PIRIN2 suppresses S‐type lignin accumulation in a noncell‐autonomous manner in Arabidopsis xylem elements

Summary PIRIN (PRN) genes encode cupin domain‐containing proteins that function as transcriptional co‐regulators in humans but that are poorly described in plants. A previous study in xylogenic cell cultures of Zinnia elegans suggested a role for a PRN protein in lignification. This study aimed to identify the function of Arabidopsis (Arabidopsis thaliana) PRN proteins in lignification of xylem tissues. Chemical composition of the secondary cell walls was analysed in Arabidopsis stems and/or hypocotyls by pyrolysis–gas chromatography/mass spectrometry, 2D‐nuclear magnetic resonance and phenolic profiling. Secondary cell walls of individual xylem elements were chemotyped by Fourier transform infrared and Raman microspectroscopy. Arabidopsis PRN2 suppressed accumulation of S‐type lignin in Arabidopsis stems and hypocotyls. PRN2 promoter activity and PRN2:GFP fusion protein were localised specifically in cells next to the vessel elements, suggesting a role for PRN2 in noncell‐autonomous lignification of xylem vessels. Accordingly, PRN2 modulated lignin chemistry in the secondary cell walls of the neighbouring vessel elements. These results indicate that PRN2 suppresses S‐type lignin accumulation in the neighbourhood of xylem vessels to bestow G‐type enriched lignin composition on the secondary cell walls of the vessel elements. Gene expression analyses suggested that PRN2 function is mediated by regulation of the expression of the lignin‐biosynthetic genes.


Introduction
Lignin is one of the most abundant biopolymers on Earth, and has been widely studied because of its ecological, physiological and economic importance. In angiosperm species, lignin is mainly composed of guaiacyl (G), syringyl (S), and p-hydroxyphenyl (H) units, which are derived from lignin monomers coniferyl, sinapyl and p-coumaryl alcohol, respectively. The monomers polymerise by oxidative coupling either as a defence mechanism in response to pathogens or as part of a developmental programme such as during vascular development (Boerjan et al., 2003;Bonawitz & Chapple, 2010). The regulation of xylem lignification is known to involve transcriptional control of the monolignol biosynthetic genes by transcription factors as well as by the transcription-related Mediator protein complex (Zhao & Dixon, 2011;Hussey et al., 2013;Bonawitz et al., 2014).
Interestingly, lignification of xylem tracheary elements continues post mortem, that is after cell death. Early experiments by Pickett-Heaps (1968), Terashima et al. (1986) and Terashima & Fukushima (1988) showed that radioactively labelled lignin monomers were incorporated into cell walls of living tracheary elements, and that the incorporation continued even after the death of the cells. Lignin incorporation into dead tracheary elements was supported by experiments in Zinnia elegans (Zinnia) in vitro xylogenic cell cultures that demonstrated that the total lignin content of the cell cultures continued to increase even though all lignifying tracheary elements were dead (Hosokawa et al., 2001;Pesquet et al., 2013). In the same in vitro system, pharmacological experiments revealed that the post mortem lignification of the tracheary elements was enabled by the production and transport of both monolignols and reactive oxygen species from the nonlignifying parenchymatic cells (Ros Barcel o, 2005;Pesquet et al., 2013). Finally, Smith et al. (2013) showed convincing evidence in intact Arabidopsis plants on how the lignification of xylem vessel elements is at least partially accomplished by the function of the cells neighbouring the vessel elements. These findings have resulted in wide acceptance of the action of ʻgood neighbours' of lignification, whereby cells that normally do not themselves lignify assist in the lignification of their neighbouring cells. Therefore, even though lignification is clearly initiated in a cell-autonomous manner during the lifetime of the tracheary elements (including vessel elements), it continues post mortem in a noncell-autonomous manner. Post mortem, noncellautonomous lignification seems to occur also in the xylary fibres of Arabidopsis, while the interfascicular fibres seem to lignify mostly in a cell-autonomous manner (Smith et al., 2017). Furthermore, living vessel elements act as good neighbours to provide monolignols to adjacent xylary fibre cells (De Meester et al., 2018).
In an earlier study, we combined pharmacology and gene expression analyses in the Zinnia xylogenic cell cultures to identify genes that are potentially involved in the regulation of noncell-autonomous lignification (Pesquet et al., 2013). One such candidate gene encoded a CUPIN domain-containing PIRIN protein (gi:219988534; Pesquet et al., 2013). PIRINs (PRNs) comprise a family of proteins that are conserved between prokaryotic and eukaryotic organisms such as mammals and plants (Wendler et al., 1997), although they seem to be functionally diverse (Adams & Jia, 2005). The Arabidopsis PRN1 protein is involved in responses to UV light, blue light and abscisic acid, in seed germination and in seedling development (Lapik & Kaufman, 2003;Warpeha et al., 2007;Orozco-Nunnelly et al., 2014), while PRN2 interacts with a protease to regulate the susceptibility to a vascular pathogen (Zhang et al., 2014). In humans, the only PRN protein was identified on the basis of its interaction with Nuclear Factor I (Wendler et al., 1997), and has since then been demonstrated to act as a transcriptional co-regulator for several different transcription factors (Dechend et al., 1999;Liu et al., 2013). In this study, we demonstrate the function of Arabidopsis PRN2 in noncell-autonomous lignification of xylem tissues, and present data suggesting that the primary function of PRN2 is to facilitate enrichment of G-type lignin in the vessel elements, possibly through regulation of the expression of the lignin-biosynthetic genes from within the vessel-neighbouring xylem elements.
When the inflorescence stems reached a height of 50 cm, the 2cm bottom part was harvested for different analyses. After that, newly formed stems were constantly removed until the hypocotyls were harvested at indicated times.

Cloning and plant transformation
For promoter-GUS constructions, 2198-, 2308-, 527-and 2427bp fragments upstream of the translational start codons of PRN1, PRN2, PRN3 and PRN4, respectively, were cloned into the pDONR207 vector by Gateway recombination using BP Clonase II (Invitrogen), followed by recombination into pMDC163 destination vectors (Curtis & Grossniklaus, 2003) using LR Clonase II (Invitrogen). To generate PRN2 complementation lines, a 3838-bp fragment of the PRN2 genomic sequence, including the promoter and the coding region was cloned (for primers see Table S1) by Gateway recombination into the pGWB16 vector (Nakagawa et al., 2007) and transformed into the prn2-1 mutant. To generate the hybrid aspen (Populus tremula 9 tremuloides) proPttPRN2:GUS construct, a 845-bp fragment upstream of the translational start of the gene (Potri.002G231900) was cloned into pDONR207 (as above) and recombined into pKGWFS7 (Karimi et al., 2002). This binary vector was used for Agrobacterium-mediated transformation of hybrid aspen (Populus tremula 9 tremuloides) clone T89.
To create the proPttPRN2::PRN2-RNAi construct, an inverted repeat of a 300-bp fragment from the PttPRN2 coding sequence, targeting the beginning of the coding sequence, was placed downstream of the 845-bp PttPRN2 promoter. The entire proPttPRN2::PRN2-RNAi sequence was synthesised by Invitrogen, then inserted into the modified pKGW + T35S (the 35S promoter was removed) binary vector, which was used for Agrobacterium-mediated transformation of hybrid aspen T89.

Histochemical GUS staining
For histochemical GUS staining of Arabidopsis, in vitro grown seedlings or cross-sections of tissue samples obtained by hand sectioning from soil-grown Arabidopsis were used. Vibratome crosssections (70 µm) of the stems of hybrid aspen plants were taken 10 cm above the soil level. Samples were fixed in 90% acetone on ice for 30 min followed by incubation at 37°C in a solution containing 1 mM X-Gluc, 1 mM K 3 Fe(CN) 6 , 1 mM K 4 Fe(CN) 6 and 0.1% Triton X-100 in 50 mM sodium phosphate buffer (Na 2 HPO 4 /NaH 2 PO 4 , pH 7.2). The samples were destained in a 70% ethanol solution and then in a 99% ethanol solution. Light microscopy images were acquired using a Zeiss Axioplan II microscope equipped with an AxioCam charge-coupled device (CCD) camera (Zeiss, Jena, Germany).

Confocal laser scanning microscopy
Arabidopsis seedlings expressing proPRN2::PRN2:GFP were stained with 10 µg ml À1 propidium iodide for 2 h. The stained seedlings were imaged using a Zeiss LSM780 inverted confocal microscope.

New Phytologist
Pyrolysis-gas chromatography/mass-spectrometry (Py-GC/ MS) Freeze-dried stems and hypocotyls were ball milled (MM400; Retsch, Haan, Germany) at 30 Hz in stainless steel jars (1.5 ml) for 2 min with one ball (7 mm diameter). In total, 60 AE 10 lg of powder was then applied to the pyroliser (PY-2020iD and AS1020E; FrontierLabs, Koriyama, Japan) mounted on a GC/ MS (7890A/5975C; Agilent Technologies AB Sweden, Kista, Sweden). Pyrolysis was conducted at 450°C. The pyrolysate was separated on a capillary column with a length of 30 m, a diameter of 250 lm and a film thickness of 25 lm (J&W DB-5; Agilent Technologies Sweden AB). The gas chromatography oven temperature programme started at 40°C, followed by a temperature ramp of 32°C min À1 to 100°C, 6°C min À1 to 118.75°C, 15°C min À1 to 250°C, and finally 32°C min À1 to 320°C. Total run time was 19 min and full-scan spectra were recorded in the range of 35À250 m/z. Data processing including peak detection, integration, normalisation and identification was carried out as described in Gerber et al. (2012). Minimum three biological replicates, each consisting of a pool of three hypocotyls or stems, were analysed.

Phenolic profiling
For metabolic profiling, Arabidopsis plants were first grown in short-day photoperiod (21°C, 9 h : 15 h, light : dark) on soil in pots 5.3 cm wide 9 5.5 cm high. Plants were moved to a longday photoperiod (21°C, 16 h : 8 h, light : dark) after 8 wk to allow bolting. Plants were grown until the stem reached on average 42 cm in height. An 8 cm inflorescence stem section (between 1 and 9 cm, above the rosette) was harvested, frozen in liquid nitrogen and deprived of all side branches, for nine biological replicates. The material was ground while frozen and extracted with 1 ml methanol (high pressure liquid chromatography (HPLC grade)) at 70°C for 10 min. After centrifugation, methanol was evaporated and samples were suspended in 200 µl water : cyclohexane (1 : 1) for extraction. A 15-µl sample of the aqueous phase was subjected to ultra-high-performance liquid chromatography-mass spectrometry (UHPLC-MS) on a Waters Acquity UHPLC system (Waters Corp., Milford, MA, USA) connected to a Synapt HDMS Quadrupole Time-Of-Flight (Q-TOF) mass spectrometer (Micromass, Manchester, UK), basically as previously described (Vanholme et al., 2013). In brief, chromatographic separation was performed on a Waters Acquity BEH C18 column (2.1 mm 9 150 mm, 1.7 µm), with the mobile phase composed of water containing 1% acetonitrile and 0.1% formic acid (A) and acetonitrile containing 1% water and 0.1% formic acid (B), all v/v. During the gradient elution, a flow rate of 350 µl min À1 was applied, with initialisation at time 0 min, 5% B, 30 min, 50% B, and 33 min, 100% B. Negative mode MS-setting, chromatogram integration and alignment via Progenesis QI software (Waters) were as previously described (Vanholme et al., 2013). The integrated peak area of the m/z values (peaks) were normalised with the function 'normalise to total peak intensity'. The list of peak areas resulting from these parameters was imported in a digital spreadsheet (Microsoft Excel) for further analysis. To select peaks for which metabolites were significantly increased or decreased in prn2 mutants compared with wild-type plants, the following criteria were used: present in all samples of at least one genotype; average normalised abundance higher than 100 counts in at least one genotype; analysis of variance (ANOVA) P-value < 0.005 on arcsinh transformed values; t-test P-value < 0.01 on arcsinh transformed values; and two-fold increased/decreased peak area in prn2 vs wild-type. Annotation of compounds matching these criteria was based on accurate m/z (AE 0.02 Da), isotope distribution and tandem MS (MS/MS) similarities according to Morreel et al. (2010a,b). For MS/MS acquisition, two biological replicates per treatment were run on the UHPLC-MS in the negative mode.

Fourier transform infrared (FT-IR) microspectroscopy
FT-IR microspectroscopic measurements and data analysis were performed as described by Gorzs as et al. (2011). Briefly, 20-lmthick transverse sections were cryo-sectioned from the hypocotyls of 8-wk-old plants. Spectra were recorded using 32 scans at a spectral resolution of 4 cm À1 on a Bruker Tensor 27 spectrometer equipped with a Hyperion 3000 microscopy accessory (Bruker Optik GmbH, Ettlingen, Germany), including a liquid nitrogen cooled 64 9 64 mercury cadmium telluride focal plane array detector (Bruker Optik). Cell-specific spectra were extracted and converted to data point tables using OPUS (v.5.0.53Àv.7.0.129; Bruker Optik), and baseline corrected (four-point linear) and total area normalised in the region 900-1850 cm À1 , using custom-built scripts programmed within the MATLAB environment (v.7.0; MathWorks Inc., Natick, MA, USA), as described by Stenlund et al. (2008). Multivariate analysis was performed using SIMCA-P+ (v.13-v.14; Umetrics AB, Ume a, Sweden), on data from three biological replicates with 10 spectra per cell type from each. Orthogonal projections to latent structures discriminant analysis (OPLS-DA) models were calculated on centred data by fixed number of components (one predictive and two orthogonal for pairwise comparison), using the default settings of SIMCA-P for OPLS-DA models. According to Umetrics, a Q2 above 0.4 can be considered as a good model, as long as it is not 2 units below R2.

Raman microspectroscopy
Here, 20-µm-thick transverse sections were cryo-sectioned from the stems and the hypocotyls of 8-wk-old plants. Raman microspectroscopy (mapping) was performed as described by Gorzs as (2017) on a Renishaw InVia Raman spectrometer equipped with a CCD detector (Renishaw plc, Wotton-under-Edge, UK), using a 514 nm Ar + ion laser with 10-100% laser power and 1 s irradiation time set in the software (WIRE, v.3.0; Renishaw), in static mode centred at 1300 cm À1 (resulting in a spectral region of c. 625-1903 cm À1 with c. 1 cm À1 spectral resolution using a 2400 lines grating). A 950 magnification lens was used, with 2 µm step sizes (XY direction), in normal or high confocal modes, recording images with a minimum of 100 spectra.
Raman shifts were calibrated using the built-in Si standard of the instrument. Cosmic rays were removed and data noise filtered using the chemometrics package of WIRE (v.3.4,Renishaw). Spectra were exported as .txt files and preprocessed using the open-source MCR-ALS script (run in MATLAB environments, v.14a-v.18a, MathWorks), as provided by the Vibrational Spectroscopy Core Facility. Spectra were cut to the 630-1900 cm À1 region to remove noise filtering artefacts, and thereafter baseline corrected (using asymmetrical least squares (Eilers, 2004); k = 100 000 and P = 0.001), smoothed (Savitzky-Golay filter (Savitzky & Golay, 1964); polynomial order = 1 and frame rate = 5) and total area normalised. After preprocessing, spectra were extracted from voxels representing pure vessel-vessel or fibre-fibre cell walls, compiled in a single Microsoft EXCEL file and imported into SIMCA-P (v.13.0-v.14.0; Umetrics AB). Multivariate models were calculated using Principal Component Analysis (PCA) and the ʻAutofitʼ options in SIMCA-P to determine the number of significant components.

2D-NMR spectroscopy
2D-NMR analysis of acetylated cell wall material was performed as described by Hedenstr€ om et al. (2009) with a few modifications. Here, 200 freeze-dried and bark-peeled 8-wk-old hypocotyls were ball milled (MM400; Retsch) at 30 Hz in stainless steel jars (1.5 ml) for 2 min with one ball (diameter 7 mm). The resulting hypocotyl powder was prepared for NMR analysis by additional ball milling and acetylation (Hedenstr€ om et al., 2009). 2D 13 C-1 H heteronuclear single quantum coherence (HMQC) spectra (pulse program hmqcetgp) were acquired at 25°C on a Bruker DRX 600 MHz instrument (Bruker Biospin GmbH, Rhenistetten, Germany) equipped with a triple-resonance cryo-probe with z-gradients. Forty-eight transients, 320 t 1 increments and an interscan delay of 1 s resulted in an experiment duration of c. 5 h for each spectrum. Spectra were manually phase-and baseline-corrected using TOPSPIN 2.0 (Bruker Biospin). The residual CHCl 3 peak was used as an internal reference (d C 77.0 ppm, d H 7.27 ppm). Peaks originating from the aromatic rings in the syringyl (S2,6; Fig. 4d,e) and guaiacyl units (G2, Fig. 4d,e) were integrated and the S : G ratio was calculated using the formula S : G = (I S2,6 9 0.5)/I G2, where I S2,6 and I G2 are the volume integrals of peaks from S and G subunits, respectively. I S2,6 is scaled down by a factor 2 because it contains signals from two protons in the aromatic ring.

Gene expression analyses
Total RNA was isolated from hypocotyls and stems using an RNeasy Plant Mini Kit (Qiagen) according to the manufacturer's protocol. All samples were treated using an RNase-free DNase Set (Qiagen) in columns to remove any remaining DNA. cDNA was synthesised by reverse transcription using oligo(dT) and random primers and 1 µg RNA as a template. Semiquantitative RT-PCRs were performed using 59 diluted cDNA with Taq polymerase for 30 cycles. The Arabidopsis Actin1 (AT2G37620) was used as a reference gene. Quantitative RT-PCRs (qPCR) were performed using 409 diluted cDNA, iQ SYBR Green Supermix (Bio-Rad), 5 pmol of each primer, an iQ5 Multicolor Real-Time PCR Detection System (Bio-Rad) and the following programme: initial denaturation at 95°C for 3 min followed by 40 cycles of 10 s denaturation at 95°C and 30 s annealing at 55°C. The Arabidopsis PDF2 (AT1G13320) and UBQ10 (AT4G05320) genes were selected as housekeeping genes after assessment by geNORM analysis (Vandesompele et al., 2002). Both housekeeping genes were included in all qPCR analysis, their geometrical mean was calculated and used in further normalisation steps. Five biological replicates were used for each analysis. Primer sequences are listed in Table S1.

Statistical analyses
For all comparisons of means between lines/treatments and the wild-type/control, Welch's t-tests were performed assuming unequal variance. For multiple pairwise comparisons of all lines/ treatments within one and the same experiment, protected (F-test performed at a = 0.05) post-ANOVA Fisher's least significant difference (LSD) tests were used.

Accession numbers
Nucleotide sequence data from this article can be found in the Arabidopsis Genome Initiative under the following accession numbers:

PRN genes show unique expression patterns in Arabidopsis vascular tissues
We demonstrated earlier in xylogenic cell cultures of Zinnia that the expression of a PRN gene was associated with tracheary element lignification in a manner suggesting a role in noncell-autonomous lignification (Pesquet et al., 2013). The Arabidopsis genome contains four genes that encode PRN proteins: PRN1, PRN2, PRN3 and PRN4. Phylogenetic analysis revealed that PRN1, PRN2, PRN3 proteins, as well as the Zinnia PRN protein, are closely related to each other, while PRN4 is somewhat more distantly related to the other PRN proteins (Fig. S1).
First of all, we analysed promoter activities for the PRN genes by histochemical b-glucuronidase (GUS) assays in transgenic lines carrying each of the PRN promoters coupled with GUS. GUS staining revealed that the promoter of PRN1 is mainly active at the seedling stage and in epidermal cells of inflorescence

Research
New Phytologist stems, but not in the vasculature of the stem or the hypocotyl of mature plants (Fig. 1a-f). In contrast, the promoters of PRN2, PRN3 and PRN4 are active in vascular tissues at both seedling and mature plant stages (Fig. 1g-x). The PRN3 promoter is active both in the phloem and xylem tissues of the stem and the hypocotyl (Fig. 1p-r). Notably, the PRN2 promoter activity is restricted to xylem cells next to the vessel elements ( Fig. 1j-l), while the promoter for PRN4 is active in both developing xylem vessels and their neighbouring cells (Fig. 1v-x). Also, promoter activity of the PRN2 homologue in Populus (PttPRN2; Potri.002G231900) is localised specifically to ray cells neighbouring the vessel elements in transgenic hybrid aspen (Populus tremula 9 tremuloides) trees carrying a proPttPRN2::GUS construct (Fig. S2).
The activity of the PRN2 promoter specifically in xylem elements neighbouring the vessels supports the function of PRN2 in noncell-autonomous lignification of the vessel elements. We next analysed the localisation of the PRN2 protein in the roots of 4-d- old seedlings carrying a proPRN2::PRN2:GFP construct. Confocal laser scanning microscopy (CLSM) revealed localisation of the PRN2:GFP fusion protein in cells next to the vessel elements but no signal in the vessel elements themselves (Fig. 2). The PRN2:GFP fusion protein appeared in this location prior to the death of the vessel elements (Fig. S3). Altogether, both the PRN2 promoter activity and the PRN2 protein localisation support the noncell-autonomous function of PRN2 in vessel lignification.

PRN2 acts as a negative regulator of S-type lignin accumulation
To investigate if the Arabidopsis PRN proteins function in xylem lignification, hypocotyl material from 8-wk-old T-DNA insertion lines (Fig. S4) was analysed for each of the four PRN genes using pyrolysis-gas chromatography/mass spectrometry (Py-GC/ MS). Py-GC/MS yields pyrolytic products that can be used to quantify the relative content of carbohydrates and the three different types (G-, S-and H-types) of lignin based on their respective MS peak area as a proportion of the cumulative area from all peaks (Gerber et al., 2012). In this analysis, 127 pyrolytic degradation products were analysed in 14 mutants, including single mutants prn1, prn2-1, prn2-2, prn3, prn4-1, prn4-2, prn4-3 and prn4-4, as well as double mutants prn1 prn2-1, prn1 prn2-2, prn2-1 prn3, prn2-2 prn3, prn2-2 prn4-1 and prn1 prn4-1, and compared with the corresponding wild-type plants (Fig. 3). Interestingly, the two prn2 single mutants and the double mutants derived from crosses with the prn2 mutants showed significantly higher relative content of S-type lignin when compared with the Col-0 wild-type (Fig. 3e) Fig. 3 The effect of the different Arabidopsis PRN proteins on the secondary cell wall composition in hypocotyls. Relative content (percentage of detected cell wall components) of carbohydrates (cellulose and hemicelluloses) (a), total lignin (b), H-type lignin (c), G-type lignin (d) and S-type lignin (e), and the S-type : G-type lignin ratio (f) in the cell walls of 8-wk-old wild-type (WT) and single and double mutants for the PRN genes, analysed by pyrolysis-gas chromatography/mass spectrometry. All mutants are in Col-0 background, except for prn4-4 which is in Ler-0 background. Error bars indicate AE SD. For each genotype, five biological replicates were analysed, each composed of a pool of three hypocotyls. The asterisks indicate statistically significant differences from the corresponding WT background, tested by Welch's t-test (*, P < 0.05; **, P < 0.01; ***, P < 0.001).

Research
New Phytologist hypocotyl and stem material. Consistent with the increased relative content of S-type lignin in the prn2 mutants, Py-GC/MS analysis showed lower relative content of S-type lignin in the PRN2-overexpressing lines, compared with the wild-type (Figs  4a,b, S5). Py-GC/MS analysis on hypocotyls of three PRN2 complementation lines (proPRN2::PRN2:myc in prn2-1 background) showed partial complementation of the S-type lignin content (Fig. 4c). Furthermore, two-dimensional nuclear magnetic resonance spectroscopy (2D-NMR) on dissolved hypocotyl cell wall material (Lu & Ralph, 2003) revealed an increased S : G ratio in the prn2-1 mutant (S : G = 0.27) compared with wildtype (S : G = 0.16) (Fig. 4d,e). While the increase in the S-type lignin of prn2 was always consistent, increased relative content of G-type lignin as well as increased total lignin content was somewhat variably found in prn2 mutants compared with the wildtype (Figs 3, 4a,b). Transgenic hybrid aspen trees with up to 40% reduction in the expression of PttPRN2 showed also a slight tendency towards higher lignin S : G ratios (Fig. S2d-f). Taken together, the results support the action of PRN2 in suppressing S-type lignin accumulation.
The effect of PRN2 on lignin accumulation was corroborated by comparative phenolic profiling of the prn2 stem material using UHPLC-MS, which enables detection of several classes of aromatic compounds including benzenoids, phenylpropanoids, flavonols and oligolignols (Vanholme et al., 2012). In total, 6414 peaks were detected. The results showed 28 and 13 peaks with a significantly higher intensity in the prn2-1 and prn2-2 mutants compared with the wild-type, respectively, of which nine were common to both alleles. No compounds were found with a significantly lower abundance in both. The nine peaks with significantly higher intensity appeared to be derived from eight compounds. Based on MS/ MS fragmentation, seven of these (compounds 1-7) could be characterised as oligolignols (Table 1), containing at least one G unit and one S unit. The structure of the two other compounds could not be determined. The aromatic region of the 2D HSQC NMR spectra of wild-type (WT) (d) and prn2-1 (e). Two hundred bark-peeled hypocotyls from each of WT and the prn2-1 mutant were harvested and pooled to obtain adequate lignin extracts for the analysis. Peaks from the S-type and G-type lignin are indicated by ʻSʼ and 'G', respectively and the numbers following these letters correspond to the positions in the aromatic rings that give rise to these peaks.

PRN2 controls lignification in a noncell-autonomous manner
Next, we investigated whether the apparent localisation of the PRN2 protein next to the vessel elements affects cell wall chemistry of the neighbouring vessels and fibres. Transverse sections from the hypocotyls were analysed using Raman microspectroscopy, which enables chemotyping of individual cell walls with a micrometre lateral resolution (Gorzs as, 2017). PCA revealed that the cell wall chemotypes of both the vessels (measured from cell walls between two vessels) and the fibres (measured from cell walls between two fibres) were different between the prn2-1 mutant and the wild-type (Fig. S6). The cell walls of the PRN2 complementation line 5 carrying proPRN2::PRN2:myc in the prn2-1 background were more similar to the wild-type than to the prn2-1 mutant (Fig. S6). The prn2-1 mutant exhibited increased fluorescence during Raman microspectroscopic measurements in both vessels and fibres to a level that disabled identification of compounds contributing to the separation between the genotypes. Nevertheless, this was indicative of lignin-related changes in the cell wall matrix of prn2-1, as lignin is primarily responsible for the fluorescence in these samples. We therefore complemented Raman microspectroscopy with the analysis of the cell walls in transverse section of hypocotyls of the wild-type and prn2-2 mutant by FT-IR microspectroscopy combined with OPLS-DA. This method generally allows robust and sensitive chemotyping of individual cells (Gorzs as et al., 2011), but has lower (diffraction limited) spatial resolution than Raman microspectroscopy. Cell-specific FT-IR spectra were collected from cell walls of vessels and fibres located next to the predicted PRN2-expressing cells (Fig. 5a). Pairwise OPLS-DA comparisons of the FT-IR spectra showed separation between prn2-2 and wild-type in both cell types (Fig. 5b,c). Lignin-related differences could be tracked in the corresponding correlation-scaled loadings plots (Fig. 5d,e) via the band intensities of the aromatic -C=Cskeletal vibrations at 1595 and 1510 cm À1 (Faix, 1991). The intensity of the band at 1595 cm À1 (often correlated with Stype lignin) was increased in prn2-2 in comparison with the wildtype in both the vessels (Fig. 5b,d) and in the fibres next to the predicted PRN2-expressing cells (Fig. 5c,e), while the band at 1510 cm À1 (indicative of more cross-linked structures and often correlated with G-type lignin) was increased in prn2-2 in comparison with the wild-type only in the vessels (Fig. 5b,d). The intensity of an additional band at c. 1460 cm À1 , which indicates changes in =CH and/or -C-H vibrations originating from either lignin or noncellulosic polysaccharides (Olsson et al., 2011), was also increased in prn2-2 (Fig. 5d,e). These data support a function of PRN2 in suppressing especially the S-type lignin accumulation in a noncell-autonomous manner. Transverse sections of inflorescence stems were chemotyped by Raman microspectroscopy (Fig. S7). The cell wall chemistry of the interfascicular fibres did not differ between prn2-2 and wildtype. In contrast, increased intensities were observed in both the vessel elements and the fibres of the vascular bundles in prn2-2 compared with the wild-type for the typical lignin-related bands at 1595 and 1660 cm À1 . These increases and the slight shift in the band at 1595 cm À1 suggest structural or compositional changes in addition to an increase in lignin content in prn2-2. Thus, PRN2 modulates lignin deposition in the vascular bundles of the inflorescence stems in accordance with its expression pattern in these tissues.

PRN2 influences the expression of the lignin-biosynthetic genes
PRN protein functions in mammals as a transcriptional co-regulator. We have earlier demonstrated that PRN2 is localised in the cytoplasm and in the nucleus (Zhang et al., 2014), supporting possible function in transcriptional control. We therefore tested whether PRN2 function on S-type lignin accumulation could involve transcriptional control of the lignin-biosynthetic genes. The expression of genes encoding lignin-biosynthetic enzymes as well as secondary cell wall biosynthesis-related transcription factors were analysed in the hypocotyls and stems of the prn2 mutants and the PRN2-overexpressor lines. Increased expression was apparent in prn2 mutants for several lignin-biosynthetic genes, including C4H (cinnamate 4-hydroxylase), 4CL1 (4coumarate:CoA ligase 1), C3H1 (p-coumarate 3-hydroxylase 1), CCoAOMT1 (caffeoyl-CoA O-methyltransferase 1), COMT (caffeic acid O-methyltransferase) and F5H1 (ferulate 5-hydroxylase All compounds differ between the prn2 mutants and the WT with statistical significance (for detailed description, see the Materials and Methods section). ʻRatioʼ stands for average peak intensity in each of the prn2 mutants divided by the average peak intensity of the WT. SD, standard deviation; RT, retention time. Nine biological replicates of Arabidopsis were analysed in each genotype. The asterisks indicate that the exact identity of these compounds could not be determined but that they were either G (

Research
New Phytologist 1) compared with the wild-type (Fig. 6a,c). It was interesting to note that the most consistent gene expression changes (increased expression in prn2 mutants and lower expression in PRN2 overexpressors in both tissue types) were present for F5H1 which is specific to S-type lignin biosynthesis. Almost all studied transcription factors except for VND6 and VND7 showed either a trend or statistically significant increase in expression in prn2 compared with wild-type especially in the stems, while suppressed expression was evident for several of the transcription factor genes in the PRN2-overexpressor lines (Fig. 6b,d). Taken together, the results support PRN2 function in suppressing the expression of both the lignin-biosynthetic genes and the secondary cell wall regulatory transcription factors.

Discussion
Lignin deposition can occur in a noncell-autonomous manner whereby lignification of xylem elements is assisted by cells neighbouring the lignifying cells (Hosokawa et al., 2001;Pesquet et al., 2013;Smith et al., 2013;De Meester et al., 2018). The underlying mechanism has, however, remained unknown. Here we identify PRN2 as a factor modulating noncell-autonomous lignification in Arabidopsis. PRN2 was shown to be strictly expressed in the xylem elements next to the lignifying vessels, and from within these PRN2-expressing cells, to suppress accumulation of lignin in the neighbouring cells not expressing PRN2. These results imply that plants have the potential to increase lignin content of xylem elements by boosting noncell-autonomous lignification even post mortem, that is after the death of the xylem elements. Noncell-autonomous lignification provides, therefore, yet another level of control as well as plasticity to modify lignin deposition of xylem elements in response to changing developmental or environmental cues.

PRN2 marks the xylem cells contributing to noncellautonomous lignification
In earlier studies, Smith et al. (2013Smith et al. ( , 2017 defined in Arabidopsis stem tissues the cell types that contribute to noncell-autonomous lignification. Modification of lignin biosynthesis in specific cell types revealed that parenchymatic, nonlignifying cells as well as xylary fibres provide lignin precursors to neighbouring vessel elements and fibres in a noncell-autonomous fashion. Lignification of the interfascicular fibres, conversely, seemed mostly cell autonomous. Furthermore, Smith et al. (2017) showed that the contribution of each cell type depends on the tissue composition of the vascular bundles. For instance, in the young part of the stem, where no fibres were yet present, only the parenchymatic cells contributed to noncell-autonomous lignification, while in the older parts of the stem both parenchymatic cells and fibres contributed. Our results on the PRN2 promoter activity in the parenchymatic cells and xylary fibres of the vascular bundles but not in the neighbourhood of the interfascicular fibres in the lower part of the stem (Fig. 1j-l) is consistent with these results, indicating that PRN2 might be a good marker for the cells contributing to noncell-autonomous lignification. Accordingly, PRN2 influenced lignification of xylem elements in the vascular bundles but not of the interfascicular fibres (Fig. S7). Noncell-autonomous lignification has not been studied earlier during secondary growth of the hypocotyl. We could show in the secondary xylem of the hypocotyl that PRN2 promoter activity was specific to the vessel-neighbouring fibres that had thick secondary cell walls and that did not in any way seem different from the rest of the secondary xylem fibres of the hypocotyl (Fig. 1k,l). Interestingly, in the secondary xylem tissues of Populus stem, the Populus PRN2 promoter was active only in the ray cells next to the vessel elements ( Fig. S2a-c). Expression of the lignin-biosynthetic genes has also repeatedly been reported in the cells that are located next to the lignifying vessel elements of the secondary xylem (Hauffe et al., 1991;Feuillet et al., 1995;Chen et al., 2000;Baghdady et al., 2006). These results suggest that it is the vessel-neighbouring ray cells that contribute to noncell-autonomous lignification during secondary growth of the stem. Arabidopsis hypocotyls do not seem to have ray cells (Chaffey et al., 2002), and it is possible that, in Arabidopsis, the PRN2-expressing xylem fibres have adopted some of the functions, including contribution to noncell-autonomous lignification, that are normally performed by the ray cells in other species.

PRN2 suppresses primarily the accumulation of S-type lignin in the neighbourhood of the vessels
The preferred expression of PRN2 next to the vessel elements suggests that noncell-autonomous lignification is more important for the vessel elements than for the fibres. This seems plausible also due to the fact that, unlike the xylem fibres, the vessel elements differentiate and die very rapidly and might have opted for a strategy relying on noncell-autonomous lignification due to their short lifetime ( 13 OE Fig. 6 PRN2 suppresses the expression of lignin-biosynthetic genes and some of the transcription factors regulating secondary cell wall biosynthesis in Arabidopsis. Expression of lignin-biosynthetic genes (a, c) and transcription factors related to secondary cell wall biosynthesis (b, d) in the hypocotyls (a, b) and the stems (c, d) of prn2-1, prn2-2 and the PRN2-overexpressor lines PRN2OE6 and PRN2OE13. The expression levels were normalised to the expression in the wild-type (WT). For each genotype, five biological replicates were analysed, each composed of a pool of three hypocotyls (a, b) or stems (c, d). PDF2 and UBQ10 were used as reference genes. Error bars indicate AE SD. The asterisks indicate statistically significant difference from the WT by Welch's t-test (*, P < 0.05; **, P < 0.01). (2020)

Research
New Phytologist vessel elements is at least partially acquired by suppression of Stype lignin accumulation in a noncell-autonomous, PRN2-dependent manner. The mechanism seems, however, somewhat unrefined as excess lignin accumulation in the prn2 mutants was not restricted to the vessel elements but was present also in fibres neighbouring the PRN2-expressing cells (Figs 5, S7). In line with this, chemotyping of Populus woody tissues revealed that changes in lignin monomer composition between the different xylem cell types were gradual, the vessel-neighbouring fibres displaying lignin composition that was intermediate between that of the vessel elements and the more distant fibres (Gorzs as et al., 2011).
Lignification is known to be regulated at three different levels: the biosynthesis of monolignols within the cell, their export through the plasma membrane into the cell wall, and polymerisation within the cell wall. To date, very little information is known about monolignol transport and how it could influence lignin composition. A transporter for p-coumaryl alcohol is known (Alejandro et al., 2012), but transporters for coniferyl and sinapyl alcohol remain to be characterised. Lignin polymerisation in xylem is, according to current knowledge, influenced mainly by the availability of the monolignols (Tobimatsu & Schuetz, 2019). Hence, the most tightly regulated step of lignification seems to be on the level of regulation of monomer biosynthesis within the cell. Our results support the function of PRN2 on Stype lignin accumulation at this level by regulating the expression of the lignin-biosynthetic genes and, in particular, the S-type specific F5H1 (Fig. 6). The human PRN protein is known to function as a transcriptional co-regulator (Dechend et al., 1999;Liu et al., 2013), supporting function of PRN2 in transcriptional regulation of xylem lignification of vascular plants.
Harnessing noncell-autonomous lignification to counteract feedstock recalcitrance Reduction in lignin content is desired in several downstream applications of the lignocellulosic raw material such as biochemical conversion of the carbohydrates to sugar moieties by enzymatic hydrolysis (i.e. saccharification). Reduction in the overall level of lignin content, however, causes undesirable side effects such as growth penalties, most likely to be due to impaired water transport in vessels that tend to collapse due to low lignin content. Efforts to reduce lignin in cell types other than vessels have also not been fully satisfactory (Yang et al., 2013;Vargas et al., 2016). The localised effect of noncell-autonomous lignification offers an interesting alternative to modify cell wall chemistry of specific xylem cell types. Maintaining noncellautonomous lignification and hence lignification of the vessel elements in plants that have fibres with a low lignin content is a desired property that is expected to reduce recalcitrance of the lignocellulose to enzymatic hydrolysis without necessarily compromising the growth of the plants. A few recent reports have given support to this approach. Smith et al. (2013) reported increased saccharification without growth penalty in Arabidopsis stems in which lignin biosynthesis was disrupted by an artificial microRNA construct for CCR1 under the control of the CesA7 promoter that reduced lignin biosynthesis in all lignifying cells but allowed lignification of the vessels through noncell-autonomous lignification by the nonlignifying, parenchymatic cells. Similarly, significant increases in saccharification without a growth penalty were obtained in Arabidopsis stems where lignification was reintroduced to the vessel elements of ccr1 mutant by transgenic expression of CCR1 under the control of a synthetic promoter containing the vessel-specific SNBE domains (De Meester et al., 2018). In both cases, it is likely that the increases in saccharification were gained because of low lignification in the interfascicular fibres that constitute the main lignocellulosic mass of Arabidopsis stem. Interfascicular fibres are not the dominating lignified cell type in major lignocellulosic feedstocks such as forest trees, and it will be interesting to study whether noncell-autonomous lignification could be harnessed also in trees to improve the biochemical conversion properties of lignocellulose. Our results on PRN2 provide a new tool to investigate this aspect. Reintroduction of lignification into the PRN2-expressing cells in trees that are hampered in lignin biosynthesis might result in low lignin lignocellulosic biomass without affecting the biomass yield.

Supporting Information
Additional Supporting Information may be found online in the Supporting Information section at the end of the article.       Table S1 All primer sequences used in this study.
Please note: Wiley Blackwell are not responsible for the content or functionality of any Supporting Information supplied by the authors. Any queries (other than missing material) should be directed to the New Phytologist Central Office.
New Phytologist is an electronic (online-only) journal owned by the New Phytologist Trust, a not-for-profit organization dedicated to the promotion of plant science, facilitating projects from symposia to free access for our Tansley reviews and Tansley insights. Regular papers, Letters, Research reviews, Rapid reports and both Modelling/Theory and Methods papers are encouraged. We are committed to rapid processing, from online submission through to publication 'as ready' via Early View -our average time to decision is <26 days. There are no page or colour charges and a PDF version will be provided for each article.
The journal is available online at Wiley Online Library. Visit www.newphytologist.com to search the articles and register for table of contents email alerts.
If you have any questions, do get in touch with Central Office (np-centraloffice@lancaster.ac.uk) or, if it is more convenient, our USA Office (np-usaoffice@lancaster.ac.uk) For submission instructions, subscription and all the latest information visit www.newphytologist.com