Position dependent control of cell fate in the Fucus embryo: role of intercellular communication

In animals, cell fate decisions during embryogenesis can arise from cellular interactions with neighbouring cells (induction) and from the asymmetric segregation of cellular factors which specify cell lineages (Gurdon, 1992). A major distinguishing feature of plant development is its indeterminate nature which, coupled to a considerable morphogenetic plasticity in response to external stimuli, allows the plant, by modifying its shape and physiology, to adapt to its environment. Thus the mechanisms underlying developmental pattern must also accommodate the plasticity of plant development. In plants, only the developmental axis and zones of continuously differentiating tissues (meristems) are laid down during early embryogenesis (e.g. Goldberg, 1994) and new organs and tissue are continuously formed throughout the life of the plant. Therefore by the end of embryogenesis overall pattern is established but morphogenesis is limited. In root and shoot apical meristems, positional information rather than cell lineage plays an essential role in fate decisions (Van den Berg et al., 1995; Steeves and Sussex, 1989). Several lines of evidence suggest that cell fate determination and pattern formation in plant embryos are also under the control of positional signalling. Transected carrot embryos are capable of restoring missing parts in a position-dependent manner (Schiavone and Racusen,1991). However, this regeneration involves de-differentiation and re-differentiation of cells at the cut surface unlike, for example, limb regeneration in amphibians in which cell migration plays an important role (Brockes, 1994). The nature of positional information in plant embryos remains obscure. At least three mechanisms have been invoked to account for positional signalling. Intercellular diffusion of morphogens may occur via plasmodesmata (Lucas, 1995) and hormonal gradients have been shown to be involved in patterning plant tissue and organs (Liu et al., 1993). Finally positional information appears to be imprinted in the cell wall of the 2-celled Fucus embryo (Berger et al., 1994).

The embryo of the brown alga Fucus presents a very simple system to study early zygotic development in plants. The axis of embryonic polarity is established in the zygote in response to external cues (e.g. Goodner and Quatrano, 1993). The early embryo is composed of only two cell types, i.e. thallus cells and rhizoid. Moreover, the embryo is free of maternal tissues and easy to manipulate. We have undertaken a systematic microsurgical approach to understand better the mechanisms of cell fate control and patterning in the developing embryo. Here we show that the cell wall is essential for maintenance of polarity in isolated cells. A gradient of positional information regulates the nature of cell regeneration in response to ablations. Later stages of embryogenesis are characterised by increased plasticity in the response to cell ablation or microsurgical dissection. After the 2-cell stage, intercellular communication involving extracellular diffusible factors appears to play an important role in fate decisions and in patterning the embryo.

Mature receptacles of Fucus spiralis were washed, blotted dry and stored at 4°C in the dark. Gametes were released into sterile filtered sea water (FSW) at 17°C over a period of 1 hour by standard procedures (Quatrano, 1980). Fertilised zygotes were settled on coverslips in culture chambers and were polarised in unilateral white light and cultured for up to 10 weeks at 17°C in FSW.

Culture chambers were placed on the stage of an inverted microscope (Nikon, Tokyo, Japan). Embryos were plasmolysed in FSW containing 1 M sorbitol. Plasmolysis produced a 10-20 μm space between the plasma membrane and the cell wall. The cell wall was removed locally using laser microsurgery (Berger et al., 1994; Taylor et al., 1996). Osmoticum was then gradually replaced by FSW containing 0.6 M sorbitol. During subsequent deplasmolysis, the rhizoid or thallus protoplast expanded through the ablated opening in the cell wall. Nucleated protoplasts which regenerated cell wall within 1-2 hours following extrusion (Berger and Brownlee, 1995) were subsequently cultured in FSW.

Specific cells in embryos were ablated with a pulsed nitrogen UV laser (Berger et al, 1994). Cells were either completely ablated (removing most of the cytoplasm) or simply irradiated (i.e. killed by laser irradiation leaving cellular debris and cell walls intact (e.g. Berger et al., 1994). Laser-treated embryos were subsequently cultured for up to 6 weeks in FSW. For photomicroscopy of embryos older than 12 days, embryos were mounted in Hydramount (Burr, Searle Scientific Services, High Wycombe, Bucks, UK) before observation. This allowed visualization of individual cells in the multicellular embryo. Typically, approximately 100 regenerating embryos (n=50-200) were counted for each type of dissection.

Rhizoid cells of 3-celled or 8-to 12-celled embryos were microinjected with FITC dextran (10 kDa) using pressure pulses applied to an intracellular electrode as described previously (Roberts et al., 1994; Taylor et al., 1996). Embryos were pre-incubated in FSW containing 0.7M sorbitol to reduce internal turgor pressure prior to injection. Following injection, embryos were transferred to FSW and cultured as described above. Dye fluorescence was recorded with a cooled CCD camera (Digital Pixel, Brighton, UK) on a Nikon Diaphot fluorescence microscope.

The first asymmetric division of the polarised zygote (Fig. 1A,B) produces two cells with striking morphological differences. The thallus cell is spherical and highly pigmented while the rhizoid cell is elongated and has fewer chloroplasts (Fig. 1C). Embryonic pattern becomes well established during subsequent divisions. The second division normally occurs in the rhizoid, producing 3 cells: thallus (T), apical rhizoid (AR), and basal rhizoid (BR) (Fig. 1D). The embryo further divides to produce apical thallus (AT), basal thallus (BT), apical rhizoid (AR) and basal rhizoid (BR1 and BR2) cells (Fig. 1E). After 48 hours, 8 cells (2 each of AT, BT, AR and BR) are visible when viewed from above although a further longitudinal division of thallus cells occurs in a plane perpendicular to the first longitudinal division to produce 8 thallus cells. Since the precise timing of this division is not certain we will refer to the 48 hours embryo as 8-to 12-celled. The thallus cell gives rise to an elongated multicellular structure which after 3 to 4 weeks usually produces filamentous hair cells at its apex (Fig. 1F). An apical meristem is subsequently laid down at the base of the hair cells (Nienburg, 1931). The primary rhizoid cell generates branches of secondary rhizoid cells which allow the embryo to attach firmly to the substratum. Thus, two different developmental programs can be distinguished in the early embryo. However these programs are not irreversibly fixed in later developmental stages, even during normal development. At about 8 weeks, changes frequently occur (in up to 90% of embryos in culture) locally in the orientation of cell divisions in the rhizoid associated with increased pigmentation of these cells. Thallus-like structures develop from these cells as confirmed by the presence of filamentous apical hairs (Fig. 1G). Occasionally these structures can be observed very close to the base of the rhizoid (Fig. 1H,I), indicating clearly their rhizoid origin. Hence the first rhizoid cell generates not only the holdfast but also a large part of the adult plant (Fig. 1J,K), whereas the initial thallus cell normally gives rise only to thallus tissue.

Laser microsurgery was used to ablate specific cells in the embryo (Berger et al., 1994). After ablation of all rhizoid cells in 8-to 12-celled embryos (Fig. 2A), thallus cells above the ablation switch their fate and re-differentiate into rhizoids in less than 3 days (Fig. 2B). Following the ablation of all thallus cells and AR cells (Fig. 2C), filamentous BR cells remaining below the ablation become spherical and pigmented, eventually forming a thallus-like structure after approximately 2 weeks (Fig. 2D). The fact that isolated cell types have the ability ultimately to regenerate embryos demonstrates the absence of long-lived cytoplasmic determinants.

Eight-week old multicellular embryos were surgically transected using a microscapel (Fig. 3A, B). After 2 weeks, rhizoid regeneration was observed from both cut surfaces in more than 80% of transected embryos (n=50) (Fig. 3C-F). This regeneration occurred independently of the level of the transection along the apical-basal axis of polarity, suggesting that at this stage of development, rhiozoid regeneration can be induced following any type of dissection.

Following sequential laser ablations of all types of basal cells in 8-to 12-celled embryos, e.g. AR+BR (Fig. 3A,B), BR (Fig. 4) or BR+AR+BT (not shown), cells immediately apical to the dissection, including thallus cells, have the ability to regenerate rhizoids from their basal pole. We took advantage of this to try to understand how rhizoid regeneration is regulated in early embryos. All cells of a specific type (i.e. AT, BT, AR or BR) were ablated (Fig. 4A) and regeneration was monitored for up to 6 weeks following ablation (Fig. 4B,C). All cell types were ultimately capable of producing rhizoids in response to ablations (Fig. 4C). However, the frequency and the timing of rhizoid regeneration were highly dependent on the position of the ablation. Three days following BR ablation all embryos exhibited new rhizoids arising from adjacent AR cells (Fig. 4B,C). In contrast, no rhizoids were observed for up to 12 days following AT or BT ablation and fewer than 25% of these embryos regenerated rhizoids even after 6 weeks, ablated cells being largely replaced by new thallus cells (Fig. 4B). Following AR ablation, rhizoid regeneration was detected from adjacent BT cells after 12 days in 70% of embryos. These results indicate the existence of an apical-basal gradient of positional information controlling rhizoid development in the embryo.

Protoplasts extruded from either thallus or rhizoid cells of 2-celled or 8-to 12-celled embryos are able to re-polarise in response to the prevailing light direction, produce a new rhizoid within 24 hours and develop into apparently normal embryos (Fig. 5A,B; see also Berger et al., 1994). In contrast, following ablations of all neighbouring cells in 8-to 12-celled (Fig. 5C) or 3-celled embryos (Fig. 5E), all types of cells conserve the ability to re-differentiate forming new rhizoids at their former basal pole and thallus tissue at their apical pole even when the polarising unidirectional light is reoriented (Fig. 5D,F). Rhizoid regeneration from BT-derived cells could be observed albeit with low frequency (25% only after 42 days) following ablation of AT cells (Fig. 5G,H, see also Fig. 4C). In this case it is unlikely that rhizoid induction is under the control of factors in the cell wall.

Rhizoid regeneration following ablation of all cells of a specific type in 8-to 12-celled embryos was monitored with respect to the presence of remaining rhizoid cells below the ablation (Fig. 6). Thus, cells adjacent to BT were sequentially ablated, (i.e. ablation of AR, AR+BR1, AR+BR1+BR2 or AT; Fig. 6A). To determine precisely the proportion of embryos in which rhizoids regenerated from BT, control ablations were also performed. For example, the control for AR ablation consisted of leaving only BR1 and BR2 alive and following rhizoid regeneration (not shown). In this case, rhizoid regeneration against the polar axis from isolated BR2 or BR1+BR2 after 21 days occurred in fewer than 5% of dissections. More than 50% of embryos regenerated rhizoids from BT within 3 days following ablation of AR+BR1+BR2, 5 days following AR+BR1 ablation and 12 days following AR ablation (Fig. 6B,C). Rhizoid regeneration from BT following ablation of AT was not observed until 12 days and reached its maximum (30% of embryos) only after 21 days (Fig. 6C). Therefore the timing and frequency of rhizoid induction from BT is correlated to the number of rhizoid cells left below the dissection. We propose that an inhibitor of rhizoid regeneration may be produced by rhizoid cells and can diffuse or is transported to the thallus where it normally prevents rhizoid formation (Fig. 6D).

Parallel experiments were carried out with 3-celled embryos (Fig. 7). When AR alone was ablated (Fig. 7A), leaving BR intact, rhizoid regeneration from T was not detected until 5 days and a majority of the embryos produced new rhizoids only after 12 days (Fig. 7B,C: left graph). All embryos regenerated rhizoids from T within 3 days after the ablation of AR+BR. Ablation of T in 3-celled embryos allowed rhizoid regeneration from AR against the normal embryonic polarity, to be monitored (Fig. 7A, right graph). Rhizoids were produced earlier and more frequently from the thallus pole of AR when BR was also ablated (Fig. 7B,C, right graph). Of course in this case new rhizoids were also produced within 3 days at the rhizoid pole of AR as early as 3 days following ablation. These results also indicate that a diffusible factor produced by rhizoid cells may inhibit thallus-to-rhizoid re-differentiation (Fig. 7D, left).

Taken together these results suggest that re-differentiation of thallus cells into rhizoid cells in response to ablation of adjacent cells is strongly influenced by the presence of pre-existing rhizoid cells. Thus, remaining rhizoid cells are able to inhibit the re-differentiation of new rhizoids from thallus cells. However, it is notable that rhizoid cells have greater potential to regenerate rhizoids from their basal pole, even in the presence of remaining rhizoid cells. Thus, following ablation of BR1 in 8-to 12-celled embryos, rhizoids were produced as quickly from AR (100% after 3 days) in the presence as in the absence of BR2 (not shown).

To test whether intercellular communication was involved in the control of rhizoid regeneration by existing rhizoid cells, FITC-dextran was injected into individual rhizoid cells of 3-celled or 8-to 12-celled embryos. FITC-dextran injected into the cytoplasm of rhizoid cells could be detected in all other cells of the embryo, including thallus cells within 24 hour (Fig. 8A-C). When sub-basal cells (e.g. BR1 in 8-to 12-celled embryos) were injected, fluorescence was detected first in more basal cells (Fig. 8A) and subsequently in more apical cells (Fig. 8B). Thus molecules at least as large as 10 kDa are able to move via symplastic connections between cells in both directions along the axis of polarity. Normal embryogenesis proceeded after injection of any cell type, resulting in multicellular embryos with all cells fluorescently labelled (Fig. 8C).

Microinjections of FITC dextran were also carried out in conjunction with laser ablations. One day after microsurgery rhizoid cells (i.e. BR) immediately basal to ablated AR cells were microinjected. Ablation of AR (Fig. 8D,G) irreversibly prevented the movement of FITC dextran to the cells above the ablation (Fig. 8E,H). Fluorescence was still detected in all cells below the ablation 1 week after the microinjection but was never observed in the cells above the ablation or their derivatives (Fig. 8E,H). The absence of fluorescence in regenerated rhizoids confirms their origin from cells apical to the ablation gap (Fig. 8E,F). The inhibitory effect on rhizoid regeneration by pre-existing rhizoid cells in microinjected embryos, despite the disruption of symplastic connections (Fig. 8E-H, see also Figs 6 and 7), suggests that symplastic communication was not involved in this control of rhizoid regeneration.

The results presented here show that a frequent response to cell ablation in 3-celled to 8 week old embryos is the production of cells that are morphologically indistinguishable from rhizoid cells. In early embryos (up to 48 hours), rhizoid production in response to cell ablation is tightly regulated in a position-dependent manner whereas in older multicellular embryos (3 to 8 weeks), rhizoid regeneration occurs independently of the level of the dissection and embryo polarity. In adult vegetative fronds of Fucus, the outgrowth of rhizoids from cortical cells has also been observed (Fries, 1984). Thus, developmental plasticity appears to be more restricted in early embryonic stages compared to later stages.

Clonal analysis in maize (Dawe and Freeling, 1991) as well as fate mapping experiments in shoot and root meristems of Arabidopsis (Van den Berg et al., 1995; Szymkowiak and Sussex, 1996) have indicated that fate determination and differentiation are dependent on positional cues. Genetic analysis of Arabidopsis mutants such as fass (Torrez-Ruiz and Jurgens, 1994), reveals that pattern formation in embryos does not require properly oriented cell divisions, suggesting indirectly that pattern and cell fate determination is position rather than lineage sensitive in plant embryos. In Fucus embryos, normal embryonic pattern also appears to be independent of the pattern of cell division since transient treatments with brefeldin A (Shaw and Quatrano, 1996) or β-oestradiol (Pollock, 1969), which lead to abnormal patterns of cell division, can produce apparently normal polar embryos.

Analysis of Arabidopsis mutants such as monopteros, which lacks the basal or central embryonic regions and subsequently the root and hypocotyl, suggests that the embryo forms from autonomous regions arising from rigid lineages in the very early embryo (Berleth and Jurgens, 1993). Here we present evidence that pattern formation in early plant embryos proceeds by positional control rather than cell lineage. Both rhizoid and thallus cells and protoplasts have the ability to regenerate the other cell type following surgical ablation. Another striking example illustrating the lack of lineages is the regeneration of apical hair and meristem cells from cells descended from BT which occurs when cells which normally give rise to these cell types (i.e. AT), are ablated in the 8-to 12-celled Fucus embryo. Finally during normal development in late embryos and young adult plants, thallus tissue can form from rhizoid cells, suggesting that rhizoid cells are not irreversibly committed to a particular fate.

Despite the evidence that positional information is an important factor in plant development (Jurgens, 1995), its nature, particularly during embryogenesis is poorly understood. Previous work with 2-celled Fucus embryos has demonstrated that rhizoid development from isolated thallus cells following laser ablation can be rapidly induced by contact of the remaining cell with the residual cell wall of the ablated cell but not if this contact was prevented (Berger et al., 1994; Kropf et al., 1993). Here we show that at later developmental stages, rhizoid production from a thallus cell can ultimately occur in the absence of intimate contact between a cell and the remaining wall of the dissected cell. Thus no differences were observed in the times and frequencies of regeneration between embryos in which cells were simply killed by laser ablation and those in which cells were completely ablated (not shown). Moreover, thallus cells which never come in contact with rhizoid cells, e.g. following dissection of AT in 8-to 12-celled embryos, are ultimately capable of re-differentiating into rhizoids in up to 25% of embryos. However, this redifferentiation from later stage embryos takes much longer in comparison with re-differentiation in ablated 2-celled embryos and only after a number of cell divisions have taken place. It can therefore be argued that different mechanisms are operating in the control of re-differentiation in embryos beyond the 2-cell stage. Furthermore isolated single cells in 3-celled and older embryos have the ability, as do isolated protoplasts, to de-differentiate and re-differentiate, though while isolated protoplasts fix a new polar axis before regenerating (Kropf et al., 1988; Berger et al., 1994), isolated single cells retain their original apicalbasal axis. However differences in patterns of re-differentiation must be considered when comparing protoplasts from 2-celled embryos with isolated cells from older embryos. Whereas the isolated protoplast re-differentiates directly, in isolated cells of older embryos re-differentiation is only apparent from daughter cells following a number of cell divisions. Together, these results suggest that after the 2-cell stage, pattern formation is not exclusively under the control of positional information in the cell wall. Therefore the extracellular matrix could provide the necessary information required for polarisation (Kropf et al., 1988) and fate specification (Berger et al., 1994) in both the zygote and the 2-celled embryo and for the retention of polarity in isolated intact cells from older embryos, whereas additional mechanisms are also likely to be involved in directing pattern in later embryonic stages.

Laser microsurgery experiments with 3-celled and older embryos indicate that all cells of the Fucus embryo have the potential ultimately to produce rhizoids in response to ablations. It is clear that cells derived from those adjacent to ablated cells take much longer to regenerate restoration of cell continuity between neighbouring cells following cell ablation, as indicated by the loss of symplastic connections. Secondly, the timing and frequency of rhizoid rhizoids against the embryonic polarity. Cells derived from rhizoid cells have greater capacities to regenerate rhizoids than those derived from thallus cells. Moreover, our results show that rhizoid regeneration from thallus tissue (but not from rhizoid tissue) is strongly influenced by the presence of residual intact rhizoid cells.

Cell contact inhibition could possibly account for the inhibition of rhizoid regeneration from apical thallus tissue in ablation experiments. However this is unlikely for the following reasons. Firstly, residual cell wall prevents complete regeneration from similar cell types, i.e. from AT following dissection of BT or from BT following dissection of AR, are fundamentally different despite a quick re-establishment of physical contact between neighbouring cells.

The frequency of rhizoid regeneration following different ablations is directly influenced by the number of rhizoid cells remaining basal to the ablation. Thus an inhibitory substance may be produced by rhizoid cells and move between ablated cell layers to thallus cells (Fig 9). The fact that the capacity of cells to regenerate rhizoids following dissections increased with time after ablation suggests that the influence of factors causing inhibition of rhizoid generation decreased with time.

Thus one possible explanation is that rhizoid inhibitory factors become progressively diluted or that intercellular communication is decreased at later embryonic stages (Fig. 9). In older embryos and young plants, control of thallus differentiation from rhizoid tissue is likely to be exerted by existing thallus tissue (Fig. 9). Thallus tissue and meristem normally differentiate from rhizoid tissue after 8 weeks. The process of re-differentiation of rhizoid cells into thallus cells was originally described as ‘adventive embryony’ (McLachlan et al., 1971). This normal stage of development in Fucus spiralis is likely to account for the ramification often observed in older embryos and adult plants (Fig. 1) (McLachlan et al., 1971). The model of control of thallus differentiation from rhizoid cells is compatible with the demonstration of apical dominance in Fucus whereby a pre-existing apical meristem inhibits the development in its vicinity of new meristematic regions (Moss, 1966).

Parallels have been established between the morphologies and division patterns of Fucus and Arabidopsis early embryos (Brownlee and Berger, 1995). The analysis of Arabidopsis suspensor mutants such as raspberry, twin or sus demonstrates the potential of the suspensor to develop into embryo-proper like structures (Yadegari et al., 1994; Vernon and Meinke, 1994; Schwartz et al., 1994). It has been proposed recently (Zhang and Sommerville, 1997) that the likely involvement of intercellular communication in patterning the Arabidopsis embryo suggests that differences may exist between the cell wall-related signals in Fucus and those involved in higher plant embryos. Here we show that in Fucus, as in Arabidopsis embryos, communication between different cell types in the embryo does indeed appear to play a role in the regulation of fate specification and pattern.

Most of our knowledge about the nature of intercellular signals in the control of cell fate and differentiation come from studies with higher plants. For example genetic and molecular analysis of pistillata (pi) mutants in Arabidopsis have shown that pi acts non-cell autonomously and that it may regulate the production of a diffusible morphogenetic signal (Bouhidel and Irish, 1996). Similarly, morphogenetic gradients are likely to be involved in regulating the development of suspensor and embryo proper in Arabidopsis as suggested by the analysis of suspensor mutants (Schwartz et al., 1994).

Several lines of evidence suggest that cell-to-cell communication via symplastic connections can influence fate decisions in plants. In fern gametophytes the distribution and number of plasmodesmata predicts cell fate suggesting that intercellular signalling may orchestrate patterning (Tilney et al., 1990). In maize, the mRNA encoding the homeodomain protein KNOTTED 1 is selectively transported through plasmodesmata to neighbouring cells (Lucas et al., 1995). The symplastic trafficking of mRNAs or proteins such as DEF or GLO can account for non-autonomous control of meristem development in Antirrhinum (Perbal et al., 1996). Segregation and gradients of transcripts have also been detected in Fucus zygotes and embryos and may be relevant to polarisation and early embryogenesis (Bouget et al., 1995, 1996). Furthermore, plasmodesmata-like structures exist in sieve plates of midrib and wing regions of adult Fucus (Fielding et al., 1987) though they have not been observed in early embryos (Brawley et al., 1977). Since microinjected molecules at least as large as 10 kDa can pass between both rhizoid and thallus cells, these cells must be symplastically connected. However laser ablation irreversibly destroys symplastic communication even when physical contact between new neighbours is re-established. Thus the inhibition of re-differentiation by thallus cells above an ablation zone into rhizoid cells by the presence of rhizoid cells below the ablation zone, suggests that the operation of cell-to-cell communication influencing cell fate can occur via the apoplast.

Although there is considerable evidence for the involvement of hormonal gradients in patterning plant tissues (Uggla et al., 1996; Carland and McHale, 1996, Masucci and Schiefelbein, 1996), the contribution of hormones to positional information in embryos is less clear. In Phaseolus vulgaris, gibberellins produced by the suspensor enhance transcription in the embryo proper (Walthall and Brady, 1986). More significantly, the effects of polar auxin transport inhibitors and the analysis of pin-1-1 mutants of Arabidopsis show that polar auxin transport is required for the establishment of bilateral symmetry during embryogenesis (Schiavone and Cooke, 1987; Liu et al., 1993). Auxin activity has long been known in Fucus (du Buy and Olson, 1937) and has been suggested to influence embryonic polarity (Jaffe and Neuscheler, 1969). Moreover, inhibitors of polar auxin transport suppress apical dominance in Fucus apices (Moss, 1964).

Overall our results support the idea that cell fate is regulated non-autonomously in Fucus embryos. In the zygote and 2-celled embryo positional information in the wall is required to fix the axis of polarity and establish the initial pattern (Berger et al., 1994). After the 2-cell stage, the cell wall is still required to retain the initial apical-basal polarity in intact isolated cells. However after the 2-cell stage intercellular communication is essential for regulation of cell fate decisions but appears to involve neither symplastic transport of inhibiting substances nor direct cell contact. Rather, apoplastic diffusible factors may be involved. This is associated with increased plasticity and decreased influence of the polar axis in patterning.

We would like to thank D. Nicholson, R. Bellé and J. Orion for assistance with photographic work, R. Williamson and A. Chrachri for Nomarski microscopy and A. Asensi and F. Corellou for fixation techniques and helpful comments. F.-Y. B. was a recipient of an EEC long term postdoctoral fellowship (BIO4CT965064). This work was also supported by the BBSRC and the Marine Biological Association, UK.