Chitotetraose activates the fungal-dependent endosymbiotic signaling pathway in actinorhizal plant species

Mutualistic plant-microbe associations are widespread in natural ecosystems and have made major contributions throughout the evolutionary history of terrestrial plants. Amongst the most remarkable of these are the so-called root endosymbioses, resulting from the intracellular colonization of host tissues by either arbuscular mycorrhizal (AM) fungi or nitrogen-fixing bacteria that both provide key nutrients to the host in exchange for energy-rich photosynthates. Actinorhizal host plants, members of the Eurosid 1 clade, are able to associate with both AM fungi and nitrogen-fixing actinomycetes known as Frankia. Currently, little is known about the molecular signaling that allows these plants to recognize their fungal and bacterial partners. In this article, we describe the use of an in vivo Ca2+ reporter to identify symbiotic signaling responses to AM fungi in roots of both Casuarina glauca and Discaria trinervis, actinorhizal species with contrasting modes of Frankia colonization. This approach has revealed that, for both actinorhizal hosts, the short-chain chitin oligomer chitotetraose is able to mimic AM fungal exudates in activating the conserved symbiosis signaling pathway (CSSP) in epidermal root cells targeted by AM fungi. These results mirror findings in other AM host plants including legumes and the monocot rice. In addition, we show that chitotetraose is a more efficient elicitor of CSSP activation compared to AM fungal lipo-chitooligosaccharides. These findings reinforce the likely role of short-chain chitin oligomers during the initial stages of the AM association, and are discussed in relation to both our current knowledge about molecular signaling during Frankia recognition as well as the different microsymbiont root colonization mechanisms employed by actinorhizal hosts.

Introduction A limited number of soil microorganisms have acquired the remarkable capacity to colonize plant roots and form mutually beneficial endosymbiotic associations. The most widespread of these are the obligate biotrophic fungi belonging to the Glomeromycota phylum, able to associate with the majority of terrestrial plants resulting in the formation of highly ramified intracellular root cortical structures known as arbuscules [1,2]. Arbuscular mycorrhizal (AM) fungi scavenge vital soil nutrients such as phosphorus, which are then delivered to the host plant via the elaborate symbiotic interface created within arbuscule-containing cortical cells, in exchange for energy-rich metabolites. The AM association is thought to have contributed to the successful establishment of the earliest terrestrial plant species over 400 million years ago [3].
In addition to this ubiquitous fungal symbiosis, certain dicot species belonging to the Eurosid 1 clade subsequently evolved additional beneficial associations with nitrogen-fixing soil bacteria. Gram-negative rhizobia primarily associate with legume Fabales hosts, whilst grampositive filamentous actinobacteria known as Frankia are able to colonize actinorhizal host species belonging to the Fagales, Cucurbitales and Rosales orders. In the majority of cases the nitrogen-fixing bacteria multiply and differentiate intracellularly within de novo specialized plant organs known as root nodules, thereby providing a valuable source of nitrogen to the host in return for photosynthate-derived resources. Interestingly, initial root colonization by nitrogen-fixing bacteria can be either intra-or intercellular depending upon the host plant [4]. Whilst intracellular root entry via infection thread compartments constructed within root hairs has been well-documented in model legume hosts such as Medicago truncatula [5,6], intercellular modes of colonization in legumes are only poorly understood at the molecular/ cellular levels. Both types of root colonization by Frankia have also been observed for actinorhizal hosts [4,7]. For example, intracellular entry via root hairs appears to be common amongst Fagales species such as Alnus glutinosa or Casuarina glauca, whereas intercellular modes of root colonization appear to be predominant for Rosales species typified by Ceanothus spp, Elaeagnus angustifolia or Discaria trinervis [8][9][10]. In the case of initial root colonization by AM fungi, experiments performed on both legume and non-legume hosts have demonstrated that AM fungal hyphae traverse atrichoblast (non-root hair) cells of the root epidermis within host-constructed transcellular compartments by a mechanism reminiscent of root hair infection thread-mediated colonization by nitrogen-fixing bacterial symbionts [2,3].
An important parameter for the establishment of both AM and nitrogen-fixing endosymbioses is the capacity for mutual recognition of the respective partners. In particular, it is now well-established that host plant perception of appropriate secreted microbial molecules leads to the activation of a Conserved Symbiotic Signaling Pathway (CSSP) required for subsequent microbe root entry [11,12]. A characteristic feature of the CSSP is the triggering of sustained nuclear-associated calcium oscillations which can be conveniently monitored in vivo using fluorescent Ca 2+ reporters [13,14]. For the large majority of rhizobial-legume associations, decorated lipo-chitooligosaccharides (LCOs) known as nodulation (Nod) factors play this key signaling function, and these LCOs are specifically recognized by host Lysin Motif (LysM) receptor like kinase (RLK) receptor complexes [15,16]. Based on research on legumes such as M. truncatula, there is evidence that chitin-based molecules are also involved in pre-infection AM fungal-host signaling. These include both AM fungal LCOs known as Myc LCOs [17] and AM fungal short-chain chitin oligomers (referred to collectively as Myc COs) and typified by chitotetraose (CO4) [18]. Although the precise roles of these chitinaceous factors remain to be determined, recent studies focused on the monocot rice have confirmed the importance of CO4 in AM fungal-host signaling [19], and identified the LysM RLK OsCERK1 as a putative component of the host perception machinery for these fungal-secreted molecules [20].
In contrast with Rhizobium-legume associations, relatively little is known about pre-infection microsymbiont-host signaling during the establishment of actinorhizal symbioses. Although strong evidence suggests that intracellular-colonizing Frankia which nodulate A. glutinosa and C. glauca do secrete symbiotic signaling factors, the precise chemical nature of these molecules has not yet been determined. The use of root hair deformation assays, nuclear-localized Ca 2+ reporters to monitor CSSP activation and the expression of symbiotic reporter genes have together led to the conclusion that these Frankia make use of alternative non-chitinaceous signals [21][22][23]. On the other hand, it is currently unknown whether chitin-based molecules are involved in AM fungal signaling to actinorhizal hosts. In order to investigate this important question, we have made use of the in vivo Ca 2+ reporter approach to evaluate CSSP activation in epidermal root tissues in response to treatment with crude AM fungal exudates as well as potential Myc CO and Myc LCO signaling molecules. The two evolutionarily distant actinorhizal hosts, C. glauca and D. trinervis, were selected as typical examples of either intracellular or intercellular modes of Frankia root colonization. Finally, it is important to underline that, in the absence of either morphological root responses or available AM-related reporter genes for the two actinorhizal hosts, this assay is the only technique currently available for assaying CSSP activation in response to potential AM fungal signaling.
Results presented in this article show that, for both actinorhizal hosts, AM fungal exudates are efficient activators of sustained nuclear Ca 2+ spiking in root atrichoblasts, the cellular targets for AM root entry. Furthermore, sub-micromolar concentrations of CO4 were able to mimic CSSP activation for the two host plants and were significantly more efficient elicitors of Ca 2+ spiking compared to the tested Myc LCOs. Interestingly, in contrast to the observed CSSP activation in AM-targeted atrichoblasts, CO4 treatment did not elicit Ca 2+ spiking in root hairs, the epidermal cells of C. glauca which are not colonized by AM fungi. Finally, although responding robustly to AM fungal signaling with CSSP activation, we show that D. trinervis atrichoblasts are unresponsive to crude Frankia supernatants. Together, these findings underline the cellular specificity of these responses and indicate that actinorhizal host plants represent excellent models for studying endosymbiont-host signaling mechanisms for both intracellular and intercellular colonization.

Plant materials and growth conditions
Transgenic C. glauca plants expressing the nuclear-localised cameleon reporter gene p35S: NUP-YC2.1 [24] were generated via Agrobacterium tumefaciens transformation and propagated by taking cuttings as previously described [22]. Transgenic composite plants of D. trinervis (also now known as Ochetophila trinervis) expressing the p35S:NUP-YC2.1 reporter in roots were obtained using ARquaI A. rhizogenes-mediated transformation according to the protocol described in [25]. Transgenic plants were grown in liquid Broughton and Dillworth (BD) medium [26]. applied to freshly excised root segments. Cell-free supernatants were prepared from "induced" cultures of Frankia strains which nodulate either C. glauca (Frankia CcI3, referred to in this article as F. casuarinae [27]) or D. trinervis (Frankia BCU110501, referred to in this article as F. discariae [28]). For both F. casuarinae (Fci) and F. discariae (Fdi) supernatants the post-fix "i" refers to the induction of the Frankia cultures by addition of the respective host plant exudates 5 days before harvesting as described in [22,29]. These induced Frankia supernatants were applied to root segments at a 100-fold dilution.

Nuclear Ca 2+ spiking responses in C. glauca and D. trinervis roots expressing the p35S:NUP-YC2.1 reporter
Assays for nuclear Ca 2+ spiking were performed using young lateral root segments (0.5 to 0.8 cm long) freshly excised from hydroponically-grown transgenic plants expressing the p35S: NUP-YC2.1 cameleon reporter. Root explants were placed in a microchamber and treated with 150 μl of the solution to be tested as described in [18]. Confocal FRET-based ratio imaging for detecting and plotting relative changes of nuclear Ca 2+ levels corresponding to YFP-to-CFP fluorescence intensity changes over time was performed according to [24] using a Leica TCS SP2 AOBS confocal laser-scanning microscope. Imaging was performed on both atrichoblasts and trichoblasts (root hairs) in the case of C. glauca. In contrast, Ca 2+ spiking responses were only studied in D. trinervis atrichoblasts since few root hairs developed on D. trinervis roots under our growth conditions. It should be underlined that despite limited root hair formation on hydroponically-grown D. trinervis roots, efficient intercellular Frankia colonization and nodulation can be observed on these roots [e.g. 25]. Experiments were repeated several times for each treatment using independent roots (detailed in S1 Table).

Initial outer root colonization by AM fungi is intracellular for both C. glauca and D. trinervis
Late stages of AM fungal colonization have been described for both C. glauca [29][30][31][32] and D. trinervis [33]. In both cases, the presence of intracellular symbiotic arbuscules located within inner root cortical cells, accompanied by axially growing intercellular fungal mycelia is consistent with classical Arum-type root colonization [34]. Since the filamentous nitrogen-fixing Frankia microsymbiont penetrates the root outer tissues of these two actinorhizal hosts via contrasting mechanisms, it was essential to examine the initial stages of AM fungal root entry for both species. The images shown in Fig 1 and the accompanying Gif animations (S1 and S2 Figs) are consistent with AM hyphopodium formation on the root surface, followed by intracellular hyphal root penetration of the outer root tissues for the two host plants. This suggests that the AM fungus crosses the epidermal and outer cortical cell layers via a transcellular apoplastic mechanism analogous to what has already been observed for both legume and nonlegume hosts [35][36][37]. Furthermore, initial AM fungal entry for the two actinorhizal hosts occurs exclusively via epidermal atrichoblasts, again similar to earlier findings for other AM hosts [38]. Thus, despite striking differences in the modes of Frankia root colonization for C. glauca and D. trinervis, it is likely that comparable cellular mechanisms operate during initial root entry of the AM fungal partner.

AM fungal spore exudates trigger sustained Ca 2+ spiking in the root epidermis of actinorhizal host plants
Our next objective was to investigate the capacity of AM fungal signals to activate the CSSP in actinorhizal host epidermal tissues, focusing primarily on AM-targeted atrichoblast cells. For this we exploited transgenic plant tissues expressing the cameleon sensor Nup-YC2.1 in order to detect nuclear-associated Ca 2+ oscillations characteristic of CSSP activation [24]. In the case of C. glauca, stable Nup-YC2.1-expressing lines were already available as described in [22]. For D. trinervis, composite plants expressing Nup-YC2.1 in root tissues were obtained via A. rhizogenes-mediated transformation (see Materials & Methods; [25,39]).
Initially, we examined whether crude exudates of germinated AM spores were able to activate the characteristic symbiotic nuclear Ca 2+ spiking response in atrichoblasts of both actinorhizal host plants. Young lateral root segments were harvested from plants growing in liquid medium and treated with a 40x concentrated solution of a R. irregularis germinated spore exudate (GSE) as described in [18]. Changes in nuclear Ca 2+ levels within root atrichoblasts were recorded over a 20 min period. Compared with H 2 O controls (Fig 2A and 2B), GSE treatment activated sustained nuclear Ca 2+ oscillations (spiking) in 80-90% of root atrichoblasts, whether from C. glauca or D. trinervis roots (Fig 2C and 2D). Having established that the CSSP can be activated by AM fungal exudates in the D. trinervis epidermis, we then evaluated responses to a diluted supernatant from the F. discariae strain that nodulates D. trinervis (supernatants from host root exudate-induced cultures were prepared as described in [22]). Strikingly, the F. discariae (Fdi) supernatant was unable to elicit Ca 2+ spiking in the D. trinervis root (Fig 2F), despite the fact that intercellular Frankia root colonization takes place between atrichoblast cells [10,25]. This contrasts with the situation for C. glauca, where, as previously shown, supernatants from the induced F. casuarinae strain (Fci) trigger sustained spiking in C. glauca root hairs, the target cells for intracellular colonization [22] (Fig 2E).

CO4 mimics the activation of symbiotic Ca 2+ spiking by fungal GSEs in the two actinorhizal hosts
The short-chain chitin oligomer chitotetraose (CO4) is able to activate the AM-dependent CSSP of both legume and non-legume hosts at sub-micromolar concentrations [18,19]. We therefore tested the capacity of CO4 to trigger nuclear Ca 2+ spiking in root atrichoblasts of C. glauca and D. trinervis at both 10 −6 M and 10 −8 M. Data presented in Fig (3A-3D) show that both concentrations of CO4 elicit pronounced and sustained Ca 2+ spiking responses similar to those observed with AM fungal GSEs (Fig 2). Percentages of responding cells ranged from 75-90% and were marginally higher when CO4 was added at μM levels. In contrast to AM fungi, F. casuarinae initially colonizes C. glauca via root hairs, and as shown above, Frankia supernatants elicit Ca 2+ spiking in these epidermal cells (Fig 2E). For this reason, we also evaluated the reactivity of C. glauca root hairs to exogenous 10 −8 M CO4. Out of the 20 root hairs examined (from three independent root segments), not a single cell responded positively to the Myc CO elicitor (Fig 3E). These findings therefore demonstrate that chitotetraose can mimic the activation of the CSSP by AM fungal GSEs in root atrichoblasts of both actinorhizal host plants, but is not able to trigger nuclear Ca 2+ spiking in C. glauca root hairs.

Myc LCOs are less efficient elicitors of nuclear Ca 2+ spiking in actinorhizal hosts as compared to CO4
In addition to Myc COs such as CO4, other chitin-based molecules have been identified as potential AM fungal symbiotic signals. LCOs, which structurally resemble rhizobial Nod factors, are also present in AM fungal exudates, and have been shown to activate the CSSP in legume hosts [17][18][19]. Myc LCOs are present in both sulphated (S-Myc LCOs) and nonsulphated forms (NS-Myc LCOs) in exudates of R. irregularis [17]. For comparative purposes, we therefore evaluated nuclear Ca 2+ spiking initiation in both C. glauca and D. trinervis root atrichoblasts in response to either S-Myc LCOs or NS-Myc LCOs. These experiments, summarized in S1 Table and  However, these data also show that Myc LCOs are significantly less active than CO4 in triggering the CSSP in both actinorhizal hosts, and that S-Myc LCOs are even less active than NS-Myc LCOs. In the case of C. glauca, the difference between CO4 and NS-Myc LCO activity can be best appreciated at the lower 10 −8 M concentration. Note also that, since spiking was only observed in 20% of atrichoblasts at 10 −6 M S-Myc LCO, experiments were not performed at the lower concentration of 10 −8 M. Although a similar overall trend was found for both actinorhizal hosts, the ability to discriminate between CO4 and NS-Myc LCOs appears greater for D. trinervis by comparison with C. glauca (Fig 4). This difference can also be appreciated in the representative Ca 2+ spiking profiles shown in S3 Fig, where 10 −6 M NS-Myc LCO elicits a response with lower spiking periodicity in the D. trinervis epidermis as compared to C. glauca. Taken together, we conclude that  (Fig 2). (e) In contrast, 10 −8 M CO4 failed to trigger Ca 2+ spiking in C. glauca root hairs (RH). Percentages of positively responding cells are indicated for each treatment with the total number of cells monitored in brackets. Note that these figures combine all CO4 treatments (see S1 Table).
https://doi.org/10.1371/journal.pone.0223149.g003 chitotetraose is a more efficient elicitor of CSSP activation in the actinorhizal host root epidermis compared to the two types of Myc LCOs evaluated in this study.

Discussion
In this article we have investigated early symbiotic signaling between AM fungi and the two distantly-related actinorhizal host plants, C. glauca and D. trinervis, examples of either intracellular or intercellular Frankia root colonization respectively. Host perception of AM fungal signals was studied for both plants using a nuclear-localized Ca 2+ reporter to monitor the activation of the conserved symbiotic signal transduction pathway (CSSP) in epidermal root tissues. This approach has revealed that AM fungal spore exudates (GSEs) elicit sustained nuclear Ca 2+ spiking in root atrichoblasts of both C. glauca and D. trinervis (Fig 2) when applied at doses equivalent to those used previously for the legume M. truncatula [18]. This finding is coherent with the observation that, for all three host plants, AM fungal root entry is intracellular, with atrichoblasts as the principal epidermal target (Fig 1).
Earlier studies had shown that short-chain chitin oligomers such as CO4/CO5 are present in AM fungal GSEs and furthermore that 10 −8 M CO4 is sufficient to mimic GSE-elicited Ca 2+ spiking in M. truncatula roots [18]. We show here that the same concentration of CO4 is able to trigger spiking responses in root atrichoblasts of both C. glauca and D. trinervis which closely resemble those observed following GSE treatment (Fig 3). Our data also reveal that, as for M. truncatula, CO4 is a significantly more active elicitor of the CSSP in roots of both actinorhizal hosts compared to either sulphated or non-sulphated Myc LCOs (Fig 4). These findings now add two actinorhizal hosts to the growing list of AM fungal host plants for which 10 −8 M CO4 is able to elicit nuclear Ca 2+ spiking [18,19]. Not only are Myc COs active on all the AM hosts examined to date, but in all cases CSSP activation is observed in non-root hair atrichoblasts, the cellular targets for AM colonization. Furthermore, for certain hosts, such as rice [19] and C. glauca (Fig 3) this response is cell-specific by comparison with root hairs (see below). Additional direct evidence favoring the role of Myc COs as signaling molecules in the initial establishment of the AM association has come from the recent finding that AM-defective Oscerk1 rice mutants are also defective in responding to CO4 [20], thus suggesting that Myc COs are perceived by a novel rice receptor complex comprising OsCERK1 associated with a second CO-binding LysM-containing membrane protein. Future research now needs to be directed towards identifying the corresponding actinorhizal receptor components capable of recognizing these AM fungal signals and thereby triggering the downstream CSSP. Finally, since LCOs do not appear to be involved in early Frankia-host signaling in the case of C. glauca (see below), the observed low-level responses to Myc LCOs are unlikely to be due to crosstalk with a Frankia-related pathway. In consequence, we interpret this Myc LCO activity as resulting from less efficient perception by the AM-associated receptor compared to Myc COs such as CO4.
As stated earlier, initial root colonization of C. glauca by F. casuarinae occurs intracellularly via root hairs [7]. A variety of host bio-assays, including the expression of transgenic Ca 2+ reporters, have together revealed that the symbiotic factors present within the Frankia exudate are unlikely to be either LCOs or COs [22,23], and hence this raises the question of the nature of the receptors for Frankia signals in C. glauca. As expected, C. glauca root hairs are the principal cellular targets for Frankia symbiotic signals [29]. The fact that short-chain COs present in AM fungal exudates can only activate the CSSP in root atrichoblasts (Fig 3) leads us to propose that, in the case of C. glauca, receptors responding to either Frankia or AM fungal signals are distinct and specifically localized to the appropriate epidermal target cell. In contrast, the situation differs for D. trinervis, since initial Frankia root colonization occurs intercellularly between adjacent atrichoblasts [10,25]. We show that, although both AM fungal exudates and CO4 trigger nuclear Ca 2+ spiking in D. trinervis atrichoblasts, spiking is not observed in these cells in response to F. discariae supernatants (Figs 2 and 3). One possible interpretation of these findings is that pre-infection symbiotic signaling leading to host CSSP activation is absent during the initial stages of Frankia colonization of D. trinervis. This question clearly merits further investigation since there is evidence from legumes that CSSP-related signaling may not be required during rhizobial intercellular "crack entry" invasion of certain host species [40,41]. In conclusion, the results presented in this article indicate that actinorhizal host plants are particularly valuable model systems for comparative studies of endosymbiont-host signaling mechanisms associated with either intracellular or intercellular root colonization.
Supporting information S1 Fig. Animated Gif of superimposed images from a site of AM fungal colonization of a D. trinervis root corresponding to (Fig 1C and 1D).