Investigating the effect of carbon source on rabies virus glycoprotein production in Pichia pastoris by a transcriptomic approach

Abstract Several factors affect protein expression in Pichia pastoris, one among them is the carbon source. In this work, we studied the effect of this factor on the expression level of rabies virus glycoprotein (RABV‐G) in two recombinant clones harboring seven copies of the gene of interest. The expression was driven either by the constitutive glyceraldehyde‐3‐phosphate dehydrogenase (GAP) promoter or the inducible alcohol oxidase1 ( AOX1) promoter. Clones were compared in terms of cell physiology and carbon source metabolism. The transcription levels of 16 key genes involved in the central metabolic pathway, the methanol catabolism, and the oxidative stress were investigated in both clones. Cell size, as a parameter reflecting cell physiological changes, was also monitored. Our results showed that when glucose was used as the sole carbon source, large cells were obtained. Transcript levels of the genes of the central metabolic pathway were also upregulated, whereas antioxidative gene transcript levels were low. By contrast, the use of methanol as a carbon source generated small cells and a shift in carbon metabolism toward the dissimilatory pathway by the upregulation of formaldehyde dehydrogenase gene and the downregulation of those of the central metabolic. These observations are in favor of the use of glucose to enhance the expression of RABV‐G in P. pastoris.

2012), efficient transcription of the transgene using strong promoter (Gasser et al., 2013), protein folding in the reticulum endoplasmic (RE) (Vanz, Nimtz, & Rinas, 2014), and protein secretion (Pfeffer et al., 2011). Additionally, bioprocess parameters such as pH, temperature, growth rate, and substrate type also affect protein expression in P. pastoris (Dragosits et al., 2009;Files, Ogawa, Scamanb, & Baldwina, 2001;Xie, Zhou, Du, Gan, & Ye, 2004). However, only few studies were focused on the influence of heterologous protein expression on yeast metabolism (Baumann et al., 2010;Çelik, Çalik, & Oliver, 2009;Nocon et al., 2014;Prielhofer et al., 2015;Xie et al., 2004). Prielhofer et al. (2015) studied the transcriptional and translational profiles of P. pastoris cultivated in shake flasks under four bioprocess conditions: (1) excess of glycerol, (2) excess of glucose, (3) limiting glucose concentration, and (4) methanol induction conditions. They showed that the carbon source affects differently, the transcription level of various endogenous genes; however, cells grown on an excess of the carbon source (glucose or glycerol) showed comparable transcriptome. They also found that the synthesis of ribosome components was not affected by methanol despite the low growth rate depicted by the cells grown under this condition. Other studies (Inan & Meagher, 2001;Xie et al., 2004) showed that the carbon source also affects the expression of heterologous genes besides endogenous genes. Xie et al. (2004) reported that different carbon sources like acetate, glycerol, sorbitol, and lactic acid used during the cultivation of recombinant P. pastoris displayed different effects on angiostatin production level. The highest angiostatin production level was achieved when lactic acid or sorbitol were used. Other carbon sources such as mannitol, alanine, and sorbitol have also been tested for the production of β-galactosidase by recombinant P. pastoris Mutclones (Inan & Meagher, 2001). All these carbon sources were able to improve β-galactosidase production level as compared to glucose or glycerol, and to reduce the amount of methanol required for the expression of the heterologous protein. The use of mixed substrates can have some attractiveness when setting up the process at large scale; it can reduce the quantity of methanol, and therefore the risk associated with the storage of large amounts of this product, and consequently can contribute to reduce the overall cost.
The carbon source can also affect the intracellular amount of the heterologous protein, even if the expression is even if the protein is secreted.
In this line, Hohenblum, Gasser, Maurer, Borth, and Mattanovich (2004) demonstrated that recombinant trypsinogen level retained in P. pastoris cells was dependent on the carbon source but not on the promoter.
In previous studies, we generated two recombinant clones of P. pastoris KM71H Mut S harboring seven copies of the rabies virus glycoprotein (RABV-G) gene (Ben Azoun, Belhaj, Göngrich, Gasser, & Kallel, 2016;. The expression of the target protein was driven either by AOX1 promoter (aox7) or GAP promoter (gap7) and directed in both clones to secretion by the alpha mating factor of Saccharomyces cerevisiae. Expression levels of RABV-G obtained by these clones were compared and a significant difference was seen in extracellular and intracellular compartments. We showed that this difference was obviously substrate and not promoter- In this work, we aim to obtain deeper insights about the expression of RABV-G in P. pastoris; we analyzed cell morphology and the transcriptional responses of cells grown in methanol and glucosecontaining media. A comparative transcriptional analysis was performed with several key genes used as indicators of the physiological state of recombinant P. pastoris clones to determine the effect of carbon metabolism on the production of RABV-G in this yeast.

| Strains and media
P. pastoris KM71H (Invitrogen, CA, USA) was used in this study.
Optimized RABV-G gene (Genbank accession number KT878717) was used for the construction of the expression cassette. The generation of multi-copy clones used in this work (gap7, aox7) was previously described in details (Ben Azoun, Belhaj, Göngrich, et al., 2016;; these clones were constructed using two different recombinant vectors pPICZαA or pGAPZαB. The expression was controlled by AOX1 or GAP as a promoter, α-mating factor of S. cerevisiae was used to direct the recombinant protein to extracellular medium. Recombinant vectors were linearized by BstXI and transformed into P. pastoris KM71H strain. Numerous transformants were generated; RABV-G gene copy number was determined by real-time quantitative PCR. Recombinant clones harboring seven copies of the RABV-G gene named gap7 and aox7 were used in this work. Empty vectors (pPICZαA and pGAPZαB) were transformed into P. pastoris KM71H; selected clones were named as NC (negative control).

| Expression of recombinant RABV-G protein in deep well plates
Recombinant P. pastoris clones were grown in YPD agar plates at 30°C.
For expression studies, a single colony of selected recombinant clones was used to inoculate 2 ml of BMGY in deep well plate (Dominique Dutscher, France), then grown overnight at 30°C and 250 rpm. After 14-16 hr, optical density was measured at 600 nm, and cells were resuspended in 2 ml of fresh BMMY medium or BGY to an initial OD 600 of 1. Cultures were performed at 250 rpm 30°C up to 72 hr; Carbon mole (C-mol) amount was kept similar in both cultures. Methanol and glucose levels were equal to 10 g L −1 and 7.2 g L −1 , respectively; they were added to 1% every 24 hr of culture. Aliquots were taken every 24 hr, cells were pelleted; supernatants and cells were stored at −20°C for further analysis.

| Analytical methods
Biomass level was determined by optical density at 600 nm. Yeast cell morphology was determined by direct examination of a drop of recombinant P. pastoris clone culture using a light microscope (Leica DFC425, Germany), and 40X objective. Cell size (μm) was estimated on 80 cells using the image-J software.
Glucose level was estimated by an enzymatic assay kit (Eurodiag, France). Methanol concentration was estimated by Gas chromatography (GC) (Shimadzu, Kyoto, Japan).

| Enzyme-linked immunosorbent assay test (ELISA)
ELISA test was performed to quantify of RABV-G expressed by the different clones. Detailed protocol was described in Ben Azoun, Belhaj, Göngrich, et al., 2016. Briefly, 100 μl per well of either the sample or the standard (inactivated and purified rabies virus) were incubated for 2 hr at 37°C. Thereafter, monoclonal antibody anti-glycoprotein TW1 (NIBSC, Hertfordshire, UK) was added to the wells and incubated for 1 hr at 37°C. Finally, anti-human antibody coupled to peroxidase (Sigma Aldrich) were added and incubated 30 min at 37°C. After tetramethylbenzidine addition, the reaction intensity was measured at 450 nm. OD values higher than 0.150 were considered as positive.

| RNA extraction
For transcript quantification, frozen cells were resuspended in Trizol reagent (Invitrogen, CA, USA) and disrupted with glass beads in FastPrepTM cell homogenizer (Thermofisher, MA, USA). Total RNA was then extracted using the RNeasy Kit from Qiagen following the manufacturer's instructions. RNA was tested in 1% agarose gel, and was quantified by measuring OD 230/260/280 using a NanoDrop (Thermo Scientific, MA, USA).

| Quantitative real-time PCR to determine gene transcriptional level
The PCR primer design was conducted using Primer 3 software (http:// www.genome.wi.mit.edu/ftp/distribution/software/). qPCR primers used in this work are listed in the supporting information Table S1.
RT-qPCR was performed using a thermal cycler (Bio-Rad, CA, USA). After preincubation at 95°C for 10 min, the thermal cycler was

| Transcription levels of RABV-G and intracellular genes
The mRNA level of RABV-G and the intracellular genes in each clone was normalized using actin gene as the endogenous control (housekeeping gene). In this case, the transcription levels of RABV-G gene and intracellular genes in the negative control (NC) (clone transformed with the empty vector) were set as control to normalize the data.
RT-qPCR reactions were run in triplicate with biological replicates (independent experiments) to allow for the statistical confidence in differential gene expression. As amplification efficiencies of the housekeeping reference gene and the target gene were similar, the relative quantification is then determined according to the following equation as described by Dheda et al. (2014).
where ∆Ct = Ct of target -Ct of reference (Actin gene) After 24 hr of culture, aox7 clone cells grown in BMMY medium containing methanol as the sole carbon source showed a cell size mRNA level = 2 −ΔCt F I G U R E 1 Cell size distribution of aox7 and gap7 recombinant clones throughout culture time. Cells were observed under a light microscope (X40) and cell size was determined by Image-J software (p-values <.05) which was 1.27-fold smaller than the gap7 clone cell size ( Figure 1).

| Cell morphology, growth profiles, and RABV-G production of recombinant clones of P. pastoris expressing RABV-G
Cell size difference during both cultures increased over time. At 48 hr of aox7 clone culture, cell size evolved to 1.44 ± 0.13 μm which was 1.46-fold lower than gap7 clone cell size (2.11 ± 0.19 μm), while at 72 hr of culture, cell size difference became more significant. At this culture time, yeast cells cultivated on methanol showed a cell size around 1.49 ± 0.5 μm, whereas for those grown on glucose, cell size was 2.68 ± 0.25 μm. These results show that aox7 clone cell size was almost constant during the culture, while for gap7 clone, cell size increased by 1.17-fold and 1.5-fold at 48 hr and 72 hr, respectively, when compared to 24 hr of culture ( Figure 1).
The use of different carbon sources also resulted in different biomass levels; cell growth profile showed a continuous increase for both clones (Figure 2a). For aox7 clone culture, OD 600 level reached 18 and 38 after 24 hr and 72 hr of induction, respectively. While for gap7 clone, biomass level increased more rapidly, OD 600 reached 21 at 24 hr and 43 at 72 hr. Final biomass level obtained during gap7 clone culture was slightly higher than that of aox7 clone.
RABV-G level increased with cell growth in both cultures. For aox7 clone, the highest expression level was obtained at 72 hr and was equal to 127 ng ml −1 . On the other hand, the secreted protein (RABV-G) concentration achieved during gap7 clone culture was 1.2fold higher (152 ng ml −1 ) than aox7 clone production level.
The difference in RABV-G production level observed between both clones can not be explained by a difference of the transcriptional level of RABV-G gene, as no significant difference in RABV-G mRNA levels was seen ( Figure S1). Hence, it appears that this difference is not due to the promoter itself, and that RABV-G expression depends on culture conditions, especially the carbon source used by the cells.
The difference in biomass level and RABV-G concentration observed during aox7 and gap7 cultures in different media containing the same carbon molar amount, suggest that cells exhibit a differential use of carbon source. Residual methanol and glucose were not detected in the culture medium for both clones (Figure 2b), showing a full uptake of the carbon source.
Correlating heterologous protein secretion to cell growth is not straightforward and the relation between those parameters is complex.
Nevertheless, Buchetics et al. (2011) reported that recombinant clones of P. pastoris producing proteins in large amounts are mainly in M and G2 phases of the cell cycle. The authors also reported that overexpression of the main mitotic cyclin CLB2, which promotes entry into mitosis and progression from the metaphase to anaphase transition, has improved antibody Fab fragment and human trypsinogen production by 1.32-and 2.1-fold, respectively, in P. pastoris recombinant clones.
Based on the impact of this protein on cell cycle phase, an analysis of the transcriptional level of the CLB2 gene was performed by RT-qPCR in the aox7 and the gap7 clones. Figure 3 showed that the transcript level of CLB2 gene in the gap7 clone was 1.63-fold higher than the level seen in clone aox7 at 72 hr of culture. This result confirms that a high secretion level of heterologous proteins in P. pastoris correlates with a high transcription level of CLB2 gene and the predominance of cells in G2 and M phases of the cell cycle.

| Transcription of central metabolism genes
To obtain better insights into host physiology, especially to identify changes in the central carbon metabolism in different culture F I G U R E 2 (a) Effect of carbon source on biomass and RABV-G production level. The gap7 and aox7 clones were cultivated on glucose and methanol, respectively. (b) Residual carbon source concentration during gap7 and aox7 culture on glucose and methanol, respectively

| Transcription of methanol metabolism genes
In order to check if there is a difference between the activation of methanol dissimilatory pathway and assimilatory pathway upon methanol induction, we determined the transcript levels of key genes involved in methanol metabolic pathway after 72 hr of induction of the aox7 clone. Genes used in this study are shown in Figure 4; transcripts of alcohol oxidize II (the second enzyme involved in methanol utilization, AOX2), formaldehyde dehydrogenase and catalase (two enzymes involved in methanol dissimilatory pathway, FLD and CAT1 to generate CO 2 and O 2 , respectively), dihydroxyacetone kinase (DAK), dihydroxyacetone synthase (DAS1/2), and triose phosphate isomerase (TPI1) were determined. These three key enzymes are involved in the methanol assimilatory pathway to provide the constituents for cell synthesis.  respectively) in the aox7 clone when compared to the negative control. Hence, the increase in the transcriptional levels of methanol assimilatory genes was lower compared to those of the genes involved in methanol dissimilatory pathway ( Figure 6). These results suggest that the dissimilation pathway which competes with the assimilation pathway for carbon flow was enhanced. Thus, methanol being incorporated into biomass might be reduced further; this explains why biomass level of aox7 clone culture was lower than that seen with gap7 clone.

| Transcription of antioxidative genes
P. pastoris cells can support ER stress to some extent without altering their metabolism. Four main genes associated in oxidative stress were chosen for testing: (1) the transcription factor YAP1 implicated in oxidative stress response and redox homeostasis; it actives the transcription of genes which encode the antioxidant enzymes.; (2) gammaglutamylcysteine synthetase (GSH1) is the ubiquitous thiol-containing reductant that maintains the intracellular redox homeostasis by decreasing cellular disulfide bonds and detoxifying harmful molecules; (3) Glutathione peroxidase (GPX1) is a essential enzyme of the antioxidant system, involved in reducing H 2 O 2 and organic hydroperoxides to H 2 O using reduced glutathione as an electron donor (Margis, Dunand, Teixeira, & Margis-Pinheiro, 2008); (4) Glutathione reductase (GLR1) is intended for the recycling of the redox buffer, it is implicated in the defense of yeast cells against oxidative stress and is regulated by YAP1. Figure 7 indicates that these antioxidative-related genes were upregulated at different levels for aox7 clone; the transcript levels of YAP1, GLR1, GPX1, GSH1 increased by 2-, 1.68-, 1.8-, 1.7-fold, respectively. However, for gap7 clone grown on glucose, the increase of transcription levels of these genes was lower than that seen for aox7 clone; they increased by 1.26-, 1.1-, 1.2-, 1.45-fold for the genes YAP1, GLR1, GPX1, GSH1, respectively.
A significant increase of GSH1 gene transcript was observed in both clones (Figure 7). This is due to the oxidative folding of RABV-G, since both clones contain a high RABV-G gene copy number.

| DISCUSSION
Although P. pastoris is an expression system widely used for protein production, its physiology during heterologous protein expression using different carbon sources has not been extensively studied as it was done for S. cerevisiae and E. coli. Only few studies describing the effect of central metabolism on recombinant protein production were reported (Çelik et al., 2009;Heyland, Fu, Blank, & Schmid, 2011;Nocon et al., 2014;Prielhofer et al., 2015).
In this work, we studied the effect of carbon source on the physiology of recombinant clones of P. pastoris expressing RABV-G. Cell size is considered as one of the most important determinant of cellular physiology; hence cell size of gap7 and aox7 clones grown on glucose or methanol, respectively, was determined. We showed that cells of gap7 clone cultivated on glucose were larger than those seen on methanol ( Figure 1). This result indicates that cell morphology of P. pastoris recombinant clone can be affected by the carbon source. Turner, Ewald, and Skotheim (2012) reported that cell size depends on extracellular conditions. In general, growth rate and nutrient conditions (type of carbon source and its concentration) are two key elements that significantly affect cell size. Yeast cells cultivated on a rich medium manifest a larger size, whereas those growing on a poor medium show a slow growth and a smaller cell size. Cipollina et al. (2008) reported that cell switch from one medium condition to another containing, for example, a different carbon source, influences cell size. When cells of S. cerevisiae were removed from ethanol (slow growth) to glucose (fast growth) they rapidly adapt to a larger size. Pluskal, Hayashi, Saitoh, Fujisawa, and Yanagida (2011) also showed that when a deficiency in glucose In this study, glucose and methanol as a carbon source were compared during the culture of two recombinant clones expressing RABV-G.
Glucose seems to be a better carbon source than methanol, resulting in higher biomass concentration, and higher RABV-G yield ( Figure 2).
However, comparable specific growth rates (average specific growth rate for both clones around 0.04 hr −1 ) were obtained for both clones which is contrary to data described in the literature. Gap7 clone should exhibit a higher specific growth rate than aox7 clone. Nevertheless, gap7 clone showed a specific productivity of RABV-G which was 20% higher than that obtained with aox7 clone, probably because of the higher transcriptional levels of the genes involved in the central metabolic pathway as explained below. Generally, it is well recognized that cell growth rate has an important role in gene regulation, and therefore in protein production (Liang et al., 2014). It appears that a high growth rate favors protein production in P. pastoris since genes involved in gene translation and expression are overexpressed. On the other hand, Buchetics et al. (2011) reported that P. pastoris recombinants clones exhibiting a high secretion level are mostly in the G2 and M phases of cell cycle. The relation between cell cycle phase and protein production has been described not only in yeasts (Frykman & Srienc, 2001) but also in mammalian cells (Charlet, Kromenaker, & Srienc, 1995). The eukaryotic cell cycle has been examined comprehensively in the model organism S. cerevisiae reviewed in Cid et al. (2002).
The appropriate progress of cell cycle is monitored by interferences of cyclins and cyclin-dependent kinases (CDK) and related regulators and phosphatases. Buchetics et al. (2011) showed that upregulation of cyclin gene CLB2 of P. pastoris led to the expected cell cycle shift toward a higher fraction of cells in the G2/M phase, and subsequently increased the product yield. In this study, the CLB2 transcript level in the two clones during their cultivation on glucose or methanol was determined. Interestingly RT-qPCR data showed that mRNA level of CLB2 in the gap7 clone cultivated on glucose were 1.63-fold higher than in the aox7 clone cultivated on methanol after 72 hr of culture.
A higher fraction of gap7clone cells in the G2/M phase compared to aox7 clone would probably be observed. Buchetics et al. (2011) also showed that cell cycle phase impacts protein secretion; co-overexpression of CLB2 mitotic cyclin in recombinant P. pastoris clones resulted in the enhancement of the production levels of human trypsinogen and Fab proteins. Thus, it seems that high level secretion of heterologous proteins correlates with the overexpression of the CLB2 mitotic cyclin. It would be sound to verify if this hypothesis can be applied as a general rule to enhance recombinant protein expression in P. pastoris.
Our results also show that co-overexpression of CLB2 is expected to increase the expression level of RABV-G in P. pastoris. Overexpression of other mitotic cyclins like CLB4 can also improve RABV-G expression in this yeast, as shown for other proteins (Buchetics et al., 2011).
In this work, the main enzymes of the central carbon metabolism pathway were studied. The aim was to investigate the effect of the carbon source on the physiology of recombinant clones of P. pastoris.
Glycolysis pathway starts with glucose conversion into glucose-6phosphate which can then be converted into biomass or to fructose-6-phosphate. In the case of methanol, the central carbon metabolism is activated from glyceraldehyde-3-phosphate dehydrogenase (GAP) to generate the energy necessary for cell growth and protein production (Çelik et al., 2009). The expression of genes related to glycolysis such as GAP and PYK in the gap7 and the aox7 clones are shown in Figures 4 and 5. We found that the transcription levels of glycolysis genes in cells growing on methanol were lower compared to those cultivated on glucose. Interestingly, similar result was observed in a recombinant clone of P. pastoris grown on methanol and expressing porcine insulin precursor (Zhu, Guo, Zhuang, Chu, & Zhang, 2011).
The transcription levels of glucose-6-phosphate dehydrogenase (ZWF1), a key enzyme in the pentose phosphate pathway, pyruvate dehydrogenase (PDH) which is an acetyl-CoA producing enzyme, and citrate synthase (CIT1) which is one of the enzymes of the TCA cycle were determined. All these genes were upregulated on glucose but downregulated on methanol. Therefore, it appears that there is a correlation between the upregulation of the central metabolic genes, high biomass concentration, and high product titer during P. pastoris growth on glucose compared to methanol. Prielhofer et al. (2015) showed that various P. pastoris genes encoding enzymes implicated in the metabolism of carbon sources are transcribed differently depending on the carbon source used, they found that P. pastoris glycolytic genes were downregulated during culture on limiting level of glucose, glycerol, and methanol compared to an excess of glucose. However, the genes encoding the gluconeogenic enzymes were downregulated on excess glucose compared to the other conditions. Nocon et al. (2014) reported that the central metabolism pathway especially the pentose phosphate pathway (PPP) affects the production level of cytosolic human superoxide dismutase in P. pastoris. They also showed that coexpression of ZWF1gene had a positive effect on intracellular production of human superoxide dismutase and bacterial ß-glucuronidase in P. pastoris (Nocon et al., 2014). Our study suggests that overexpression of ZWF1 would also improve the expression level of heterologous proteins targeted to secretion in this yeast. It will be also useful to study if co-overexpression of the 6-gluconolactonase (SOL3), the second enzyme in the PPP would improve the secretion of RABV-G, as demonstrated for intracellular expression by Nocon et al. (2016).
Methanol metabolism is another remarkable physiological change in the aox7 clone. After 72 hr of induction, the transcript levels of genes involved in methanol dissimilatory pathway (FLD1 and CTA1) were highly induced compared to genes involved in assimilatory pathway (DAS1/2, DAK2, TPI1). These observations correlate with those described by Zhu et al. (2011), they showed that the genes involved in dissimilatory pathway were upregulated in high copy clones expressing porcine insulin precursor (PIP) under the control of AOX1 promoter. The significant activation of methanol assimilation pathway compared to central carbon metabolism pathway could be the reason for reduced RABV-G protein production in the aox7 clone compared to gap7 clone, particularly when we take into account that RABV-G transcript level was similar for both clones (Ben Azoun, Belhaj, Göngrich, et al., 2016; Ben Azoun, Belhaj, & Kallel, 2016). These data indicate that methanol does not offer enough carbon and energy to strongly sustain cell growth and RABV-G production. One way to alleviate these limitations could be the co-overexpression of the proteins involved in methanol assimilation pathway.
Methanol catabolism generates the production of H 2 O 2 , needing the action of antioxidants. YAP1, the oxidative stress response transcription activates the expression of genes related to the glutathione redox system such as GLR1, GPX1, and GSH1. This study showed that these genes were significantly highly expressed in cells growing on methanol compared to those growing on glucose. The upregulation of GSH1 transcript level was observed in both conditions suggesting that cysteine was limited in both recombinant clones (Nisamedtinov et al., 2011).
The rise of oxidative stress genes in the aox7 clone indicates that cellular redox potential gradually shifted to a more oxidizing state throughout culture on methanol. An explanation might be that the antioxidant gene GSH1 was upregulated during RABV-G production on methanol, due to oxidative folding of the foreign protein, meaning that the cells become highly exposed to formaldehyde accumulation. Thus, P. pastoris induced the up-expression of FLD1 to detoxify formaldehyde (Sunga & Cregg, 2004;Zhu et al., 2009) and to maintain healthy redox balance in this metabolic state.
In conclusion, in this study, we studied the impact of glucose or methanol as a carbon source on the physiology of P. pastoris recombinant clones expressing RABV-G through the transcriptional analysis of genes involved in carbon metabolism pathways. Our results suggest that when glucose was used, an increase in biomass and RABV-G production levels were obtained when compared to the recombinant clone grown on methanol. Furthermore, the use of methanol as the sole carbon source resulted in a raise of oxidative stress. Hence, glucose appears to be the preferred carbon source for the expression of RABV-G in P. pastoris and can be utilized to set-up the high cell density production process in a bioreactor.

ACKNOWLEDGMENTS
This work is supported by the Tunisian Ministry of Scientific Research and Higher Education (grants LR11IPT01 and "Valorisation des Résultats de la Recherche").