Effect of the CALHM1 G330D and R154H Human Variants on the Control of Cytosolic Ca2+ and Aβ Levels

CALHM1 is a plasma membrane voltage-gated Ca2+-permeable ion channel that controls amyloid-β (Aβ) metabolism and is potentially involved in the onset of Alzheimer's disease (AD). Recently, Rubio-Moscardo et al. (PLoS One (2013) 8: e74203) reported the identification of two CALHM1 variants, G330D and R154H, in early-onset AD (EOAD) patients. The authors provided evidence that these two human variants were rare and resulted in a complete loss of CALHM1 function. Recent publicly available large-scale exome sequencing data confirmed that R154H is a rare CALHM1 variant (minor allele frequency (MAF)  = 0.015%), but that G330D is not (MAF  = 3.5% in an African American cohort). Here, we show that both CALHM1 variants exhibited gating and permeation properties indistinguishable from wild-type CALHM1 when expressed in Xenopus oocytes. While there was also no effect of the G330D mutation on Ca2+ uptake by CALHM1 in transfected mammalian cells, the R154H mutation was associated with defects in the control by CALHM1 of both Ca2+ uptake and Aβ levels in this cell system. Together, our data show that the frequent CALHM1 G330D variant has no obvious functional consequences and is therefore unlikely to contribute to EOAD. Our data also demonstrate that the rare R154H variant interferes with CALHM1 control of cytosolic Ca2+ and Aβ accumulation. While these results strengthen the notion that CALHM1 influences Aβ metabolism, further investigation will be required to determine whether CALHM1 R154H, or other natural variants in CALHM1, is/are associated with EOAD.


Introduction
Alzheimer's disease (AD) is a progressive neurodegenerative disorder leading to the most common form of dementia in elderly people. Histological studies of the AD brain have revealed pathological changes triggered by two classical lesions, the senile plaques and neurofibrillary tangles [1,2]. Senile plaques result from the accumulation of amyloid-b (Ab), a series of peptides produced by sequential endoproteolysis of the amyloid precursor protein (APP) by band c-secretases [3,4]. APP is genetically linked to early-onset familial forms of AD and Ab is considered to be a causative factor in AD [5]. The etiology of AD is influenced by a strong genetic heterogeneity. Rare autosomal dominant mutations cause early-onset familial AD, whereas complex interactions between different genetic variants and environmental factors modulate the risk for the vast majority of late-onset AD cases [6][7][8].
Calcium homeostasis modulator 1 (CALHM1) [9] was identified by a tissue-specific gene expression profiling approach [10,11] that screened for genes located on susceptibility loci for AD and that are preferentially expressed in the hippocampus, a brain region affected early in AD [12]. CALHM1 is a plasma membrane Ca 2+permeable ion channel regulated by voltage and extracellular Ca 2+ (Ca 2+ o ) levels [9,[13][14][15]. The exact function of CALHM1 in the brain is not completely understood, but recent evidence has revealed that CALHM1 controls neuronal intracellular Ca 2+ (Ca 2+ i ) homeostasis and signaling, as well as Ca 2+ -dependent neuronal excitability [13,14].
We initially proposed that the naturally occurring P86L variant in CALHM1 (rs2986017) was associated in European cohorts with both AD risk and an earlier age-at-onset of AD [9]. Reports of both refutation and confirmation of the association with AD risk in independent genetic studies followed the original results [16]. A meta-analysis of all published studies has now shown that CALHM1 per se has no significant impact on AD risk and is thus likely not a robust independent risk gene for AD [16]. However, the meta-analysis confirmed the association of CALHM1 with AD age-at-onset [16]. In support of the idea that CALHM1 might be involved in the pathological process of AD, we have reported that CALHM1 activation triggers a Ca 2+dependent pathway that suppresses extracellular Ab accumulation in cell lines [9]. Moreover, two independent genetic studies have showed that the CALHM1 P86L variant influences Ab levels in human cerebrospinal fluid [17,18], but see also [19]. In vitro functional studies further demonstrated that the P86L variantthrough a mechanism yet to be determined [14] -caused a partial loss of function by inhibiting the effect of CALHM1 on Ca 2+ influx and Ab repression [9,13,14,20,21]. Altogether these results support the notion that CALHM1 controls Ab metabolism and AD pathogenesis.
In a recent study [20], Rubio-Moscardo et al. reported the identification of two natural variants in CALHM1 that occurred in early-onset AD (EOAD) patients. The two CALHM1 variants, G330D and R154H, were identified by sequencing CALHM1 coding regions in three independent series comprising a total of 284 EOAD patients and 326 controls. The authors found that the G330D and R154H variants were associated with a complete loss of CALHM1 control of Ca 2+ influx in cell lines, suggesting that these variations interfere with CALHM1 function and may thus contribute to the risk of EOAD [20]. In the current study, we have reassessed the functionality of the CALHM1 G330D and R154H variants on channel gating and Ca 2+ permeability by using electrophysiological recordings and Ca 2+ i measurements in CALHM1-expressing oocytes and mammalian cells, respectively. While there was no effect of the G330D mutation on CALHM1 function in any of our experimental approaches, we found that the R154H mutation resulted in a partial inhibition of CALHM1dependent Ca 2+ uptake in cell lines.
In the context of characterizing CALHM1 function, Rubio-Moscardo et al. failed to observe an inhibition of Ab accumulation by CALHM1 expression using a modified version of the originally described protocol of CALHM1 activation (hereafter referred to as the ''Ca 2+ add-back'' protocol [9]), which consists of transiently lowering Ca 2+ o and then adding it back at physiological concentration [13,14,21,22]. Here, we have also reassessed the effect of CALHM1 activation on Ab levels in cell lines by testing different Ca 2+ add-back protocols, including the modified version used by Rubio-Moscardo et al. In all conditions, we confirmed the robust and sustained repressing effect of CALHM1 activation on Ab accumulation. In line with the effect of the two CALHM1 variants on CALHM1-mediated Ca 2+ influx, we further demonstrate that the G330D variant has no functional effect on the control of Ab levels by CALHM1, whereas the R154H variant is associated with a partial loss of CALHM1 function.

CALHM1 variant frequency in a large-scale exome sequencing database
We interrogated the NHLBI Exome Sequencing Project (ESP) database to determine the frequency of the CALHM1 R154H and G330D variants (see Methods). The R154H variant (10:105218048, rs371900329) was found in a single individual across the entire ESP cohort (minor allele frequency (MAF) = 0.015%). In contrast, the G330D variation (10:105215071, rs114015468) was found in one individual of European-American ethnicity and in 151/4,255 individuals of African-American ethnicity (a 3.55% MAF in the latter population). Therefore, sequencing analysis of these large human cohorts show that CALHM1 R154H is a rare variation, but that CALHM1 G330D, with a MAF above 1%, is not.  [14,15]. Rubio-Moscardo et al. observed that the R154H and G330D variants resulted in a complete loss of CALHM1-dependent Ca 2+ influx in transfected mammalian cells [20]. To determine if R154H or G330D alters CALHM1 gating or Ca 2+ permeability, we recorded whole cell currents from Xenopus oocytes injected with wild-type (WT)-CALHM1, R154H-CALHM1, and G330D-CALHM1 cRNA. In the presence of 2 mM Ca 2+ o , membrane depolarization of oocytes expressing WT-CALHM1 or the CALHM1 variants resulted in slowlyactivating outward currents that deactivated upon stepping to hyperpolarized voltages (Figs. 1A-C). Using the same conditions, similar currents were not observed in water-injected oocytes (data not shown, see also Ref. [14]). This indicates that R154H-CALHM1 and G330-CALHM1 form functional voltage-dependent channels.
Lowering Ca 2+ o resulted in a left shift in the apparent conductance-voltage (G-V) relationship for CALHM1, allowing channels to open at resting membrane potentials (Fig. 1D, see also Ref. [14]). To determine if R154H-CALHM1 and G330-CALHM1 are also regulated by Ca 2+ o , we recorded currents from CALHM1-expressing oocytes in divalent-free and 2 mM Ca 2+ o containing solutions, and analyzed the apparent G-V relationships. For WT-CALHM1, 2 mM Ca 2+ o right shifted the apparent G-V relationship by 136.7 mV from a half-activation voltage (V 0.5 ) of -75.9 mV in 0 mM Ca 2+ o to +60.8 mV in 2 mM Ca 2+ o (Fig. 1D). Similarly, the presence of 2 mM Ca 2+ o caused 145.2 mV and 141.6 mV right shifts in the apparent G-V relationships for R154H-CALHM1 and G330D-CALHM1, respectively (Fig. 1D). These results suggest that R154H-CALHM1 and G330D-CALHM1 are regulated by both voltage and Ca 2+ o , with gating properties indistinguishable from WT-CALHM1.
We then asked whether the lack of Ca 2+ influx observed by Rubio-Moscardo et al. in cell lines expressing R154H-CALHM1 and G330D-CALHM1 was due to alterations in Ca 2+ permeability. To assess if R154H-CALHM1 and G330D-CALHM1 are permeable to Ca 2+ , we measured reversal potentials (E rev ) in solutions containing 0 or 2 mM Ca 2+ o with constant NaCl. The presence of Ca 2+ o resulted in a shift in E rev to more depolarized voltages for WT-CALHM1, R154H-CALHM1, and G330D-CALHM1 (Fig. 1E). This shift in E rev was not significantly different for the two variants compared to WT-CALHM1, suggesting that relative Ca 2+ permeability was not altered. In conclusion, we were unable to detect any difference in the gating or Ca 2+ permeability for the R154H and G330D variants, compared to WT-CALHM1.

Effect of the CALHM1 G330D and R154H variants on Ca 2+ influx in mammalian cells
To further investigate the effect of the G330D and R154H mutations on CALHM1 function, we evaluated CALHM1mediated Ca 2+ influx by measuring Ca 2+ i levels in transfected hippocampal HT-22 cells loaded with the fluorescent Ca 2+ indicator Fluo-4. Removal and subsequent add-back of Ca 2+ o results in a CALHM1-mediated increase in Ca 2+ i levels [9]. Upon Ca 2+ add-back stimulation, we found that G330D-CALHM1 produced the same elevation in Ca 2+ i levels as WT-CALHM1 ( Figs. 2A and 2B). In contrast, R154H significantly inhibited CALHM1-mediated Ca 2+ influx in these conditions ( Figs. 2A and  2B). R154H-CALHM1 only exhibited a partial loss of function compared to the dead mutant W114A-CALHM1 that, as expected [13], fully inhibited CALHM1-mediated Ca 2+ influx (Figs. 2A and 2B). These results show that the G330D mutation did not alter CALHM1 function, while the R154H mutation resulted in a partial loss of function in mammalian cells.

Reevaluation of the effect of CALHM1 activation on Ab levels
In our original study [9], we showed that Ca 2+ add-back in two CALHM1 expressing neuroblastoma cell lines had an inhibitory effect on extracellular Ab accumulation. Expression of CALHM1 in N2a cells decreased Ab levels, while silencing of CALHM1 in SH-SY5Y cells increased Ab levels, upon Ca 2+ add-back [9]. The decrease in Ab levels observed in CALHM1-transfected cells was accompanied by enhanced secretion of sAPPa [9], the product of the non-amyloidogenic processing of APP [23]. In the context of characterizing the functionality of the identified CALHM1 variants, Rubio-Moscardo et al. failed to observe any effect of CALHM1 expression on extracellular Ab accumulation [20]. The authors tested CALHM1-transfected HEK293 cells and used a modified version of the original Ca 2+ add-back protocol, which extended the Ca 2+ omission step to 30 min [20], compared to 1-10 min in the previous studies [9,13,14,21,22]. A more significant difference in Rubio-Moscardo et al. protocol relates to the Ab secretion time period after Ca 2+ add-back: 6 hr [20] vs. 60 min in our protocol [9]. In addition, Rubio-Moscardo et al. performed the Ca 2+ add-back protocol in the presence of 1% FBS, an ingredient that was omitted in our previous study.
In this context, we reassessed the effect of CALHM1 expression on Ab levels not only in N2a cells, but also in HEK293 cells, and by using Ca 2+ add-back conditions that included: (i) a 30 min Ca 2+ omission, (ii) 1% FBS, and (iii) Ca 2+ add-back secretion time periods that extended to 6 hr. In all tested conditions, Ca 2+ addback in CALHM1 expressing cells consistently inhibited the accumulation of extracellular Ab (Fig. 3). Specifically, the amount of time allowed for secretion following Ca 2+ add-back (Figs. 3A and 3C) and the presence of 1% FBS (Figs. 3B and 3D) did not affect CALHM1-mediated inhibition of Ab accumulation. In addition, the previously reported CALHM1-mediated stimulation of sAPPa release [9] was observed in nearly all tested conditions (Fig. 3), with the exception of the N2a cells incubated without FBS (for the 6 hr secretion time point, see Fig. 3A) and at all time points with FBS present (Fig. 3B). These data show that, in different cell lines, CALHM1 expression and its activation by the Ca 2+ add-back condition generate a potent and long-lasting repressing effect on extracellular Ab accumulation. The differences in the Ca 2+ add-back protocol used by Rubio-Moscardo et al. can therefore not explain the disparity in the results observed their study [20]. We note that all transfections performed by Rubio-Moscardo et al. did not include assessment of protein expression efficiency after transfection, raising concerns about the functional expression of CALHM1 in their system. Indeed, Ca 2+ measurements are very sensitive and could have detected low CALHM1 expression (e.g., in the presence of poor transfection efficiency) that would fail to translate into an effect on Ab levels.

Effect of the CALHM1 G330D and R154H variants on Ab levels
To determine whether the R154H and G330D variants affect Ab accumulation, we exposed transfected N2a cells to the 1 hr Ca 2+ add-back protocol, as in Fig. 3A. We observed that G330D-CALHM1 produced the same inhibitory effect on Ab accumulation as WT-CALHM1 (Figs. 4A and 4B). In contrast, expression of R154H-CALHM1 was less effective in reducing Ab levels, compared to WT-CALHM1 (Figs. 4A and 4B). In addition, WT-CALHM1 and G330D-CALHM1 produced comparable stimulation of sAPPa secretion, while there was less sAPPa secreted from cells expressing R154H-CALHM1 (Fig. 4A). These effects on APP metabolism did not result from significant changes in the expression levels of the different CALHM1 mutants (Fig. 4A). These data, which are in line with the effect of the two CALHM1 variants on CALHM1-mediated Ca 2+ influx in mammalian cells (Fig. 2), further demonstrate that the G330D variant has no obvious functional effect on CALHM1, whereas the R154H variant is associated with a partial loss of CALHM1 function.
Previous work has shown that Ca 2+ i disturbances can affect APP processing and Ab production [24,25]. Notably, expression of BACE1 b-secretase is controlled by different Ca 2+ -dependent transcriptional pathways [26,27]. To determine whether CALHM1-mediated Ca 2+ influx influences BACE1 expression,  Effect of CALHM1 Human Variants on Cytosolic Ca 2+ and Ab Levels PLOS ONE | www.plosone.org we measured BACE1 levels by western blotting (WB) in cells transfected with WT or mutated CALHM1. No significant differences in BACE1 levels were observed in these conditions (Fig. 4A), showing that CALHM1 did not affect Ca 2+ -dependent BACE1 expression. Therefore, CALHM1 influenced Ab levels by a mechanism independent of BACE1 transcriptional control.
Together our data show that, while R154H-CALHM1 exhibits normal channel properties at the plasma membrane ( Fig. 1), it impairs the overall control by CALHM1 of Ca 2+ cell uptake (Fig. 2) and APP metabolism (Figs. 3 and 4). These results suggest that R154H does not prevent the formation of a fully functional channel, but instead may act by interfering with CALHM1 trafficking and/or its maturation to reduce the expression levels or stability of CALHM1 at the plasma membrane. The effect of R154H on CALHM1-mediated Ca 2+ influx could also indirectly be influenced by other plasma membrane Ca 2+ channels, such the store-operated Ca 2+ (SOC) channels. Our previous work has failed, however, to observe an effect of CALHM1 activation on SOC entry [14] and SOC channel inhibition did not interfere with CALHM1-mediated Ca 2+ influx [9]. Selective blockage of other common plasma membrane Ca 2+ channels in excitable cells, such as the voltagegated Ca 2+ channels (VGCCs), was also without effect on CALHM1-mediated Ca 2+ influx [9]. Collectively, these data suggest that CALHM1 does not influence SOC channels or VGCCs. The effect of the CALHM1 R154H mutation is thus unlikely to be due to defects in the response of these other channels. Nevertheless, we cannot exclude the possibility at this point that other unidentified pathways of Ca 2+ re-entry might be involved in this mechanism. The effect of the R154H mutation on CALHM1 is therefore likely to be complex and further investigation will be required to determine the precise mechanisms whereby this mutation interferes with CALHM1 function.

Conclusions
In the present work, we used electrophysiology and cell biology approaches to show that: (i) CALHM1 G330D is a frequent variant with no functional consequences on CALHM1 channel properties and is therefore unlikely to contribute to the risk of EOAD; and (ii) CALHM1 R154H is a rare variant that causes a partial loss of CALHM1 function and increases Ab levels in cell lines. These results motivate further investigation aimed at determining whether CALHM1 R154H, or other functional natural variants in CALHM1, is/are associated with EOAD.

Cell lines and transfections
The cell lines used in this study were described previously [13,23].

WB analyses
Cell extracts (5-20 mg, depending on the primary antibody used) were analyzed by SDS-PAGE using the antibodies listed above. Secreted total Ab and sAPPa levels were analyzed by WB, as described previously [28,29]. Briefly, 20 mL of medium was electrophoresed on 16.5% Tris-tricine gels (for Ab WB) or 10% Tris-HCl gels (sAPPa WB) and then transferred onto nitrocellulose membranes. For Ab WB, membranes were microwaved for 5 min in PBS. Membranes were then blocked in 5% fat-free milk in TBS and incubated overnight at 4uC with 6E10 antibody (1:1000 in SuperBlock Blocking Buffer, Thermo Fisher Scientific). A standard ECL detection procedure was then used.

Electrophysiology in Xenopus oocytes
All procedures involving the use of Xenopus laevis were approved by the University of Pennsylvania Institutional Animal Care and Use Committee. Adult female Xenopus laevis obtained from Xenopus One were maintained in existing university facilities. To obtain oocytes, Xenopus laevis were first anaesthetized by immersion in 0.15% tricaine for 15 minutes. Once anaesthetized, frogs were placed supine, a 1.0 cm incision was made in the lower abdominal quadrant through which part of an ovary was raised and removed and then the incision was closed with 4-0 Reli sutures (MYCO Medical). Oocytes were defolliculated with type 2 collagenase (Worthington Biochemical). No Xenopus laevis frogs or embryos were sacrificed for this study.
Standard bath solutions (pH 7.2) contained (in mM) 100 Na + , 100 Cl 2 , 2 K + , and 10 HEPES with 2 Ca 2+ or 0.5 EGTA and 0.5 EDTA to create the divalent cation-free solution. To determine conductance-voltage relations in the presence and absence of Ca 2+ o , currents at 280 mV were measured following a series of voltage pulses, as described before [14,15]. For the permeability experiments, standard bath solution (pH 7.2) contained (in mM) 17 Na + , 10 Cl 2 62 Ca 2+ . Instantaneous I-V recordings were analyzed to determine the reversal potential, as described before [14,15]. Recordings were analyzed with Igor Pro.

Variation allele frequency analysis
We interrogated the NHLBI Exome Sequencing Project (ESP) release ESP6500SI-V2 using the Exome Variant Server [30]. The ESP measured genotypes in a population of 6,503 individuals across several cohorts using an exome-sequencing assay. Exome sequencing makes it possible to observe genomic sequence over the covered exonic regions and to call rare variants in these regions.