Asymmetric cell division in plants: mechanisms of symmetry breaking and cell fate determination

Abstract

Asymmetric cell division is a fundamental mechanism that generates cell diversity while maintaining self-renewing stem cell populations in multicellular organisms. Both intrinsic and extrinsic mechanisms underpin symmetry breaking and differential daughter cell fate determination in animals and plants. The emerging picture suggests that plants deal with the problem of symmetry breaking using unique cell polarity proteins, mobile transcription factors, and cell wall components to influence asymmetric divisions and cell fate. There is a clear role for altered auxin distribution and signaling in distinguishing two daughter cells and an emerging role for epigenetic modifications through chromatin remodelers and DNA methylation in plant cell differentiation. The importance of asymmetric cell division in determining final plant form provides the impetus for its study in the areas of both basic and applied science.

Introduction

Coordination of cell division and expansion is critical for the generation of different tissues in a multicellular organism. Whereas symmetric divisions generally amplify homogenous cell populations, asymmetric cell division (ACD) allows for the production of diverse cell types, while still maintaining stem cell populations for additional growth [1]. An asymmetrically dividing cell must polarize molecular components such as mRNA, proteins, cell-surface markers, or organelles unequally to create distinct cellular domains [2]. Once cellular asymmetry is achieved, coordination of the division plane with the segregation of these components is needed so that the resulting daughter cells are not equal. Although polarization is a common prerequisite for asymmetric division, daughter cell asymmetry can also be achieved without generating visible physical differences via polarization. A seemingly symmetric division can result in identity changes in daughter cells through positional or external cues.

Factors that cause asymmetries during division can be described as “intrinsic” (from within) or “extrinsic” (from outside). Intrinsic factors are those that display a recognizable asymmetry in the parental cell prior to cytokinesis. Extrinsic refers to differences in local external signaling that produce cell identity changes after division. The use of intrinsic processes to produce daughter cells with different cell fates is generally favored when cells experience variable environments due to migration or in unicellular organisms where surrounding conditions can change rapidly. In contrast, extrinsic control of asymmetry is predominant when cells are exposed to highly ordered and predicable environments, where migration is limited or altogether restricted [3]. Recent works in both plants and animals suggest that both intrinsic and extrinsic processes can play a role in polarization, division, and cell fate. In a complex organism, both processes are important with one type playing a more prominent role over the other under specific conditions. Regardless of process or organism, the outcome of all divisions needs to be both predicable and reliable.

Although many aspects of plant and animal development are homologous at the cellular level, these two groups have evolved different mechanisms for generating asymmetries during division. The mechanistic divergence likely results from the independent evolution of multicellularity between plants and animals as well as a major structural difference between plant and animal cells, the cell wall. The cell wall is a rigid, cellulosic extracellular matrix which provides support and space for intercellular communication [4]. Plant cell proliferation within this matrix restricts cell movement and prevents contact between plasma membranes of neighboring cells. This starkly contrasts the mobility and intercellular contact among animal cells. Based on this difference, it may not be surprising that the identification of homologous asymmetric division components is lacking. Despite the differences, some mechanistic similarities exist between the two systems. Here we present classic examples of ACD and cell fate determination mechanisms in animals and highlight similarities and novel differences used in plants.

Asymmetric divisions in animals

Animal intrinsic regulation

For over two decades, the Caenorhabditis elegans embryo has been an important model used to study key components of intrinsic ACD. Within the C. elegans zygote, prior to division, a clear inequality in the segregation of a highly conserved group of proteins called PAR (partitioning defective) proteins can be observed. PAR proteins (PAR1-PAR6 and PKC-3) were identified from genetic screens resulting in defective anterior–posterior polarity; producing daughter cells with altered fate or size [5–7].

Initial symmetry breaking of the zygote begins with the association of the sperm-derived centrosome with the cortex, which defines the posterior pole of the embryo [8]. This event is followed by the establishment of cortical and cytoplasmic asymmetries. Several PAR proteins, which are initially localized uniformly, begin to concentrate in the posterior or anterior end of the zygote to direct the segregation of cell fate determinants (Fig. 1a). PAR3 and PAR6 (PDZ-containing proteins) and protein kinase C (PKC-3) form a complex at the anterior end of the zygote. In contrast, PAR2 (ring-finger protein) and PAR1 (serine-threonine kinase) localize to the posterior pole. These PAR proteins engage in complex interactions with one another to help establish and stabilize the physically and functionally distinct PAR domains. Once inequality is stabilized, PAR proteins activate downstream effectors to bisect the zygote into two cells of unequal size, a larger anterior and smaller posterior cell. Localized cullin-dependent protein degradation can also contribute to the unequal segregation of cell fate determinants. Specifically, the CCCH zinc-finger proteins MEX5/6 are enriched in the anterior side [9] while another zinc-finger protein, PIE-1, is degraded in the anterior side of the zygote and thus retained in higher concentration in the posterior cell after division [10] (Fig. 1a). Currently, there is no evidence for extrinsic signaling in polarizing these events.

Fig. 1.

Intrinsic and extrinsic pathways determine asymmetric cell fate in animals and plants. a An intrinsic polarity pathway in animals is represented by PAR proteins that are differentially segregated (orange and green) in one cell stage C. elegans embryos. A anterior, P posterior. Polarized PAR proteins induce unequal degradation of the PIE-1 differentiating transcription factor. Arrows indicate positive regulation and blocked lines indicate negative regulation. b The organization of stem cell niche (SCN) in Drosophila female germ lines. Secreted morphogenes (BMP) from cap cells are perceived by the receptors (Tkv/Punt) in stem cells to inhibit the expression of the differentiation gene Bam. The expression of Bam in the daughter cells outside of the niche drive fate differentiation. c The intrinsic polarity pathway in plants is represented by the BASL polarity module in stomatal asymmetric division in Arabidopsis. The polarity complex is composed of BASL, the MAPKKK YDA, MPK3/6 and POLAR and is inherited in the stomatal lineage ground cell (SLGC) but not the meristemoid (M). Extrinsic signals (arrowhead) are hypothesized to trigger BASL polarization through the YDA MAPK pathway. The SPCH transcription factor is a direct substrate of MAPK for degradation. An unequal expression of SPCH is therefore hypothesized to associate with the BASL polarity complex. d The organization of stem cell niche in Arabidopsis root apical meristem. The WOX5 transcription factors in the quiescent center maintains the neighboring stem cell via the ACR4 receptor, which delivers CLE40 signals from the differentiating columella cells, to suppress WOX5 expression. The negative feedback loop between WOX5 and ACR4 maintains stem cell homeostasis in the root

A second prominent example of intrinsic polarization involves the establishment of unequal Notch signaling activation between daughter cells. Notch signaling is an evolutionarily conserved pathway for converting information from the exterior of the cell into a transcriptional response in the nucleus [11]. During sensory organ precursor (SOP) production in Drosophila, the Notch signaling inhibitor protein, Numb, is directed by the PAR proteins to be asymmetrically inherited in the sensory organ lineage and segregates only to one daughter cell [12]. The Numb-negative daughter cell has comparatively high Notch signaling, resulting in differential identities. Consistently, Numb loss-of-function mutants display inappropriate differentiation of daughter cells along with other severe defects [13]. Similarly in zebrafish (Danio rerio), the apical daughter cell of a neuronal stem cell retains higher levels of Notch ligands to ensure more Notch signaling response in the differentiating daughter cell [14].

Another important mechanism for differentially restricting Notch signaling in daughter cells is through endosomal trafficking routes. Endosomes fine-tune many processes in the cell via sorting, recycling, and storing of cellular cargo. In addition to the asymmetric inheritance of Numb described above, Notch signaling is differentially increased in one daughter cell during SOP production through directed sorting of Sara protein-marked endosomes [15]. Internalized Delta ligands and Notch receptors are asymmetrically sorted after endosomal packaging to influence Notch signaling in daughter cells. Polarized endosomal trafficking to produce biased Notch signaling has been quickly generalized to other stem cell systems including the Drosophila intestine [16] and the spinal cord of zebrafish [17]. The directionality of Sara endosome movement is controlled by activities of microtubule binding proteins, Klp98A (kinesin motor protein) and Klp10A (MT depolymerizing kinesin) along with its antagonist, Patronin [18].

Animal extrinsic regulation

Stem cells are often housed in a specialized and stable microenvironment called a ‘niche’, which provides extracellular cues to nurture and maintain stem cells that undergo self-renewing divisions [19, 20]. As the same basic paradigms govern stem cells in both flies and mammalians [21], we focus on some members of the core machinery of stem cell maintenance in the Drosophila germline stem cell (GSC) niches.

Drosophila GSCs divide perpendicular to hub cells/cap cells (the male and female niche, respectively). The orientation of this division ensures that one cell remains in contact with the niche and continues as a stem cell, while the other loses direct contact and differentiates. The direct contact between GSCs and the niche cells provides an attachment to anchor the stem cells and sets up local asymmetric signaling to repress differentiation [22]. The failure of GSCs to adhere to niche cells results in loss of stem cell recruitment and maintenance [23].

The specific signal secreted from niche cells in the Drosophila ovary and testis is BMP (bone morphogenic proteins) [24, 25]. BMP molecules are sensed by the GSC receptors, Thickveins (Tkv) and Punt, which ultimately suppress the expression of the master differentiation gene Bag of marbles (Bam) to maintain GSC identity [21, 26] (Fig. 1b). BMP diffusion beyond the GSC is restricted by extracellular matrix (ECM) collagens [27] and by the ligand-stabilizing molecule Dally (division abnormally delayed) produced by cap cells [28]. Differences in BMP responses are reinforced in the differentiating daughter cells through the expression of Bam [21].

In addition to active suppression of differentiation genes, the niche also controls the non-random segregation of organelles. In male GSCs and vertebrate neural stem cells, cells in contact with the niche preferentially retain the parental centrosome as long as they remain in the niche, while the daughter centrosome migrates to the opposite pole of the cell and is inherited by the daughter cell fated to differentiate [29, 30]. However, the association of parental centrosomes with the stem cell is not universal; Drosophila neuroblasts and female GSCs inherit the daughter centrosome during cell division [31, 32]. The non-random segregation of centrosomes contributes to asymmetric segregation of many cellular components, including sister chromatids [33], the midbody [34, 35], and the mitotic spindle that positions cleavage furrow [36, 37]. Interestingly, non-random sister chromatid segregation provides a means for asymmetric distribution of histone modifications to daughter cells. Newly synthesized histones (H3) are preferentially inherited by Drosophila differentiating goniablasts [38] and specific phosphorylation (H3T3) by the Haspin kinase distinguishes parental H3 to be retained in male GSCs [39].

Overall, the view on cell fate specification in specialized niches has focused on unique environmental signals for polarizing divisions and defining stem cells [21]. However, studies in Drosophila have identified proteins called Wicked (a component of the nucleolar RNP complex) and Under-developed (component of RNA pol II complex) that are asymmetrically inherited in a cell-intrinsic manner. These proteins are inherited independent of BMP signaling to the daughter that will retain stem cell identity [40]. This implies that variation in general translation machinery (Wicked and Under-developed) between daughter cells provides yet another mechanism to regulate GSC cell identity and maintenance. The variety and complexity of mechanisms employed by animals to ensure inequality during ACD highlight the importance of this process in the proper creation of form.

Generation of asymmetric division in plants

Cell structure in plants differs in key ways from animals. Specifically, the cell wall is immobilizing and makes plants reliant on precise coordination of divisions for the production of organs. In addition, the majority of biomass in a mature plant is produced de novo post-embryonically resulting in more developmental plasticity in plants compared to animals. It is clear from the sequenced genomes of several plant species that obvious homologs of intrinsic (PAR proteins) and extrinsic (Notch/Delta signaling) do not exist in plants (Arabidopsis genome initiative, [41]). The rigid organization of plant cells favors the use of external cues to establish polarity. The acquisition of cell fate relies on positional information and cellular communication. However, there are examples of novel polarity proteins in plants that appear to mimic the localization behaviors of the PAR proteins, although they likely perceive external signals [42–44]. Here, we review specific examples of asymmetric division and identity determination in plants and discuss the known molecular components involved and how they relate to intrinsic or extrinsic signaling.

Division polarity proteins

Although there are asymmetrically localized proteins in plants, only a few are specifically polarized in response to or in preparation for cellular division [45–47]. Stomatal development has become an excellent model to investigate division polarity proteins, as several novel examples have been identified [43, 44, 48]. Stomata allow the free exchange of gases and water vapor between the plant and the environment. These specialized structures are produced through a series of asymmetric divisions of a dedicated cell lineage in the epidermis [49, 50]. The transcription factors controlling cell identity of the major cell types are generally conserved among diverse plant groups and are not asymmetrically inherited during ACD [51, 52]. In contrast, proteins that display intrinsic or extrinsic polarization during ACD show little conservation, even within the angiosperm lineage.

Stomatal complexes are significantly different between the well-studied dicot and monocot models, Arabidopsis and Zea mays, respectively. A stomatal complex in Z. mays refers to a stoma (two dumbbell shaped guard cells) flanked by a pair of specialized subsidiary cells (SCs). In both monocots and dicots, a guard mother cell (GMC) is produced through an asymmetric division via the action of the bHLH transcription factor SPEECHLESS (SPCH) [51]. In Z. mays, the cells flanking the GMC are called subsidiary mother cells (SMC). SMCs position their nuclei proximal to the GMC and undergo an asymmetric division to produce subsidiary cells adjacent to the GMC. Two LRR-RLK-like proteins called PANGLOSS 1 (PAN1) and PAN2 accumulate in the SMC at the junction between the GMC and SMC [48, 53]. In addition to stomatal development, these two proteins may function in additional contexts to regulate polarization of membrane components during division [54]. Although the specific signal is not known, research supports the idea that the SMC division is induced and polarized by a signal coming from the GMC. PAN1 and PAN2 physically interact with Rho-family GTPases (ROPs) to promote nuclear migration in the SMC toward the GMC, which is dependent on the formation of an F-actin patch. Actin nucleation is achieved through a SCAR/WAVE complex believed to act both upstream and downstream of these components to stimulate localization of PANs and ROPs and promote actin polymerization and nuclear migration [55]. Although components that are physically polarized in the SMC are well studied, determinants of SC identity and how they are regulated remain unknown.

In contrast to Z. mays, Arabidopsis stomatal complexes do not contain dedicated subsidiary cells and none of the several known LRR-RLK receptors involved display polar localization [56, 57]. Instead, Arabidopsis appears to utilize asymmetries in ligand concentrations to provide orientation cues [58–61]. The stomatal lineage starts with an ACD initiated through the action of SPCH, which results in the production of a meristemoid and a stomatal lineage ground cell (SLGC) [62] (Fig. 1c). Many meristemoids divide reiteratively until the expression of the bHLH protein MUTE is triggered causing the transition to GMC [42]. Meristemoid and GMC-derived small peptide ligands of the EPF (EPIDERMAL PATTERNING FACTOR) family provide an extrinsic signal to direct the orientation of new divisions [58–60].

An intrinsic regulator of meristemoid division, BREAKING OF ASYMMETRY IN THE STOMATAL LINEAGE (BASL), exhibits dynamic changes in localization during stomatal lineage ACD [44]. BASL polarity orientation is disrupted when the LRR-LRK receptor TOO MANY MOUTHS (TMM) or EPF1 ligand is absent [44], implying that polarity positioning is modulated by external cues. BASL migrates to the cell cortex distal the division plane during asymmetric division, resulting in SLGC inheritance of BASL after division (Fig. 1c). BASL acts as a scaffold for the Mitogen-activated protein kinase kinase kinase (MAPKKK), YDA, causing increased YDA activity in the SLGC [43]. A second protein, POLAR, also localizes distal to the division plane [42]. However, its function and relation to the YDA MAPK module remains unexplored. The YDA MAPK module is hypothesized to suppress SPCH function ensuring higher SPCH activity in the meristemoid to promote continued division [63] (Fig. 1c). The YDA MAPK pathway is pivotal to a diverse number of developmental programs outside of stomatal development [64, 65]. Currently, there is no evidence to indicate BASL spatially organizes the YDA MAPK signaling in other tissues to regulate stem cell divisions.

Although there are almost no homologous ACD components to compare between plants and animals, Arabidopsis stomatal development provides thematic similarities to classic animal ACD mechanisms. BASL and PAR polarization are both clear and precise indicators of cellular asymmetry prior to cytokinesis. Although the downstream events resulting from their polarization are different, they represent a common theme in “prepping” cells for ACD. Similarly, the animal stem cell niche signals provide external orientation cues to dividing stem cells just as the EPF-family ligands provide similar orientation information to plant cells.

Outstanding questions

The crosstalk between the intrinsic BASL-POLAR polarity module and the YDA MAPK pathway in stomatal asymmetric division is in contrast to the more insulated intrinsic PAR polarity systems in animals. The separation of PAR domains requires phosphorylation-mediated mutual exclusion at the cell cortex [6]. Although BASL polarity formation requires MAPK-mediated phosphorylation [43], the molecular machinery that delivers and maintains BASL polarity is unknown. The identification of new polarity proteins, especially those complementary to the BASL polarity pole, is needed to shed light onto the core genetic network. In the PAN system, the WAVE/SCAR complex polarizes prior to and is required for PAN1/2 polarization. What positional cues from the maize GMC and how they trigger the early polarization event of WAVE/SCAR are important questions in maize stomatal ACD. The identification and characterization of many polarity proteins in plants has been achieved through low-throughput screens. Higher discovery rates of these dynamic proteins may be in our future with the use of automated time-lapse imaging techniques [66]. Additionally, the increased use of single-cell transcriptome analysis in cells that undergo asymmetric division is likely to yield transcripts and potential candidates essential for the process [67]. Currently, there appears to be a diverse number of mechanisms controlling the movement and localization of polarity proteins in plants [68]. Identification of additional members using both high- and low-throughput methods will help to shed light on the commonalities that may exist among them.

Mobile factors and gradients of cell fate determinants

Root apical meristem

The root apical meristem (RAM) contains a group of slowly replicating cells called the quiescent center (QC) that signal surrounding cells to remain undifferentiated. The organization of the QC and stem cells is remarkably similar to that of animal stem cell niches (Fig. 1d). Root stem cells surrounding the QC orient their divisions to produce two daughter cells. One remains in contact with the QC and maintains stem cell function, while the other becomes an initial that will proliferate into the major tissues of the root [69]. The daughter cells produced by RAM stem cells are named based on the tissues they will ultimately generate. Here we specifically describe two types: columella and cortex/endodermis stem cells, which have been studied rigorously with regard to understanding unique strategies for division and differentiation [70–73].

Columella stem cells divide anticlinally to generate self-renewing stem cells and columella (root cap) cells (Fig. 1d). The process of columella stem cell maintenance is controlled by several parallel, but crosstalking pathways [74]. The transcription factor WOX5 is produced in the QC and functions as a traveling signal to maintain stem cell identity in adjacent cells [75] (Fig. 1d). Antagonistically, the small peptide ligand, CLE40, is transcribed in differentiated columella cells and secreted into the apoplast, which provides a decreasing ligand gradient toward the QC. CLE40 acts through the leucine-rich repeat (LRR) receptor CLAVATA1 and the non-LRR receptor ARABIDOPSIS CRINKLY4 (ACR4) to repress WOX5 expression and promote differentiation [76] (Fig. 1d). Recent work suggests that a plant U-box (PUB) protein, PUB4, functions downstream of CLE signaling likely through activation of CYCD6;1 to promote cell proliferation and the timing of asymmetric division in the RAM [77]. Additionally, two AP2 class transcription factors PLETHORA1 (PLT1) and PLT2 display a graded distribution with maxima in the QC [78, 79]. PLT function is dose dependent, where high levels of PLTs promote stem cell fate [79].

The two NAC domain transcription factors, FEZ (“little cap”) and SOMBRERO (SMB, “big cap”) directly regulate the columella stem cell division [71] and represent an excellent example of oscillating expression during asymmetric division (Fig. 2a). Specifically, FEZ functions to promote columella stem cell division and is expressed in columella stem cells, but disappears from apical daughter cells after ACD. FEZ expression reappears in the stem cell prior to ACD initiation forming an oscillating pattern of expression. In contrast, SMB represses division and promotes columella cell differentiation. The two genes form a feed back loop where FEZ drives division in stem cells and activates SMB transcription in daughter cells. Once activated, SMB downregulates FEZ, which in turn inhibits division and promotes the differentiation [71] (Fig. 2a). SMB functions together with its close homologs, BEARSKIN1 (BRN1) and BRN2, to promote root cap maturation [70]. The purpose of the oscillating expression pattern of FEZ is not clear; however, technological improvements in time-lapse imaging in recent years are likely to provide additional examples of cyclic expression and the role it plays in developmental processes.

Fig. 2.

Mobile factors control division and cell fate decisions. a Columella stem cell division. WOX5 moves from the QC (quiescent center) to the columella SC (stem cell). FEZ expression dynamically oscillates during rounds of asymmetric division. FEZ activates SMB in differentiating columella to produce a feedback loop to ultimately inhibit the expression of FEZ in the CC (columella cell). b Division of endodermis/cortex initials. Cell types are color coded as indicated. The diffusible transcription factor SHR moves from the stele to the endodermis where it is sequestered by SCR. SHR movement into the cortex is prevented and a positive feedback loop promotes endodermal cell fate. c Hypophysis division. MP activates TMO7 in provasculature (orange), which moves to the hypophysis (dark green). MP activation results in basally directed auxin flow through the up-regulation of PIN1. Both auxin response and TMO7 activity is necessary to specify hypophysis identity and initiate division

Cortex/endodermal stem cells divide to produce an initial (cortex/endodermis initial, CEI) that divides asymmetrically to form clonally related cortex and endodermis cells (Fig. 2b). This asymmetric division is regulated by the activities of the plant-specific GRAS family transcription factors, SHORT-ROOT (SHR) and SCARECROW (SCR) [80–83]. Instead of two single layers of endodermis and cortex, a layer of ground tissue with mixed endodermis/cortex identity is produced in strong loss-of-function mutations in either gene [80, 81]. CEI division and fate determination of endodermis and cortex cells relies on cell non-autonomous movement of SHR protein from the stele into adjacent cells where it activates SCR transcription. A reduction in SHR is observed prior to endodermis division, suggesting that SHR works in a dosage-dependent manner to regulate division in the endodermis [84]. In endodermal cells, SCR sequesters SHR in the nucleus to ensure one cell layer movement of SHR [73]. This sequestration involves a second protein, the zinc-finger JACKDAW (JKD), which also limits SHR movement from the endodermis into the cortex. A loss of JKD function leads to movement of SHR outside the endodermis and ectopic cell divisions in the ground tissue and an increased number of cell layers in the root [85]. In the CEI, SHR and SCR directly activate specific cell cycle genes, e.g., CYCD6;1, to promote cell division and layer patterning in the root [86]. Therefore, the response to SHR/SCR function in the endodermis is to initiate differentiation, whereas SHR/SCR response in the CEI results in division. Specific mechanisms that drive different responses in each cell types remain to be determined.

Hypophysis

The root meristem in Arabidopsis is initiated in the globular-stage embryo by the asymmetric division of the hypophysis, the uppermost cell of the suspensor. The hypophysis undergoes a stereotypical asymmetric division to produce a single lens-shaped cell, the precursors of the quiescent center (QC) and the lower-tier stem cells of the root meristem. Two proteins work antagonistically to control this process. The BODENLOS (BDL)/IAA12 protein negatively regulates the auxin responsive transcription factor, MONOPTEROS (MP, also known as AUXIN RESPONSIVE FACTOR 5, ARF5) (Fig. 2c). Auxin levels in the provasculature result in the degradation of BDL and the subsequent action of MP. MP positively regulates the auxin efflux carrier, PIN1, and several other downstream targets including the bHLH transcription factor, TARGET OF MONOPTEROS7 (TMO7) [87]. Loss-of-function TMO7 mutants result in aberrant hypophyseal divisions producing some rootless seedlings similar to an mp mutant [87]. The current model suggests that MP/ARF5 causes an increase in basally directed auxin movement into the hypophysis and directly activates TMO7 in the proembryo. TMO7 is a mobile transcription factor that mediates asymmetric division in the hypophysis in concert with auxin-related events to generate the primary root.

Additional downstream targets of MP/ARF5 include a group of zinc-finger transcription factors, including NO TRANSMITTING TRACT (NTT) and two closely related genes WIP DOMAIN PROTEIN 4 (WIP4) and WIP5 [88]. These genes are completely redundant, but triple mutants have no roots as a result of incorrect or loss of hypophyseal division, similar to mp mutants. MP/ARF5 binds to conserved AuxRE sites in conserved introns of NTT, WIP4 and WIP5 to promote gene expression [88]. All of the mentioned proteins play an essential role in mediating an auxin signal to direct hypophyseal division (Fig. 2c). The process of embryonic root initiation is a clear example of how the unique response to hormone signaling is dependent on the types of transcription factors being expressed.

Outstanding questions

Mobile transcription factors are repeatedly used in the regulation of ACD in plants, but are certainly not restricted to this function. The use of non-cell-autonomous transcription factors are prominently used in other aspects of development including shoot growth, flower induction, floral organ development and epidermal patterning [89]. However, we still know very little about how the directionality of their movement is regulated. Movement through plasmodesmata, endosomes, and sieve elements are all recognized pathways for non-cell-autonomous short- and long-range movement of transcription factors in many different plant systems (reviewed in [90]). How this movement is restricted and how mobile transcription factors integrate signaling is less understood and is key to understanding the complexities of multicellular development. Based on our current knowledge, extrinsic and intrinsic factors likely participate in altering protein stability or restricting transcription factor movement. Reporter gene expression and protein localization studies will surely identify additional proteins that function non-cell-autonomously in the next decade. Future dedicated studies focused on identifying proteins that bind these mobile transcription factors to restrict or alter their function may provide some commonalities among this very diverse group of proteins.

Signaling and expression asymmetries during ACD

Similar to animals, plant embryogenesis has been an important model for examining mechanisms of ACD [91, 92]. The predicable cell division and gene expression patterns demonstrated during embryogenesis have revealed several components involved in polarity and cell identity [93]. After fertilization, the zygote divides asymmetrically to give rise to a smaller apical cell and a larger basal cell that initiate different developmental programs. The small apical cell produces the major components of the embryo proper (proembryo) and the larger basal cell develops into an extra embryonic anchor called the suspensor. Although many details remain unknown, extrinsic activation of signaling pathways to promote or inhibit gene expression is an important component of ACD during plant development [94].

YDA signaling is required for the first ACD of the zygote and establishment of basal cell identity. In a yda mutant, the first division is more symmetrical and the suspensor often takes on embryonic character [95]. Consistently, mutations in the downstream signaling components, MPK3/MPK6, result in similar defects in the embryonic division [96]. Activation of YDA-dependent signaling in the zygote is triggered by sperm-provided Pelle-like kinase SHORT SUSPENSOR (SSP) transcripts (Fig. 3a) [97]. Loss of SSP function results in defects in establishing the apical and basal lineages, similar to yda mutants [97]. Additional activators of YDA signaling are the EMBRYO SURROUNDING FACTOR 1 (ESF1) family of cysteine-rich peptides [98]. These peptides are produced by the central cell of the female gamete, supporting the idea that external cues instruct zygote development. The specific receptors that perceive ESF1 and how their signals are delivered to the YDA pathway remain unknown. Downstream of YDA is the nuclear protein GROUNDED (GRD), which works cooperatively with Wuschel-Related Homeobox (WOX) proteins to establish apical–basal polarity in the zygote [99, 100] (see below).

Fig. 3.

Auxin and cell wall components in asymmetric cell division. a Zygotic asymmetric division. WRKY2 promotes zygotic elongation and activation of WOX8/9 that is required for WOX2 expression in the apical domain. Increased level of auxin in the apical cell is mediated by PIN7 activity. In tobacco zygotic division, AGPs are enriched in the apical cell. In Arabidopsis, the YDA MPK3/6 cascade transmits the upstream SSP and ESF1 signaling to the downstream transcription factor GRD to promote zygotic elongation and asymmetric division. GRD is not a direct substrate of MPK3/6 (dashed arrow) and may function with an unknown partner (question mark) downstream of the YDA MPK3/6 cascade. b SPCH initiates stomatal asymmetric division of the meristemoid mother cell (MMC) to produce a small daughter cell meristemoid (M) and a large daughter cell, stomatal lineage ground cell (SLGC). After a few rounds of M asymmetric division (arrow), MUTE expression turns on guard cell (GC) differentiation. High auxin level is associated with stomtal asymmetric division and auxin depletion is mediated by PIN3 in Ms and is associated with GC differentiation. The negative regulation of auxin on stomatal production is achieved by TIR1/AFB-mediated suppression of a positive regulator Stomagen

After division of the zygote, clear separation of mRNA expression domains of a group of WOX transcription factors (WOX2/8/9) is observed [101–103]. These genes are critical for determining the different cell fates of the apical and basal lineages, but it is unclear if expression asymmetry is established in the zygote or occurs very rapidly after division [104]. WOX2 is expressed specifically in the apical daughter cell, whereas WOX8/9 are expressed in the basal cell (Fig. 3a) [101, 103]. Expression of these transcription factors defines distinct transcriptional domains along the basal–apical embryo axis. The WRKY2 transcription factor plays a role in zygotic polarization by directly binding and activating WOX8/9 [102]. WRKY2 or WOX8/9 mutants results in loss of WOX2 expression in the apical lineage, indicating that WOX8/9 function in the basal lineage is required for WOX2 expression in the apical lineage and that signaling between the two cells is critical for embryo development. In addition to the importance of the WOX transcriptional domains in determination of apical and basal lineages, the fates of these cells types are also dependent on auxin signaling (see below).

Auxin and cell wall components

Auxin

Hormones play important roles in controlling stem cell specification, maintenance and differentiation in animals and plants [105]. Among phytohormones, auxin plays a role in a broad variety of biological processes [106–108]. Recent reviews have provided in-depth discussion on genetic circuitries and hormonal crosstalk in setting up stem cell niche, regulating stem cell activity and homeostasis in plants [109–113]. Here we focus on a few examples demonstrating that unequal auxin distribution helps to differentiate cell types during development.

Directional auxin flow is controlled by a variety of transporters, including the PIN efflux carriers [114], the influx carriers AUXIN RESISTANT1/LIKE AUX1 (AUX1/LAX) [115] and the P-GLYCOPROTEIN (PGP/ABCB) transporters [116]. Immediately after the first asymmetric division of the zygote, PIN7 is localized to the apical side of the basal cell resulting in directional auxin flow into the apical cell [117]. It was found that both the polar distribution of efflux carriers and the auxin responsive factors are critical for specifying apical and basal lineage fates (Fig. 3a) [118, 119]. In this process, potential connections between the YDA MAPK signaling cascade (discussed earlier) and elevated expression of auxin biosynthetic enzymes and auxin levels were revealed by proteomic analyses of yda loss and gain-of-function mutants [65]. Because endogenous auxin levels are dramatically elevated in plants expressing constitutively active YDA [65], it is appealing to hypothesize that YDA controls zygote elongation by promoting auxin signaling. Further evidence of the importance of auxin in zygotic divisions comes from mutations in the ARF guanine-nucleotide exchange factor (ARF/GEF) GNOM [120]. GNOM is required for the endosomal recycling of the auxin efflux carrier PIN1 and loss of GNOM produces symmetric zygotic division [121]. The loss of specific localization of PIN1 and directional auxin movement results in impaired asymmetric division in early embryogenesis [122, 123].

PIN1-mediated auxin transport and auxin responsive genes are required for the specification and division of the hypophysis [87, 120, 124, 125] (Fig. 2b). Additional regulation of this process involves the protein phosphatases PP2C, POLTERGEIST (POL) and POLTERGEIST LIKE 1(PLL1). Mutations in these proteins cause failures in hypophysis division due to reduced expression of PIN1 protein and WOX5 expression in the lens-shaped daughter cell of the hypophysis [126]. Biochemical data suggests that ACR4 kinase and POL and PLL1 might function antagonistically to alternate the phosphorylation status of WOX5 to control its expression [127, 128]. However, the direct regulation between POL/PLL1 and WOX5 remains to be established.

Auxin also modulates cell fate during stomatal asymmetric division [129]. In vivo live imaging of the auxin input marker (DII-VENUS) and output marker (DR5:VENUS) in developing leaves indicates that auxin signaling is depleted from meristemoids before differentiation into GMCs (Fig. 3b). Consistently, meristemoids have higher PIN3-GFP fluorescence compared to the SLGC after division. Therefore, a decrease in auxin concentration may trigger the meristemoid to transit from asymmetric to symmetric cell division pattern [130]. Overproduction of stomata and clustered stomatal patterning are observed in triple and quadruple mutants of pin1, 2, 3, 4, 7, and auxin receptor mutants of tir1 afb1, 2, 3 [130]. Auxin stabilizes the interaction of the auxin co-receptors, TIR1/AFB and AUX/IAA proteins, which activate MP/ARF5 transcription. MP/ARF5 directly binds to the STOMAGEN promoter to suppress its expression [131]. STOMAGEN functions antagonistically with two negative stomatal regulators from the same family, EPF1 and EPF2 [61], to regulate the number of stomata (Fig. 3b). It is known that the bHLH protein MUTE triggers meristemoid differentiation into the GMC, but how auxin depletion is linked to MUTE expression requires further investigation.

Outstanding questions

The complexity of the regulators of auxin biosynthesis, transport and activity makes it challenging to provide a unified and mechanistic view on how auxin regulates ACD. Auxin plasma membrane receptors are arguable and whether auxin locally affects cell wall elasticity and cell growth in the process of cell polarization remains a long-standing question. Currently, the lack of cytoplasmic and extracellular sensors that closely monitor auxin levels is a significant obstacle to resolving these questions. After ACD, the mechanism that triggers PIN activities to distribute unequal auxin levels to daughter cells and how ARFs crosstalk with other cell fate regulators to control cell identity asymmetry are major unresolved questions. Based on examples in lateral root formation, multiple ARF/IAA-ARF modules may cooperatively regulate the steps during asymmetric division [132]. Detailed phenotypic characterization of loss- and gain-of-function mutants, in combination with high-resolution 4-D imaging and computational analysis [133], may provide insight into this process. Improved auxin sensors DR5v2 and R2D2 hold the promise to disclose more cell divisions that produce asymmetric auxin signaling in progenies in different developmental contexts [134].

Cell wall components

Cell walls were originally thought to play a mostly structural role in plant development, but detailed functional research has revealed a more complex role of the cell wall during development [135–137]. Experimental approaches and computational modeling supports the role of cell wall components, including proteins and polysaccharides, in instructing cell polarity and influencing asymmetric cell division in development.

Arabinogalactan proteins (AGPs) belong to one of the most complex and highly diverse cell wall proteoglycans [138]. AGPs polarize at the apical pole of tobacco zygotes and disrupting AGP activity with chemical inhibitors alters zygotic ACD (Fig. 3a) [139]. AGP function appears to be deeply conserved. In brown algae, AGP-like proteins were found in cell wall extracts and biochemical disturbance of their function led to impaired embryogenesis [140]. More genetic evidence from Arabidopsis supports the importance of the sugar side chains of AGPs [141, 142]. A glycosyltransferase of CAZY family GT31 (AtGALT31A) is expressed in the suspensor and can galactosylate AGP side chains. Mutations in AtGALT31A cause embryo arrest at the globular stage and observable defects in the hypophyseal ACD during embryo development [141].

During stomatal development, asymmetric cell division requires the polarized BASL activity [44]. Induced regional cell expansion due to over accumulation of polarized BASL supports a role for BASL or the associated YDA MAPK module in modifying the cell wall and promoting cell growth [43, 44]. MAPKs can regulate the microtubule cytoskeleton via their binding proteins [143] to modify cell shape and division orientation [144]. Computational modeling suggests that postmitotic polarity-switching may occur through substances in the newly formed cell wall, which orients BASL polarity in the newly born meristemoid [145]. A group of glycosylphosphatidylinisitol (GPI)-anchored proteins called EARLY NODULIN-LIKE PROTEIN (ENODL) show enriched localization at the newly formed cell walls. Triple mutants of enodl13, 14, 15 exhibit defects in stomatal divisional patterning, mimicking the phenotype of a basl mutant [146]. Whether BASL localization or function is regulated by the ENODL proteins is an interesting question for our future studies.

Epigenetic regulation in cell fate determination in plants

Studies of animal and human embryonic stem cells revealed that stem cell genomes have more open chromatin configuration and dynamic association with chromatin proteins [147, 148], while differentiation is accompanied by progressive deposition of repressive histone marks to compact chromatin [149]. In animals, the expression of pluripotency factors (Oct4, Nanog, and Sox2) reorganizes chromatin long-distance connectivity to induce stemness in differentiated cells [150]. Here, we review specific examples of the essential role of chromatin remodeling in reprogramming transcription to regulate stemness and differentiation in plants.

Histone modification

Histone H3 methylation (H3K4me3) is generally associated with gene activation and is important for RAM niche maintenance. Arabidopsis SET DOMAIN GROUP 2 (SDG2) belongs to the evolutionary conserved H3K4-methyltransferases (Trithorax group, TrxG) proteins and contributes to genome-wide H3K4me3 deposition [151]. Consistently, loss of SDG2 resulted in reduced H3K4me3 modification in the root stem cell niche [152]. The WOX5 transcription factor travels from the QC into the columella stem cells where it maintains stem cell identity (Fig. 2a). The WOX5 promoter region is demarcated with H3K4me3, likely mediated by SDG2, to promote WOX5 expression (Fig. 4a). Once expressed, WOX5 recruits the transcription repressor TOPLESS (TPL) and a histone deacetylase HDA19. These factors together repress the expression of a differentiating factor CYCLING DOF FACTOR 4 (CDF4) in stem cells [153]. A PHD domain protein REPRESSOR OF WUSCHEL1 (ROW1) specifically recognizes and binds to H3K4me3 in the WOX5 promoter region to repress its expression outside the QC [154].

Fig. 4.

Epigenetic regulations in stem cell asymmetric cell division in Arabidopsis. a Histone modification events in the root stem cell niche. The SWI/SNF remodeling ATPase BRM directly interacts with the chromatin regions of PIN1, 2, 3, 4, 7 to promote their expression by suppressing the polycomb group (PcG) protein binding, which deposits the repressing mark H3K27me3. The expression of WOX5 is positively regulated by SDG2, a TrxGH3K4-methyltransferases, that marks the chromatin region with activating H3K4me3. b Active DNA methylation and histone modification in stomatal fate differentiation. The expression of EPF2 peptide is suppressed by the RdDM pathway-induced DNA methylation in the promoter region of EPF2, which is actively erased by the ROS1 DNA glycosylase. The terminal differentiation of guard cells is secured by the FAMA-RBR complex that suppresses the expression of stomatal early genes (SPCH and others). RBR can recruit the PcG proteins that deposit H3K27me3 repressive marks to the FAMA-binding regions

Directional auxin flow is also controlled by chromatin modifiers through the epigenetic regulation of PIN proteins. BRAHMA (BRM), a SWI/SNF chromatin remodeling ATPase, is expressed in the root apical meristem [155] (Fig. 4a). Loss-of-function brm mutants produce short roots with defective QC formation [156]. Chromatin-immunoprecipitation determined that BRM directly binds to the chromatin of PIN1, 2, 3, 4, 7 genes and disrupts auxin distribution [156]. BRM is hypothesized to antagonize the functions of Polycomb Group (PcG) proteins, which establish and maintain H3K27me3 repressive mark in plants [157]. Indeed, in a brm mutant, increased levels of H3K27me3 were found in PIN1 and PIN2 chromatin regions [156].

RETINOBLASTOMA-RELATED (RBR) is a plant homolog of the tumor suppressor Retinoblastoma (pRb), a key cell cycle regulator that suppresses G1-S transition [158]. In animals, pRb associates with various chromatin-modifying proteins, including the PcG repressing complex [159, 160], to regulate cell proliferation and differentiation. Similarly, the interaction between RBR1 and the PcG complex (POLYCOMB REPRESSIVE COMPLEX 1, PRC1) and PRC2, has been established in plants [161, 162]. The role of RBR in cellular differentiation is exemplified through its binding to the LxCxE domain of FAMA, the bHLH transcription factor that promotes stomatal terminal differentiation [163, 164]. Plants expressing a modified version of FAMA produces a “stomata-in-stomata” (SIS) phenotype that results from the reinitiation of SPCH and MUTE expression in mature stomata. As SPCH and MUTE are normally suppressed at this stage by H3K27me3 modification by the function of PRC2, overexpression of a PRC2 subunit CURLY LEAF (CLF) rescued the SIS phenotype [165]. Thus, the function of FAMA-RBR has been proposed to recruit PRC2 to suppress the early genes in stomatal development and to ensure stomatal terminal differentiation (Fig. 4b).

DNA methylation

Arabidopsis stomatal development has revealed a few key genes that are regulated by DNA methylation and transcriptional silencing. Plants grown under low humidity conditions show reduced stomatal production, as a result of elevated DNA methylation on the SPCH locus [166]. REPRESSOR OF SILENCING 1 (ROS1)/DEMETER-LIKE 1 encodes a bifunctional 5-methylcytosine DNA glycosylase/lyase and is critical for active DNA demethylation in most tissues of Arabidopsis [167, 168] (Fig. 4b). In ros1 mutants, the promoter region of the EPF2 negative regulator is hypermethylated via the RNA-dependent DNA methylation (RdDM) pathway to suppress its expression [169] (Fig. 4b). Loss of ROS1 and related genes produced more stomatal lineage cells, phenocopying that of an epf2 mutant [169].

Outstanding questions

Our knowledge about the epigenetic regulation of stem cell division and differentiation is fragmented. The global organization of chromatin and nuclear architecture related to stemness of plant cells is not known. However, the discovery of transcription factors, e.g. WOX5 and AP2 [153, 170], recruiting histone modifiers in the regulation of cell identity provides a promising direction. Whether key transcription factors commonly use epigenetic factors to regulate developmental processes in plants has not been widely studied. Epigenetic modifications to regulate PIN and EPF2 expression might reflect the plasticity of plant development in adapting to environmental changes. It is possible that the asymmetric segregation of histone modification demonstrated in Drosophila GSCs may play a role in the regulation of plant cell asymmetric division. The phosphorylation of H3T3 and the Haspin kinase are conserved in plants and mutations in Haspin kinase causes pleiotropic defects in Arabidopsis development, including alterations in the first embryonic cell division [171]. Elucidating whether asymmetric segregation of H3T3 phosphorylation is commonly employed in both animals and plants to differentiate daughter cell fates is an important future endeavor.

Cell division: cytokinesis

Animal cells use cytoskeletal components and associated proteins to constrict the cell membrane to produce new daughter cells after mitosis. In contrast, plant cells rely on two specialized cytoskeletal structures, the pre-prophase band (PPB) and the phragmoplast, to identify the division plane and to create a new cell wall between the daughter nuclei of a dividing plant cell. The location of the cell plate is determined prior to mitosis in late interphase. Just prior to mitosis, the PPB composed of a cortical array of filaments and associated proteins is assembled at the location of the future division plane. In late anaphase, the microtubule array called the phragmoplast is produced and oriented perpendicular to the division plane. During cytokinesis, the phragmoplast guides vesicles containing lipids, proteins and carbohydrates to the midzone of the phragmoplast where they fuse to produce the cell plate at the position previously occupied by the PPB.

How the division plane is orientated is a fundamental question in studying plant organogenesis. Arabidopsis roots, stomatal lineage cells and embryos are major model systems in the field. Recent reviews discussed core regulators and pathways that direct division plane control in plants [172–177]. Here we briefly highlight a general rule for plant cell division and consider how cell polarity might participate in determining the division plane.

Geometry has been shown to be determining factor in the placement of the new cell wall during symmetric divisions, which leads to halving the volume of the mother cell using the shortest path as predicted by Errera’s rule [178, 179]. However, during asymmetric division, polarity cues are thought to override the geometry-based default mechanism. For instance, hormone or polarity protein distribution may directly impact the location of the PPB to result in asymmetric placement of the new cell wall during division. Support for this comes from the clear disruption of ACD when polarity proteins such as PAN1 (maize) and BASL (Arabidopsis) are mis-localized or absent. Recent work has established that the PPB-localized kinesin ARK3/KINUa and the polarity protein, BASL, are regulated by SPCH [180]. The downregulation of ARK3/KINUa disrupts ACD (basl-like phenotype) suggesting a potential direct link between polarity proteins and establishment of the PPB. Future research should focus on clarifying how BASL and ARK3 interact to direct PPB positioning in asymmetric cell division. This will likely rely on the identification of additional components in the network surrounding SPCH.

Concluding remarks and future perspectives

The perception of external positional cues for polarizing events during division and differentiation of cell types is an emerging theme in the field of plant ACD. However, our understanding of any complete pathway from perception to cytokinesis is fragmentary and similarities among different plant cell types are lacking. It is likely that the commonalities between plant and animal ACD will not be based on homologous components, but on common premises. The fundamental differences between plant and animal cellular structure, signaling molecules, and division patterns highlight the dissimilarities between the two systems. However, the noteworthy use of a non-cell-autonomous regulation system incorporating mobile transcription factors provides a unique platform from which to specifically study plant ACD. SHR action and movement in the Arabidopsis root has been most extensively studied and identifying components of the circuit involved in SHR degradation and movement may provide broad insight into this plant-specific mechanism of ACD regulation and provide a future means of regulating this process for applied agricultural uses.