The miR156 juvenility factor and PLETHORA 2 form a regulatory network and influence timing of meristem growth and lateral root emergence

In Arabidopsis thaliana, progression from the juvenile to the adult state is regulated by a network of microRNA genes, including those that code for miR156, which is expressed at high levels in juvenile plants, and miR172, which increases in level as plants mature (Wu et al., 2009). The change in miR156 expression that occurs with age involves an increase in repressive H3K27me3 histone marks associated with miR156A and miR156C (Xu et al., 2016b, 2018a,b). Mutants that are impaired in their ability to place these marks show delayed phase transitions (Xu et al., 2016b).

Expression of miR156 affects plant development by targeting mRNAs for destruction. Specifically, miR156 transcripts target genes in the SQUAMOSA PROMOTER BINDING PROTEIN-LIKE (SPL) family, decreasing SPL expression (Yu et al., 2012; Xu et al., 2016a). Among the genes induced by SPLs are the miR172 family of microRNA genes. Thus, reduced expression of miR156 leads to an increase in miR172 gene expression, which promotes adult characteristics in the shoot (Wu et al., 2009). Although most well understood in A. thaliana, this network of small RNAs appears to control phase transitions in many species (Wang et al., 2011; Poethig, 2013).

Root development is also affected by the miR156 family of genes. The miR156 family comprises genes that code for nearly identical mature miRNAs, with individual genes being differently expressed; several miR156 genes are expressed in the root apical meristem and in lateral root meristems (Yu et al., 2015; Barrera-Rojas et al., 2020; Cheng et al., 2021). However, overexpression of miR156 reduces the number of cortical cells in the meristem (Barrera-Rojas et al., 2020). Supporting the observation that miR156 can reduce meristem size are data showing that levels of miR156 are low in areas of active cell division (Cheng et al., 2021). Additionally, overexpression of miR156 affects root development by accelerating the emergence of initiated root primordia (Yu et al., 2015). Downstream of miR156, changes in expression of SPL10, a target of miR156, have also been shown to affect growth of the primary root (Gao et al., 2018; Barrera-Rojas et al., 2020). These effects of miR156 expression share the common thread of affecting the meristem and growth zone of primary or lateral roots.

Another gene family that profoundly affects meristem development is the PLETHORA (PLT) family of transcription factors. Root meristems form at many points during development; the primary root meristem forms in the embryo and other meristems arise after germination. When meristems develop from callus or during lateral root formation, three PLETHORA family transcription factors, PLT3, PLT5 and PLT7 are expressed early in the process. These PLTs are required for the subsequent expression of PLT1 and PLT2 (Kareem et al., 2015; Hofhuis et al., 2013; Du and Scheres, 2017). Expression of PLT1 and PLT2 activates genes required for the progression of meristem formation, and also allows the stem cells and meristem to persist (Du and Scheres, 2018; Aida et al., 2004; Galinha et al., 2007). In established primary root meristems, PLT proteins contribute to a protein complex that induces WOX5, which, in turn, maintains neighboring cells as stem cells (Shimotohno et al., 2018; Sarkar et al., 2007; Pi et al., 2015).

As roots age, both the meristem and elongation zone enlarge (Beemster and Baskin, 1998), an increase that could be related to changes in PLT gene expression. High levels of PLT expression in the root tip are present in the stem cells and these levels progressively decrease toward the shootward end of the elongation zone (Galinha et al., 2007). High levels of PLT2 increase the length of the meristem and repress elongation of individual cells (Mähönen et al., 2014). Shootward of the stem cell niche, the level of PLT2 reaches a threshold at the point where cell division slows and expression of some markers of differentiation begins (Salvi et al., 2020). A negative regulatory loop between PLT2 and the cytokinin response factor ARR12 participates in setting the location of this boundary between the meristem and the transition zone (Salvi et al., 2020). Growth of the meristem is also restricted by the cytokinin response factor ARR1 (Moubayidin et al., 2010, 2016).

Here, we consider the possibility that the enlargement of the growth zone that occurs during the first few days after germination may be driven by increases in PLT2 expression. We examine the extent to which this and other changes in root developmental traits are coordinated by miR156 and discover that miR156 regulation of PLT2 is partially responsible for some of the phenotypes associated with high levels of miR156. The data also uncover a regulatory network through which miR156C and PLT2 repress one another.

Overexpression of miR156 has highly pleiotropic effects on shoot growth, including more rapid production of rosette leaves (Wu and Poethig, 2006; Wu et al., 2009). To determine the extent to which overexpression of miR156A specifically affects root development, 35S::miR156A plants were reciprocally grafted to wild type and grown to maturity in soil-filled rhizotrons. Plants in which both roots and shoots were genotypically 35S::miR156A were substantially larger than wild type; the shoots had more leaves and the roots were more densely packed (Fig. 1A,B). Grafting wild-type rootstock onto a 35S::miR156A shoot resulted in an intermediate root system phenotype; the root system was less dense than when 35S::miR156A was self-grafted, but somewhat more dense than when wild type was self-grafted (Fig. 1C). This indicates that the shoot system plays some role in establishing the density of the root system. To determine whether the increased root system resulted entirely from a larger shoot, 35S::miR156A rootstock was grafted onto wild-type shoots. Once again, the root systems took on an intermediate phenotype (Fig. 1C). 35S::miR156A roots with wild-type shoots were less dense than those of 35S:miR156A self-grafts but denser than those produced by wild-type self-grafts, indicating that the level of miR156A expression in the root affects root development.

To examine the pattern of miR156 expression in the days immediately following germination, we used previously described lines expressing pmiR156A::GUS and pmiR156C::GUS reporter constructs (Barrera-Rojas et al., 2020). For the time-course experiments described here and throughout this paper, seeds were examined at 12 h intervals after plating to determine the time of germination; all ages are reported as time since germination.

The level of miR156A and miR156C expressed in the root tip declined from day 1, when roots were almost entirely stained, to day 9 (Fig. 2A,B). In the root cap, expression of both miR156A and miR156C decreased until staining was no longer visible. Staining in the growth zone (meristem plus elongation zone) also decreased, but remained visible. For miR156A, expression shootward of the growth zone became patchy before disappearing (Fig. 2C,D), whereas for miR156C, staining shootward of the growth zone persisted, particularly in the central vasculature (Fig. 2E-H). Expression of miR156A and miR156C during lateral root formation was variable. However, miR156A staining was generally not visible in outer layers of the primordia from the time primordia contained three layers of cells through emergence (Fig. 2D). For miR156C, staining was visible in stage 1 and stage 2 lateral root primordia (Fig. 2E,F) and not visible in recently emerged lateral roots (Fig. 2G); intermediate stages showed variable levels. Longer emerged lateral roots showed expression of miR156C behind the tip (Fig. 2H).

We followed the expression of PLT3, a gene that is expressed early in lateral root development, by using a 4.5 kb PLT3::GUS transcriptional reporter. On day 1, β-glucuronidase (GUS) staining in the primary root was strong throughout the meristem and root cap (Fig. 3A, top row). Expression of PLT3::GUS shootward of the quiescent center declined rapidly as the roots matured. To determine whether this decline in PLT3 expression is driven by increasing maturity of the root, the PLT3::GUS reporter was crossed to 35S::miR156A (Wu and Poethig, 2006). Overexpression of miR156A did not change the expression of PLT3::GUS detected in the root tip during this period (Fig. 3A,B). Consistent with these results, real-time PCR levels of PLT3 in the miR156A-overexpressing line on day 2 were observed to be 1.3±0.2 (mean±s.d.) times those of wild type.

Expression of a typically late-expressed PLT gene (PLT2) was observed in root tips of a wild-type line carrying a PLT2-Venus fluorescent fusion protein driven by the native promoter (Fig. 3C, top row). On day 1, PLT2-Venus was expressed mainly in the root cap and the stem cell niche. On day 2, expression expanded shootward. The intensity of expression also increased, reaching a peak around day 4 (Fig. 3C,D). The level of PLT2-Venus then fell from around day 4 to day 7.

We crossed the PLT2-Venus reporter into the 35S::miR156A line. In this line, the level of PLT2-Venus in 35S::miR156A was modestly lower than that of wild type at day 1, and the rapid increase that occurred in wild type was delayed, resulting in substantially lower levels around days 3 and 4 compared with wild type (Fig. 3C,D). Similar to wild type, levels of PLT2-Venus in 35S::miR156A fell between about 4 and 7 days (Fig. 3D). At day 4, the level of PLT2-Venus fluorescence in 35S::miR156A plants was still rising whereas the level in wild type was not. Thus, 35S::miR156A reduced the initial level of PLT2-Venus expression and delayed and reduced its subsequent increase.

The spatiotemporal pattern in which signal from pmiR156C::GUS declined (Fig. 2A) raised the question of whether PLT2 might modulate expression of this small RNA. In the wild-type background on day 4, high levels of pmiR156C::GUS staining were observed just shootward of the quiescent center (Fig. 4B,C). Expression of miR156C then declined somewhat through the meristem, increasing again near the start of the elongation zone. Rootward of the quiescent center, expression of miR156C also declined before increasing again in the outer layers of columella cells (Fig. 4C). The regions near the quiescent center that had lower levels of staining were also regions in which PLT2 was expressed (Fig. 3C).

To check whether miR156C is affected by PLT2 expression, we crossed pmiR156C::GUS to the plt2-3 CRISPR allele and observed GUS staining in F3 seedlings on day 4 (Fig. 4A,C). Throughout the measured region, GUS stained plt2-3 plants somewhat more strongly than wild-type lines, although the difference was larger in the meristem and in the region just shootward of the meristem. The increased staining in pmiR156C::GUS roots in the plt2-3 background indicates that PLT2 represses expression of miR156C in the root tip, at least on day 4 when PLT2 levels in wild-type roots are high.

Because 35S::miR156A plants had reduced PLT2-Venus expression (Fig. 3), we hypothesized that increasing the dosage of PLT2 by crossing in PLT2-Venus might limit the effect of 35S::miR156A on root architecture. To assess root architecture, we grew plants to maturity in rhizotrons (in this case without grafting). As before, 35S::miR156A plants had substantially denser root systems than did wild-type plants (Fig. 5). Wild-type plants expressing PLT2-Venus had root systems that appeared to be a bit shorter than those of wild-type plants without the added copy of PLT2, but these shallower root systems did not appear to be substantively different from wild type in terms of their root density (Fig. 5B). In contrast, 35S::miR156A plants carrying PLT2-Venus had root systems that were noticeably less dense than those lacking the added copy of PLT2. Thus, the line carrying PLT2-Venus partially reversed the miR156 overexpression phenotype. Crossing in PLT2-Venus also affected the shoot system, with flowering occurring earlier than for 35S::miR156A alone. Conversely, introducing the plt2-3 mutant into 35S::miR156A slightly increased the average root density in this experiment (Fig. 5B), which is consistent with lower levels of PLT2 expression conferring some of the 35S::miR156A overexpression phenotype. Together, these data suggest that a substantial portion of the increased density of 35S::miR156A root systems is attributed to PLT2.

To examine the effect of miR156 on development of the root system in more detail, we analyzed growth kinematically, obtaining parameters of interest from the velocity profile (Baskin and Zelinsky, 2019; see Materials and Methods). The length of the meristem was of particular interest because overexpression of miR156 is associated with lower levels of PLT2, and in several plt mutants, including plt1plt2 double mutants, the meristem becomes shorter and can be lost as the plants mature (Galinha et al., 2007).

In wild-type plants, the length of the meristem, the length of the growth zone and overall root growth increased substantially from day 2 to day 6 (Fig. 6). In miR156A-overexpressing plants, the lengths of the meristem and growth zone were initially longer than wild type (Fig. 6C,D). The rate of cell proliferation, measured in a separate experiment, was generally higher than in wild type, although day 1 did not show a significant effect (Fig. S1). However, in 35S::miR156A, the longer meristem was not maintained and by day 6 both the meristem and the growth zone were notably shorter than those of the wild type. Adding a copy of PLT2-Venus to 35S::miR156A plants largely reversed this effect, restoring the initial length of these zones to that of wild type and permitting the zones to increase in length from day 2 to day 4 (Fig. 6). These data support the idea that the effect of miR156 on the length of the meristem acts at least partially through its effect on PLT2 expression.

To test further whether elongation of the growth zone might be regulated by the phase of the plant, we investigated the tem1tem2 double mutant, which has elevated levels of the miR172 maturity factor (Aguilar-Jaramillo et al., 2019). The growth rate of these plants increased faster than wild type (Fig. 6A). Relative expansion rate remained similar to that of wild type, but the length of the meristem, as well as the total length of the growth zone, increased more rapidly than in the wild type (Fig. 6B,D). These data support the idea that longer growth zones are a feature of more mature plants.

In a separate set of experiments, the rate of cell production was calculated from the ratio of root growth rate to the length of mature cortical cells (Baskin, 2013). The higher-than-wild-type growth rate of the tem1tem2 double mutant was associated with a higher-than-wild-type rate of cell production (Fig. S1). Because cell production rate also increased as wild-type plants aged, these data also support the idea that the meristems of tem1tem2 plants accelerate developmental transitions seen in the wild-type root growth zone.

Having examined the effect of phase regulators on development of the root apical meristem, we were interested in knowing the extent to which the formation of lateral root meristems might be affected by the maturity of the plant. To separate effects of plant maturity from the developmental progression undergone by individual lateral root primordia, we marked the position of the root tip each day for 10 days, numbering the segments between marks from 1 (oldest) to 10 (youngest). We followed expression of PLT3 in the forming lateral root meristems using the PLT3::GUS reporter line. Because GUS staining is lethal, we removed a sample of roots each day and counted the numbers of lateral root primordia and emerged laterals in each segment. Data from one such experiment are shown in Fig. 7A.

As expected, we observed that the number of PLT3::GUS expressing spots – presumptive lateral root primordia – increased during the first 2 days after a segment formed (Fig. 7A, dark-blue bars). The number of PLT3::GUS-expressing spots then declined on the third or fourth day after a segment formed. Although some of this decline was accounted for by primordia developing into emerged lateral roots (light-colored bars), emergence does not account for all of the loss. In fact, based on examination through a compound microscope, segments that were at least 3 days old had numerous unstained lateral root primordia, indicating that PLT3 expression is repressed in some primordia after a few days.

To determine whether the maturity of the plant affects the progression of lateral root development, we investigated whether the time at which lateral root primordia lose their PLT3::GUS expression varied with the age of the plant. Based on three trials, of which the data in Fig. 7A are one, the average number of stained areas in segment 1 tended to decline steadily over time, reaching a minimum 7.3 days after the segment formed. In subsequent segments, the loss of staining reached its greatest extent approximately 4 days after the segment formed. Specifically, the number of PLT3::GUS-stained spots reached a minimum 4±1.0 days after segment 2 formed, and 3±1 days, 4.3±1.5 days and 4±0 days after segments 3, 4 and 5 formed, respectively. These data indicate that, with the notable exception of segment 1, the decrease in the number of PLT3::GUS-stained spots appears to be a sequential part of the process of forming a new lateral root, little affected by the age of the plant.

In addition to the number of PLT3::GUS-expressing spots, the size of each spot also changed as the plants matured and this feature was found to vary with the age of the plant. Unlike the previous metric, which was followed in individual segments over time, the area occupied by PLT3::GUS stain was measured on the segment of root that formed in the most recent 24 h, i.e. the root tip. For a plant that was 2 days post-germination, we looked at segment 2, and for a plant that was 3 days post-germination, we looked at segment 3, and so on. In this way, the youngest tissue in plants of one age was compared with the youngest tissue in plants at the next age. In this youngest tissue, the area of individual PLT3::GUS-stained spots increased as the age of the plant increased (Fig. S2A). The stained area increased steadily for roughly the first 4 days, and then appeared to plateau.

In the segment of root that formed 1 day before the youngest segment (e.g. segment 3 on a plant that is 4 days-post germination), the area of each stained spot increased slightly between days 1 and 2, and then declined (Fig. S2B). The decline in the size of PLT3::GUS-stained spots in lateral root primordia mirrored the decline in PLT3::GUS seen in the primary root tip (Fig. 3B). Evidently, the area of PLT3::GUS staining in lateral root primordia is affected by the age of the plant, with a notable inflection around day 4, although the specific patterns depended on the age of the segment.

The kinetics of lateral root emergence in wild-type roots can be visualized by replotting the data in Fig. 7A to show the percentage of emerged laterals for segments of interest (Fig. 7B). The length of time required for lateral root primordia to emerge decreased as roots matured. This can be seen in Fig. 7B by comparing the slopes of the lines representing segment 1 (light blue) and segment 4 (black). On segment 1, the rate of lateral root emergence was relatively slow and laterals continued to emerge even by day 10, at which time about 30% of the primordia on the segment had emerged. As the root aged, the rate of emergence rose, so that by segment 4, roughly half of the primordia on the segment emerged in a single day.

Despite the fact that lateral roots emerged more rapidly as roots matured (Fig. 7B), plants that overexpress miR156 emerge lateral roots more rapidly than wild type (Yu et al., 2015). We confirmed this finding by counting the number of emerged lateral roots on segments 1, 2 and 3 over time in wild type and 35S::miR156A (Fig. 8). This established a paradox as both more mature root segments and roots that overexpress a juvenility factor emerge lateral roots more rapidly. We wondered if this situation might arise from PLT2, insofar as PLT2 is expressed as a bell-shaped curve with lower levels in both older wild-type plants and in younger plants that overexpress miR156A (Fig. 3D). To check, we compared the rate of lateral root emergence in 35S::miR156A lines with and without PLT2-Venus crossed in, and found that the additional PLT2-Venus gene effectively restored the rate of lateral root emergence essentially to that of the wild type in all three of the root segments examined (Fig. 8).

If PLT2-mediated repression of lateral root emergence occurred predominately during the brief peak of PLT2 expression around day 4, it might shift the age at which lateral roots emerge. To see if this were the case, emerged lateral roots were counted in wild type and plt2-3. For this analysis, only the portion of primary root that grew by the end of day 3 (segments 1-3) was considered. The average number of emerged roots that formed on this portion of the root in wild type and plt2-3 appeared similar at day 8; however, on days 4 and 5 the plt2-3 loss-of-function mutant had more lateral roots than did wild type (Fig. 9A). Thus, the presence of PLT2 delayed the age at which emerged lateral roots appear.

Having identified roles for PLT expression in the timing of root development, we wondered whether PLT genes might also regulate the timing of any developmental events in the shoot. As A. thaliana plants mature, the shape of successive leaves gradually progresses from round to oblong (Wu and Poethig, 2006). Our data confirm this progression from round leaves (aspect ratio near 1) to a more elongated form (aspect ratio near 2; Fig. 9B). A progression similar to that of the wild type was seen in a loss-of-function mutant of PLT4 (babyboom 1-1). As previously described, the first two leaves of a plt3plt5plt7 triple mutant show no difference from wild type (Radhakrishnan et al., 2020). However, in older leaves, the plt3plt5plt7 triple mutant took on a more mature aspect more rapidly than did wild type (Fig. 9B). Evidently, absence of these three ‘early’ PLTs accelerates maturation of leaf shape. Thus, in the shoot, these PLTs retard this aspect of maturation. These data, and the data showing that PLT2 affects the age at which emerged lateral roots appear, indicate that PLT genes affect the age at which some developmental events occur throughout the plant.

We investigated the possibility that roots undergo phase change by characterizing how a pivotal regulator of plant phase, miR156, and a known regulator of meristem activity, PLT2, influence root development. Overexpression of miR156 increases the density of the root system (Figs 1, 5). Our data further show that expression of PLT2 and the miR156 family are linked by a mutually repressive regulatory network (Figs 3-5). Because PLT2 is expressed at high levels for a brief period of time, its effects on meristem development (Fig. 6) and lateral root emergence (Figs 8, 9) are restricted to specific developmental stages. Our model of how these factors work together to regulate root development is shown in Fig. 10.

The data presented here indicate that PLT2 and miR156 form a regulatory loop in which each represses expression of the other. Repression of PLT2 by miR156 is shown in Fig. 3 and is consistent with the observation that PLT2-Venus partially rescues the effects of miR156A overexpression (Figs 5, 6, 8). Mature miRNA transcripts target mRNAs for destruction by binding to regions of mRNA with complementary sequences. Although PLT2 and miR156C have a 10-base-pair-long region of alignment, there is as yet no evidence that this pairing has a functional significance. The reported targets of miR156 are in the SPL family, which suggests that the effect of miR156C on PLT2 is indirect.

Evidence that PLT2 represses miR156C is seen in the expression of miR156C::GUS (Fig. 4). PLT2 binds genomic DNA 1.1 kb upstream of another miR156, miR156D (Santuari et al., 2016). Like miR156C, miR156D is expressed in root meristems (Yu et al., 2015). Although motifs resembling the PLT2-binding consensus do exist near miR156C, binding was not reported to be above the designed threshold for the experiment (Santuari et al., 2016). In addition, PLT4 (BBM) has been shown to bind the miR156C and miR156D genes (Horstman et al., 2015, 2017). Thus, PLTs may affect expression of several miR156 genes in the root meristem.

PLT2 may also induce expression of some miR172 genes, a microRNA family associated with shoot maturation. A whole-genome study of PLT2 binding identified a PLT2-binding site 446 bp before the transcription start site of miR172E, and about 1.5 kb after the start sites of miR172A and miR172D (Santuari et al., 2016). In addition, overexpressing PLT2 reduced the level of the miR172 targets SNZ, SMZ and TOE1 (Santuari et al., 2016). PLT2 binding was not reported near SNZ or SMZ (Santuari et al., 2016) further strengthening the idea that PLT2 represses these genes by inducing miR172. Our attempts to confirm a difference in levels of pre-miR172 in wild type and plt2-3 by qPCR in day 6 root tips were hampered, at least in part, by overall low levels of expression. However, the published data raise the possibility that PLT2 may serve as a switch, repressing the miR156 juvenility factor and inducing miR172.

It appears that the level of PLT2 expression affects the balance between promoting growth at the present moment and generating the potential for future growth, and the optimal trade-off between these states may vary as the plant matures. Because levels of miR156C are high at germination, expression of PLT2 is expected to be strongly repressed. Indeed, levels of PLT2 protein were low in the first day or two after germination (Fig. 3D). High levels of PLT2 repress the rate of cell elongation (Mähönen et al., 2014). Therefore, low levels of PLT2 around the time of germination should allow the rate at which existing cells elongate to increase, promoting elongation of the radicle and allowing the root to emerge from the seed and reach the soil more rapidly, a factor that could be advantageous at this early stage of the life cycle. However, if the total level of PLT expression remains low for too long, the meristem may fail to maintain its length (Galinha et al., 2007). The regulatory relationship between miR156C and PLT2 provides for a subsequent increase in PLT2 expression that allows the meristem to enlarge, setting the stage for a robust level of future growth (Fig. 10).

Considerable evidence supports the idea that a pulse of auxin is transported from the shoot to the root shortly after germination, and it is interesting to consider how that pulse might interact with the developmental changes observed here. Two separate studies detected sharp increases in the auxin concentration in the root tip around day 8 (Marchant et al., 2002; Bhalerao et al., 2002). These increases are too late to influence the growth zone between days 2 and 6. In one of the studies, using long-day plants, such as the ones used here, the increased concentration of auxin in the root tip on day 8 was preceded by an increase in auxin concentration in the root near the root-shoot junction at day 6 (Bhalerao et al., 2002). A pulse of auxin in the shootward segments arriving at day 6 is also too late to explain the bursts of lateral root emergence in segments 1 to 3 (Figs 7, 8). The rate of lateral root emergence was affected around day 7 on segments 2 and 4 (Fig. 7B), an effect that could be temporally correlated with the arrival of the auxin pulse, but it is not clear how repeatable this is. It would be interesting to investigate whether the arrival of the auxin pulse contributes to ending the first period of growth zone development. A change in auxin level could be related to a change in PLT expression, either as cause or effect, because PLT expression promotes auxin synthesis and auxin promotes PLT expression over the long term (Santuari, et al., 2016; Mähönen et al., 2014). Long-term exposure to auxin also increases expression of miR156 (Yu et al., 2015).

Phase change can be defined in many ways, but the general idea is that plants, like animals, go through distinct stages of development that can be characterized by a co-occurring group of attributes that are regulated by a dedicated system. The idea that phase change may occur in the root is perhaps surprising as the type of organs produced by the root remains unchanged; however, age-dependent changes in root development have been widely observed. Many root crops, such as radish, undergo conspicuous secondary thickening as plants mature (Hoang et al., 2020). Even in A. thaliana, periclinal division of the cortical cells gives rise to a second cortical layer during the second week after germination (Dolan et al., 1993; Cui, 2016; Bertolotti et al., 2021). Another example of differential root development is found in maize, in which many of the genes that affect brace root formation also affect plant phase (Hostetler et al., 2021).

Here, a burst of PLT2 expression that occurs around day 4 coordinates growth of the meristem and a delay in lateral root emergence (Figs 6, 9). These events are regulated by miR156. Because miR156 regulates phase, being regulated by miR156 is effectively synonymous with being regulated by the phase of the plant. Additional support for the involvement of the classical phase-change pathway in these coordinated developmental events in the root comes from prior work indicating that a miR156-resistant form of SPL10 (pSPL10:rSPL10) prevents emergence of lateral roots for nearly 15 days (Yu et al., 2015). An miR156-resistant SPL10 line also produces roots with an apparently longer meristem (Barrera-Rojas et al., 2020). Taken together, we argue that roots undergo phase change, and that the burst of PLT2 that occurs around day 4 coordinates a brief phase of development between germination and juvenile vegetative growth, during which the meristem expands. We suggest that this period be called the seedling phase (Fig. 10).

The data presented here support the hypothesis that establishment of root meristems is a phase-regulated attribute of plant development, similar to leaf shape or trichome number in the shoot. The presence of phase-regulated aspects of development in the root further implies that roots undergo phase change. Being aware that roots have distinct phases is crucial to the design of research projects. Additionally, knowing that young seedlings have higher levels of PLT3 expression and lower PLT2 has implications for efforts to increase regeneration in older recalcitrant plants, and may provide ideas for regulating the rate at which the first lateral roots emerge, a factor that can impact seedling establishment and susceptibility to broadcast herbicides. The massive density of the 35S::miR156A root system is of particular interest, especially as some of the increase is maintained even when the root system is grafted onto other shoot systems, as this may allow for the generation of plants that can more effectively store carbon in the soil.

A. thaliana L. (Heynh) ecotype Columbia were grown on vertically oriented agar plates (with 1/2× MS, pH 5.8 with MES, 1% sucrose and 1.5% agar) at 23.8°C with 16 h light/8 h dark, unless otherwise noted. The 35S::miR156A seed was a gift from Scott Poethig (University of Pennsylvania, PA, USA); tem1tem2 (Osnato et al., 2012) was a gift from Soraya Pelaz (CRAG Centre for Research in Agricultural Genomics, Barcelona); the PLT3 promoter used for PLT3::GUS was described by Galinha et al. (2007), and its GUS fusion, as well as PLT2-Venus were gifts from Yujuan Du [Institute of Plant Sciences Paris-Saclay (IPS2), France]; and plt3plt5plt7 triple mutants (Prasad et al., 2011) and bbm1-1 (Galinha et al., 2007) were from Viola Willemsen (Wageningen University & Research, The Netherlands).

CRISPR/Cas9-mediated mutagenesis was used to generate the PLT2 knock-out allele plt2-3. A construct was generated using Golden Gate cloning (Engler et al., 2014; Nekrasov et al., 2013), harboring two sgRNAs. Unless otherwise stated, plasmids originated from the MoClo Toolkit and Plant Parts kit (Addgene plasmid #1000000044 and plasmid #1000000047). pAGM4723-FASTR-RPS5a::aCas9-PLT2sgRNA2-PLT2sgRNA3 was generated by using spacer sequences TGTGAAGAGTGAATGTGAGG and CTTAGGAGTGAGCAAATCGG to design forward sgRNA primers and amplify corresponding PLT2sgRNA2 and PLT2sgRNA3 using the pICH86966::AtU6p::sgRNA_PDS construct (Addgene plasmid #46966) as a template. PCR products were combined with AtU6-26 promoter from level 0 plasmid pICSL90002 (Addgene plasmid #68261) into level 1 vectors pICH47751 and pICH47761, respectively. Subsequently, level 1 vectors harboring sgRNAs were combined with pICH47732-FAST_R (red seed selection) and pICH47742-RPS5a::aCas9 and the end linker pICH41780 into level 2 binary vector pAGM4723. Plasmid pICH47732-FAST_R was generated by Golden Gate cloning of the pFAST-R selection cassette from pICSL7008 (monomeric tagRFP from Entacmaea quadricolor fused to the coding sequence of AtOLE1) into pISCH47732. Plasmid pICH47742-RPS5a::aCas9 was generated by assembling pICH41233-RPS5a, pICH41308-aCas9 and pICH41421(nosT) into pICH47742. The RPS5a promoter was amplified using pRPS5AF-BpiGGAG (TGTGAAGACAAGGAGCTCAACTTTTGATTCGCTATTTG) and pRPS5AR-BpiTACT (TGTGAAGACAAAGTAGGCTGTGGTGAGAGAAACAGA) followed by Golden Gate cloning into pICH41233. A plasmid harboring the A. thaliana codon optimized aCas9 was kindly provided by the Puchta lab (Fauser et al., 2014) and amplified using a Cas9F-BpiAATG (TGTGAAGACAAAATGGATAAGAAGTACTCTATCGGACTC) and a Cas9R-BpiGCTT (TGTGAAGACAAAAGCTCAAACCTTCCTCTTCTTCTTAGG) followed by Golden Gate cloning into pICH41308.

Columbia plants were transformed by floral dip (Clough and Bent, 1998), and T1 transgenic seeds selected under a fluorescence binocular. Inflorescences of T1 plants were genotyped for induced mutation by PCR using primers plt2-3F (CCAAACTTGCGTTTCTCAAA) and plt2-3R (AGAGGCACAAGTGACGACTG) followed by sequencing (Table S1). Seeds of mutant plants were selected for absence of the CRISPR/Cas9 construct followed by genotyping for homozygosity of the mutation. The plt2-3 allele is characterized by a single A nucleotide insertion at 80 bp into the second exon thereby creating a frameshift and an early stop codon 14 aa downstream.

Seeds were sterilized in 70% ethanol, followed by 12-15 min in 2.6% sodium hypochlorite, and four washes in sterile deionized water, prior to being stored at 4°C for about 72 h. The morning following seeding, plates were placed under a dissecting microscope and those with emerged radicles were marked as germinated. Germination checks continued at 12 h intervals until a sufficiently large cohort of seeds germinating in one 12 h window was obtained. This time was defined as time zero, and only that cohort was followed for the rest of the experiment, ensuring that plants harvested on a given day would be at the same developmental age. Roots were excluded from the experiments based on pre-established guidelines. Roots were excluded if they: (1) grew into the agar rather than on top of it, (2) grew off the agar into the air, (3) fell off the agar, (4) ran into the edge of the plate, (5) died in part or whole, or (6) were on plates that were visibly infected.

The position of the root tip was marked at daily intervals. A plate holder in the growth chamber was set to eye level, and scratches were made on the back of the plate with a razor blade to indicate the position of the root tip. At the end of the experiment, plates were photographed using a digital camera. Root lengths and angles were measured using ImageJ (Schneider et al., 2012).

For experiments involving GUS stain, plates were placed in a horizontal position and substrate solution (Willemsen et al., 1998) was gently pipetted over each plant. Plates were wrapped in foil and left at 37°C for 45 min to 16 h, with the time varying by experiment, prior to being placed on a glass microscope slide in water, or, for miR156C::GUS in Fig. 4, in ice-cold chloral hydrate. Roots were photographed on a Zeiss Axioscope or an Olympus SZX7 dissecting microscope with cellSens Standard software. The intensity of PLT3::GUS staining was determined using ImageJ to calculate mean intensity of the inverted red channel for a region of interest that fit within the meristem (Fig. S3).

Cell length was measured in propidium iodide-stained roots under a Zeiss LSM Pascal 5 confocal microscope, using the built-in image analysis software. This population was germination checked at 24 h intervals. Cortical cell measurements began at the first cell shootward of an emerged root hair, and images were taken along the root in the shootward direction until approximately 20 cortical cells were obtained; in younger roots, the number of cortical cells that could be measured in a given root was often fewer than 20. Measured cells were averaged for each root. Reported values reflect the averages and s.e.m. of values for eight (or in one case nine) roots per genotype per day. Cell production rates were calculated as: average root elongation rate (μm/day) divided by average mature cell length (µm/cell)=cells/day (Baskin, 2013). Errors were propagated by calculating the fractional error for each initial value and then taking the square root of the sum of the squares. The resulting value was then multiplied by the cell production rate to obtain the error (http://www.upscale.utoronto.ca/PVB/Harrison/ErrorAnalysis/).

For kinematic analysis, the root tip was imaged through a 10× objective. For each root, eight images were acquired, separated by 10 s. The time that any plate spent in the horizontal position was restricted to 10 min or less. After imaging, the plate was returned to the growth chamber. Later, velocity profiles were generated for each pair of images separated by 40 s using Stripflow software (Baskin and Zelinsky, 2019). For each root, the four velocity profiles thus obtained were averaged. Then, the average profile was fitted to a modified sigmoidal function (Peters and Baskin, 2006). Fitted values for four parameters are reported here. First is maximal velocity, which is equivalent to root growth rate. The second is relative elongation rate (often called ‘cell elongation rate’), obtained from the slope of the profile within the elongation zone where velocity increases steeply and almost linearly. The third is meristem length, taken as the position where the velocity profile transitions between regions of gradual and steep increase; this length includes a short region adjacent to the meristem where cells no longer divide. Finally, the fourth is growth zone length, taken as the position where the velocity profile transitions between regions of steep increase and constancy; constant velocity means zero relative elongation and hence represents the mature zone (Baskin and Zelinsky, 2019).

Plants were germinated on agar plates without sucrose, and seedlings were grafted when the first leaves were emerging as described by Melnyk (2017). Approximately 1 week after grafting, grafts were tested by gently pulling on the shoot: those that held together and showed no signs of adventitious root formation were transferred to agar plates or rhizotrons and grown until maturity (∼6 weeks). Rhizotrons were prepared by placing a thin foam spacer between two sheets of safety glass and wrapping three edges of the glass in Micropore tape (3M). The space between the glass was filled with dry soil (Just Natural Organic All Purpose Garden Soil, Oldcastle Lawn & Garden, Inc., Atlanta, GA, USA) that had been passed through a sieve to remove large particles and baked overnight at 60°C. Packing of soil was promoted by repeatedly pounding the sealed bottom of the rhizotrons on the floor; additional soil was then added to the top until the structure was full. Filled rhizotrons were submerged overnight in water containing approximately one capful per 66 l water of Miracle-Gro All Purpose Plant Food (Miracle-Gro Lawn Products Inc., Marysville, OH, USA). Rhizotrons were then wrapped in foil, leaving an opening along the narrow strip of soil at the top, planted, and placed in the growth chamber at about a 45° angle. To assay root density, foil was removed and rhizotrons were imaged from the previously lower side against which the roots had grown. The perimeter of each root system was drawn by eye, and the average mean gray value of the pixels inside the resulting shape was determined using ImageJ (Schneider et al., 2012). To control for between-rhizotron differences, the density of each root system is reported relative to wild type. For Fig. 5B bottom (35S::miR156A versus 35S::miR156 plt2-3), the sample size was increased from five to 15 after the initial group size proved to be underpowered.

Confocal laser-scanning microscopy images of PLT2-Venus were obtained on a Zeiss LSM 880 Airyscan confocal laser-scanning microscope using an LD LCI Plan Apochromat 40×/1.2 immersion-corrected DIC objective. Roots were placed on a coverslip in a drop of water or propidium iodide (1 µg/ml, to stain the cell walls and visualize the root meristem). The root tip was imaged using the tile scan function in Zen Black. Single-channel grayscale images were exported from ZEN Black and opened in ImageJ. Mean gray intensity was measured over a rectangular region of interest that covered the stem cell niche (Fig. S3).

Plants were grown in soil until they had nine leaves, when they were uprooted and inverted. Leaves were removed from the base of the shoot (first formed) to the innermost (upper, most recently formed) with the aid of a dissecting microscope, and then affixed to a sheet of clear Plexiglas using double-sided tape. Leaves were scanned on a Konica Minolta Bizhub 654e machine, and images quantified using ImageJ (Schneider et al., 2012). Maximum length and width were recorded.

We thank the National Science Foundation for funding the purchase of the confocal (MRI 1828041); Greg Lee for helpful conversations; Zofia Stanley for statistical advice; Laura Green for help drawing figures; Andrea Paterlini for teaching us how to graft; Davy Opdenacker and Tom Beeckman along with Kavya Yalamanchili and Ikram Blilou for showing us how to make mini-rhizotrons; Scott Poethig for 35S::miR156A; Soraya Pelaz for tem1tem2; Carlos Barrera-Rojas and Fabio Tebaldi Silveira Nogueira for pmiR156A::GUS and pmiR156C::GUS lines and technical advice; Ben Scheres' lab for PLT mutant and reporter lines; and Connie Estevez and undergraduate students in Oberlin College BIOL 204 and BIOL 221 for preliminary experiments.