the and gaps are introduced for optimal alignment and maximum similarity between all compared sequences. The identical amino acids among all the aligned sequences are indicated as '* ' whereas ":" conserved substitutions and empty space represents a non-conserved substitution. and the identical amino acids with FIP-gsi are shown in a gray background. The key residues in FIP-fve (W24, T28, D34, T90, I91 and W111) and the corresponding residues in FIP-Lrh (W25, D35, I92 and W112) present in the carbohydrate binding module (CBM) are shown as a solid line ( ). A Neighbour-joining phylogram (with real Branch length, distance corrected) was generated using the Omega Clustal program with FIPs from the representatives of genera Lignosus, Ganoderma, Flammulina, Volvariella and Trametes. p53 activation pathway 35 whereas FIP-gmi inhibited the metastatic ability of A549 cells 36 . Suppression of tumor cell growth by these FIPs is associated with an increase in poly functional T cells that secrete multiple effector cytokines, such as IL-2, IFN-γ and IL-17 [37][38][39] .
The activity of a protein may be influence by proper protein folding, and post-transcriptional carbohydrate modification. Native FIPs (except LZ-8) were reported to lack carbohydrate modifications 8 . Recombinant FIPs, for example the rFIP-gts produced in insect cells 40 , rLZ-8 and rLZ-9 in yeast cells 41 , FIP-tvc 15 and FIP-fve (with a His-tag) 26 in E. coli lacks glycosylation albeit significant biological activity, indicating that carbohydrate modification of FIP is not necessary for biological activity. In accordance, the 6xHisFIP-Lrh expressed in E. coli showed substantial haemagglutination and cytotoxic properties. The underlying mechanism for its anti-tumour effects may be associated with immune cells activation through binding of CBM on their cellular glycans which need further experimental verification.

Conclusion
A new FIP-Lrh cDNA was isolated from L. rhinocerotis, with closest protein sequence identity to G. lucidum (64.55%) and very similar predicted 3-D structure to FIP-fve. Like other FIPs, it is a sugar binding protein, and the more positively charged putative CBM pocket of FIP-Lrh predicted a stronger interaction with N-acetylgalactosamine and N-acetylglucosamine. A functional recombinant 6xHisFIP-Lrh was successfully      System (Promega) and recombinant plasmids from four positive transformants were sent for sequencing of both DNA strands at MyTAGC Bioscience Sdn. Bhd. One of the positive transformants, designated as FIP-pGEM-T_3, was used as a template for PCR cloning into the NheI and HindIII site of pET28a(+ ) expression vector. Ligation mixture was transformed into BL21 strain E. coli cells and selected on Luria Bertani (LB) agar supplemented with 50 μ g/mL kanamycin. Recombinant clones were then verified by PCR screening and DNA sequencing as previously described.
Expression and purification of 6xHisFIP-Lrh. For expression study, 1 mL overnight culture of BL21 cells containing FIP-Lrh in pET-28a(+ ) expression vector (FIP-Lrh_pET/BL21) was used to inoculate 20 mL of LB broth (50 μ g/ml Kanamycin) and allowed to grow at 37 °C with shaking at 150 rpm. At optical density (OD 600nm ) of 0.6, 1 mL of uninduced (t = 0 hr) culture sample was obtained and expression was induced by addition of 1 mM IPTG. Induced culture samples were obtained at hourly time points (t = 1 to 4 hr), the cells were pelleted by centrifugation at 12100 × g for 2 min and used for total lysate and solubility test as previously described 48  with Agilent 6550 Accurate-Mass Q-TOF LC/MS. Desalted protein sample was reconstituted with 7 μ l of solution A (0.1% formic acid in water). Sample injection volume was 2 μ l. After sample pre-concentration, analytical separation was accomplished over a 75 μ m × 150 mm analytical column at 0.4 μ l/min in an initial gradient of 5% solvent B, 50% solvent B at 11 min, 70% solvent B at 15 min, 70% solvent B at 18 min and 5% solvent B at 19 min. Solvent A consists of 0.1% formic acid in water while solvent B contains 100% acetonitrile in solvent A. Total run time was 25 min including post-run of 6 min. For subsequent MS (rate: 8 spectra/s, time: 50 ms/spectrum) and MS/MS (rate: 4 spectra/s, time: 125 ms/spectrum) analyses, spectra were acquired in a MSMS mode with scan range from 200 to 3000 m/z and 50 to 3200 m/z, respectively. Capillary and fragmentor voltage were 1800 V and 175 V, respectively with drying gas flow rate of 5.0 l/min at 290 °C. Generated raw data were searched against L. rhinocerotis genome database using Agilent Spectrum Mill MS Proteomics Workbench software packages. The following parameters and filters were implemented for protein and peptide identification: MH+ scan range from 100 to 3200 Da, carbamidomethylation of cysteines was set as a fixed modification, protein score > 20, fast discovery rate < 1%.
Haemagglutination assay. In compliance with ethics standards, the methods used to obtain human and animal blood were performed in accordance with the guidelines and experimental protocols approved by the ethics committee for the Faculty of Applied Sciences, UCSI University. Informed consent was obtained from human blood donor and assay was performed according to Hsu et al. 50 with a few modification. Briefly, a total of 100 μ L whole blood was collected using a heparinized capillary tube from Balb/c mouse via retro-orbital route while human blood was collected via finger pricking. The blood samples were washed three times with equal volume of PBS and collected by centrifugation at 200 g for 10 min. A total 100 μ L of 2% human or mouse blood in PBS was respectively co-incubated with 6xHisFIP-Lrh (0 to 2.01 μ M) in a 96-well plate. The plate was shaken for 30 sec and incubated in a CO 2 incubator for 2 hr at 37 °C and degree of haemagglutination was observed under an inverted microscope.
Cytotoxicity assay on cancer cell lines. Human cervical cancer HeLa cell line (ATCC CCL-2) was cultured and maintained in Dulbecco's Minimum Essential Medium (DMEM, Gibco) supplemented with 5% Fetal Bovine Serum (FBS). Human breast cancer MCF-7 cell line (ATCC HTB-22) and human lung cancer A549 cell line (ATCC CCL-185) were cultured and maintained in Roswell Park Memorial Institute 1640 medium (RPMI-1640, Biowest) supplemented with 10% FBS. All media were supplemented with penicillin (100 U/mL) and streptomycin (100 μ g/mL) in cytotoxicity study. Cells were cultured at 37 °C in a 5% CO 2 incubator.
Cells at 70% confluent were harvested by trypsinization and used for cytotoxicity assay as previously described 51 . Briefly, cells seeded in triplicate wells at 1 × 10 4 cells/200 μ l/well were added with 50 μ l of 6xHisFIP-Lrh ranging from 0-3.36 μ M. The positive control was DMSO (0% to 20%) and negative control was untreated cells. Cells were incubated for 24 hr overnight after which 50 μ l of MTT solution (5 mg/mL) was added into each well and incubated further for 4 hr. The supernatant was then discarded, 200 μ l of DMSO was added and optical density (OD) at 570 nm was measured. IC 50 of 6xHisFIP-Lrh against respective cell lines was calculated from triplicate assays.