Samples for quantitative real-time PCR (qPCR) were grown from April to June 2015 in a field owned by Nagoya University, Japan. Twenty Japanese cypress (C. obtusa) saplings (about 100 cm in height) were planted in plastic pots filled with a mixture of red soil and compost. The saplings were loosely fixed to a stake using wire to maintain vertical stem growth. Four of these saplings were grown vertically to generate normal wood. In order to prepare compression wood of different levels of development, we grew 16 other saplings, with four saplings each being grown at four different tilt angles (5˚, 10˚, 20˚, and 30˚). Sampling was conducted during the most active period of cambial growth in June. After removing the bark, the stems of the saplings were cut into segments and the differentiating xylem tissues in the cut portions were harvested using a chisel. The lower sides of the stems (compression wood) were collected from the inclined saplings, and both sides of the stems were collected from the vertical saplings (normal wood). Immediately after harvesting, these tissues were frozen in liquid nitrogen and stored at −80˚C for total RNA extraction. We also prepared samples to be used for tissue observations for measuring circularity of the transverse section of the xylem tracheid. After harvesting the differentiating xylem for total RNA extraction, we sampled the stem by using a saw to cut the surface where the tilt angle of the stem was constant in each sapling.

Samples for immunolabeling and protein extraction were grown from April to June 2016. Twenty Japanese cypress (C. obtusa) saplings (about 100 cm in height) were grown at the same condition of a previous experiment. Eight of these saplings were grown at an angle (i.e. with non-vertical stems) to generate compression wood. The other twelve saplings were grown vertically to generate normal wood. Each of the saplings with compression/normal wood was used for immunolabeling. Selected pieces of stem that included differentiating compression/normal wood were harvested from the saplings and cut with a razor blade into small square blocks measuring several millimeters on each side. The other saplings with compression/normal wood were used for protein extraction. The differentiating xylem tissues were collected using the same harvesting method for total RNA extraction. Immediately after harvesting, these tissues were frozen in liquid nitrogen and stored at −80˚C for protein extraction.

Total RNA was extracted from 50 mg of differentiating xylem sample using an RNeasy Plant Mini Kit (Qiagen, Valencia, CA, USA) according to the manufacturer’s protocol. The RNA concentration was estimated spectrophotometrically using Gene Quant (Amersham, Buckinghamshire, UK). The RNA samples were treated with DNase I (TaKaRa Bio, Otsu, Japan) to remove contaminating genomic DNA. The removal of genomic DNA was confirmed by agarose gel electrophoresis after PCR amplification of 5 ng of the RNA samples. Total RNA was converted into cDNA using PrimeScript RT Master Mix (TaKaRa Bio). The cDNA products were used as a template in the qPCR. Based on the contig sequence homologous to the PME gene obtained by analysis of gene expression using a next-generation sequencer [27], gene-specific primers were designed using Primer 3 Plus (http://www.bioinformatics.nl/cgi-bin/primer3plus/primer3plus.cgi/) [28]. The primer sequences were as follows: Hinoki_PME_F, 5’-TGTCAGAGCCAGTGAAGAGAGAG-3’; Hinoki_PME_R, 5’-ATCGCCACAAACACAAGGAG-3’. The quantitative reaction was performed on a StepOnePlus Real Time PCR System (Life Technologies, Carlsbad, CA, USA) using the POWER SYBR Green PCR Master Mix (Life Technologies). The reaction mixture (20 μL) contained 2× POWER SYBR Master Mix, 0.2 μM each of the forward and reverse primers, and 5 ng of template cDNA. PCR amplification was performed under the following conditions: 95˚C for 10 min, followed by 40 cycles at 95˚C for 15 s and 58˚C for 60 s and was performed three times per sample. Gene expression was normalized against ubiquitin as an endogenous gene [27] [29]. Relative gene expression was calculated using the comparative CT method [30].

Harvested stem segments were fixed in 3% glutaraldehyde in 67 mM phosphate buffer (pH 7.0) for 1 week at 4˚C. Subsequently, fixation in phosphate buffer was repeated. Transverse sections, 10 μm thick, were prepared from the segments using a sliding microtome. The sections were stained with 1% safranin and dehydrated in an increasing ethanol series. After soaking in xylene, the sections were mounted on glass slides with EntellanNeu (Merck, Darmstadt, Germany) and observed under a light microscope (BX60; Olympus, Tokyo, Japan). The observations were captured using a camera (DP70; Olympus, 1360 × 1024 pixels). Ten cells were selected for each sample, and the area and the perimeter of the intracellular cavity of the transverse section of the xylem tracheid were measured using the image processing software ImageJ. Following, the circularity (R) was calculated. Circularity is a parameter showing how close a shape is to a circle, and as circularity approaches 1, the closer a shape is to a true circle. Circularity is defined by the following formula:

R = 4 π S L 2

(S is area, L is the perimeter).

To create the anti-PME antibody, we used cDNA cloning and sequencing to identify the accurate base sequence of the PME gene that was expressed in differentiating xylem cells. The reverse transcribed cDNA of total RNA obtained from differentiating xylem cells of C. obtusa normal wood was amplified by PCR. Primers to amplify the Open Reading Frame (ORF) region of the contig sequence were designed using Primer 3 Plus. The primer sequences were as follows: Hinoki_PME_ORF_F, 5’-AATTAACGGCAAGCAAATCG-3’; Hinoki_PME_ORF_R, 5’-GAGTGCGCGAGTATTCCTTC-3’. We then amplified the ORF region using the primer pair and TaKaRa Ex Taq® Hot Start Version (TaKaRa). The amplified cDNA fragments were inserted into the pGEM-T Easy Vector (Promega, Madison, WI, USA) using T4 DNA Ligase, and their sequences were determined. The new sequences were submitted to the DNA Data Bank of Japan (Accession number: LC348962).

We grinded portions (5 - 10 g) of the differentiating xylem in liquid nitrogen to produce a powder. To obtain the soluble and ionically bound proteins, we extracted the differentiating xylem overnight in 50 mM sodium acetate-acetic acid buffer (pH 5.0) containing 1 M CaCl₂, polyvinyl pyrrolidone (3% of sample weight), 0.5 mM phenylmethylsulfonyl fluoride, and 5 mM β-mercaptoethanol at 4˚C. After extraction, the buffer was exchanged with 50 mM sodium acetate-acetic acid buffer (pH 5.0) by dialysis. The protein was concentrated using a Vivaspin 20 sample concentrator (GE Healthcare, Buckinghamshire, UK).

We determined protein concentrations by the Bradford procedure using bovine serum albumin (BSA) as the standard [31].

Rabbit anti-PME immunoglobulin G (IgG) antibody was designed based on the amino acid sequence of the PME protein (ASEGSNGNEN); the peptide of this amino acid sequence was used as the antigen to raise the rabbit IgG antibody. To absorb the anti-PME antibody into the antigen, we mixed the anti-PME antibody and antigen (peptide of ASEGSNGNEN). Following, we incubated the mixture at 4˚C for 2 days. The proteins extracted from the differentiating xylem were separated by sodium dodecyl sulfate-PAGE (SDS-PAGE) (10%) and transferred to a polyvinylidine difluoride membrane (Fluorotrans W-F; Port Washington, NY, Pall Corporation). After washing the membrane in 20 mM Tris-HCl (TBS), pH 7.6, containing 0.1% (v/v) Tween 20 (TBS-T) buffer, we immersed it in TBS-T containing 2% (w/v) ECL Blocking reagent for 1 h at room temperature to block nonspecific antibody binding and washed in TBS-T. The membrane was immersed overnight in anti-PME antibody or absorbed anti-PME antibody diluted 100-fold with Can Get Signal Solution 1 (TOYOBO, Osaka, Japan) at 4˚C, and then washed thrice for 20 min in TBS-T. Subsequently, we immersed the membrane for 4 h in anti-rabbit IgG, HRP-linked whole Ab goat (MBL, Aichi, Japan) diluted 20,000-fold with Can Get Signal Solution 2 at room temperature and washed thrice for 20 min in TBS-T containing 0.5 M NaCl and 0.02% SDS. ECL Prime Western Blotting Detection Reagent (GE Healthcare) was used to generate chemiluminescence signals, which were detected with an LAS-1000 plus luminescent image analyzer (FUJIFILM, Tokyo, Japan).

Square stem blocks, each several millimeters on a side, were frozen rapidly by immersion in liquid chlorodifluoromethane (HCFC-22) cooled with liquid nitrogen. The blocks were transferred into an acetone solution containing 0.7% glutaraldehyde cooled to −80˚C, and then incubated for > 3 days to obtain a frozen substitute. The blocks were incubated at −20˚C overnight, then at 4˚C overnight. After equilibrating the blocks to room temperature, we immediately washed them thrice for 10 min in acetone. Subsequently, the block was immersed thrice for 1 h in 100% ethanol to replace acetone with ethanol. After this, the block was immersed overnight in a solution of ethanol:LR White resin = 1:1. They were embedded in LR White resin (London Resin Co., Basingstoke, UK). The resin was cured at 50˚C overnight in airtight gelatin capsules.

Immunolabelling was conducted according to procedures described previously [32] [33]. Transverse thin (0.5 µm) sections for immunofluorescence light microscopy were cut using a rotary microtome (HM 350; MICROM, Walldorf, Germany) with a diamond knife and then mounted on MAS-GP typeA slides (Matsunami, Osaka, Japan). Each section was first immersed in 1% (w/v) BSA/TBS-T for 1 h to block nonspecific binding. The sections were then washed thrice for 15 min in TBS-T.

The anti-PME antibody used for western blotting was employed for immunolabeling. Each section was incubated in the anti-PME antibody diluted 10-fold in 1% BSA/TBS-T at 4˚C for 2 days, then washed thrice for 15 min in TBS-T. The sections were incubated in the goat anti-rabbit IgG secondary antibody, Alexa Fluor 647 (Thermo Fisher Scientific, MA, USA) diluted 300-fold in 1% BSA/TBS-T at 35˚C for 4 h. We shaded the sections from light during and after this treatment. The sections were washed thrice for 15 min in TBS-T and mounted in Fluoromount/Plus medium (Diagnostic BioSystems, CA, USA). We detected Alexa 647 fluorescence in the sections by confocal laser scanning microscopy (Fluoview FV10i; Olympus, Tokyo, Japan). Controls were prepared by replacing with 1) 1% BSA/TBS-T or 2) solutions of the anti-PME antibody absorbed into the antigen (diluted 10-fold).

The monoclonal antibodies LM19, LM20 (PlantProbes, Leeds, UK) were also employed for immunolabeling of homogalacturonans. Each section was incubated in LM19 or LM20 diluted 10-fold in 1% BSA/TBS-T for 2 days at 4˚C, then washed thrice for 15 min in TBS-T. The sections were incubated in the goat anti-rat IgM secondary antibody, Alexa Fluor 647 (Thermo Fisher Scientific) diluted 100-fold in 1% BSA/TBS-T at 35˚C for 4 h. The subsequent procedure was similar to the immunolabeling procedure of the anti-PME antibody. Controls were prepared by replacing with 1) 1% BSA/TBS-T or 2) solutions of the LM19 and LM20 absorbed into their respective antigens (LM19 diluted 10-fold in a solution of 1 mg of polygalacturonic acid per mL, LM20 diluted 10-fold in a solution of 1 mg of pectin per mL) [34].

We measured the circularity of transverse sections of differentiating xylem tracheid in samples grown at five different tilt angles of the stem. As a result, we observed circularity in transverse sections of the tracheid, which is characteristic of compression wood, in samples from stem tilt angle greater than 10˚ (Figure 1(c)-(e)). Circularity of transverse sections of the xylem tracheid increased with

greater tilt angles of the stem. For each sample, we then compared the PME gene expression level in differentiating xylem cells quantified by quantitative real-time PCR and the circularity of the transverse section of the xylem tracheid. The results showed greater circularity of the transverse section of the xylem tracheid for samples with relatively low level of PME gene expression (Figure 2). For samples with average circularity near 0.95, i.e. samples with circular transverse sections of the xylem tracheid, the relative level of PME gene expression was less than 1/10 of that observed in samples with a square-shaped transverse section of the xylem tracheid.

To determine whether the anti-CoPME antibody could specifically recognize the C. obtusa PME, we performed western blotting of the anti-CoPME antibody against proteins extracted fro/m the differentiating xylem of normal and compression wood. The anti-CoPME antibody reacted to proteins extracted from differentiating xylem of C. obtusa. We detected two signals from proteins extracted from normal wood (Figure 3(a)). A strong signal was seen in the molecular weight range of 55 and 60 kDa, while a weak signal was observed between 37 and 39 kDa. Similar results were observed from proteins extracted from compression wood (Figure 3(b)). However, the anti-CoPME antibody absorbed by antigen did not react to proteins extracted from the differentiating xylem of C. obtusa (Figure 3(a), Figure 3(b)).

The anti-CoPME antibody was used for immunolabeling to observe the transverse

section of differentiating xylem tracheid from the cambium to the xylem tracheid. In sections of normal wood, labeling was seen in the compound middle lamella from the cambium to the expansion zone (Figure 4(b)). There was labeling across the entire perimeter of the cell wall after the expansion zone ended and thickening of the secondary wall started. There was also labeling in the cell corner. Furthermore, in sections of compression wood, there was sparse labeling on the cell walls from the cambium to the mature cell (Figure 5(b)). From the

cambium to the mature cell, the sections of normal wood showed more intense labeling than compression wood. In both types of sections, there was no labeling in the secondary wall at the stage when thickening was completed. When the anti-CoPME antibody absorbed with antigen was used as a control, there was no labeling either in sections of normal wood or compression wood (Figure 4(d), Figure 5(d)). There was no labeling in either sections of normal wood or compression wood even when using a control with the anti-CoPME antibody removed (Figure 4(f), Figure 5(f)).

The anti-homogalacturonan antibodies LM19 and LM20 were used for immunolabeling to observe the transverse section of differentiating xylem tracheid from the cambium to the xylem tracheid. LM19 recognizes low methylated homogalacturonan, while LM20 recognizes highly methylated homogalacturonan. Observation of normal wood sections using the LM19 antibody showed labeling in the compound middle lamella from the cambium to the expansion zone (Figure 6(b)). There was particularly intense labeling in the radial wall. After the end of the expansion zone and start of the thickening of the secondary wall, labeling in the radial wall weakened, and there was no labeling in the tangential wall. No labeling was observed in the cell corner. There was no labeling in the secondary wall at the stage when thickening was completed. Furthermore, in compression wood sections, the middle lamella between the cambium and expansion zone showed weaker labeling than in sections of normal wood (Figure 7(b)). After the end of the expansion zone and start of thickening of the secondary wall, similar to normal wood sections, labeling in the radial wall weakened, while there was no labeling in the tangential wall of sections of compression wood. Using the LM19 antibody absorbed with 1 mg/mL polygalacturonic acid solution, neither normal wood nor compression wood sections showed labeling (Figure 6(d), Figure 7(d)).

When the LM20 antibody was used to observe normal wood sections, there was labeling in the middle lamella between the cambium and expansion zone (Figure 8(b)). There was particularly intense labeling in the radial wall. After the end of the expansion zone and start of the thickening of secondary wall, there was labeling in the radial wall, but labeling in the tangential wall had weakened. There was intense labeling in the cell corner. There was no labeling in the secondary wall at the stage when thickening had completed. Similar labeling patterns as normal wood sections were seen in compression wood sections (Figure 9(b)). Moreover, using the LM20 antibody absorbed with 1 mg/mL pectin solution, neither normal wood nor compression wood sections showed labeling (Figure 8(d), Figure 9(d)). Neither normal wood nor compression wood sections showed any labeling using a control that had LM19 and LM20 removed (Figure 6(f), Figure 7(f)).

The graph, which shows the relationship between relative PME gene expression

level in C. obtusa differentiating xylem cells and the average circularity of the transverse section of xylem tracheid, indicated that lower PME gene expression levels are correlated to greater circularity of the transverse section of the xylem tracheid. Therefore, circularity of the transverse section of xylem tracheid may be caused by expressional inhibition of the PME gene. In this study, we have quantified the gene expression levels by sampling differentiating xylem cells from the region between the cambium to the expansion zone. For this reason, we believe that expression of the PME gene may have been suppressed prior to circularity of the transverse sections of the tracheid.

We used Western blotting to test the specificity of the anti-CoPME antibody. In the normal wood sample and compression wood sample, we detected a strong signal near molecular weights of 55 - 0 kDa. We converted the cDNA sequence of the PME gene functioning in the differentiating xylem, obtained by sequencing, to an amino acid sequence and analyzed the sequence using the EXPazy-ProtParam tool (https://web.expasy.org/protparam/). This revealed that the total molecular weight of this protein was at least 53 kDa. Therefore, this suggests that the anti-CoPME antibody may have recognized and bound to the PME present in the differentiating xylem.

Furthermore, in both the normal wood sample and compression wood sample, we detected a signal near molecular weights of 37 - 39 kDa. We believe that this signal recognized the PME cleaved at a certain site in the amino acid region by proteases. However, it is also possible that the anti-CoPME antibody is non-specifically bound to a different protein with a similar sequence to the peptide sequence (ASEGSNGNEN). In order to identify the protein detected by the anti-CoPME antibody, it is necessary to recover the protein bound to the anti-CoPME antibody and investigate the amino acid sequence of said protein. However, anti-CoPME antibody absorbed by antigen did not recognize any protein extracted from the differentiating xylem. For this reason, we believe that the signal detected was due to recognition of PME present in the differentiating xylem by the anti-CoPME antibody.

Through immunolabeling using the anti-CoPME antibody, we examined CoPME localization in transverse sections of the differentiating xylem tracheids. CoPME labeling was observed in the compound middle lamella between the cambium and expansion zone in sections of normal wood and compression wood. Labeling was also mainly observed in the compound middle lamella, from the expansion zone to the mature cell. As a result, we inferred that CoPME was present in the compound middle lamella between the cambium and mature cells. It has been shown that in the differentiating xylem of Populus euramerica branches, PME is present in the compound middle lamella of the radial wall and in the cell corner [35]. This study produced similar results. When immunolabeling was carried out using anti-CoPME absorbed with antigen, there was no labeling. Therefore, we believe that the labeling indicated CoPME recognition present in the differentiating xylem tracheids by the anti-CoPME antibody.

Through immunolabeling using the anti-homogalacturonan antibodies LM19 and LM20, we examined homogalacturonan localization in transverse sections of the differentiating xylem tracheids. LM20 recognizes highly methylated homogalacturonan, which serves as a substrate when PME is functioning. LM19 recognizes low methylated homogalacturonan, which is produced when PME functions.

LM20 labeling was observed in compound middle lamella between the cambium and expansion zone, in both normal wood and compression wood sections. Labeling was also observed in compound middle lamella of the radial wall, from the expansion zone to the mature cell. This result suggested that highly methylated homogalacturonan, which is the substrate when PME is functioning, is present in abundance in the compound middle lamella between the cambium and mature cell. In the differentiating xylem of Populus euramerica branches, highly methylated homogalacturonan is reportedly present in the middle lamella of the radial wall and tangential wall in the cambium [34] [36], and similar results were observed in this study. No labeling was observed when immunolabeling was performed using the LM20 antibody absorbed with pectin solution. Therefore, we believe that labeling showed recognition of highly methylated homogalacturonan present in the differentiating xylem tracheids by the LM20 antibody.

LM19 labeling was observed in compound middle lamella between the cambium and expansion zone in both normal wood and compression wood sections. The intensity of labeling weakened from the expansion zone to the mature cell. In the differentiating xylem of Populus euramerica branches, low methylated homogalacturonan is reportedly weakened significantly as the secondary wall was thickening, and similar results were observed in this study [36]. LM19 antibody may not have recognized the low methylated homogalacturonan, which composed the pectin gel that functions as an adhesive between cell walls. In the differentiating xylem of Populus euramerica branches, low methylated homogalacturonan is reportedly present in the intermediate layer of the radial wall in the cambium [34], but this study did not produce similar results. No labeling was observed when immunolabeling was performed using the LM19 antibody absorbed with polygalacturonic acid solution. Therefore, we believe that labeling indicated recognition of low methylated homogalacturonan present in the differentiating xylem tracheids by the LM19 antibody.

It is also conceivable that highly methylated homogalacturonan is present in greater quantity than low methylated homogalacturonan in the cell walls of mature cells. In mature cells, highly methylated homogalacturonan may be the majority form of homogalacturonan present, and low methylated homogalacturonan may only be present in small amounts. However, the result in this study, in which far more labeling of highly methylated homogalacturonan than low methylated homogalacturonan in the compound middle lamella was found, was similar to the observations made from immunohistochemical staining of homogalacturonan in mature xylem of Pinus sylvestris [37]. Several hypotheses have been proposed to explain the greater amount of highly methylated homogalacturonan than low methylated homogalacturonan in cell walls after lignification. It has been reported that the presence of Ca²⁺, which is said to crosslink the carboxyl groups in low methylated homogalacturonan, causes one-electron oxidation of coniferyl alcohol and is involved in the lignification process [38]. In other words, the liberation of Ca²⁺ from homogalacturonan might be necessary for lignification to start after the end of the expansion zone. It has also been reported that in the presence of Ca²⁺, low methylated homogalacturonan has affinity to isoperoxidase, which is responsible for polymerization of lignin [39]. These reports suggested that liberated Ca²⁺ and low methylated homogalacturonan may be involved in lignification of the cell wall.

When the distribution of localized CoPME was compared between normal wood and compression wood sections, we observed that normal wood had more intense labeling than compression wood in the compound middle lamella from the cambium to expansion zone. This result suggested that the amount of CoPME present in the differentiating xylem of normal wood was greater than compression wood. This result also supports the relationship between increased circularity of transverse sections of the tracheid and decreased PME expression levels in

differentiating xylem cells. Further, we observed labeling around the entire perimeter of the cell wall in normal wood sections. Conversely, compression wood sections did not show labeling around the perimeter of the cell wall and, instead, showed sparse labeling. This result suggested that while CoPME was present all around the cell wall of normal wood, the same was not true for compression wood. We believe that cell shape becomes circular according to the following process of cell wall formation in compression wood. In places where CoPME functions, pectin gel formation allows sufficient adhesion between cell walls. However, in places where CoPME does not function, we believe that there is inadequate adhesion between cell walls. At the end of the expansion zone, the volume of the cell decreases due to a decrease in the turgor pressure of the tracheid. Further, due to moisture shrinkage of the tracheid, the adhesion begins to peel off in places of inadequate adhesion between cell walls, resulting in cell gaps and, thereby, generating a circular cell shape (Figure 10).