ABSTRACT
As a prerequisite for constant growth, plants can produce vascular tissues at different sites in their postembryonic body. In particular, the formation of vascular tissues during longitudinal and radial expansion of growth axes differs fundamentally with respect to its anatomical configuration. This raises the question to which level regulatory mechanisms of vascular tissue formation are shared throughout plant development. Here, we show that, similar as primary phloem formation during longitudinal growth, the cambium-based formation of secondary phloem depends on the function of SMXL genes. Using promoter reporter lines, we observe that SMXL4 and SMXL5 activities are associated with different stages of secondary phloem formation in Arabidopsis stems and the specific loss of SMXL5 function results in the absence of secondary phloem. Interestingly, the additional disruption of SMXL4 activity increases cell proliferation rates in the cambium region without that secondary phloem is formed. Based on genome-wide transcriptional profiling and expression analyses of phloem-related markers we conclude that early steps of phloem formation are impaired in smxl4;smxl5 double mutants and that additional cambium-derived cells fail in establishing any phloem-related feature. Our results show that molecular mechanisms determining primary and secondary phloem share important features but differ slightly with SMXL5 playing a more dominant role in the formation of secondary phloem.
INTRODUCTION
Bodies of higher vascular plants, such as trees, acquire an impressive amount of biomass that solely relies on post-embryonic growth. Continuous formation of organs and tissues is driven by distinct stem cell niches called meristems (Greb and Lohmann 2016). Longitudinal growth of shoots and roots depends on shoot and root apical meristems (SAM, RAM), respectively. These proliferating tissues establish primary shoot and root axes which harbour vascular bundles consisting of two types of primary vascular tissues: The water and nutrient-conducting xylem and the photoassimilate-allocating phloem (Ruonala et al. 2017). Naturally, both xylem and phloem are life-essential conduits for long-distance transport that determine plant fitness (De Rybel et al. 2016). After establishment of primary growth axes, axes of dicotyledonous species and conifers grow radially with the help of a lateral meristem, called the vascular cambium. The cambium produces secondary phloem and xylem in a bidirectional manner (Campbell and Turner 2017). Therefore, although functionally equivalent, primary and secondary vascular tissues originate from different cellular environments. This raises the question to which extend molecular circuits instrumental for forming primary and secondary vascular tissues are comparable (Jouannet et al. 2014).
Formation of primary vascular tissues is extensively studied in root tips of Arabidopsis thaliana (Arabidopsis) (De Rybel, et al. 2016) and a large body of literature emphasizes the tight intercellular crosstalk during vascular patterning (Baima et al. 2001, Bishopp et al. 2011, Carlsbecker et al. 2010, Miyashima et al. 2019, Smetana et al. 2019). Especially, a mutual negative interaction between auxin and cytokinin signalling is essential for establishing domains of xylem and phloem production (Bishopp, et al. 2011, De Rybel et al. 2014, Mähönen et al. 2006, Ohashi-Ito et al. 2014). For the promotion of procambial divisions between both domains, phloem-derived or locally expressed DNA-BINDING ONE ZINC FINGER (DOF) transcription factors promote cell proliferation (Miyashima, et al. 2019, Smet et al. 2019). DOFs are counteracted by CLASS III HOMEODOMAIN-LEUCINE ZIPPER (HD-ZIP III) transcription factors which are expressed in early xylem cells where they promote xylem differentiation and instruct cambium organization in a non-cell autonomous manner (Miyashima, et al. 2019, Smetana, et al. 2019).
Except promoting HD-ZIP III gene activity, the phloem-borne DOF transcription factor PHLOEM EARLY DOF2 (PEAR) directly activates the SUPPRESSOR OF MAX2 1-LIKE 3 (SMXL3) gene which not only stimulates procambial cell divisions (Miyashima, et al. 2019), but also determines phloem specification (Wallner et al. 2017). In this context, SMXL3 acts together with its closest homologues SMXL4 and SMXL5 as key regulators of primary phloem formation (Wallner, et al. 2017). This role is reflected in the absence of primary phloem in smxl3;4;5 triple mutants, while double mutants show specific defects in (proto)phloem initiation and differentiation within the RAM (Wallner, et al. 2017). Although SMXL3/4/5 function has not yet been related to other phloem regulators (Anne and Hardtke 2018) and their molecular role is still obscure, early expression of all three genes in phloem precursor cells and their profound role in phloem formation define SMXL3/4/5 genes as central promoters of phloem formation (Wallner, et al. 2017). These features define SMXL3/4/5 genes as ideal subjects to compare the mechanisms underlying the development of primary and secondary phloem in different anatomical environments.
Here we report a specific role of SMXL5 in promoting secondary phloem formation during radial stem growth in Arabidopsis. By investigating genome-wide and promoter reporter-based gene expression studies in grafted plants, we observe that secondary phloem formation is disrupted in smxl5 single and smxl4;smxl5 double mutants. In contrast to SMXL4 and SMXL5, which are expressed at different stages during secondary phloem formation, SMXL3 is not expressed during secondary phloem formation in the Arabidopsis stem. Based on our observations, we conclude that there are functional similarities and differences within the SMXL3/4/5 sub-clade and that the formation of primary and secondary phloem shares essential features.
RESULTS
SMXL4 and SMXL5 promoters are active during secondary phloem formation
SMXL5 promoter activity marks a cambium sub-domain which is located distally to bifacial cambium stem cells and has been associated with phloem formation during radial plant growth (Shi et al. 2019). To see whether there is an association of the primary phloem-promoting SMXL3/4/5 genes with secondary phloem, we investigated SMXL3 and SMXL4 promoter activities in cross sections of stems displaying secondary tissue conformation and compared those activities to the promoter activity of SMXL5. Secondary tissue conformation is established by the formation of the interfascicular cambium (IC) between vascular bundles of stems (Altamura et al. 2001, Mazur et al. 2014, Sanchez et al. 2012, Sehr et al. 2010). The resulting cambium cylinder produces xylem proximally (towards the stem centre) and phloem distally (towards the stem periphery, Figure 1a). Therefore, vascular tissues found in interfascicular regions are clearly cambium-derived and of secondary nature.
When analysing transgenic lines expressing endoplasmic reticulum (ER)-localized YFP (YELLOW FLUORESCENT PROTEIN) under the control of the respective SMXL promoters, we found SMXL5:YFP-ER activity, as expected, in undifferentiated cambium cells and distally from these cells in developing and mature phloem (Figure 1b, c). In comparison, SMXL4:YFP-ER activity was observed exclusively in differentiated primary and secondary phloem (Figure 1d, e) whereas only weak SMXL3:YFP-ER activity was observed within the primary phloem of vascular bundles (Figure 1f, g). This was in accordance with previous findings of SMXL3 activity being strong in the seedling root but weak in above-ground tissues (Wallner, et al. 2017). Together, these observations indicated different roles of SMXL3/4/5 genes in vascular development.
SMXL4 and SMXL5 suppress cambium activity in the stem
Since SMXL3 activity was not found in secondary phloem of stems, we focused on SMXL4 and SMXL5 to determine their roles in secondary phloem formation. To this end, we compared the extension of cambium-derived tissue (CDT) in interfascicular regions of wild type, smxl4, smxl5 and smxl4;smxl5 mutants. smxl4 and smxl5 single mutants did not differ in CDT extension (Figure 2a-c, e) from wild type plants. However, analysis of the smxl4;smxl5 double mutant revealed a significant increase in CDT production compared to wild type and the single mutants (Figure 2d, e). Because smxl4;smxl5 plants showed a weaker growth habit overall (Figure 2f), possibly due to defects in root growth (Wallner, et al. 2017), we investigated stem anatomy after grafting wild type roots to smxl4;smxl5 shoots and vice versa. Self-grafts of wild type and smxl4;smxl5 served as controls. Interestingly, smxl4;smxl5 stems showed an increase in CDT production independently from the root genotype (Figure 3a, b, d, e, f, h, i). In contrast, stem diameter and overall growth of smxl4;smxl5 shoots supported by a wild type root were fully restored while wild type stems supported by a smxl4;smxl5 root showed small stem diameter and weak overall growth (Figure 3a-d, j, k). We therefore concluded that deficiency of SMXL4 and SMXL5 leads to increased cambium proliferation in stems independently from root or shoot growth.
SMXL5 promotes differentiation of secondary phloem
To see whether increased CTD production in smxl4;smxl5 mutants is associated with altered formation of secondary phloem, we visualized callose, a marker for differentiated sieve elements (SEs) (Barratt et al. 2011), in wild type, smxl4, smxl5 and smxl4;smxl5 stems as well as in our grafts by aniline staining. These analyses revealed callose deposition in 98.6 % and 94.2 % of the analysed interfascicular regions in wild type and smxl4 plants, respectively (Figure 4a, b, e, f, i). In contrast, 85.2 % of interfascicular regions analysed in smxl5 stems were deprived of callose deposition (Figure 4c, g and i). Similarly, 76.1 % of all interfascicular regions in smxl4;smxl5 stems showed no signs of callose deposition (Figure 4d, h, i). Of note, we obtained similar results when analysing grafted samples. Independently from the root genotype, smxl4;smxl5 stems were largely deprived of secondary phloem formation within interfascicular regions, while wild type stems supported by a smxl4;smxl5 root remained unaffected (Figure S1). This indicated that increased CDT production is not accompanied by increased phloem differentiation and that the formation of secondary phloem is indeed SMXL5-dependent. Importantly, primary (proto)phloem formation in primary vascular bundles (Figure 4b, c, f, g) and the RAM of smxl5 or smxl4 single mutants (Figure S2) resembled wild type plants. Thus, we concluded that molecular mechanisms determining primary and secondary phloem shared important features but differed slightly with SMXL5 playing a more dominant role in the formation of secondary phloem. Moreover, disruption of SMXL4 function in a SMXL5-deficient background leads to increased proliferation of the cambium in stems.
Phloem markers are absent in interfascicular regions of smxl5 stems
To characterize the anatomical alterations in smxl4;smxl5 mutant stems, we compared the transcriptomes of smxl4;smxl5 stem bases supported by a wild type root to those of wild type self-grafts. Application of an adjusted p-value cut-off of 0.01 and a fold change > 1.5 and < 0.5, identified 1343 downregulated and 273 upregulated genes in smxl4;smxl5 stems, respectively (Figure 5a and Table S1). Since we were especially interested in vasculature-related responses within the smxl4;smxl5 mutant stem, we first compared our datasets to a set of 280 vascular genes described previously (Endo et al. 2014). From those 280 genes, we found 98 to be downregulated in smxl4;smxl5, while only 13 were upregulated (Figure 5a). As expected, the downregulated fraction contained several important phloem regulators, such as ALTERED PHLOEM DEVELOPMENT (APL) (Bonke et al. 2003), SIEVE ELEMENT OCCLUSION RELATED 1 (SEOR1) (Froelich et al. 2011), BREVIS RADIX (BRX) (Mouchel et al. 2004, Rodriguez-Villalon et al. 2014), BARELY ANY MERISTEM 3 (BAM3) (Depuydt et al. 2013), NAC DOMAIN CONTAINING PROTEIN 86 (NAC086) (Furuta et al. 2014), JULGI1 and JULGI2 (Cho et al. 2018) or CELLULOSE SYNTHASE (CESA) subunits 4 and 8. CESA4 and CESA8 are important during secondary cell wall formation of differentiating SEs (Taylor et al. 2003). To see whether phloem development is disrupted at a distinct state, we next compared our identified genes with four groups of genes (modules) associated with four different stages of phloem formation (Kondo et al. 2016). As a result, we found strong enrichment of our downregulated genes in all four modules. In contrast, only one of the upregulated genes (AT5G04310) was part of module IV (Figure 5b and Table S1). Collectively, our RNA sequencing data showed downregulation of a large spectrum of phloem-related genes reinforcing our finding of impaired secondary phloem differentiation in smxl4;smxl5. Moreover, the observation that, in addition to ‘late’ modules, a large fraction of the genes comprising module I was downregulated in smxl4;smxl5 mutants implying that already early events of phloem formation depend on the function of SMXL4 and SMXL5.
Interestingly, the upregulated fraction mostly included genes involved in stress responses, such as GLUTAREDOXIN S family members (GRXS) (Moseler et al. 2015) or RAPID ALKALIZATION FACTOR (RALF) (Sharma et al. 2016). Despite more CDT was produced in smxl4;smxl5 (Figure 2, Figure 3), though, genes associated with cambium stem cells, such as PHLOEM INTERCALATED WITH XYLEM (PXY), SMXL5, WUSCHEL HOMEOBOX RELATED 4 (WOX4), AINTEGUMENTA (ANT) or HIGH CAMBIAL ACTIVITY 2 (HCA2) (Etchells et al. 2013, Guo et al. 2009, Shi, et al. 2019, Smetana, et al. 2019), were not significantly changed (Table S1). These observations suggested that the number of cambium stem cells was not increased in smxl4;smxl5 mutants and that additional cambium-derived cells acquired an identity unrelated to vascular cell fates.
To confirm these assumptions, we crossed cambium and phloem markers into smxl4, smxl5 and smxl4;smxl5 plants and investigated activity patterns and intensities by confocal microscopy. Promoter activities of PXY:mTurquoise-ER and SMXL5:VENUS-ER mark the proximal and distal cambium domains, respectively (Shi, et al. 2019), and we expected an increase especially in the number of SMXL5-positive cells if the additional cells were arrested at early stages of phloem development. However, the extension of the cambium domains marked by PXY:mTurquoise-ER and SMXL5:VENUS-ER did not differ between smxl4;smxl5, wild type and the smxl single mutants in non-grafted or grafted plants (Figure 6 and Figure S3). This confirmed our expectation that the actual stem cell pool including the phloem-forming distal cambium domain was not altered in smxl4;smxl5. Moreover, impaired phloem formation in SMXL5-deficient plants did not affect cambium organization.
To determine the state of cells present distally from cambium stem cells, we analysed promoter-reporter activities of SMXL4 and SIEVE ELEMENT OCCLUSION 2 from Medicago truncatula (MtSEO2) as early and late phloem markers, respectively (Froelich, et al. 2011, Knoblauch and Peters 2010, Wallner, et al. 2017). Interestingly, neither marker was active in IC regions of secondary smxl5 and smxl4;smxl5 stems grafted onto wild type roots (Figure 7 and Figure S4). In contrast, the activities of SMXL4 and MtSEO2 promoter-reporters were abundant in secondary phloem of wild type and smxl4 and detected in primary phloem of all analysed genotypes (Figure 7 and Figure S4). This indicated that differentiation of secondary phloem and particularly SEs derived from the cambium in Arabidopsis stems was promoted by SMXL5. In addition, we did not find indications that additional cambium-derived cells in SMXL4/5-deficient plants established phloem-related signatures.
DISCUSSION
Growth of vascular plants, which represent the major source of food and biomass on earth (Nieminen et al. 2012), requires a fine-tuned balance between stem cell maintenance, cell divisions and differentiation (Hong et al. 2015, Sozzani and Iyer-Pascuzzi 2014). Despite their fundamental importance, molecular mechanisms behind vascular development are still not fully understood. Here we identified a dual role of SMXL5 in promoting secondary phloem formation and, together with SMXL4, in suppressing CDT production.
SMXL4 and SMXL5 fulfil roles as major phloem regulators across different organs and growth stages (Wallner, et al. 2017). By using the SE-specific promoter-reporter MtSEO2:GFP-ER we determined that SE formation during secondary phloem formation is largely dependent on SMXL5. Moreover, more cells were produced by the cambium in smxl4;smxl5 double mutants, but cells generated at the distal phloem side failed to establish any phloem-related features. This finding is consistent with previous observations: Differentiation, including enucleation, of SEs is absent in the RAM of smxl4;smxl5 double mutants and primary phloem within vascular bundles of smxl4;smxl5 stems accumulated undifferentiated, still nucleated cells (Cho, et al. 2018, Wallner, et al. 2017). Interestingly, we observed defects in secondary phloem formation already in smxl5 single mutants, which showed normal SE differentiation within the RAM and the primary phloem of stems. Given the differences in expression domains and levels of SMXL3/4/5 gene activities in secondary stems, we propose specialized roles of those otherwise redundantly acting phloem regulators during radial plant growth.
Comparing SMXL3/4/5 function in phloem formation in the RAM and the IC may shed light on cambium regulation. Already in the embryo the quiescent centre (QC) gives rise to four types of so-called initials (Scheres 2007, Sozzani and Iyer-Pascuzzi 2014). Those are root stem cells with pre-determined fate towards e.g. cortex, vasculature, epidermis or columella (Sozzani and Iyer-Pascuzzi 2014). They are maintained by the otherwise quiescent QC and divide asymmetrically into a stem cell and a daughter cell (De Rybel, et al. 2016, Scheres 2007). Detached from the organizing QC, the daughter cells start to differentiate into specific tissue types (Sozzani and Iyer-Pascuzzi 2014). Phloem daughter cells undergo important anticlinal and tangential cell divisions to initiate more specific cell lineages in a highly controlled spatio-temporal manner (Bonke, et al. 2003, Scacchi et al. 2010). In smxl4;smxl5 mutants, we found the second tangential cell division within the phloem cell lineage to be delayed (Wallner, et al. 2017). Consequently, more cells were produced that failed in differentiating into phloem cells (De Rybel, et al. 2016, Wallner, et al. 2017). We made a similar observation for the cambium: More cambium-derived cells were produced and, towards the distal cambium side, those cambium descendants failed to differentiate. As in the RAM, the nature of these cells is still unknown. Neither the proximal cambium marker PXY:mTurquoise-ER nor the distal cambium marker SMXL5:VENUS-ER showed an extended activity within smxl5 or smxl4;smxl5 stems and other phloem-related markers were fully absent. We hypothesize that those cambial daughter cells are stuck in an undifferentiated state, similar to cells in the RAM of smxl4;smxl5 mutants.
Exploring the roles of SMXL genes in secondary phloem formation, we found indications that SMXL genes belong to the earliest players in phloem formation. We found genes downregulated in smxl4;smxl5 which belong to the earliest genes transcriptionally activated during phloem induction (Module I) (Kondo, et al. 2016). Of note, SMXL5 is also part of genes in the Module I (Kondo, et al. 2016), which highlights the importance of SMXL5 during early stages of phloem formation. Overall, we discovered that SMXL5 plays a key role in a so far unknown regulatory network instructing secondary phloem formation. Therefore, our observations support the conclusion that molecular mechanisms acting in primary and secondary phloem share important features.
MATERIAL AND METHODS
Plant material and growth conditions
Arabidopsis thaliana (L.) Heynh. Col-0 plants were used in all cases. Genetic analysis was performed with mutant alleles smxl5-1 (SALK_018522) and smxl4-1 (SALK_037136). Genotyping was done by PCR using primers as described previously (Wallner, et al. 2017). Before sowing seeds directly on soil, they were vapour sterilized for 2 h in a sealed container by Hydrogen chloride, which was produced by mixing 40 ml 13% sodium hypochlorite with 1.2 ml 30% hydrochloric acid. For breaking seed dormancy and providing contemporaneous germination, seeds were stratified on soil at 4 °C for 3 days in the dark. Plants were first grown in short day (SD) conditions (8 h light and 16 h dark) with 60 % humidity and 22 °C for 3 weeks and then transferred to long day (LD) conditions (16 h light and 8 h dark) at 22 °C and 60 % humidity.
For grafting experiments and root studies, seeds were sterilized by 70 % ethanol supplemented with 0.2 % Tween-20 for 15 min, washed twice with 100 % ethanol and air dried under sterile conditions. Seeds were sown on Murashige and Skoog (MS) medium-plates containing 0.8 % phyto agar and 1 % sucrose. Seeds were stratified on plates for 3 days at 4°C in the dark. Seedlings were grown in SD conditions at 22°C for 5 days prior to grafting. Freshly grafted seedlings were kept for 6 days on wetted filter paper in sterile conditions to ensure reconnection of the vasculature and then transferred for another 7 days on MS-plates for recovery. Well-connected grafts were transferred to soil and kept in SD for another 3 days. After 21 days (3 weeks) plants were transferred to LD conditions.
Grafting
Grafting was performed on 5 day-old seedlings grown in SD as previously described (Melnyk and Meyerowitz 2015).
Generation of plasmids and transgenic lines
Plasmids SMXL3:YFP-ER, SMXL4:YFP-ER and SMXL5:YFP-ER were generated via a two-step cloning procedure using the vector pGreen0229 (Hellens et al. 2000). The exact procedure was described previously (Wallner, et al. 2017). Construction of the double-marker PXY:mTurquoise-ER;SMXL5:VENUS-ER was done by using the Greengate system as described in (Shi, et al. 2019). Col-0 seeds transformed with MtSEO2:GFP-ER were obtained from Michael Knoblauch (Froelich, et al. 2011).
Microscopy and tissue staining
For analysing reporter activities or aniline staining in stems, the stem base was embedded in small polystyrol blocks and hand sections of approximately 0.5 mm were taken by a razor blade (Wilkinson Sword). Reporter activities were analysed by a Leica TCS SP5II (Leica) system equipped with the Leica Application Suite X software. Prior to imaging, fresh stem sections were counterstained for 5 min by 5 µg/ml propidium iodide (PI, Merck) dissolved in tap water or 0.01 % Direct Red 23 (Sigma-Aldrich) dissolved in 1× PBS. mPS-PI staining of roots was carried out as described before (Truernit et al. 2008, Wallner, et al. 2017). YFP (yellow fluorescent protein) was excited by an argon laser at 514 nm and the emission detected at 520-540 nm. PI was excited at 561 nm (DPSS laser) and detected at 590-690 nm. Aniline staining was performed as described previously (Schenk and Schikora 2015, Wallner, et al. 2017). Stained stem sections were analysed using the epifluorescence microscope Zeiss Axio Imager.M1 (Carl Zeiss). Fluorescence of aniline was excited by the UV lamp and emission was collected using the DAPI filter.
Histological analyses
1-cm stem fragments just above the uppermost rosette leaf and 1-cm fragments including the hypocotyl and the uppermost root region were harvested from 15 – 20 cm tall plants and infiltrated by 70 % ethanol for at least 3 days. Samples were processed using the Tissue dehydrator LOGOS Microwave Hybrid Tissue Processor (Milestone) and the LEICA ASP200S (Leica). After embedding in paraffin, 10 μm sections were produced using a microtome, deparaffinised, stained with 0.05% toluidine blue (Applichem), and fixed by Entellan (Merck). Slides were scanned using a MIRAX Slide Scanner (3DHistech) or the Axio Imager.M1 microscope (Carl Zeiss). Scans were analysed using the Pannoramic Viewer (Histech) or ImageJ 1.49d (Schindelin et al. 2012).
RNA extraction and sequencing
RNA was extracted from 1-cm stem segments including the stem base. Three stem segments were pooled into one biological replicate. Tree biological replicates were collected per genotype. RNA was extracted by phenol/chlorophorm as described (Mallory and Vaucheret 2010) and treated with RNase-free DNase (Fermentas). Single reads of 50 nucleotides in length were sequenced using HiSeqV4 SR50. Reads were aligned to the Arabidopsis genome (TAIR10) using TopHat2 v2.0.14 (Kim et al. 2013) and the statistical analysis was done using the DESeq2 package from the R/Bioconductor software (Love et al. 2014). Mutant genotypes were compared to wild type by applying a stringency of adjusted p-value < 0.01. Data processing included the use of VENNY (Oliveros 2007) and alignment to The Arabidopsis Information Resource (TAIR)-databases (Huala et al. 2001) by VirtualPlant 1.3 (Katari et al. 2010).
Statistical analysis
Statistical analyses were performed using IBM SPSS Statistics for Windows, Version 23.0. Armonk, NY: IBM Corp. Means were calculated from measurements with sample sizes as indicated in the respective figure legends. Error bars represent ± standard deviation. All analysed datasets were prior tested for homogeneity of variances by the Levene statistic. Statistically different groups were determined by a One-way ANOVA with a confidence interval (CI) of 95 %. If variances were homogeneous (equal), a post-hoc Bonferroni test was applied for datasets n < 5 and a post-hoc Tukey HSD test was applied for datasets with n > 4. For comparison of two datasets a Student’s t-test (for equal variances) was performed. Graphs were generated in GraphPad Prism version 6.01 (GraphPad Software), or Excel (Microsoft).
Accessing RNA sequencing data
Raw data discussed in this publication have been deposited in NCBI’s Gene Expression Omnibus (Barrett et al. 2013) and are accessible through GEO Series accession number GSE129190 at https://www.ncbi.nlm.nih.gov/geo/.
Author Contributions
Conceived and designed the experiments: EW, TG. Performed the experiments: EW Analysed the data: EW, VJ. Wrote the paper: EW, TG.
Conflict of Interest
The authors declare that they have no conflict of interest.
Acknowledgements
This work was supported by the Deutsche Forschungsgemeinschaft (DFG) through the grant GR2104/4-1, the SFB873, and a Heisenberg Professorship [GR2104/5-1] to T.G.. Library preparation and next-generation-sequencing was performed at the Deep Sequencing Core Facility provided by the Exzellenzcluster CellNetworks at the University of Heidelberg (http://www.cellnetworks.uni-hd.de/483065/Deep_Sequencing_Core_Facility).