Spatiotemporal cytokinin response imaging and ISOPENTENYLTRANSFERASE 3 function in Medicago nodule development

Abstract Most legumes can establish a symbiotic association with soil rhizobia that trigger the development of root nodules. These nodules host the rhizobia and allow them to fix nitrogen efficiently. The perception of bacterial lipo-chitooligosaccharides (LCOs) in the epidermis initiates a signaling cascade that allows rhizobial intracellular infection in the root and de-differentiation and activation of cell division that gives rise to the nodule. Thus, nodule organogenesis and rhizobial infection need to be coupled in space and time for successful nodulation. The plant hormone cytokinin (CK) contributes to the coordination of this process, acting as an essential positive regulator of nodule organogenesis. However, the temporal regulation of tissue-specific CK signaling and biosynthesis in response to LCOs or Sinorhizobium meliloti inoculation in Medicago truncatula remains poorly understood. In this study, using a fluorescence-based CK sensor (pTCSn::nls:tGFP), we performed a high-resolution tissue-specific temporal characterization of the sequential activation of CK response during root infection and nodule development in M. truncatula after inoculation with S. meliloti. Loss-of-function mutants of the CK-biosynthetic gene ISOPENTENYLTRANSFERASE 3 (IPT3) showed impairment of nodulation, suggesting that IPT3 is required for nodule development in M. truncatula. Simultaneous live imaging of pIPT3::nls:tdTOMATO and the CK sensor showed that IPT3 induction in the pericycle at the base of nodule primordium contributes to CK biosynthesis, which in turn promotes expression of positive regulators of nodule organogenesis in M. truncatula.


Introduction
Legume species acquired the capacity to interact symbiotically with rhizobium bacteria to fix atmospheric dinitrogen, allowing their growth without fertilizers on nitrogendeprived soils. This interaction involves the development of specific organs, the root nodules, that host rhizobia and provide them with carbon sources and the microenvironment required for nitrogen fixation, a reaction catalyzed by the bacterial enzyme nitrogenase. In most rhizobium-legume associations, the perception of bacterial lipochitooligosaccharides (LCOs) in the epidermis, commonly known as Nod factors, initiates a signaling cascade that is transmitted to the inner cell layers activating cell division, with simultaneous rhizobial infection of the host root. In Medicago truncatula, rhizobial infection is initiated within a curled root hair tip and the subsequent formation of a transcellular apoplastic compartment called the infection thread (IT). The IT traverses the epidermis, cortex, and ramifies within the confines of the nodule primordium, developed by organized cell divisions in the root endodermis, cortex, and pericycle (Roy et al., 2020). These coordinated mechanisms of root infection and nodule organogenesis ensure that nodule maturation occurs in perfect coordination with nodule colonization by rhizobia (Xiao et al., 2014). M. truncatula produces indeterminate nodules characterized by a longitudinal gradient of differentiation with a persistent distal apical meristem and older proximal layers (Ferguson et al., 2010). Like other developmental processes, nodulation is modulated by phytohormones (Buhian and Bensmihen, 2018).
The plant hormone cytokinin (CK) is involved in various aspects of plant growth and development. CK signaling consists of a phosphorelay mediated by a two-component system comprising a sensor and a response regulator (RR). The site of CK perception is thought to be the lumen of the endoplasmic reticulum. The CK-induced phosphorelay causes transcriptional changes in the nucleus mediated by type-B and type-A RRs, which play positive and negative roles in this regulation, respectively (Kieber and Schaller, 2018). Type-B RRs typically bind to target genes at the consensus sequence (A/G)GAT(T/C) enriched in their cis-regulatory regions. Synthetic CK sensors called two-component signaling sensors (TCS) containing concatemeric versions of this sequence have been developed for plants (Zürcher et al., 2013). CK plays essential role during nodule formation (Ferguson and Mathesius, 2014;Gamas et al., 2017). In M. truncatula, CK accumulates in the section of the root susceptible to rhizobial infection as early as 3 h after LCO treatment (van Zeijl et al., 2015). The TCS is activated by rhizobium in the cortical cells of M. truncatula that go on to form indeterminate nodules, and those in Lotus japonicus which form determinate nodules (Held et al., 2014;Jardinaud et al., 2016). In soybean (Glycine max), a regulatory feedback loop involving auxin and CK governs proper determinate nodule development (Turner et al., 2013). Rhizobia also induce the expression of CK biosynthetic and signaling genes in the epidermis of M. truncatula (Liu et al., 2015;Damiani et al., 2016;Jardinaud et al., 2016). The pMtRR9::GUS transcriptional reporter, a CK RR type-A (RRA), was rapidly detected in the root epidermis, in addition to other root tissues, in response to LCOs (Op den Camp et al., 2011). A more recently developed CK signaling sensor termed TCS new (TCSn) (Zürcher et al., 2013), driving GUS expression, enabled detection of the activation of a CK response in the M. truncatula root epidermis and the outer cortex 8 h after the LCO treatment or Sinorhizobium meliloti inoculation (Jardinaud et al., 2016). In L. japonicus, the TCS reporter was activated first in the cortex and only later in the epidermis by rhizobia (Held et al., 2014;Reid et al., 2017). The differing sequence of activation of CK responses during early symbiotic stages is likely a reflection of differences in the process of nodule development between them (Gamas et al., 2017).
CK plays an antagonistic role during root infection at the epidermis and nodule formation in the cortex (Gamas et al., 2017). The positive regulation of CK on nodule formation was first reported by physiological studies, which showed that exogenous CK induces nodule-like structures on the roots of several legumes (Heckmann et al., 2011). Further evidence for the positive role of CK in nodule inception (NIN) has come from the analysis of nodulation-defective mutants altered in CK receptors, LOTUS HISTIDINE KINASE 1 (LHK1) in L. japonicus, and CK RESPONSE 1 (CRE1) in M. truncatula (Gonzalez-Rizzo et al., 2006;Murray et al., 2007;Tirichine et al., 2007;Plet et al., 2011), and other CK receptors in both legumes (Held et al., 2014;Boivin et al., 2016). Moreover, a gain-of-function LHK1 line generates spontaneous nodules in the absence of the rhizobia (Tirichine et al., 2007;Madsen et al., 2010). Transcriptomic analyses in M. truncatula identified symbiotic genes that are rapidly induced by exogenous CK on roots (Ariel et al., 2012), such as NIN. NIN and the CRE1dependent pathways are connected by a positive feedback loop, with NIN binding to the CRE1 promoter and activating its expression (Vernié et al., 2015). Similarly, CRE1 is required for CK-induced NIN expression (Plet et al., 2011). In L. japonicus roots, exogenous CK treatment also induces NIN specifically in root cortical cells (Heckmann et al., 2011). Moreover, NIN ectopic expression leads to root cortical cell divisions and nodule-like structures in both L. japonicus and M. truncatula (Soyano et al., 2013;Vernié et al., 2015), through the activation of transcription factors, such as NUCLEAR FACTOR Y SUBUNIT A1 (NF-YA1) and B1 (NF-YB1) (Soyano et al., 2013;Laloum et al., 2014;Hossain et al., 2016;Shrestha et al., 2021). In contrast, in the epidermis of M. truncatula, CK negatively regulates root infection (Gamas et al., 2017). Indeed, depletion of the epidermal CK pool obtained by expressing a CK OXIDASE/DEHYDROGENASE enzyme under an epidermis-specific promoter caused an increased number of ITs and nodules (Jardinaud et al., 2016). Additionally, exogenous CK treatment inhibited the induction of the LCO response and pre-infection marker EARLY NODULIN 11, in a MtCRE1-dependent fashion (Jardinaud et al., 2016). Recently, a link between the epidermis-derived CK and cortical cell divisions was established (Jarzyniak et al., 2021). M. truncatula ATP-binding cassette transporter 56 (MtABCG56) transports CK from the epidermal to cortical cells, activating the CRE1-dependent CK responses, including the RRA4 (Jarzyniak et al., 2021). These downstream responses trigger further CK biosynthesis required for nodule development (Mortier et al., 2014;van Zeijl et al., 2015;Vernié et al., 2015).
Several CK biosynthesis genes, including ISOPENTENYL TRANSFERASE 3 (IPT3) and IPT1, CYP735A1, LONELY GUY 1 (LOG1), and LOG2, are upregulated in response to LCOs or during nodulation in L. japonicus and M. truncatula (Chen et al., 2014;Mortier et al., 2014;Azarakhsh et al., 2015;van Zeijl et al., 2015;Reid et al., 2017;Azarakhsh et al., 2018;Schiessl et al., 2019). The expression of MtIPT3, MtLOG1, and MtLOG2 transcriptional GUS reporters was also detected in the nodule primordium (Mortier et al., 2014;Azarakhsh et al., 2020). Decreasing LOG1 expression leads to impaired nodulation in M. truncatula (Mortier et al., 2014). All these studies highlight the importance of CK biosynthesis during root infection and nodule development. Transcriptional fusions using GUS gene reporter allowed identifying CK signaling at a tissue-specific level in M. truncatula roots during these biological processes. However, these studies have been limited in their temporal resolution. A detailed spatial and temporal characterization of the CK response in M. truncatula roots should clarify the role of this hormone in nodule induction and organogenesis.
In this study, we present the spatiotemporal regulation of CK response in rhizobia-inoculated roots using a fluorescence-based CK signaling sensor, pTCSn::nls:tGFP. To further explore the potential of this sensor, we employed it along with the transcriptional fusion of the CK biosynthetic gene IPT3 during nodule development. Simultaneous monitoring of the pIPT3::nls:tdTOMATO reporter and the CK sensor activities during nodule development suggested that IPT3 induction in the pericycle at the base of nodule primordium contributes to CK biosynthesis, which in turn promotes nodule organogenesis in M. truncatula. Furthermore, we analyzed the loss-of-function mutant of ipt3 and found that it is required for nodule development in M. truncatula.

Results
A fluorescent protein-based CK sensor is activated in root epidermal and cortical cells upon CK treatment in M. truncatula CK responses have been studied in response to LCOs and S. meliloti in M. truncatula roots, using transcriptional reporters with RRAs or the synthetic TCSn promoters fused to the GUS gene (Op den Camp et al., 2011;Plet et al., 2011;Jardinaud et al., 2016;Fonouni-Farde et al., 2017). In soybean, fluorescent protein-based auxin and CK transcriptional reporters have been successfully used to monitor and determine their cellular level ratios in root and nodule tissues (Fisher et al., 2018). In L. japonicus, a tissue-specific timecourse experiment following the activity of pTCSn::nls:GFP showed that CK response occurs in cortical cells before expanding to the epidermis (Reid et al., 2017).
In this work, we designed a fluorescent protein-based CK transcriptional reporter that addresses the limitations of the GUS reporter system. This CK sensor consists of the TCSn promoter (Zürcher et al., 2013), driving the expression of the turbo green fluorescent protein (tGFP) fused to a nuclear localization signal (nls) peptide (pTCSn::nls:tGFP). This alternative approach to the GUS reporter system allows continuous, nondestructive monitoring of CK signaling throughout plant development by live imaging, and co-imaging with other fluorescence-based reporters.
Before evaluating the CK sensor activity, we characterized the timing of CK transcriptional responses in M. truncatula roots by analyzing the expression profiles of three RRA genes, RRA3, RRA4, and RRA11, after 1, 8, 24, and 48 h of 6benzylaminopurine (6-BAP) treatment. We found that RRAs reached their maximum gene expression after 24 h of the 6-BAP treatment ( Figure 1A). Based on the time of CK signaling activation, we assessed the tissue-specific CK response using the pTCSn::nls:tGFP CK sensor in M. truncatula transgenic roots after 24 h of BAP treatment. In nontreated roots, tGFP was primarily detected in the columella, root apical meristem (RAM), and elongation zone (EZ) of the root tip, as previously described for the pTCSn::GUS reporter ( Figure  1B; Jardinaud et al., 2016;Fonouni-Farde et al., 2017). Very few cells showed nuclei-localized fluorescence in the differentiation zone (DZ) of the root ( Figure 1C), indicating the absence of CK response in the DZ in nontreated roots. In contrast, roots treated with 6-BAP for 24 h exhibited a strong signal of nuclei-localized fluorescence in the epidermis and cortex in the DZ of the root (Figure 1, D and E). 4 0 ,6-diamidino-2-phenylindole (DAPI) counterstain confirmed that tGFP fluorescence was localized to the nuclei (Supplemental Figure S1). These observations indicate that the pTCSn::nls:tGFP sensor constitutes a suitable molecular tool to investigate the CK response in M. truncatula and that CK signal transduction occurs in M. truncatula epidermal and cortical cells after the application of CK to the root.
Tissue-specific time-course of CK signaling during the early symbiotic interaction with S. meliloti in M. truncatula roots Characterization of CK response during indeterminate nodule development remains limited to a few time points during the process by using RRA promoters fused to GUS reporters in M. truncatula (Op den Camp et al., 2011;Plet et al., 2011). To obtain a high spatiotemporal resolution of CK response during indeterminate nodule development in M. truncatula, we analyzed the activity of pTCSn::nls:tGFP in a time-course experiment using transgenic roots after S. meliloti inoculation. Prior to inoculation, very low pTCSn::nls:tGFP activity was detected in the cell layers of the susceptibility zone (SZ; Figure 2A). At 4 hours after inoculation (hai), pTCSn::nls:tGFP activity started in the epidermal Figure 1 A reporter of CK signaling based on the pTCSn::nls:tGFP transcriptional fusion is activated in the epidermal and cortical cells after CK treatment. A, RT-qPCR analyses of RRAs gene expression after 1, 8, 24, and 48 h of 1 mM 6-BAP or mock treatment. The Student's t test was performed, and asterisks represent statistically significant differences between 6-BAP and mock treatments in each time point. *P 5 0.05, **P 5 0.01. Values are the means ± SE of fold-changes of two biological replicates (n = 2). B, pTCSn::nls:tGFP activity in the root tip of nontreated M. truncatula transgenic root. The tGFP signal from the nuclei of the RAM and EZ is shown in green. In magenta, the signal emitted by cell wall polysaccharides bound to Calcofluor white M2R is shown. C, pTCSn::nls:tGFP activity in the DZ of nontreated transgenic root and (D, E) in DZ of 1 mM 6-BAP treated transgenic root after 24 h in the epidermis (ep) and cortex (co). D, tGFP signal (green) and (E) merged image showing tGFP and Calcofluor white stained (magenta) signals. Scale bar: 100 mm (B and C) and 50 mm (D and E). cells of the SZ, indicating that rhizobia-induced CK response occurs very early in the epidermis of M. truncatula ( Figure 2B). At 24 hai, nuclei-localized fluorescence was still observed in the epidermal cells but was also present in outer cortical cells of the SZ ( Figure 2C) and by 48 hai, strong fluorescence was widespread in the outer and inner cortical cell layers of the SZ ( Figure 2D). Thus, CK signaling is activated first in the epidermis, reaches the outer cortical cells within 24 hai and extends to most cortical cell layers within 48 hai.

Tissue-specific time-course of CK signaling during the indeterminate nodule development in M. truncatula
We also monitored the pTCSn::nls:tGFP activity throughout nodule development from the first cell divisions in the pericycle, the endodermis, and the cortex, until the mature nodule formation. Moreover, we associate the tissue-specific pTCSn::nls:tGFP activity time-course with the sequential cell division program characterized previously during nodule formation (Xiao et al., 2014).
At 3 d after inoculation (dai), the pTCSn::nls:tGFP signal that was widely distributed across the cortical cell layers of the SZ disappears ( Figure 2D), giving rise to a robust and more localized signal at the pericycle and dividing cortical cell layers C3-C5. These cells are related with the nodule primordium initiation at the developmental Stages II and III ( Figure 2E; Xiao et al., 2014). This nodule primordiumspecific pattern of the CK response allowed us to clearly distinguish a nodule primordium from a lateral root primordium (LRP) during their early developmental stages, where tGFP expression was weaker and limited to the vasculature and developing meristem (Supplemental Figure S2). At 4 dai, we found that the Stages IV and V nodule primordia showed the CK signaling activation extending to most of the dividing cortical cell layers ( Figure 2F). At 5 dai, the nodule primordium emerges from the main root and becomes a true nodule when the meristem starts functioning (Stage VI). At this point, the CK response was localized to the C3, the C4-to C5-derived cells that form the multi-layered nodule meristem (C3) and the nonmeristem zone immediately below (C4 and C5), respectively ( Figure 2G). At 6 dai, the nodule is in an advanced developmental stage, with the vascular bundles starting to surround the nodule meristem. At this stage, the CK signaling was strongly activated in the central zone of the nodule, including the nodule meristem and the C4/5-derived cells that will be colonized by rhizobia ( Figure 2H).

CK biosynthesis by IPT3 is required for nodule development in M. truncatula
CK and its downstream responses are critical regulators of nodule initiation and development. However, the molecular mechanism of the local CK biosynthesis during nodule organogenesis remains poorly characterized in M. truncatula. It has been proposed that KNOX3 directly activates the transcription of CK biosynthesis genes, IPT3, LOG1, and LOG2, to promote CK biosynthesis during nodule organogenesis  (Azarakhsh et al., 2015(Azarakhsh et al., , 2020. However, the genetic characterization of CK biosynthesis genes in M. truncatula remains limited to LOG1 (Mortier et al., 2014). IPT3 expression is induced at 72 hai in the root, reaching a maximum at 5 dai (Schiessl et al., 2019;Supplemental Figure S3). Moreover, the pIPT3::GUS reporter showed that IPT3 is expressed in the nodule primordium of M. truncatula (Azarakhsh et al., 2020). These observations indicate that IPT3 represents an excellent candidate to investigate the role of CK biosynthesis during nodule organogenesis.
We identified three M. truncatula Tnt1 lines showing an insertion in the single exon of IPT3 (Tadege et al., 2008). These mutants were named ipt3-1; ipt3-2, and ipt3-3 ( Figure 3A). Isolation of homozygous individuals of ipt3 was confirmed by reverse transcription quantitative polymerase chain reaction (RT-qPCR) analyses ( Figure 3B). Then, ipt3 mutants and wild-type plants were used to perform nodulation assays. At 14 dai, all the ipt3 mutants showed a significantly lower nodule number than the control (Figure 3, C and D). Moreover, between 40% and 55% of the ipt3 mutants did not develop any nodule, compared to 15% in the wild-type background ( Figure 3E). Root dry weight measurements revealed no significant differences between wild-type plants and ipt3 mutants after 14 dai, indicating that the reduction in the number of nodules is not due to defects in root growth (Supplemental Figure S4A), while shoot dry weight significantly decreased only in ipt3-1 (Supplemental Figure S4B). Cross-sections of wild-type and ipt3 mutant nodules showed normal structure and colonization at 14 dai (Supplemental Figure S5A). At 21 dai, the number of nodules was reduced in all three mutant backgrounds (Supplemental Figure S5B). These nodules developed normally, and there was no difference in nitrogen fixation in ipt3 mutant nodules compared with wild-type plants (Supplemental Figure S5C). To validate that the mutation in IPT3 is causing the defects in nodulation phenotype, we complemented the ipt3-3 mutant with the functional IPT3 driven by its native promoter (pIPT3::IPT3). At 14 dai, the number of nodules in the ipt3-3 roots transformed with the empty vector (EV) was significantly lower than the wild-type transformed with the EV. At the same time, no significant differences were observed between ipt3-3 roots transformed with the pIPT3::IPT3 construct and wild-type transformed with the EV (Figure 3, F and G), confirming that functional IPT3 rescues the nodulation defects observed in the ipt3-3 background and showing that CK biosynthesis by IPT3 is required for nodulation in M. truncatula.
IPT3 expression is induced in the stele at the base of the developing nodule primordium To obtain further insights into the role of IPT3 during nodule organogenesis, we investigated IPT3 expression by monitoring the accumulation of the tdTOMATO fluorescent protein driven by the IPT3 promoter in a time-course experiment. This strategy allowed us to monitor IPT3 expression and CK signaling simultaneously after the inoculation with S. meliloti through live imaging in M. truncatula transgenic roots. To achieve this goal, we cloned the IPT3 promoter in frame with the tdTOMATO fluorescent protein fused to a nls peptide, resulting in the pIPT3::nls:tdTOMATO construct. Before inoculation with rhizobia, IPT3 expression and CK response overlapped at the root stele (Supplemental Figure S6, A-C). At 24 h after S. meliloti inoculation, IPT3 expression was similar to the control treatment and still localized in the vasculature (Supplemental Figure S6, D-F). In contrast, the CK response was observed in the epidermis and outer cortical cells of the SZ, as described above (Supplemental Figure S6, G-I). At 2 dai, CK response expanded to most of root cortical layers of the SZ, while pIPT3::nls:tdTOMATO activity was still mainly found in the stele with a similar expression level to that of the control and 24 hai (Figure 4, A-C; Supplemental Figure S6, E and H). These results suggest that IPT3 is not involved in the activation of CK signaling during the early symbiotic interaction (48 hai).
At 3 dai, nodule primordium was initiated, and CK response was mainly localized to the dividing cortical cells and stele ( Figure 4D). IPT3 expression was strongly induced in the stele at the base of dividing cortical cells of the nodule primordium ( Figure 4E). This result is consistent with prior reports of the induction of the IPT3 72 h after S. meliloti inoculation (Supplemental Figure S3; Schiessl et al., 2019). The induction of the CK response and IPT3 expression overlapped at the root vasculature and the base of the nodule primordium ( Figure 4F). At 4 dai, the CK response extended to more cortical cell layers of the developing nodule primordium ( Figure 4G), whereas IPT3 expression levels remained high but localized in the stele below the dividing cortical cells ( Figure 4H), showing lower overlap with CK signaling in the stele ( Figure 4I). At 6 dai, the CK response was mainly localized at the central zone of the nodule ( Figure 4J), while IPT3 was expressed in the stele at the base of the nodule ( Figure 4K). At this stage, the overlap between the CK response and IPT3 expression declined, with each one showing a specific spatial pattern ( Figure 4L). At 7 dai, the CK response was mainly localized to the apical part of the nodule ( Figure 4M) and IPT3 expression was detected in the developing nodule vasculature ( Figure 4N), in line with the pIPT3::GUS activity in nodule vascular bundles recently reported (Azarakhsh et al., 2020). At this stage, the CK response was restricted to the nodule meristem, and IPT3 expression was localized at the base of the nodule and vascular bundles ( Figure 4O). Together, these results suggest that the IPT3 induction at the base of the nodule primordium contributes to the biosynthesis of CK, which in turn triggers CK signaling during nodule organogenesis.
IPT3 expression is activated in the pericycle and pericycle-derived cells during the indeterminate nodule development To precisely characterize the cell type-specific activation of IPT3 expression, we monitored the pIPT3::nls:tdTOMATO activity in cross-sections of roots inoculated with S. meliloti. pIPT3::nls:tdTOMATO activity was induced at 3 dai, based on the observation of a higher fluorescence signal of the tdTOMATO protein compared with the control and roots at 2 dai, confirming the transcriptional activation of IPT3 at 3 dai described above ( Figure 5, A-C). The observation of the fluorescence in cross-sections indicated that IPT3 is induced in the pericycle at 3 dai ( Figure 5, D-F). Later in nodule development, pIPT3::nls:tdTOMATO activity is localized at the base of nodule primordia, in the stele ( Figure 5, G-I). Cross-sections of nodule primordium revealed that IPT3 expression was restricted to the pericycle and the pericycle-dividing cells at 4 dai ( Figure 5J). At 5 and 6 dai, pIPT3::nls:tdTOMATO activity was still found in the pericycle, but also in the new cell layers likely derived from the pericycle-dividing cells ( Figure 5, K and L). At 8 and 9 dai, pIPT3::nls:tdTOMATO activity was detected in the cells surrounding the developing vascular bundles of nodules ( Figure 5, M-O). We also explored the pIPT3::nls:tdTOMATO , and NF4651 Tnt1 lines have one insertion in the single exon region (black) flanked by 5 0 -and 3 0 -untranslated regions (gray) of the gene and were renamed ipt3-1, ipt3-2, and ipt3-3, respectively. Black bar, 100 bp. B, RT-qPCR analysis of IPT3 gene expression in R108 genotype and ipt3 homozygous mutants. Values indicate means ± SE for three biological replicates (n = 3). P-values were calculated by analysis of variance (ANOVA) followed by Tukey's post hoc testing. Groups of significant difference (P 5 0.05) are indicated with different letters. C, Representative image of 3week-old R108 and ipt3 mutant plants at 14 dai with S. meliloti 1021. Scale bar: 3 cm. D, Nodule number in R108 and ipt3 mutants at 14 dai. Boxplots center line, median; the box extends from the 25th to 75th percentiles; whiskers, 1.5 interquartile range; points out of the whiskers, outliers. Statistical analysis was performed using ANOVA followed by Tukey's post hoc testing. Groups of significant difference (P 5 0.05) are indicated with different letters. E, Percentage of wild-type (n = 19), ipt3-1 (n = 15), ipt3-2 (n = 18), and ipt3-3 (n = 18) mutant plants showing 1 or more nodules or none at 14 dai. F, Number of nodules in transgenic roots of EV/R108, EV/ipt3-3, and pIPT3::IPT3/ipt3-3 mutant at 14 dai. Boxplots center line, median; the box extends from the 25-75th percentiles; whiskers, 1.5Â interquartile range; points out of the whiskers, outliers. Statistical analysis was performed using ANOVA followed by Tukey's post hoc test. Groups of significant difference (P 5 0.05) are indicated with different letters. G, Representative nodules of EV/R108 (n = 28), EV/ipt3-3 (n = 29), and pIPT3::IPT3/ipt3-3 (n = 18). Blue color indicates that the nodules are infected (upper) and tdTOMATO fluorescence is used to identify the transformed roots (lower). Scale bar: 1 mm.  activity during lateral root development. We found that IPT3 expression is restricted to the stele of the primary root during the formation of LRP with a similar fluorescence intensity to the one observed in the stele of the primary root with no LRP (Supplemental Figure S7, A and B). In a fully emerged lateral root, pIPT3::nls:tdTOMATO activity was detected in the newly formed stele (Supplemental Figure  S7C). Cross-sections showed that IPT3 is expressed in the pericycle and the cells surrounding the developing vasculature of a recently formed lateral root (Supplemental Figure  S7, D and E). Accordingly, IPT3 expression is activated after 72 h during lateral root formation, when the lateral root fully emerges and the vasculature is already visible, and not in LRP (Supplemental Figure S7F; Schiessl et al., 2019). These results show that IPT3 gene expression induction is restricted to the pericycle and the pericycle-derived cells during the indeterminate nodule development of M. truncatula.

IPT3 is required for the induction of symbiotic genes during nodule initiation
Nodule initiation is dependent on crucial regulators that promote cortical cell division, including the CK receptor CRE1, the transcription factor NIN, and its targets LBD16 and NF-YA1 (Gonzalez- Rizzo et al., 2006;Plet et al., 2011;Laporte et al., 2014;Vernié et al., 2015;Schiessl et al., 2019). The ipt3 mutants show impairment of nodule development ( Figure 3D), suggesting that the biosynthesis of CK precursor by IPT3 is required to activate CK-induced positive regulators of nodule development. To test this hypothesis, we compared expression of these essential regulatory genes between the wild-type and two ipt3 mutant lines, ipt3-2 and ipt3-3. The ipt3-1 mutant was excluded due to the previously mentioned dwarf phenotype. Three-day-old M. truncatula plants were inoculated with S. meliloti, and the SZ of the root was harvested at 4 dai for gene expression analyses. We found that expression of NIN, LBD16, and NF-YA1 was upregulated in wild-type plants compared to noninoculated control plants. In contrast, in the ipt3 mutants, these genes were not significantly induced (Figure 6), indicating that the transcriptional activation of positive regulators of nodulation requires IPT3 at 4 dai. CYCLINA;3 (CYCA;3), a CK-induced gene and a cell division marker during nodule initiation (Schiessl et al., 2019), did not show significant induction either in the wild-type plants or ipt3 mutants with respect to the control treatment ( Figure 6). Besides CRE1, we found significant transcriptional induction of CK signaling genes, such as RRA3 and RRA11, which was affected in ipt3 mutants compared to the control (Figure 6), indicating IPT3 is required for the activation of CK signaling during nodule initiation. On the contrary, RRA4 was not induced in wild-type and ipt3 mutants by rhizobia compared to the control ( Figure 6). IPT3 was upregulated in wild-type plants at 4 dai with respect to the control (Figure 6), consistent with the rhizobia-induced expression pattern observed in previous work and with the visual reporter (Supplemental Figure S3; Schiessl et al., 2019; Figure 5). To better evaluate the impact of IPT3 on CK biosynthesis and signaling during nodulation, we examined pTCSn::nls:tGFP activity in wild-type and ipt3-3 mutant roots, after S. meliloti inoculation. As expected, the ipt3-3 mutant showed lower number of nodules. However, the spatiotemporal characterization of CK sensor showed that CK signaling follows the same pattern in wild-type and ipt3-3 mutant roots (Supplemental Figure S8), indicating that normal CK signaling occurs in the ipt3-3. These results suggested that rhizobia-induced IPT3 expression contributes to CK biosynthesis, which promotes transcriptional activation of positive nodule development regulators. However, the similar CK signaling pattern in wild-type and ipt3-3 mutant nodules suggests that other IPT genes are induced by rhizobia (Supplemental Figure S3) and can semi-redundantly participate in CK biosynthesis during nodule development. Figure 6 Rhizobia-dependent induction of nodulation regulators and CK signaling genes is impaired in ipt3 loss-of-function mutants. RT-qPCR analyses of nodulation regulators (NIN, LBD16, and NF-YA1), cell cycle (CYCA;3), and CK signaling genes (CRE1, RRA3, RRA4, and RRA11) after mock treatment or at 4 dai in wild-type, ipt3-2, and ipt3-3 mutants. Values indicate means ± SE of fold-changes of three biological replicates (n = 3). P-values were calculated by ANOVA followed by Tukey's post hoc test. Groups of different significance (P 5 0.05) are indicated with different letters.

Discussion
In this study, we developed a tGFP-based TCSn reporter that allowed us to perform a live imaging tissue-specific time course, with a high temporal resolution of rhizobia-induced CK signaling. This reporter system allowed us to gain further insights into the CK signaling induction during rhizobia perception and nodule formation in M. truncatula. Our data indicate that the activation of CK signaling occurs in multiple discrete stages, initially in the epidermis of the root SZ and expanding across the cortex during the first 48 h. After 48 h, this widespread CK signaling activation disappears, giving way to the second wave of CK signaling activation in the cortex, localized in the specific cells that will give rise to the nodule primordium. The characterization of ipt3 mutants suggests that IPT3 contributes to the CK biosynthesis that triggers this second wave of signaling activation.
It has been proposed recently that the MtABCG56 transporter, which is transcriptionally induced between 6 and 24 h after the LCO treatment, exports bioactive CKs from the root epidermis to the cortex, promoting CRE1dependent cortical CK responses (Jarzyniak et al., 2021). In agreement with this model, we observed that CK signaling was activated at 24 hai in the outer cortical cells ( Figure 2C), and it extends to most of cortical cell layers at 48 hai in the SZ ( Figure 2D). Thus, MtABCG56-dependent CK transport from epidermis to cortex may contribute to the first wave of CK signaling activation in the cortex (Jarzyniak et al., 2021; Figure 7). These results are consistent with CK activation patterns previously observed using pTCSn::GUS, showing GUS activity localized in the epidermis and outer cortical cells at 8 hai and the inner cortical cells at 72 hai, together with in situ hybridization data which detected expression of RRA4 widely in the root cortex at 48 hai (Vernié et al., 2015;Jardinaud et al., 2016). Here, a live imaging time course allowed us to precisely elucidate the timing of the CK signaling activation pattern observed during the early symbiosis interaction. In contrast, in L. japonicus, CK signaling activation in cortical cells precedes epidermal CK responses (Held et al., 2014;Reid et al., 2017). These results suggest that the spatiotemporal CK signaling activation may differ between determinate and indeterminate nodulating species. The difference in rhizobia-triggered CK signaling patterns between these species highlights the need for high-resolution tissue-specific characterization of the CK responses in other legumes.
The second wave of CK signaling activation in the cortex requires de novo CK biosynthesis. It has been reported that CK biosynthesis genes, such as LOG1, LOG2, and IPT3, are expressed in the nodule primordium of M. truncatula and that their expression is promoted by KNOX3 (Mortier et al., 2014;Azarakhsh et al., 2020). By monitoring the pIPT3::nls:tdTOMATO and the CK sensor simultaneously, we resolved the spatiotemporal pattern of IPT3 expression and its interaction with the CK signaling during the indeterminate nodule development in M. truncatula. We found that IPT3 is induced in the stele and adjacent cells to the first dividing cortical cells at 3 dai ( Figure 4E), overlapping with CK signaling activation at the base of the nodule primordium ( Figure 4F). Cross-sections of inoculated roots and developing nodules showed that IPT3 is induced in the pericycle and the pericycle-derived cells (Figure 5), suggesting that CK biosynthesis in the pericycle by IPT3 promotes cell division. This is unlike prior reports of pIPT3::GUS activity in the central zone of the nodule primordium 3 d postinoculation (Azarakhsh et al., 2020). This discrepancy may be explained by a higher sensitivity of the GUS reporter, which could reveal the promoter activity even with very low expression levels. This outcome might also be derived from the diffusion of the GUS reaction product to adjacent cells, especially when extended incubation periods are required. The ipt3-3 mutant complementation with the pIPT3::IPT3 construct suggests the promoter region chosen in this study contains the regulatory elements required for the proper expression of IPT3. Similar to what was reported with pIPT3::GUS 7-to 12-d postinoculation (Azarakhsh et al., 2020), we found that IPT3 is expressed in those cells surrounding the vascular bundles of the nodules (Figure 5, J-L). Our exploration of pIPT3::nls:IPT3 activity during lateral root development indicates that IPT3 expression remains restricted to the stele during LRP formation and only later is activated in the newly formed stele of a fully emerged lateral root (Supplemental Figure S7). Cross-sections of young lateral roots showed that IPT3 is expressed in the pericycle and developing vasculature (Supplemental Figure S7). Thus, IPT3 is transcriptionally activated during nodule primordium development but not in LRP. Accordingly, CK responses appear higher in a nodule primordium compared to an LRP (Supplemental Figure S2). The identity of a nodule may be determined by a higher CK/auxin ratio in the nodule primordium than an LRP, as it was observed in soybean (Fisher et al., 2018). IPT3-specific activation in the pericycle may contribute to this higher CK/auxin ratio and consequently determine nodule identity.
IPT enzymes catalyze the formation of iP riboside 5 0 -diphosphate or iP riboside 5 0 -triphosphate, which are precursors for bioactive CK biosynthesis by LOG family enzymes (Sakakibara, 2006). In M. truncatula, IPT genes belong to a small gene family with six members, MtIPT1-5 and MtIPT9, whose gene expression is altered by S. meliloti inoculation (Supplemental Figure S3; Schiessl et al., 2019). As CK signaling differentially regulates both root colonization and nodule formation, it is expected that CK biosynthesis genes present different tissue-specific expression patterns in response to rhizobia. For example, IPT2 and LOG3 are induced in the epidermis after the LCO treatment or S. meliloti inoculation (Jardinaud et al., 2016;Jarzyniak et al., 2021), whereas LOG1 and LOG2 are expressed in the central zone of developing nodule primordia, overlapping with our observations of the CK signaling activation (Mortier et al., 2014;Azarakhsh et al., 2020). Functional characterization of the other IPTs and their cell-type expression analysis will be critical to understanding the complete landscape of the CK biosynthesis during nodulation. Our observations suggest that the activation of IPT3 in the pericycle at 3 dai, likely in cooperation with other IPT genes (Supplemental Figure S3), provides the substrate for LOG family enzymes to synthesize bioactive CKs which promote nodule primordium development (Kurakawa et al., 2007). Accordingly, LOG1 RNAi plants show lower number of nodules than control plants in M. truncatula (Mortier et al., 2014), similarly to the ipt3 loss of function mutants. The same effect has been reported in L. japonicus IPT3 RNAi plants, which produced fewer ITs and nodules than wild-type (Chen et al., 2014).
It has been proposed that LjIPT3 also participates in the generation of shoot-derived CK precursors, which are involved in the autoinhibition of nodulation (AON) mechanism in L. japonicus (Sasaki et al., 2014). Whether IPT3 participates in AON in M. truncatula remains an open question. IPT3 expression is activated in the shoot at 7 dai in M. truncatula in a SUPER NUMERARY NODULES (SUNN)-dependent manner (Azarakhsh et al., 2018). We found that rhizobia-dependent IPT3 activation in the root occurs at 3 dai in the pericycle ( Figure 5F), coinciding with the timing of AON induction (Kassaw et al., 2015). These findings suggest that IPT3-derived CKs may be involved in different mechanisms in shoots and roots. The different timing of IPT3 activation in different organs could explain its dual role during nodulation in L. japonicus and M. truncatula.
Rhizobia-dependent transcriptional activation of CKinduced positive regulators of nodule development, such as NIN, CRE1, LBD16, and NF-YA1, was affected in ipt3 mutants at 4 dai ( Figure 6). However, the tissue-specific temporal characterization of the CK sensor in ipt3-3 mutant reveals that CK signaling occurs in the epidermis, cortex, and developing nodules (Supplemental Figure S8). This observation indicates that the nodules formed in ipt3 mutants present the same CK signaling activation as the wild-type plants. IPT1, which shows a marked transcriptional activation in response to rhizobia inoculation (Supplemental Figure S3), may provide the substrate to promote CK production and consequently CK signaling activation in developing nodules of ipt3-3 mutant. RT-qPCR analyses of CK signaling genes showed a substantial overall reduction of gene expression ( Figure 6), according to the lower number of nodules in ipt3 mutants ( Figure 3).
Together, these results suggested that de novo CK precursor synthesis is required for the CK-mediated induction of key nodule development regulators and the proper nodule development. We propose a high-resolution model for the tissue-specific temporal activation pattern of CK signaling during indeterminate nodule development in M. truncatula (Figure 7). Furthermore, we show that IPT3 contributes to CK biosynthesis, which in turn promotes the expression of positive regulators of nodule development, such as NIN and CRE1.

Plant material and growth conditions
Medicago truncatula R108 is the wild-type background for experiments involving the Tnt1 mutant lines; Jemalong A17 was used for all other assays. Seeds were scarified for 8 min in sulfuric acid and sterilized for 4 min in bleach (12% [v/v] sodium hypochlorite). After rinsing with sterilized water, seeds were sown on 1% (w/v) agar plates supplemented with 1 lM gibberellic acid and stored at 4 C for 3 d before incubating overnight at 24 C in the dark. Germinated seedlings were transferred to square plates (22.5 Â 22.5 cm) containing modified Fahräeus medium (Boisson-Dernier et al., 2001) supplemented with 15 mM NH 4 NO 3 and grown vertically at 24 C under long-day conditions in a growth chamber (16-h light/8-h dark; 150 lmol m -2 s -1 light intensity).

Genotyping of Tnt1 insertion lines
Medicago truncatula R108 Tnt1 transposon insertion lines utilized in this research project, which are jointly owned by the Centre National De La Recherche Scientifique, were obtained from the Noble Research Institute. Three different Tnt1 transposon insertion lines, namely NF5762 (ipt3-1), NF3757 (ipt3-2), and NF4651 (ipt3-3), were genotyped by PCR using Tnt1-specific primers (Cheng et al., 2011) combined with MtIPT3 gene-specific primers encompassing the insertions (Supplemental Table S2). The expression of MtIPT3 in homozygous plants was tested by RT-qPCR to confirm they were knockout mutants of the MtIPT3 gene (Supplemental Table S2).

Complementation of ipt3-3 mutant
One-day-old wild-type and ipt3-3 seedlings were transformed with Agrobacterium rhizogenes MSU440, as previously described (Boisson-Dernier et al., 2001). Three-weekold composite plants showing transgenic roots were transferred to growth pouches (https://mega-international.com/ tech-info/) containing Modified Nodulation Medium (Chakraborty et al., 2021). The plants were acclimated for a week and then inoculated with phemA::lacZ S. meliloti 1021 (Leong et al., 1985). The nutrient medium was replenished every week. Two weeks after inoculation, transformed roots expressing p35S::ER:tdTOM::tNOS were selected based on the presence of the tdTOMATO fluorescence and stained for lacZ (5 mM potassium ferrocyanide, 5 mM potassium ferricyanide, and 0.08% (w/v) X-gal in 0.1 M PIPES, pH 7) for 4 h at 37 C. The total nodule number from transgenic roots per plant was scored under an Olympus MVX10 fluorescence stereo microscope.
Agrobacterium rhizogenes-mediated transformation and in vitro nodulation assay The constructs described above were introduced into A. rhizogenes MSU440 electrocompetent cells and used to generate transgenic roots in M. truncatula (Boisson-Dernier et al., 2001). The transgenic roots were selected based on the fluorescence emitted by the pTCSn::nls:tGFP construct at the root tip, visualized under an Olympus MVX10 fluorescence stereo microscope. For the in vitro nodulation assay timecourse experiment, 3-week-old transgenic roots were transferred to Buffered Nodulation Medium (Ehrhardt et al., 1992) supplemented with 0.1 mM AVG (aminoethoxyvinyl glycine hydrochloride; Sigma-Aldrich, St Louis, MO, USA) to reduce ethylene production. After 5 d of acclimation, transgenic roots were treated with a suspension of S. meliloti 1021 (OD 600 = 0.02) supplemented with 3 mM of luteolin to activate LCO production (Sigma-Aldrich) and transgenic roots were collected at different timepoints. To characterize pTCSn::nls:tGFP activity in ipt3-3 mutant, 3-week-old transgenic roots were grown in pouches as described above. After S. meliloti 1,021 inoculation, transgenic roots were harvested at different timepoints for live imaging.

In vitro CK treatment and S. meliloti inoculation
Three-day-old M. truncatula seedlings were transferred to Fahräeus medium supplemented with water (mock treatment) or 1 mM of 6-BAP (Sigma-Aldrich) and maintained under the same growth conditions. Five roots from different seedlings were collected at 1, 8, 24, and 48 h after CK treatment and immediately frozen in liquid nitrogen for RNA extraction. To analyze pTCSn::nls:tGFP activity after CK treatment, M. truncatula transgenic roots were submerged in a solution of 1 mM 6-BAP for 5 min. After 24 h, transgenic roots harboring the pTCSn::nls:tGFP construct were harvested and used for microscopy analysis. For RT-qPCR studies of nodule development regulators and CK signaling genes in wild-type and ipt3 mutants, germinated seedlings were transferred to nitrogen-free Fahräeus medium and grown under the same conditions described above. Threeday-old seedlings were inoculated alongside the root with 200 mL/root of S. meliloti 1021 (OD 600 = 0.02) resuspended in liquid nitrogen-free modified Fahräeus medium or with Fahräeus medium (mock treatment). Four days after inoculation, root segments from the SZ (3 cm sections from 2 cm above the root tip) were harvested and pooled (10 plants per sample) for RNA extraction.

Gene expression analysis
RNA extraction was performed as previously described (Chang et al., 1993). RNA samples were digested with DNase I and purified using the RNA Clean & Concentrator-5 kit following the manufacturer's instructions (Zymo Research, Irvine, CA, USA). RT-qPCR analyses were performed using the Luna Universal Probe One-Step RT-qPCR Kit (New England Biolabs) following the manufacturer's instructions using 200 ng of total RNA. The EF1a gene was used as a housekeeping gene and the average of two technical replicates was obtained to calculate relative gene expression (DDCt method). Primers used for RT-qPCR experiments are listed in Supplemental Table S2.

Microscopy live imaging
Before microscopy imaging, transgenic roots were placed in a staining solution with 0.1% (w/v) Calcofluor white M2R (Sigma-Aldrich) in PBS 1X (Corning) for 10 min at room temperature and then rinsed with PBS 1X before imaging. A Leica TCS SP5 confocal microscope was used for live imaging. For cross-sectioning, roots and nodules were submerged in a freshly prepared solution of 4% (w/v) formaldehyde in PBS 1X, vacuuming three times for 5 min followed by overnight fixation at 4 C. The fixed roots and nodules were embedded in 6% (w/v) agarose and sectioned (50 lm of thickness) in PBS 1X using a Leica VT1000S vibratome. A confocal microscope Leica TCS SP5 was used for live imaging. At least ten different roots for treatment or time points were analyzed. Images were acquired by a HCX PL APO CS 20.0X0.70 dry and HCX PL APO lambda blue 63.0X1.40 Oil. tGFP was excited with 488-nm laser (Argon) set at 80% of intensity and the emission was collected at 497-557 nm. tdTOMATO was excited with 514-nm (Argon) and 543-nm lasers (HeNe 543) set at 80% of intensity and the emission was collected at 570-643 nm. Calcofluor white M2R was excited with 405-nm Diode laser set at 40% of intensity and the emission was collected at 450-480 nm. Fluorescent signals are presented with green (tGFP), red (tdTOMATO), and magenta or cyan (calcofluor white M2R). For DNA staining, transgenic roots were incubated with 5 lg/mL DAPI in PBS 1Â for 10 min at room temperature and then rinsed with PBS 1X before imaging. DAPI was excited with 405-nm Diode laser set at 40% of intensity and emission was collected at 450-480 nm. The confocal gain and pinhole were set to 700 and 60 mm, respectively, and the images were acquired at 1,024 Â 1,024 pixels resolution.

In vivo nodulation assay
For in vivo nodulation assay, seedlings were germinated as described above. Then, plants were grown in pots (9 Â 9 Â 9 Â cm) containing pre-sterilized calcined clay, Turface (Profile Products, Buffalo Grove, IL, USA), and sand (2:2 v/v), where Turface was placed at the bottom and on the top of a layer of sand. Plants were watered with modified Fahräeus medium supplemented with 0.5 mM of NH 4 NO 3 and covered with a lid. After 1 week of acclimation, Fahräeus medium supplemented with 0.5 mM of NH 4 NO 3 was removed entirely from the tray and replaced with a nitrogen-free modified Fahräeus medium. Plants were treated by pouring 10 mL of S. meliloti 1,021 suspension (OD 600 = 0.02) into each pot. Plants were watered using nitrogen-free modified Fahräeus medium every 2-3 d. After 2 weeks of inoculation, nodule number was assessed by inspecting plant roots under an Olympus MVX10 fluorescence stereo microscope. After counting nodules, roots and shoots for each individual were separated and dried in paper bags. After 48h of drying at 65 C, root and shoot dry weight were recorded.

Nodule sectioning
Plants were grown in growth pouches as described above and treated with phemA::lacZ S. meliloti 1021 (Leong et al., 1985). After 2 weeks, nodules were stained for lacZ as described above overnight at 37 C. Next, nodules were fixed in 4% (w/v) formaldehyde in PBS 1X (Corning), by vacuuming three times for 5 min followed by 4 h at 4 C. The fixed nodules were embedded in 6% (w/v) agarose and sectioned (50 lm of thickness) in PBS 1X using a Leica VT1000S vibratome. Images of nodule sections were taken with a Zeiss Axioplan 2 microscope attached to a QImaging Retiga EXi Fast 1394 camera.

Acetylene reduction assay
Seedlings were germinated as described above then grown in pots (6 Â 2.5 Â 2.5 cm) containing 4:1 pre-sterilized turface and sand. After 5 d of growth, 5 mL of phemA::lacZ S. meliloti 1021 (OD 600 = 0.02) was added to each pot. Plants were fertilized weekly with 5 mL of Plant Prod 0-15-40 medium (nitrogen free, pH 6.8; Master PlantProd Inc.; Brampton, Ontario, Canada). Three weeks after inoculation, plants were gently removed from the substrate without washing, and an acetylene reduction assay was performed as described previously (Oke and Long, 1999). After this, plants were stained for lacZ as described above to visualize infected nodules, and nodule number counted under a stereo microscope.

Accession numbers
Accesion numbers of the genes studied in this work can be found in the Medicago Genome version 4.0 available in Phytozome database, as follows: EF1a (Medtr1g101870), IPT1

Supplemental data
The following materials are available in the online version of this article.
Supplemental Figure S3. IPT expression levels in a timecourse experiment after S. meliloiti inoculation in M. truncatula roots.
Supplemental Figure S4. Root and shoot dry weight measurements of wild-type and ipt3 mutants at 14 dai.
Supplemental Figure S5. Nodules of ipt3 mutants show normal rhizobial colonization and nitrogen fixation capacity.
Supplemental Figure S6. IPT3 expression is not induced after 24 h of S. meliloti inoculation and is localized in the stele of M. truncatula root.
Supplemental Figure S7. IPT3 is expressed in the stele of LRP and lateral root of M. truncatula.
Supplemental Figure S8. Rhizobia-dependent CK signaling activation is still occurring in epidermis, cortex, and developing nodules of ipt3-3 mutant.
Supplemental Table S1. List of plasmids used in this study.
Supplemental Table S2. List of primers used in this study.