Research Article

An Evolutionarily Conserved Mechanism Delimiting SHR Movement Defines a Single Layer of Endodermis in Plants

See allHide authors and affiliations

Science  20 Apr 2007:
Vol. 316, Issue 5823, pp. 421-425
DOI: 10.1126/science.1139531

This article has a correction. Please see:


Intercellular protein movement plays a critical role in animal and plant development. SHORTROOT (SHR) is a moving transcription factor essential for endodermis specification in the Arabidopsis root. Unlike diffusible animal morphogens, which form a gradient across multiple cell layers, SHR movement is limited to essentially one cell layer. However, the molecular mechanism is unknown. We show that SCARECROW (SCR) blocks SHR movement by sequestering it into the nucleus through protein-protein interaction and a safeguard mechanism that relies on a SHR/SCR-dependent positive feedback loop for SCR transcription. Our studies with SHR and SCR homologs from rice suggest that this mechanism is evolutionarily conserved, providing a plausible explanation why nearly all plants have a single layer of endodermis.

Stem cell renewal and patterned differentiation of their progeny are fundamental processes in the development of multicellular organisms. The root of Arabidopsis thaliana is particularly suitable to study these processes, because it has a simple and stereotyped cellular organization (fig. S1) (1). SHR and SCR are key regulators of root radial patterning (2, 3) and stem cell maintenance (4). In shr and scr mutants, the cortex/endodermis initial (CEI) cell, which normally gives rise to two files of ground-tissue cells (an inner layer of endodermis and an outer layer of cortex), produces only a single cell layer (fig. S1) (2, 3, 5). SHR is a transcription factor (6) expressed in the stele that moves into the adjacent cell layer where it controls SCR transcription and endodermis specification (6). By contrast, the SCR protein is absent from the stele, is predominantly expressed in the endodermis, the CEI cell, and the quiescent center (QC), and is required for the asymmetric cell division that gives rise to the cortex and endodermis (3, 7). SHR protein does not move beyond a single layer comprising the endodermis, CEI cell, and QC. This is in sharp contrast to moving signal proteins in animals (8). Endodermis and cortex in the root are derived from the same initial cells through asymmetric cell divisions. Notably, although the number of cortex cell layers varies considerably, nearly all plant species examined so far have only one layer of endodermis, suggesting an evolutionarily conserved mechanism to form this single cell layer. SCR has been found to play a role in restricting SHR movement (9, 10), but the underlying mechanism has remained unclear.

Positive feedback control of SCR transcription. SHR and SCR belong to the GRAS family of transcription factors (11). In both animals and plants, transcriptional regulation is known to play a key role in development (12, 13). To elucidate the mechanism by which SHR and SCR control root radial patterning, we therefore first dissected their transcriptional circuits. Previously, it has been shown that SHR directly controls SCR transcription (14). However, there was also indication for SCR autoregulation (9). Using a chromatin immunoprecipitation–polymerase chain reaction (ChIP-PCR) assay (15), we found that SCR binds to its own promoter (Fig. 1A and fig. S2) but not to the promoters of SHR and a SIN3-like gene (At5g15020), which does not appear to be regulated by SHR (14). By reverse transcription PCR (RT-PCR), we confirmed the previous finding that SCR expression is reduced in both the shr and scr backgrounds (9) (fig. S2). Our results thus demonstrated that SCR is controlled by a SHR/SCR-dependent positive feedback loop.

Fig. 1.

SCR binds its own promoter and other SHR targets. (A and B) ChIP-PCR assay with the use of an antibody to GFP, showing binding by GFP-SCR and SHR-GFP to the promoters of SCR, MGP, and SCL3. Fold enrichment values in all panels, from (A) to (F), as determined by quantitative real-time PCR (QPCR), are means ± SE from technical replicates. pSHR, SHR promoter; SCR, SCR coding sequence; IN, input DNA; M, Mock ChIP. (C) SHR-GFP binding to some of its targets is abolished in scr mutant background. (D to F) SHR-GFP and GFP-SCR bind to the promoters of NUC (D), MGP (E), and RLK (F), as revealed by promoter scanning. The asterisks mark the positions of putative binding sites.

SHR and SCR are functionally interdependent. Recently, we identified a number of additional putative direct targets of SHR (14). To determine whether these are also direct targets of SCR, we assayed binding by ChIP-PCR. Our initial results showed that SCR binds to the promoters of MAGPIE (MGP) and SCR-LIKE 3 (SCL3) (Fig. 1A and fig. S3), which are also bound by SHR (Fig. 1B and fig. S3). We then checked SHR and SCR binding to other putative direct SHR targets by promoter scanning (15) (Fig. 1 and fig. S4). We found that SHR and SCR both clearly bind to the promoters of NUTCRACKER (NUC), a receptor-like kinase (RLK), and MGP (Fig. 1, D to F). Their binding sites on the NUC and MGP promoters were located in regions relatively far upstream of the translation start sites (TSS) (Fig. 2, D and E), which explains why we were unable to confirm NUC as a direct SHR target by ChIP-PCR using PCR primers that amplify more proximal sequences (14). Notably, most binding sites for SHR and SCR at these promoters appear to coincide.

Fig. 2.

SCR and SHR directly interact. (A) SHR-GFP is detected in SCR immunoprecipitates (SCR-IP). (B) SCR is coimmunoprecipitated with SHR-GFP. (C) SCR does not coimmunoprecipitate with GFP-SCR (GFP-IP). The SCR-IP assay shows that SCR is expressed in the transgenic plants expressing GFP-SCR. (D) Yeast two-hybrid assay showing direct interaction between SHR and SCR. AD-SHRΔ, which lacks the N-terminal 120 amino acids of SHR, still interacts with SCR. β-Gal, β-galactosidase; Gal4 UAS, Ga14 binding sites; BD, Ga14 DNA binding domain; AD, Ga14 activation domain; BD-53 and AD-T as a pair are used as a positive control, whereas the BD-lamin and AD-T pair is a negative control. BD-53, fusion between BD and the p53 protein; BD-lamin; fusion between BD and lamin; AD-T, fusion between AD and the T protein.

The observation that SHR and SCR bind to a common set of genes suggests functional interdependence between these two transcriptional regulators. We therefore examined SHR binding to some of its targets in an scr background. In the absence of SCR, SHR binding to these targets is abolished (Fig. 1C). Expression levels of all these genes are reduced in the shr and scr mutants (fig. S2), indicating that SCR is required for SHR to regulate these genes. To determine the extent of overlap between SHR and SCR targets, we performed genome-wide expression analysis in shr and scr mutants. Nearly all putative direct SHR targets that we previously identified (14) show significant reduction in their expression in both mutant backgrounds (Table 1). Moreover, a large portion of SHR indirect targets also showed reduced expression in the scr background (table S1).

Table 1.

Expression levels of SHR direct targets in shr and scr mutants, relative to WT, as measured by whole-genome Affymetrix ATH1 microarray. FC, fold change (reduction).

shr (FC)P valuescr (FC)P value
NUC 2.8 6.4 × 10-27 1.6 9.2 × 10-5
MGP 2.5 3.7 × 10-19 1.5 1.7 × 10-3
SCR 2.5 5.0 × 10-8 4.0 4.3 × 10-7
Br60×2 1.9 7.4 × 10-23 1.8 3.0 × 10-4
RLK 1.4 2.8 × 10-5 1.4 3.8 × 10-1
SCL3 1.3 6.3 × 10-3 1.4 7.6 × 10-2
Tropinone reductase (TRI) 1.2 2.7 × 10-1 1.2 7.0 × 10-1
SNEEZY (SNE) 1.0 6.6 × 10-1 0.8 5.9 × 10-5

SHR and SCR proteins directly interact. Functional interdependence between SHR and SCR could be achieved through their cooperative binding to the same promoter or through direct interaction. To determine whether SHR and SCR form a complex, we performed coimmunoprecipitation. Reciprocal pull-down experiments showed that SCR and SHR are in a complex (Fig. 2, A and B). In yeast cells, SHR and SCR interact directly (Fig. 2D), and the central domain spanning the two leucine heptad repeats and the VHIID (Val-His-Ile-Ile-Asp) motif is responsible for this interaction (fig. S5). However, SCR does not appear to interact with itself (Fig. 2C). The finding that SCR physically interacts with SHR provides a molecular basis for their functional interdependence. However, clearly not all aspects of SHR activity rely on interaction with SCR, because the mutant ground-tissue layer in scr still expresses endodermal markers that are not detected in the shr mutant background. One hypothesis is that the SHR/SCR complex controls some aspects of SHR function, such as asymmetric cell division, QC specification, and stem cell maintenance, whereas complexes formed between SHR and other proteins fulfill other aspects of SHR function, particularly endodermis specification.

SCR affects SHR subcellular localization and movement. As the SHR protein with a strong nuclear localization signal is no longer capable of moving (7), the finding that SHR and SCR directly interact suggests that one role for SCR might be to sequester SHR into the nucleus, thus preventing its movement. Indeed, the fusion protein between green fluorescent protein (GFP) and SHR (SHR-GFP) becomes largely cytoplasmic in the mutant cell layer of scr (scr-1 in Fig. 3C) (10), in contrast to its exclusive nuclear localization in the endodermis of wild-type (WT) roots (Fig. 3C). However, because of the low amount of SHR-GFP in the mutant cell layer, it is unclear whether SHR moves out of the mutant layer into the epidermis.

Fig. 3.

SCR determines SHR subcellular localization and its range of movement. (A) SCR transcript levels in two independent SCR RNAi lines (SCRi-1 and SCRi-2), relative to that in WT, as determined by RT-QPCR. (B) Root lengths of the SCR RNAi lines and WT 6 days after germination. Error bars in (A) and (B) indicate SD. (C) Confocal images of 6-day-old roots of WT, scr-1, SCRi-1, and SCRi-2 seedlings, showing their structure [propidium iodide (PI) staining], an endodermal marker expression (pSCR::GFP), and SHR-GFP localization (pSHR::SHR-GFP). The insets in the bottom panels are enlarged images of the framed areas. C, cortex; E, endodermis; M, mutant cell layer; S, supernumerary cell layers. Scale bars, 10 μm.

A large pool of SCR would be required to completely block SHR movement. The positive feedback loop for SCR transcription could provide such a mechanism (16). To test this hypothesis, we examined the effect of reduction in SCR expression on SHR movement using an RNA interference (RNAi) construct. We reasoned that, if SCR levels were reduced below a threshold level, some SHR protein might be able to move into the presumptive cortex where it would activate SCR transcription and endodermis specification. Asymmetric cell division would also occur, giving rise to an additional layer of ground tissue. This process could be repeated until free-moving SHR was exhausted. In support of our hypothesis, plants from the RNAi transgenic lines that we generated produced multiple layers of cells. Two lines that have different levels of SCR transcript were further examined (Fig. 3A). As shown in Fig. 3C, the extra cell layers in both lines express the endodermal marker pSCR::GFP, and SHR-GFP expressed in the stele is also present in these supernumerary cell layers. Notably, SHR-GFP is detected in both daughter cells of the CEI cell, whereas its levels appear to decrease in the outer cell layer after each additional cell division (Fig. 3C, insets). Moreover, the number of supernumerary cell layers is inversely correlated with the level of SCR transcript in the two independent transgenic lines. Furthermore, SHR-GFP is primarily nuclear-localized in the supernumerary layers of SCRi-1, the weaker RNAi line, but is largely cytoplasm-localized in SCRi-2, the stronger RNAi line (Fig. 3C). The two lines also showed reduced root length that correlates with the strength of RNAi (P values are 2.9 × 10–5 and 2.8 × 10–13, respectively; Student's t test, n = 39 roots), although the QC, CEI, and other initials appear normal (Fig. 3, A and B). These results demonstrate the critical role of the positive feedback mechanism for SCR in restricting SHR movement, root radial patterning, and root growth.

Our results support a mechanism by which SCR tightly restricts SHR movement, as described below. On the one hand, SCR sequesters SHR into the nucleus through protein complex formation, making SHR incapable of further movement. On the other hand, the SHR/SCR-dependent positive feedback loop for SCR transcription ensures no free-moving SHR can escape from the endodermis by driving a rapid buildup of SCR that does not self-interact but rather preferentially interacts with SHR. This mechanism would require a basal level of SCR expression to initiate the feedback loop. Notably, a substantial level of SCR mRNA is still detectable in both the shr and scr backgrounds, and its specific radial expression pattern is largely unaltered. This SHR/SCR-independent basal SCR transcription may be one of the key factors defining the boundary for SHR movement.

The model that we propose for SCR to restrict SHR movement could also account for the fact that different cell fates are rapidly acquired by the progeny of the daughter cells of the CEI cell (9). After this asymmetric cell division, the concentration of the SHR/SCR complex will remain high in the inner cell of the endodermal lineage driven by a sustained supply of SHR from the stele, which activates the SCR feedback loop. This high concentration of the SHR/SCR complex would maintain the expression of SCR as well as other downstream patterning genes. By contrast, the SHR/SCR concentration in the other cell of the cortex lineage would drop rapidly, resulting from the inability of SHR to move beyond the endodermis coupled with protein turnover and the dilution accompanying cell division. Indeed, although SCR is detected in both cells immediately after the asymmetric cell division, SCR and other endodermal markers are only expressed in the endodermis soon thereafter (9).

Interaction and expression of SHR and SCR homologs in rice. The observations that nearly all plants examined so far have only a single layer of endodermis (even though the number of cortex layers can be highly variable) and that SCR orthologs are exclusively expressed in the endodermis (1719) suggest that the mechanism described above is likely to be evolutionarily conserved. However, the only SHR homolog cloned so far, which is claimed to be the closest SHR homolog from rice, shows an expression pattern that is distinct from SHR in Arabidopsis (19), thus casting doubt on this hypothesis.

Database searches revealed that there are, in fact, two close rice homologs for both SHR (Os03g31750 and Os07g39820) and SCR (Os11g03110 and Os12g02870). We named the more similar SHR and SCR homologs OsSHR1 (Os07g39820) and OsSCR1 (Os11g03110) and the more dissimilar ones OsSHR2 (Os03g31750) and OsSCR2 (Os12g02870), respectively (table S2).

The rice genes that were previously reported as homologs of SHR and SCR are OsSHR2 and OsSCR1 (19). We therefore cloned OsSHR1 and analyzed its expression in rice roots by in situ hybridization. As shown in Fig. 4A, OsSHR1 and OsSCR1 are both expressed in tissues analogous to those of their counterparts in Arabidopsis. OsSCR1 and OsSHR1 interact in yeast as strongly as Arabidopsis SHR and SCR do (Fig. 4B). They also interact equally well with SHR and SCR, but no interaction was observed between OsSHR2 and OsSCR1 (Fig. 4B). These results strongly suggest that OsSHR1 and OsSCR1 are functional homologs of SHR and SCR in rice. They further suggest that the functional relationship between SHR and SCR, as well as their role in radial patterning in higher plants, is evolutionarily conserved.

Fig. 4.

Analysis of SHR and SCR homologs from rice. (A) In situ hybridization showing the expression patterns of OsSCR1 and OsSHR1 in rice root. The framed areas in the left panels are shown at a higher magnification (middle panels). Scale bars, 20 μm. (B) Yeast two-hybrid assay examining the interaction between OsSHR1 or OsSHR2 with OsSCR1, as well as their interaction with SHR and SCR. Ade, adenine.

Proteins that move as signaling molecules play a critical role in both animal and plant development (8, 20, 21). Although the list of transcription factors that are able to move is growing (2228), little is known about the mechanisms regulating intercellular movement. Decapentaplegic (Dpp), for example, a well-characterized example from animals, moves passively by diffusion and forms a gradient across multiple layers of cells as a result of unregulated binding to and internalization by its receptors located on the surface of the cells that it passes through (8, 29). By contrast, both SHR movement and its range of action are actively regulated, and the mechanism that we have uncovered in this study is quite distinct from those previously described. Although some aspects of this mechanism have been reported for other proteins, this is the first example where both protein-protein interaction and transcriptional control are involved to achieve tight control of protein movement. This difference may extend to other moving plant proteins and indicate a fundamental difference between plant and animal signaling during development.

Supporting Online Material

Materials and Methods

Figs. S1 to S5

Tables S1 to S3


References and Notes

View Abstract

Navigate This Article