From one cell to many: Morphogenetic field of lateral root founder cells in Arabidopsis thaliana is built by gradual recruitment

Significance

A single plant can produce from hundreds to millions of lateral roots (LRs), which facilitate water and mineral uptake and transport. LRs originate in a tissue called the pericycle, but the underlying mechanism is obscure. With the aid of clonal analysis, we deduced that LR initiation begins with a single pericycle founder cell (FC) which then recruits the neighboring pericycle cells to become FCs. A time-lapse analysis validated this conclusion and also showed that FC recruitment is an auxin-dependent and gradual process that takes place in an unpredictable pattern. Thus, LR initiation is not a one-step process, as previously considered.

Abstract

The reiterative process of lateral root (LR) formation is widespread and underlies root system formation. However, early LR primordium (LRP) morphogenesis is not fully understood. In this study, we conducted both a clonal analysis and time-lapse experiments to decipher the pattern and sequence of pericycle founder cell (FC) participation in LR formation. Most commonly, LRP initiation starts with the specification of just one FC longitudinally. Clonal and anatomical analyses suggested that a single FC gradually recruits neighboring pericycle cells to become FCs. This conclusion was validated by long-term time-lapse live-imaging experiments. Once the first FC starts to divide, its immediate neighbors, both lengthwise and laterally, are recruited within the hour, after which they recruit their neighboring cells within a few hours. Therefore, LRP initiation is a gradual, multistep process. FC recruitment is auxin-dependent and is abolished by treatment with a polar auxin transport inhibitor. Furthermore, FC recruitment establishes a morphogenetic field where laterally peripheral cells have a lower auxin response, which is associated with a lower proliferation potential, compared to centrally located FCs. The lateral boundaries of the morphogenetic field are determined by phloem-adjacent pericycle cells, which are the last cells to be recruited as FCs. The proliferation potential of these cells is limited, but their recruitment is essential for root system formation, resulting in the formation of a new vascular connection between the nascent and parent root, which is crucial for establishing a continuous and efficient vascular system.

In plants, organogenesis is maintained throughout the lifespan of an organism due to the activities of their meristems. Shoot and root branching result in the continuous production of new apical meristems. In Arabidopsis thaliana, lateral roots (LRs) arise deep within the root, in the pericycle, the single layer of cells constituting the most external tissue of the vascular cylinder. Xylem pole pericycle (XPP) cells continue to divide after leaving the root apical meristem (1, 2) and maintain morphogenetic potential. Before LR primordium (LRP) initiation, XPP cells become primed (3, 4) and competent to form LRs. This competence is restricted in time and space (5, 6). Despite the wide use of A. thaliana as a model organism, we still lack an understanding of the cellular bases of LR initiation and do not know how the pericycle founder cells (FCs), which give rise to an LR, become involved in the initiation process, whether certain pericycle cells are predetermined to become FCs and if so, in what pattern, and whether the number of FCs is strictly defined.

During LR formation, the plant hormone auxin acts as a morphogenetic trigger that specifies certain XPP cells as LR FCs (7). In most cases, two longitudinally adjacent FCs are thought to give rise to the entire primordium in the longitudinal plane (SI Appendix, Fig. S1) This type of LRP initiation, known as longitudinal bicellular initiation (8), has been identified in a number of angiosperm species (9) and is considered to be the prevalent type of LRP initiation which is well documented in anatomical and time-lapse studies in A. thaliana (8, 10–13). However, another type of LRP initiation, the longitudinal unicellular type, has been reported in which a single FC in a cell file gives rise to the entire LRP in the longitudinal plane (7, 8). It is unclear how widespread each type of initiation is and whether either type of initiation can occur in a given LRP. As it is difficult to predict an LRP initiation site, even time-lapse studies may not provide definitive answers to these questions (14). In the transverse plane (SI Appendix, Fig. S1), three contiguous XPP cells are commonly considered to be involved in LRP initiation (8, 15, 16). Recently, it was reported that five to seven cell files of contiguous XPP cells give rise to an LR (14). To clarify which type of LRP initiation is most common in A. thaliana, we performed both a clonal analysis and time-lapse experiments.

Clonal analysis provides insight into the morphogenetic components of organ development, such as the distribution, orientation, and abundance of cell divisions, cell fate, and an estimation of the number of cells giving rise to an organ (17, 18). However, a thorough clonal analysis of LR formation is laborious and was last performed more than half a century ago (19). In the current study, with the aid of clonal analysis, we deduced and subsequently validated using time-lapse tracking that, in most cases, LR initiation starts with a single FC. Neighboring XPP cells in other cell files are gradually recruited as FCs and participate in LRP formation. In dicots, phloem-pole pericycle (PPP) cells are not thought to participate in LRP formation (1, 2, 15, 16). Unexpectedly, we established that the last recruited FCs are PPP cells, and they are also induced to divide, participating in the formation of the vascular connection between the nascent and parent root. The pattern of pericycle cell division related to LR initiation is more variable than generally considered, and LR initiation is a multistep process.

Results

Clonal Analysis Suggests That Lengthwise LR Initiation Starts with a Single FC.

During plant organ development, no cell rearrangement occurs, and progeny cells do not move, which facilitates clonal analysis. In this study, we implemented a classical clonal analysis of LRP initiation using a previously described system (20). A heat shock-induced transposition of a DS1 transposon takes place in 35S-DS1-H2B:YFP; HS-Ac seedlings and results in the production of a yellow fluorescent protein (YFP)-labeled cell, which after division results in labeled cells (20). Twenty-five XPP clones, consisting of three to six cells, were found outside LRPs or LRs, confirming that XPP cells maintain proliferation in the differentiation zone (1, 2); these clones were excluded from the analysis. A total of 84 clones were identified within the LRP or LR (SI Appendix, Table S1). In the longitudinal plane, clones with a single (Fig. 1A), pair (Fig. 1B), or group of five (Fig. 1C) closely spaced YFP-marked cells were identified. As no other YFP-marked pericycle cells were found in the shootward and rootward direction of these clones, the data suggest that longitudinally only one FC participates in LRP formation. Other cases supporting this scenario were found (Fig. 1 D–F). These clones support the previously proposed longitudinal unicellular type of LRP initiation (8). Clones spread in the longitudinal plane approximately half the length of the LR thickness were also found (Fig. 1H). These clones suggest that two FCs in the longitudinal plane could give rise to the LR, corresponding to the longitudinal bicellular type of LRP initiation (7, 8). Alternatively, these clones could have resulted from a division of a single cell (in the longitudinal plane), with Ds transposition occurring later and only in one of the daughter cells. In some LRs, two-thirds of the LR thickness in the longitudinal plane was marked (Fig. 1 J and K). This scenario could correspond to the unequal participation of two longitudinally adjacent FCs in LRP formation. Other type of clones with one-third the central portion of the LR thickness (in the longitudinal plane) not marked (Fig. 1 M and N) or with approximately one-third of the central portion YFP-marked (Fig. 1P) were found. These clones could be interpreted in two ways: either three progenitor cells in the longitudinal direction gave rise to the LR, or Ds transposition occurred after LRP initiation.

Fig. 1.

Clonal analysis suggests that predominantly a single FC in the longitudinal plane gives rise to an LR. (A–F) Clones that correspond to the longitudinal unicellular type of LRP initiation. (A) A single YFP-marked XPP cell. (B) A clone of two longitudinally adjacent cells resulted from the first cell division of the FC. The borders of the clone are indicated with arrowheads. (C–E) Clones distributed along the entire length of the LRP: (C) stage I, (D) stage II, and (E) stage V. (F) An LR with a clone distributed along its entire thickness in the longitudinal plane. (G) Schematic depiction of the clones corresponding to one FC in the longitudinal plane giving rise to the LRP or LR. (H and I) A clone and its schematic depiction distributed along half of the thickness of an emerged LR in the longitudinal plane. (J and K) A clone and its schematic depiction distributed along two-thirds of the thickness of the emerged LR in the longitudinal plane. The dashed line separates images of the same LR. (L) stage IV LRP and (M) emerged LR containing clones of YFP-marked cells at their flanks. (O) Emerged LR with a clone distributed across approximately one-third of its thickness in the longitudinal plane. (N and P) Schematic depiction of the clones shown in L, M, and O, respectively. Numbers of each type of clones out of the total number of clones analyzed are indicated for G, I, and K and N and P. Seedlings were analyzed at 6 d after germination (dag). Magnification in A–D and L, M, and O is the same; p, pericycle; px, protoxylem. (Scale bar, 40 µm.)

Thus, the clonal analysis showed that three types of clones could correspond to models in which longitudinally one (Fig. 1G), two (Fig. 1 I and K), and three (Fig. 1 N and P) FCs give rise to the LR in 70, 20, and 5% of the total of 84 clones, respectively (Fig. 1 and SI Appendix, Table S1). This suggests that overall the longitudinal unicellular type of initiation is more common. Therefore, the clonal analysis strongly supports the scenario in which a single FC along the root participates in LRP initiation as proposed earlier (8).

Anatomical and Clonal Analyses Suggest That after the First FC Starts to Divide, Neighboring Pericycle Cells Become Founders.

To deduce the number of FCs participating in LRP formation in the transverse plane, Z-sections were collected, and clonal and anatomical analyses were combined to establish what fraction of a developing LRP is composed of a clone of YFP-marked cells. Three-dimensional analysis permitted the identification of LRPs with a single longitudinal file of YFP-marked cells and two adjacent cell files with fewer cells within the LRP (Fig. 2A). This suggested that LR initiation started in the FC that formed a clone and that the neighboring pericycle FCs entered cell division at a later time point. We hypothesize that, after the first FC division, neighboring pericycle cells in the transverse plane (i.e., laterally located) become recruited as FCs. The number of cells arising from each FC can be used to deduce the hypothetical time sequence of FC recruitments involved in LRP formation. Using this approach, we can hypothetically identify the first FC (designated as FC0) and the first, second, etc. sequentially recruited FCs (1, 2, etc. in Fig. 2 A–C). In some clones, three (Fig. 2B) or four (Fig. 2C) FCs produced clones of YFP-marked cells. The probability of clones forming simultaneously in neighboring cells is low. A possible model is that an XPP YFP-marked cell divided by a tangential division (perpendicular to the radius of the parent root; SI Appendix, Fig. S1) within the root apical meristem, and then the YFP-marked daughter cells left the meristem and were specified as FCs. Based on clonal analysis, we deduced that the number of FCs that gave rise to the LRP was 1, 3, 4, and 6 FCs in 15.7, 23.5, 37.3, and 2.0% of cases, respectively (n = 51; SI Appendix, Table S1).

Fig. 2.

Clonal analysis suggests that LRs are initiated from a single FC, which recruits neighboring FCs. (A and B) stage I and (C) stage II LRPs composed of progeny of three (A and B) and four (C) FCs. Clones of YFP-marked cells were formed from one (A), three (B), and four (C) FCs. Z-sections are shown in A and B. (D) A three-dimensional (3D) reconstruction and a transverse section of a stage V LRP showing a sector of marked cells corresponding to one-quarter of the LRP in the transverse plane. (F and H) A 3D reconstruction of emerged LRs containing a clone of YFP-marked cells that occupies a central (F) or peripheral (H) sector of approximately one-sixth of its thickness in the transverse plane. Details of the clone depicted in F are shown in SI Appendix, Fig. S2. (E, G, and I) Schematic depictions corresponding to D, F, and H, respectively. Arrows in G show the lateral view (LV) and frontal view (FV); see also SI Appendix, Fig. S1. Cell 0 in A–C is hypothetically the first FC that started LR initiation; the hypothetical sequence of FC recruitment is indicated by a number near arrows. All panels, except C, show LRPs or LR in lateral view; C shows the LRP in frontal view. Seedlings were analyzed at 6 dag. The dashed line separates images of the same LRP in A, B, and D or LR in H. px, protoxylem. (Scale bar, 40 µm.)

By analyzing clones at later stages in the transverse plane, the number of FCs was deduced based on the fraction of the LRP occupied by a clone. For example, in the clone depicted in Fig. 2D, approximately one-quarter of the annulus sector of the LRP was occupied by a clone of YFP-marked cells, suggesting that four FCs gave rise to the LRP. The YFP-marked clones were also spread to approximately one-sixth of the LR annulus sector. These clones were formed in a central (Fig. 2 F and G and SI Appendix, Fig. S2) or peripheral (Fig. 2 H and I) portion of the sector viewed in the transverse plane. Therefore, we deduced that six FCs gave rise to the LR in the transverse plane. Overall, our clonal analysis showed that 3, 4, and 6 FCs gave rise to the emerged LR in 3.0, 12.1, and 54.5% of cases, respectively (n = 33; SI Appendix, Table S1). The average number of deduced FCs increased from 3.2 ± 1.2 in the young LRPs (n = 40) to 5.5 ± 0.9 in emerged LRs (n = 23; mean ± SD; P < 0.001, Mann-Whitney rank sum test). This analysis supports the hypothesis of FC recruitment during LR formation.

The alf4-1 (aberrant lateral root formation 4–1) mutant is deficient in activation of cell division in pericycle FCs (7, 21). As presumptive FC specification does occur in this mutant, but LRPs are not formed, we hypothesized that the recruitment of new FCs is blocked and that only a small number of presumptive FCs exist in the transverse plane. To this end, we analyzed the number of presumptive FCs in alf4-1 DR5 seedlings as described (7). Indeed, in most cases, only one presumptive FC in the transverse plane was specified (number of presumptive FCs in the transverse plane was 1, 2, and 3 in 62.7, 32.6, and 4.7% of cases, respectively; n = 43 preinitiation events in 11 roots; SI Appendix, Fig. S3). Therefore, these data are consistent with the anatomical and clonal analyses and strongly suggest that first only one FC is specified after which it recruits neighboring pericycle cells to become FCs.

Time-Lapse Experiments Confirm That the Longitudinal Unicellular Type of LR Initiation Is the Most Common.

To validate the conclusion that the LRP is initiated from a single FC in the longitudinal plane, we implemented a time-lapse analysis of p35S::H2B-RFP pUBQ10::NPSN12-YFP (22, 23) seedlings. As the exact position of the LR initiation event is difficult to predict (4, 5), we designed an approach of “confocal hunting” for the FCs (SI Appendix, Fig. S4). To this end, we detected the most distal LRP and programmed the image acquisition of a confocal laser scanning microscope in a way that a series of Z-sections in sequential and overlapping fields toward the apex was captured every 30 min during a 24-h period. The time series of images were analyzed retrospectively, and when division of the first FC was detected and could be followed until LRP stages II to IV (24) over 24 h, the LRPs were analyzed. The first pericycle FC whose descendants contributed to an LRP was denominated as FC0. In LRPs where more than one FC0 was identified (see next section), only the FC0 that made the greater contribution to the LRP was included in the analysis of types of initiation.

Both the longitudinal unicellular and bicellular type of initiation were found (Fig. 3 A and B), representing 88 and 12% of cases (15 and 2 LRPs, respectively; n =17). Interestingly, in LRPs #16 and 17, which were initiated by the bicellular type, we observed that, in laterally adjacent cell files, there was a single FC that started to divide at the same time as longitudinally adjacent FCs (SI Appendix, Fig. S5). Therefore, these cases were considered to be of mixed type of initiation. To compare the two types of initiation, we estimated the cell cycle duration of each descendant of FC0. When FC0 entered mitosis, it was completing its first cell division cycle. This time was recorded as time 0 for the second cell division cycle, and thus the duration of the second, third, and fourth division cycle was determined. In this way, the FC0 progeny genealogical trees were constructed (Fig. 3 C and E). We observed that in both types of LRP initiation, the second cycle of FC0 descendants was shorter than that of the third and fourth cell cycle (Fig. 3 D and F), and the overall average cycle time (for cycles 2–4) was relatively short, i.e., 5.3 h (n = 85 for 15 LRPs) and 5.8 h (n = 38 for 2 LRPs), for uni-and bicellular types, respectively (P > 0.05, Student’s t test). Similarly, no differences between types of initiation were found for each of the second, third, and fourth cycle of descendants of FC0 (SI Appendix, Table S2), suggesting that the previously unrecognized longitudinal unicellular type of initiation indeed leads to LRP development and that no differences in outcome exist between the types. Thus, our confocal hunting method permitted us to validate the clonal analysis deduction that in most cases a single FC in the longitudinal plane initiates the formation of an LRP.

Fig. 3.

Primordia initiated by unicellular and bicellular types of initiation are similar in the duration of the first cell cycles of FC descendants. (A and B) The first LRP FC (FC0) and its progeny in p35S::H2B-RFP pUBQ10::NPSN12-YFP seedlings. (A) Unicellular and (B) bicellular types of LR initiation. Time 0 corresponds to the FC0 cell division. Arrowheads indicate the end walls of FC0. Red signal is shown in magenta. Schemes of FC0 with nuclei resulting from the first, second, third, and fourth cell cycle are color-coded in black, red, green, and blue, respectively. (Scale bar, 50 µm.) (C and E) Cell genealogy trees for FC0 shown in A and B, respectively. (D and F) Duration of the first cell cycles of FC0 descendants in 24-h time-lapse experiments; circles are color-coded as in A and B. Lines indicate median values; n = 18–38 (unicellular) and 8–16 (bicellular) cell cycles from 15 and 2 LRPs, respectively. Different letters indicate a statistically significant difference at P < 0.05, Kruskal-Wallis one-way analysis of variance on ranks. Seedlings were analyzed at 6 dag. See also SI Appendix, Table S2.

To establish whether the first FC0 division was a formative (asymmetric) or a proliferative division, we analyzed the lengths of FC0 daughter cells and their contribution in the LRPs initiated by the unicellular longitudinal type. The first FC0 division was close to symmetrical and asymmetrical in 38 and 62% of cases, respectively (n = 21, SI Appendix, Table S3). Importantly, the daughter cells always contributed to the LRP, with the predominant contribution of a shorter cell to the LRP central domain and a longer cell to the flanking domain. Therefore, the first division of the FC0 should be considered developmentally asymmetric.

After the First FC Initiates LRP Formation, Neighboring Pericycle Cells Rapidly Become Founder Cells.

The FCs by definition are those pericycle cells that contribute to LRP formation (8). The anatomical and clonal analyses performed here suggest that first only one FC is specified after which it recruits neighboring pericycle cells to become FCs. To validate this conclusion, we performed time-lapse experiments. After the FC0 has started to divide, the process of conversion of neighboring pericycle cells in LR FCs is recognized here as FC recruitment. Using our confocal hunting approach (SI Appendix, Fig. S4), we analyzed the number of FC recruitments and the pattern and time of the first divisions of each FC contributing to LRP formation in the transverse plane in four dimensions (4D). As the time of the first division of FC0 in each LRP was different, the extent of FC recruitment varied, and the number of pericycle cell files contributing to the LRP during a 24-h period ranged from four to seven, with the majority of LRPs (65%, 11 of 17) having five cell files (Fig. 4 A and C and SI Appendix, Fig. S5).

Fig. 4.

The first FC division is followed by dynamic FC recruitment. (A) An example of a time-lapse experiment showing FC recruitment dynamics in p35S::H2B-RFP pUBQ10::NPSN12-YFP seedlings. Red signal is pseudocolored in green. Asterisks indicate the division of FC0 and each subsequently recruited FC; arrowheads indicate end walls of FCs; Z-section projections are marked as a–e, each 2.5 μm thick. Time 0 corresponds to division of FC0. (B) Schematic depiction of an LRP shown in A. The sequence of each FC recruitment is color-coded. (C and D) LRPs were analyzed by stages in 24-h time-lapse experiments. The total number of pericycle cell files contributing to the LRP and the number of cell files viewed in the transverse plane (TP) in which the FC0 is found (C) and the number of LRPs with the maximum number of FCs in the longitudinal plane (LP) (D). (E) The shortest time of FC recruitment within each new cell file in the TP and LP planes. (F) Dynamics of subsequent FC recruitment irrespective of direction. Circles: median value; vertical bars: 95% confidence interval; Kruskal-Wallis one-way analysis of variance on ranks showed statistical difference (asterisks) at P < 0.05 (Dunn’s method). Data on C–F, n = 17. Seedlings were analyzed at 6 dag. (Scale bar, 50 µm.)

As FCs in more than one cell file can simultaneously participate in LRP initiation (8), we first established how common this scenario is. Out of 17 LRPs analyzed, 10 (59%), 6 (35%, LRP # 1, 2, 6, 9, 16, 17), and 1 (6%, LRP# 4) LRPs started their initiation from a single, or simultaneously two or three pericycle cell files, respectively (Fig. 4D and SI Appendix, Fig. S5). Thus, in most cases, LRP development starts with the division of a single FC followed by FC recruitment in both the transverse and longitudinal planes. The recruitment process was very fast. If the time of the first division of FC0 was considered 0 h, the shortest time of FC recruitment in each of three cell files adjacent to the one containing FC0 took place on average at 0.8, 2.2, and 4.4 h for the first, second, and third cell files contributing to the LRP, respectively, showing a steady and gradual increase in time (Fig. 4E). Similar results were obtained when each subsequent FC recruitment was evaluated (Fig. 4F). It is important to note that a single FC was recruited in each cell file at a time (e.g., yellow cell, Fig. 4B) or a pair of longitudinally adjacent cells were recruited simultaneously (e.g., blue cells, Fig. 4B). The latter case resembles the longitudinal bicellular type of LRP initiation; however, this represents an FC recruitment step in development. Over a 24-h period, up to seven pericycle cell files participate in LRP formation (Fig. 4C).

Next, we analyzed the features of FC recruitment in the longitudinal plane. Interestingly, when the LRP was initiated by the longitudinal unicellular type, the first FC recruitment in the longitudinal plane was present in 8 out 17 LRPs (LRP # 1, 2, 5, 9, 10, 12, 14, 15; SI Appendix, Fig. S5). Also, a simultaneous recruitment of two adjacent cells in the longitudinal plane, similar to the blue cells depicted in Fig. 4B, was frequently observed (LRP# 1, 4, 5, 7, 12, to 15; SI Appendix, Fig. S5). In addition, in 9 LRPs, three or four FCs were identified in the longitudinal plane (Fig. 4D). Commonly, in these cases, the most distal and proximal FCs in a file contributed to the LRP cells, but, being peripheral FCs, their contribution was minimal. FC recruitment in the longitudinal plane is also a rapid process, taking an average of 2.4 h (Fig. 4E). When in the same file, more than one FC was specified, the recruitment in the longitudinal plane predominantly occurred in the rootward (acropetal) direction (12 cases vs. 4 cases in the opposite direction, SI Appendix, Fig. S5), the same direction in which the new initiation is taking place (5), suggesting it is important for the rapid establishment of a young LRP and prevention of the next LRP initiation in too close a proximity. This analysis shows that FC recruitment in the longitudinal plane is a common process similar to that in the transverse plane.

Overall, our analysis of the timing of the FC recruitment sequence, irrespective of direction, shows that the first three FCs are recruited very rapidly, within on average 1.1 h after FC0 division (Fig. 4F). Subsequent FC recruitments were significantly later (Fig. 4F), but still overall recruitment was quick (FC7 and FC8 were recruited on average by 3.7 and 4.6 h). Not all cell files in which FC was specified had the same contribution to the developing LRP. The master cell file has been recognized as the file that has “a leading role in the development of the primordium” (14). In our experiments, FC0 had the greatest contribution to LRP formation in 10 out of 17 LRPs analyzed (SI Appendix, Fig. S5). Overall, this analysis shows that FC recruitment is a gradual process and that LRP initiation is continuous and not a one-time event.

FC Recruitment Is Auxin-Dependent.

Auxin acts as a morphogenetic trigger for FC specification (7). We thus asked whether auxin also participates in FC recruitment. To this end, we implemented a time-lapse confocal hunting approach using a maximum auxin response reporter pDR5rev::3XVENUS-N7 (25) crossed with pUBQ10::NPSN12-YFP (23). We found that while in neighboring pericycle cells no signal was present, each FC division was preceded by the appearance of DR5 signal, which was spread across every FC file within 6 h of the appearance of the DR5 signal in FC0 (Fig. 5F). There was a strong positive correlation between the time of appearance of the DR5 signal in each FC and the time of its subsequent division (SI Appendix, Fig. S6). Furthermore, the relative pDR5::3XVENUS-N7 signal intensity measurements showed that the highest auxin response was detected in FC0 and FC1 nuclei just before cell division. In adjacent FC files, the DR5 signal was progressively weaker (Fig. 5B), and overall the signal propagation was gradual (SI Appendix, Fig. S7A). PIN1 expression was also gradually propagated during FC recruitment (SI Appendix, Fig. S7B). This suggests that the auxin response is involved in the recruitment. As FCs at the periphery of the LRP viewed in the transverse plane had a lower impact on LR formation, the extent of the FC auxin response before their division aligned well with the proliferative potential of the FCs. Importantly, the FC0 daughters always maintained DR5 activity.

Fig. 5.

FC recruitment is auxin-dependent and spreads toward the PPP cells, which are also recruited as FCs. (A and B) Analysis of FC recruitment in the pDR5::3XVENUS-N7 pUBQ10::NPSN12-YFP line over a 24-h period. (A) Timing of the appearance of the DR5 signal and subsequent division in each individual FC (n = 10 LRPs). FC0 is the first FC per LRP that entered mitosis. FC1 to FC4 are subsequent recruitments. FC4 was not specified during the experiment in LRP 3 and 9–11. (B) Relative signal intensity in subsequently recruited FCs. Signal intensity was measured in FCs just 0.5 h before they entered cell division. The maximum value per LRP was assigned a value of 1. Bars show mean value ± SD, n = 10 LRPs; FC4 was not present in all LRPs. In A and B, the same order of LRPs is maintained for each FC. (C) Number of FC cell files determined for p35S::H2B-RFP pUBQ10::NPSN12-YFP seedlings in 48-h time-lapse experiments; images acquired each hour (n = 6). Inset shows clones of one LRP at the end of the experiment. Each clone includes cells formed from FCs within a cell file irrespective of the number of FCs. (D) Time-lapse analysis of changes in PPP cells in UBQ10::NPSN12-YFP pS17 (At2g22850)::H2B-2xmCherry seedlings. The first cell division of the PPP FC in St I LRP is shown; time 0 is the time before the PPP cell division; n = 10. (E and F) Phloem-loaded tracer experiments. (E) The same A. thaliana F1 cross as in D, 60 min after cotyledon loading; Inset shows tracer-loaded phloem near the LRP. (F) The 2x35S::Lti-tdTomato root 60 min after cotyledon loading; n = 7. Note, the phloem strand (marked with red squares) on E and F is adjacent to nuclei of derivatives of the PPP cells. Arrowheads and arrows indicate end walls and recently formed cell walls, respectively. Seedlings were analyzed at 6 dag. (Scale bar, 50 µm.)

To validate the importance of auxin for FC recruitment, we analyzed the effect of the polar auxin transport inhibitor N-1-Naphthylphthalamic acid (NPA). As NPA blocks LRP initiation (6, 10), we analyzed already initiated stage I LRPs and evaluated the timing of the appearance of the DR5 signal in FC0 neighboring pericycle cell files. Indeed, no propagation of DR5 signal was found (SI Appendix, Fig. S7C and Table S4). This analysis suggested that auxin transport is required for FC recruitment. However, as no new cell division was found in the developing LRPs after the NPA treatment, the recruitment blockage could be a consequence of arrested cell proliferation. Importantly, seedlings treated with a potent inhibitor of cell division, oryzalin, did not show an arrest in propagation of the DR5 signal, and the number of files with presumptive FCs was similar to that in untreated roots (SI Appendix, Fig. S7D and Table S4). This analysis shows that auxin signaling and transport are essential for FC recruitment.

Formation of the FC Morphogenetic Field Is Restricted by Protophloem-Adjacent Pericycle.

The pericycle cell layer consists of two subpopulations with different developmental fates, XPP and PPP (1, 2, 15, 16). In A. thaliana, PPP cells do not participate in LR formation (15, 16). As FC0 is specified from an XPP cell, and the FC morphogenetic field is formed toward the PPP, we hypothesized that the PPP cells should behave as a restrictive factor for the defining FC morphogenetic field boundary. To this end, we counted the number of pericycle cells in the transverse plane in a semicircle corresponding to the distance from one protophloem pole to another at the level of stage I and II LRPs (SI Appendix, Fig. S8) and found there to be 8.1 cells (n = 11). To evaluate the total number of FC files participating in the FC morphogenetic field formation, we ran time-lapse experiments over a 48-h period and established that a total of 7.1 FC cell files (n = 6) gave rise to an LR (Fig. 5C), suggesting that PPP cells can also become FCs. To verify this possibility, we used the PPP marker line pS17 (At2g22850)::H2B-2xmCherry (26, 27). PPP cells were recruited as FCs, but their proliferation potential was limited (Fig. 5D). Using a phloem-mobile symplastic tracer (28) applied to A. thaliana cotyledons, we detected the progeny of PPP cells and confirmed that these cells were recruited as FCs (Fig. 5 E and F). From these experiments, we conclude that 1) the formation of the FC morphogenetic field is restricted by the PPP, 2) the PPP cells are capable of becoming FCs, and 3) the cell progeny of PPP FCs should participate in the vascular connection between the parent root and LR to maintain continuity of phloem elements.

Discussion

Until now, we did not know how many cells begin LR initiation in plants. In this work, based on our clonal analysis of A. thaliana LR development, we deduced that, in most cases, the LRP is initiated from a single FC. Live imaging of roots and cell tracking in 4D confirmed this conclusion. Subsequently, FC0 recruits neighboring pericycle cells to become FCs in both the transverse and longitudinal planes. The first FC0 division is developmentally asymmetric and, regardless of daughter cell lengths, does not result in the formation of a stage I LRP. Therefore, this developmental stage should be recognized as stage 0 in LRP formation. Pericycle FCs do not behave as stem cells, as all their progenies become an LRP. To be considered stem cells, one of the daughters would have to maintain pericycle cell identity. This underlines the differences between stem and founder cell identities.

In contrast to the situation in animals, FC recruitment in roots is not accompanied by cell movement, but only by changing pericycle cell fate. This is a highly dynamic process that depends on the propagation of an auxin signal. This hormone behaves as a morphogenetic trigger for FC specification (7) and is also involved in the FC recruitment process (this study). FC recruitment is involved in building a morphogenetic field (concept reviewed in refs. 29–32), and when the central portion of this field has entered LRP stage I, the surrounding cells continue to be involved in the field. The gradual nature of FC recruitment and the steady spread of the auxin signal clearly show that LR initiation is not a one-step process.

Cell doubling time in a young A. thaliana LRP is less than 4 h (8). The time-lapse analysis performed in the current study confirmed that rapid cycling of the FCs is a common feature of young developing LRPs, with cell cycles being as short as 1.5 h (Fig. 3D). Cell proliferation in the early developing LRP has one of the shortest cell cycles, if not the shortest, in plants (33). Rapid establishment of the LRP cell population probably evolved as a mechanism of lateral inhibition that prevents the formation of too closely spaced LRPs. Interestingly, when FC recruitment occurs in a lengthwise orientation, in most cases, it is observed in the rootward (acropetal) direction (SI Appendix, Fig. S4), which could be involved in preventing new initiation events from taking place in the same direction (5). FC recruitment and cycling of previously recruited FCs are simultaneous processes. The exact distance from the root tip to where the FC0 starts to divide was not measured. Based on the protoxylem differentiation state and root hair development, we deduced that the first division takes place in the young differentiation zone, just after the completion of cell elongation, within a region of ∼0.8–2 mm from the root tip. The DR5:Luciferase activity in the oscillation zone of a growing root starts in the basal portion of the root apical meristem and in the elongation zone (4), and, based on images in this previous study, its maximum is found at 0.8–1.1 mm from the tip. As the root region with auxin-dependent FC recruitment coincides with the location of the DR5:Luciferase maximum, and FC recruitment is substantially advanced within 3–6 h of FC0 starting to divide, a time frame similar to the period during which DR5:Luciferase activity increases (4), we hypothesize that FC recruitment could mechanistically underlie the oscillation. This question remains to be addressed.

Root developmental plasticity results from a number of components, including a variable number of participating FC files, nonstereotypical pattern formation, and a flexible morphogenetic response to available water (13, 14, 34). In this study, we revealed an additional plasticity component related to the absence of a specific pattern and sequence of FC recruitment. Future work should examine whether each subsequent FC recruitment is involved in determining the LR emergence vector toward a mineral or water source depending on the soil microenvironment.

Materials and Methods

Seeds were surface sterilized, and seedlings growth conditions were as described (6). For clonal analysis, a heat-shock–inducible, Activator (Ac) –Dissociator (Ds) lineage-marking technique was used (20), and 218 seedlings with YFP-labeled cells were selected from >1,192 and were analyzed in detail using a Zeiss LSM 510 Meta laser-scanning confocal microscope. For time-lapse analysis, imaging was performed using the confocal system described (35). Experiments were commonly 24 and 48 h in duration, and images were acquired each 0.5 and 1 h, respectively. The detailed description of all plant materials, plant growth conditions, clonal and time-lapse analyses, microscopy, cell tracking and image analysis, phloem-loading experiments and treatments, and statistical analysis is given in SI Appendix, SI Materials and Methods.

Data Availability.

The raw data that support the findings of this study are available from the corresponding author upon reasonable request.