SLO BK Potassium Channels Couple Gap Junctions to Inhibition of Calcium Signaling in Olfactory Neuron Diversification

The C. elegans AWC olfactory neuron pair communicates to specify asymmetric subtypes AWCOFF and AWCON in a stochastic manner. Intercellular communication between AWC and other neurons in a transient NSY-5 gap junction network antagonizes voltage-activated calcium channels, UNC-2 (CaV2) and EGL-19 (CaV1), in the AWCON cell, but how calcium signaling is downregulated by NSY-5 is only partly understood. Here, we show that voltage- and calcium-activated SLO BK potassium channels mediate gap junction signaling to inhibit calcium pathways for asymmetric AWC differentiation. Activation of vertebrate SLO-1 channels causes transient membrane hyperpolarization, which makes it an important negative feedback system for calcium entry through voltage-activated calcium channels. Consistent with the physiological roles of SLO-1, our genetic results suggest that slo-1 BK channels act downstream of NSY-5 gap junctions to inhibit calcium channel-mediated signaling in the specification of AWCON. We also show for the first time that slo-2 BK channels are important for AWC asymmetry and act redundantly with slo-1 to inhibit calcium signaling. In addition, nsy-5-dependent asymmetric expression of slo-1 and slo-2 in the AWCON neuron is necessary and sufficient for AWC asymmetry. SLO-1 and SLO-2 localize close to UNC-2 and EGL-19 in AWC, suggesting a role of possible functional coupling between SLO BK channels and voltage-activated calcium channels in AWC asymmetry. Furthermore, slo-1 and slo-2 regulate the localization of synaptic markers, UNC-2 and RAB-3, in AWC neurons to control AWC asymmetry. We also identify the requirement of bkip-1, which encodes a previously identified auxiliary subunit of SLO-1, for slo-1 and slo-2 function in AWC asymmetry. Together, these results provide an unprecedented molecular link between gap junctions and calcium pathways for terminal differentiation of olfactory neurons.


Introduction
The nervous system generates a tremendous diversity of cell types that enable formation of functional neural circuits for information processing and mediating behaviors. Cellular diversity is especially important in the developing sensory system as it allows animals to detect different cues in the environment. However, the molecular mechanisms that generate neuronal diversification are only partly understood. One way to generate cellular diversity in the nervous system is to specify different fates and functions of individual cell types across the left-right axis. Left-right asymmetry in the nervous system is present throughout the animal kingdom [1][2][3]. For example, anatomical and functional asymmetries in the human nervous system have been described, such as the greater size of the planum temporale in the left hemisphere, and the localization of language to the left hemisphere of the brain [4]. Defects in brain asymmetry have been correlated with various neurological diseases such as dyslexia and schizophrenia [5]. In the C. elegans nervous system, two pairs of head sensory neurons display molecular and functional asymmetries: the ASE taste neurons and the AWC olfactory neurons [6][7][8][9].
The left and right AWC olfactory neurons appear symmetric at the anatomical and morphological level. However, the two AWC neurons differentiate asymmetrically into two distinct subtypes, one default AWC OFF and one induced AWC ON , at both molecular and functional levels in late embryogenesis [10][11][12]. The AWC ON subtype expresses the G-protein coupled receptor (GPCR) gene str-2 and functions to detect the odorant butanone [11,12]. The AWC-OFF subtype expresses the GPCR gene srsx-3 and functions to sense the odorant 2,3-pentanedione [12,13]. AWC asymmetry is stochastic, such that the AWC ON subtype is induced on the left side of the animal in 50% of the population and on the right side of the animal in the other 50% [11]. AWC asymmetry is maintained throughout the life of an animal [11,14,15].
The ky389 and ky399 alleles were identified from a forward genetic screen for mutants with two AWC ON neurons (2AWC ON phenotype) [11]. The ky389 and ky399 mutations were revealed as gain-of-function (gf) alleles of slo-1 in a study demonstrating a central role of slo-1 in behavioral response to ethanol [23]. slo-1 encodes a conserved voltage-and calcium-activated large conductance BK potassium channel [24,25]. Activation of SLO-1 (Slo1) channels causes hyperpolarization of the cell membrane, thereby reducing cellular excitability and limiting calcium entry through voltage-gated calcium channels [26]. The 2AWC ON phenotype of slo-1(gf) mutants suggests a sufficient role of slo-1(gf) in promoting AWC ON . However, the effect of slo-1 loss-of-function mutations on AWC asymmetry and the mechanism by which slo-1 functions to control AWC asymmetry remained unaddressed. Here we demonstrate that both slo-1 and slo-2 BK channels are necessary for the establishment of AWC asymmetry. We show that slo-1 and slo-2 act redundantly downstream of nsy-5 (innexin gap junction protein) and in parallel with nsy-4 (claudin) to antagonize the function of unc-2 and egl-19 (voltagegated calcium channels) in the induced AWC ON subtype. Asymmetric expression of slo-1 and slo-2 in the AWC ON neuron, which is dependent on NSY-5 and NSY-4, is necessary and sufficient for AWC asymmetry. In addition, SLO-1 and SLO-2 BK channels localize close to UNC-2 and EGL-19 voltage-gated calcium channels, suggesting that SLO channels may inhibit calcium channels through functional coupling and negative feedback. Our results also suggest that slo-1 and slo-2 may regulate AWC communication to control AWC asymmetry through modulating UNC-2 synaptic puncta and synaptic vesicle clustering. Thus, our study identifies an unprecedented role of SLO BK potassium channels in mediating transient gap junction signaling for inhibition of a calcium channel-activated kinase cascade in terminal differentiation of olfactory neurons.
Although slo-1(gf) mutants caused a strong 2AWC ON phenotype, we found that loss-offunction (lf) mutations in slo-1 did not display any defects in AWC asymmetry (Fig 1B and  1C). This suggests that slo-1 may function redundantly with other genes to establish AWC asymmetry. Since slo-2 encodes the only other calcium-activated SLO-like potassium channel in C. elegans and its expression overlaps with slo-1 [24,27,28], we hypothesized that slo-1 and slo-2 may function redundantly to control AWC asymmetry. Similar to slo-1(lf) mutants, slo-2 (lf) mutants did not exhibit abnormalities in AWC asymmetry (Fig 1B and 1D). However, slo-1 (eg142lf); slo-2(ok2214lf) double mutants had a complete penetrance of two AWC OFF neurons (2AWC OFF phenotype): the expression of the AWC ON marker str-2 was lost and the AWC OFF marker srsx-3 was expressed in both AWC neurons (Fig 1Aiii and 1B). Together, these results suggest that slo-1 and slo-2 have essential and redundant roles in promoting the AWC ON subtype.
To determine whether slo-1 and slo-2 affect general AWC fate, we examined the expression of two general AWC markers, the guanylyl cyclase gene odr-1 and the homeodomain protein encoding gene ceh-36, both of which are expressed in both AWC neurons in wild-type animals. Both slo-1(ky399gf) and slo-1(eg142lf); slo-2(ok2214lf) double mutants displayed normal expression of odr-1 and ceh-36 (Fig 1B), suggesting that general AWC identity is not affected by the mutations.
slo-1 and slo-2 act downstream of nsy-5 to antagonize calcium channelmediated signaling in promoting the AWC ON subtype The 2AWC ON phenotype of slo-1(gf) mutants and the 2AWC OFF phenotype of slo-1(lf); slo-2 (lf) double mutants (Figs 1Aii, 1Aiii, 1B and 2A) indicate that the two BK potassium channels function to promote the induced AWC ON subtype. To shed light on how slo-1 and slo-2 promote AWC ON , we investigated where they are located within the AWC asymmetry pathway by generating double and triple mutants of slo-1, slo-2, and other genes previously implicated in AWC asymmetry (Fig 2A).
Since our genetic results put slo-1 and slo-2 (BK potassium channels) at a position similar to unc-2/unc-36 and egl-19/unc-36 (voltage-gated calcium channels) in the AWC asymmetry (C) and SLO-2 (D) proteins with mutation sites and resulting residue changes indicated for the mutants used. Asterisks indicate missense or nonsense mutations. Orange and blue lines indicate regions deleted in slo-1(lf) and slo-2(lf) mutants.
To determine if slo-1 and slo-2 are expressed asymmetrically in AWC neurons, we compared their respective expression level in AWC left (AWCL) and AWC right (AWCR). Although slo-1 and slo-2 are expressed in both AWC neurons in the majority of wild-type animals, both slo-1 and slo-2 are asymmetrically expressed in AWCL or AWCR in a stochastic manner (Fig 3B and 3D, AWCL>AWCR versus AWCL<AWCR are indistinguishable). Random asymmetry of slo-1 and slo-2 expression in AWC neurons is consistent with the stochastic nature of AWC asymmetry. In contrast, nsy-5(ky634lf) and nsy-4(ky627lf) mutants exhibited a significant increase in the percentage of animals that expressed slo-1 and slo-2 symmetrically in AWCL and AWCR (Fig 3B and 3D, AWCL = AWCR). This data suggests that nsy-5 (innexin) and nsy-4 (claudin) are required for the asymmetric expression of slo-1 and slo-2 in AWC neurons, and is consistent with our genetic analysis demonstrating that nsy-5 acts in parallel with nsy-4 to promote AWC ON through slo-1 and slo-2 ( Fig 2B). As a control, we examined the asymmetric expression of slo-1 and slo-2 in unc-43(n498gf)/CaMKII mutants, which cause a 2AWC OFF phenotype, similar to that caused by nsy-5(ky634lf) and nsy-4(ky627lf). We found that the asymmetric expression of both slo-1 and slo-2 was unaffected by the unc-43(n498gf) mutation (Fig 3B and 3D). This suggests that unc-43 (CaMKII) does not regulate the and slo-2p::GFP (C) at a higher level in AWCR (bottom panels) than in AWCL (top panels). Both AWCL and AWCR were labeled by odr-1p::TagRFP. The cell body of both AWC cells is outlined by dashed lines. Scale bar, 5 μm. Anterior is at left and ventral is at bottom. (B, D) Quantification of asymmetric expression of slo-1p::GFP (B) and slo-2p::GFP (D) in AWCL and AWCR in wild type and mutants defective in AWC asymmetry. The single focal plane with the brightest fluorescence in each AWC was selected from the acquired image stack and compared for fluorescence intensity. The fluorescence intensity of slo-1p::GFP and slo-2p::GFP was compared using visual quantitative scoring between AWCL and AWCR in each animal, as previously performed [7,22,58]. expression of slo-1 and slo-2, and is consistent with our genetic results, which place slo-1 and slo-2 upstream of unc-43/CaMKII. The result also supports that the effect of nsy-4 and nsy-5 loss-of-function mutations on asymmetric expression of slo-1 and slo-2 was not due to the 2AWC OFF phenotype.
We also compared expression level of slo-1 and slo-2 in AWC ON and AWC OFF , and found that slo-1 and slo-2 are expressed predominantly in the AWC ON cell (Fig 3E-3H). These results are consistent with the hypothesis that slo-1 and slo-2 promote AWC ON in a cell-autonomous manner.

slo-1 and slo-2 act cell autonomously to specify AWC ON
To determine the site of slo-1 and slo-2 function in promoting AWC ON , we performed genetic mosaic analysis in slo-1(lf); slo-2(lf) mutants containing an integrated AWC ON marker (str-2p:: GFP) transgene and the extrachromosomal array odr-3p::slo-1(overexpressor (OE)); odr-1p:: DsRed or odr-3p::slo-2(OE); odr-1p::DsRed. Both odr-3p::slo-1(OE) and odr-3p::slo-2(OE) transgenes rescued the 2AWC OFF phenotype in slo-1(lf); slo-2(lf) mutants and also caused a slight 2AWC ON overexpression phenotype (Fig 4A and 4C). Since extrachromosomal transgenes are unstable and can be randomly lost at each cell division, the co-injected marker odr-1p::DsRed (normally expressed in both AWC) was used to indicate the presence of the slo-1(OE) or slo-2 (OE) array in AWC. Specifically, we determined if retention of the slo-1(OE) or slo-2(OE) array in only a single AWC cell causes a bias of AWC ON choice in that cell when the mosaic animals exhibited a wild-type 1AWC ON /1AWC OFF phenotype. We found that the slo-1(OE); slo-2(lf) AWC became AWC ON and the slo-1(lf); slo-2(lf) AWC became AWC OFF in the majority of mosaic animals in which slo-1 was expressed only in a single AWC neuron ( Fig 4B). Similarly, the slo-1(lf); slo-2(OE) AWC became AWC ON and the slo-1(lf); slo-2(lf) AWC became AWC OFF when mosaic animals expressed slo-2 only in a single AWC neuron ( Fig 4D). Together, these results support that slo-1 and slo-2 act cell autonomously to specify AWC ON . We did observe a very small percentage of mosaic animals in which the slo-1(lf); slo-2(lf) AWC became AWC ON (Fig 4B and 4D). This suggests that although slo-1 and slo-2 have a largely cell-autonomous role in promoting the AWC ON fate, they may also have a nonautonomous role. This is similar to other genes in the AWC asymmetry pathway, such as nsy-5 and nsy-4 which display both autonomous and nonautonomous roles in AWC asymmetry [20,21].
Mosaic analysis was also performed in transgenic lines in which slo-1(T1001Igf), containing the ky389gf mutation, was overexpressed in a wild-type background, resulting in a strong 2AWC ON phenotype ( Fig 4E). When the transgene was retained in only one of the two AWC cells, the slo-1(gf) cell became AWC ON and the wild-type cell became AWC OFF in the majority of mosaic animals ( Fig 4F). This result is consistent with a largely cell-autonomous function of slo-1 in promoting AWC ON , and also suggests that the AWC with slo-1(gf) activity may become hyperpolarized, allowing the cell to reduce calcium influx and take on the AWC ON subtype.
If no obvious difference in fluorescence intensity between the two AWC cells was observed, the animal was categorized as AWCL = AWCR. If an obvious difference in fluorescence intensity was observed between AWCL and AWCR, the animal was assigned to AWCL > AWCR or AWCR > AWCL. For both slo-1p::GFP and slo-2p::GFP, the visual quantification of fluorescence was performed by the same individual. Only animals with visible expression in both AWC neurons were used in the analysis. p values were calculated using Fisher's exact test. ns, not significant. Error bars indicate standard error of proportion. (E, G) Representative images of wild-type L1 animals expressing slo-1p::2xnlsGFP (E) and slo-2p::GFP (G) in AWC ON (bottom panel) but not in AWC OFF (top panel). Both AWC neurons were marked with ceh-36p::myrTagRFP. AWC ON cells were marked by str-2p::2xnlsTagRFP, and AWC OFF neurons were defined by lack of str-2p::2xnlsTagRFP. Scale bar, 5 μm. Anterior is at left and ventral is at bottom. (F, H) Quantification of slo-1p::2xnlsGFP (F) and slo-2p::GFP (H) expression in AWC ON and AWC OFF . The single focal plane with the brightest fluorescence in each AWC was selected from the acquired image stack and compared for fluorescence intensity. The fluorescence intensity of slo-1p::2xnlsGFP and slo-2p::GFP was compared using visual quantitative scoring between AWC ON and AWC OFF in each animal, as previously performed [7,22,58]. Each animal was categorized into one of three categories: AWC ON = AWC OFF , AWC ON > AWC OFF , and AWC OFF > AWC ON based on the comparison of GFP intensities between AWC ON and AWC OFF cells of the same animal. p values were calculated using a Z-test. Error bars indicate standard error of proportion.  AWC phenotypes in wild type, slo-1(eg142lf); slo-2(ok2214lf), and slo-1 (eg142lf); slo-2(ok2214lf) expressing extrachromosomal transgenes odr-3p::slo-1(OE); odr-1p::DsRed (A) or odr-3p::slo-2(OE); odr-1p::DsRed (C). (B) AWC phenotypes of slo-1(eg142lf); slo-2(ok2214lf) mosaic animals containing the extrachromosomal transgene odr-3p::slo-1(OE) in only one AWC cell, inferred by the presence of the coinjection marker odr-1p::DsRed (normally expressed in both AWC). The data was derived from a subset of data in (A). (D) AWC phenotypes of slo-1(eg142lf); slo-2(ok2214lf) mosaic animals containing the extrachromosomal transgene odr-3p::slo-2(OE); odr-1p::DsRed in only one AWC cell. The data was derived from a subset of data in (C). (E) AWC phenotypes in wild-type animals expressing the extrachromosomal transgene nsy-5p:: SLO-1 and SLO-2 BK potassium channels are localized in the vicinity of UNC-2 and EGL-19 voltage-gated calcium channels in AWC axons SLO-1 and SLO-2 have overlapping expression patterns and have been suggested to potentially form heteromeric channels [24,28,29]. In addition, it has been shown that BK channels and Ntype voltage-gated calcium channels localize in close proximity to achieve functional coupling of these channels [30]. To determine if SLO-1, SLO-2, UNC-2 (N/P/Q-type calcium channels), and EGL-19 (L-type calcium channels) localize in close proximity in AWC, we generated single copy transgenes expressing functional translational reporters driven by the AWC odr-3 promoter using Mos1-mediated single copy insertion [31][32][33]. The tagged proteins expressed in these transgenes were functional in rescuing respective mutant phenotypes [34](S1 Fig, Materials and Methods).
These single copy insertion transgenes showed that GFP::UNC-2, SLO-1::TagRFP, SLO-1:: GFP, SLO-2::TagRFP, and GFP::EGL-19 were mainly localized on the plasma membrane of AWC cell bodies and also displayed a punctate pattern along AWC axons (Fig 5 and S2 Fig), similar to the previously shown localization pattern of GFP::UNC-2 in AWC [34]. Since these channels were localized throughout the plasma membrane of the AWC cell body and had distinct punctate patterns in AWC axons, we focused on analyzing their localization in relation to each other in AWC axons. We found that both SLO-1::TagRFP and SLO-2::TagRFP were localized adjacent to GFP::UNC-2 and GFP::EGL-19 in AWC axons (Fig 5A and 5B, S2A and S2B Fig). In addition, SLO-2::TagRFP is located close to SLO-1::GFP in AWC axons ( Fig 5C). The Coloc 2 plugin in Fiji was used to quantify colocalization of these proteins in AWC axons using three different algorithms (Pearson's correlation coefficient, Spearman's rank correlation coefficient, and Li's ICQ). Each of the algorithms displayed positive correlation indices (Fig 5D and  S2C Fig). This further supports that UNC-2 and EGL-19 localize close to SLO-1 and SLO-2, and that SLO-1 and SLO-2 are localized in close proximity as well.
These results support the notion that BK potassium channels (SLO-1 and SLO-2) and voltage-gated calcium channels (UNC-2 and EGL-19) may function in close proximity for rapid activation of SLO-1 and SLO-2 channels by locally increased calcium levels near UNC-2 and EGL-19 calcium channels.

slo-1 and slo-2 regulate the localization of synaptic markers in AWC neurons
It has been shown that communication between the pair of AWC neurons via chemical synapses in axons is important for induction of the AWC ON subtype [11]. Our genetic data suggests that slo-1 and slo-2 are required for the specification of the induced AWC ON subtype. In addition, SLO-1 and SLO-2 displayed distinct punctate localization patterns in AWC axons. Thus, we examined whether slo-1 and slo-2 regulate localization of synaptic markers in AWC neurons. To do so, we generated Mos1-mediated single copy insertion transgenes expressing fluorescently tagged synaptic markers, GFP::UNC-2 and YFP::RAB-3, driven by the AWC odr-3 promoter (Figs 5A, 5B and 6). UNC-2 is localized to presynaptic active zones and RAB-3 is a synaptic vesicle marker [34].
In wild type, GFP::UNC-2 was localized in the AWC axon and cell body, and YFP::RAB-3 was mainly localized in a punctate pattern in the AWC axon as shown previously [34] (Fig 6).
doi:10.1371/journal.pgen.1005654.g004 were localized in AWC cell bodies (arrows) and in a punctate pattern along AWC axons (arrowheads). In AWC axons, SLO-1::TagRFP was localized next to GFP::UNC-2 (A); SLO-2::TagRFP was adjacent to GFP::UNC-2 (B); and SLO-2::TagRFP was localized near SLO-1::GFP (C). Insets show higher magnification of the outlined areas that exemplify localization of two translational reporters in close proximity. Scale bar, 5 μm. Anterior is at left and ventral is at bottom. (D) Quantification of mean correlation coefficient between SLO-1 and UNC-2, SLO-2 and UNC-2, as well as SLO-1 and SLO-2 using three algorithms of the Coloc 2 plugin in Fiji: Pearson's correlation coefficient, Spearman's rank correlation coefficient, and Li's ICQ. For each colocalization class, images of three animals were used for quantification. Positive values of each coefficient indicate positive correlation, values close to zero indicate no correlation, and negative values indicate anti-correlation. Pearson's correlation coefficient ranges from -1 to +1; Spearman's rank correlation coefficient ranges from -1 to +1; Li's ICQ value ranges from -0.5 to +0.5. A schematic diagram of the AWC cell body, axon, dendrite, and cilia that represents the approximate region of images in A-C is shown in S2D Fig. doi:10.1371/journal.pgen.1005654.g005 In slo-1(ky399gf) animals, the intensity of GFP::UNC-2 or YFP::RAB-3 was not significantly affected in the AWC axon and cell body (Fig 6). However, slo-1(eg142lf); slo-2(ok2214lf) mutants displayed significant reduction in intensity of GFP::UNC-2 and YFP::RAB-3 in the AWC axon and cell body (Fig 6). These results suggest that slo-1 and slo-2 are required for localization and/or stability of synaptic markers, UNC-2 and RAB-3, in AWC neurons, which may contribute to the 2AWC OFF phenotype caused by the slo-1(eg142lf); slo-2(ok2214lf) mutations. Our genetic mosaic analysis suggests a minor role of nonautonomous function of slo-1 and slo-2 in establishing AWC asymmetry (Fig 4), which is consistent with a possible role of slo-1 and slo-2 in regulating synaptic communication of AWC neurons. In addition, autofluorescence of the gut found in wild-type animals was visibly decreased in slo-1(eg142); slo-2 (ok2214lf) mutants (S3A Fig), suggesting that the SLO channels are required for gut autofluorescence.
As a control, the intensity of GFP expressed from the transgene odr-3p::GFP was analyzed in wild-type and mutant backgrounds, and no significant effect was observed in slo-1(ky399gf) and slo-1(eg142lf); slo-2(ok2214lf) mutants (S3B Fig). This result rules out the possibility that the activity of the odr-3 promoter is regulated by slo-1 and slo-2, and also supports the notion that the effect of slo-1(lf); slo-2(lf) mutations on UNC-2 and RAB-3 is mainly at the subcellular localization level. It is also possible that slo-1(lf); slo-2(lf) mutations may affect unc-2 and rab-3 at post-transcriptional levels, such as translation efficiency, mRNA and/or protein stability.
Previous studies have shown that slo-1(lf) or slo-2(lf) mutations result in increased neurotransmitter release at the neuromuscular junction in the ventral nerve cord [25,35]. However, a recent study showed that UNC-2 localization is not affected at the presynaptic terminals of neuromuscular junctions in slo-1(lf) mutants [36]. Thus, previous findings did not demonstrate a correlation between increased neurotransmitter release and increased localization of UNC-2 or RAB-3 at presynaptic sites of the neuromuscular junction in slo-1(lf) or slo-2(lf) mutants. To examine whether the localization of UNC-2 and RAB-3 is affected in ventral cord motor neurons in slo-1(lf); slo-2(lf) mutants, we quantified the intensity of GFP::UNC-2 and RAB-3:: mCherry driven by the unc-25 promoter, which is expressed in ventral cord motor neurons [34]. We examined the axons located anterior to VD5 and DD3 neurons in wild type and slo-1 (lf); slo-2(lf) mutants at the L4 stage, but no significant difference was observed (S4 Fig). This suggests that slo-1 and slo-2 do not play an apparent role in the localization of these presynaptic markers in the ventral nerve cord. The different effects of slo-1 and slo-2 mutations on the localization of synaptic markers in AWC neurons and ventral cord motor neurons suggest that slo-1 and slo-2 take on a different function in AWC neurons than in the ventral cord motor neurons.
Although no apparent effect of slo-1(lf); slo-2(lf) mutations on the localization of UNC-2 and RAB-3 was observed in ventral cord motor neurons, the effect of slo-1 and slo-2 mutations on locomotion was performed by analyzing the wavelength and wave width of body wave tracks of wild type, slo-1(lf), slo-2(lf), and slo-1(lf); slo-2(lf) animals. We found that the wavelength of the worm track was not affected in the mutants, however the wave width was significantly increased in slo-1(lf), slo-2(lf), and slo-1(lf); slo-2(lf) mutants (S5 Fig). These results suggest that slo-1 and slo-2 are required for normal locomotion.
bkip-1 is required for slo-1 and slo-2 function in promoting AWC ON Previous studies have identified several modulators of SLO-1 activity in muscles using forward genetic screens. Since genes may interact in similar pathways in different tissues, we chose these candidate genes to determine whether they may also modulate SLO-1 activity in AWC neurons. bkip-1 mutants were identified from a screen for suppressors of the lethargic phenotype of slo-1(gf) mutants. BKIP-1 (BK channel Interacting Protein), a single pass membrane protein, functions as an auxiliary subunit of SLO-1 to assist in regulating neurotransmitter release and regulate the surface expression of the channel [37]. Similar to bkip-1, ctn-1 (αcatulin), identified from two independent screens for suppressors of the slo-1(gf) lethargic phenotype, also regulates surface localization of SLO-1 in both muscles and ventral nerve cord motor neurons [38,39]. In addition, components of the dystrophin-associated protein complex (DAPC), including dys-1 (dystrophin), dyb-1 (dystrobrevin), stn-1 (syntrophin), and dyc-1 (Cterminal PDZ-domain ligand of nNOS), control the localization of SLO-1 in muscles but not in neurons [40,41]. Furthermore, islo-1, encoding a transmembrane protein, functions as an adaptor protein that links the DAPC to SLO-1 for SLO-1 localization in muscles [40].
Previous work demonstrates a role of bkip-1 in regulating the surface expression of SLO-1 in muscle dense bodies and the nerve ring [37]. We therefore determined whether bkip-1 affects SLO-1 localization in AWC neurons by examining a functional SLO-1::GFP translational reporter driven by the AWC odr-3 promoter in wild type and bkip-1(zw2) mutants ( Fig  7B). We found that in bkip-1(zw2) mutants, SLO-1::GFP intensity was significantly reduced in AWC axons (Fig 7B and 7C) but is not significantly affected in cell bodies (Fig 7D). This suggests that bkip-1 is required for appropriate localization of SLO-1 in AWC axons but not in the AWC cell body. Consistent with the result suggesting that bkip-1 is not the only factor required for slo-1 function in AWC asymmetry (Fig 7A), this result also suggests that slo-1 activity could be required in both the AWC axons (dependent on bkip-1) and cell bodies (independent of bkip-1). We also examined whether bkip-1(zw2) mutants display altered the localization of SLO-2::GFP in AWC axons, but did not find a significant effect (S6 Fig). This result suggests that slo-2 may require bkip-1 in a manner independent of appropriate localization. bkip-1 may be required for appropriate slo-2 expression levels, or BKIP-1 may physically interact with SLO-2.
Together, our results showed that bkip-1 is the only one of the known modulators of slo-1 activity in muscles to be also required for slo-1 and slo-2 function in AWC asymmetry. Thus, our results suggest that slo-1 and slo-2 need a different set of regulators for their function in AWC asymmetry.

Additional voltage-gated potassium channel EGL-2 (EAG) plays a role in AWC asymmetry
The voltage-dependent activation of SLO-1 and SLO-2 channels is modulated by calcium (for SLO-1 and SLO-2) and chloride (for SLO-2) [24,26]. To determine whether any chloride channels or other voltage-gated potassium channels might be involved in establishing left-right AWC asymmetry, we examined AWC asymmetry in mutants of selective channels that have been shown to be expressed in the nervous system (WormBase). Although the majority of mutants examined did not display an AWC asymmetry defect (S7A Fig), a gain of function mutation in unc-103 (ERG voltage-gated potassium channel) resulted in a slight 2AWC ON phenotype (S7B Fig). In addition, a gain of function mutation in egl-2 (EAG voltage-gated potassium channel) caused a high penetrance of the 2AWC ON phenotype (S7B Fig), as previously shown [11]. We found that the egl-2(n693gf) mutation suppressed the 2AWC OFF phenotype observed in slo-1(eg142); slo-2(ok2214) double mutants, nsy-5(ky634lf), unc-43(n498gf), and tir-1(ky648gf) single mutants (S7B Fig). This suggests that egl-2 may function downstream of these genes to promote the AWC ON fate. Alternatively, it is possible that egl-2 may function at the same level as slo-1 and slo-2; and that the production of 2 AWC ON neurons by egl-2(gf) in the slo-1(eg142); slo-2(ok2214) mutants is because egl-2(gf) is sufficient to cause enough membrane hyperpolarization to induce AWC ON even in the absence of slo-1 and slo-2. Like slo-1(lf) and slo-2(lf) mutants, loss-of-function mutations in egl-2 did not cause a significant effect on AWC asymmetry nor did slo-1(lf); egl-2(lf) double mutants (S7B Fig), suggesting that egl-2 may act redundantly with other factor(s) in promoting AWC ON .

Discussion
Here we identify an essential role of SLO BK potassium channels in asymmetric differentiation of one pair of olfactory neurons. Our findings reveal a functional link between gap junctions and SLO channels in inhibition of voltage-gated calcium channels for diversification of olfactory neurons. To the best of our knowledge, stochastic AWC asymmetry is the first system in which SLO channels are implicated in terminal neuron differentiation, stochastic cell fate determination, and left-right patterning.
Our results suggest antagonistic and parallel functions of BK potassium channels (SLO-1 and SLO-2) and voltage-gated calcium channels (UNC-2/UNC-36 and EGL-19/UNC-36) downstream of NSY-5 gap junctions in AWC asymmetry. UNC-2/UNC-36 and EGL-19/ UNC-36 activate a CaMKII-MAP kinase cascade to specify the default AWC OFF subtype, while SLO-1 and SLO-2 inhibit the calcium channel-activated kinase cascade to promote the induced AWC ON subtype (Fig 8). Calcium and voltage are potential signals that mediate intercellular communication between the two AWC neurons and other neurons in the NSY-5 gap junction network to coordinate stochastic AWC asymmetry [19,20]. In addition, both SLO BK channels and voltage-gated calcium channels generate voltage and calcium signals, and are subject to calcium-and voltage-dependent activation and inactivation [26,42]. The regulatory loop between gap junctions, SLO BK channels, and voltage-gated calcium channels can potentially generate sustained differences in calcium-regulated signaling outputs between the two AWC cells through positive and negative feedback mechanisms, leading to asymmetric differentiation of AWC cells. This extends the previous model of NSY-5 function in AWC asymmetry by identifying SLO BK channels as the mediators of transient gap junction signaling for antagonizing voltage-gated calcium channel pathways.
Signaling via NSY-5 gap junctions may lead to transcriptional regulation of slo-1 and slo-2 in order to ensure that these genes are expressed asymmetrically in the AWC neurons. Studies have shown that connexin gap junction proteins are capable of regulating gene expression. For example, gap junction communication mediated by Cx43 is required for ERK phosphorylation of the transcription factor Sp1, which in turn leads to appropriate expression of an osteoclastin transcriptional element [43]. It has been suggested that gap junctions may allow diffusion of second messengers such as calcium and cyclic nucleotides, which subsequently can influence gene transcription [43,44]. It has also been suggested that C-terminal tails of connexins may bind to particular proteins, which can then contribute to regulating gene expression [44]. It is possible that NSY-5 gap junctions use similar mechanisms to regulate slo-1 and slo-2 gene expression.
SLO-1 is 55% identical to its mouse orthologue Slo1 and SLO-2 is 41% identical to its mammalian orthologue Slack, while SLO-1 is only 18% identical to its nematode paralogue SLO-2 along the entire channel peptide [28]. SLO-1/Slo1 and SLO-2/Slack have overlapping expression patterns and may form heteromeric channels [24,28,29]. However, functional relationships between SLO-1/Slo1 and SLO-2/Slack have not yet been demonstrated in any biological contexts. Our results show that SLO-1 localizes in close proximity to SLO-2 in AWC neurons. In addition, our results suggest that slo-1 and slo-2 have complete functional redundancy in AWC asymmetry, since loss-of-function mutations in either gene alone did not cause any defects in AWC asymmetry while slo-1(lf); slo-2(lf) double mutants displayed a complete penetrance of the 2AWC OFF phenotype. Functional redundancy between SLO-1/Slo1 and SLO-2/ Slack may represent one of the general mechanisms for their roles in other systems. The voltage range of activation of BK channels is modulated by different intracellular factors including calcium (for SLO-1, Slo1, and SLO-2), chloride (for SLO-2 and Slack), sodium (for Slack), pH, and phosphorylation [24,26]. None of the mutants of chloride channels we examined displayed any AWC asymmetry defects. In addition, although SLO-2 shares a complete redundant function with SLO-1 in AWC asymmetry, it has not been shown that the activation of SLO-1 channels is sensitive to chloride. These findings suggest that SLO-2's redundant role with SLO-1 in establishing AWC asymmetry may be more dependent on sensitivity to calcium than to chloride.
Calcium-activated BK channels and voltage-gated calcium channels have been shown to localize in close proximity to ensure selective and rapid activation of BK channels by a local increase in cytosolic calcium level [30]. The sensitivity of vertebrate Slo1 channels to calcium provides an important negative feedback for calcium entry in many cell types. For example, activation of Slo1 channels causes transient membrane hyperpolarization, which limits calcium entry through voltage-gated calcium channels to control the burst of calcium action potentials in cerebellar Purkinje cells and to regulate synaptic transmission in presynaptic terminals [26]. Our genetic results and findings that SLO-1 and SLO-2 localize close to UNC-2 and EGL-19 voltage-gated calcium channels are consistent with the physiological roles of vertebrate Slo1 channels in inhibiting voltage-gated calcium channels through functional coupling and negative feedback. By analogy to functional coupling between Slo1 and voltage-gated calcium channels in vertebrates, SLO-1 and SLO-2 may couple with UNC-2/UNC-36 and EGL-19/UNC-36 to generate oscillation of cytosolic calcium and voltage signals to coordinate stochastic AWC asymmetry through a feedback loop. In this hypothetical feedback loop, an increase in voltage triggers voltage-gated calcium channels to open, leading to an increase in intracellular calcium levels. High calcium levels allow the coupled calcium-activated BK channels to open, resulting in a decrease in voltage. The decreased voltage causes the voltage-gated calcium channels to close, leading to a decrease in intracellular free calcium levels and the subsequent closure of calcium-activated BK channels and an increase in voltage. This would initiate another cycle of calcium and voltage oscillation. Previous studies identified two forms of intercellular communication important for AWC asymmetry: one is mediated by NSY-5 gap junctions between the cell body of AWC and other neurons in a network [19,20]; the other is by synaptic connection between two AWC axons [10,11]. Since SLO-1 and SLO-2 are localized in proximity to UNC-2 and EGL-19 at the AWC axons, functional coupling between BK channels (SLO-1 and SLO-2) and voltage-activated calcium channels (UNC-2 and EGL-19) may occur at AWC axons.
Our study has revealed that SLO-1 and SLO-2 have different functions and interacting partners in AWC olfactory neurons than in ventral cord motor neurons. Our genetic analysis suggests that BK potassium channels (SLO-1 and SLO-2) act to antagonize calcium channels (UNC-2/UNC-36 and EGL-19/UNC-36) to promote the AWC ON identity. A recent report suggests that in M4 motor neurons, UNC-2 and UNC-36 function to activate SLO-1, which in turn antagonizes the EGL-19 calcium channel to inhibit synaptic transmission at the M4 neuromuscular junction [45]. A recent study showed that UNC-2 localization is not affected at the presynaptic terminals of neuromuscular junctions in slo-1(lf) mutants [36]. However, our results suggest that slo-1 and slo-2 are required for appropriate localization or stability of presynaptic markers UNC-2 and RAB-3 in AWC axons. We also show that SLO-1 and SLO-2 localize in close proximity to both UNC-2 and EGL-19 calcium channels in AWC neurons, in contrast to a report that SLO-2 exclusively couples with EGL-19 but not with UNC-2 in ventral cord motor neurons [35].
BK channels are ubiquitously expressed and have a staggering repertoire of functions in different tissues. To achieve functional diversity, BK channels, which assemble as tetramers of pore-forming α-subunits, can form complexes with various auxiliary β-subunits. For example, the β1 subunit changes gating and calcium sensitivity of Slo1 α subunits, and β2 subunits promote fast inactivation of Slo1 channels [26]. In addition, functional diversity of Slo1 channels can be achieved by alternative splicing, posttranslational modifications, and heteromultimer formation [26]. In C. elegans, several modulators have been identified for surface expression and activity of SLO-1 channels in muscles and neurons [37][38][39][40][41]. Our results show that the auxiliary subunit BKIP-1 is the only previously identified modulator of SLO-1 to be required for SLO-1 and SLO-2 function in asymmetric AWC differentiation. AWC asymmetry may provide an effective model system to identify novel modulators of SLO BK channels in vivo due to the ease of unbiased forward genetic screens in identifying biologically relevant genes and robust phenotypic readouts of SLO channel activity.

Germ line transformation
Transgenic strains were generated by injecting DNA constructs into the syncytial gonad of adult worms (P 0 ) as previously described [54]. F 1 worms expressing fluorescent transgenes were picked and cloned (1 worm per plate). The F 1 clones that have F 2 progeny containing fluorescent transgenes were selected as transgenic lines and analyzed.

Imaging of transgenic worms expressing fluorescent proteins
Transgenic strains expressing fluorescent markers or fluorescently tagged proteins were mounted onto 2% agarose pads and anesthetized with 5mM sodium azide (Sigma) or 7.5mM levamisole (Sigma). Z-stack images were acquired at room temperature (20-22°C) using Zeiss Axio Imager Z1 or M2 microscopes, each of which is equipped with a motorized focus drive, a Zeiss objective EC Plan-Neofluar 40x/1.30 Oil DIC M27, a Piston GFP bandpass filter set (41025, Chroma Technology), a TRITC filter set (41002c, Chroma Technology), and a Zeiss AxioCam CCD digital camera (MRm for Z1 and 506 mono for M2) driven by the Zeiss imaging software (AxioVision for Z1 and ZEN for M2). For comparison of fluorescence intensity, all animals in each set of experiments were subjected to the same exposure time.

Genetic mosaic analysis
Genetic mosaic analysis was performed with various unstable extrachromosomal transgenic arrays in either wild type or slo-1(eg142lf); slo-2(ok2214lf) mutants. Three different experiments were performed to determine the sites of slo-1 and slo-2 function in AWC asymmetry. odr-3p:: slo-1 was injected into slo-1(lf); slo-2(lf) mutants to determine whether slo-1 acts cell autonomously or nonautonomously to rescue the 2AWC OFF mutant phenotype. A similar experiment was performed using the odr-3p::slo-2 extrachromosomal array. In the third experiment, nsy-5p::slo-1(T1001Igf) was injected into wild-type animals. In all three experiments, the odr-1p:: DsRed marker (expressed in AWC and AWB) was included in the injection mix to serve as an indicator for presence or absence of the extrachromosomal transgene in AWC. The AWC ON and AWC OFF neurons were determined using expression of a stable integrated str-2p::GFP (AWC ON marker) transgene. Transgenic strains were passed for minimum of six generations to allow the transgenes to stabilize before scoring for mosaic animals.

Colocalization analysis
Colocalization was quantified using the Coloc 2 plugin (http://fiji.sc/Coloc_2) in Fiji [56]. Three different algorithms were used: Pearson's correlation coefficient, Spearman's rank correlation coefficient, and Li's ICQ. For each colocalization class, images of at least three animals were used for quantification. Positive values of each coefficient indicate positive correlation, values close to zero indicate no correlation, and negative values indicate anti-correlation. Pearson's correlation coefficient ranges from -1 to +1; Spearman's rank correlation coefficient ranges from -1 to +1; Li's ICQ value ranges from -0.5 to +0.5

Locomotion analysis
Locomotion analysis was performed on L4 animals in wild type, slo-1(eg142), slo-2(ok2214), and slo-1(eg142); slo2(ok2214) animals. Single animals of each genotype were placed on a bacterial lawn and allowed to make tracks. The worm tracks as well as individual worms were imaged and analyzed in ImageJ [57]. All animals were placed on the same batch of NGM plates seeded with the same batch of HB101 and were imaged on the same day. Wavelength was measured as the distance between wave peaks, and at least 3 wavelengths were measured and averaged per animal. The wavelength was normalized by the body length of the animal. Wave width was measured as the distance from the peak to the trough of the worm wave. At least 3 wave widths were measured and averaged per animal. The wave width was normalized by the body length of the animal.  A, B) were localized in AWC cell bodies (arrows) and in a punctate pattern along AWC axons (arrowheads). In AWC axons, SLO-1::TagRFP was localized next to GFP::EGL-19 (A); SLO-2:: TagRFP was adjacent to GFP::EGL-19 (B). Insets show higher magnification of the outlined areas that exemplify localization of two translational reporters in close proximity. Scale bar, 5 μm. Anterior is at left and ventral is at bottom. (C) Quantification of mean correlation coefficient between SLO-1 and EGL-19 as well as SLO-2 and EGL-19 using 3 algorithms of the Coloc 2 plugin in Fiji: Pearson's correlation coefficient, Spearman's rank correlation coefficient, and Li's ICQ. Images of four (SLO-1 and EGL- 19) or six (SLO-2 and EGL-19) animals were used for quantification. Positive values of each coefficient indicate positive correlation, values close to zero indicate no correlation, and negative values indicate anti-correlation. Pearson's correlation coefficient ranges from -1 to +1; Spearman's rank correlation coefficient ranges from -1 to +1; Li's ICQ value ranges from -0.5 to +0.5. (D) Schematic diagram of the AWC cell body, axon, dendrite, and cilia. The outlined region represents the approximate region of images shown in Fig 5A, 5B 6) slo-1 and slo-2 mutants alter gut autofluorescence but have no effect on expression of GFP from an AWC odr-3 promoter in AWC cells. (A) Representative images taken at identical exposure times of wild-type, slo-1(ky399gf), and slo-1(eg142lf); slo-2 (ok2214lf) animals expressing the single copy insertion odr-3p::GFP::unc-2 (partial head images of the same animals were shown in Fig 6A). The gut autofluorescence of the worm is noticeably decreased in slo-1(eg142lf); slo-2(ok2214lf) mutants as compared to wild-type and slo-1 (ky399gf) mutants. Scale bar, 20 μm. (B) Left panels: Images of wild type, slo-1(ky399gf), and slo-1(eg142lf); slo-1(ok2214lf) mutants expressing odr-3p::GFP in AWC in L1. Right panel: Quantification of GFP fluorescence intensity in AWC cell bodies. For each animal, GFP intensity was quantified from the single focal plane with the brightest GFP expression in the AWC cell body and subtracted by background fluorescence intensity. Scale bar, 5 μm. Student's t-test was used for statistical analysis. ns, not significant (p = 0.6). Error bars, standard error of the mean. AU, arbitrary unit. (TIF) cell body. Anterior is at left and ventral is at bottom. Student's t-test was used for statistical analysis. ns, not significant. Error bars, standard error of the mean. AU, arbitrary unit. (TIF) S7 Fig. Other voltage-gated potassium and chloride channels in AWC asymmetry. (A) Analysis on the effect of mutations in additional voltage-gated potassium channels and chloride channels. 2AWC ON , both AWC cells express str-2; 1AWC OFF /AWC ON , only one of the two AWC cells expresses str-2; 2AWC OFF , neither AWC cell expresses str-2. (B) Analysis on the effect of mutations in egl-2 (EAG voltage-gated potassium channel) and unc-103 (ERG voltage-gated potassium channel) on AWC asymmetry. (TIF)