Genome-wide identification and characterization of CKIN/SnRK gene family in Chlamydomonas reinhardtii

The SnRK (Snf1-Related protein Kinase) gene family plays an important role in energy sensing and stress-adaptive responses in plant systems. In this study, Chlamydomonas CKIN family (SnRK in Arabidopsis) was defined after a genome-wide analysis of all sequenced Chlorophytes. Twenty-two sequences were defined as plant SnRK orthologs in Chlamydomonas and classified into two subfamilies: CKIN1 and CKIN2. While CKIN1 subfamily is reduced to one conserved member and a close protein (CKIN1L), a large CKIN2 subfamily clusters both plant-like and algae specific CKIN2s. The responsiveness of these genes to abiotic stress situations was tested by RT-qPCR. Results showed that almost all elements were sensitive to osmotic stress while showing different degrees of sensibility to other abiotic stresses, as occurs in land plants, revealing their specialization and the family pleiotropy for some elements. The regulatory pathway of this family may differ from land plants since these sequences shows unique regulatory features and some of them are sensitive to ABA, despite conserved ABA receptors (PYR/PYL/RCAR) and regulatory domains are not present in this species. Core Chlorophytes and land plant showed divergent stress signalling, but SnRKs/CKINs share the same role in cell survival and stress response and adaption including the accumulation of specific biomolecules. This fact places the CKIN family as well-suited target for bioengineering-based studies in microalgae (accumulation of sugars, lipids, secondary metabolites), while promising new findings in stress biology and specially in the evolution of ABA-signalling mechanisms.

SCIEntIfIC RePoRts | (2019) 9:350 | DOI: 10.1038/s41598-018-35625-8 metabolism, gene regulation, and stress response, together with hexokinases (HXK) and Sucrose-Phosphate Phosphatase (SPP) 12,13 . This protein family is also interconnected with TOR 14 , epigenetic pathways 15 and with the activation/repression of entire metabolic branches in other organisms 16 , making it a good candidate for further studies in microalgae. The first described SnRK family member was the yeast Snf1, being well-known by its role in lipid accumulation 17 and glucose repression, regulating carbon metabolism 12 . Globally, AKIN/SnRK1/Snf1/AMPK kinases concentrate divergent stress signals by activating specific enzymes and transcription factors, related to metabolic regulation, protein biosynthesis and cell organization, as part of a complex system aimed to increase cell survival under unfavourable energetic balance 13,18 . AMPK regulates through direct phosphorylation or associated γ subunits ADP-AMP levels sensing 19 , with also associated β subunits regulating substrate specificity and cell location 20 . These kinase-regulatory subunits protein complexes are remarkably well conserved across Eukarya and deeply rooted into the life tree. Orthologs to AMPK and γ subunits have been found into Bacteria and Archaea kingdoms, being β subunits exclusive from eukaryotes 21 . Plants are an exception of this family conservation showing unique γ functionally equivalent βγ proteins and γ-like subunits without direct SnRK1 interaction 22 . This exceptionality comes also at regulation level, with a sugar phosphate mediated regulation of the energy sensing kinase 23 . Moreover, although AMPK duplication events are common in plants and animals 21 , plants SnRK family underwent an extensive duplication and diversification event, giving origin to three subfamilies: SnRK1/AKIN (the closest to Snf1), SnRK2 and SnRK3 24 . All these plant SnRK subfamilies share a common Ser/Thr kinase domain, followed by a UBA and a KA1 domain in SnRK1. In SnRK2 an osmotic stress activation domain I is present after the kinase domain, while SnRK3 contains a NAF/FISL domain 25 . There is strong evidence that subfamilies SnRK2 and 3 evolved after gene duplication of SnRK1 in order to enable plants to develop networks capable of linking stress, ABA, and calcium signalling with metabolic and epigenetic responses 13 . In plants, some SnRK2 are also key transductors in ABA-mediated responses to salt and other abiotic stresses 25 , presenting specific ABA interaction acidic motifs known as domain II 26 placed after the characteristic osmotic stress activation domain I.
In Arabidopsis thaliana, thereafter Arabidopsis, the SnRK family is composed by 38 members and 3 subfamilies: SnRK1 (3 genes), SnRK2 (10 genes), and SnRK3 (25 genes) 27 . SnRK1 has been linked to increased tolerance to nitrogen stress 28 and to energy sensing and gene regulation 14 . It showed to be implied in plant response to starvation and energy deficit by coordinating ATP/cAMP, redox status and C/N ratios to regulate broad metabolic branches, either directly by phosphorylation of enzymes or indirectly by phosphorylation of transcription factors 13,29,30 . SnRK1 has been considered a potential target to improve plant performance under unfavourable conditions 25 . SnRK2 and SnRK3 showed to have also a key role in signalling pathways that regulate plant response to nutrient limitation, drought, cold, salt, and osmotic stress 25 . The SnRK2 subfamily has an essential role in gene expression regulation through the activation of bZIP transcription factors and SWI/SNF/helicase complexes 12,31 tightly connected to epigenetic mechanisms that perfectly control gene activation or repression. However, little is known about the structure or the role of this protein kinase family in microalgae stress response mechanisms and its relation with biotechnological processes, such as the accumulation of high value-added molecules like energetic molecules (sugars and lipids) 32 or pigments (astaxanthin, lutein, and β-carotene).
The Chlorophyceae Chlamydomonas reinhardtii, thereafter Chlamydomonas, shares common ancestry with vascular plants 33 . Therefore, it is expected that many of its responses to limiting conditions would be similar 34 . Few studies suggest the involvement of SnRK family, named CKIN in Chlamydomonas 35 , in stress response, namely under sulphur 36,37 and nitrogen deprivation 5 , and cold stress 35 . Gonzalez-Ballester et al. 36 reported the existence of eight putative SnRK2-like genes in Chlamydomonas, while Valledor et al. 35 found three cold stress-responsive proteins showing sequence homology with Arabidopsis AKIN10/11 family (CKIN1, CKIN2, and CKIN3). Both authors suggested that, like in plants, Chlamydomonas abiotic stress response is mediated by CKINs. Notably, it has been recently reported by Sato et al. 32 that SAC1 and SnRK2.2 act as positive regulators of DGTT1, enhancing TAG synthesis under Sulphur starvation in Chlamydomonas. Chlamydomonas SnRKs similarity to land plants is conceivably not only limited to direct stress SnRK induction, but also ABA-SnRK interaction. ABA showed to be involved in Chlamydomonas cell signalling during osmotic stress response 38 . However, microalgae ABA-mediated responses seem to be less complex than in land plants as little or no homology was found between most land plants ABA receptors/effectors and the Chlamydomonas proteome 39,40 . Considering that SnRKs control entire branches of the metabolism in Arabidopsis and other studied models, the identification of CKIN stress-specific dynamics, will potentially reveal new targets for further bioengineering research aiming to accumulate economically relevant biomolecules.
Therefore, in the present study, we aim to fully describe the entire set of genes belonging to the CKIN family in Chlamydomonas and its potential implication in specific stress response mechanisms and in ABA-mediated responses. The combination of Chlamydomonas and other microalgae genome mining, plant protein-protein interaction databases, and quantitative reverse transcription PCR (RT-qPCR) allowed not only the definition of this family and its evolutive history, but also defining its interacting networks and testing its expression levels under exogenous ABA addition, ABA synthesis inhibition, and a wide-range of stressful conditions. The results herein presented represent a great advance in microalgae and stress biology research, defining a new set of potential targets for biotechnological improvement. Although SnRK are a key group of protein kinases for biotechnology, this family was never fully characterized in microalgae.

Results
Identification of SnRK protein orthologs in Chlamydomonas. Initial Table 1). Out of these, 10 genes were previously described by Gonzalez-Ballester et al. 36 and Valledor et al. 35 , while 12 were found in Chlamydomonas for the first time. Protein alignment, phosphorylation site identification, functional prediction of unannotated domains, expression and protein interaction analyses were performed for the curation of the identified sequences as described below. M-Coffee alignment of 4 Chlamydomonas CDPK along the 19 identified Chlamydomonas catalytic CKIN sequences allowed the definition of three putative catalytic functional clusters (Fig. 1a). A fourth cluster was conformed with the 3 identified CKIN1 regulatory sequences aligned with their Arabidopsis, Homo sapiens and Saccharomyces cerevisiae orthologs (Fig. 1b). Although closely related to CKIN2, CKINL was excluded from the kinase group during alignments curation. The first identified catalytic cluster involved the SnRK1/AKIN complex, including Chlamydomonas catalytic subunit α, CKIN1, (Serin/Threonin Kinase (PTHR24343), UBA, and KA1/ αCTD domains) and CKIN1L (Serin/Threonin Kinase (PTHR24343) and UBA domains). CKIN1L displayed also unique features as a long N-terminal unconserved sequence and lacked conserved Thr189, key into CKIN1 activation 41 , and the regulatory KA1/αCTD domain (Supplementary Figure S1). The second cluster, SnRK1 regulatory subunits, included those non-catalytic subunits of the SnRK1 complex: CKIN β (Immunoglobulin E-set/ CBM and βCTD/ASC/AMKβI domains), CKIN βγ (Immunoglobulin E-set/CBM and CBS domains) and related CKIN γ (CBS domains). CKIN γ showed more identity to plant γ subunits and Saccharomyces cerevisiae γ-like SDS23 42 than to true γ-acting proteins as plant βγ and human γ subunits (Fig. 1b). The sequences belonging to these two clusters were conserved across evolution as shown by its curated alignments using M-Coffee (Fig. 1c). Sequences belonging to the Chlamydomonas CKIN2 subfamily were characterized by a SRK2 Serin/Threonin Kinase domain, which is shared with CKIN1. M-Coffee alignment tree arranged CKIN2 sequences into two clusters, CKIN2 (A) and CKIN2 (B) (Fig. 1a). The Serin/Threonin Kinase domain (PTHR24343) was present in all Chlamydomonas CKIN2 sequences and its Arabidopsis orthologs (  Table S3). In plants, these kinases have two conserved serine or threonine phosphorylable residues into the activation loop required for gaining kinase activity 43 . Although Serin/Threonin Kinase domain showed slight variations across Chlamydomonas CKINs, it can be considered that activation region is conserved and functional since this site has been revealed as differentially phosphorylated in various environments (Supplementary Figures S2, S3) after reanalyzing available phosphoproteomic datasets [44][45][46] .
Chlamydomonas CKIN2 (A) has regions homologous to Arabidopsis osmotic stress-dependent activation domain I after its kinase domains, where CKIN2 (B) sequences showed a conserved region after kinase domain but less similar to the plant domain I (Supplementary Figures S2, S3). In CKIN2.2 and 2.6-2.8 ABA-dependent activation domains II-like sequences also followed domain I, but were smaller and less rich in acidic residues than its Arabidopsis counterparts 36 (Supplementary Figure S2). CKIN2 subfamily was further analyzed along CKIN1 and its regulatory subunits to determine gene duplicities and evolutive relations between the members of this family (Fig. 2). CKIN genes mapped to 12 chromosomes, being CKIN2.10 and 2.12, and CKIN2.1, 2.3 and 2.11 close in their respective chromosomes. Homology results showed that CKIN2.2 and CKIN2.5-2.7 evolved by duplication of CKIN2.8, the sequence exhibiting higher homology to higher plants and closer to CKIN1. On the other hand, CKIN2.1, 2.3 and 2.4, CKIN2.14 and 2.10, and CKIN2.12 and 2.13 evolved from three different ancestors no longer conserved in Chlamydomonas.
The employment of the strategy described above over Ostreococcus lucimarinus, Chlorella variabilis, Coccomyxa subellipsoidea, Volvox carteri, and Dunaliella salina, showed divergent SnRK/CKIN family structures between land plants and microalgae (Supplementary Table S4). All studied species had one SnRK1 catalytic subunit, but had different elements belonging to SnRK2 and 3 subfamilies. Chlamydomonadaceae species lacked SnRK3 subfamily orthologs but had large and diverse SnRK2/CKIN2 (A) groups composed by CKIN2 proteins with uncharacterized insertions and long C-terminal extensions, and others similar to plant SnRK2s as it was observed in Chlamydomonas. Chlorella and Coccomyxa shared with Chlamydomonadaceae a SnRK2/CKIN2 (B) group and a reduced SnRK/CKIN2 (A) group making a SnRK2 subfamily more similar to Arabidopsis compared to Chlamydomonoceae. In turn, Chlorella and Coccomyxa had SnRK3 elements. Interestingly, Ostreococcus, the closer genome to land plants among the analyzed, contains only a small number of Arabidopsis-like sequences not including SnRK3, making the smallest SnRK families among the studied ones (Fig. 3, Supplementary Table S4).
Moreover, the use of iTAK database 47 and kinase classification 48 over all Chlamydomonas, Volvox, Dunaliella and Ostreococcus genomes showed similar grouping than proposed following our approach. SnRK1 sequences Figure 2. CKIN family evolution in Chlamydomonas. Chlamydomonas CKIN family genes were represented along chromosomes and gene duplications showed as links between duplicated elements. Link colour and thickness show BLASTP e-value and % identity based duplication confidence. Red thicker links joins genes coming from highly probable duplication events with e-values lower than 10 −50 and more than 50% identity, blue links joins genes coming from mid probable duplication events with e-values lower than 10 −45 and more than 45% identity and green links joins genes coming from low probable duplication events or ancient duplication events with e-values lower than 10 −40 and more than 40% identity. were classified as CAMK_AMPK and almost all found SnRK2 sequences fell into SNF like CAMK_OST1L kinase group. In Chlamydomonas, the divergent CKINL and CKIN1L sequences made an exception being classified respectively into CAMK_CAMKL-PASK-PIM and CAMK_Cr-1, close to SNF microalgae exclusive groups. Ortologs to these sequences were also found in Volvox, Dunaliella, and Chlorella. . Subfamily distribution between species shows SnRK3 restricted to land plants and basal core Chlorophytes. SnRK/CKIN2 subfamily resulted highly variable between species, from the reduced and close to land plants Ostreococcus subfamily to the diverse and large Chlamydomonadaceae subfamilies.

Modification of basal media
Nitrogen deprivation (-N) Substitution of ammonium chloride by potassium chloride.
Carbon deprivation (-C) Removal of carbon sources from the media.  Expression profiles of Chlamydomonas CKINs under abiotic stress. RNA was isolated from CC-503 strain after 48 h of exposure to different stressful situations described in Table 2. The application of these stresses significantly affected cell growth, reducing multiplication rate in all cases, except for phosphorous limitation and UV irradiation (Supplementary Figure S4). These results were complemented with a targeted analysis of available RNAseq datasets studying the response of this organisms to nitrogen 49,50 , sulphur 37 , iron and carbon dioxide 51 deficiencies and hydrogen peroxide-induced oxidative damages 52,53 . The expression of Chlamydomonas CKIN genes greatly varied under the abiotic stresses tested, with different ranges of overexpression/repression in function of the analysed gene and stress (Fig. 4, Supplementary Table S5). PEG-induced osmotic and UV stresses triggered the strongest responses of this family, with an average CKIN abundance increase of 3.9-and 4.5-fold respectively and compared to control. Contrarily, the transition from mixotrophic to autotrophic conditions, H 2 O 2 -induced oxidative damage, and nitrogen limitation did not induce overall abundance change of the CKIN family genes. Although this reflects the dynamics of the entire CKIN family under several stresses, it is more interesting to analyse the dynamics of specific CKIN genes to look for possible targets for future bioengineering studies.
CKIN genes also showed different ranges of response within each treatment, from the small abundance variations of CKIN2.14 and CKIN1L (2.28-to −10-fold and 2.94-to −2.43-fold change; maximum and minimum change considering all stresses, respectively) to the strongest response of CKIN2.1 and CKIN2.12 (440-to −1.47-fold and 354-to −1.28-fold change respectively). Salt, temperature, UV and PEG-induced osmotic stress caused the overexpression of all genes of these family at different levels, except for CKIN2.14, downregulated under salt, low-temperature and UV stress and CKIN1L, downregulated by PEG. On the other hand, carbon limitation (both the reduction of available CO 2 and the transition from mixotrophic to autotrophic growth) led to the overexpression of genes belonging to the CKIN1 complex, CKIN2.3, CKIN2.10 and CKIN1L, while nutrient deficiencies (S, N, P, Fe) induced gene-specific responses. Although Arabidopsis γ and AKIN10/11 or SnRK1.3 interaction have not been identified, Chlamydomonas CKIN γ (close to plant γ and Saccharomyces SDS23) showed a close expression pattern to CKIN1 and its regulatory subunits clustering in the same group. Moreover, CKINL differential expression pattern under tested stresses supported its exclusion from the CKIN2 subfamily.   STRING and ChlamyNet based CKINs interaction and co-expression networks. CKINs showed to be a part of a complex interaction network, as shown into the STRING 54 and ChlamyNet 55 databases. STRING encompasses a collection of predicted and experimentally proven protein-protein interactions in Chlamydomonas and other species. The Chlamydomonas CKINs present in the STRING database turned out to be associated between them and with several biochemical and regulatory pathways ( Fig. 6; Supplementary Figure S5). Three main CKIN clusters were defined based on the interactions observed. The first one, the hub of this network, includes CKIN α, β, and βγ subunits of the CKIN1 complex. This cluster is directly linked to CKIN2, carbon and nitrogen metabolism, and mRNA splicing. In addition, it was indirectly linked through Ca-dependent mechanisms to autophagy and DNA remodelling and maintenance mechanisms. No interactions were found for the non-interacting CKIN γ subunit. Second and third CKIN clusters comprise CKIN2 subfamily, being linked to different Protein Phosphatase 2 C (PP2C) family phosphatases. CKIN2.4, 2.9, 2.13, 2.16, and CKINL only showed interaction with PP2CF, PP2C3, and A8IIX1 while the other set of CKIN2 and CKIN1L was also interacting with 9 other PP2C and related phosphatases. ChlamyNet database comprises a Chlamydomonas transcript correlation network. Eight CKINs were identified in this database being part of several clusters (Supplementary Figure S6). CKIN β, CKIN2.12, and CKIN1L were into the same group, directly connected to carbon (macromolecule and hexose metabolism, photosynthesis) and nitrogen metabolism (amino acid metabolism and protein turnover). CKIN2.6 formed a second cluster directly related with autophagy elements. CKIN γ conformed the third major cluster interacting with transcriptional regulation elements, autophagy and protein folding/assemblage. CKIN 2.3 and 2.16 were also related to transcriptional regulation and CKIN 2.7 to autophagy, heat response and protein folding. Other proteomic and metabolomic interaction networks, as the Chlamydomonas nitrogen starvation and recovery response STRING based network presented by Roustan et al. 56 , and the nitrogen starvation response sPLS correlation network published by Valledor et al. 5 , were surveyed for CKIN or CKIN related elements. Although no CKINs were identified into the first network, interactants as the flagellar associated protein (FAP121) and the protein kinase CDPKK2 were shared between CKIN and nitrogen stress response network. The second network only included CKIN γ.

Discussion
The SnRK family is conserved in all eukaryotes, as SnRK1/SNF1/AMPK/CKIN1, functioning as a sensor of cell energetic status 18,57,58 . In plants, this family evolved into two more subfamilies, 2 and 3, which are key players in ABA-dependent and -independent stress response mechanisms 25,31,59 . In Chlamydomonas, SnRK/CKINs were initially described under sulphur deprivation 60 , and later related to a wide range of stress responses 5,35-37 and indicated as potential targets for increasing the production of industry-demanding biomolecules 32 . However, these studies focused on the functional characterization of some CKINs rather than defining the whole family, explaining why only 10 CKINs were described in Chlamydomonas so far, when in land plants like Arabidopsis this family comprises 38 members 27 .
Mining Chlamydomonas genome using homology searches allowed the definition of a large number of candidate genes, most of them only homologous to the Ser/Thr kinase domain. Furthermore, CKIN and CDPK protein families showed great similarity, being the absence of a Ca-binding EF-hands domain the key to distinguish these families 27 . In consequence, all BLAST hits were later curated by protein domain analyses. This strategy was later expanded to a domain-based genome mining looking for sequences containing all the domains required for a protein to be classified as CKIN. This proved to be useful allowing the incorporation of two novel CKIN sequences (CKIN2.12, and CKIN2.15) otherwise lost during data mining.
Overall, in silico genomic analyses resulted in the characterization of 21 proteins belonging to the CKIN family. As this family has not been deeply studied in other algae, comparisons are difficult to perform but recently, one of these analyses described only 2 elements belonging to SnRK1/CKIN1 subfamily in Volvox carteri 61 . Out of these, only one sequence can be undoubtedly classified as SnRK1, since its kinase, UBA, and KA1/αCTD domains are highly conserved across evolution 62 . The other sequence cannot be assigned to this group since regulatory residues were not present, being more similar to CKIN1L. Conversely, any SnRK2/CKIN2 elements were described in this species. The application of the described above workflow to the analysis of Volvox, Dunaliella, Coccomyxa, Chlorella and Ostreococcus genomes allowed the definition of progressively larger and more complex SnRK2s subfamilies as the species were closer to Chlamydomonas. Identified SnRK kinase members in Chlamydomonas, Volvox, Dunaliella and Ostreococcus were identified as such into the iTAK database 47 supporting Chlamydomonas CKINL exclusion. These analyses validated the method capabilities to discover new SnRK/ CKINs, highlighting the need of specific methodologies beyond homology searches, such as protein domain analyses, for the fully characterization of a protein family when dealing with families showing atypical evolution such as CKIN2 subfamily in algae 63 . Interestingly, from the genomes analysed, only land plants and humans had more than one SnRK1. SnRK1/AMPK duplication is common between plants and vertebrates whereas no more than one Snf/CKIN1 subfamily member has been identified in fungi and invertebrates 21 . This suggests a probable complexity related feature. Vertebrates and plants needed to diversify this regulatory mechanism to cope with their more complex life cycles and somatic structures, or more stringent environments 64,65 . This increasing somatic/environmental complexity fits the functional diversification found in some land plants and vertebrates SnRK1/AMPKs. Supporting this is the Arabidopsis SnRK1.3 reproductive tissues focused expression 66 , same species AKIN10/11 differentiated response under stress 64 and muscle/liver focused expression of human AMPK α2 which is related to osteogenesis and adipogenesis 67 .
On the other hand, and despite the conservation of SRK2 Serin/Threonin kinase domain of the CKIN2 subfamily, more efforts are needed to be paid to unequivocally classify these elements. If CKIN2 subfamily originated from duplication and divergence of CKIN1, a common ancestor or "founder like" to all SnRK2/CKIN2s would diverse this subfamily in different groups ranging from CKIN1 to land plants SnRK2s. Both CKIN2 subgroups (A/B) were conserved within core Chlorophytes, while those closer to SnRK1, CKIN2 (B), were lost or changed in land plants. SnRK2/CKIN2 (A), including plant SnRKs, was larger and more diverse in core Chlorophytes, with almost all this sequence diversity being Chlamydomonadaceae exclusive. In Chlamydomonas, CKIN2.2 and 2.8 can be considered the common ancestors of most of its CKIN2 (A) sequences, despite some ancestor of this group long CKIN2s may be lost along CKIN2 (B) group ancestor. Furthermore, despite being present in Chlorella and other Chlorophytes, this ancestor CKIN2 (B) group has been lost in Ostreococcus, exhibiting only plant-like CKIN2s, this may also be happening due to its reduced genome size. Although atypical and different from land plants, algal CKINs are supposed to be engaged in most of the functions land plants SnRKs are involved. This is sustained by the gene balance hypothesis, which explains the conservation observed, predicting that transcription factors and proteins belonging to signalling networks are more likely to be retained 68 .
In plants, SnRK1-based sugar signalling acts as a master regulator of carbon/energy ratios and is strongly linked to hormonal regulation and different signal transduction pathways 69,70 . In Chlamydomonas this complex resulted to be very sensitive to most of the imposed stresses. However, the suppression of carbon in culture media did not caused the highest overexpression rates of this complex, being more responsive to salt and extreme temperatures. The damages that these stresses represent to algae causing an energy shock are probably greater than those from removing acetate from media (that will still interchange CO 2 with air). The metabolic remodeling required to survive under extreme conditions mediated by SnRKs has been reported in different plant systems 35,71 , being the interaction of the SnRK1 complex with specific transcription factors and kinases a basic mechanism to trigger proteome and metabolome remodelling. CDPKK2, a Chlamydomonas ortholog to Arabidopsis GRIK1, known for activating the land plant SnRK1 72 is one of the kinase intermediaries highlighted into presented STRING based SnRK network. The kinase was also present as central node into Chlamydomonas nitrogen depletion and recovery stress network 56 supporting the complex activity under abiotic stresses other than carbon depletion.
Conversely to CKIN1, CKIN2 proteins were more difficult to classify due to its partial divergence to its land plant counterparts. Previously described CKIN2 elements, falling all into the CKIN2 (A) group, exhibited a high similarity to its Arabidopsis SnRK2s orthologs. On the contrary, newly discovered elements, all CKIN2 (B) sequences and the Chlamydomonadaceae exclusive SnRK2 (A) sequences, were more dissimilar. However, all Chlamydomonas CKINs exhibited a Ser/Thr kinase domain and, in the case of CKIN2 (A) sequences a land plant-like regulatory domain I which is less conserved in Chlamydomonadaceae sequences CKIN2.11 and CKIN2.15-16. This domain is related to osmotic and salt stress responses in plants 26,63 . The domain I function seems to be mostly conserved in Chlamydomonas, as all elements containing it, except CKIN2.14, were overexpressed under hyperosmotic and saline stresses. Interestingly CKIN2 (B) group sequences exhibited a conserved region after kinase domain which is different than land plants domain I. CKIN2 (B) sequences and CKIN2 (A) CKIN2.11 and CKIN2.15 were also overexpressed under hyperosmotic and saline stresses probably due to the existence of an algae specific domain and other sensing/signalling mechanisms respectively.
The presence of different regulatory sequences, some of them elongated, may explain the differential classification of CKIN2 after M-Coffee analysis, which focused on most conserved regions. CKIN2 (A) grouped the closest sequences to the common land plant-algae ancestor and all Chlamydomonadaceae exclusive ones. Some of these sequences were almost identical in length and sequence to Arabidopsis while most, including the closest Chlamydomonas CKIN2 to Arabidopsis SnRK2s (CKIN2.8), had long C-terminal ends and/or extra sequence loops. These features are probably related to a multiple environmental stress response capacity. CKIN2 (A) sequences have a heterogeneous response pattern under different stresses. Besides this, their expression levels under sulphur or nitrogen starvation, phosphorous limitation and oxidative stress were higher than CKIN2 (B) sequences.
CKIN2 (B) contained core Chlorophyte exclusive sequences. Sequences of this smaller group are more homogeneous, with a conserved region after kinase domain in place of regulatory domain I and long C-terminal ends (excepting CKIN2.9). Within the CKIN2 (B) cluster, CKIN2.13 and 2.10 were the most overexpressed CKIN2 under carbon starvation and autotrophic growth respectively, surpassing CKIN1 complex. This correlates this CKIN2 (B) "ancestor" group sequences with an energy stress related function commonly associated to CKIN1 complex.
Response heterogeneity and group size makes difficult to propose a common regulatory mechanism for CKIN2 (A). Moreover, despite they share with CKIN2 (B) and land plants a well conserved kinase-dependent activation loop 43,73 , most of its sequences also display extra sequence loops/elongations whose function is probably diverse. Furthermore, the presence of phosphorylation sites, coiled-coil protein interaction motifs, glycine and glutamine rich regions in CKIN2 (A) sequences points to a probable CKIN2 regulatory function. Plant SnRK2 protein ends usually have an important role in their regulation trough phosphorylation 74 . Moreover, coiled-coil structures, know by their function in protein-protein interactions 75 , hold different kind of protein interactions in plant SnRKs, such as the interaction with PP2Cs 76 . In addition to the activation loop and these C-terminal potential regulatory regions, the regulation of SnRK2 in Arabidopsis requires the presence of two regulatory domains. All CKIN2 (A) had a domain I with different degrees of degeneration, but domain II like regulatory sequences could only be recognized into CKIN2.2 and CKIN2.6-2.8, the closest CKIN2s to land plants SnRK2 subfamily. Although regulatory domain II is characteristic of some CKIN2 (B), these sequences are shorter that those present in land plants. In land plants this domain, with its characteristic acidic patch, is present in highly ABA responsive elements, being named ABA box, despite some SnRK2s can respond to ABA without this structure 26,63,77 .
Interestingly, CKIN2.14 and CKIN2.5 were the only responsive elements to exogenous ABA, regardless of not having acidic patches in their sequences. ABA regulatory mechanisms are well described in land plants, where the acidic domain II mediates the inactivation of the kinase through a SnRK2-PP2C interaction 74 . ABA, interacting with PYR /PYL/RCAR ABA receptors inhibits PP2C, allowing the kinase activation 78,79 . Although all of these elements are not present in microalgae or have a very low similarity to higher plants 39,40 , string-based protein-protein analyses demonstrated the interaction between SnRK2s and PP2C, protein phosphatase Mg2+/ Mn2+ dependent (PPM) and Protein phosphatase type 1 isoform K (PP1K) phosphatases, and CAB and CDP kinases, all core players in ABA signalling 80 . Therefore, CKIN2.14 and 2.5 might be elements in an ABA regulatory pathways in Chlamydomonas, highlighting the need of novel SnRK-PP2C interaction sequences and alternative ways in which ABA breaks this interaction. This hypothesis is also supported by the effect of exogenous application of fluridone, which increased the expression of CKIN2.1, 2.2, 2.5 and 2.10 despite it can also be considered as an oxidative stress induction since CKIN β subunit was also induced. Results clearly showed low ABA signalling dependence of CKIN2s in Chlamydomonas. In lesser extent, ABA-independent regulation of SnRKs occurs in land plants, as in the case of Arabidopsis SnRK2. 1, 2.4, 2.5, 2.9 and 2.10, induced by osmotic stress following an ABA-independent pathway 63 . Arabidopsis ABA-independent SnRKs regulate transcripts of stress related genes under hyperosmotic conditions, thus complementing ABA-dependent SnRK2s function 81 . In Chlamydomonas, ABA-independent CKIN2s were also responsive to osmotic stress, as in land plants. But contrary to plants, specific responses to low nitrogen 5 or sulfur deprivation 36 seem to be triggered in an ABA-independent manner. These results support the idea that the plant-specific SnRK2/CKIN2 subfamily plays a crucial role in stress response signaling both in Arabidopsis and Chlamydomonas.
These mechanisms are not entirely related to energy-saving decisions, but lead to a complex remodelling of cell metabolism, as demonstrated by the interactions with DNA repair and maintenance pathways and TOR in Arabidopsis 82 , and in Chlamydomonas to a similar and complex interaction network as it was proposed by STRING and ChlamyNet analyses. The fact that the expression of a large number of CKIN2s is induced by a single stress suggests a great compensatory effect or pleiotropy within this family in Chlamydomonas. It is well-known that the consequence of most stresses is oxidative damage and/or low energy syndrome 83,84 but it is striking the high number of elements triggered in Chlamydomonas under most of the studied stresses given their low intensity. On the other hand, the high responsiveness to osmotic, salt, and UV stresses of this family is crucial for a freshwater alga in natural environments, since they cannot control neither water quality nor environmental UV irradiation. Thus, a fast, flexible, and efficient mechanism is required to ensure algae survival under unfavourable environmental conditions.
The apparent lack of specificity of CKIN confronts to other plant systems, with elements with very specific functions and stress responsiveness. Land plants also had a third SnRK subfamily, comprising proteins kinases interacting with calcineurin B-like calcium binding domains 63,85 , mostly involved in drought and salt resistance, being the SOS (salt overly sensitive) the best-known mechanism 86 . The characteristic NAF/FISL 27 domain of SnRK3 is not present in Chlamydomonas, but Chlorella and Coccomyxa have one element. This suggests that SnRK3 is not only characteristic of land plants but is also present in the last common ancestor and lost in some core Chlorophytes. Homologs of this sequence were not found in Ostreococcus (the closest microalgae to land plants) but this would be easily explained by its reduced genome. Moreover the reduced SnRK3 subfamily size in microalgae makes easy their lost through mutation, being probably compensated in their function by SnRK2s, specific CDPK proteins 87 or by SNF/SKP1/Ubiquitin ligase complexes, already identified as key elements of hormone, sugar, and stress responses 88 . In line with this, Chlamydomonas, Volvox, and Dunaliella have larger and more diverse SnRK2/CKIN2 subfamilies than Chlorella and Coccomyxa, both containing one SnRK3 element.
The genome-wide approach used in this work over Chlamydomonas for the identification of its CKIN family have completed previous work in this species while extending it to other microalgae. This kinase family description has shed light to its unique structure and sequence features in Chlamydomonadaceae, highlighting the conserved abiotic stress sensitivity of the Chlamydomonas members. Thus, paving the way for the description of novel CKIN and probably ABA mediated stress responsive pathways in microalgae. The shared and unique core Chlorophytes SnRK/CKIN family structure make Chlamydomonas a suitable system for novel stress response mechanisms identification and a better fitted model than land plants for the identification of algae specific targets for biofuels and secondary metabolites production enhancement in these species.

CKIN sequence identification and classification in Chlamydomonas. Chlamydomonas CKIN fam-
ily genes were initially defined by BLASTP comparison against the Chlamydomonas proteome v5.5 89 available at Phytozome 33 employing Chlamydomonas 35,36 and Arabidopsis 25 previously identified CKINs and SnRKs as query (Supplementary Table S1). Homology was considered for e-values lower than 10 −25 generating a first uncurated sequence list (Supplementary Table S2).
On a second step, Inter Pro Scan 90 was used to define the domain structure of all candidate sequences, filtering out those proteins with no SnRK domain structure (e.g. CDPKs). Furthermore, BLASP query sequences domains (Supplementary Table S3) were used as reference to search for potential CKINs into the Chlamydomonas genome using BIOMART 91 . Protein sequences with characteristic SnRK/CKIN domains, were aligned with M-Coffee 92 .
Maximum likelihood (ML) sequence trees were built into PhyML 93 platform employing M-Coffee general alignment distances after a TCS (Transitive Consistency Score) 94 alignment filtration. One hundred bootstrap replicates were done into the same PhyML platform over the filtered alignment data to assess tree consistency. Different tree topologies were evaluated including a different set of ortholog SnRK proteins sequences (Supplementary Table S6) and Chlamydomonas CDPK sequences. Same tree-group sequences were aligned together using M-Coffee to validate sequence adhesion to their group through distinctive sequence motif conservation. These block alignments were curated using g-blocks 95 . COILS 96 and different phosphoproteomic datasets were used for alignment enrichment with coiled-coil predicted regions and phosphosites. Gene duplication was inferred from tree topologies and intra SnRK sequence comparison by BLASTP 97 . Intra tree group comparisons of BLASTP e-values and identity % were used as parameters for duplicity consideration. Three confidence thresholds for high, medium and low duplication origin probability were respectively defined at 10 −50 , 10 −45 , and 10 −40 for e-values, and 50, 45, and 40 for identity %.
Volvox, Dunaliella, Coccomyxa, Chlorella and Ostreococcus sequences homologous to Chlamydomonas and Arabidopsis CDPKs and SnRKs were obtained from Phytozome 98-101 employing the methods previously applied to Chlamydomonas genome mining as described above. SnRK/CDPK sequence groups found in Chlamydomonas, Volvox, Dunaliella and Ostreococcus were curated both manually and through iTAK application 47 . Identified SnRK/CKIN sequences in microalgae species and Arabidopsis along with Chlamydomonas CDPK as an outgroup, were aligned using M-Coffee. A ML sequence tree was built into PhyML platform from the TCS filtered M-Coffee global alignment distances. One hundred bootstrap replicates were done into the same PhyML platform over the filtered alignment data to assess tree consistency. Transfer bootstrap expectation (TBE) bootstrapping method 102 with 100 replicates was used along conventional Felsenstein's method for improving deep branches bootstrap support.
Chlamydomonas culture and stress response characterization. Chlamydomonas strain CC-503 cells were grown on a closed incubator (25 °C, 120 rpm, 16 h light:8 h dark photoperiod and 190-200 µE m 2 s −1 light intensity provided by warm white LEDs) in liquid HAP culture media 103 supplemented with 10 mM sodium acetate at an initial cell density of 3-5 × 10 5 cells mL −1 . Basal media composition and growth conditions were changed according to Table 2 to test different abiotic stresses.
Fifty mL culture samples were collected at the beginning of the experiment (0 h) and 48 h after the start of the assay. Samples were centrifuged (4000 rpm, 6 min) and resulting cell pellet masses were estimated gravimetrically and frozen using liquid nitrogen. Four biological replicates of each stressful scenario were performed.

Expression analysis of CKIN genes under a variety of environmental stresses.
A precise estimation of SnRK abundance based on RT-qPCR was performed over 9 experimental situations ( Table 2). RNA was extracted from the previously frozen pellets using the method described by Valledor et al. 104 . 1.7 µg of DNA free RNA was used for cDNA synthesis.
RT-qPCR analysis was performed using the CFX96 Touch Real-Time  Table S7). Each individual experiment was performed with two analytical replicates for each biological replicate from each condition tested. UBQ and RCK1 were selected as endogenous controls after testing the expression stability of IDA5, UBQ, TUB, and RCK1 with the geNorm software 105 . Gene expression of each SnRK was evaluated by calculating ΔΔCq values (Supplementary Table S5b) according to the recommendations proposed by Hellemans et al. 106 .
NGS-based transcriptomic datasets of Chlamydomonas reinhardtii available at the AlgaePath repository 107 were mined to perform a complementary characterization of the CKIN family expression under stressful conditions. Five different datasets were analysed: sulphur depletion 37 , nitrogen deprivation 50 , low CO 2 content 51 , oxidative stress and iron deprivation 53 . Each CKIN gene was searched using their gene accession, and its fold change variation under stress was obtained by comparison to non-stressed abundance values. In all cases, abundances within each dataset were normalized against its corresponding abundance in controls (Supplementary Table S5a). The integration of NGS and RT-qPCR abundances was done based on the z-score of the fold change variation between control and stress situation.
ABA-induced regulation of CKINs. HAP medium was supplemented either with 500 µM ABA or 50 µM fluridone (carotenoid and ABA synthesis inhibitor through Phytoene desaturase inhibition) following previous works 38,108 . CC503 cell cultures were grown under the previously described basal conditions for 48 h, and compared to its corresponding controls. RNA extraction and RT-qPCRs were performed over 3 biological replicates each treatment. The effect of ABA and fluridone was monitored by quantifying the expression of the ABA-inducible genes Beta-Carotene Hydroxylase (CHX) and Ascorbate peroxidase (APX) 109 .
Bioinformatic and statistical analyses. All the procedures for the identification and classification of Chlamydomonas SnRK were performed locally employing the bioinformatics suite Geneious v7 (Biomatters Inc.), with the exception of Inter Pro Scan 90 and BIOMART 91 searches that were performed at the European Bioinformatics Institute (ebi.ac.uk) and Phytozome (phytozome.jgi.doe.gov) websites, respectively.
Protein-protein functional interactions were identified by using STRING v10 54 and ChlamyNet 55 databases. CKINs protein sequences were uploaded to STRING application and database was queried considering Chlamydomonas as a reference organism. Two related networks were made, one showing highest confidence (over 0.85 STRING interaction score) known interactions. The other network including also high confidence (over 0.7 STRING interaction score) known and predicted interactions. CKINs protein sequences were also uploaded to ChlamyNet application, containing a Chlamydomonas transcript-based correlation network. A smaller network was made out of the original containing only first level CKIN interactions. STRING and ChlamyNET resulting networks were represented using Cytoscape v3.4 (Cytoscape Consortium 2016).
R v. package for R 110 . When mining available datasets in Algaepath, the abundance of each CKIN under the different stress situations was compared to its corresponding non-stressed controls considering fold change.

Data Availability
Sequence data analysed in this work, with their IDs listed at Supplementary