3. Cell-Free Protein Synthesis of AaeUPO
For our study on the cell-free production of a UPO, we used the
upo1 gene from
Cyclocybe (Agrocybe) aegerita, which encodes the first reported UPO
[17]. The extracellular heme-thiolate enzyme contains an N-terminal signal peptide followed by a propeptide and has a C-terminal intramolecular disulfide bond as well as six potential
N-glycosylation sites
[36]. For easier detection, the UPO gene was tagged with a C-terminal polyhistidine peptide. To reduce the complexity of the CFPS system, the mRNA of
upo1 was synthesized independently from the translation system and provided externally. The CFPS reactions were tested for the formation of functional cell-free UPO with an LC-MS-based propranolol assay since the formation of 5-hydroxypropranolol (5-OHP) from propranolol is characteristic of
AaeUPO
[12][37]. By using this assay, we could exclude the possibility of measuring false positive activities from potentially interfering endogenous monooxygenases (cytochrome P450 enzymes). Nevertheless, negative controls that contained no mRNA templates were additionally included in all experimental setups.
Cell-free production of
AaeUPO was evaluated with lysates from
N. crassa and
A. niger, respectively. Essentially, the reactions were set up according to Wu et al. with minor modifications
[33]. The mRNA carrying the coding sequence for the UPO was applied at 28 nM since higher concentrations had adverse effects on the synthesis of active enzyme (
Figure A1). The formation of active cf
AaeUPO was observed over a period of 24 h (
Figure 2). With both lysates, significant UPO activity could be detected: 10,498 ± 28 mU L
−1 using the
N. crassa lysate and 783 ± 18 mU L
−1 with the
A. niger lysate. The different yields of active enzyme correlated with the results for the reporter luciferase (
Figure 1b). Compared to the lysate from
N. crassa, the yield of active enzyme with the
A. niger lysate was 10-times (after 5 h incubation) to 15-times (after 24 h incubation) lower. Interestingly, in contrast to
N. crassa, the UPO activity with the
A. niger lysate was higher after 5 h of incubation than after 24 h. This decrease in activity could indicate residual proteolytic activity in the lysate, despite the application of protease inhibitors during the preparation process.
Figure 2. Time-dependent activity of cell-free synthesized AaeUPO with lysates from N. crassa and A. niger. The 20-µL reactions were prepared in separate tubes, incubated at 18 °C for the indicated time, and tested for UPO activity by measuring the formation of 5-OHP from propranolol by LC-MS analysis. Values are means ± standard deviation.
Western blot analyses were done to get information about the localization of the synthesized protein and its relative amount (
Figure 3). As a secretory protein, cf
AaeUPO was expected to be present in the microsomes, i.e., in ER-derived vesicles. In the case of the
A. niger lysate-based CFPS, a band was detected in the microsomal fraction that corresponded to
AaeUPO, which has a theoretical molecular weight of 41.5 kDa in its unprocessed form. Overall, the amount of detectable protein was rather low, which explains the relatively low activity. With the
N. crassa lysate, two prominent bands could be seen in the whole CFPS reaction mixture. As demonstrated by deglycosylation (
Figure 3d), the lower band corresponds to the non-glycosylated (marked with an asterisk) and the upper band to the glycosylated (marked with an arrowhead) form of cf
AaeUPO. The larger part of the glycosylated protein was located in the microsomes, while the non-glycosylated UPO was almost exclusively found in the supernatant. Activity measurements of the fractions indicated that the UPO activity stems from the glycosylated form as the ratio of the activities between the fractions correlated with the distribution of synthesized protein (
Figure A2). The residual glycosylated cf
AaeUPO seen in the supernatant fraction may be the result of an incomplete fractionation process, similarly described in an article by Thoring et al.
[38]. They showed that—after centrifugation with 16,000×
g—smaller microsomes were still present in the supernatant. The rather high portion of non-glycosylated protein might indicate an insufficient translocation capacity. Nonetheless, the successful expression of active
AaeUPO proves that our fungal lysates contain functional microsomes that allow appropriate post-translational modification of secretory proteins.
Figure 3. Localization of cell-free AaeUPO synthesized with lysates from A. niger and N. crassa, respectively. CFPS was done in 40-µL reactions, incubated at 18 °C for 5 h and fractionated; 10 µL of the whole CFPS reactions, supernatant, and microsomal fractions, respectively, were subjected to SDS-PAGE and subsequent Western blot analysis using a polyhistidine tag antibody. (a) Schematic representation of the cell-free synthesis of AaeUPO. (b) CFPS samples from A. niger. (c) CFPS samples from N. crassa. (d) Deglycosylation of cfAaeUPO produced with lysate from N. crassa. (−)—without deglycosylation enzyme mix; (+) with deglycosylation enzyme mix; M—protein marker, total—whole CFPS reaction; S—supernatant; MF—microsomal fraction; NTC—no-template control; arrowhead—glycosylated cfAaeUPO; asterisk—non-glycosylated cfAaeUPO.
Unfortunately, it was not possible to use the polyhistidine tag on the cf
AaeUPO for purification from the CFPS reaction with sufficient purity, due to the presence of a larger amount of contaminating fungal proteins. Thus, an accurate protein concentration of the synthesized cf
AaeUPO could not be determined. In future experiments, the use of more appropriate affinity tags, such as the Strep II tag, should be considered
[39].
The good results for the long incubation times with the lysate from
N. crassa (
Figure 2) were unexpected since, in batch CFPS reactions, the synthesis process usually stops after a few hours. For example, Hodgman and Jewett reported a “long synthesis time of 120 min” for luciferase using a yeast CFPS-batch reaction system, albeit at a slightly higher incubation temperature of 21 °C
[40]. The main reasons for the termination of synthesis were seen in a rapid decline in energy-providing components and an accumulation of inhibitory molecules, such as inorganic phosphate
[41]. To clarify whether the apparently long synthesis time in our experiments was actually due to better protein synthesis performance of the lysate, we expressed the reporter protein luciferase using CFPS with the lysate of
N. crassa and measured the activity over a 24-hour period (
Figure 4a). The activity of luciferase increased over the first three hours of CFPS but decreased thereafter, suggesting that the process of protein synthesis was not maintained throughout the incubation period. To confirm that this was also true for CFPS of
AaeUPO, samples from the time-course experiment (
Figure 2) were subjected to Western blot analysis (
Figure 4b). As in the previous Western blot, two prominent bands appeared in the gel. Their intensity and thus the amount of cf
AaeUPO produced increased until the third hour of incubation and then remained nearly constant (or decreased slightly) for the next hours. The 24-hour samples appeared to have lower cf
AaeUPO protein levels than the samples from the previous time points, although activity was highest after 24 h. It might be that nonfunctional, e.g., misfolded UPO protein was degraded by remaining proteases in the lysate, but this needs to be clarified in future experiments.
Figure 4. Time-dependent cell-free protein synthesis of luciferase and AaeUPO with N. crassa lysate. (a) Detection of luciferase activity in CFPS-batch reactions at different time points. The 20-µL reactions were prepared in separate tubes, incubated at 18 °C for the indicated time, and tested for luciferase activity by measuring the oxidation of luciferin via luminescence detection. Values are means ± standard deviation. (b) Western blot analysis of CFPS samples from different incubation times. The 20-µL reactions were prepared in separate tubes, incubated at 18 °C for the indicated time, and subjected to SDS-PAGE and, subsequently, Western blot analyses using a polyhistidine tag antibody. The semi-quantitative analyses were done individually for the glycosylated and non-glycosylated proteins. The densitometric values for t0 were subtracted from the values of other samples and the highest values were set to 100%. M—protein marker; arrowhead—glycosylated cfAaeUPO; asterisk—non-glycosylated cfAaeUPO.
Since the capacity to produce protein was limited to 3 h in our setup, an intrinsic maturation period of the UPO protein could be an explanation for the apparently long synthesis time. Such a maturation time was observed for a bacterial P450 monooxygenase that was expressed in a cell-free approach using lysate from
E. coli [25]. In this case, the highest enzyme activity was reached 150 min after protein synthesis had stopped. However, since the exact process of heme incorporation and protein folding is still unknown for UPOs, it must be clarified in the future whether a longer maturation time is indeed necessary. Here, the CFPS might prove to be a valuable tool for gaining insight into the genesis of UPOs. It should be relatively simple to change the conditions in CFPS, for example, by adding mRNA for the co-expression of chaperones. The effects on UPO formation would be directly measurable and could be evaluated in a short time.
Table 1 gives a brief comparison of the volume activities and production times of
AaeUPOs obtained from homologous, heterologous, or cell-free approaches. Attempts to express wt
AaeUPO heterologously in the standard expression hosts
S. cerevisiae or
P. pastoris with appreciable yields failed
[42][43]. Rounds of directed evolution were necessary to increase the production in the two yeasts that do not contain endogenous UPO genes
[20][43]. This highlights an advantage of the cell-free approach: the possibility to use translationally active lysates from naturally UPO-coding (and producing) fungi. Thereby, all essential components should be available for the in vitro UPO synthesis. However, it has to be mentioned that the evolved mutant variant r
AaeUPO PaDa-I expressed in
P. pastoris in a bioreactor in a six-day cultivation period reached much higher expression levels than the homologous or cell-free systems. Nevertheless, cell-free protein production is by far the fastest method to gain significant amounts of functional UPO.
Table 1. Comparison of AaeUPO produced by different approaches.
If the heterologously produced PaDa-I variant of r
AaeUPO is included in the comparison of the three expression systems, the yield obtained with CFPS appears to be the lowest. However, in the present study, we have not yet fully exploited the potential of the cell-free approach using fungal lysates. There are general aspects in CFPS that have a high impact on the performance and would need to be optimized, for example, biomass production, lysate preparation, and buffer composition
[44][45]. Also, the template design is essential for an optimal translation initiation, which can be triggered by adding elements like IRES
[46]. As indicated by the Western blot analyses (
Figure 3), the production of active UPO in our setup may be limited by a non-optimal translocation efficiency. Co-expressing components of the secretory apparatus could solve this problem, as this approach was already successful in living cells, for example, for the production of secreted antibodies in CHO cells
[47]. Additionally, there is the possibility to change the CFPS reaction from a batch to a continuous process, which would result in an increase in protein yield
[48]. Other influential parameters or components are specific for the expression of UPOs. The availability and fit of the heme cofactor are crucial for the functioning of UPOs, therefore the formulation and amount of heme in CFPS reactions will have to be optimized. This could be achieved on different levels: (i) during growth of the fungi, e.g., with the addition of hemin, which worked well in the expression of horseradish peroxidase in
P. pastoris [49]; and (ii) by adding heme directly to the CFPS reaction, which was successful in the cell-free production of manganese peroxidase as well as horseradish peroxidase
[23][24][27]. Since
AaeUPO and other UPOs have at least one internal disulfide bond and other UPOs form functional dimers via disulfide bridges, it might also be beneficial to add protein disulfide isomerases like PDI or DsbC to CFPS reactions
[23][27].
3.3. Substrate Screening with Cell-Free AaeUPO
The cell-free protein production approach should offer the possibility to establish a rapid screening platform for interesting biocatalysts. Thus, we tested the cell-free UPO produced with N. crassa lysate directly in a substrate screening without the need for time-consuming purification steps. To this end, the conversion (oxygenation, oxidation) of typical UPO substrates, which are also used in activity assays, as well as of three common pharmaceuticals was performed with simply prepared CFPS reaction solutions. The enzyme that is partly located in microsomes was efficiently released by ultrasonic treatment (Figure A3). After centrifugation, the supernatant was used for photometric UPO assays with ABTS, veratryl alcohol, 5-nitrobenzodioxole, and naphthalene, and for LC-MS analyses with the pharmaceuticals propranolol, diclofenac, and clopidogrel.
The β-adrenergic blocker propranolol, the non-steroidal anti-inflammatory drug diclofenac, and the antithrombotic prodrug clopidogrel are known to be metabolized by cytochrome P450 monooxygenases in the human body. In previous studies, it was shown that wt
AaeUPO is able to perform the same reactions to give the (major) human drug metabolites 5-hydroxypropranolol, 4′-hydroxydiclofenac, and 2-oxoclopidogrel, respectively (
Figure 5)
[15][37][50].
Figure 5. Conversion of the pharmaceuticals propranolol, diclofenac, and clopidogrel by cell-free prepared UPO (cfAaeUPO).
The activities of cfAaeUPO were compared with those of its wild-type counterpart (taking into account that wtAaeUPO did not contain a polyhistidine tag). For the purpose of comparison, we used enzyme samples with a normalized volume activity of 100 U L−1 (based on veratryl alcohol oxidation) in all reactions. The results are listed in Table 2.
Table 2. Substrate conversion by homologous wtAaeUPO and cfAaeUPO. All reactions contained a UPO activity of 100 mU mL−1 (based on the oxidation of veratryl alcohol). The enzyme activities are given in mU mL−1 (a) or µU mL−1 (b).
All substrates were converted by both enzyme preparations, although quantitative differences in the activities were found in some cases. For NBD, naphthalene, and propranolol, the measured activities were similar; however, for ABTS, diclofenac, and clopidogrel, the activities of the cell-free variant were two to four times lower. ABTS is a substrate that is oxidized via one-electron abstraction by the peroxidative activity of UPO resulting in a cation radical product (‘classic’ peroxidase activity without oxygen transfer). The fungal lysate used for CFPS may have contained metabolites with antioxidant properties that could have interfered with the detection of the ABTS radical
[51]. Thus, substrates like ABTS might only be suitable in screening approaches with unpurified UPO, if the yield of cell-free enzyme is relatively high. A possible explanation for the lower conversion of diclofenac and clopidogrel could be the presence of the C-terminal polyhistidine tag, which may have impaired activity, e.g., by shielding the heme-access channel. Such an effect was already reported for a mutant variant of
AaeUPO (PaDa-I)
[52]. In addition, negative effects of potentially different glycosylation patterns cannot be excluded.
In summary, it took less than two days from the DNA template carrying the
AaeUPO gene to obtain the substrate screening results with expressed cf
AaeUPO. This is much faster compared to conventional screening systems based on heterologous expression in host microbes
[53]. Moreover, if transcription and translation were coupled in one CFPS reaction, the turnaround time of the entire process could be reduced even further.
4. Conclusions
In the present study, we evaluated a novel approach to produce unspecific peroxygenases (UPOs) by using a cell-free synthesis system based on fungal lysates. The UPO from Cyclocybe (Agrocybe) aegerita was successfully expressed with lysates derived from N. crassa or A. niger. Activities of this cfAaeUPO of up to 105 U L−1 (based on veratryl alcohol oxidation) could be achieved within 24 h, where the lysate from N. crassa showed a significantly higher translational performance compared to A. niger. The unpurified CFPS reaction solutions were subjected to a substrate screening, in which the performance of the cell-free enzyme (cfAaeUPO) was comparable to that of the homologously produced enzyme (wtAaeUPO). The procedure presented here highlights the general potential of CFPS for rapid screening of new UPOs or UPO variants, including the estimation of their catalytic performance. Nevertheless, for large-scale production of suitable candidates, hetero-logous production in fungal hosts will continue to be the method of choice, as cell-free synthesis is currently not competitive in terms of yield. Last, but not least, it must be pointed out that the full potential of lysates from filamentous fungi is far from being tapped, as their use in CFPS is still in its infancy.