Seeds of the barley (H. vulgare) cv. Morex were husked and sterilized by washing the seeds in 70% EtOH shortly and in 5% sodium hypochlorite for 30 min on a shaker at 4°C. The seeds were then rinsed with autoclaved dH₂O twice and plated on square plastic plates (120 × 120 mm) containing growth agar (0,5 g MES hydrate, 2,2 g Murashige & Skoog medium per liter, pH 5.7). Per plate, 5 seeds were placed 2 cm from the top of the plate into the agar. The plates were then stored for at least 2 days in darkness at 4°C for stratification and placed in a 16°C phytochamber with 24 h light at a 45° angle to the shelf for growth. Arabidopsis thaliana (Col-0) seeds were treated and grown as described in Stahl et al. (2009).

The synthetic peptides were acquired from Thermo Fisher Scientific and Centic Biotec with the following amino acid sequences: HvCLE402p (MLOC_3686.1) REVPTGPDPIHH; AtCLE40p RQVHypTGSDPLHHK (Hyp = hydroxyproline); mCLE40p LPQHPHGRSDVT. The peptides were added to the growth medium at a final concentration of 1 μM and the seeds were grown on these plates as described above for 5 days after germination (DAG).

Probes for the HISTONE H4 (AK357536) mRNA were prepared from the whole coding sequence. The DNA was cloned into the pGGC000 entry vector of the GreenGate cloning system (Lampropoulos et al., 2013) and amplified including the T7 and SP6 promoter sites by PCR. RNA probes were produced as described in Hejátko et al. (2006). RNA in situ hybridisations were performed on roots of plants 8 DAG as described in Jackson (1991), except for the following changes: after fixing the tissue over night at 4°C in 4 % para-formaldehyde, 0.1% tween-20, 0.1% triton-x-100 in PBS, a Leica ASP 300 tissue processor was used for embedding with the following protocol: 1 h 50% Ethanol (EtOH), 1 h 70% EtOH, 1 h 95% EtOH plus Eosin Y, 1 h 100% EtOH plus Eosin Y, 1 h 100% EtOH, 1 h 100% EtOH, three times 1 h 100% Xylol, 20 min paraplast at 60°C, 10 min paraplast at 60°C. 10 μm sections were made at the microtome.

Modified pseudo-Schiff propidium iodide (mPS-PI) staining was performed as described for floral stalks in Truernit et al. (2008) on root tips of plants 8 DAG. The staining with Schiff reagent and PI was carried out using vacuum. The samples were examined with either a 25x oil objective with a numeric aperture (NA) of 0.8 using a Zeiss laser scanning microscope (LSM) 510 Meta or a 40x water objective with a NA of 1.20 using a Zeiss LSM 780. PI was excited with a 561 nm Argon laser with emission detection at 566–718 nm.

For cross sections of the root hair zone, roots were embedded in melted 5% agarose and sectioned manually with a sharp razor blade. Endodermis staining with berberine hemisulfate was carried out as described in Lux et al. (2005). The samples were examined with a 40x water objective with a NA of 1.20 using a Zeiss LSM 780. Green fluorescence was excited with a 488 nm Argon laser with emission detection at 490–544 nm. Transmitted light pictures were taken with a transmitted light detector (T-PMT).

EdU staining was performed with the Click-iT EdU Imaging Kit (Invitrogen) and the fluorophor Alexa568 as described in the manufacturer's manual with the following modifications: root tips of plants 8 DAG were covered with 10 μM EdU in dH₂O and placed in the phytochamber for the respective incubation time. Root tips were fixed for 1 h under vacuum and permeabilized for 1 h at room temperature. The Click-iT reaction was carried out for 1 h under vacuum in darkness. DNA-counterstaining was performed with 1 μg/ml DAPI in PBS for 1 h in darkness under vacuum. The samples were cleared for around 14 days at 4°C in clearing solution described in Warner et al. (2014). The roots were examined with a 40x water objective with a NA of 1.20 using the Zeiss LSM 780. DAPI was excited with a 405 nm Diode with emission detection at 410–560 nm, Alexa568 was excited with a 514 nm Argon laser and emission was detected at 545–697 nm in a separate track. The pinhole was set to 2,05 Airy units. Pictures were taken with the tile scan function with 10% overlap, a threshold of 0.70 and automatically stitched by the microscope software.

RNA in situ hybridizations were examined with a plan-neofluar 20x objective with a NA of 0.50 or a plan-neofluar 40x objective with a NA of 0.75 using a Zeiss Axioskop light microscope.

For the alignment of the CLE-motifs, the 12 amino acid CLE-motif of Arabidopsis CLE40 (Hobe et al., 2003) was used for a BLAST search of barley CLE homologs. Barley homologs were obtained from http://webblast.ipk-gatersleben.de/barley/ using the default BLAST parameter settings among high and low-confidence genes (Mayer et al., 2012). Alignments were performed using MEGA7 (Molecular Evolutionary Genetics Analysis version 7.0) for bigger datasets and a MUSCLE alignment.

Measurements of the meristem length, counting of the distal stem cells and segmentation of the z-stacks were carried out in Fiji (Schindelin et al., 2012). For meristem length measurements, the border between meristem and elongation zone was defined by the first cell in the outermost cortex cell layer that doubled in cell length compared to its distal neighbor and analysis was carried out qualitatively from direct observation (as described in Dello Ioio et al., 2007). Segmentation of cells in the z-stack of the stem cell niche was performed with the MorphoLibJ plugin and morphological segmentation (Legland et al., 2016). All plots were created in R (R Core Team, 2015). Statistical significances of the meristem length and the distal stem cell differences were determined by a two-tailed Student's T-Test with the indicated p-value. For image compilation, Adobe Photoshop was used. In the microscope images of the meristems contrast and brightness were changed in parts of the images, as the images were composed of single microscope images.

In the first days of seedling development, the seminal roots are the main root type involved in water uptake (Knipfer and Fricke, 2011), so they are of particular importance for development of the plant in water stress conditions. We therefore focused our study on this root type. There is no consensus in the literature on the number of seminal roots and the appearance of a primary root in different barley cultivars (Hackett, 1968; Luxová, 1986). Under our experimental conditions, the root system of the barley cv. Morex consists of 2–6 seminal roots that arise during the course of the first five DAG. Around 10 DAG, adventitious roots arise from the shoot. This is in accordance with the data of Knipfer and Fricke (2011) and Hackett (1969). Luxová (1986) distinguished primary and seminal roots of barley by position in the embryo, however, this was not possible for the cv. Morex without dissecting the embryo. Because we could not detect a phenotypical difference between any of the first roots, we made no distinction between primary and seminal roots. Within the first 16 DAG, the roots grew to an average length of approximately 5 cm in our growth conditions (Figure 1A). After germination, the number of cells in the meristem increases, until the meristem reaches its full size when the rates of cell division and cell elongation are balanced. However, problems in visualizing meristem cells in barley result from the thickness and size of the roots which make microscopy impossible without clearing. By adapting the technique of mPS-PI staining (Truernit et al., 2008) which was previously described for Arabidopsis flower stalks to barley, we could stain the root cell walls and starch granules in the root tip for confocal microscopy without previous sectioning. To measure the meristem size, we analyzed the outermost layer of the cortex cells next to the epidermis and defined the transition zone between the meristem and the elongation zone as prior exemplified for Arabidopsis. There, the transition between the meristem and the elongation zone is defined as the region where the first cortical cell doubles in size compared to its distal neighbor (Dello Ioio et al., 2007). The observation that the different seminal roots of a single barley seedling emerge over the course of up to 5 days following germination complicates this type of analysis, as the measured roots will differ slightly in age and, accordingly, developmental stage. This is reflected in the variation in root meristem lengths at each time point (Figures 1B,C). The meristem continued to grow, but growth slowed down approximately 6 DAG, while root length steadily increased in the monitored time window (Figures 1A–C).

We first characterized the number and identity of root cell files. Because cells are not yet differentiated in the root meristem, they cannot be distinguished only on the basis of their morphology. We therefore stained the suberized tissue in a cross section of the differentiated part of the root (root hair zone) with the fluorescent dye berberine hemisulfate (Figure 2). In this region of the root, the endodermis, the exodermis and the epidermis have suberized cell walls as diffusion barriers (Nawrath et al., 2013). We found that the barley root consists of one layer of epidermis, one layer of exodermis derived from the cortex, four layers of cortical cells and one endodermal layer. The cortex cells can be categorized into inner cortex (small cells) and outer cortex (larger cells) on the basis of their morphology. The central cylinder of the barley cv. Morex root comprises one large central and eight smaller peripheral metaxylem vessels, confirming similar observations on seminal roots of the cv. Golf by Knipfer and colleagues (Knipfer and Fricke, 2011). As the exodermis only forms in the differentiating zone of the root in a layer of cortical cells beneath the epidermis, it can be considered as cortex cell file in the root meristem (Enstone et al., 2003).

In many plant species analyzed, the QC serves to maintain adjacent stem cells in an undifferentiated state by short range signaling (Dolan et al., 1993; van den Berg et al., 1997; Ni et al., 2014; Kerk and Feldman, 1994). QC and stem cells of maize and Arabidopsis are characterized by their slower cell division rate in comparison to the surrounding cells (Clowes, 1984; Dolan et al., 1993). This quiescence could be necessary to protect the QC from DNA damage caused by DNA replication and allows to provide a pool of cells with an error-free genome for renewing the surrounding stem cells (Heyman et al., 2014). To investigate the cell division rate in the barley stem cell niche, we made use of two approaches, EdU staining and RNA in situ hybridization to detect expression of Histone H4, which is specifically expressed during S-phase. EdU is a thymidine analog that is incorporated into the DNA during DNA synthesis and hence labels cells in the S-phase of the cell-cycle (Kotogány et al., 2010). In Arabidopsis, a 24 h incubation with EdU labeled all cells in the stem cell niche except for the QC cells (Vanstraelen et al., 2009). In the barley root meristem, a 24 h treatment with EdU marked almost all nuclei in the stem cell niche (Figure 3B), while after a 12 h incubation with EdU 60 % of the roots (9/15) carried cells lacking EdU label in the putative stem cell niche (Figure 3A). Results of a 6 h EdU treatment were even more striking and 70% of all roots (30/44) lacked EdU incorporation at this position (Figure 4). Importantly, we could not detect any group of cells that was non-dividing over more than 24 h. To identify the QC cells, we analyzed the cell division rate in different subsets of the stem cell region in roots that were treated with EdU for 6 h (Figure 4). Notably, the cell division rates in the subsets revealed that a quiescence gradient in the QC region exists, with the highest quiescence, i.e. the lowest cell division rates in the cell layer adjacent to the root cap (subset 1) and increasing cell division rates in subsets 2, 3 and 4 (Figures 4B–I). Subset 3, with the most striking difference in cell division rate in comparison to the other subsets, is displayed in Figures 4F,G and includes around 9 cells in longitudinal sections (blue frame). The cell division rate in subset 3 ranged from 0 to 30%, while the cell division rate in the surrounding cells (orange region in Figure 4F) was predominantly in a range from 20 to 60% (Figure 4G). We therefore suggest that subset 3 cells represent the QC region. However, in 14/40 roots, more than 30% of the QC cells were EdU stained within a 6 h period (Figure 4G). To confirm this cell division pattern, we performed RNA in situ hybridisation with a probe detecting Histone H4. Cell division rates in subset 3 (blue frame in Figure 5B) ranged from 0–20 and10–50% in the surrounding region (orange frame in Figures 5B,C), supporting the previous identification of the QC in subset 3. Assuming that the QC has a hemispherical shape, we calculated that the entire QC consists of around 30 cells.

We studied mPS-PI stained barley roots to identify the clonal origin of the epidermal, the five cortical and the endodermal cell layers. By tracing the outlines of cells, we found that the endodermis and a variable number of cortex cell files share the same founder cell which locates adjacent to the QC (in 17/23 roots at 5–8 DAG). In most cases, a QC abutting inner cortex endodermis initial (ICEI) gave rise to the endodermal cell layer and the inner cortex cell layer, whose descendants remained distinctly smaller than the outer cortex cell layers (see Figure 2). However, in contrast to the model organism Arabidopsis, cell division patterns are less regular. We often found that cells at a distance to the QC underwent periclinal cell divisions, thereby generating additional cortex cell files (white arrow heads in Figure 6A). The outer cortex cell layers derived from a distinct outer cortex initial (OCI) that first generated two cortex cell files by alternating between anticlinal and periclinal divisions. Further periclinal division in descendants give rise to additional cortex layers. Serial optical sections confirm this cell division pattern (Figure 6B, Supplementary Movie 1). This pattern of cell layer generation differs from that in rice, where a ground tissue stem cell abutting the QC undergoes four rounds of asymmetric divisions to generate endodermis, cortex, exodermis and sclerenchyma cell layers (Rebouillat et al., 2009; Ni et al., 2014). In maize, the position and cell division pattern of stem cells has not yet been analyzed in detail (Hochholdinger et al., 2004a; Jiang and Feldman, 2005). In barley, the epidermis can be traced back to a dedicated stem cell adjacent to the QC (in 20/26 roots at 5–8 DAG) (Figure 6A). The origin of the vascular system and division patterns of vascular stem cells could not be traced back unequivocally by our mPS-PI staining method. Cell division patterns appeared highly variable, and reporter lines marking cell clones would be required to determine number and behavior of vascular initials. However, a single file of metaxylem cells is prominent in all roots, which can be identified at a distance of more than 5 cells from the QC (Figures 2, 6A).

At the root tip, we observed a distinct border that separates the starch-containing columella cells and the undifferentiated columella precursors from the proximal part of the root (Figure 6A). Differentiated columella cells contain starch grains that are necessary for gravitropism (Kiss et al., 1996). In rice, maize and Arabidopsis all cells of the columella and lateral root cap except for the columella stem cells contain starch grains (Figure 10; Dolan et al., 1993; Lim et al., 2000; Wang et al., 2014). In barley cv. Morex, we detected starch granules only in the five to six distal layers of the columella. The lateral root cap and on average four cell layers proximal to the differentiated root cap cells, however, lacked starch granules (Figures 1D, 6A, 9E) and might act as columella stem cells. EdU treatment revealed that the first cell layer distal to the QC had a high cell division potential, which declined in the three more distal layers (Figures 6C,D). We occasionally also observed cell divisions in differentiated root cap cells (gray arrowhead in Figure 6C), which was also noted for columella cells in Arabidopsis, but not for rice (Wang et al., 2014; Hong et al., 2015). The lateral root cap originates from periclinal and oblique (in 6/30 roots at 5–8 DAG) cell divisions of lateral columella stem cells (Figure 6A, cells framed with orange), and lateral root cap cells maintain high division activity (marked by orange line in Figure 6C). Extended EdU staining over 48 h confirmed that most lateral root cap cells divided at least once within this time frame. Histone H4 expression analysis supported the conclusions drawn from EdU stainings, showing that (1) all cells of the root cap remain division active, and (2) that the columella stem cell layer proximal to the QC maintains the highest divisional activity (blue arrow head in Figure 5A). The cell division pattern and cell wall arrangement at the position of the epidermis initial (EPI) indicates that the epidermal cell layer and the lateral root cap of barley are of independent origin, as it is typical for monocot roots (Figure 6A; Clowes, 2000; Rebouillat et al., 2009).

In Arabidopsis, a constant population of columella stem cells is maintained through a negative feedback regulation, involving the differentiated columella cells and the QC. The QC promotes columella stem cell fate in adjacent cells due to a non-cell autonomous function of the mobile transcription factor WOX5 (Pi et al., 2015). Differentiated columella cells, which are the descendants of the columella stem cells, express the secreted peptide CLE40 which acts via receptor kinases to confine WOX5 expression (Stahl et al., 2009, 2013). An excess of CLE40 causes a rapid differentiation of stem cells toward columella cell fate. Additionally, CLE40 also regulates the size of the proximal meristem: here, increased levels of CLE40 can induce stem cell differentiation and loss of meristem activity (Hobe et al., 2003; Fiers et al., 2005). We now asked if a similar mechanism is acting in barley, and identified 21 CLE family peptides encoded in the available barley genome (Mayer et al., 2012). In Arabidopsis, CLE peptides are involved in a variety of developmental processes, but only two of them, CLE40 and CLAVATA3 (CLV3), act in meristem maintenance pathways. In rice, FLORAL ORGAN NUMBER2 (FON2) and FON2-LIKE CLE PROTEIN1 (FCP1) encode highly homologous CLE peptides, and of all CLE genes analyzed, only FCP1, FON2, CLV3, and CLE40 carry two introns, reflecting a common evolutionary origin (Fletcher et al., 1999; Hobe et al., 2003; Chu et al., 2006; Suzaki et al., 2006, 2008). Our search for CLE genes in the barley genome uncovered MLOC_3686 (named now HvCLE402) (Mayer et al., 2012), which carries two introns and encodes a CLE peptide highly related to FCP1 (Figures 7, 8). Treating barley plants with 1 μM synthetic HvCLE402 peptide consisting of the 12 amino acid CLE motif (HvCLE402p) or the Arabidopsis AtCLE40 peptide led to a significantly reduced root meristem length in comparison to untreated or mock treated plants (Figures 9A,B). We found that the number of proximal meristem cells was severely reduced (Figure 9C), indicating that the reduction of meristem size is likely caused by premature differentiation of proximal meristematic cells. This is consistent with observations made for the response of rice or Brachypodium to increased CLE peptide levels (Kinoshita et al., 2007; Czyzewicz et al., 2015). In the distal root meristem, 4 layers of columella stem cells give rise to the starch granule containing columella cells. In Arabidopsis, columella stem cell fate and number is negatively feedback regulated by CLE40 peptide, which is generated by differentiated columella cells. We asked if a similar regulation takes place in barley, and counted columella stem cell layers in peptide treated and untreated roots. Treatment with high levels of HvCLE402 or AtCLE40 peptides caused a minor, albeit statistically significant reduction of stem cell number, indicating that regulation of stem cell fate cannot depend solely on HvCLE402 (Figures 9D,E). Furthermore, we found that the HvCLE402 peptide can trigger both differentiation of columella stem cells and the proximal meristem cells in Arabidopsis, like the AtCLE40 peptide (Figure 10, Stahl et al., 2009), indicating that the CLE40 pathway controlling proximal meristem maintenance is highly conserved between monocots and dicots, while the distal meristem is only partially regulated through a CLE40 pathway.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.