Purification and characterization of a cytochrome c with novel caspase-3 activation activity from the pathogenic fungus Rhizopus arrhizus

Background Members of Rhizopus species are the most common cause of mucormycosis, a rare but often fatal fungal infection. Host induced pathogen apoptosis and pathogen induced host cell apoptosis are often involved in fungal infections. In many organisms, the release of mitochondrial cytochrome c can trigger apoptosis by activating caspase proteases, but the role of fungal cytochrome c in apoptosis remains unknown. Results DNA sequence encoding Rhizopus arrhizus cytochrome c was cloned and expressed in E. coli. Both native and recombinant cytochrome c were purified using ion exchange followed by gel filtration chromatography. The identities of purified proteins were confirmed by MALDI-MS and UV-Visible spectroscopy. For the first time, we demonstrated that Rhizopus arrhizus cytochrome c could activate human capspase-3 in HeLa cell extracts. We also found that Rhizopus arrhizus cytochrome c has redox potential, peroxidase activity, and spectral properties similar to human and horse cytochrome c proteins. Conclusions Rhizopus arrhizus cytochrome c can activate human caspase-3 in HeLa cell extracts and it possesses similar physical and spectral properties as human and horse cytochrome c. This protein was found to have a previously unknown potential to activate human caspase-3, an important step in the apoptosis cascade. Electronic supplementary material The online version of this article (doi:10.1186/s12858-015-0050-9) contains supplementary material, which is available to authorized users.


Background
Mucormycosis are rare but often life-threatening infections seen in immunocompromised, diabetic, and organ transplant patients [1]. These infections are difficult to treat and with an increase in the number of diabetic patients and organ transplants, in the future such infections are likely to increase [1,2]. Rhizopus spp. accounts for the majority of the mucormycosis infections [1]. Results in recent decades suggest that apoptosis may play an important role in both establishment and clearance of fungal infections [3]. In response to fungal infections, the oxidative burst by host immune cells could help in infection clearance by triggering apoptosis in the fungus [4,5]. Also, pathogen-induced apoptosis in host immune cells could help the pathogen to establish the infection [6]. Thus, the success of an infection partly depends on the outcome of such host pathogen interactions. It has been shown that extracts from the filamentous fungus could induce apoptosis in human cells [6,7], but the identity of the protein/factor responsible for the induction of apoptosis remains unknown. In recent decades, it has also been established that in many organisms cyt c plays an important role in apoptosis [8,9]. During apoptosis, cyt c is released from the mitochondria into the cytoplasm and binds to apoptosis protease-activating factor-1 (Apaf-1) thus triggering a caspase activation cascade [8]. For example, in humans, cyt c can activate caspase-3 in cell-free activation assay [9]. The first evidence of the existence of apoptosis in fungi came from the study by Medeo et al. of a Saccharomyces cerevisiae mutant that showed signs of apoptosis [10]. Since then, many homologous mammalian apoptotic proteins were discovered in budding yeast [11]. In an important finding, using deletion mutants which were unable to produce functional cyt c, Silva et al. presented the first evidence that cyt c is involved in hyperosmotic stress induced yeast apoptosis [12]. However, the nature of the role cyt c plays in apoptosis of filamentous fungi like R. arrhizus remains unknown.
To characterize R. arrhizus cyt c, we cloned the gene for cyt c from R. arrhizus in E. coli. The purified cyt c was compared with two mammalian cyt c (horse and human). We selected R. arrhizus cyt c as a candidate for this study mainly for two reasons. First, this fungal cyt c is distantly related to mammalian cyt c and its biochemical properties are largely unknown.
Secondly, R. arrhizus is a pathogenic fungus and any information gained on the pro-apoptotic activity of cyt c may also contribute to the identification of better therapeutic targets. We found that cyt c of R. arrhizus has similar biochemical properties to mammalian cyt c.

Results and discussion
A 19-fold purification was achieved for recombinant R. arrhizus cyt c using a three step process involving ammonium sulphate precipitation, cation exchange and gel filtration chromatography (Fig. 1, Table 1). The purified recombinant cyt c protein eluted from gel filtration column with absorption ratio 410/280 nm of >4.0 (Fig. 1b, Table 1). Purity was also checked by SDS electrophoresis. The eluted cyt c produced one band, which had an apparent MW of ca. 14 kDa (Fig. 1c). Since this is the first report on purification of recombinant R. arrhizus cyt c, we compared our purification results with those reported by others using similar plasmids. We obtained~9 mg of cyt c with a 410/280 ratio of >4 from 1 L of E. coli culture. Our yield was similar to the reported yield for human cyt c of >8 mg L −1 [13]. Patel et al. reported a yield of~15 mg L −1 for horse cyt c [14]. However, in that work the authors only performed a two-step purification (ammonium sulphate precipitation and a cation exchange column) [14]. Our yield was only marginally lower if we compare our yield after the second purification step (Table 1).
In the case of the native R. arrhizus cyt c (protein isolated form commercial preparation of lipase) the final protein eluted from a Superdex 75 column as a single peak (Fig. 2a), but the 410/280 nm ratio was only 0.65. Efforts to further purify the native protein using an additional anion exchange column to trap impurities did not result in any improvement.
This concurs with a previous study that reported difficulties in purifying this protein from R. oryzae (syn. to R. arrhizus) to high purity [15,16]. The relative molecular weight determined for the native R. arrhizus cyt c by SDS-PAGE was 12.55 ± 1.27 kDa (Fig. 2b) in agreement with expectations for its molecular weight. Identity of both, the native and the recombinant protein bands was unequivocally confirmed by MALDI tandem mass spectroscopy ( Fig. 3 and Additional file 1: Figure S1).
The UV-vis spectra of the oxidized native R. arrhizus cyt c show peaks at 409 and 529 nm and in the reduced state, α and β peaks are prominent at 549 and 520 nm, respectively (Fig. 2c). Similar values were seen for the recombinant R. arrhizus cyt c. These values are close to other type-c cytochromes, thus suggesting a similar heme environment ( Table 2).
The presence of an absorption maximum at~700 nm in both recombinant and the native R. arrhizus ferri cytochrome (Additional file 2: Figure S2A  of the axial heme ligands. Additionally, sequence alignment shows that residues corresponding to Met-80 and His-18 are conserved in 14 fungal cyt c [17], as well as in human and horse cyt c. The redox potential of native and recombinant R. arrhizus cyt c was found to be similar to that of human and horse cyt c, respectively, with no statistically significant difference ( Fig. 4 and Additional file 3: Figure S3). The redox potential measurements for human and horse cyt c with 263.43 and 268.64 mV, respectively, were in close agreement with earlier reported values validating our technique [18,19]. The redox potential measurements of native and recombinant R. arrhizus cyt c were 266.90 and 270.04 mV and statistically the same. These results further support structural similarity between the recombinant and native R. arrhizus cyt c proteins.
A major portion of cyt c remains loosely associated with the inner mitochondrial membrane via ionic interactions with the negatively charged mitochondrion-specific phospholipid, cardiolipin. Kagan et al. had shown that cardiolipin specific peroxidase activity of cardiolipin bound cyt c may play an important role in triggering apoptosis by aiding in the release of pro-apoptotic proteins, including cyt c, from the matrix of the mitochondria [20]. Furthermore, peroxidase activity assay provides an indirect measure of heme accessibility and we therefore compared the peroxidase activities of mammalian and R. arrhizus cyt c.
The peroxidase activity of native R. arrhizus cyt c was found to be low compared to other known peroxidases, which is a general characteristic of type-c cytochromes. Recombinant cyt c showed a higher K cat and V max compared to native R. arrhizus cyt c. These variations are most likely due to the higher purity of the recombinant protein compared to native one ( Table 2, Table 3). Also, a high peroxidase activity could be related to partial denaturation and thus a more accessible heme in the non-recombinant cyt c. A partial denaturation can involve the breaking of the methionine coordination that leaves the heme more accessible for catalysis [21]. However, this is ruled out by the UV-spectra of the recombinant cyt c because they show the presence of the charge transfer band thus indicating that methionine coordination is intact (Additional file 2: Figure S2).
Native R. arrhizus cyt c showed similar activity as horse cyt c while it was higher than human cyt c (Table 3). On the other hand, judging from K M values the affinity of R. arrhizus cyt c seems to be lower than human cyt c, and very similar compared to horse cyt c ( Table 3). The K M value measured for horse cyt c was in close agreement with Kim et al. [22] but varied from the one reported by Randi et al. [23]. It is important to note that we used nonlinear regression to analyze the kinetics while the two cited studies used Lineweaver-Burk plots, a method considered less reliable [24].
We noted the presence of the cyt c in good amounts in a commercial preparation of a secreted lipase of R. arrhizus (Fig. 2). Interestingly, in an earlier study, the authors have indicated the presence of a "soluble factor"  in R. arrhizus extract that can activate caspase, but they did not identify this factor [7]. Based on these observations, we tried to verify whether this unknown watersoluble factor could be cyt c present in the supernatant of the R. arrhizus cultures or not. Efforts to locate the cyt c in supernatant fractions with heme staining (data not shown) and using an antibody against cyt c were not successful (Additional file 4: Figure S4). These results suggest that under our test conditions, the protein is not secreted into the supernatant. However, the possibilities that the protein might be secreted in response to some special cue, or in trace amounts, could not be ruled out.  In the cell free caspase-3 activation assay, compared to all cyt c tested, the recombinant R. arrhizus cyt c showed the lowest activity. Its activity was similar to the negative control, a cell lysate with asparaginase-II (Fig. 5). The native R. arrhizus cyt c showed a statistically significant (P < 0.05) higher caspase-3 signal but under the same conditions no statistically significant difference was observed in the signals of recombinant R. arrhizus cyt c and the negative asparaginase-II control (Fig. 5).
In our in vitro caspase-3 assay, we did not observe any caspase-3 activity in the aqueous extracts prepared from the R. arrhizus culture (Additional file 5: Figure S5). This result indicate that the source of previously reported activity in the aqueous extract is unlikely to be cyt c [7].
Compared to recombinant cyt c, higher caspase-3 activation by native R. arrhizus cyt c could be due to the presence of other unidentified factor/s. Identity of these factors is a matter of further investigation and beyond the scope of the present study.
Additionally, the activation differences between native R. arrhizus and the horse cyt c could be due to the variations in distribution of positive patches on the R. arrhizus cyt c surface, which are known to be critical for cyt c/Apaf-1 interaction. Cyt c and Apaf-1 interact through an extensive region. Many lysine residues (7,8,25, 39 and 72) spread across the surface of cyt c are known to be important for this interaction. Disruption of these residues by site directed mutagenesis has been shown to abolish or lower the ability of horse and human cytochrome to activate caspases [25,26]. In the case of native R. arrhizus, the reduction in caspase-3 activation was seen when compared to horse cyt c. Sequence comparison with horse cyt c revealed that in the native R. arrhizus cyt c two of these lysine residues (7 and 25) are occupied by alanine (Fig. 6). We propose that these changes to hydrophobic non-polar groups (Ala) could decrease cyt c affinity for Apaf-1. The lower caspase-3 activation by the native R. arrhizus cyt c is a potential reflection of this effect. Indeed, these two lysine residues (7 and 25) were demonstrated to be essential for Apaf-1 binding, since a 10-100 fold drop in caspase activity was observed by Yu et al. when these amino acids were mutated in combination with other Apaf-1 interacting residues in horse cyt c [25].
In Saccharomyces cerevisiae cyt c, a tri-methylation at lysine 72 was attributed for being responsible for its lower caspase activation potential [26]. We reasoned that it is possible that a similar tri-methylation on the K72 in   R. arrhizus cyt c could also partly account for its lower caspase activity as seen in our assays (Fig. 5). However, our MALDI data does not support the presence of Lys-72 methylation (Additional file 6: Figure S6). We also found four additional lysine substitutions in R. arrhizus cyt c (K5A, K22E, K88A & K100E) that have not been studied before, and could potentially contribute to its reduced caspase-3 activation (Fig. 6). It will be interesting to see if an R. arrhizus mutant with A7K and A25K mutations shows increased caspase-3 activation to the level similar to that of horse cyt c, since this would help to elucidate the role of other lysine substitutions seen in R. arrhizus cyt c.

Conclusions
We have reported here a comparative characterization of R. arrhizus cyt c. To the best of our knowledge, this is the first report on recombinant purification and biochemical characterization of the R. arrhizus cyt c.
The mitochondrial proteins are attractive targets for new antifungal drugs [27]. A better understanding of the role of specific sequences in cyt c's ability to induce apoptosis and its differences with mammalian cyt c could lead to the identification of methods for exclusive targeting of this fungal pathogen. It will be interesting to see if the amino acid differences in cyt c of R. arrhizus could provide it with sufficient specificity to induce apoptosis exclusively in fungal cells. If this turns out to be the case, designing of fungi specific apoptosis inducing drugs may be possible. Although results of targeting human cancer cells by actively inducing apoptosis using mammalian cyt c delivery have been promising [28,29], replicating them in the fungal cells could be very challenging due to the presence of their cell wall. Our results show the ability of R. arrhizus cyt c to activate caspase-3. This finding indicates that this pathogen could potentially use a similar mechanism in vivo to establish infections. To explore the possibility of existence of such mechanism in vivo, a study of clinical isolates and samples from infected patients would be helpful. At present such studies are impeded by the lack of our knowledge about the role of R. arrhizus cyt c in apoptosis. This study will help catalyze exploration into the possible involvement of cyt c in establishing mucormycosis infections.

Gene cloning and protein purification
Recombinant R. arrhizus cyt c was expressed from a construct made by modifying the pBTR1 plasmid [13]. pBTR1 was a gift from Gary Pielak (Addgene plasmid # 22468). It The numbering position of residues is with respect to horse cyt c. b Protein models of R. arrhizus and horse cyt c showing positions of 3 important lysine residues involved in Apaf-1 binding. From left to right: R. arrhizus model (generated using Swiss-model homology modeling server) and horse cyt c (PDB ID-1HRC). In R. arrhizus, two residues important for Apaf-1 binding, lysine 7 and 25 (red) are substituted by alanine. Another important lysine residue (blue) is conserved in both R. arrhizus and horse cyt c. Both images were generated using PyMOL contained human cyt c gene and a heme lyase gene from Saccharomyces cerevisiae. The lyase gene is essential for heme incorporation into cyt c protein. In the modified plasmid (pBRA), human cyt c was replaced with the gene encoding R. arrhizus cyt c. The sequence coding for R. arrhizus cyt c was commercially synthesized (GenScript) and ligated into pBTR1 vector (without the human cyt c gene) using Gibson Assembly (NEB). The primers utilized and steps performed to obtain pBRA are provided in the Additional file (primer and vector details are provided in Additional file 7: Figure S7). pBRA was transformed into BL21(DE3)T1 R cells (Sigma). The bacterial cultures were grown in TB media (12 gmL −1 tryptone, 24 gmL −1 yeast extract, 8 mL glycerol, 2.3 gL −1 KH 2 PO 4 and 12.5 gL −1 K 2 HPO 4 ) containing ampicillin (100 mgL −1 ) at 37°C under constant shaking at 220 rpm. A 5 mL overnight culture grown from a single colony was used to inoculate a 500 mL culture. The culture was harvested after 18 h of growth at 37°C by centrifuging at 7000 g at 4°C. The pellet obtained was incubated with lysozyme 3 gL −1 and 3 mg of DNase 1 in lysis buffer (50 mM Tris-Cl pH 6.8 and 1 mM EDTA) for 8 h at 4°C with stirring. The lysis mixture was sonicated on ice for 5 min. Lysed cells were centrifuged at 7000 g on a Sorvall RC 6 plus (Thermo Scientific) at 4°C. Ammonium sulfate was added slowly to the supernatant at 4°C under constant stirring to a final concentration of 350 gL −1 . The resulting precipitate was centrifuged at 7000 g for 30 min at 4°C. The supernatant containing cyt c was dialyzed using a 3.5 kDa dialysis membrane overnight against 20 mM sodium phosphate buffer at pH 6.8. Dialyzed protein was loaded on a HiTrap SP Sepharose column (GE Healthcare) equilibrated with buffer A (40 mM sodium phosphate buffer at pH 6.8). Bound protein was eluted with elution buffer B (40 mM sodium phosphate at pH 6.8 and 1 M NaCl) using a linear gradient from 0 % B to 100 % B over a 7 column volume. Peak fractions with an absorbance ratio of 410/280 nm > 2.5 were pooled and loaded on a Superdex 200 gel filtration column (GE Healthcare). Protein elution from the Superdex 200 column was followed by absorption at 410 and 280 nm. Peak fractions were analyzed by loading onto a 15 % SDS-PAGE. Native R. arrhizus cyt c was purified using lipase of R. arrhizus obtained from Sigma using a similar scheme as that of the recombinant protein.
Human and horse cyt c were used as controls in all the assays performed unless otherwise stated. Human cyt c was expressed in E. coli using plasmid pBRT1 and purification was done as described by Olteanu et al. [13]. Horse heart cyt c was purchased from Sigma and used without further purification.

UV-vis spectroscopy
All spectra were measured using a UV-Vis NanoDrop 2000/2000c (Thermo Scientific) spectrophotometer except for the charge transfer band (CTB). The CTB region was measured on a UV-2450 Shimadzu spectrophotometer, with a slit width of 0.5 nm. The peak location was determined by using the picking algorithm of the UV-Probe 2.33 software (Shimadzu).

Trypsin digestion and mass spectrometry
In-gel trypsin digestion was performed using 500-1000 ng of both native and recombinant R. arrhizus cyt c as described [30]. In brief, the gel-eluted peptides resulting from trypsin digestion were first vacuum dried and later purified on a C18 reverse phase tip column (Millipore) using the instructions supplied by the manufacturer. Trypsin-digested peptides were eluted in a total volume of 5 μl of acetonitrile and 0.1 % trifluoro acetic acid solution (50:50 v/v). The purified peptides were directly spotted onto the MALDI plate following the dried droplet method [31]. The matrix solution used contained 5 mg/mL αcyano-4-hydrooxycinnamic acid (HCCA, Sigma) in acetonitrile and 0.1 % trifluoro acetic acid (50:50 v/v). Tandem mass spectrometric analysis was performed using an ABSCIEX 4800 Plus MALDI TOF/TOF™ Analyzer in Top 6 mode. The spectra were collected in positive ion reflector mode with 500-1000 laser shots per spectrum. Protein identification was performed using the Mascot server (Matrix Science, http://www.matrixscience.com/ search_form_select.html).

Redox potential
The redox potential of R. arrhizus cyt c was determined as described [18] with some minor modifications. Briefly, 100 mM stock solutions of potassium ferrocyanide (fer-roCN), potassium ferricyanide (ferriCN), and (+)-sodium L-ascorbate were prepared in 100 mM potassium phosphate buffer at pH 7.0. Prior to use, all solutions were vacuum degasified and purged with nitrogen for 2-3 min.

Peroxidase assay
Cyt c peroxidase activity was measured as described [22]. Briefly, each reaction was performed in a total reaction volume of 150 μL with 50 μM 2,2′-azino-bis(3-ethylbenzothiazoline-6-sulphonic acid) (ABTS), 2.5 μM for human and horse cyt c, and 1.5 μM for native R. arrhizus cyt c. The concentration of cyt c was determined using the absorption coefficient of 29.5 mM −1 cm −1 at 550 nm [32]. Each reaction was followed for 40 s by recording the absorption of oxidized ABTS at 405 nm. The concentration of oxidized ABTS was calculated using the extinction coefficient of 36.8 mM −1 cm −1 at 405 nm. For each reaction with varying H 2 O 2 concentrations (ranging from 1.0 to 28 mM), initial rates were measured using the linear part of the absorption curve. All absorption readings were made using a Shimadzu UV-2450 spectrophotometer using a cell of 1 cm path length at 25°C. The Michaelis-Menten constants K M , K cat and V max were determined by nonlinear regression curve fitting using Prism 5 (Graph-Pad Software).

Cell free caspase-3 activation assay
HeLa cells were grown to 90 % confluency, harvested, and disrupted as described [29]. Prior to disruption, approximately a total of 2 × 10 7 cells were suspended in 2 mL of lysis buffer consisting of 20 mM HEPES at pH 7.5, 10 mM KCl, 1.5 mM MgCl 2 , 1 mM Na-EDTA, 1 mM Na-EGTA, 1 mM DTT, 250 mM sucrose, and 0.1 % v/v protease inhibitor cocktail (2 mM AEBSF, 0.3 μM aprotinin, 130 μM bestatin, 14 mM E-64, 1 mM leupeptin, 1 mM EDTA). The HeLa cell extract (lysate) was stored at −80°C for at least 5 days before use. Total protein content in the lysate was determined by the Bradford method. Each cyt c sample was incubated with the lysate in a total volume of 50 μL at 37°C for 1 h using a Mastercycler (Eppendorf ). This incubation mixture consisted of 1 mM dATP, 4 mg/mL total protein from lysate, and 10 μM cyt c; cyt c concentration was determined using the absorption coefficient 29.5 mM −1 cm −1 at 550 nm [32]. Immediately thereafter, caspase-3 activation was performed as per the manufacturer's protocol (CaspACE™ Colorimetric Assay System, G7220). Briefly, 20 μL from the incubation mixture was added to 78 μL of a mixture containing 128.2 mM HEPES at pH 7.5, 12.82 % w/v sucrose, 0.1282 % w/v CHAPS, 2.56 % v/v DMSO, and 12.8 mM DTT. Afterwards, 2 μL of 10 mM caspase-3 substrate (Ac-DEVD-pNA) was added. The plate was incubated overnight, and the absorbance measured at 405 nm using a microplate reader (BioTek-Synergy H1 Hybrid Reader). All measurements were performed in triplicate. The activation by horse cyt c was considered 100 %, and other results were normalized relative to horse cyt c. In these assays, a negative control that contained HeLa cell lysate with asparaginase-II, an unrelated non-apoptotic protein, was used. Asparaginase-II was purified from E. coli as described earlier [33]. Additionally, an aqueous extract of R. arrhizus was also made and tested for caspase-3 activation (Additional file 5: Figure S6).