Plant Cyclic Nucleotide-Gated Channels: New Insights on Their Functions and Regulation1[OPEN]

Calcium (Ca21) signaling is crucial for all aspects of plant physiology, including defense, abiotic stress responses, and development; recent research has elucidated the role of plant cyclic-nucleotide–gated channels (CNGCs) in Ca21 signaling and downstream processes. CNGCs belong to the superfamily of voltagegated ion channels. Like voltage-gated K1 channels, animal CNGCs, and hyperpolarization and cyclic nucleotide-regulated channels, plant CNGCs are tetrameric and have six transmembrane domains, with a cytosolic N-terminal (NT) and C-terminal (CT) region per subunit (Jegla et al., 2018). The first plant CNGC isoforms were identified as calmodulin (CaM)-binding proteins in 1998 (Köhler and Neuhaus, 1998; Schuurink et al., 1998); over the past five years, pioneering work has established CNGCs as Ca21-permeable channels involved inCa21 oscillations andpossibly receptor-mediated signaling. The spatio-temporal variation in cytosolic Ca21 concentrations affects a wide range of cellular responses (Webb et al., 1996). For example, Ca21 flux across the plasma membrane is an early signaling step in establishing symbiosis and immunity (Zipfel and Oldroyd, 2017). Moreover, Ca21 affects many developmental processes: repetitive spiking or oscillations in cytosolic Ca21 concentration entrain circadian rhythms, underlie polar expansion of root hairs and pollen tubes, occur in response to the application of auxin to elongating root cells, and control stomatal movements in response to CO2 and abscisic acid (Felle, 1988; McAinsh et al., 1995; Holdaway-Clarke et al., 1997; Allen et al., 2001; Love et al., 2004; Monshausen et al., 2008). The production of Ca21 oscillations requires positive and negative feedback regulation, and theoretical modeling of Ca21 oscillations in plants has been successfully applied to some model systems (Martins et al., 2013; Liu et al., 2019). However, understanding Ca21 dynamics on the molecular and quantitative levels in plants has been hampered by lack of knowledge about the

Calcium (Ca 21 ) signaling is crucial for all aspects of plant physiology, including defense, abiotic stress responses, and development; recent research has elucidated the role of plant cyclic-nucleotide-gated channels (CNGCs) in Ca 21 signaling and downstream processes. CNGCs belong to the superfamily of voltagegated ion channels. Like voltage-gated K 1 channels, animal CNGCs, and hyperpolarization and cyclic nucleotide-regulated channels, plant CNGCs are tetrameric and have six transmembrane domains, with a cytosolic N-terminal (NT) and C-terminal (CT) region per subunit (Jegla et al., 2018). The first plant CNGC isoforms were identified as calmodulin (CaM)-binding proteins in 1998 (Köhler and Neuhaus, 1998;Schuurink et al., 1998); over the past five years, pioneering work has established CNGCs as Ca 21 -permeable channels involved in Ca 21 oscillations and possibly receptor-mediated signaling.
The spatio-temporal variation in cytosolic Ca 21 concentrations affects a wide range of cellular responses (Webb et al., 1996). For example, Ca 21 flux across the plasma membrane is an early signaling step in establishing symbiosis and immunity (Zipfel and Oldroyd, 2017). Moreover, Ca 21 affects many developmental processes: repetitive spiking or oscillations in cytosolic Ca 21 concentration entrain circadian rhythms, underlie polar expansion of root hairs and pollen tubes, occur in response to the application of auxin to elongating root cells, and control stomatal movements in response to CO 2 and abscisic acid (Felle, 1988;McAinsh et al., 1995;Holdaway-Clarke et al., 1997;Allen et al., 2001;Love et al., 2004;Monshausen et al., 2008). The production of Ca 21 oscillations requires positive and negative feedback regulation, and theoretical modeling of Ca 21 oscillations in plants has been successfully applied to some model systems (Martins et al., 2013;Liu et al., 2019). However, understanding Ca 21 dynamics on the molecular and quantitative levels in plants has been hampered by lack of knowledge about the molecular nature and regulation of the channels that allow Ca 21 entry.
In this Update, we summarize recent advances in physiological, biochemical, and electrophysiological characterization of CNGCs, giving new insight into the molecular functions and regulation of plant CNGCs, focusing on subunit assembly, phosphorylation, and CaM binding.

ARE CNGCS TRULY "CYCLIC-NUCLEOTIDE-GATED" CHANNELS?
Progress in understanding the assembly, activation, and regulation of plant CNGCs has been slow. This may be due in part to the pronounced differences to their animal counterparts: In contrast to early assumptions that CNGCs were non-selectively permeable to cations (Talke et al., 2003), new research shows that several CNGCs conduct Ca 21 but often do not allow K 1 to cross. Table 1 summarizes our current knowledge about the regulation by cyclic nucleotide monophosphates (cNMPs) and CaM of distinct CNGC subunits expressed in heterologous expression systems, such as Xenopus laevis oocytes and human embryonic kidney (HEK) cells.
Controversial findings regarding the requirement of elevated cyclic adenosine monophosphate (cAMP) or cyclic guanosine monophosphate (cGMP) levels indicate that our current assumptions about CNGC gating may require major revision in the near future. As an example, Arabidopsis (Arabidopsis thaliana) CNGC2 was among the first cloned family members (Köhler and Neuhaus 1998). Initial electrophysiological characterization in X. laevis oocytes suggested that CNGC2 is a voltage-dependent K 1 -permeable channel activated by cAMP or cGMP (Table 1; Leng et al., 1999Leng et al., , 2002. In comparison, CNGC4, which together with CNGC2 forms subgroup B of clade IV of Arabidopsis CNGCs (Mäser et al., 2001), was reported to encode a voltage-independent cAMP-and cGMP-gated channel (Balagué et al., 2003). Recent work suggests that at least in X. laevis oocytes, both channels work as a hyperpolarization-activated calcium-permeable channel in a heteromeric assembly, without requirement for the elevation of cNMPs levels .
Similarly, patch-clamp recordings on plasma membranes of plant cell protoplasts have detected cNMPdependent stimulation of hyperpolarization-activated Ca 21 -permeable channels, and these could in some cases be attributed to distinct CNGC isoforms, as in case of e.g. CNGC5 and CNGC6 in Arabidopsis guard cells (Wang et al., 2013). However, depending on the expression system and experimental condition, there seems to be no absolute requirement for the elevation of cNMP levels above the resting state (Table 1), and cNMP affinities and binding dynamics have in most cases not been well studied. The usage of genetically encoded reporters for cAMP or cGMP (Isner and Maathuis, 2013;Jiang et al., 2017) and precise biochemical analysis may provide more definite answers regarding the physiological importance of cNMPs to activate CNGCs in vivo. In light of recent advances toward the characterization of phosphorylation and CaM binding as gating agents or modifiers, the regulation by cNMPs will have to be reevaluated. cNMPs might act only on a subset of CNGC subunits, they may act as a cofactor rather than a true trigger for ligand-activation, or they may modify the voltage-dependence or affinities toward other regulators. In addition, several recent reports showed that the universal calcium sensor protein CaM plays a more complex and significant role in regulating CNGCs than previously thought (see below for details). Therefore, despite the significant progress in recent years, the permeability, functional regulation, and nature of ligands of plant CNGCs still need further studies.

REGULATION OF CNGC ACTIVATION AND TURNOVER BY PHOSPHORYLATION
In animals, phosphorylation is one way to regulate CNG and HCN channels (Kaupp and Seifert, 2002;Herrmann et al., 2015). For example, the vertebrate CNGCs CNGA1 and CNGB2 function as hetero-tetrameric channels in rod photoreceptors and the phosphorylation status of Tyr residues in these channels controls their activity (Molokanova et al., 2003). Likewise, Tyr phosphorylation alters the gating of the HCN4 pacemaker channel (Li et al., 2008). Early pharmacological studies showed that protein kinase inhibitors prevent the activity of hyperpolarization-dependent calcium channels in plant cells (Köhler and Blatt 2002;Stoelzle et al., 2003), indicating that protein phosphorylation plays a critical role in stimulus-specific Ca 21 signaling.
In recent years, regulation via direct phosphorylation by Ca 21 -dependent protein kinases (CDPKs/CPKs), has been documented for a number of plant ion channels, including the K 1 channel KAT1, SLOW ANION CHANNEL-ASSOCIATED1, and TWO-PORE CHAN-NEL1 (Geiger et al., 2009;Maierhofer et al., 2014;Ronzier et al., 2014;Kintzer and Stroud, 2016;Bender et al., 2018). In an extensive survey of CPK substrates in Arabidopsis, Curran et al. (2011) identified CNGC6, CNGC7, CNGC9, and CNGC18 as potential targets of CPK1, CPK10, or CPK34. So far, a specific CPK-CNGC interaction has only been shown for the kinase domain of CPK32 and CNGC18 by yeast two-hybrid assays (Y2H) and Förster resonance energy transfer analysis (Table 2; Zhou et al., 2014). Coexpression of the constitutively active form of CPK32 in X. laevis oocytes strongly enhanced CNGC18 channel activity, although actual phosphorylation was not shown and the phosphorylation sites in CNGC18 were not identified . Positive regulation of CNGCs by CDPKs opens the possibility that an initial Ca 21 influx may precede activation of CNGCs by CDPKs. In this scenario, CNGCs may amplify or modify a Ca 21 response initiated by a different channel or from an internal calcium store, because some CDPK activation requires   elevated Ca 21 concentration. So far, negative regulation of CNGC activity by CDPKs has not been reported, but is possible. In any case, this notion supports the idea that CNGCs are part of larger protein complexes that include other channels, pumps, and decoders such as CDPKs, or are localized in proximity to these players. This notion is discussed in detail later in this article ( Fig. 1) Considering the plasma membrane localization of most plant CNGCs, receptor kinases, or receptor-like kinases (RLKs) are likely candidates for the kinases that phosphorylate CNGCs. Indeed, Ladwig et al. (2015) reported that CNGC17 binds to the Arabidopsis H 1 -ATPases (AHA), AHA1 and AHA2, as well as to BRASSINOSTEROID INSENSITIVE1-ASSOCI-ATED RECEPTOR KINASE1 (BAK1). BAK1 is a Leurich repeat RLK (LRR-RLK), which can associate with various pattern recognition receptors (PRRs) as a coreceptor, forming functional receptor complexes that regulate a wide variety of physiological responses from growth to immunity (Kim and Wang, 2010;Ranf, 2017;Liang and Zhou, 2018). The growth-regulating phytosulfokine (PSK) receptor PSKR1, another LRR-RLK superfamily member, also binds to AHA1, AHA2, and BAK1, suggesting that CNGC17, PSKR1, BAK1, and AHAs may form a protein nanocluster to initiate downstream signals ( Fig. 1; Ladwig et al., 2015). In addition, these interaction data suggest that BAK1 or other LRR-RLKs phosphorylate plant CNGCs.
In 2019, three studies revealed that LRR-RLKs and related kinases phosphorylate CNGCs, and examined the physiological relevance of this phosphorylation (Tian et  Arabidopsis. An increase of the cytosolic Ca 21 concentration ([Ca 21 ] cyt ) is essential for the oxidative burst after recognition of PAMPs/MAMPs such as the bacterial elicitor peptide flagellin22 (flg22) or fungal chitin (Seybold et al., 2014;Kadota et al., 2015). Upon flg22 recognition by the PRR kinase FLAGELLIN SENSING2 (FLS2), a receptor complex consisting of FLS2, BAK1, and BOTRYTIS-INDUCED KINASE1 (BIK1) forms, leading to transphosphorylation of these kinases and release of BIK1, which activates downstream signaling (Couto and Zipfel, 2016). A well-studied downstream target of BIK1 is the membrane-localized NADP oxidase (NADPH), RESPIRATORY BURST OXIDASE HOMOLOG D, which is responsible for the oxidative burst during PAMP-triggered immunity (PTI; Li et al., 2014;Kadota et al., 2015). Tian et al. (2019) reported that CNGC4, but not CNGC2, also interacts with BIK1, and is phosphorylated at the CT cytosolic domain upon flg22 recognition via FLS2 (Fig. 1). BIK1 is a cytoplasmic kinase that is a central component of PRR complexes with components such as FLS2, EF-TU RECEPTOR, PERCEPTION OF THE ARABIDOPSIS DANGER SIGNAL PEPTIDES, and CHITIN ELICITOR RECEP-TOR KINASE1 (Couto and Zipfel, 2016). CNGC2 and CNGC4 interact and form a functional heteromeric channel, which is inhibited in the presence of CaM (Chin et al., 2013;Tian et al., 2019). BIK1 can activate this CNGC2-CNGC4 heteromeric channel in the presence of the inhibitory CaM, possibly via phosphorylation of CNGC4, and was therefore suggested to induce CNGC2-CNGC4-mediated Ca 21 influx in response to PAMP recognition. Tian et al. (2019) showed CNGC2 (also known as DEFENSE NO DEATH1 [DND1]) and CNGC4 (also known as DND2/HYPERSENSITIVE RESPONSE-LIKE LESION MIMIC1) are positive Figure 1. Model of a CNGC-containing signal complex (nanodomain/channelosome, shown in darker gray). A heterotetrameric CNGC channel is part of a sensing receptor complex containing PRRs, their clients (e.g. BIK1, RESPIRATORY BURST OXIDASE HOMOLOG D, RLCK), various pumps (e.g. proton ATPase and Ca 21 pumps), and decoders (e.g. CPKs and CaM). The formation of such a signal complex can be permanent or temporal upon recognition of specific stimuli (transient signaling complex) and the combination of specific players can contribute to generate precise spatiotemporal Ca 21 signals. Recruitment of CNGCs in a specific signaling complex may be achieved by MLO proteins. Phosphorylation plays significant roles to activate CNGCs or induce their turnover by E3 ubiquitin ligases and the 26s proteasome. V, Vesicle.
regulators of PTI only under specific calcium concentrations (i.e. 1.5 mM [Ca 21 ] ext ), as their null mutants showed reduced PTI under this condition, but behaved like wild type under lower calcium concentrations (i.e. 0.1 mM [Ca 21 ] ext ). They reported that sufficient [Ca 21 ] ext is essential to activate calcium-dependent PTI. However, both cngc2 and cngc4 single null mutants are hypersensitive to calcium and have pleiotropic phenotypes (Yu et al., 1998;Clough et al., 2000;Chan et al., 2003;Wang et al., 2017). Furthermore, cngc2 (dnd1) mutants experience Ca 21 stress under normal Ca 21 levels (Chan et al., 2008), raising the possibility that they cannot respond normally to many triggers, including PAMPs. Therefore, future studies should clarify which channels mediate the Ca 21 response under low Ca 21 supply in cngc2 (dnd1) mutants and whether the compromised PTI in these mutants is a result or independent of other pleiotropic phenotypes.
In another recent study, Wang et al. (2019) examined the role of CNGC phosphorylation in rice (Oryza sativa) PTI and programmed cell death (PCD). The null mutants for CNGC2 (dnd1) and CNGC4 (dnd2/hypersensitive response-like lesion mimic1) show complex and contradictory phenotypes such as autoimmune phenotypes with constitutive elevation of salicylic acid levels and expression of pathogenesis-related (PR) genes, but reduced PCD in the hypersensitive response (Yu et al., 1998;Clough et al., 2000;Moeder et al., 2011). In the absence of pathogens, Arabidopsis cngc2 and cngc4 mutants also show conditional spontaneous lesions. A very similar lesion mimic phenotype was observed for the barley null mutant of CNGC4, necrotic leaf spot1 (Rostoks et al., 2006), and the rice mutant, cell death and susceptible1, which lacks a functional OsCNGC9 gene . The rice cell death and susceptible1 mutant shows impaired blast fungus resistance and reduced calcium influx, oxidative burst, and PTI-related gene expression, indicating that OsCNGC9 has a significant role in PTI . Furthermore, the rice receptor-like cytoplasmic kinase (RLCK) OsRLCK185 physically interacts with and phosphorylates OsCNGC9 and this phosphorylation activates channel function (Tables 1 and 2; Wang et al., 2019). OsRLCK185 and OsRLCK176 interact with the chitin receptor OsCHITIN ELICITOR RECEPTOR KINASE1 (Miya et al., 2007). Therefore, CNGCs represent a common downstream target of phosphorylation during PTI. Because the closest Arabidopsis homolog of OsCNGC9 is CNGC14, which has been associated with Ca 21 entry during auxin-regulated growth (Shih et al., 2015;Zhang et al., 2017;Dindas et al., 2018;Brost et al., 2019), and not CNGC2 or 4, it will be interesting to investigate the functional diversification of CNGCs in different plant species.
In the third new study from 2019, Yu et al. (2019) used a suppressor screen to identify an Arabidopsis CNGC as a key player in the mediation of cellular homeostasis, which is regulated by BAK1 and its closest homolog, SOMATIC EMBRYOGENESIS RECEPTOR-LIKE KINASE4 (SERK4). Like CNGC2 and CNGC4, the two closely related group-I channels CNGC11 and CNGC12 have been implicated in immunity and PCD (Yoshioka et al., 2006;Moeder et al., 2011). Likewise, a role in immunity, wound signaling, and insect resistance has been proposed for CNGC19 and CNGC20, which comprise subclade IVa in the CNGC family (Moeder et al., 2011;Meena et al., 2019). BAK1 and SERK4 are involved in a wide variety of physiological phenomena and the bak1-4 serk4-1 double mutant develops severe postembryonic lethality due to hyperactivation of PCD (He et al., 2007;de Oliveira et al., 2016). Yu et al. (2019) conducted a nonbiased suppressor screen of cell death in RNA interference-BAK1/ SERK4-silenced plants and found that a knockout mutant of CNGC20 suppressed this cell death phenotype. Furthermore, they showed that BAK1 phosphorylates the CT cytosolic domain of CNGC20 (Table 2). These sites are conserved in CNGC19, which can also be phosphorylated by BAK1. This breakthrough study revealed a novel mechanism of CNGC regulation in which phosphorylation of the C terminus regulates CNGC19 and CNGC20 protein stability. According to their model, CNGC19 and CNGC20 form a heterotetrameric channel that positively regulates cell death, and BAK1/SERK4 phosphorylation accelerate CNGC19/20 turnover to maintain cellular homeostasis (Fig. 1).
These three recent studies showed that phosphorylation of CNGCs plays key roles in their regulation (Table 1). One important question will be whether phosphorylation itself activates/inactivates the channel or whether a conformational change will alter the accessibility or sensitivity for cNMPs or CaM. Further investigation of other CNGCs may identify common patterns and will tell us whether each CNGC subunit undergoes unique regulation of either activity or turnover by phosphorylation.

CNGC GATING BY THE CALCIUM-SENSOR PROTEIN CAM AND ITS ROLE IN SHAPING CA 21 OSCILLATIONS
The partial overlap of cNMP-binding domains and the CaM binding site suggested that these signaling compounds competitively regulate CNGCs (Arazi et al., 2000;Hua et al., 2003b;Kaplan et al., 2007). However, a second CaM-binding domain adjacent to the cNMP-binding domain, which is formed by an Ile-Gln (IQ) motif and is conserved among CNGCs (Fischer et al., 2013), challenged this model. Moreover, additional CaM binding sites were mapped within the NT cytosolic domain and distal of the IQ-domain, and both positive and negative regulation of CNGCs by CaM was shown for CNGC12 ( Fig. 2; Chin et al., 2010;DeFalco et al., 2016). The finding that the IQ motif binds CaM in a Ca 21 -independent manner and is required for channel function paved the way for revisiting the gating mechanisms of CNGCs, because CaM could serve as a built-in subunit that senses local changes in Ca 21 concentration upon channel opening, and induces rapid, Ca 21 -dependent feedback regulation (DeFalco et al., 2016;Fischer et al., 2017;Demidchik et al., 2018). However, additional data are required, especially about the role of cNMPs and their possible interplay with CaM, to gain a better understanding of the gating mechanisms of CNGCs.

Negative Regulation by CaM Binding
In X. laevis oocytes, CaM binding to the C terminus of CNGC14 and to the CNGC2/CNGC4 complex inhibits ion channel function Zeb et al., 2020). Unfortunately, it is not clear which CaM binding site is involved and whether this inhibition required an elevation of [Ca 21 ] cyt . Interestingly, in both of the abovementioned studies, CaM7 functioned as a negative regulator (Fig. 2) and CaM7, but not CaM2, specifically inhibited currents through CNGC14 (Zeb et al., 2020). The mechanism by which CaMs exert their isoform-specific regulation remains unclear, because all Arabidopsis CaM isoforms interacted with the C terminus of CNGC14 and CNGC6 in Y2H assays (Fischer et al., 2017).
The CaM2.1 and CaM7.1 isoforms have identical protein sequences except for one conserved K to R change, while the CaM2.2 splice variant used by Zeb et al. (2020) contains 12 additional residues. CaM is highly conserved across kingdoms, which poses questions about the expression levels and physiological relevance of the extended splice variant described by Zeb et al. (2020). If the negative regulation of CNGCs depends on Ca 21 -loading of the respective CaM, this will assist in shaping the Ca 21 signature in vivo. In the case of CNGC14, this will relate to Ca 21 oscillations in root hairs and auxin-dependent growth (Shih et al., 2015;Dindas et al., 2018;Brost et al., 2019).
In vivo, Ca 21 -dependent binding of CaM to the NT binding site of CNGC12 negatively regulates its channel activity (DeFalco et al., 2016). Ectopic expression of CNGC12 with a mutated NT domain, which cannot bind Ca 21 -CaM, constitutively induced PCD, similar to the phenotype produced by a constitutively active channel. Therefore, CaM may regulate channel activity via binding to the NT and CT domains. As channel activity is dependent on heteromeric subunit assembly Tian et al., 2019), one urgent task in Ca 21 signaling research is determining the stoichiometry of natural channel assemblies, including their associated CaM subunit(s).

Positive Regulation by CaM Binding
CNGC12 contains three CaM-binding sites: the NT-CaMBD and the CT-CaMBD (which interact with Ca 21 -CaM), and the IQ-CaMBD (which associates with apo-CaM; DeFalco et al., 2016). A mutation in a CT motif that had been shown to be crucial for CaM binding (Arazi et al., 2000) resulted in a loss-of-function of CNGC12, indicating CaM binding to this site positively regulates CNGC function Chin et al., 2010). Mutation of the core IQ sequence to DA disrupts the interaction with CaM and channel function. When CNGC11/12, a chimeric channel composed of the NT half of CNGC11 and the CT half of CNGC12, is expressed in Nicotiana benthamiana leaves, PCD is induced by constitutively activated Ca 21 flux (Yoshioka et al., 2006. By contrast, expression of the CNGC11/12 DA mutant does not induce PCD (DeFalco et al., 2016). If only the Ca 21 -dependent interaction of  . In B and C, the pore is closed and the CNBD is separated from the membrane via the C-linker, to illustrate the closed state.
CaM with the CNGC12 IQ domain was disrupted, channel function could be partially retained, suggesting that the IQ domain-calcium-free CaM (apo-CaM) complex supports channel function (Fischer et al., 2017). This conclusion was further substantiated by heterologous expression of CNGC11 and CNGC12 in X. laevis oocytes. CNGC12-mediated hyperpolarization-dependent Ca 21 currents were enhanced by ;3-fold upon coexpression with CaM1 or apo-CaM1, which was kept in the apo state by mutating all four Ca 21 -binding sites . In comparison, CNGC11 was inactive as a channel in X. laevis oocytes, both in the presence and absence of CaM . Only CaM1, which is identical to CaM4, was able to activate CNGC12 in X. laevis oocytes, but CaM6 was not . This again points to isoform-specific CaM functions, despite the ability of CaM2 and CaM6 to bind to the C terminus as well as to the isolated IQ domain of CNGC12 in yeast (Fischer et al., 2017). The CaM1 and CaM6 protein sequences (protein models CaM1.1 and CaM6.1) differ in five positions with conserved exchanges (E/D; K/R; T/S; I/V); this led us to question how such subtle differences in protein sequence produce the observed functional differences.
Furthermore, interaction of apo-CaM to CNGCs suggest the concept that CaM may function as a built-in Ca 21 sensor of CNGCs. CaM has two lobes (C and N) with two EF-hands each connected by a flexible linker. Both lobes bind Ca 21 with different affinities, which contributes to the ability of CaM to regulate many target proteins (Villarroel et al., 2014). Apo-CaM can interact with the IQ domain of CNGCs via its C-lobe, indicating that apo-CaM attaches to CNGCs in the resting state and plays a role as a Ca 21 -sensing subunit for the channel complex (Fischer et al., 2017). Indeed, apo-CaM association is required for Ca 21 sensing and for channel opening, at least in some channels such as CNGC12 (DeFalco et al., 2016;Fischer et al., 2017;Zhang et al., 2019) and CNGC8/CNGC18 heteromers ( Fig. 2; Pan et al., 2019), where the channel-CaM complex may support sustained Ca 21 oscillations during pollen tube growth. As both proximal and distal regions of the core IQ motif play critical roles in CaM accommodation (Fischer et al., 2017), the observation that some CNGCs are activated by CaM (probably by binding of apo-CaM to their IQ domain), while others are not, poses new questions about the complexity of the interaction of CaM with different CNGC subunits and heteromeric CNGC complexes (Fig. 2). Therefore, more quantitative and dynamic analyses of CaM isoform-specific interactions with CNGCs are required to improve our understanding of any CaM-induced gating mechanism.

Role in Shaping Ca 21 Oscillations
Many recent publications have shed light on the role of CNGCs as central elements of plant Ca 21 oscillators. This is not unexpected, because their intrinsic CaM-binding properties make CNGCs function as Ca 21 -feedbackregulated elements (Fig. 2).
In pollen tubes, CNGC18 is essential for guidance and tip growth (Frietsch et al., 2007;Gao et al., 2016), while CNGC7 and CNGC8 have partially redundant functions in controlling pollen tube growth (Tunc-Ozdemir et al., 2013). In different heterologous expression systems, CNGC18 mediates hyperpolarization-activated calcium currents, but the regulation of channel activities appears to be complex (Table 1). In HEK293T cells, addition of cAMP and cGMP activated CNGC7, CNGC8, and CNGC18 to produce inward calcium currents at hyperpolarized potentials (Gao et al., 2016). In another report, CNGC18 expressed in X. laevis oocytes could be activated by coexpression of a constitutively active form of the Ca 21 -dependent protein kinase CPK32 . The authors therefore suggested that Ca 21 -dependent feed-forward stimulation of calcium entry occurs via CPK32 during Ca 21 oscillations in growing pollen tubes . In later experiments, CNGC18 was highly active in X. laevis oocytes in the absence of plant kinases and without addition of membrane-permeable cyclic nucleotides , leaving us to question the impact of CNGC18 regulation by CPK32 and cyclic nucleotides in pollen tubes.
A recent study presented a novel mechanism for regulation by heteromeric channel assembly and CaM in the absence of elevated levels of cyclic nucleotides (Table 1; Pan et al., 2019). In X. laevis oocytes, CNGC18 currents were inhibited by coexpression of CNGC7 or CNGC8, and this inhibition was relieved in the presence of CaM2. By contrast, CNGC7 and CNGC8 produced nonfunctioning homomeric channels in the presence and absence of CaM2. Biochemical studies revealed that the CNGC C-termini interacted with each other and with apo-CaM2 or Ca 21 -CaM2. Ca 21 loading of CaM2 lowered the affinity for the CNGC8 and CNGC18 C-termini from 50 to .800 nM, suggesting that Ca 21 induced the dissociation of CaM2 from the heteromeric channel complex, which leads to channel inactivation. In this scenario, the heteromeric CNGC18-CNGC8 complex would be active at low [Ca 21 ] cyt , when apo-CaM is associated, but a rise in [Ca 21 ] cyt would trigger CaM release and channel closure. Pan et al. (2019) thus suggest a new model in which the dissociation of Ca 21 -CaM2 induces inhibition of the channel complex (Fig. 2).
This type of Ca 21 feedback regulation perfectly meets the theoretical expectations for the situation in growing pollen tubes. However, no oscillatory calcium current (or free-running membrane potential) was measured in oocytes, where the "oscillator" had been reconstituted. Despite the presence of high extracellular Ca 21 concentrations of 30 mM, current amplitudes in the presence of CaM2 with nonfunctional EF-hands (CaM2 1234 ) were only ;20% higher than those with Ca 21 -sensitive CaM2. The study by Pan et al. (2019) provides new and essential data for future modeling of the Ca 21 oscillator, if the suggested mechanism can be validated in vivo.
Modeling of Ca 21 oscillations could then also integrate knowledge about feedback-control by membrane voltage, as well as on-and off-rates of CaM binding.
Loss of CNGC14 causes root hair defects, including swelling and branching, as well as bursting of the root hair tip Brost et al., 2019), indicating its role in the regulation of cell integrity during polar growth. CNGC5, CNGC6, and CNGC9 also contribute to the robustness of unidirectional cell expansion and stability of cytosolic Ca 21 oscillations (Brost et al., 2019;Tan et al., 2020). Loss of CNGC14 had the strongest effect by destabilizing the calcium oscillations and inducing growth defects. The typical Ca 21 -oscillation period of ;30 s found in wild-type root hairs was not established in the cngc14 cngc6 and cngc14 cngc9 double mutants (Brost et al., 2019). However, this period was still present in cngc6 cngc9 double mutants and cngc9 single mutants, although with much less robustness, identifying CNGC14 as the major pacemaker in vivo. Finally, under the experimental conditions used, the cngc6 cngc9 cngc14 triple mutants initiated root hair bulges, which rapidly burst after transition to the rapid growth phase. In another study, growth defects of the cngc5 cngc6 cngc9 triple mutant could be complemented by overexpression of each of the CNGC subunits, indicating similar functions of these channels (Tan et al., 2020). Similar to the results for CNGC14, heterologous expression of CNGC5-, CNGC6-, or CNGC9-induced hyperpolarization-activated Ca 21 currents in HEK293T cells, although the role of cyclic nucleotides required for channel activation differs between individual studies (Table 1; Gao et al., 2016;Tan et al., 2020). In addition to cytosolic Ca 21 oscillations, the participation of CNGCs in nuclear Ca 21 oscillations has also been reported (Charpentier et al., 2016;Leitão et al., 2019). CNGC15 homologs from Medicago truncatula and Arabidopsis are the only CNGCs so far that are localized to the nuclear envelope, where they participate in nuclear Ca 21 oscillations, which are crucial for root growth and symbiosis establishment (Charpentier et al., 2016;Leitão et al., 2019).

CNGC HETERO-TETRAMERIZATION AND LOCALIZATION TO MEMBRANE NANODOMAINS
Ca 21 signals participate in many physiological responses; therefore, one important question is how specific stimuli generate unique signals to maintain signaling specificity. The rates of Ca 21 entry and export, Ca 21 buffering and binding to target proteins, and the respective reaction volumes determine the shape of the "Ca 21 signature" (Clapham, 2007;McAinsh and Pittman, 2009). Hetero-tetramerization of plant CNGCs thus provides a versatile tool to generate unique patterns of Ca 21 signatures. Based on the examples described above, it is reasonable to hypothesize that each subunit has a unique mode of regulation by phosphorylation and CaM binding, but also a certain degree of functional redundancy.
Another possible mechanism for creating unique Ca 21 signatures is the formation of protein complexes at the plasma membrane acting as specific sensing modules (Fig. 1). This idea is supported by the observation of interactions of CNGCs with receptor kinases and other membrane-localized proteins as discussed above (Table 2; Ladwig et al., 2015;Wang et al., 2019;Yu et al., 2019;Meng et al., 2020). Following the membrane raft hypothesis proposed by Simons and Ikonen (1997), subcompartmentalization of plasma membrane proteins in nanodomains or microdomains may produce signaling hubs that give specificity in plant signaling (Keinath et al., 2010;Demir et al., 2013;Jaillais and Ott, 2020). For example, the FLS2 coreceptor BAK1 is also a coreceptor of the major brassinosteroid (BR) receptor BRI1. Upon sensing BR, BRI1 forms an active receptor complex with BAK1, thereby initiating BR signaling (Kim and Wang, 2010). Interestingly, Wang et al. (2015) showed that BRI1 localizes to membrane nanodomains and that this partitioning of BRI1 is essential for proper BR signal transduction. Furthermore, Bücherl et al. (2017) showed that FLS2 and BRI1 localize to distinct plasma membrane nanodomains and such spatiotemporal separation of two receptor kinases could contribute to their signaling specificity in immunity and growth regulation. Thus, it is plausible to hypothesize that CNGC hetero-tetrameric channels are parts of sensing complexes, together with specific receptors and downstream decoder proteins, to create specific downstream outputs (Fig. 1).

CONCLUDING REMARKS
As discussed above, recent studies have substantially enriched the field of CNGC research (see Advances). As expected, these new data and concepts raise further questions for deepening our understanding of this channel group and its role in plant calcium signaling (see Outstanding Questions).
The interaction of CNGCs with receptor-like kinases and other membrane-localized proteins, as recently reported for Mildew Locus O (MLO) proteins (Meng et al., 2020), would allow specific CNGCs together with CaMs to be part of different nanodomains associated with their respective receptors. These membrane domains may include decoder proteins such as CPKs. The exploration of how such "channelosomes" generate stimulus-specific Ca 21 signatures that are decoded instantly by the attached decoder proteins will be an exciting future direction for CNGC research (Fig. 1). Structural modeling using solved animal CNGC/HCN structures has improved our understanding (Hua et al., 2003b;Baxter et al., 2008;Niu et al., 2020) but cannot accurately predict the structure of the important CT CaM binding domains. Therefore, resolving high-resolution structures of plant CNGCs will help our understanding of their gating and regulation mechanisms.
The first CNGC family member was identified in 1998 in a screen for CaM binding targets in a complementary DNA library from barley (Hordeum vulgare) aleurone cells, and this CNGC was therefore named H. vulgare CaM Binding Transporter 1, HvCBT1 (Schuurink et al., 1998). Around the same time, two Arabidopsis genes homologous to animal CNGCs were identified and named CNGC1 and CNGC2 (Köhler and Neuhaus, 1998).
The CNGC nomenclature was adopted for future family members, following the suggestion of Mäser et al. (2001). Indeed, within the presumed C terminus of each channel, a cyclic nucleotide-binding domain represents the most conserved sequence. Despite this clear domain classification, binding affinities of cAMP or cGMP to this site have not been measured, and the exact role of these nucleotides, both for channel opening and for physiological functions, is still unclear. Furthermore, there is a fierce, ongoing debate about the production of cNMP in plants (Qi et al., 2010;Ashton, 2011). In light of recent advances in understanding CNGC assembly, function, and regulation by CaM, it is time to conduct more quantitative analyses by using (genetically encoded) reporters for cNMPs, as well as biochemical methods, single-channel recordings, and structural approaches to assess (CaM and cNMP) ligand affinities, gating behavior, and composition of membrane domains containing CNGCs. These analyses will provide us with a better understanding of the role of cNMP for CNGC regulation.
In this Update, we summarized exciting new findings on the molecular functions of plant CNGCs and discussed their significance. With this remarkable progress, we are entering a new era of research on CNGCs and calcium signaling, and we anticipate that more advances in this research field will emerge in the near future.