Here we report the expression of human alkaline phosphatase by the ciliate T. thermophila. A codon-adapted artificial gene encoding the full-length precursor enzyme is correctly translated into the functional hiAP enzyme. The heterologous expressed enzyme exhibited phosphatase activity and became N-glycosylated during its passage through the ER/Golgi compartments, suggesting that both, the human signal peptide as well as the human GPI anchor signal, were correctly recognized by the T. thermophila protein biosynthesis machinery. Furthermore, the expression cassette encoding the human full-length protein, including the C-terminal GPI anchor showed a membrane associated localization on the surface of the ciliate cells, whereas the C-terminal truncated protein, lacking this GPI anchor recognition signal was secreted into the medium.
The distinct localization pattern of the stained cells (see Figure 3C and Additional file 1 and 2) raises the interesting question which signals determine protein targeting and localization within the ciliates' membranes. Previous results from the Tiedtke group (University of Münster, Germany) showed that hydrolases of the phagolysosomal compartments (e.g. glycosidases, proteinases and lipases) are secreted into the surrounding medium [26, 29, 30]. Tetrahymena possesses three secretory pathways, namely regulated exocytosis of dense-core granules, constitutive secretion of lysosomal enzymes and phagocytosis . At least, the first two of these routings can be used for recombinant protein secretion [10, 14]. However, the exact sites of secretion remain to be determined and therefore it is not really known whether or not different secretory and/or surface proteins use different exocytotic pathways.
In previous experiments we could show that fusion proteins of the T. thermophila PLA1 signal peptide with the human DNase I became secreted and correctly processed . Interestingly, the results we show here clearly indicate that also human signal peptides are sufficient for proper secretion. As recently demonstrated by the Turkewitz group (Department of Molecular Genetics and Cell Biology, University of Chicago) T. thermophila provides many excellent features as a model system to analyze regulated secretion in more detail [32, 33]. The truncated hiAP used in this study provides an additional instrument to address such questions, because truncated hiAP fused to signal peptides or to parts of proteins that are secreted continuously (e. g. lysosomal hydrolases) or in a stimulus dependent manner (e. g. proteins of the GRL family) might allow deeper insights into the secretory pathways and their regulation. Illustrating the underlying factors that regulate expression and secretion is of paricular importance for high yield recombinant protein expression, because the efficiency of such processes correlates to the yield of biologically active protein. However, scientific studies that focus on basic research rather than on biotechnological aspects will be necessary to elucidate these questions.
We performed fermentation experiments and used the truncated "hiAP without GPI" system to monitor several modifications of the fermentation process. We found that extracellular hiAP activity increased only in the first 24 to 48 h after induction of the MTT1 promoter. Of course on the one hand this observation emphasizes that truncated hiAP provides a powerful tool to quantify the secretion efficiency in living ciliate cells by a very sensitive assay. But on the other hand this result raises the question why the hiAP activity reached a stable level although the T. thermophila cells grew further to titers of 2.8 × 106 cells/ml. One reasonable explanation might be that recombinant hiAP is probably increasingly degraded by the ciliates' proteases of the cathepsin L family. This is in agreement with studies that demonstrated that several endogenous hydrolases including proteases accumulate during the late stages in continuous fermentation processes [26, 27]. However, the hiAP activity remained stable upon incubation in cell free medium from late log cultures for at least six hours (data not shown). An additional explanation is probably the induction characteristic of the applied MTT1 promoter. The promoter of the metallothionein gene MTT1 of T. thermophila that was first introduced in 2002 can be rapidly activated by simply adding cadmium ions to the growth medium [23, 34–37]. Shang et al. stated that MTT1 mRNA reaches a maximum level at about 45 min after cadmium addition to the culture medium . However, in the same study it was shown that best expression levels of a recombinant antigen which expression was driven by the MTT1 promoter were detectable after 9 h of cadmium induction. This observation fits to the finding that highest hiAP expression level lagged behind the published data regarding endogenous MTT1 mRNA level and that protein yield decreases after a certain time of induction. Multiple addition of cadmium did not lead to a "pseudo-constitutive" promoter activity in the cultures, suggesting that the inducible effect disappears the longer the fermentation lasts. Furthermore, it has been shown in T. pyriformis that a substantial portion of the added cadmium could accumulate inside the cells by chelating metallothioneines . A combination of both effects, protease degradation as well as the adaptation to the cadmium stress is also possible.
Interestingly, the recombinant expression of human alkaline phosphatases shows a further fascinating aspect. Poelstra et al. (1997) and Bentala et al. (2002) could show that human alkaline phosphatases are able to reduce mortality in mice infected with Gram-negative bacteria [38, 39]. Additionally, Heemskerk et al. (2009) were able to demonstrate in a phase IIa study that alkaline phosphatase treatment improved renal function in severe sepsis or septic shock patients . These studies show that alkaline phosphatases, usually catalyzing the hydrolysis of monoesters with release of inorganic phosphate, are also capable of reducing toxicity of endotoxins/lipopolysaccharides (LPS), components of the outer membrane of Gram-negative bacteria.
So far only few tools are available to fight against the clinic features of sepsis and the currently available therapies focused on antagonizing the most pro-inflammatory cytokines or by inactivation of the LPS cascade for example by neutralizing anti LPS antibodies .
Unfortunately, the studies also revealed that high dosage of alkaline phosphatases are necessary to attenuate LPS toxicity and reduce mortality. As human sources for alkaline phosphatases are limited, different approaches are tried to obtain sufficient amounts of this enzyme to proceed with the clinical tests. Ciliates that secrete large amounts of recombinant human phosphatases might provide a solution for this problem. A general comparison of Tetrahymena with common expression systems is given in Additional file 3.
Nam, Ermonval and Sharfstein (2007) demonstrated production levels of 200 U/liter hiAP with CHO cells in a fed-batch bioreactor with a fermentation time of 5 to 6 days . Later, Nam et al. (2008) produced 700 - 2,300 U/liter with CHO cells in a fed-batch bioreactor within 5 days . Furthermore, Chen, Chang and Chang (2004) described a 5 liter fermentation of the yeast Pichia pastoris with a duration time of at least 5 days and measured 4,000 Units/liter hiAP in the supernatant .
We could demonstrate here that up to 14,000 U/liter hiAP could be produced within 2 days fermentation time in a bench top bioreactor. Moreover, it has to be considered that typical steps for a process optimization, like the improvement of the yield by strain optimization and the development of an elaborated fermentation process, has not been undertaken so far. Thus, it should be possible to increase the production of hiAP significantly. Compared to the production methods which have been described in the past, the here used Tetrahymena based expression system seems to be an attractive alternative for the production of hiAP.