**3. Discussion**

TAG lipases responsible for the degradation of the lipids accumulated in oil bodies are attractive knockdown targets for the enhancement of the lipid productivity and storage in microalgae. Nonetheless, considering the numerous data available on bacterial, terrestrial plant and animal lipases those from algae and more especially microalgae have been relatively neglected. Therefore, more emphasize has to be given to the characterization of algal lipases, and hence, further work is needed in these aspects. Future approaches to maximize the enzymatic potential of microalgae are likely to focus on three different strategies: (i) the use of ever-increasing amounts of available omics data to optimize microalgal strains for the production of valuable products, through the overexpression of one or more enzymes by the use of genome editing tools; (ii) the identification and subsequent characterization of metabolic pathways involving the production of specific enzymes, such as lipases which are still poorly characterized; (iii) the search for genes with direct biotechnological applications in microalgal genomes and transcriptomes datasets. The feasibility of employing any of the aforementioned approaches or a combination of them will be directly influenced by progress in growth and genetic manipulation of microalgae.

In this study, we have used computational approach to identify lipase genes and to classify the respective lipases from a *C. vulgaris* strain. Lipases operate usually at the interface between lipid and water. An important feature of many lipases that is used for lipase classification is the presence of a mobile subdomain lid or flap located over the active site [22]. Among the 14 putative TAG lipases identified after *C. vulgaris* genome analysis, 10 have high identity in ESTHER database with Lipase\_3 domain-containing protein. Family 3 of lipolytic enzymes are widely distributed in animals, plants and prokaryotes and possess the conserved consensus sequence GXSXG. Members of this family were demonstrated to be very closely related and exhibit the canonical α/β-hydrolase fold as well as the typical catalytic triad. Enzymes of this class exhibit also high activities at low temperature (less than 15 ◦C) believed to originate from a conserved sequence motifs they display [23]. Four lipases out of the 10 aforementioned were predicted to be either cytoplasmic, chloroplastic or extracellular. The six remaining could be anchored to a membrane with a distinct N-terminal transmembrane domain formed by at least four transmembrane helices (Figures 1 and 2). Lip\_4551 and Lip\_2999 were predicted with quite similar 3D models composed of three domains: a catalytic domain containing the catalytic triad and a one helix lid, an N terminal transmembrane domain formed by long helices and a C terminal domain with mainly α helices (Figure 2). It has been reported that 10 additional modules can be attached to the core domain including lid modules, cap modules, N-terminal or C-terminal domains. Accordingly, superfamilies could be assigned to five groups (core, lid, cap, one additional domain or two additional domains) [24]. Predicted transmembrane

domains by bioinformatic tools were already reported for microalga lipases [25]. Some authors characterized and used as a self-immobilized lipase for esterification reactions membrane bound lipase from microalga [21]. The membrane localization could be in intracellular or extracellular counterparts or even in lipid droplets (LD). In eukaryotes, some TAG lipases and their cofactors have been demonstrated to localize to LDs [26]. For example, Diatom Oleosome-Associated Protein 1 (DOAP1) is translocated from the ER to LDs in *Fistulifera solaris* [27].

**Figure 1.** 3D models of seven putative lipases without transmembrane domains. Four of them (Lip\_1704, Lip\_1795, Lip\_4364, Lip\_6297) display only the core module with a Rossman Fold architecture. Lip\_3448 presents a C2 N-terminal domain, while Lip\_5800 and Lip\_5999 present a C-terminal alpha helices module. Lids are shown in dark blue and active site serine in yellow sticks.

As for Lip\_3448, the N terminal module is a PLAT domain found in a variety of lipid-associated proteins. It forms a β-sandwich composed of two β-sheets of four βstrands each, which is known as a C2 domain in pfam classification. Interestingly, two predicted lipases have a C terminal module only composed of α helices. These two proteins (Lip\_5800 and Lip\_5999) are shown to be closely related in cladogram of sequence similarity. Hence, the predicted lipases could be classified into a main core with Rossman fold architecture lipases (Lip\_1704, Lip\_1795, Lip\_4364, Lip\_6297, Lip\_5462, Lip\_3928, Lip\_4575), two domain lipases (Lip\_4232, Lip\_3448, Lip\_3076, Lip\_5800, Lip\_5999) and three domain lipases including a transmembrane domain (Lip\_2999 and Lip\_4551).

**Figure 2.** (**a**) Three predicted membrane-associated lipases with a transmembrane module shown in lilac; lids are shown in dark blue and active site serine in yellow sticks. (**b**) Gene annotation and domain boundaries of Lip\_4551 (left panel), Qmeanbrane result for transmembrane localization for Lip\_4551 (right panel).

Oxyanion holes are crucial for high-energy oxyanion intermediate stabilization. They consist of two residues, which donate their backbone amide protons to stabilize the substrate in the transition state. In fact, during hydrolysis, a negatively charged tetrahedral intermediate is generated and the oxygen ion formed modulates the catalytic efficiency of the enzyme [28]. The first residue is located in the structurally conserved nucleophilic elbow. As a consequence, its backbone amide is positioned identically in all lipases. In contrast, the second oxyanion hole residue is not located in a region with conserved sequence and structure between lipases, but in a loop between the β3-strand and the αA-helix in the core module [29,30]. Consequently, lipases are classified into three classes according to their oxyanion hole type: GX, GGGX and Y [31]. In all lipases, the first oxyanion hole is a conserved glycine which contacts the nucleophilic elbow (highlighted with a star in Figure 3). When the oxyanion hole is formed by the amide backbone of the C-terminal neighbor X of this conserved glycine, it is termed as 'GX type,' with X being the second oxyanion hole residue. In our case, the inspection of the multiple sequence alignment of the 14 lipases demonstrates they belong all to the GX class with the conserved glycine (G) residue followed by an alanine (A), cysteine (C) or serine (S) residue (Figure 3). The lipases with GX oxyanion hole type are widely distributed and diverse, and they usually prefer hydrolyzing medium and long chain substrates [32]. The type of amino acid X is conserved inside the superfamilies; for example, it is hydrophilic in *Candida antarctica* like lipases (T), filamentous fungi lipases and cutinases (S, T), and hydrophobic in *Moraxella* (F), *Mycoplasma* (F, W) and *Pseudomonas* lipases (L, F, M) [29].


**Figure 3.** Multiple sequence alignment of putative lipases showing the conserved lipase 3 motif GXSXG and the conserved G residue for GX classification highlighted with orange star.

According to the shape of the binding site cavity, lipases can be divided into three categories: (i) lipases with a funnel-like binding site (lipases from the mammalian pancreas and cutinase), (ii) lipases with tunnel-like binding sites (lipases from *Candida rugosa*, and *Candida antarctica* A) [33] and (iii) lipases with a crevice-like binding site (lipases from *Rhizomucor* sp. and *Rhizopus* sp.) [34]. It should be noted that most of the template structures used for 3D modeling are lipases from *Rhizomucor miehei.* In addition, the inspection of predicted open lid models like Lip\_1704 showed a crevice-like cavity shape as shown in Figure 4.

**Figure 4.** (**a**) Slabbed close up view of the active site cavity for Lip\_1704 showing a crevice-like shape (**b**) A surface top view with DEPTH showing the shape of substrate entrance in the same protein.

The amphipathic nature of the lid is crucial for the substrate specificity. It provides new insight into the structural basis of lipase substrate specificity and a way to tune the substrate preference of lipases. Based on the type of lid domain, lipases were also classified into three groups, such as lipases without lids, lipases with one loop or one helix lids and lipases with two or more helix lids. It has been reported that high temperature lipases contain larger lid domains with two or more helices, and that all mono- and diacylglycerol lipases have a small lid with a form of loop or helix [22]. As shown in Figures 1 and 2, almost all lipases found in *C. vulgaris* have small lids with one loop (Lip\_4364) or one helix (Lip\_1704). However, Lip\_5462 displays an entire cap domain with three small helices lid covering a deep cavity of 15.6 Å and shows, surprisingly, 40% of sequence identity with human lysosomal acid lipase (LAL). In fact, it has been demonstrated that, in addition to the direct association of lipases to oil bodies, macro-autophagy (referred to as lipophagy) plays a critical roles in lipid catabolism in eukaryotes [35]. During this type of autophagy, autophagosomes containing a portion of an oil body are merged with lysosomes containing LAL, which could contribute to TAG degradation [36]. Transcriptomic analysis of *Neochloris oleoabundans* (an oleaginous microalga) reveals up regulation of an LAL encoding gene under nitrogen starvation condition [37]. Accordingly, the in silico prediction method used for lipases of *C. vulgaris* allowed the identification of Lip\_5264, which could be transported to lysosomes. This enzyme was predicted to have a signal peptide and 40% of sequence identity with the human LAL. It consists of a core domain belonging to the classical α/β hydrolase-fold family with a classical catalytic triad (Ser-161, His-378, Asp-347), an oxyanion hole and a "cap" domain, which probably regulates substrate entry to the catalytic site (Figure 5). LAL breaks down cholesteryl esters (CEs) and TGs into free cholesterol, glycerol and fatty acids (1–3). Defective LAL have been associated with two autosomal recessive diseases in humans: Wolman's disease and CE storage disease [38,39]. The gene of Lip\_5264 consists of 8 exons spread over almost 4 kb, while human LAL consists of 10 exons spread over 36 kb. Lip\_5264 encodes a 445 amino acid mature protein following the cleavage of 24 signaling peptide residues, with an expected molecular mass of 50 kDa whereas human LAL encodes for 378 residues with a signal peptide of 21 amino acids and a molecular mass of 43 kDa. The two compared proteins are glycosylated and share high structure identity, as shown in Figure 5c with some differences, including the lid helices, which contain a cluster of highly conserved Cys residues C 236 and C 243 (Lip\_5264 numbering) (Figure 5d). The lysosomal proteins in microalga have not ye<sup>t</sup> been fully investigated, and it remains unclear how lipophagy contributes to lipid degradation. These should be an attractive research topic in a future work. Microalgae are a good source of nutrients for human nutrition. However, they are also rich in various biomolecules, which may have a potential in promoting human health. Defective or diminished LAL activity of human LAL has been associated with some mutations and the molecular mechanisms of these loss-of-function mutants leading to WD and CESD have ye<sup>t</sup> to be explored. Some study demonstrated that these mutations could be located in the signal peptide or in the lid domain [40]. A complete physicochemical characterization of this *C. vulgaris* LAL combined with a deep structure–function relationship investigation of the probable mutation effect using a structure-based molecular model speculating the loss of function could be of interest. The current treatment options for CESD phenotypes are limited to diets excluding cholesterol and lipid-rich food, cholesterol lowering drugs such as statins and ultimately liver transplantation. Recombinant LAL replacement therapy has been shown to be effective in animal models and human clinical trials and was recently authorized in Europe and the United States [41].

**Figure 5.** (**a**) 3D model of putative Lysosomal acid lipase Lip\_5462 lid is shown in dark blue and catalytic serine in yellow sticks. (**b**) Crystal structure of human Lysosomal acid lipase PDB ID: 6V7N. Lid is shown in orange and catalytic serine in yellow sticks. (**c**) The overlay of the two aforementioned structures showing high structure similarities and lid differences. (**d**) A close up view of the three helices lid of Lip\_5462 showing conserved cysteine residues in red sticks.

#### **4. Materials and Methods**
