Epidermal patterning genes are active during embryogenesis in Arabidopsis

Epidermal cells enclose the cortex and differentiate in a position-dependent manner along the root-hypocotyl axis of Arabidopsis. In the root, epidermal cells located over two cortical cell files develop as hair cells while those overlying a single cortical cell file develop as non-hair cells (Dolan et al., 1994; Galway et al.,1994). In the hypocotyl, epidermal cells lying over two files of cortical cells develop as stomatal complexes while those epidermal cells overlying a single cortical cell file develop as elongated cells(Gendreau et al., 1997; Berger et al., 1998a). The root meristem, from which the seedling root is derived, forms at the basal end of the root-hypocotyl axis and the cellular organisation of the meristem is in place by the late heart stage of embryogenesis(Scheres et al., 1994). A combination of genetic determinants and positional signals specifies the identity of epidermal cells in the seedling root(Galway et al., 1994; Lee and Schiefelbein, 1999; Masucci et al., 1996; Wada et al., 1997; Berger et al., 1998b). Enhancer-trap and promoter-reporter gene studies indicate that the root epidermal pattern is established during embryogenesis(Berger et al., 1998b; Lin and Schiefelbein,2001).

Characterisation of mutants with defects in the specification of epidermal cell types has defined a pathway for epidermal patterning in the root and in the hypocotyl of the seedling. GLABRA2 is a homeodomain protein required for non-hair fate in the root, and elongated epidermal cell fate in the hypocotyl(Masucci et al., 1996; Di Cristina et al., 1996; Berger et al., 1998a; Hung et al., 1998). Positive regulators of GL2 transcription in seedlings include WEREWOLF (WER), which encodes a MYB protein(Lee and Schiefelbein, 1999)and TRANSPARENT TESTA GLABRA (TTG), which encodes a WD40 protein (Galway et al., 1994; Walker et al., 1999). GL2 and WER are expressed in non-hair cells in the root and elongated cells in the hypocotyl of the seedling(Masucci et al., 1996; Di Cristina et al., 1996; Lee and Schiefelbein, 1999)and the accumulation of GL2 transcript in these cells requires WER function (Lee and Schiefelbein, 1999; Lee and Schiefelbein, 2002). CAPRICE (CPC) is a MYB protein and is required for the specification of root hair cell identity(Wada et al., 1997), yet it is expressed in non-hair cells (Lee and Schiefelbein, 2002; Wada et al., 2002). In cpc mutants only a few hairs are formed on the root (Wada et al., 1997),but no mutant phenotype has been described for the hypocotyl. TRIPTYCHON (TRY) encodes a CPC-related MYB protein and acts in a partially redundant manner with CPC to specify root hair cell identity (Schellmann et al.,2002).

It has been shown, on the basis of double mutant phenotypes and promoter-reporter gene studies, that a reciprocal transcriptional control between WER and CPC contributes to the establishment of the position-dependent expression of GL2 in the seedling root(Lee and Schiefelbein, 2002). WER positively regulates the expression of both GL2 and CPCin non-hair cells, while CPC moves into hair cells and represses the transcription of WER and GL2 in these cells(Lee and Schiefelbein, 2002; Wada et al., 2002). Lee and Schiefelbein (Lee and Schiefelbein,1999) proposed that a WER:CPC protein ratio could determine the fate of epidermal cells in the root. A high WER:CPC ratio would result in the specification of a non-hair cell while a low ratio would result in the formation of a hair cell. In the seedling, positional cues may be responsible for the expression of WER in non-hair cells and a high WER/CPC ratio,which initiates the cascade of molecular events that leads to position-dependent cell type specification in the root epidermis(Lee and Schiefelbein, 2002). The WER, CPC and GL2 genes have been shown to be required for the maintenance of pattern in the post-embryonic seedling(Berger, 1998a; Masucci et al., 1996; Di Cristina et al., 1996; Lee and Schiefelbein, 1999; Lee and Schiefelbein, 2002). Lin and Schiefelbein (Lin and Schiefelbein, 2001) have shown that GL2 is expressed during embryogenesis and is controlled by WER and TTG. Here we show the embryonic expression of WER, CPC and GL2 and describe how the complex transcriptional interaction between these genes starts in the embryo and defines the specific expression of GL2,which leads to epidermal cell differentiation in the root-hypocotyl axis of the seedling.

Wild-type Landsberg erecta (Ler) ecotype and gl2-1 (in Ler) were obtained form the Nottingham Stock Centre, cpc (in WS) was kindly provided by K. Okada (Kyoto University, Japan), and wer-1 (in Col-o) was kindly provided from J. Schiefelbein (University of Michigan, USA).

Tissue fixation and wax embedding were performed as described by Long and Barton (Long and Barton,1998). Tissue sectioning and in situ hybridisation of digoxigenin-UTP-labelled RNA probes were performed as described by Coen et al.(Coen et al., 1990). Antisense GL2 probe was generated using T7 RNA polymerase to transcribe the pGL5 cDNA clone that spans exon 5 of the GL2 gene after linearising with BamHI. Sense GL2probe was transcribed with T3 RNA polymerase from the same clone linearised with EcoRI. Antisense WER probe was generated using T7 RNA polymerase to transcribe the pWER1 clone, containing a 450 bp WERcDNA fragment that spans exons 2 and 3, after linearisation with BamHI. Sense WER probe was synthesised with T3 RNA polymerase from the same clone linearised with EcoRI. Antisense CPC probe was generated using T7 RNA polymerase, from the full-length CPC cDNA clone linearised with EcoRI, and the CPCsense probe was generated by transcription with T3 RNA polymerase, of the PCR amplified insert of the same clone. Antisense GFP probe was generated using T7 RNA polymerase to transcribe the clone containing the GFP5cDNA, after linearising with XhoI, and the sense GFP probe was transcribed with T3 RNA polymerase from the same clone, but linearised with BamHI. All the clones are in pBluscript SK+ (Stratagene).

For each gene, serial sections through embryos at different developmental stages were probed by in situ hybridisation and the experiments were repeated on independent samples.

Digital images were captured with a Nikon Cool Pix 995 using differential interference contrast (DIC) optics and assembled using Photoshop 5.

Embryos at different developmental stages were dissected from the seed coats with a hypodermic needle and aligned on the surface of a thin layer of Murashige and Skoog medium supplemented with 0.5% (w/v) Phytagel and 1% (w/v)sucrose to avoid embryo desiccation. Embryos were then imaged with a MRC600 Biorad confocal microscope using a 488 nm excitation line, and a 523 nm short-pass filter to monitor the expression of GFP. Optical sections of 2.5-3μm were collected, processed with the NIH Image program(http://rsb.info.nih.gov/nih-image/)and assembled using Photoshop 5. At least ten embryos for each of the heart,torpedo and mature developmental stages of the J2301 marker line and of the homozygous gl2/gl2 J2301/J2301, cpc/cpcJ2301/J2301, and wer/wer J2301/J2301 lines were analysed.

In the seedling GL2 is expressed in non-hair cells of the root and elongated cells of the hypocotyl (Hung et al., 1998; Berger et al.,1998a). These two cell types overlie the wall of single cortical cells and GL2 is required for their position-dependent specification(Masucci et al., 1996). To investigate the earlier temporal and spatial pattern of GL2expression during embryogenesis, GL2 mRNA was detected by in situ hybridisation on transverse and longitudinal sections of the root and hypocotyl of wild-type embryos at different developmental stages, as classified by Jürgens and Mayer(Jürgens and Mayer,1994). We consistently detected GL2 expression in heart stage embryos (Fig. 1). Series of consecutive longitudinal sections through wild-type heart stage embryos were hybridised with the GL2 antisense probe and alternate sections displayed a faint GL2 expression in protodermal cells at the base of the embryo (arrows in Fig. 1A,B). This suggests that protodermal cells, which are not in contact with each other, first express GL2 at the basal end of the embryo during the heart stage. During the transition between heart and torpedo stage, periclinal divisions in the protodermal cells give rise to an outer lateral root cap layer of cells and an inner epidermal layer(Scheres et al., 1994). At the torpedo stage GL2 transcript is detectable throughout root cap,lateral root cap and epidermal layer in transverse sections of the root(Fig. 1C,Da). In the hypocotyl, GL2 is detected in most of the epidermal cells that overlie a single cortical cell when viewed in transverse section(Fig. 1C,Db). At the mature stage, in both root and hypocotyl, GL2 is expressed in epidermal cells that overlie a single cortical cell and will become non-hair cells in the root and elongated cells in the hypocotyl(Fig. 1E,F). These data suggest that GL2 expression is first established in isolated protodermal cells at the base of the heart stage embryo, and by the torpedo stage the expression spreads to every cell of the root cap and epidermis of the root. Between the torpedo and the mature stage GL2 expression is restricted to those cells that will become non-hair cells in the seedling. In contrast,in the hypocotyl, GL2 expression is already established in an alternate pattern by the torpedo stage, which is then maintained throughout the subsequent stages. The temporal difference in the establishment of the cell-specific expression pattern of GL2 between the root and hypocotyl suggests that during embryogensis the two epidermal cell types are defined earlier in the hypocotyl than in the root.

WER is required to repress the formation of hairs and stomatal complexes in those cells that contact a single underlying cortical cell in the root and the hypocotyl, respectively (Lee and Schiefelbein, 1999). To define when WER expression is established during embryonic development, in situ hybridisation studies were performed on wild-type embryos at different developmental stages. WERexpression was not detected in globular or heart stage embryos(Fig. 2Aa,Ab), but was first detected in torpedo stage embryos. At this stage WER mRNA is consistently present in consecutive transverse sections throughout the root cap, the lateral root cap and the epidermal layer in the root(Fig. 2B,Ca) and throughout the epidermal layer in the hypocotyl (Fig. 2B,Cb). Later in embryogenesis, in mature embryos, no WERexpression is detected in any cell type of either the root or the hypocotyl(Fig. 2D,E). Therefore, high levels of WER expression may be required for only a short period during embryogenesis.

CPC is required for hair cell specification in the root(Wada et al., 1997) and so far its function has not been characterised in the hypocotyl. Based on the role and expression of the other genes involved in epidermal patterning in the root and the hypocotyl we suspected that CPC might also be expressed during embryogenesis. CPC expression was first detected in the torpedo stage embryo and no expression was detected in earlier embryonic stages (Fig. 3Aa,Ab,Ac). At the torpedo stage, CPC mRNA was detected throughout the root in sections close to the basal end (Fig. 3B,Ca). CPC expression was also detected in epidermal and cortical cell layers of the hypocotyl (Fig. 3B,Cb). In the mature embryo CPC mRNA was present in the root cap and lateral root cap of the root and was also detectable at a low level throughout the root (Fig. 3D,Ea). In the hypocotyl CPC mRNA was only present in the vascular tissues (Fig. 3D,Eb).

It has previously been shown that WER positively regulates GL2 expression in seedlings (Lee and Schiefelbein, 1999). To test if WER controls GL2 in the embryo, the expression of each gene was determined in wer and gl2 mutant embryos by in situ hybridisation. No GL2 mRNA was detected in the root or hypocotyl of mature werembryos (Fig. 4A,B) but we found that in wild-type embryos GL2 is expressed in epidermal cells that overlie single cortical cells (Fig. 1F). This indicates that WER is required for the establishment and maintenance of GL2 expression in the embryo. Surprisingly, WER expression was detected throughout the epidermal cell layer of gl2 mutant mature embryos in both root and hypocotyl(Fig. 4C,D), whereas no expression of WER was previously detected in mature wild-type embryos(Fig. 2D,E). This suggests that GL2 is required to repress WER transcription in the mature embryo. Even though WER expression was not detectable in early embryonic stages, i.e. globular and heart stage(Fig. 2A), we wanted to test if WER is active and required for the onset of GL2 expression at the heart stage. No GL2 mRNA was detected in wer heart stage embryos (Fig. 4E,F). This data indicate that at the heart stage the level of WER expression is below detection but that the gene is active and promotes GL2expression.

To determine if CPC regulates GL2 expression in the developing embryo we examined the expression of GL2 in cpcmutant embryos. In cpc heart stage embryos GL2 expression is detected in protodermal cells at the basal end of the embryo, as observed in wild type (Fig. 4G,H; Fig. 1A,B) and in some protodermal cells in the central region (future hypocotyl) of the embryo. In mature cpc mutant embryos GL2 is expressed at a low level in the lateral root cap and in every epidermal cell of the root(Fig. 4I,Ja). This contrasts with the pattern of GL2 expression in wild-type in which epidermal GL2 expression is restricted to the future non-hair cells(Fig. 1Fa). To determine if GL2 can control CPC, the expression of CPC was examined in mature gl2 mutant embryos. The pattern of CPCmRNA localisation in mature gl2 mutant embryos is very similar to that found in wild-type embryos, i.e. it is located in the lateral root cap(Fig. 4K,La; Fig. 3D,E). Therefore, GL2 does not control the accumulation of CPC mRNA whereas CPC is required for the establishment of the specific pattern of GL2 expression in the root of the mature embryo but not for the initial expression of GL2 at the heart stage.

In the hypocotyl of mature cpc embryos GL2 mRNA was present in some epidermal cells that were located over the wall of single cortical cells (Fig. 4I,Jb), as in wild type (Fig. 1Fb). In gl2 embryos, CPC expression was present in the vascular precursor cells as in wild-type embryos(Fig. 4K,Lb; Fig. 1Fb). This indicates that GL2 does not require the activity of CPC for the maintenance of its expression in the hypocotyl of the mature embryo. Furthermore, as it is found in the root, GL2 does not control CPC expression in the hypocotyl.

To determine if WER controls CPC we examined the expression of CPC in the root and in the hypocotyl of werembryos. In wer torpedo stage embryos, the same CPCexpression was detected as in wild-type mature stage embryos(Fig. 4M,N; Fig. 3D,E). In wermature stage no CPC mRNA was detectable in any sections through wer mutant embryos hybridised with the CPC antisense probe(Fig. 4O), in contrast to wild-type mature embryos where CPC is expressed in the lateral root cap and in the provascular tissue in the hypocotyl(Fig. 3E) and WER is not expressed (Fig. 2E). These data indicate that WER positively regulates CPC transcription. To examine if there is a reciprocal control between WER and CPC we analysed the expression of WER in cpc mature embryos; no WERmRNA was detectable, either in the root or in the hypocotyl(Fig. 4P). As WER was also not expressed in the wild-type mature embryo(Fig. 2E), this suggests that CPC does not negatively regulate WER transcription at this stage.

We have shown that GL2, CPC and WER are expressed and active in the embryo. To further investigate their role during embryogenesis and to confirm that they are functional in the wild-type embryo, we analysed the GFP expression pattern of the root-hypocotyl epidermal marker line J2301 in gl2, cpc and wer mutants.

In seedlings the GFP enhancer trap line J2301(http://www.plantsci.cam.ac.uk/haseloff/Home.html)expresses GFP in non-hair cells of the root and in elongated cells of the hypocotyl (Berger et al.,1998a). Berger et al. (Berger et al., 1998b) indicated that GFP is expressed at the basal end of the mid-heart stage embryo in the derivatives of the hypophysis in J2301 enhancer trap embryos. Later, from the mid-torpedo stage onwards, GFP is expressed in the root epidermis and root cap cells. We repeated this analysis by confocal microscopy, but consistently detected no GFP expression in any embryonic stage prior to the torpedo stage(Fig. 5Aa). Instead we consistently found that isolated cells expressed GFP at the torpedo stage(Fig. 5Ab,Ad). It was difficult to determine if these GFP-expressing cells were located in the base of the hypocotyl or in the upper part of the root(Fig. 5Ab,Ac). By the end of the torpedo stage, GFP was expressed in alternate longitudinal files of two to three cells (Fig. 5Ad). At the root end of mature stage embryos every cell expressed GFP and above this zone cells expressed GFP in an alternative pattern(Fig. 5Ae).

It was not possible by confocal analysis to determine with certainty the identity of the cells that expressed GFP along the radial axis. It was also difficult to resolve the root from the hypocotyl. Therefore, we analysed the expression pattern of GFP by in situ hybridisation. Consecutive transverse sections from the root tip to the hypocotyl of late torpedo stage embryos were hybridised using an antisense GFP mRNA probe. In the root we detected GFP mRNA exclusively in the lateral root cap(Fig. 5B,Ca), while in the hypocotyl we detected GFP mRNA in those epidermal cells that overlay single cortical cells (Fig. 5B,Cb). These results indicate that the enhancer trap J2301 is first expressed in the embryo at the torpedo stage in isolated cells, and later, GFP expression continues in the lateral root cap cells of the root and in those cells of the hypocotyl that overlie the anticlinal cortical cell walls.

After characterising the GFP expression pattern of the J2301 enhancer trap we generated plants that carried the enhancer trap in gl2, cpc and wer mutant backgrounds. No GFP expression was detected by confocal microscopy in any embryonic stage of gl2 mutants(Fig. 6A), similarly, no GFP expression was detected in cpc and wer mutants (data not shown). To confirm these results we analysed the GFP expression by in situ hybridisation on longitudinal sections of mature embryos of GFP marker line J2301 in wild type, gl2 and cpc mutants. GFPmRNA was detected in the lateral root cap and hypocotyl of wild type(Fig. 6B,C), but none was detected in gl2 (Fig. 6D,E) or cpc mutants(Fig. 6F,G). These data indicate that GL2, CPC and WER genes are required to establish the expression of the GFP in the J2301 enhancer trap in both the root and the hypocotyl of the mature embryo. Therefore, these genes not only control their reciprocal expression in both the root and the hypocotyl, but also the expression of the epidermal marker line J2301. The results further prove that GL2, CPC and WER are functional during embryogenesis in the epidermis and lateral root cap of the root and in the epidermis of the hypocotyl.

We show that the WER, CPC and GL2 genes, which are required for the patterned differentiation of cells along the root-hypocotyl axis in the seedling (Galway et al.,1994; Masucci et al.,1996; Di Cristina et al.,1996; Wada et al.,1997; Berger et al.,1998a; Hung et al.,1998; Lee and Schiefelbein,1999; Walker et al.,1999; Wada et al.,2002), are expressed and active during embryogenesis. We show that these genes are expressed along the root-hypocotyl axis of the developing embryo and propose that a complex interaction of transcriptional regulation exists among WER, CPC and GL2 genes during the formation of epidermal pattern in the embryo.

We propose a model based on the regulatory interactions between WER,CPC and GL2 that occur during embryogenesis(Fig. 7). We have shown that WER positively regulates GL2 expression at the heart stage. Then, by the torpedo stage, GL2 expression has spread to all cells in the future epidermis. WER and CPC expression is then detectable and WER promotes GL2 expression throughout the epidermis. CPC is in turn required for the preferential accumulation of GL2 transcript in future non-hair cell, perhaps by negatively regulating GL2 transcription in hair cells position. In the mature embryo GL2 negatively regulates WER transcription and WER positively regulates CPC expression from the torpedo to mature stages. These events result in GL2 being expressed at high levels in the future non-hair cells and absent from the future hair cells in the mature embryo. The pattern of GL2 expression is then maintained in the root of the seedling and accounts for the pattern of hair cells and non-hair cells in the root epidermis, where GL2 negatively regulates hair formation in cells located in the non-hair position. The demonstration that WER activity is required for the transcription of GL2at the heart and mature stage and for the expression of CPC at the mature stage is at odds with the fact that we cannot detect WER mRNA at either stage by in situ hybridisation. It is possible that WERtranscript accumulates at levels that are too low to be detected by in situ hybridisation or that the protein is stable and persists despite the fact that the WER mRNA has disappeared. Further experiments are required to explain this observation.

In the hypocotyl, GL2 is required to specify the fate of elongated epidermal cells (Berger et al.,1998a; Hung et al.,1998; Lin and Schiefelbein,2001) and from the early stages of embryogenesis, GL2expression is restricted to those epidermal cells that will become elongated cells (this study). WER is required for the control of GL2expression in the hypocotyl of the embryo(Lin and Schiefelbein, 2001)(this study) and of the seedling (Lee and Schiefelbein, 1999). We show that CPC may also regulate GL2 transcription at the heart stage(Fig. 4G,H). Further support for this role of CPC comes from the observation that the J2301 enhancer trap is not expressed in the cpc hypocotyls in the mature embryo (Fig. 6F,G). This indicates that during embryogenesis, both WER and GL2 are required for the specification of epidermal cell identity throughout the entire length of the root-hypocotyl axis. In contrast CPC does not control the position-dependent expression of GL2 expression in the embryonic hypocotyl, as it does in the root. It is therefore likely that other MYB-related proteins are required for the restriction of GL2expression in the hypocotyl and that CPC may have other functions not related to epidermal cell specification in the hypocotyl since it is also expressed in the provascular region.

During embryogenesis the different domains of the embryo are progressively laid down and by the late heart stage, the cellular organisation of the future root is in place (Scheres et al.,1994) and GL2 mRNA is detectable. Tissues are concentrically organised in the root (Dolan et al., 1993) and it is likely that the radial organisation of the tissue layers is laid down first, and then later, the epidermal patterning is superimposed on the pre-existing radial elements. The epidermal pattern is then completed when GL2 expression is restricted to the cells in the non-hair cell position of the root at the mature stage and to the cells in the elongated cell position of the hypocotyl at the torpedo stage. In the seedling, the pattern established during embryogenesis is then propagated in a position-dependent manner, under the control of information that might be located in the cell wall of the underlying layer of cortical cells(Berger et al., 1998b). The model that we propose for the establishment of the root and the hypocotyl epidermal pattern differs from those reported by Lin and Schiefelbein(Lin and Schiefelbein, 2001)and Berger et al. (Berger et al.,1998b). Based on GL2 promoter-reporter gene expression studies, Lin and Schiefelbein proposed that the epidermal cell patterning mechanism in the root initiates during the early heart stage and it occurs before the establishment of a functional meristem. This interpretation is clearly correct for the patterning of the hypocotyls but contentious for the patterning of the root because it is impossible to resolve the precise patterns of expression in the root pole from the published images. Furthermore, Berger et al. (Berger et al.,1998b) proposed that domains of positional information in the root epidermis are established by the torpedo stage. This study was based on a confocal microscope analysis of the GFP expression pattern of the J2301 line,an enhancer trap that is expressed in non-hair cells of the root and elongated cells of the hypocotyl in the seedling. We have now shown that in the mature stage embryo the enhancer trap J2301 expresses GFP in the lateral root cap and epidermis of the hypocotyl. Hence, the gene expression pattern observed by Berger et al. (Berger et al.,1998b) was probably in the epidermis of the basal end of the hypocotyl and could not have been in the root epidermis. This is consistent with our view that the epidermal cells of the hypocotyl are patterned by the torpedo stage and the root pattern is established later.

We can conclude from the results of these two previous studies and this one that in the hypocotyl the initiation of epidermal pattern occurs in the heart stage and is completed by the torpedo stage. The development of cell patterning in the root epidermis is also initiated in the heart stage embryo,but it is not completed until the embryo reaches the mature stage. The difference between root and hypocotyl in the timing of the establishment of epidermal pattern may reflect the presence of a gradient of positional information along the apical-basal axis of the embryo. Alternatively, it may reflect the presence of an equivalent source of positional information along the apical-basal axis that it is interpreted at different developmental stages in the root and in the hypocotyl.

GL2 mRNA is first present in isolated epidermal cells at the root pole in the heart stage embryo. By the torpedo stage GL2 is present in every epidermal cell and by the end of the mature stage it is restricted to cells that will develop as non-hair epidermal cells. This restriction of GL2 expression to future non-hair cells is regulated by genes whose transcripts do not accumulate in the same cell-specific manner as GL2, i.e. in non-hair cells. For example CPC transcripts are found in a number of different cell layers and in every cell of the future epidermis. This suggests the spatial distribution of WER and CPC transcript alone cannot account for the pattern of GL2transcript accumulation and suggests that cell to cell movement of a signal that is controlled by CPC or WER may play an important role in the establishment of epidermal pattern. Since the cell to cell movement of CPC protein in the epidermis of seedling roots has been demonstrated, CPC is a strong candidate for such a mobile signal(Wada et al., 2002).

The restriction of GL2 transcript to non-hair cells indicates that the development of the pattern in the epidermis is progressive – there is early widespread GL2 expression throughout the epidermis that is later restricted to cells in the future non-hair location. This progressive restriction of gene expression is analogous to the process that occurs during the development of sensory bristles in Drosophila. Early in development the proneural genes are expressed in large fields of cells only to become restricted to individual cells later(Simpson et al., 1999). The restriction of proneural expression to individual cells is mediated by cell-cell interactions. It is therefore likely that cell-signalling events may act in the Arabidopsis embryo between the heart and torpedo stages. The molecular basis of these interactions remains to be elucidated.

We thank Des Bradley for help, advice and critical reading of the manuscript, Fred Berger for advice and Nigel Kilby for providing the CPC cDNA clone. This research was supported by the Gatsby Charitable Foundation and the BBSRC.