A dynamic genetic-hormonal regulatory network model explains multiple cellular behaviors of the root apical meristem of Arabidopsis thaliana

The study of the concerted action of hormones and transcription factors is fundamental to understand cell differentiation and pattern formation during organ development. The root apical meristem of Arabidopsis thaliana is a useful model to address this. It has a stem cell niche near its tip conformed of a quiescent organizer and stem or initial cells around it, then a proliferation domain followed by a transition domain, where cells diminish division rate before transiting to the elongation zone; here, cells grow anisotropically prior to their final differentiation towards the plant base. A minimal model of the gene regulatory network that underlies cell-fate specification and patterning at the root stem cell niche was proposed before. In this study, we update and couple such network with both the auxin and cytokinin hormone signaling pathways to address how they collectively give rise to attractors that correspond to the genetic and hormonal activity profiles that are characteristic of different cell types along A. thaliana root apical meristem. We used a Boolean model of the genetic-hormonal regulatory network to integrate known and predicted regulatory interactions into alternative models. Our analyses show that, after adding some putative missing interactions, the model includes the necessary and sufficient components and regulatory interactions to recover attractors characteristic of the root cell types, including the auxin and cytokinin activity profiles that correlate with different cellular behaviors along the root apical meristem. Furthermore, the model predicts the existence of activity configurations that could correspond to the transition domain. The model also provides a possible explanation for apparently paradoxical cellular behaviors in the root meristem. For example, how auxin may induce and at the same time inhibit WOX5 expression. According to the model proposed here the hormonal regulation of WOX5 might depend on the cell type. Our results illustrate how non-linear multi-stable qualitative network models can aid at understanding how transcriptional regulators and hormonal signaling pathways are dynamically coupled and may underlie both the acquisition of cell fate and the emergence of hormonal activity profiles that arise during complex organ development.


Table 1. Comparison of the GOF simulations with experimental evidence GOF
In silico phenotype Experimental phenotype Recovery Reference CK ARR1 is ectopically active in some of the attractors and SHY2 is ectopically active only in the provascular TD attractors.
The expression of SHY2 is enhanced after trans-zeatin treatment, but is still confined to the pro-vascular tissues of the TD. The size of the PD domain is reduced.
No such line has been analyzed in the RAM. NC --SHY2 No QC, No ARF5 activity in the PD attractors.
A SHY2 GOF line has a smaller PD than wild-type plants. SHY2   The PD domain is larger in roots where CK is degraded in the TD or in the pro-vascular. This is a multicellular phenotype that is not comparable with our single-cell simulation results.
NC [1] ARR1 ARR1 and SHY2 are not active in any attractor.
arr1 mutants have longer meristems than wild-type plants. This is a multicellular phenotype that is not comparable with our results.

NC [1]
SHY2 SHY2 is not active in some attractors. shy2-31 loss of function mutants have a larger PD of the RAM than wild-type roots. This is a multicellular phenotype not comparable with our results.
NC [2] AUXIAA The components of CK signaling pathway are not active in any of the attractors, and the components of auxin signaling are constitutively active.
There is a high degree of AUXIAA redundancy in the RAM, making it impossible to compare this results with experimental data. There is a high degree of redundancy of the ARFs in the RAM, making it impossible to compare this results with experimental data.
WOX5 is ectopically active in the central and peripheral pro-vascular PD attractors.
Roots are agravitropic because columella cells do not differentiate correctly. WOX5 is still confined in its regular position, but QC25 expression domain is slightly expanded in a MIR160 overexpression line. Quantitative variations in auxin levels along the RAM might be reason this phenotype is not comparable with our results.
NC [16,17] ARF5 No QC attractor. mp mutants do not express WOX5 since embryonic development, and endoreduplication starts early in the RAM.
PR [6,13] AUX No QC and no PD attractors. Disruption of auxin signaling in the RAM anticipate endocycle onset.
The QC, the cortex/endodermis, the pro-vascular tissues are mis-specified in the scr mutant.
R [11,13,18,19] SHR Most of the attractors are lost, with the exception of two of the Central provascular, and the Root Cap ones.
shr roots have defects in the QC specification, endodermis specification, and in the pro-vascular pattern. Ectopic metaxylem forms in place of protoxylem in the pro-vascular.
In phb-mu resistant line, PHB is not degraded by MIR165 and it is expressed ubiquitously in the RAM specifying all the pro-vascular cells as metaxylem. The loss of the QC cell type has not been described experimentally and constitutes a novel prediction of our model.
In phb loss of function mutants there is ectopic specification of the protoxylem in the pro-vascular, and the synthesis of CK in the TD is compromised. R [14] Summary of the comparison of the recovered attractors in the LOF simulations with the experimental mutant phenotypes. Abbreviations are as in Table 1.