Spores of fertile individuals of S. latissima and L. digitata, collected at Perharidy (near Roscoff, 48.73°N, 4.00°W) were released onto cover slips using the hanging-drop technique (Wynne, 1969). The cover slips with settled spores were cultivated at 14°C and exposed to 20 μmol photons s^–1m^–2 of white light, with a 12 h light/dark cycle. The developing sporophytes were raised in 100 mL Petri dishes with weekly changes of culture medium. For all cultures, filtered autoclaved natural seawater (FSW) was enriched with Provasoli solution (10 mL Provasoli solution/L seawater) (Provasoli, 1968). After 4 weeks, the sporophytes were detached from the cover slips and transferred to 10 L bottles connected to an aeration system. Culture medium in the 10 L bottles was changed weekly. The culture conditions were the same as for the production of sporophytes from settled spores.

Cultures of the filamentous brown algae L. elsbetiae, Laminariocolax tomentosoides, and Microspongium tenuissimum were obtained from the Bezhin Rosko culture collection. They were kept in Petri dishes at 14°C and 5 μmol photons s^–1m^–2 with monthly changes of culture medium.

Both S. latissima and L. digitata species were co-cultivated with the endophyte for 2 weeks as described below.

Fifteen 2 L bottles were filled with 1.5 L sterile Provasoli-enriched FSW and connected to an aeration system (Supplementary Figure 1A). One kelp sporophyte was transferred to each of the bottles. A hole was punched in the kelp sporophytes at 1 cm distance from the basal meristem using a pipet tip. In the following experiment, the longitudinal growth of the sporophyte blade was measured by monitoring the distance of the hole from the basal meristem with a ruler (Parke, 1948) (Supplementary Figure 1B). The first measurement was done after 3–5 days to ensure that the growth behavior of all sporophytes was similar. Subsequently, a filament of either L. elsbetiae or M. tenuissimum of similar size was added to 5 bottles each (N = 5). M. tenuissimum – a filamentous brown alga which is not endophytic in S. latissima and L. digitata - was used as a control to test a nutrient competition effect. Nothing was added to the remaining 5 bottles (control cultures of sporophytes without endophyte).

After the addition of the filaments (day 0), growth of S. latissima was measured on days 3, 6, 9, 11, and 14. Growth of L. digitata was measured on days 3, 6, 10, and 14. To ensure a sufficient nutrient supply, an amount of 0.5 mL of Provasoli solution per day of experiment was added after each measurement. All co-cultivation experiments were performed at 14°C and 20 μmol photons s^–1m^–2 with a 12 h light/dark cycle.

The maximum quantum yield of photosystem II (Fv/Fm) was measured on the same days using a JuniorPAM fluorometer (Walz, Germany). The sporophytes were dark-acclimated for 20 min prior to the measurement. After the last measurement, the kelp sporophytes were frozen in liquid nitrogen and kept at −80°C for the molecular detection of the endophyte in the kelp tissue.

The growth curves and Fv/Fm graphs were drawn with GraphPad prism (GraphPad Prism Software, Inc., United States) and SPSS was used for statistical analyses (IBM Corp. Released 2015. IBM SPSS Statistics for Windows, Version 23.0. Armonk, NY, United States: IBM Corp.). Normality of the data and homogeneity of variances were tested with the Shapiro-Wilk test and the Levene test, respectively. Subsequently, data were analyzed with one-way ANOVAs. Significant differences were evaluated with the Tukey post hoc test.

Direct microscopic observation was not adapted to the observation of endophyte spore settlements after 2 weeks, as first tested on two L. digitata plantlets, co-cultivated with endophytes. We therefore used a qPCR-based molecular approach: DNA extraction and qPCR quantification of endophytes in the hosts were performed as described previously (Bernard M. et al., 2018). In total, up to 15 individuals of S. latissima and 13 individuals of L. digitata from three independent co-cultivation bioassays with L. elsbetiae as well as 3 randomly chosen control sporophytes of each species were analyzed by qPCR for the detection of L. elsbetiae DNA using specific ITS primers. Cycle threshold (CT) values were calculated with the LightCycler 480 Software (Roche, Germany): endophyte DNA was amplified with a maximum of 33 cycles, and above 34 cycles, it was considered as undetectable.

Sixteen L. digitata sporophytes raised in laboratory culture were transferred to small glass beakers, filled with 50 mL autoclaved FSW. 150 μg/ml of oligoguluronate blocks (GG, prepared from L. hyperborea according to Haug et al. (1974) were added to 8 sporophytes. All beakers were placed on a shaker for 3 h (100 rpm) and the occurrence of an oxidative burst was measured as described below. After the incubation, the sporophytes were washed by transferring them to new beakers containing 50 mL autoclaved FSW and shaking for another 15 min. This washing step was repeated twice. A hole was punched in 1 cm distance of the meristem in the kelps and they were transferred to 2 L bottles. The first measurement was done after 3 days to assure that growth behavior of all sporophytes (control and GG-treated) was similar. Then, filaments of L. elsbetiae were added to 4 of the GG pre-treated and to 4 of the untreated L. digitata sporophytes. Growth was measured as described above on days 3, 7, 10, and 14. Statistical analysis were performed as described above.

The net production of H₂O₂ in seawater surrounding kelp-endophyte co-cultures was determined using a luminol chemiluminescence method (Küpper et al., 2001). After measuring the fresh weight of young sporophytes of S. latissima and L. digitata, they were transferred to glass beakers containing 50 mL seawater and placed on a shaker (100rpm). The experimental set-up consisted of a control (only S. latissima or L. digitata), both kelps co-cultivated with the endophytes L. elsbetiae or L. tomentosoides, and 50 mL of seawater containing only filaments of L. elsbetiae or L. tomentosoides. As a positive control, 150 μg/mL of GG were added to another glass beaker containing 50 mL of seawater and a sporophyte of either S. latissima or L. digitata.

150 μL of seawater were taken as sample for each measurement. Measurements were done before starting the treatment (t = 0), and 2, 4, 6, 8, 10, 15, 20, 25, and 30 min after the addition of the endophytes or GG. For each measurement, 50 μL of 20 U.mL^–1 horseradish peroxidase, dissolved in pH 7.8 phosphate buffer, and 100 μL of 0.3 M luminol (5-amino-2,3-dihydro-1,4-phtalazinedione) were added automatically to the sample by two injectors of the GloMax 20/20 Luminometer (Promega, United States). Chemiluminescence was measured immediately after the injection with a signal time of 1 s. A standard calibration curve from 0.1 μM to 20 mM H₂O₂ was drawn to determine the concentration of H₂O₂ in the seawater samples. H₂O₂ production by the kelp per g fresh weight was estimated by integrating the total amount of H₂O₂ monitored over 30 min and expressed as log2-transformed fold changes between control and treatments. The experiment was repeated 3 times and a one-sample-t-test was used for statistical analysis.

For transcriptomic analysis, 16 bottles were filled with 1.5 L autoclaved Provasoli enriched FSW and adapted to an aeration system. One sporophyte of S. latissima (3–5 cm) was added to each bottle. After 24 h of acclimation time, filaments of L. elsbetiae were added to 8 of the bottles. The four individuals of the control group and 4 individuals co-cultivated with L. elsbetiae were taken after 24 and 48 h. The kelp sporophytes were blotted dry with tissue paper, frozen in liquid nitrogen and stored in −80°C until RNA extraction. The same experimental set-up was used for L. digitata sporophytes (Supplementary Figure 1C).

RNA was extracted as described by Heinrich et al. (2012) with a combination of a classical CTAB-based method and the RNeasy Mini kit (QIAGEN, Hilden, Germany) including an on-column DNA digestion. Quantity and purity of the extracted RNA were tested on a NanoDrop^TM spectrophotometer (Thermo Fisher Scientific Inc., Waltham, United States) and on a 2% agarose gel.

Based on the quality and concentration, three replicates of each condition and each kelp species were chosen for commercial library preparation and Illumina paired-end sequencing (HiSeq3000) at the Plateforme Génomique du Genopole Toulouse Midi-Pyrénées GeT (France).

The quality of the Illumina reads was checked using FastQC (Andrews, 2010). Reads were cleaned by removing adapters, low quality reads (Phred score < 33) and short reads (<50 nucleotides) with Trimmomatic (Bolger et al., 2014) and residual rRNA was removed with SortMeRNA (Kopylova et al., 2012). Another quality check was performed with FastQC on the processed reads to ensure that high quality reads were obtained through the cleaning steps. An additional cleaning step was done by removing reads of the endophyte by mapping to the transcriptome of L. elsbetiae, produced in the context of the Phaeoexplorer project¹.

A de novo transcriptome assembly was created for both kelp species separately based on the pooled processed control reads using Trinity (Grabherr et al., 2011) with the default options. The quality of the assembly was assessed by re-mapping the cleaned reads using the bowtie2 aligner (Langmead and Salzberg, 2012). Transcript abundance was estimated in TPM with PERL scripts implemented in Trinity. The quality of the transcriptome assemblies was evaluated by using BUSCO v2.0 (Simão et al., 2015) with eukaryote dataset and redundancy further reduced according to the TPM value.

Gene annotation was performed with a Blastx search against the NCBI-NR (Release 239) and the Uniprot databases (Release 2020_02) with an E-value cut-off of 10^–5. Furthermore, genes were assigned to 2nd level GO subcategories within the three root categories molecular function, cellular component and biological process using Blast2GO (Conesa et al., 2005).

Differential gene expression between the control and the co-cultivation treatments was determined separately for the 24 and 48 h samples using DESeq2 (Love et al., 2014). Log2 fold change values ≥ 0.7 and ≤ −0.7 with a p-value < 0.01 were considered to be up- and downregulated, respectively. Heat maps were plotted using the R package pheatmap.

The genes that were differentially expressed in both species were compared with Blastn (E-value cut-off of 10^–5) against each other in order to identify common DEGs.

No significant differences in growth occurred within 2 weeks of co-cultivation of S. latissima with the non-endophytic M. tenuissimum and endophytic L. elsbetiae (Figure 1A and Supplementary Table 1).

In the case of L. digitata, a significant difference between the treatments occurred 6 days after the addition of L. elsbetiae (Figure 1B and Supplementary Table 1). The growth of L. digitata decreased significantly (Figure 1B) as compared to the other treatments after 6 days of co-cultivation (Supplementary Table 1, one-way ANOVA, p = 0.013) and the difference persisted until the end of the experiment (Supplementary Table 1). There was no significant effect of co-cultivation with M. tenuissimum on the growth of L. digitata.

No significant differences in Fv/Fm occurred within S. latissima or L. digitata alone or in co-cultivation with M. tenuissimum and L. elsbetiae (Supplementary Table 1).

At the end of the experiment, i.e., after 2 weeks of co-cultivation, DNA of the endophyte was detected in 7 out of the fifteen S. latissima DNA samples using qPCR Laminarionema specific primers, whereas it was only detected in 4 out of 13 samples of the L. digitata DNA samples. No L. elsbetiae DNA was detected by qPCR in any of the controls.

Co-cultivation of L. digitata sporophytes with the endophyte L. elsbetiae resulted in a significant decrease of growth from day 3 until day 14 (red line in Figure 2 and Supplementary Table 2), as already described before (control as green line in Figure 2). However, the addition of L. elsbetiae did not have any effect on the growth of L. digitata sporophytes that had been pre-treated with GG 3 days before the co-cultivation was started (golden line, Figure 2). There was no effect of the GG elicitation pre-treatment alone on the growth of L. digitata (gray line in Figure 2).

As previously demonstrated by Küpper et al. (2001), oligoguluronate (GG) blocks triggered an oxidative burst in both kelp species, which is indicated by a significant fold change of H₂O₂ release as compared to the control (Table 1).

The addition of L. tomentosoides to S. latissima resulted in a slight increase of H₂O₂ concentration in the seawater (log2FC = 0.21, one-sample t-test, p = 0.09, Table 1). When added to L. digitata, on the other hand, L. tomentosoides caused a significant decrease of the H₂O₂ concentration in the seawater (log2FC = −0.42, one-sample t-test, p = 0.04, Table 1). No significant changes in the H₂O₂ concentration were observed after the addition of L. elsbetiae to both kelp species.

The cleaned RNA sequencing reads of S. latissima and L. digitata were de novo assembled by Trinity and the general features of these two de novo transcriptomes are presented in Supplementary Table 3. 90% of total expression was present in 24,733 and 34,251 transcripts, respectively. The results of the BUSCO analysis revealed a near-complete gene sequence information for two transcriptomes with 82.4 and 84.4% complete BUSCO matches with eukaryotic dataset in S. latissima and L. digitata, respectively.

Overall, the distribution of GO terms for the three root categories “Molecular Function,” “Cellular Component,” and “Biological process” were similar for the assembled transcriptomes of S. latissima and L. digitata (Supplementary Figure 2). Within the molecular function category, most hits were assigned to catalytic activity and binding. Within the biological process root, most genes belonged to metabolic and cellular processes (Supplementary Figure 2). Overall, 47.34% of the obtained genes of S. latissima could be annotated whereas the annotation rate was slightly lower in L. digitata (45.66%).

The comparison between control and co-cultivated conditions identified 107 and 155 differentially expressed genes (DEG) at both time points in S. latissima and L. digitata, respectively (Supplementary Figure 3). Only six DEG were detected in S. latissima in presence of L. elsbetiae after 24 h, whereas 46 genes were significantly differentially expressed in L. digitata between control condition and co-culture with the endophyte (Supplementary Table 4). At 24 h, the majority of DEGs were downregulated in both species (S. latissima, 5 genes, i.e., 83.33% of DEG, L. digitata: 29 DEG, i.e., 63%), with stronger repression for those identified in S. latissima and for 16 DEG of L. digitata (Figure 3A). Unlike S. latissima, where only one gene was moderated upregulated, 11 genes featured log2FC value above 5 compared to control in L. digitata (Figure 3A).

After 48 h, more changes in gene expression occurred and most detected DEG were upregulated in both species (Figure 3B). In S. latissima, log2FC values of the 101 identified DEG ranged from 11.41 to −28.27 (Supplementary Table 4), but the majority (78.22%) showed moderate expression log2FC between 2 and −2 (Figure 3B and Supplementary Table 4). In L. digitata, log2FC values of the 109 DEG ranged from 16.51 to −14.28 (Supplementary Table 4): 52 DEGs (47.71%) showed moderate log2FC between 2 and −2 and 29 DEGs (26.61%) were strongly upregulated (log2FC > 6) (Figure 3B, Supplementary Table 4). Among 262 DEGs identified in S. latissima and L. digitata in the 2-days cocultures with the endophyte, only 21 homologous genes were shared by both kelps.

Similar to the whole transcriptome analysis, an important part of the differentially expressed genes of both kelp species (between 39.0 to 45.0%) did not have a match through Blastx search in the available protein databases, as represented on Figure 3. In the case of S. latissima, after 24 h of co-cultivation with the endophyte, only one gene had a match through Blastx search as a conserved unknown Ectocarpus protein (Figure 3A, Supplementary Table 4). In L. digitata, among the 23 DEGs annotated based on at least one database after 24 h of the addition of L. elsbetiae, one putative respiratory burst oxidase homolog protein was significantly strongly upregulated (Table 2), and 9 genes corresponded to conserved unknown or hypothetical proteins in Ectocarpus genome (Supplementary Table 4).

After 48 h of co-cultivation of S. latissima with the endophyte, 19 DEGs corresponded to conserved unknown or hypothetical Ectocarpus proteins. A functional putative annotation was retrieved for only 38 genes out of 101 DEG (see Supplementary Table 4). Among those DEGs, several genes related to cell wall metabolism were significantly up-regulated, including five mannuronan C-5-epimerases, an endo-1,3-beta-glucanase, and an alginate lyase (Table 3). Two vanadium-dependent bromoperoxidases and a PAP2/haloperoxidase-like protein were also up-regulated, which could be related to antioxidative responses. In addition, two defense-related genes were identified as DEGs. One upregulated gene encoding a putative respiratory burst oxidase homolog protein might be involved in the oxidative burst and another gene encoding a putative LRR receptor-like serine/threonine-protein kinase related to PAMPs recognition was downregulated in the presence of L. elsbetiae. Additionally, three genes that might be involved in oxylipin signaling pathways were up-regulated, including one gene encoding lipoxygenase and two genes encoding lipases (Table 3). In L. digitata, a putative annotation was obtained for 60 DEGs (Supplementary Table 4). 31 DEGs were annotated as conserved unknown or hypothetical Ectocarpus proteins whereas a putative functional annotation was retrieved for 25 DEGs (see Supplementary Table 4). Among those DEG putative annotations, there were three up-regulated genes that might be related to defense responses such as metacaspase, serine carboxypeptidase-like and carbohydrate 4-sulfotransferase genes (Table 2). The co-cultivation with endophyte repressed the expression of genes involved in photosynthesis and carbohydrate metabolism such as a light harvesting protein lhcf6, a carbonic anhydrase and a cellulose synthase. However, a photosystem II reaction center protein D1 gene and two GDP-mannose dehydrogenase genes were up-regulated during the co-cultivation. The co-cultivation also induced the down-regulation of a gene involved in fatty acid metabolism, a long-chain acyl-CoA synthetase putative gene. Furthermore, a gene coding for a protein homologous to one of those upregulated in the Ectocarpus imm-mutant and a gene with high similarity to the Ectocarpus virus gene (EsV 1–7), encoding for a cystein-rich protein, were strongly upregulated in the co-cultivation treatment with L. elsbetiae (Supplementary Table 4).

Previously, it was reported that algal endophytes can reduce the growth of their hosts by up to 70%, as it has been shown for the red algal endophyte Hypneocolax stellaris in its host, the rhodophyte Hypnea musciformis (Apt, 1984). Here we show that an effect on growth depends on the host species, as different physiological responses occurred between the two hosts S. latissima and L. digitata challenged by the endophyte L. elsbetiae.

The growth of the main host S. latissima was not affected by the endophyte during the 2 weeks of co-cultivation, and L. elsbetiae DNA was detected in half of the sporophytes at the end of the experiment. Although this detection does not prove that endophytic filaments were already growing inside of the kelp thallus, at least spores of the endophytes were likely to be attached to the kelp tissue by the end of the experiment. Thus, a direct and tiny contact between the endophyte and the kelp had been established, as already observed (Heesch and Peters, 1999), but without directly affecting the growth of S. latissima.

In contrast, the growth of the occasional host L. digitata was significantly reduced after a few days of co-cultivation, when only one third of the plantlets featured L. elsbetiae DNA contamination. As the filamentous brown alga, M. tenuissimum, did not have any effect on both kelps, nutrient competition could be excluded as a possible cause of the growth reduction. On the transcriptome level, the expression of several growth-related genes in L. digitata was impacted by the co-cultivation with L. elsbetiae. One gene encoding a putative cellulose synthase was significantly down-regulated 48 h after the addition of L. elsbetiae, and this might decrease cell elongation and division rate in young sporophytes. In Arabidopsis thaliana, lack of cellulose synthase resulted in a reduced cell growth rate of young seedlings (Hu et al., 2018). A gene coding for carbonic anhydrase, an essential enzyme for CO₂ fixation, was also down-regulated in L. digitata, suggesting an impact on carbon uptake and photosynthesis. It has been reported that endophytic pathogens can impair the efficiency of energy transfer from the light harvesting complexes to the reaction center of PS II in land plants (Luque et al., 1999; Guidi et al., 2007) and in the seagrass Zostera marina (Ralph and Short, 2002). In L. digitata co-cultivated with the endophyte, we also observed differential expression of photosynthesis-related genes. The down-regulation of a light harvesting protein gene lhcf6 suggests photoinhibition, which may be counteracted by the up-regulation of a photosystem II reaction center protein D1 gene as a key step in the repair of photodamaged PSII (Murata et al., 2007). Indeed, PAM measurements did not indicate any impact on the performance of photosystem II of the two kelp species, during the 2 weeks of co-cultivation. According to our data, the presence of L. elsbetiae did not strongly affect its main host, while it had a significant impact on photosynthesis and growth physiology of L. digitata, its occasional host. The decreased growth rate could be due to the trade-offs between algal defense and growth metabolisms during biotic interactions, as already shown in land plants (Züst and Agrawal, 2017).

In our study, the endophyte-detection results showed that the association with its host tends to be lower for L. digitata than for S. latissima after 2 weeks of co-cultivation, indicating that the mechanism of early resistance to the infection with L. elsbetiae may differ between the two hosts. In addition to different regulations of growth or photosynthesis pathways, the RNA-seq data provide insights of the defense-related transcriptional mechanisms occurring during the first infection step. Only a small portion (0.4%) of both transcriptomes showed significant expression differences, as already observed during the early steps of biotic interactions such as grazing in brown algae (Flöthe et al., 2014; Ritter et al., 2017) or bacterial infection in red algae (de Oliveira et al., 2017). Distinct transcriptomic regulation patterns appeared between the two kelps species, according to time and intensity of gene regulation. After 24 h of co-cultivation with the endophyte, only few genes were differentially regulated in S. latissima, whereas L. digitata already featured a stronger transcriptomic response. Although most of the DEGs were down-regulated in both species, a small portion of genes in L. digitata were already highly up-regulated after 24 h. After 48 h, almost the same amount of DEGs were found in both kelp species. Our results therefore suggest that the first contacts with the endophyte induced a faster and stronger transcriptomic response in the occasional host, L. digitata, than in the main host, S. latissima. The majority of differentially expressed genes were unique in the two hosts and only twenty-one genes were commonly differentially expressed in both kelps. This confirms that overall the two kelps react differently to the contact with the endophyte.

The overall rate of functional annotations was very low, as it is usually the case for non-model organisms (Armengaud et al., 2014). It is therefore difficult to fully understand the molecular responses of the two kelps toward an infection with L. elsbetiae, and further investigations will be necessary to characterize the functions of some conserved or specific unknown genes. However, some interesting gene candidates were annotated that are related to defense responses. Putative respiratory burst oxidase homolog protein (rboh) genes were strongly upregulated in the L. digitata samples after 24 h of the addition of the endophyte (Figure 4A), and in S. latissima after 48 h, but with lower log2FC value (Figure 4B). Rbohs are involved in oxidative bursts of plants (Torres and Dangl, 2005) and have been shown to be induced by elicitors and other stress conditions in some macroalgae (Luo et al., 2015; Wang et al., 2018). The local oxidative burst is known to be the earliest conserved response following recognition of the attack(er) in the plant and macroalgal innate immunity systems and the reactive oxygen species produced by Rbohs act as toxic compounds and/or as defense signals (Potin et al., 2002). The comparison of the expression of the putative rboh genes in the two species suggests that the immunity responses in L. digitata could be much quicker and stronger than the one in S. latissima after the first contact with the endophyte. The 24-h delay of transcriptomic regulations in S. latissima might be due to the repression of the attack(er) recognition in this species. A gene encoding LRR receptor-like serine/threonine-protein kinase was indeed downregulated in S. latissima, and not in L. digitata, after 48 h of the co-cultivation with L. elsbetiae. LRR receptor-like serine/threonine-protein kinases play a central role in the signaling of pathogen recognition in land plants (Afzal et al., 2008). It has been shown that viruses and bacteria have developed mechanisms to suppress the expression of genes involved in pathogen recognition in plants before an infection (Stack et al., 2005; Akira et al., 2006; Boller and He, 2009). The downregulation of this gene in S. latissima could result in an incomplete or inaccurate recognition of L. elsbetiae as an external biotic attack, avoiding the induction of immunity defense responses and leading to stronger infection. In S. latissima, no other putative known defense-related genes were found to be upregulated. Based on transcriptomic regulations, a different scenario seems to occur in L. digitata, where several defense-related genes were upregulated. For instance, a putative carbohydrate sulfotransferase gene, upregulated after 48 h, might have an effect on the recognition between host and endophyte. Carbohydrate sulfotransferases act by adding sulfonyl groups from 3′phosphoadenosine-5′phosphosulfate (PAPS) to a glycoside receptor and the sulfated carbohydrate can serve as a source of extracellular biological information to cells, influencing recognition events and signaling pathway (Bowman and Bertozzi, 1999). Metacaspase belongs to a cysteine protease family and has an essential role in programmed cell death caused by the immunity response of plants (Spoel and Dong, 2012). The upregulation of these genes were also observed in terrestrial plants during the contact with pathogens (Hoeberichts et al., 2003; Jayaswall et al., 2016). Furthermore, a serine carboxypeptidase-like protein (SCPL) was upregulated in L. digitata. SCPLs comprise a large family of protein hydrolyzing enzymes that play roles in multiple cellular processes. In addition, some SCPLs are involved in the biosynthesis of antimicrobial compounds and disease resistance in land plants (Liu et al., 2008; Mugford et al., 2009). The SCPL we found might be also involved in the biosynthesis of defense-related compounds. Their strong upregulation in L. digitata suggests that - unlike S. latissima - L. digitata recognizes the endophyte as a thread and activates defense reactions and efficient immunity responses, which could explain the lower infection patterns in natural L. digitata populations (Bernard M. et al., 2018; Bernard M.S. et al., 2018). In young plantlets, the activation of defense reactions could consume energy, resulting in the decrease of growth rate during the first contacts with endophytes.

Above we hypothesized that the growth of L. digitata in the presence of L. elsbetiae was slowed down due to an activation of energy-costing defense reactions. Growth behavior of L. digitata in co-cultivation with the endophyte after GG elicitation, however, was similar to the controls. An elicitation with GG has been shown to strongly induce defense-related genes in L. digitata (Cosse et al., 2009). The pre-treatment could therefore restore normal growth behavior of the kelp in co-culture with the endophyte, due to the activation of the kelp defense reactions prior to the co-cultivation. Previously, resistance of L. digitata sporophytes against the filamentous algal endophyte L. tomentosoides has been induced by a GG pre-treatment (Küpper et al., 2002). The authors suggested that the oxidative burst caused by the addition of GG activated secondary, long-term defense mechanisms in the kelps that in a second time lead to a strengthening of their cell wall, thereby building up a mechanical barrier against the endophyte. In the field, the protection of juvenile sporophytes by GG elicitation was also observed in S. latissima, which reduced the density of endobionts and the number of bacterial cells on sporophytes (Wang et al., 2019).

Our experimental set-up, on the other hand, was rather monitoring the initial steps of kelp-endophyte interactions, mainly during the spore settlement and germination. The results suggest rapid and direct defense mechanisms that may have been enhanced through a GG-induced priming effect, as already observed in L. digitata (Thomas et al., 2011). Future studies using the qPCR bioassay could test a potential effect on long-term resistance of the kelp against the endophyte.

In the co-cultivation with L. elsbetiae, many cell-wall modification-related genes were upregulated in S. latissima. Five mannuronan C-5 epimerase (MC5E) were upregulated in S. latissima. MC5E is catalyzing the last step of alginate biosynthesis, i.e., the conversion of mannuronic to guluronic acids (Michel et al., 2010). The upregulation of MC5E indicates activation of alginate synthesis and modification of the cell wall toward strengthening in S. latissima upon co-cultivation (Figure 4B). Other genes associated with cell-wall metabolism were endo-1,3-beta-glucanase and alginate lyase. Endo-1,3-beta-glucanase catalyzes the hydrolysis of 1,3-beta-D-glucosidic linkages in callose, laminarin, and various carbohydrates found in the cell wall of plants (Casu et al., 2007). Alginate lyase can degrade alginate by cleaving the glycosidic bond through a β-elimination reaction, generating an oligomer with a 4-deoxy-L-erythro-hex-4-enepyranosyluronate at the non-reducing end (Kim et al., 2011). The functions of both genes are related to the degradation of the cell wall. The fact that these enzymes were upregulated 48 h after the infection with the endophyte suggests an enhanced activity of either decomposing the host cell wall or degrading the endophyte cell wall. Two DEGs in S. latissima, imm upregulated 3 and EsV 1–7, were previously identified as members of a putative regulatory cascade with their potential life-cycle-related roles in E. siliculosus (Peters et al., 2008; Macaisne et al., 2017). The strong up-regulation of these two genes might lead to the regulation of the cell-cycle in order to repair damaged cell wall, which was also observed in the wounding responses of land plants (Dombrowski et al., 2020). The other gene regulations occurring after 48 h could be related to the accumulation and perception of cell wall degradation products, such as oligoalginates (Figure 4B). Indeed, some stress-response related genes were upregulated in S. latissima such as vanadium dependent bromoperoxidases (vBPO) and vanadium dependent iodoperoxidases (vIPO) involved in halide metabolism of brown algae and upregulated during abiotic and biotic oxidative responses (Cosse et al., 2009; Strittmatter et al., 2016; Salavarría et al., 2018). Other upregulated genes were putatively involved in calcium and oxylipin signaling pathways, two key pathways activated upon biotic and abiotic stress (Knight, 1999; Eckardt, 2008; Zhang et al., 2014).