Chlorophyta microalgae as dietary protein supplement: a comparative analysis of productivity related to photosynthesis

Microalgae are studied as innovative sources of a wide range of highly valuable products, including proteins for the food/feed sectors. However, protein content varies depending on algal species, culture conditions and harvesting period. The Phylum Chlorophyta includes most of the described species of green algae. Due to their remarkable peculiarities, four Chlorophyta species belonging to two distinct classes were selected for the research: Chlorella vulgaris and Chlorella protothecoides as Trebouxiophyceae, and Neochloris oleoabundans and Scenedesmus acutus as Chlorophyceae. The algae were studied to obtain comparative results about their growth performance, and total protein content and profile under the same culture conditions. Since photosynthetic process directly influences biomass production, photosynthetic pigment, PSII maximum quantum yield and thylakoid protein content and profile were analysed. SDS-PAGE and 2D BN/SDS-PAGE were performed to expand information on the organization and assembly of the resolved thylakoid complexes of samples. Despite the algal species showed similar growth rates and photosynthetic efficiency, S. acutus showed the highest capability to accumulate proteins and photosynthetic pigments. Regarding the thylakoid protein profile, the two Trebouxiophyceae showed very similar pattern, whereas different amounts of LHCII occurred within the two Chlorophyceae. Finally, the separation of thylakoid protein complexes in 2D BN/SDS-PAGE revealed a more complex pattern in S. acutus as compared to the other species. Overall, it is suggested that a higher stability of the photosynthetic membranes can result in higher biomass and protein production. Altogether, results have highlighted the metabolic uniqueness of each strain, resulting in a non-obvious comparison with the other species.


Introduction
The potential of microalgae as a source of renewable energy and high-value bio-products has been widely explored (Khan et al. 2018;Camacho et al. 2019). For nutritional purposes, proteins and lipids, especially omega-3 polyunsaturated fatty acids (PUFAs), are the most interesting molecules; moreover, vitamins, oxidation-protecting pigments (i.e. carotenoids, Cars) and other compounds contribute to the nutritional value of microalgae Wells et al. 2017;Khan et al. 2018;Camacho et al. 2019;Parisi et al. 2020). The advantage of cultivating these organisms on non-arable land, extreme environments up to space, in microgravity conditions (Stankovic 2018), lets us consider them a "crop for the future" (Torres-Tiji et al. 2020). Among the aforementioned compounds, proteins are one of the most important and can represent up to 50% of the total microalgal biomass dry weight (DW) (Lourenço et al. 2004;Hayes et al. 2017;Camacho et al. 2019). Moreover, various microalgal species are studied as innovative sources of proteins in food sectors, not only for their high protein Costanza Baldisserotto and Alessandra Sabia contributed equally to this work.

3
content, but also for the nutritional quality, which is comparable to that of other referenced food proteins (Spolaore et al. 2006;Hayes et al. 2017;Khan et al. 2018;Jayappriyan et al. 2021). Among the protein-rich microalgae, Chlorella and Arthrospira are the most widely used both as food and as a nutraceutical supplement. The success of these algae for the market is linked to their high content in different types of proteins, some of which positively impact health, for example by reducing cholesterol levels Khan et al. 2018). However, other microalgae species, like those belonging to Scenedesmus, Dunaliella and Aphanizomenon genera, are proposed to produce proteins with nutraceutical potential (Buono et al. 2014). In this perspective and taking into account the increasing protein demand for nutrition (Henchion et al. 2017;FAO 2020), the identification of the best microalgal strains is a current and challenging issue. It is well known that protein content and pattern in microalgae are highly influenced by culture conditions, growth phases and morphological and structural characteristics of each species (Meijer and Wijffels, 1998;Lourenço et al. 2004;Tran et al. 2009;Laurens et al. 2014). Consequently, a direct comparison of protein content in different microalgae is very difficult. Furthermore, several methods are employed for protein extraction and quantification; thus, especially when comparing data from literature, this makes the comparison between species even more complex (Meijer and Wijffels, 1998; Barbarino and Lourenço 2005;Ursu et al. 2014).
Taking these issues into account, this study analyses the protein content and pattern of four microalgae species belonging to the same Phylum (Chlorophyta) and cultivated under identical culture conditions. Chlorophyta include most of the described species of green algae which are characterised by great morphological diversity and by the ability to grow in a wide range of environmental conditions (Leliaert et al. 2012). Due to these remarkable peculiarities and on the basis of their phylogenetic position, four Chlorophyta species were selected: Chlorella vulgaris and Chlorella protothecoides (both Trebouxiophyceae), and Neochloris oleoabundans and Scenedesmus acutus (both Chlorophyceae) (Fig. 1). Although a 2-to-2 comparison based on four species cannot provide totally comprehensive information, it is nevertheless useful for improving knowledge about Chlorophyta and highlights possible differences between the two classes with the highest biotechnological potential, Trebouxiophyceae and Chlorophyceae (Garrido-Cardenas et al. 2018). Modern molecular studies state that algae belonging to these two classes, even if share similar morphological shape, are characterised by a remarkable diversity of physiological and biochemical constraints, since they have evolved in independent phylogenetic lineages because of convergent evolution within Chlorophyta (Fang et al. 2017). Among the Trebouxiophyceae, C. vulgaris is one of the best-studied microalgae and is considered a model organism in plant science (Krienitz et al. 2015). Due to its high adaptability to a variety of environmental conditions and high productivity in large-scale cultivation systems, this strain is widely exploited for several biotechnological applications including food, feed, cosmetics and pharmaceuticals production, wastewater treatments or as renewable energy source (Khan et al. 2018;Camacho et al. 2019). Moreover, the European Food Safety Authority (EFSA) considers C. vulgaris a functional safe food due to its high protein, vitamin and mineral content (Chacón-Lee and González-Mariño, 2010). Chlorella protothecoides (now named Auxenochlorella protothecoides) has been chosen due to its capability to produce large quantities of lipids, suitable for biodiesel production, and PUFAs, employed in the food industry as supplements (Heredia-Arroyo et al. 2010;Patel et al. 2018). Moreover, based on the availability of its genome, this species could become a model oleaginous microalga, also suitable for the application of genetic engineering tools (Gao et al. 2014;Wu et al. 2015). Among Chlorophyceae, Neochloris oleoabundans (now named Ettlia oleoabundans) is an attractive candidate for biodiesel production because of its high capability to accumulate lipids under stress or mixotrophic conditions (Pruvost et al. 2009;Baldisserotto et al. 2016). Moreover, the bioactivity of extracellular polysaccharides released by this alga has been recently ascertained (Li et al. 2020). Finally, the green alga Scenedesmus acutus (now named Tetradesmus obliquus) was selected for its ability to produce large amounts of proteins and it is considered a good candidate to be source of essential and non-essential amino acids (da Silva et al. 2020). Besides, S. acutus is used as a model organism to produce lipids for biodiesel and other industrially important co-products, such as β-carotene, omega-3 fatty acids, glycerol and bioethanol (Damiani et al. 2014;Patnaik and Mallick 2015). Overall, there are many publications on lipid production by the four Chlorophyta listed above, while relatively less information is available on their content of proteins, especially in relation to photosynthetic proteins (for example, thylakoid proteins) or photosynthesis in general, which supports the primary metabolisms in these unicellular microorganisms.
In the present work, the four selected Chlorophyta species were cultivated at a laboratory scale under identical culture conditions to obtain comparative results on their growth performances and protein content and profile, which could be propaedeutic in view of large-scale cultivations for industry. Furthermore, photosynthetic pigment content, photosystem II (PSII) maximum quantum yield (F V /F M ratio) and the organization and assembly of the resolved thylakoid complexes were analysed, since the photosynthetic mechanism, which these parameters provide an indication of, directly influences algae growth and consequently biomass production. For the latter issue, thylakoid proteins were analysed by sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE) and Blue Native (BN)-PAGE in second dimension (2D BN/SDS-PAGE) electrophoresis. Indeed, research on the organization and assembly of the photosynthetic proteins in thylakoids of green microalgae still needs progress. Studies on this field have mainly concerned the model green alga Chlamydomonas reinhardtii (Nama et al. 2015(Nama et al. , 2019Devadsu et al. 2021), while only a few studies have examined other species of green algae, like S. obliquus, Botryococcus braunii or N. oleoabundans under mixotrophic conditions (Kantzilakis et al. 2007;Giovanardi et al. 2017;van den Berg et al. 2020). On the other hand, exhaustive knowledge of the photosynthetic machinery represents a valid tool for a better exploitation of photosynthetic organisms. In this regard, it is supposed that high biomass and protein production by plant organisms, including microalgae, are directly linked to the whole photosynthetic process, also involving aspects on the thylakoid membrane architecture and stability.
For morpho-physiological analyses and photosynthetic pigment quantification, aliquots of algae were periodically collected during an 18-day-long growth period. For protein analyses (both total and thylakoid proteins), aliquots of samples were harvested when the microalgal cultures reached the stationary phase of growth (18th day of cultivation). Experiments were performed at least in triplicate.

Growth evaluations: cell density, optical density and dry biomass
Growth evaluations were carried out on algal samples at different times of cultivation during 18-day-long experiments. Cell density was evaluated using a Thoma haemocytometer (HBG, Giessen, Germany) and plotted on a logarithmic scale to obtain growth kinetics. Cell density was also employed to estimate the specific growth rates during the exponential phase (μ, day −1 ) (Giovanardi et al. 2013;Baldisserotto et al. 2020). Parallelly, growth of samples was also estimated by measuring the OD 750 .
For dry biomass determination (DW of algal biomass per litre; g DW L −1 ), aliquots of samples were filtered through pre-dried and pre-weighed GF/F glass-fibre filters (0.7-µm pore diameter; Whatman). Filters with cell pellets were rinsed with distilled water, dried for 72 h at 60 °C to constant weight and weighed.
During the growth period, total pH of culture media was periodically monitored using a bench pH meter.

Photosynthetic pigment extraction and quantification
For all algae species, extractions of photosynthetic pigments were performed according to Baldisserotto and coworkers (2014) with absolute methanol for 10 min at 80 °C. All manipulations were performed under a dim-green light to avoid photodegradation. Pigment concentrations were evaluated according to equations reported in Wellburn (1994) and expressed on a cell basis (nmol PIG 10 −6 cells), by counting cells with the Thoma's chamber described above, and/or percentage of algal dry weight (%DW), by dividing pigment content by biomass concentration.

Maximum quantum yield of PSII measurements
In order to determine the in vivo chlorophyll fluorescence of photosystem II (PSII), an ADC OS1-FL Pulse Amplitude Modulated (PAM) fluorometer (ADC Bioscientific Ltd., UK) was used with the following settings: measuring light (ML), 655 nm; saturation pulse (SP), 0.6 s of white light (350-690 nm) at 15,000 µmol photons m −2 s −1 . For the measurements of the maximum yield of PSII, samples were dark-adapted for 15 min and prepared according to that reported in Ferroni et al. (2018). The PSII maximum quantum yield was reported as F V /F M ratio, where F V is the variable fluorescence (F M − F O ), F M is the maximum fluorescence measured flashing the samples with a saturating light pulse and F O is the original, basal fluorescence of samples in the dark (Lichtenthaler et al. 2005;Kalaji et al. 2014).

Total protein extraction and quantification
The total protein content was evaluated by extracting two different fractions of soluble proteins: an easily extractable fraction (fraction 1, F1) and a less-easily extractable protein fraction (fraction 2, F2). Fraction 2 contains less soluble proteins, which include also transmembrane proteins. In detail, aliquots of algal samples were centrifuged for 10 min at 500 × g and treated according to Baldisserotto et al. (2016) (initial set up based on Ivleva and Golden 2007), with some modifications. Pellets were resuspended in 2 mL of soluble proteins buffer (soluble-PB: 2 mM Na 2 EDTA, 5 mM ε-aminocaproic acid, 5 mM MgCl 2 , 5 mM dithiothreitol dissolved in PBS buffer 1x; PBS buffer (stock solution 10x): 80 g NaCl, 2 g KCl, 14.4 g Na 2 HPO 4 × 2H 2 O, 2.4 g KH 2 PO 4 dissolved in 1L of distilled water), transferred into Eppendorf tubes and then centrifuged (10 min, 2000 × g) (washing step). Subsequently, pellets were resuspended in 200 μL of soluble-PB. For three times, samples were frozen in liquid N 2 for 2 min and subsequently heated at 80 °C for other 2 min, then rapidly frozen in liquid N 2 and kept at − 20 °C overnight. The following day, samples were added with glass beads (0.40-0.60-μm diameter; Sartorius, Germany) and vigorously vortexed for 10 min (mixing cycles of 30 s followed by cooling on ice for 30 s). After the addition of 100 μL of soluble-PB, samples were centrifugated (10 min, 1500 × g) and the supernatants (F1 fraction with easily extractable, soluble proteins extract) were harvested, rapidly frozen in liquid N 2 and kept at − 20 °C until analyses. The remaining pellets were re-extracted with 1 mL of less-soluble proteins buffer (less-soluble PB: 0.1 M NaOH, 1% sodium dodecyl sulphate, 0.5% β-mercaptoethanol dissolved in distilled water), by vortexing tubes for 2 min and subsequently boiling them at 60 °C for 15 min. After centrifugation (10 min, 1500 × g), supernatants were harvested (I extract), while pellets were re-extracted in 0.5 mL of lesssoluble PB as described above. Samples were then centrifuged (10 min, 1500 g) and the supernatants (II extract) were added to the I extract obtaining a fraction with lesssoluble proteins (F2), which was rapidly frozen in liquid N 2 , and kept at − 20 °C until analyses. The quantitative estimation of proteins, expressed in terms of percentage of algal dry weight (%DW) and of content per algal culture volume (g DW L −1 ), was determined with a modified Lowry's method, suitable for membrane protein determination (Markwell et al. 1981). Bovine serum albumin (BSA) was used as the standard (Sigma, USA).

Isolation of thylakoid membranes
Thylakoid membranes of microalgal samples were isolated according to Giovanardi and coworkers (2017), based on a slightly modified protocol reported by Järvi et al. (2011). For extraction, after 1 h of dark incubation, an aliquot of 300 mL of algal cultures in the stationary phase of growth (18 days of cultivation) was harvested by centrifugation at 600 × g for 10 min at 4 °C. Pellets were transferred into an ice-cold mortar containing sand quartz and cells were grinded in the presence of liquid N 2 . Subsequently, the lysate was resuspended in a grinding P1 buffer (330 mM sorbitol, 50 mM Tricine-KOH pH 7.5, 2 mM Na 2 EDTA pH 8.0, 1 mM MgCl 2 , 5 mM ascorbate, 0.05% bovine serum albumin, 10 mM NaF) and moved into 15-mL tubes. Samples were centrifuged at 300 × g for 5 min at 4 °C and then at 700 × g for 5 min at the same temperature, in order to remove sand quartz and cell debris. Pellets were discarded, while the thylakoids contained in the supernatant were collected by centrifugation at 7000 × g for 10 min at 4 °C. After that, the supernatant was discarded, and thylakoids were resuspended in 1 mL of P2 shock buffer (5 mM sorbitol, 50 mM Tricine-KOH pH 7.5, 2 mM Na 2 EDTA, 5 mM MgCl 2 , 10 mM NaF) and centrifuged at 7000 × g for 10 min at 4 °C. Then, the supernatant was removed and around 100 μL of P3 storage buffer (100 mM sorbitol, 50 mM Tricine-KOH pH 7.5, 2 mM Na 2 EDTA pH 8.0, 5 mM MgCl 2 , 10 mM NaF) was added to the pellet. Thylakoid samples were rapidly frozen in liquid N 2 and stored at − 80 °C until further analyses. Manipulation of samples was always performed on ice and in very dim safe light. Quantification of chlorophylls (Chls) in thylakoid samples was performed according to Porra et al. (1989).

Denaturing gel electrophoresis
The two total protein fractions (i.e. F1 and F2) and the thylakoid membrane proteins were characterised by SDS-PAGE (15% acrylamide resolving gel containing 6 M urea), according to Laemmli (1970) with minor modifications. For SDS-PAGE gels, 10 or 15 μg of proteins was loaded in each lane respectively, whereas, for thylakoid membrane proteins, samples were loaded on equal Chls basis (2 μg in each lane). After electrophoresis, proteins were visualised by Coomassie staining (0.1% Coomassie Brilliant Blue R250, 7% acetic acid, 40% MeOH in distilled water) following routine protocols. ImageJ software (Version 1.52e, National Institutes of Health, Bethesda, USA) was used for quantitative densitometric analysis of gel bands. Main thylakoid protein component (PSI, CP43, CP47, D2, D1 and the Light-Harvesting Complex of PSII, LHCII) bands were identified on the basis of the molecular weight. The quantitative analysis of the electrophoretic gels was performed using ImageJ software (imagej.nih.gov), following the manufacturer's instructions. In particular, it was used to generate lane profile plots and to quantify the band intensities. To this purpose, a horizontal rectangular selection of each band of interest was done across four neighbouring lanes, and the background due to Coomassie staining was subtracted before measuring the band intensity.

Data treatment
Data were processed with GraphPad Prism 9 (Graph Pad Software, USA). In each case, means ± standard deviations for n number of samples are given. The statistical significance of differences was determined by one-way ANOVA followed by a multiple comparison test (Tukey's test). A significance level of 95% (p < 0.05) was accepted.

Evaluation of growth
The four green algae, C. protothecoides, C. vulgaris, S. acutus and N. oleoabundans, cultivated under identical culture conditions, showed similar growth kinetics ( Fig. 2a and its insert). Despite cultures started from different cell densities as the obvious consequence of algae inoculation, which was set up on an OD 750 basis, all samples promptly entered the exponential phase and grew with very similar growth rates, ranging between 0.6 and 0.7 day −1 (time interval for all cultures, 0-5 days; Fig. 2a with its insert, and 2b). After this period, all cultures underwent a 5-7-day-long late exponential phase, and then entered the stationary phase of growth, reaching different, but not significant, final cell densities, with higher values for N. oleoabundans (about 14 × 10 6 cells mL −1 ) as compared to the other cultures (about 7-9 × 10 6 cells mL −1 ; ANOVA, p > 0.05) (Fig. 2a). These different final cell densities did not coincide with different OD 750 values (Fig. 2a, insert; ANOVA, p > 0.05). Moreover, despite C. protothecoides, C. vulgaris and S. acutus reached lower cell densities than those observed for N. oleoabundans, they yielded higher biomass (61, 32 and 50% higher than that of N. oleoabundans, respectively) (Fig. 2c). The pH of culture media showed the following differences between samples: (1) C. vulgaris, S. acutus and N. oleoabundans gave pH curves almost perfectly overlapping the shape of growth curves, with Neochloris maintaining lower values of pH with respect to the other two algae; (2) at 10 days of cultivation, C. protothecoides showed one pH peak, which was followed by a reduction to stable values around 10 during the following days of cultivation (Fig. 2d). Overall, the highest pH values were recorded for S. acutus cultures at 14-18 days of cultivation.

Photosynthetic pigment content
On the whole, concentration of photosynthetic pigments inside cells of all samples increased with cultivation time, as expected for growing algae cultures due to self-shading, and reached different final concentrations (Fig. 3a-d).
Interestingly, despite the general increasing trend, S. acutus showed an initial evident reduction in pigment concentration in the time interval 0-5 days of cultivation. Regardless of this initial and temporary decline, this Trebouxiophycea showed the highest Chla concentration as compared to the other species (1.3, 1.7 up to 2.5 times higher values than in C. protothecoides, C. vulgaris and N. oleoabundans, respectively, at time 18 days) (Fig. 3a). Differently, the highest Chlb content was observed in both S. acutus and C. protothecoides (Fig. 3b). Overall, S. acutus showed the highest total Chls (Chl TOT ) content as compared to the other species (ca. 1.2, 1.6 and 2.3 times higher than in C. protothecoides, C. vulgaris and N. oleoabundans, respectively) (Fig. 3c). Similar to that observed for Chla and Chl TOT content, S. acutus also contained the highest Car concentrations as compared to the other strains . ANOVA, p < 0.05. Different superscripts denote significant (p < 0.05) differences between samples (ca. 1.3-1.4 times higher than in C. protothecoides, C. vulgaris, and ca. 2.4 times higher than in N. oleoabundans) (Fig. 3c, d). The general increase in pigment content resulted in relatively stable and species-specific photosynthetic pigment molar ratios (Fig. 3d, e). The Chla/b ratio was generally higher in S. acutus (ca. 3) as compared to the other species (ca. 2.5) (Fig. 3e). In parallel, during the entire cultivation time, S. acutus showed almost a constant Chl TOT /Car ratio (ca. 6), while in the other species, the same molar ratio was subjected to an evident increase until the 5th day. Subsequently (5-18 days), in C. protothecoides and in N. oleoabundans, Chl TOT /Car ratio tended to settle at similar values (about 5.5-5.8), also similar to those of S. acutus, while in C. vulgaris, it gradually decreased down to about 4.5 (ANOVA, p < 0.05) (Fig. 3f).
At the end of the cultivation time, when also total and thylakoid proteins were analysed, the pigment content was also evaluated as percentage of algal dry biomass (%DW; Fig. 3g, h) to give a more exhaustive datum for biotechnological applications of the algal biomass. Chl TOT contents ranged between 3.8%DW, as the lowest value, for N. oleoabundans, and 5.8%DW, the highest one, for C. vulgaris and S. acutus; C. protothecoides showed an intermediate value (4.9%DW) (Fig. 3g). Differently, the Car content showed very evident differences among samples (Fig. 3h). In detail, C. vulgaris reached the lowest Car content (0.10%DW) as compared to the other species (ca. 3.5, 4 and 7 times lower than values obtained from C. protothecoides, N. oleoabundans and S. acutus, respectively) (Fig. 3h).

Maximum quantum yield of PSII
The maximum quantum yield of PSII (F V /F M ratio) was monitored during the experiment to compare the photosynthetic efficiency of each species (Fig. 4). No significant differences were reported among samples at 5, 7 and 10 days of cultivation (Fig. 4). In detail, C. protothecoides tended to give higher values of F V /F M as compared to the other algae, but the values differed significantly only at time 0, 14 and 18 days (ANOVA, p < 0.05). In particular, at the 14th day of cultivation, while C. protothecoides was characterised by a F V /F M of about 0.78, C. vulgaris and S. acutus showed F V / F M values around 0.74-0.75, and N. oleoabundans gave the lowest one, ca. 0.71 (ANOVA, p < 0.05). At the end of the experiment, C. protothecoides still showed the highest values of the PSII maximum quantum yield (0.78) as compared to all other samples (0.71-0.72; ANOVA, p < 0.05). Despite the abovementioned differences, overall results showed that all selected species had F V /F M values that, during the cultivation time, varied in a short range of values from 0.65 to 0.75, not attributable to stressed cultures (Fig. 4).

Total protein content and pattern
In present research, due to the different morphology and cell characteristics of the four examined microalgae (not shown), preliminary tests using various protocols for protein extraction were performed to find the more suitable and employable one for all the four Chlorophyta (for details, see: Meijer and Wijffels 1998;Ivleva and Golden, 2007;Pruvost et al. 2011;Popovich et al. 2012;Serive et al. 2012;Baldisserotto et al. 2016). Based on the results obtained from different extraction methods (data not shown), the procedure proposed in the present work allowed obtaining a successful extraction of total proteins from all samples, independent of their morphological and biochemical characteristics.
In this study, a fraction containing easily extractable soluble proteins (F1) and a fraction with less-easily extractable ones (F2), including transmembrane proteins, were extracted from the four Chlorophyta species.
As shown in Table 1, the four green microalgae showed some differences in the final total protein content. In detail, S. acutus showed the highest value of total protein content (53.2%DW) as compared to the other species (47, 9 and 8% higher than in C. protothecoides, C. vulgaris and N. oleoabundans, respectively). Scenedesmus acutus was also characterised by the highest total protein content in the culture (13.3 g L −1 ) ( Table 1).
Both fractions were also screened with SDS-PAGE analyses to obtain a qualitative comparison of protein extracts for the four Chlorophyta. Coomassie-stained gels revealed that the number and intensity of protein bands were different among the four species, independent of the fraction (i.e. with more or less easily extractable soluble proteins). Moreover, the F2 fraction gave the best results in terms of quality and resolution of protein bands as compared to the F1 fraction (Fig. 5). In all samples, the extracted proteins were resolved into distinct bands, mainly from 76 to 12.3 kDa. Based on literature data, the distinct band with a molecular mass of about 55-56 kDa was assignable to the ribulose-1,5-bisphosphate carboxylase/oxygenase (RuBisCO) large subunit (Spreitzer 1993;Park et al. 1999) and the protein bands below 30 kDa mainly corresponded to subunits of the LHCII (Bennett 1991) (Fig. 5). Interestingly, the pattern of the easily extractable soluble protein fraction (F1), which represented only a very minor component in all algae cultures (Table 1) and appeared poorly defined, showed differences in bands among all algae (Fig. 5a). However, the main results derived from the SDS-PAGE analysis of F2 fraction, with the less easily extractable soluble proteins (Fig. 5b). In this sample, the protein pattern was overall similar within the same algal class. The two Chlorophyceae (C. protothecoides and C. vulgaris) showed similar protein profile each other; the same observation was applicable to the two Trebouxiophyceae (N. oleoabundans and S. acutus) as well. Indeed, in algae from both classes, resolution of LHCII was peculiar (Fig. 5b). In C. protothecoides and C. vulgaris, the LHCII was mostly represented by one evident band, while it was resolved into two bands with similar intensity in the other two algae.

Thylakoid protein complexes
The SDS-PAGE and 2D BN/SDS-PAGE analyses were performed to obtain information on thylakoid protein complexes and their native organization inside the photosynthetic membrane in the four Chlorophyta (Figs. 6 and 8). The assignment of major bands belonging to abundant thylakoid proteins conserved over the entire green lineage was based on approximate molecular mass (Minagawa and Takahashi 2004;Dekker and Boekema 2005); furthermore, it is consistent with relatively recent thylakoid protein profiles and immunodetections in N. oleoabundans (Giovanardi et al. 2017). On the basis of SDS-PAGE analysis, the densitometric profile was obtained for each species (Fig. 6a,  b). In each lane, the first conspicuous band corresponded to the major photosystem I (PSI) polypeptides PsaA/PsaB (around 62 kDa) (Fig. 6a, b). As regards PSII, the band around 43 kDa was assignable to its light-harvesting protein CP43, and the two homologous proteins of nearly 30 kDa corresponded to D1 and D2 subunits of the reaction centre of PSII core. Finally, the LHCII was resolved in four or five separated bands below 30 kDa. In general, the two Trebouxiophyceae, C. vulgaris and C. protothecoides, shared a very similar thylakoid membrane protein pattern as compared to the other two species, which instead showed differences mainly in the bands corresponding to the LHCII proteins (Fig. 6). For semi-quantitative comparison between the four green algae, the relative amount (band volume ratio) of different photosynthetic proteins in the thylakoid complexes was quantified. The band volumes were used to calculate band volume ratios (Fig. 7). The LHCII/D2 ratio, which reflects the abundance of LHCII as compared to PSII cores, was comparable within the same classes of Chlorophyta: it was lower in the two Trebouxiophyceae than in the two Chlorophyceae, with values around 6 and 8, respectively (Fig. 7a). The PSI/D2 ratio, which provides information about stoichiometric variations involving the two photosystems, was markedly lower in S. acutus as compared to all other species (− 40 to − 45%; ANOVA, p < 0.05) (Fig. 7b). Finally, the LHCII/PSI band volume ratio, which reflects the potential LHCII availability also for PSI, was subjected to species-specific variability (Fig. 7c). The two Chlorophyceae, N. oleoabundans and S. acutus, showed values of about 4 and 2.2, respectively, while the two Trebouxiophyceae, C. vulgaris and C. protothecoides, recorded values below 2. Due to the abovementioned differences in availability of LHCII for the photosystems, it was evaluated if the stoichiometry of photosynthetic proteins had an impact on native interactions; thus, thylakoid membranes were mildly solubilised using dodecyl β-D-maltoside and separated by BN-PAGE.
All the four green microalgae exhibited the presence of four distinct bands, ordered from I to IV, and a heavier and less defined band, called megacomplex (MC), on the uppermost part of the gel (Fig. 6c). The latter MC band was particularly evident in S. acutus (Fig. 6c). BN-PAGE profile appeared different between the two Chlorophyceae, while it was very similar for the two Trebouxiophyceae (Fig. 6c). Identification of complexes of the Chlorophyceae N. oleoabundans was done based on a published BN-PAGE profile from the same alga and supported by 2D BN/SDS-PAGE pattern (Giovanardi et al. 2017) (Fig. 8a). In detail, the lowest band (I) was attributed to LHCII monomers, band II to LHCII trimers, band III to PSII monomers and band IV to supercomplexes of PSI and LHCs (Fig. 6c). In N. oleoabundans, the LHCII was mainly found in monomeric form (Figs. 6c and 8a). In the 2D gel, the PSII was revealed by its subunits CP47, CP43, D2 and D1 (Fig. 8a). PSII was found almost completely as a monomer in band III (Fig. 8a). However, as already observed by Giovanardi et al (2017), a very faint amount of PSII was present as a dimer co-migrating with PSI (band III I in Fig. 8a), slightly lighter than the PSI-LHCI complex. Some faint PSII components also co-migrated with PSI in band IV: these are most probably ascribed to PSII dimers associated with one LHCII trimer (as shown in Giovanardi et al. 2017, band IV has approximately the same weight of C 2 S complex of Arabidopsis). The high molecular weight region including MC actually was comprised of a series of complexes formed by PSI and/or PSII with their antennae. The lightest components of the series were probably assignable to co-migrating individual supercomplexes of PSI or PSII. Conversely, the heaviest components could include both photosystems in native megacomplexes. A clear spot aligned with PSI-LHCI at about 130 kDa was also resolved from bands IV and, to a lesser extent, MC (Fig. 8a). Because of its weight and alignment with PSI, it corresponded to PsaA/PsaB heterodimer, as also observed Values are means ± s.d. (n = 3). ANOVA, p < 0.05. Different superscripts denote significant (p < 0.05) differences between the samples ◂ in Chlamydomonas sp. UWO (Nelson and Ben-Shem 2004;Szyszka-Mroz et al. 2015). Finally, on the right side of the 2D BN/SDS-PAGE, large quantities of free proteins with different molecular weights were evident (Fig. 8a).
Overall, the 2D BN/SDS-PAGE profiles highlighted some evident differences comparing the samples (Fig. 8).
In the other Chlorophycea, S. acutus, the presence of PSII subunits was more included in the MC band, which was much more represented than in the other algae (Figs. 6c and 8b). In particular, the heaviest complex was very well resolved with its components of PSI, PSII and LHCs. The PSI-LHCI band in the 2D gel was mainly formed by PsaA/PsaB heterodimer (Fig. 8b). LHCII was almost equally distributed between the monomeric and the trimeric forms (Fig. 8b). Solubilised stripes from S. acutus thylakoids produced a large amount of free single proteins at the right side of the SDS-PAGE gel in second dimension likewise in N. oleoabundans (Fig. 8b).
The two Trebouxiophyceae shared a very similar BN-PAGE profile (Fig. 6c); as a reference sample, C. vulgaris stripe was solubilised and run in 2D for native band analysis (Fig. 8c). As compared to N. oleoabundans samples, in C. vulgaris, LHCII was more represented in the trimeric form than in the monomeric one, PSII was similarly represented mostly by monomers, while co-migrating PSI with PSII dimers (from a III I band) and MC were not abundant (Fig. 8c). PsaA/PsaB heterodimer of 130 kDa and free proteins on the right side of the gel also characterised the 2D profile in this alga (Fig. 8c).

Discussion
Modern molecular studies have revealed that the green algal classes Chlorophyceae and Trebouxiophyceae, which are characterised by similar morphological shape but also by a remarkable diversity of physiological and biochemical constraints, evolved in independent phylogenetic lineages as the result of convergent evolution within the Chlorophyta Phylum (Leliaert et al. 2012;Krienitz et al. 2015). As highlighted by the present research study, the specific variability (and similarities) of each strain clearly emerges comparing not only growth but also the biochemical profile and characteristics of the examined algae.
The monitoring of microalgal growth and the knowledge of biological characteristics of these micro-organisms are fundamental issues to improve the performance of cultures in biotechnological fields (Havlik et al. 2013;Maglie et al. 2021). Growth, photosynthetic pigment and protein content represent species-specific aspects of microalgae but are influenced by culture conditions. Thus, our comparative study on four Chlorophyta species cultivated in the same culture conditions was helpful to better highlight biological differences among those species. However, it should be underlined that the present study was set up employing laboratory culture conditions; for an industrial mass cultivation Values are represented as means ± s.d. (n = 3). ANOVA, p < 0.05. Different superscripts denote significant (p < 0.05) differences between the samples within each day of cultivation Table 1 Protein content, expressed as percentage of dry weight biomass (%DW) and gram per litre (g L −1 ), at the 18th day of cultivation in C. protothecoides, C. vulgaris, N. oleoabundans and S. acutus samples. Values are represented as means ± s.d. (n = 3). ANOVA, p < 0.05 Different superscripts denote significant (p < 0.05) differences between the means in columns of each protein fraction

Strains
Easily extractable soluble proteins, F1 Less-easily extractable soluble proteins, F2 Total proteins Chlorella protothecoides 3.4 a ± 0.4 0.91 a ± 0.13 32.7 a ± 4.9 8.7 a ± 0.3 36.1 a ± 5.2 9.6 b ± 0.5 Chlorella vulgaris 1.8 b ± 0.5 0.41 b ± 0.08 46.8 ab ± 5.9 10.5 acd ± 1.9 48.6 ab ± 6.5 10.9 ab ± 3.7 Neochloris oleoabundans 3.8 a ± 0.6 0.63 ab ± 0.13 45.7 ab ± 7.8 7.6 bd ± 0.55 49.5 ab ± 8.1 8.2 b ± 0.6 Scenedesmus acutus 1.1 b ± 0.1 0.26 c ± 0.04 52.1 b ± 4.2 13.0 c ± 0.9 53.2 b ± 4.3 13.3 a ± 0.1 of these algae, specific scale up tests are needed to adjust cultivation settings and systems. In this perspective, our work gives useful basic biological information of the tested algae in view of future massive productions. In biotechnological applications of microalgae, growth is usually quantified by measuring the optical density (OD at different wavelength: usually 750 or 680 nm) or the biomass concentration as DW per litre of algal culture; direct cell counts at the microscope are less common (Moheimani 2013). In addition, the monitoring of pH values of culture media is an indirect way to monitor algal growth. In this regard, the pH of culture media is attributed to the speciesspecific assimilation and release of CO 2 during photosynthesis and respiration, respectively, and is correlated to the growth of algae (Khalil et al. 2010;Moheimani 2013). OD is frequently used as a rapid and non-destructive indirect measurement of biomass or cell density, but its use to indirectly estimate cell density or biomass can bring artefacts, since the OD of living cells varies during growth phases and among treatments. DW determination is a convenient, but time-and material-consuming method for growth evaluations. Conversely, the direct count of cells is time consuming and needs expertise by the operator. OD and DW methods generate mistakes in the growth estimation being closely dependent on culture conditions, different growth phases and morphological state of the cells, namely cell size and shape (Griffiths et al. 2011;Baldisserotto et al. 2016). So, for biological studies, it is appropriate to integrate OD and In a, on each lane, 2 μg of Chl was loaded. Molec-ular weight (kDa) marker is reported on the left side of the gel. In c, purified thylakoids were solubilised with 1.5% (v/v) dodecyl β-Dmaltoside, loaded on an equal chlorophyll (8 μg) basis and separated on 5-12.5% gradient native gel. The position of major complexes is indicated by labels on the left side of the gel DW with cell counts. This is clearly confirmed in this study, where all parameters have been analysed for growth estimation of algae. The four selected green algae showed similar growth kinetics and rates. Values of the latter parameter are close to the results reported for some other Chlorophyta species cultivated at irradiance similar to those used in this study (Singh and Singh 2015). However, by comparing OD, cell density and DW at the end of the experiment, it emerges that these parameters are not directly correlated with each other, being, for example, the same OD produced by different cell densities and biomass. This discrepancy descends from different morphological characteristics of the microalgae (example, cell size and shape) (Leliaert et al. 2012), but can have important implications when studying microalgae, especially in view of the potential application of their biomass, and when comparing literature data on growth. Moreover, in all cases, the pH value of culture media during cultivation displayed a diverse pattern among samples. According to previous studies (Moheimani 2013;Difusa et al. 2015;Sabia et al. 2015), under unregulated pH conditions, microalgae showed an overall increase in the pH of the medium with time, which is attributed to the consumption of CO 2 during the photosynthetic process and to protons uptake (Villarejo et al. 1995;Zerveas et al. 2021). For a better study of algal growth, the PSII maximum quantum yield was also measured. The photochemical activity of PSII is widely used to estimate the physiological state of microalgae, being very sensitive to stressful conditions (White et al. 2011;Schuurmans et al. 2015), and, therefore, is also employed to follow and optimise microalgae growth performance (Schuurmans et al. 2015). The results stated that the selected four Chlorophyta maintained a nearly invariable and optimal PSII photochemistry throughout the experiment, thus excluding the occurrence of major stresses (White et al. 2011). So, cultures were further analysed in relation to their biotechnological impact and biological characteristics.
For the exploitation of microalgae especially in the food/ feed sectors, besides growth, it is important to quantify the added value of biomass. Under autotrophic conditions, the more abundant molecules accumulated inside algae are photosynthetic pigments and proteins. Chlorophylls and carotenoids characterise all photosynthetic organisms, while RuBisCO is the major enzymatic protein in photosynthetic plant cells (including microalgal cells) and LHCII is the most abundant membrane protein on Earth (Crepin and Caffarri 2018). Both these classes of molecules have a double importance in a biotechnological perspective, since they represent a high percentage of the whole biomass Fig. 7 Semi-quantitative analyses of (a) LHCII/D2, (b) PSI/D2 and (c) LHCII/PSI band intensities (band volume ratio) in thylakoid membranes of C. protothecoides (CP; white), C. vulgaris (CV; light grey), N. oleoabundans (NO; dark grey) and S. acutus (SA; black). Values are represented as means ± s.d. (n = 3). ANOVA, p < 0.05. Different superscripts denote significant (p < 0.05) differences between the samples ▸ (contributing to improve the nutritional and antioxidant properties of food, feed and supplements) and participate in photosynthesis (Crepin and Caffarri 2018;Di Stefano et al. 2018;Amorim et al. 2021). The pigment composition of photosynthetic microorganisms is linked to the phylogenetic position of the organism and to environmental constraints (Deblois et al. 2013). Differences in photosynthetic pigments among the tested Chlorophyta were mainly attributed to phylogenetic position since algae were cultivated under the same culture conditions. In detail, the specific variability of each strain clearly emerged comparing the time-course analysis of the photosynthetic pigment content on a cell basis, suggesting that each species developed a kind of a unique metabolic framework, certainly determined by their phylogenetic characteristics, together with their specific morpho-physiological state (Deblois et al. 2013). Additionally, differences in pigment content could also relate to the adaptive evolution to the original sampling locations, as especially inferred by the conspicuous differences observed between S. acutus and N. oleoabundans, which belong to the same phylogenetic branch. Indeed, N. oleoabundans was originally isolated from sand samples of the Saudi Arabian desert, while S. acutus from wastewater samples harvested in Northern Italy (Chantanachat and Bold 1962; https:// ccala. butbn. cas. cz/ en). Interestingly, from the biotechnological point of view, S. acutus produced the biomass with the highest content in Car, which can be oriented to various industrial applications. It is important to consider new sources of natural Car, because the global market of these pigments was projected to reach about 1.53 billion US$ in 2021 and considering that the use of synthetic Car has started to decline due to potential toxic effects of synthetic molecules (Ambati et al. 2019).
Parallel to photosynthetic pigments, the protein content and profile of the four Chlorophyta species were evaluated to highlight differences among samples. It is well known that a comparison of protein content in different microalgae is very difficult owing to the morphological and biochemical characteristics of each microalgae species but also to the several methods employed for protein extraction and quantification (Meijer and Wijffels 1998;Barbarino and Lourenço 2005;Safi et al. 2013;Ursu et al. 2014). As a preliminary step, several protein extraction methods were tested. The protein extraction procedure originally proposed by Ivleva and Golden (2007) was modified and allowed a better comparison of protein content. According to Buono et al (2014), disruption of the cell by vortexing with glass beads in the presence of SDS was an essential and effective extraction step to lyse the cell walls. The values of protein content here reported are consistent with literature data for green microalgae (Barka and Blecker 2016). Interestingly, S. acutus not only showed the highest value of protein content (53.2%DW) but also the highest total protein accumulation in the final culture (13.3 g L −1 ) as compared to the other samples. The BN-PAGE strips were treated with 10% (v/v) SDS and 5% (v/v) β-mercaptoethanol and denatured subunits were separated using SDS-PAGE. Second dimension gels were silver stained. Marker molecular weight (kDa) is indicated on the right side of the gels The SDS-PAGE analysis of total proteins showed a different pattern among algae, both considering the more and the less easily extractable fractions, F1 and F2. The main differences concerned LHCII subunits. Thus, the thylakoid membrane protein profile was studied more in detail by BN-PAGE and SDS-PAGE in first and second dimensions. The two Trebouxiophyceae shared a very similar thylakoid membrane protein pattern as compared to the other two species, which instead showed differences mainly in the bands corresponding to the LHCII proteins. To evaluate if the selected species showed significant variations in fundamental thylakoid complexes stoichiometry, the relative amount (band volume ratio) of some key proteins in the thylakoids were compared. Photosynthetic organisms regulate the stoichiometry of their photosynthetic complexes in response to variations of light intensity and spectral quality. To avoid light excess and to profitably make use of limiting light, plants and green algae have evolved different acclimation mechanisms that include both short-term and long-term responses (Anderson et al. 1995). Long-term acclimation responses generally involve changes in the stoichiometry of PSI versus PSII and the modulation of the LHCII (Humby and Durnford 2006;Wobbe et al. 2016). Among green algae, studies on acclimation mechanisms to light intensity are mainly on C. reinhardtii, which is widely explored as a photosynthetic model organism (Durnford et al. 2003;Mettler et al. 2014). However, contrasting results on acclimation behaviour have been reported in other green algae (Falkowski 1980;Finazzi and Minagawa, 2014) suggesting that acclimation mechanisms in green microalgae can be species-specific.
The PSI/D2 ratio showed that S. acutus had a markedly lower value as compared to all other species. The photosystems stoichiometry is modulated to balance the excitation between PSI and PSII; in a denser culture, such as that of S. acutus, the reciprocal shading of cells is expected to increase, leading not only to less available light intensity, but also to a relative enrichment in far-red. Since far-red is mainly absorbed by PSI, cells are expected to downregulate PSI in favour of PSII, as observed in S. acutus. More in general, the PSI/D2 ratio was negatively related to the biomass yields, indicating that this ratio reflected the acclimation to the available light quality, which in turn depends on the culture density (Büchel 2015;Wobbe et al. 2016). In the short-term, the excitation regulation of the two photosystems requires modulable associations of LHCII with PSI and PSII. The relative abundance of LHCII as compared to PSII cores was similar within the same class of Chlorophyta, but lower in the two Trebouxiophyceae than in the two Chlorophyceae. The LHCII/PSI ratio, which may reflect the availability of LHCII also for PSI (Falkowski 1980), had a species-specific variability, but with higher values for the two Chlorophyceae than for the two Trebouxiophyceae. In general, the two Chlorophyceae revealed a higher availability of LHCII antennae for both photosystems. In this paper, the functional distribution of LHCII between PSII and PSI was not determined. However, the BN/SDS-PAGE results are strongly suggestive of extensive and stable interactions with not only PSII, but also PSI, and, very probably, both photosystems at the same time. The heaviest MC, which is larger than known PSII-LHCII supercomplexes or PSI-LHCII state transition complexes, could be interpreted as an accidental aggregate (Galka et al. 2012). However, this is in contrast with ever-accumulating evidence about the existence of PSI-PSII-LHCII megacomplexes, which may mediate the interconnectivity between PSI and PSII, and play physiological roles in energy transfer between the two PSs, not necessarily linked to state transitions (Järvi et al. 2011;Grieco et al. 2015;Yokono et al. 2015Yokono et al. , 2019Ferroni et al. 2016;Giovanardi et al. 2017Giovanardi et al. , 2018Furukawa et al. 2019).
Applications of BN/SDS-PAGE technique in green microalgae have been mainly limited to C. reinhardtii and only seldom used for other microalgae, sometimes with uncertain results (Tran et al. 2009;Kantzilakis et al. 2007;Szyszka-Mroz et al. 2015;Giovanardi et al. 2017;Zhao et al. 2017;Garcìa-Cerdàn et al. 2019;van den Berg et al. 2020). However, by comparison with many published profiles of thylakoid complexes in higher plants (Kügler et al. 1997;Ferroni et al. 2014;Albanese et al. 2016), those of the tested microalgae were singularly poor of large complexes. This may depend on the harsh extraction protocol, required because of the robust cell walls of microalgae. Nevertheless, based especially on C. reinhardtii, the complexity of native thylakoid components in green algae can be as high as in land plants (Zhao et al. 2017;Umetani et al. 2018;Garcìa-Cerdàn et al. 2019;Kubota-Kawai et al. 2019;Nama et al. 2019;Calvaruso et al. 2020;van den Berg et al. 2020), different to that assumed or reported in previous years (Dekker and Boekema 2005;Kouřil et al. 2012). Therefore, the abundant free proteins on the right side of the gel are the unavoidable result of a harsh thylakoid isolation method. Although native interactions were not completely preserved, we highlighted some differences in the organization of the resolved thylakoid complexes of the four investigated green algae. By comparing the 2D BN/SDS-PAGE, each species showed some peculiarities in their thylakoid protein pattern. One evident difference was the prevalence of monomeric LHCII over the trimeric form in N. oleoabundans. Although it was reported that a population of monomeric LHCII exists in the photosynthetic membrane, where it acts as a dissipator of excess energy under high light conditions (Bielczynski et al. 2016), it is more realistic that very abundant monomeric forms of LHCII in green algae are caused by the isolation protocol and derive from the dissociation of trimers. Therefore, the relative abundance of monomers and trimers would reflect the recalcitrancy of microalgae to thylakoid isolation. A similar reason can explain the prevailing monomeric PSII form, instead of the dimers. It is plausible that the characteristics of the cell wall could have influenced the extraction procedure. Indeed, in N. oleoabundans, the cell wall is formed by two layers (a thin outer layer and a thick inner one) (Rashidi and Trinidade 2018;Spain et al. 2021), while for Scenedesmus and Chlorella species, the cell wall can be composed by a third one (Voigt et al. 2014;Dixon and Wilken 2018;Spain et al. 2021). Furthermore, the cell wall composition is very different in these algae (Abo-Shady et al. 1993;Rashidi and Trinidade 2018). However, the attribution of different thylakoid complexes patterns only to methodological issues is too simplistic. In particular, it does not explain the resolution of high molecular weight complexes, which should be the most labile and in Angiosperms are not even preserved when dodecyl-maltoside is used as detergent (Järvi et al. 2011).
Only minor amounts of PSII dimers were observed/preserved in C. vulgaris, S. acutus and N. oleoabundans, and were well visibly organised in MC only in S. acutus, where the lighter complexes of the MC band were very likely due to co-migration of PSII-LHCII supercomplexes and PSI-LHCI. Differently, the heaviest complexes visible on the left side of the 2D BN/SDS-PAGE gel in S. acutus corresponded to various assemblages of PSI, LHCI and PSII-LHCII, and therefore were attributed to the PSI-(PSII)-LHCII megacomplexes. The comparison with MC bands of Arabidopsis reported in Giovanardi et al. (2017) supports this interpretation, since MC bands of algae samples were heavier than that of C 2 S 2 M 2 supercomplexes found in Arabidopsis. These results agreed with SDS-PAGE analysis and further showed that the higher LHCII/PSI ratio found in S. acutus can be linked to a quite complex interaction of LHCII with PSI, as discussed above. In particular, the pattern of protein complexes was richer in S. acutus as compared to the other species, which however showed the presence of PSI-(PSII)-LHCII megacomplexes. Finally, each species was characterised by a high molecular mass protein (> 130 kDa), which was assigned to components of a PSI-associated pigmentprotein complex (Szyszka-Mroz et al. 2015).
Collectively, the results provide useful basic information for the exploitation of C. vulgaris, C. protothecoides, N. oleoadundans and S. acutus in biotechnological applications, such as the food/feed sector due to their richness in proteins. Even if only four species were studied, results have also highlighted the metabolic uniqueness of each strain, resulting in a non-self-explanatory comparison with the others, even the most related ones. It can be concluded that S. acutus is the most promising species to be used for biotechnological applications, due to its capability to produce satisfactory quantities of biomass, rich in proteins and photosynthetic pigments (mainly Car). Overall, it is shown that, in S. acutus, the major stability of the photosynthetic membranes very probably supports the productivity of a high-value biomass, confirming the importance of the organization of photosynthetic membranes to sustain the efficient growth of microalgae also in a biotechnological context. Funding Open access funding provided by Università degli Studi di Ferrara within the CRUI-CARE Agreement. This study was supported by the University of Ferrara through the Futuro In Ricerca grant (FIR-2018 granted to SP) and the Fondo di Ateneo per la Ricerca Scientifica (FAR granted to SP and CB).

Data availability
The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.

Conflict of interest The authors declare no competing interests.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.