Purification and characterization of recombinant CRLAAO

Recombinant CRLAAO was purified in a two-step process (Fig. 30). Upon ion exchange chromatography the active fractions that have LAAO activity and absorption at 460 nm were pooled and concentrated. Gel filtration of this material yielded pure recombinant CRLAAO (inset of Fig. 30 B). The purification is summarized in Table 6. LAAO activity was assessed using various L-amino acids.

Fig. 30. Purification of CRLAAO expressed in Pichia pastoris. A. Mono-Q chromatography:

Culture supernatants were concentrated ≈100-fold and then applied to the column (8mL) pre-equilibrated with 20 mM Tris-Cl pH 8.0. Elution was with a NaCl gradient (0-1 M). B. Gel filtration: Active and western positive fractions after Mono-Q chromatography were pooled, concentrated and applied to a Superdex G200 (23.5mL) gel filtration column (B). The insert shows the SDS PAGE of fractions 13 and 14.

Table 6. Purification of Pichia pastoris derived CRLAAO

Purification step Protein^b (mg/mL)

Volume (mL)

Total protein

(mg)

Total activity^c

(U)

Sp.activity (U/mg)

Supernatanta 0.08 900 72 9 0.01

Ultra filtration 1.5 9 13.5 2.43 0.18

Ion exchange

chromatography 0.9 1 0.9 2.4 2.6

Gel filtration 3.15 0.25 0.8 2.5 3.15

a900 mL of supernatant from the culture medium was used for the purification of secreted LAAO. ^bProtein content was estimated using the standard Bradford assay. ^cUnits were determined using the peroxidase coupled assay as described in materials and methods. Yield of CRLAAO was typically between 0.2 - 0.4 mg/L of fermentation broth.

When the purified CRLAAO was treated with PNGase-F; the increased electrophoretic mobility of the deglycosylated CRLAAO is similar to that of authentic CRLAAO (Fig. 31A). There is no change in the electrophoretic mobility upon treatment with sialidase (Fig. 31A) indicating the absence of sialic acids.

LAAO activity was measured for various amino acids and the obtained substrate specificity pattern is closely similar to that reported by Wellner and Meister [11] for Crotalus adamanteus LAAO (Fig. 31B).

Fig. 31. Characterization of expressed CRLAAO. A. Western blot analysis of the effect of CRLAAO treatment with PNGase-F and sialidase. B. Various amino acids were tested for LAAO activity in duplicate assays and the average values were normalized to the readings of L-Met. Inset: Dependence of LAAO activity from the L-Leu concentration Each point indicates the average of duplicate values obtained from a 96 well plate assay (200 µL). The recombinant CRLAAO showed the highest activity at 1-2 mM [L-Leu] and a typical substrate inhibition pattern similar to that of native CRLAAO.

3.5 Discussion

Our initial attempts at the heterologous expression of CRLAAO was based on the report by Torii et al. that LAAO from Crotalus atrox could be expressed in an active form in mammalian cells [27]. To verify this a mammalian expression system based on 293-T cells was developed that contained the native signal sequence and should promote secretion and expression of CRLAAO (data not shown). Amounts of recombinant protein detectable by Western blot were secreted into the medium.

However, upon 24 h of post transfection no significant activity was observed. It should be pointed out that Torii et al. used the same assay [27], however their

incubation time (for a single assay) was >60 min. In our opinion this type of assay is very prone to artefacts at longer incubation times and we thus conclude that the activity reported by these authors is questionable. Consequently serious doubts on the validity of this report [27] are justified.

Using various strains of E. coli the CRLAAO protein was expressed in good quantities. However, the protein was found in the inclusion bodies (data not shown).

Production of soluble, active LAAO has never been observed under these circumstances. Attempts to generate active CRLAAO from the inclusion bodies were not successful. This failure to obtain active CRLAAO was taken as suggesting that (active) LAAO is toxic for the host cell. Since toxicity is mediated, at least in part, by the production of H2O2, and since LAAO activity and degradation of L-amino acids might be greatly reduced in the absence of oxygen, we have attempted the expression of the CRLAAO gene under anaerobic conditions (N2 atmosphere). Also in this case no production of soluble CRLAAO was observed. This was taken as an indication that the post-translational modification consisting in glycosylation affects decisively the solubility of CRLAAO. Based on the work by Nishizawa et al, [44], we have developed an E.coli expression system that should co-over express chaperonins (GroEL/ES) and CRLAAO. This yields small quantities (0.1-0.2 mg/L) soluble CRLAAO when the IPTG induction is carried at low temperatures (16 to 20 °C).

However this soluble form did not show any LAAO activity.

It should be stated that our observations coincide very well with the experiences of Geueke and Hummel [43]. These authors have reported recently that attempts to express a Rhodococcus opacus L-amino acid oxidase in E. coli only yields inclusion bodies, while expression in a Streptomyces lividans strain yields soluble and active enzyme, albeit at rather low yields [43]. The failure to obtain soluble and active CRLAAO in a prokaryotic system is attributed to the lack of glycosylation. Based on these negative results we have resorted to a yeast expression system (Pichia pastoris). CRLAAO was cloned after the α-signal sequence that promotes secretion. Expression of the CRLAAO gene was repressed using glycerol as the sole carbon source, and until a large biomass had been collected after fermentation (≈ 20 g/L). Subsequent induction of expression by switch to methanol

as C-source leads to secretion of recombinant CRLAAO with yields ranging from 0.2 - 0.5 mg/L culture medium. Pure and active recombinant CRLAAO was obtained from the culture supernatant after ion exchange chromatography and gel filtration.

The purity of the protein is >95% based on SDS-PAGE analysis. This method yields

≥1 mg recombinant CRLAAO / 8L culture medium. It should be noted that the expression in Pichia pastoris of achacin, an antimicrobial glycoprotein from the giant African Snail Achacina fulica Férussac that has LAAO activity, has been reported [51]. In this case the expression yields were also rather low with 0.2 mg/L.

The higher molecular size of the achacin expressed in Pichia pastoris could be attributed to the extra two N-glycosylation sites when compared to CRLAAO (Asn 371 and Asn172). Although we have obtained similar yields from the culture supernatants in agreement with the hypothesis put forward above that glycosylation is important for the secretion, solubility and consequently also for the activity of recombinant CRLAAO. This is inferred from the fact that the presence of tunicamycin leads to either very low or no secretion of unglycosylated and inactive CRLAAO (Fig. 29).

Finally it should be stated that the heterologous expression of a potentially cell toxic protein that is soluble and active, and that also requires glycosylation constitutes a delicate approach in which several parameters must be optimized.

Accumulation of expressed LAAO in some organelles of the host, even if its expression construct should lead to export into the medium has raised further problems. Since we have never observed growth inhibition of Pichia pastoris upon induction, this seclusion might be an important factor in preventing toxicity. The localization of expressed LAAO in P. pastoris cells is currently under investigation.

4 CHAPTER: III

Mechanisms of Cell Death Induction by L-Amino Acid Oxidase, a Major Component of Ophidian Venom

Sudharsana Rao Ande*, Phaneeswara Rao Kommoju*, Sigrid Draxl, Michael Murkovic, Peter Macheroux, Sandro Ghisla, and Elisa Ferrando-May.

* These authors contributed equally to this work.

• Manuscript accepted for publication in Apoptosis journal

• The following experiments of the indicated sections were performed by myself: 4.3.5.

Measurement of rate of H2O2 production; 4.3.6. Normalization of enzyme activites to get similar rates of H2O2 production in the cell culture medium; 4.3.7. Amino acid analysis. (Preparation of samples and analysis of the data); 4.3.8. Preparation of the medium depleted of L-amino acids by pH and freeze inactivation method; 4.3.10. Desialylation of LAAO; 4.3.11. Generation of antibodies specific for LAAO; 4.3.12. Western blotting; 4.4.3. Sialic acid estimation by alkali Ehrlich´s method and 4.4.4, Cell binding experiments

4.1 Abstract

L-amino acid oxidase (LAAO) from the Malayan pit viper induces both necrosis and apoptosis in Jurkat cells. Cell death by necrosis is attributed to H2O2 produced by oxidation of α-amino acids. In the presence of catalase that effectively scavenges H2O2, a switch to apoptosis is observed. The major factors contributing to apoptosis are proposed to be: (i) generation of toxic intermediates from fetal calf serum (ii) binding and internalization of LAAO. The latter process appears to be mediated by the glycan moiety of the enzyme as desialylation reduces cytotoxicity. D-amino acid oxidase (DAAO), which catalyzes the same reaction as LAAO but lacks glycosylation, triggers necrosis as a consequence of H2O2 production but not apoptosis in the presence of catalase. Thus induction of cell death by LAAO appears to involve both the generation of H2O2 and the molecular interaction of the glycan moiety of the enzyme with structures at the cell surface.

4.2 Introduction

Apoptosis is a controlled and regulated form of cell death that plays an important role in the development and maintenance of higher organisms. It is defined by several morphological and biochemical hallmarks, like the exposure of phosphatidylserine to the outer leaflet of the plasma membrane, nuclear condensation, and chromatin cleavage into oligonucleosomal fragments. By contrast, necrosis represents an accidental form of cell demise resulting in early cell lysis, spillage of intracellular contents into the surrounding tissue, and inflammation.

Experiments performed in cell culture show that one and the same toxic insult, e.g.

the exposure to a prooxidant, can trigger either apoptosis or necrosis depending on its dose and duration of exposure [64-66]. Intracellular energy levels and mitochondrial function are primarily involved in the determination of the shape of cell death [67].

In fact, ATP and NAD⁺ were shown to be rapidly depleted in necrotic but not apoptotic cell death [64]. Moreover, under low ATP conditions typical apoptotic stimuli may induce necrosis [68, 69]. Recently, an elevation of the cytosolic ATP level was shown to occur upon apoptosis induction and to be required for the manifestation of apoptotic hallmarks [70].

One of the major regulators of cell death by apoptosis is the Bcl-2 protein, which exhibits a protective function against a wide array of stimuli and treatments, including chemotherapeutic agents, oxidative stress, growth factor withdrawal, or neurotoxins. Several mechanisms have been proposed for Bcl-2’s antiapoptotic effects, the best characterized being the control of mitochondrial integrity and the prevention of the release of apoptogenic molecules from the mitochondrial intermembrane space (reviewed in [71]). More recently Bcl-2 has been shown to operate upstream of mitochondria in the regulation of pathways that emanate from the endoplasmic reticulum. This involves the control of calcium homeostasis and of calcium fluxes between the ER and the mitochondria (see e.g. [72]). Further, Bcl-2 has been shown to protect from oxidative stress via upregulation of antioxidant enzymes and by elevating the concentration of cellular thiols [73-77].

It has been reported that L-amino acid oxidases (LAAOs), present in the venom of various snakes, such as crotalids, elapids and viperids, possess antibacterial properties [49] as well as an anti-HIV activity [58] and can induce cell death [6, 36, 55, 78]. These enzymes are members of the well-studied family of flavin (FAD) dependent oxidases. They catalyze the dehydrogenation of L-amino acids and the concomitant generation of H₂O₂ according to the sequences shown in Scheme 1.

In some snake species LAAO constitutes up to 30% of the total venom proteins [10]. Several related proteins with LAAO activity have been described that exhibit antimicrobial, antineoplastic or apoptosis-inducing activity [50]. The sources of these proteins are quite diverse ranging from the giant African snail (achacin from Achatina fulica Férussac [50]), to sea hares (dolabellanin A and aplysianin A from Dolabella auricularia and Aplysia kurodai, respectively [53]), and parasite infected fish (Apoptosis-Inducing Protein, AIP from Chub mackerel [6]). Interestingly the substrate preference of these enzymes is distinct from that of the snake venom LAAO, which is selective for aromatic (Phe, Tyr) and hydrophobic (Ile, Leu, Val) amino acids. Both aplysianin A and AIP are highly specific for the basic amino acids Lys and Arg, whereas achacin and dolabellanin A have a broader substrate specificity that includes aromatic, hydrophobic and basic amino acids [6].

Despite these differences in substrate specificity, it is assumed that the general mechanism of toxicity of these enzymes is based on the generation of cytotoxic amounts of H₂O₂ [21]. This hypothesis is reinforced by the protective effects of catalase, a scavenger of H₂O₂. However, D-amino acid oxidase (DAAO), which generates the same products from the oxidation of D-amino acids, was reported to lack antibacterial activity [49]. On the other hand the apoptosis-inducing activity of achacin and AIP have also been ascribed to the consumption of the amino acid substrates in the medium, e.g. Lys in the case of AIP [6] and aromatic and basic amino acids in the case of achacin [55]. A further factor that might play a crucial role in induction of cell death could be specific interactions of LAAO at the cell surface, as first suggested by Suhr and Kim [36]. Since the publication of that report, we have solved the 3D-structure of LAAO [18]. It shows a ≈20Å long and narrow channel

that connects the active center to the “outside” of the protein, and provides a route for substrate entry and product release [18]. In addition LAAO was found to be N-glycosylated at Asn361 and Asn172. The two glycan residues are remarkably homogeneous, consisting of a bis-sialylated, biantennary, core-fucosylated dodecasaccharide [17]. Intriguingly, the glycans are located in the vicinity of the channel [17]. This observation led us to posit the hypothesis that the proapoptotic effect of LAAO may involve interaction with the cell surface via sialic acid binding immunoglobulin-like lectins (siglecs). The effect could be twofold. Firstly, binding of LAAO to siglecs could target the exit of the active site channel to the plasma membrane resulting in high local concentrations of H₂O₂ that would be directly delivered to the cell and thus escape detoxification by catalase present in the medium. Secondly, LAAO may act via crosslinking of siglecs, a process that has been shown to trigger apoptosis in eosinophils and neutrophils [79].

Here, we report on the effects of LAAO from the venom of the Malayan pit viper (Calloselasma rhodostoma) on the viability of Jurkat cells and the influence of catalase, of LAAO´s glycan moiety, and of growth medium composition. Our results have important implications for understanding the mechanisms of cell death triggered by this family of enzymes

4.3 Materials and methods