M. polymorpha of male accession Takaragaike-1 (Tak-1) was cultivated through thallus fragmentation on Jiffy pellets (Jiffy Products International AS, Stange, Norway) or maintained asexually through gemma growth on half-strength Gamborg’s B-5 basal medium (GB_1/2) with minimal organics medium (Sigma-Aldrich, Merck, Darmstadt, Germany) supplemented with 1% sucrose and 1.4% agar with a pH of 5.5. The plants were placed in growth chamber at 22°C under a 8 h dark/16 h light photoperiod and at 100 μE m^–2 s^–1 light intensity. The 2 week-old thalli were only grown in vitro. The 3 week-old thalli were grown for 2 weeks in vitro, and then for 1 week on Jiffy pellets. The 5 week-old thalli were grown for 4 weeks in vitro, and then for 1 week on Jiffy pellets. Only the tips of the thalli, i.e., the actively growing tissues, were used for the experiments.

About 3.5–4 g of fresh material were used for 2 week-old thalli, 5–7 g for 3 week-old thalli and 35–40 g for 5 week-old thalli. Extraction of CWPs started with purification of cell walls as described with slight modifications (Feiz et al., 2006). The plant material was placed on nylon mesh (25 μm pore size) and washed by 2 L of sodium acetate buffer 5 mM pH 4.6, then with 1 L of sodium acetate buffer 5 mM pH 4.6 containing 0.4 M sucrose. The plant material was ground at 4°C for 30 s in a blender containing 400 mL/g fresh material sodium acetate buffer 5 mM pH 4.6 containing 0.4 M sucrose supplemented with 20 mM DTT, 10 mM TCEP [Tris (1-carboxyethyl) phosphine hydrochloride] (Sigma-Aldrich), protease inhibitors cocktail (Sigma-Aldrich) and poly (vinylpolypyrrolidone) (PVPP, Sigma-Aldrich). This step was followed by centrifugation at 30,000 g for 15 min at 4°C. It was repeated twice, but successively with sodium acetate buffer 5 mM pH 4.6 containing 0.6 M sucrose and 1 M sucrose. The resulting pellet was washed on nylon mesh (pore size: 25 μm²) with 3 L sodium acetate buffer 5 mM pH 4.6. It was finally ground in liquid nitrogen using a mortar with pestle until obtaining a homogeneous fine powder. This fraction, so-called “purified cell walls,” was lyophilized until complete desiccation.

Extraction of proteins was done as described (Irshad et al., 2008). Twelve milliliter of 5 mM sodium acetate buffer at pH 4.6 containing 0.2 M CaCl₂ were added to 1 g lyophilized purified cell walls together with 9 μL/g protease inhibitors cocktail (Sigma-Aldrich). After 20 min of agitation, the mixture was centrifuged at 45,000 g for 30 min at 4°C and the supernatant was collected. The proteins were successively extracted with 5 mM sodium acetate buffer pH 4.6 containing (i) 0.2 M CaCl₂ (6 mL/g purified cell walls), (ii) 2 M LiCl (10 mL/g purified cell walls) in the presence of 10 μL of protease inhibitors cocktail, and (iii) 2 M LiCl (6 mL/g purified cell walls). All the supernatants were pooled and subjected to a last centrifugation step to eliminate cellular residues. The protein solution was desalted on Econo-Pack^® 10 DG columns (BIO-RAD, Hercules, CA, United States) equilibrated with 100 mM ammonium formate buffer. The protein solution was lyophilized and resuspended in deionized water. The proteins were quantified using the BCA method (INTERCHIM UPTIMA, Montluçon, France) and separated by 1 D-SDS PAGE.

This procedure was realized as described (Minic et al., 2007) with some modifications. The thalli (5 g) were frozen in liquid nitrogen and ground in a mortar with a pestle to a fine powder. The resulting powder was suspended in 50 mL of ice-cold extraction buffer (25 mM potassium phosphate buffer pH 7.4, 0.5 M NaCl, 20 mM DTT, 5 mM TCEP (Sigma-Aldrich), 1 g PVPP, 250 μL plant protease inhibitors cocktail (Sigma-Aldrich) and were ground in a mortar with pestle for 5 min on ice. The material was centrifuged at 40.000 g for 15 min at 4°C and the proteins were precipitated with ammonium sulfate (80% saturation) to concentrate the protein extract. The pellet was resuspended with 3 mL of ConA buffer (20 mM Tris-HCl pH 7.4, 1 mM CaCl₂/MnCl₂/MgCl₂, 0.5 M NaCl). The proteins were desalted on Econo-Pack 10 DG (BIO-RAD) columns. Then, the protein concentration was measured using the BCA method (INTERCHIM UPTIMA). Five mg of proteins were incubated with 1.5 mL ConA resin (20 mg/mL agarose conjugate) (Sigma-Aldrich) for at least 3 h at 4°C. Then, the ConA suspension was poured into a small plastic disposable column (Thermo Fisher Scientific, Rockford, IL, United States). The column was washed thrice with the ConA buffer, followed by elution with 0.2 M methyl-α-glucopyranose (Sigma-Aldrich) in the same buffer. The proteins were quantified using the BCA method (INTERCHIM UPTIMA) and separated by 1 D-SDS PAGE.

Sixty microgram of each protein extract were loaded on 12% polyacrylamide gel and migrated for 1 h at 35 V. Gels were stained overnight with Coomassie blue PageBlue™ Protein Staining Solution (Thermo Fisher Scientific). Prior to silver staining, this step was followed by washing of gels with UHQ water until the complete disappearance of the blue color. Silver staining was performed as described (Shevchenko et al., 1996).

The extraction procedure was designed after (Wang et al., 2016). One g of lyophilized powdered thalli was extracted with 25 mL 80% ethanol using an ultrasonicator (Elmasonic S ultrasonic, Elma Schmidbauer GmbH, Singen, Germany) for 30 min at 50°C. This step was repeated twice. After each extraction, the solution was vortexed for 10 min in dark. Both extracts were combined and filtered through a 0.45 μm membrane (Merck Millipore, Darmstadt, Germany). The filtrate was evaporated using a rotary evaporator at 55°C for 20 min. The dry residue (crude extract) was dissolved in 80% ethanol. The total phenolics content was determined according to the Folin and Ciocalteu method (Kaur and Kapoor, 2002) with some modifications. Twenty microliter from each sample were added to 1.58 mL deionized water and 100 μL Folin and Ciocalteu’s phenol reagent (Sigma-Aldrich). After 2 min of mixing, 300 μL of sodium carbonate solution (0.2 g/L) were added and the mixture was homogenized. The samples were incubated at 40°C for 30 min. Finally, the absorbance was measured at 765 nm. The total phenolics content expressed as mg gallic acid/g dry mass was determined by comparison to a standard curve determined using various concentrations of gallic acid (0–500 mg/mL).

Sixty μg of each protein extract were solubilized in 30 μL Laemmli buffer (Laemmli, 1970). They were vortexed during 3 min prior being heated at 70°C during 5 min. They were shortly separated by 1D-SDS-PAGE at a constant voltage of 200 V during 12 min. The gels were stained with Coomassie blue (BioSafe™, BIO-RAD) according to the purchaser’s instructions except that a step of fixation in 10% acetic acid, 40% ethanol was added before the rinsing step. The gel was cut into three or six pieces for the proteins extracted from purified cell walls, or the N-glycoproteins, respectively. Each sample was then digested according to the in-gel digestion protocol described in Supplementary Table 1A. Then, the samples were analyzed by LC-MS/MS following the procedure described in Supplementary Table 1B. The mass spectrometer was a Q-Exactive Plus (Thermo Fisher Scientific) coupled to a NanoLC 425 System (Sciex, Villebon sur Yvette, France). The raw data produced via the XCalibur™ 4.0.27.19 software (Thermo Fisher Scientific) were converted using msconvert (Proteowizard 3.0.9576)² before the database search. The V3.1 version of the M. polymorpha genomic data was downloaded from Phytozome.³ Database search was perform using an in-house database collecting usual contaminants and X!Tandem Piledriver (2015.04.01.1) (Craig and Beavis, 2004; see Supplementary Tables 1B, 2, 3 for details). X!TandemPipeline (Release 0.4.2) (Langella et al., 2017) was used for MS data processing (see Supplementary Table 1B for details). The validation of the identification of a protein was only confirmed when at least two peptides were found with an e-value lower than 1.00E-02 and an overall e-value for the protein lower than 1.00E-05. Only proteins identified with at least two different specific peptides in the same biological replicate and found in at least two biological replicates were validated. The quantification of the CWPs extracted from the purified cell walls has been performed with MassChroQ (version 2.2.29) as described (Valot et al., 2011) using the XIC (extracted ion chromatogram) values. The MS/MS are publicly available at ProticDB (Langella et al., 2013;⁴ projects “cell_wall_marchantia_proteome” and “cell_wall_marchantia_glycoproteome”). The data related to CWPs can be found at WallProtDB (see text footnote 1) (San Clemente and Jamet, 2015).

Two types of bioinformatic predictions have been performed: the sub-cellular localization of proteins and the presence of functional domains using the ProtAnnDB in-house tool⁵ (San Clemente et al., 2009). Proteins having a predicted signal peptide, but lacking intracellular retention motifs or having less than two transmembrane domains have been called CWPs.

The usual procedure to isolate cell walls starts with a rough grinding of plant tissues (Feiz et al., 2006). At the beginning of our experiments with 5 week-old thalli, we have faced difficulties to avoid strong oxidation reactions which gave a dark brown to nearly black color to the extracts. It was not possible to get rid of these compounds even after gel filtration on Econo-Pack 10 DG columns. The presence of high amounts of phenolics in M. polymorpha tissues was previously described (Schwartner et al., 1996; Soriano et al., 2019). The risk was to favor their unspecific oxidative fixation on proteins and thus to prevent the proper extraction and the identification of proteins by MS and bioinformatics. We have quantified these compounds at the three different stages of development of M. polymorpha of interest (Figure 1): 2 week-old thalli undergoing active growth and exhibiting scarce rhizoids, 3 week-old thalli representing an intermediate stage of development with more rhizoids, and 5 week-old thalli carrying gemma cups. We have compared the results to those obtained with aerial organs of A. thaliana and B. distachyon which have both provided easy material for protein preparation as shown in previous studies (Douché et al., 2013; Hervé et al., 2016; Duruflé et al., 2017), The starting material (named “raw material”) was compared to purified deproteinized cell walls (Figure 2). The amount of phenolic compounds correlated with the age of M. polymorpha thalli showing an increase by 100% between 5 week-old thalli compared and 2 week-old ones, in both types of materials. By comparison, this amount was similar to that found in the leaves of B. distachyon and a little bit higher than that found in A. thaliana rosettes. The difference remained high between the values obtained with the purified deproteinized cell walls, especially with those of A. thaliana. Since our difficulties appeared to be specific to M. polymorpha, we assumed that (i) the phenolics compounds were less easily extracted during the cell wall purification procedure and remained at the subsequent steps of protein extraction and (ii) the oxidation reactions were more important in M. polymorpha cell wall extracts. Then, we have modified the cell wall purification procedure by adding an antioxidant, 10 mM TCEP, at the very beginning of the experiment. It should be mentioned that the addition of DTT alone did not allow solving this issue. Although the amount of phenolic compounds could not be reduced, the oxidative reactions could be prevented. The protein extracts then appeared in a light brown color.

We have used a two-step protocol to prepare the protein extracts from purified cell walls. It includes the preparation of a purified cell wall fraction followed by an extraction of proteins with CaCl₂ and LiCl salt solutions (Irshad et al., 2008). This means that only proteins weakly interacting or ionically bound to cell wall components have been extracted. The extraction of covalently bound proteins would have required a different strategy such as the use of trypsin on purified cell walls (Chen et al., 2015).

The amount of purified cell walls as well as that of the extracted proteins was quantified (Table 1 and Supplementary Figure 1). The yield of cell wall preparation (mass of purified cell walls per mass of fresh material) strongly increased from 2 to 5 week-old thalli, from 8.8 ± 1.8% to 23.3 ± 5.8%. Conversely, the yield of proteins extracted from cell walls with salt solutions (CaCl₂ 0.2 M followed by LiCl 2 M) decreased from 2 to 5 week-old thalli, from 90.9 ± 16.2 to 22.3 ± 0.3 μg proteins/g fresh material. Regarding the N-glycoproteins purified by affinity chromatography on ConA, the elution of the proteins was checked by 1D SDS-PAGE and three fractions (E1-E3) were retained for further analysis and protein quantification (Supplementary Figure 2). The yield of the purification procedure (mass of total proteins per mass of proteins eluted from the ConA column) was similar at all three stages of development, i.e., it ranged between 11 and 13% of the total protein amount.

The next step was to check the quality of the protein extracts. This was performed by analyzing equivalent amounts of proteins (60 μg) by 1D SDS-PAGE (Figure 3 and Supplementary Figures 3, 4). The patterns showed distinct protein bands, indicating no protein degradation, and they were similar for each set of biological replicates (proteins extracted from purified cell walls and N-glycoproteins), indicating a good reproducibility of the experiments (Supplementary Figures 3, 4). Note that it was necessary to stain the N-glycoproteins with silver nitrate, probably because the Coomassie blue staining was not efficient due to the N-glycosylations. The comparison of the different stages of development showed little variations between the protein patterns (Figure 3). Then, the protein extracts were ready for MS analysis and protein identification. After protein identification, all the identified proteins have been annotated to predict their sub-cellular localization (see section “Materials and Methods,” Supplementary Tables 2, 3). Only the proteins predicted to be secreted and not retained in an intracellular compartment were considered as CWPs. Based on XIC values, 191 out of the 385 CWPs extracted from purified cell walls could be quantified. A principal component analysis showed the good grouping of the three biological replicates for each developmental stage (Supplementary Figure 5).

The number of predicted N-glycosylation sites was determined for each set of CWPs to assess the efficiency of the ConA affinity chromatography (Supplementary Figure 6). The percentage of proteins with no predicted N-glycosylation sites was much higher for the CWPs extracted from purified cell walls (17 vs. 4%), than for the CWPs purified by ConA affinity chromatography. In addition, 36% of the CWPs extracted from cell walls contained one or two predicted N-glycosylated sites, whereas 35% of the CWPs purified by affinity chromatography on ConA contained two or three predicted N-glycosylated sites. These results showed a clear enrichment in N-glycosylated proteins for the CWPs purified by ConA affinity chromatography.

Altogether, 1239 different proteins were identified in the protein extracts from purified cell walls among which 376 (30.3%) were predicted to be CWPs (Table 2). The total number of identified proteins was the highest in the 2 and 3 week-old samples (966 and 942, respectively) and the lowest in the 5 week-old sample (764). Conversely, the proportion of CWPs was the lowest in the 2 and 3- week-old samples (29.5 and 28.3%, respectively) and the highest in the 5 week-old sample (43.1%).

The number of identified N-glycoproteins was lower with 427 different proteins, but the proportion of CWPs was higher (51.7%). The comparison between both proteomes showed that 189 proteins were identified in both of them whereas 35 CWPs were only identified in the N-glycoproteome (Supplementary Figure 6). Overall, 410 different CWPs were identified.

The 410 CWPs were distributed into the nine functional classes described in the Introduction according to the presence of predicted functional domains (Jamet et al., 2008; see section “Materials and Methods,” Supplementary Tables 2, 3). Considering the weak contribution of rhizoids and scales to the material analyzed compared to thalli, we have made comparisons with the already described cell wall proteomes of A. thaliana and B. distachyon leaves which have be obtained using the same protocol (see WallProtDB; see text footnote 1). Several features could be highlighted (Figure 4A): (i) the proportion of PAC was lower in M. polymorpha thalli than in A. thaliana and B. distachyon leaves (18.8 vs. 25.1 and 25.6%); the proportion of oxido-reductases (OR) was close to that found in B. distachyon, but higher than in A. thaliana (16.8 and 15.3 vs. 11.8%); the proportion of proteases (P) was lower than in A. thaliana and B. distachyon (9.8 vs. 13.9 and 17.4%); the proportion of proteins of unknown function (UF) was higher in M. polymorpha (13.2 vs. 10.5 and 6.0%); as well as that of M (18.5 vs. 11.2 and 13.5%).

In the PAC functional class, the major proteins families were: (i) the glycoside hydrolases (GH) with five GH16 (endoxyloglucan transferases), 15 GH17 (b-1,3-glucosidases), five GH18 (chitinases), seven GH19 (chitinases/lyzozymes), three GH5 (endo-b-1,4-glucanase), and two GH28 (polygalacturonases); (ii) the expansins (7); and (iii) the carbohydrate esterases (CEs) with six pectin methylesterases (PMEs, CE8) and two pectin acylesterases (PAEs, CE13). Among the OR, class III peroxidases (CIII Prxs) were predominant with 39 identifications, whereas 12 polyphenol oxidases were identified. The major proteins of the proteins related to lipid metabolism (LM) functional class were the lipases acylhydrolases of the GDSL family (23) whereas the non-specific lipid transfer proteins (nsLTPs) were less represented (4). Besides, most of the identified lectins were predicted to bind D-mannose (eight out of 12). Among the signaling proteins (S functional class), six fasciclin arabinogalactan proteins were identified. A closer look at the miscellaneous (M) functional class revealed the presence of a large proportion of germins (25), dirigent-like proteins (23) and phosphate-induced 1 proteins (phi-1/EXORDIUM-like) (10) compared to A. thaliana and B. distachyon, but a lower proportion of purple acid phosphatases (PAPs) (6) and thaumatins (4) (Figure 4B). Finally, 52 proteins with yet unknown functions were identified.

The cell wall proteomes obtained at the three developmental stages were compared: 2 week-old young actively growing thalli exhibited scarce rhizoids and scales whereas the 5 week-old thalli showed more rhizoids, more scales and gemma cups. It should be noted that only the tips of the thalli were analyzed, i.e., the tissues undergoing active growth. The quantitative data were used to refine the pattern of accumulation whenever possible (see Supplementary Figures 7–9 for a full description, and Supplementary Figure 10 for a few protein families).

Altogether, 253 CWPs were common to the three developmental stages, 20 were specific to the 2 week-old thalli (“early proteome”), 10 to the 3 week-old, and 61 to the 5 week-old (“late proteome”) (Figure 5). Overall, the distribution of the CWPs in functional classes was not modified between the three stages of development (not shown). To refine this comparison, a quantitative analysis was performed using the three biological replicates corresponding to CWPs. A heatmap calculated with the 191 CWPs which were quantified clearly showed the grouping of the biological replicates for each developmental stage (Figure 6 and Supplementary Figure 7). For each of the three developmental stages, a specific pattern of CWPs accumulation could be found. The cell wall proteomes of 2 and 3 week-old thalli segregated drastically from that of 5 week-old thalli. The latter broadly exhibited an opposite profile compared to that of 2 week-old thalli with the CWPs accumulated at a low level becoming accumulated at a high level, and vice versa. Six clusters of CWPs could be identified depending of their kinetics of accumulation: increasing or decreasing level of accumulation, transient increase or decrease (Supplementary Figure 8).

For different protein families, the distribution of the members followed a similar pattern as shown in Figure 7 for the dirigent-like protein family: some members were only present at 2 weeks, or gradually disappeared; others remained present during the time course of the experiment; and others were only identified at the latest stage of development. Other examples of this typical pattern are shown in Supplementary Figure 10 for the following protein families: GH17, D-mannose-binding lectins, GDSL lipases/acylhydrolases, phi-1/EXORDIUM-like proteins, CIII Prx, polyphenol oxidases, and germins.

Based on the number of specific peptides, a short list of the most abundant CWPs at the three stages of development could be established (Table 3). It comprises the only identified GH3 (exo-acting β-D-glucosidase), three PMEs, three CIII Prxs, four polyphenol oxidases, one PAP and one protein of yet unknown function. These proteins could perform particular basic functions in M. polymorpha cell walls.

In this study, we could identify 410 CWPs of M. polymorpha. To evaluate the coverage of the whole expected cell wall proteome, we have determined the size of a few gene families using the corresponding PFAM entries⁶ and we have compared it to the number of identified CWPs (Table 4). For three largest gene families (CIII Prx, germins and polyphenol oxidases, 189, 73, and 59 members respectively), the percentage of coverage was between 20 and 25%. For the other protein families, it was between 30 and 55%. Altogether, we can evaluate the coverage of the expected M. polymorpha cell wall proteome around 33%. The missing proteins could be (i) accumulated at other stages of development like during the reproductive cycle, (ii) present in minute amount and thus not detected by the used methods, or (iii) lost or not extracted with the used methods.