The leucine zipper EF‐hand containing transmembrane protein‐1 EF‐hand is a tripartite calcium, temperature, and pH sensor

Abstract

Leucine Zipper EF‐hand containing transmembrane protein‐1 (LETM1) is an inner mitochondrial membrane protein that mediates mitochondrial calcium (Ca²⁺)/proton exchange. The matrix residing carboxyl (C)‐terminal domain contains a sequence identifiable EF‐hand motif (EF1) that is highly conserved among orthologues. Deletion of EF1 abrogates LETM1 mediated mitochondrial Ca²⁺ flux, highlighting the requirement of EF1 for LETM1 function. To understand the mechanistic role of this EF‐hand in LETM1 function, we characterized the biophysical properties of EF1 in isolation. Our data show that EF1 exhibits α‐helical secondary structure that is augmented in the presence of Ca²⁺. Unexpectedly, EF1 features a weak (~mM), but specific, apparent Ca²⁺‐binding affinity, consistent with the canonical Ca²⁺ coordination geometry, suggested by our solution NMR. The low affinity is, at least in part, due to an Asp at position 12 of the binding loop, where mutation to Glu increases the affinity by ~4‐fold. Further, the binding affinity is sensitive to pH changes within the physiological range experienced by mitochondria. Remarkably, EF1 unfolds at high and low temperatures. Despite these unique EF‐hand properties, Ca²⁺ binding increases the exposure of hydrophobic regions, typical of EF‐hands; however, this Ca²⁺‐induced conformational change shifts EF1 from a monomer to higher order oligomers. Finally, we showed that a second, putative EF‐hand within LETM1 is unreactive to Ca²⁺ either in isolation or tandem with EF1. Collectively, our data reveal that EF1 is structurally and biophysically responsive to pH, Ca²⁺ and temperature, suggesting a role as a multipartite environmental sensor within LETM1.

1. INTRODUCTION

Leucine zipper EF‐hand containing transmembrane protein 1 (LETM1) is an inner mitochondrial membrane protein, whose exact function remains a point of contention since its identification in 1999. ¹ , ² , ³ Initially thought to exchange potassium (K⁺) for protons (H⁺) due to increased mitochondrial swelling associated with knockout of the LETM1 yeast homolog MDM38 and subsequent rescue with a K⁺ ionophore, it has now been shown in multiple studies that LETM1 directly exchanges calcium (Ca²⁺) for H⁺. ² , ⁴ , ⁵ , ⁶ The constitutively expressed LETM1 gene was initially identified as part of a hemizygous microdeletion on the small arm of chromosome 4P16.3 found in patients with Wolf‐Hirschhorn syndrome, a rare chromosomal disorder characterized by microcephaly, growth retardation, intellectual disability, and early onset of epileptic seizures. ³ , ⁷ LETM1 deficiency in yeast causes mitochondrial swelling and is lethal under non‐fermentable conditions, and in mouse studies, homozygous LETM1 deletion is embryonically lethal within 6.5 days, indicating that a fundamental importance of LETM1 function is conserved from lower eukaryotes to higher vertebrates. ⁴ , ⁸ Thus, it is not surprising that LETM1 was identified as 1 of approximately 2,000 essential genes by two independent studies. ⁹ , ¹⁰

LETM1 is a 739‐residue type 1 single transmembrane protein that resides in the inner mitochondrial membrane (IMM), with an amino (N)‐terminus facing the intermembrane space (IMS) and carboxyl (C)‐terminus facing the matrix (Figure 1(a)). ¹⁷ Studies employing a Tyr labeling technique and detection via mass spectrometry suggest LETM1 contains a second previously unpredicted transmembrane domain, resulting in both the N‐ and C‐termini being oriented in the matrix. ¹¹ Yet, this proposed second domain is too short to span the membrane as a helix and is not predicted using TMpred, ¹⁸ TMHMM ¹⁹ or Phobius ²⁰ servers. The N‐terminal region contains a coiled‐coil (CC1) and a PTEN‐induced kinase 1 (PINK1) phosphorylation site, ¹⁴ whereas the C‐terminal region contains a ribosome binding domain (RBD), three coiled‐coils (CC2, CC3, CC4), and two conserved putative EF‐hands (Figure 1(a)). Several LETM1 null expression experiments have been published, showing impaired mitochondrial Ca²⁺ and K⁺ flux and disrupted mitochondrial bioenergetics and morphology. ⁴ , ⁵ , ⁸ , ²¹ , ²² , ²³ , ²⁴ Nevertheless, the mechanistic significance of the EF‐hand(s) to LETM1 function has remained unclear and relatively unexamined. As previous studies have shown LETM1 is able to directly facilitate Ca²⁺/H⁺ exchange activity, ² , ⁵ , ⁶ the presence and function of EF‐hand(s) may be integral to either buffering or sensing Ca²⁺ in the regulation of LETM1 antiporter activity.

FIGURE 1.

Domain architecture and sequence alignments of LETM1 proteins. (a) Domain architecture of human LETM1. The relative locations of the mitochondrial targeting sequence (MTS; blue), coiled‐coil 1, 2, 3, 4 (CC1/2/3/4, red), transmembrane 1, 2 (TM1/2, green), ¹¹ , ¹² ribosome binding domain (RBD, magenta) and EF‐hand motif 1, 2 (EF1/2, yellow) ³ , ¹³ are shown. The position of the PINK1 Thr192 phosphorylation site is indicated (cyan sphere). ¹⁴ The residue ranges are shown directly above and below each domain, labeled based on UNIPROT annotations and previous publications, as indicated ³ , ¹¹ , ¹² , ¹³ , ¹⁴ . At top, the orientation of each domain relative to the IMM are shown for 1 × TM (solid lines) and 2 × TM (dashed lines) LETM1 architectures. (b) Multiple sequence alignment of the putative EF‐hand motif containing region of several LETM1 orthologues. Sequences for human (NCBI accession NP_036450.1), mouse (AAH61115.1), fruit fly (NP_726453.1), roundworm (CAB03150.1), thale cress (OAP05849.1); chimpanzee (JAA40971.1), bovine (AAI20275.1), wild boar (NP_001231877.1), frog (AAI21319.1) and zebrafish (NP_001038673.1) were aligned using Clustal Omega. ¹⁵ Residue conservation is shown below the alignment as fully conserved (*), highly conserved (:) or partially conserved (.). For each sequence, helix (orange) and β‐strand (yellow) predictions by PSIPRED ¹⁶ are shown. The locations of the EF‐hand binding loop sequences derived from previous publications ³ , ¹³ are underlined. At top, the EF‐hand constructs engineered in the present study are indicated. The location of the Phe (red box) used to rationalize the EF1_Y668ins mutant (red arrow) is indicated

To begin teasing out the mechanistic role of the EF‐hand(s) in human LETM1 function, we characterized the biophysical and biochemical properties of the putative EF‐hand motifs in isolation and tandem. Unexpectedly, we found that the sequence identifiable EF‐hand (EF1) binds Ca²⁺ with low affinity (i.e., high equilibrium dissociation constant, K_d). In contrast, a second LETM1 region, previously suggested to be an EF‐hand but lacking a consensus loop sequence, was unresponsive to Ca²⁺. Remarkably, EF1 showed robust structural, conformational and stability sensitivity to changes in pH (6.0–8.5) as well as both high (> 37°C) and low (< 37°C) temperature. Specifically, small changes in these environmental conditions not only altered the Ca²⁺ binding affinity, secondary structure levels and thermal stability, but also modulated solvent exposed hydrophobicity and oligomerization. Collectively, our data reveal that EF1 has a non‐mutually exclusive tripartite sensing ability, suggesting multiple different environmental inputs may regulate LETM1 function via integrated EF1‐mediated sensing coupled with altered protein–protein interactions.

2. RESULTS

We used PSIPRED ¹⁶ to predict the distribution of secondary structure components around the sequence identifiable canonical EF1 loop. As expected, PSIPRED revealed helices upstream and downstream of the 12‐residue loop; however, a second, shorter helix, immediately upstream of the putative entering helix was also predicted (Figure 1(b)). Given that EF‐hand motifs are typically found in pairs that are adjacent in sequence space, the absence of any sequence identifiable second EF‐hand motif in human LETM1 and proximity of this predicted helix to the EF1 entering helix, we generated a construct encompassing both helices preceding the loop and the putative exiting helix (i.e., residues 643–699). Indeed, we were able to express and isolate this EF1 protein to high levels and purity. In contrast, a construct encompassing only the single putative entering and exiting helices (i.e., residues 664–699; Figure 1(b)) did not yield appreciable protein under similar expression and purification conditions.

2.1. Ca²⁺ binding, pH and temperature modulate EF1 folding and stability

Far‐UV circular dichroism (CD) spectroscopy was used to evaluate the secondary structure of EF1 at varying pH, temperatures and Ca²⁺ levels. At pH 7.8, representing the slightly alkaline environment of the mitochondrial matrix where this EF‐hand motif resides, ²⁵ physiological temperature (i.e., 37°C) and without Ca²⁺, EF1 displayed mean residue ellipticity (MRE) minima at ~205 and 222 nm, consistent with α‐helical structure. Unexpectedly, decreasing the temperature to 20°C and 4°C resulted in a systematic loss in α‐helicity, with the 4°C spectrum exhibiting a single intense minimum at ~200 nm, indicative of random coil (Figure 2(a)‐(c)). The addition of Ca²⁺ promoted α‐helical structure, evidenced by more intense negative ellipticity at ~208 and 222 nm, at all three temperatures (Figure 2(d)‐(f)). The specific increase in α‐helicity upon Ca²⁺ addition, evaluated as the change in MRE at 222 nm, was most pronounced at 4°C (Figure 2(g)‐(i)). Given possible pH fluctuations in the matrix of respiring mitochondria, ²⁵ , ²⁶ we next characterized the effect of pH on EF1 secondary structure. At pH 8.5 and in the absence of Ca²⁺, EF1 showed very similar spectra as observed at pH 7.8. In contrast, the spectra acquired at pH 6.0 not only showed more negative ellipticity at both minima but also a rightward shift of the first minimum toward ~208 nm, indicating increased α‐helicity at lower pH (Figure 2(a)‐(c)). This robust pH effect was observed at all temperatures in the absence of Ca²⁺. In the presence of Ca²⁺, there was no significant difference in the spectra at pH 6.0, 7.8 and 8.5, at any temperature (Figure 2(d)‐(f)).

FIGURE 2.

Secondary structure and thermal stability of EF1. Far‐UV CD spectra are shown for EF1 buffered at pH 8.5 (green), pH 7.8 (blue) and pH 6.0 (red) in the absence of Ca²⁺ at (a) 4°C, (b) 20°C and (c) 37°C, and in the presence of 30 mM Ca²⁺ at (d) 4°C, (e) 20°C and (f) 37°C. As an indicator of α‐helicity, a comparison of the EF1 222 nm MRE signals at pH 8.5, 7.8 and 6.0 in the absence (red) and presence (blue) of 30 mM Ca²⁺ at (g) 4°C, (h) 20°C and (i) 37°C are shown. Thermal melts based on the fractional change in MRE at 222 nm as a function of temperature in the absence (red) and presence (blue) of 30 mM Ca²⁺ are shown at (j) pH 6.0, (k) pH 7.8 and (l) pH 8.5. All Far‐UV CD experiments were performed using 30 μM protein. Data are means ± SEM of n = 3 separate protein expression samples. Different lowercase letters indicate means that are significantly different at p < .05, following a two‐way ANOVA and Tukey's post‐hoc test. In J‐L, the fit to a two‐state native to unfolded model is shown (black solid line)

Having observed a clear temperature sensitivity in the secondary structure of EF1, we next assessed the thermal stability by monitoring the change in MRE at 222 nm as a function of temperature. The thermal melts acquired in the absence of Ca²⁺ confirmed a remarkable ability of EF1 to cold unfold at temperatures below ~37°C; moreover, this cold unfolding was similarly observed at pH 6.0, 7.8 and 8.5. Ca²⁺‐free EF1 at all pH conditions also exhibited heat denaturation at temperatures above ~42°C (Figure 2(j)‐(l)). Addition of Ca²⁺ to the samples drastically stabilized EF1, indicated by decreased and increased apparent cold and hot unfolding midpoints at all pH values, respectively. In the presence of Ca²⁺, cold unfolding was not clearly visible between 20°C and 37°C; further, a hot denaturation baseline was not observed between 37°C and 95°C at any pH (Figure 2(j)‐(l)). Since denaturation at high and low temperature was fully reversible, we fit the Ca²⁺‐free thermal melts to two‐state equilibrium unfolding models, extracting midpoints of temperature denaturation (T_m), specific heat capacity of unfolding (∆C_p), enthalpy of unfolding (ΔH) and thus, Gibbs free energy of unfolding (∆G) (Table 1). This thermodynamic analysis revealed the highest ∆G and T_m for the pH 6.0 condition, consistent with the significantly increased α‐helicity levels observed in our far‐UV CD spectra. Thermodynamic parameters were not reliably extracted from the Ca²⁺ loaded datasets due to the absence of discernable baselines. Nevertheless, we took the temperature where the fractional change in ellipticity (222 nm) was 0.5 (T_0.5) as an indicator of stability, finding similar apparent stabilities at all pH conditions (Table 1).

TABLE 1.

Summary of the LETM1 EF1 thermal and thermodynamic stability parameters

	pH 6.0		pH 7.8		pH 8.5
	‐Ca2+	+Ca2+	‐Ca2+	+Ca2+	‐Ca2+	+Ca2+
aTm/ b T0.5 (°C)	56.6 ± 0.2	75.3 ± 0.9	55.7 ± 0.2	73.3 ± 0.3	54.8 ± 0.3	71.7 ± 0.2
aΔCp (cal·mol‐1·K‐1)	919 ± 7.8		1236 ± 6.9		1181 ± 10.5
aΔH (kcal·mol‐1)	14.9 ± 0.3		16.4 ± 0.3		16.2 ± 0.5
cΔGunfolding (cal·mol‐1)	‐4891 ± 282		‐8312 ± 316		‐7033 ± 460

Note: Data are means ± SEM of n=3 separate protein expression samples.

^aT_m, ΔCp and ΔH were extracted from the Ca²⁺‐free thermal melts using a two‐state equilibrium unfolding model.

^bT_0.5 was extracted from the Ca²⁺‐loaded thermal melts as the temperature where the fractional change in ellipticity was 0.5.

^cΔG was calculated at a reference temperature of 20°C using the Gibbs‐Helmholtz equation.

Collectively, these data reveal that in the absence of Ca²⁺, physiological relevant deviations from pH 7.8 and thermal variations from 37°C modulate folding of EF1; further, in the presence of Ca²⁺, folding and stability are equalized across pH conditions.

2.2. EF1 exhibits weak but selective Ca²⁺ binding affinity that is influenced by pH and temperature

Given that Ca²⁺ induced robust folding and enhanced the stability of EF1, we next used the changes in the MRE at 222 nm as a function of increasing Ca²⁺ to estimate the binding affinity. Under each pH condition at 20°C, the far‐UV CD spectra exhibited systematically more negative ellipticity with increasing Ca²⁺ concentration, which began to saturate at ~10–15 mM. Fitting the relative change in MRE versus total Ca²⁺ concentration to a single binding site model that accounts for protein concentration, revealed apparent equilibrium dissociation constants (K_d) between ~2–4 mM. Interestingly, the apparent affinity was significantly lower (i.e., higher K_d) at pH 6.0 compared to pH 7.8 and 8.5 (Figure 3(a), (b); Table 2). Additionally, consistent with the cold denaturation phenomenon, systematic decreases in K_d (i.e., increases in affinity) were observed with increasing temperature to 37°C, assessed at pH 7.8 (Figure 3(c), (d); Table 2).

FIGURE 3.

Ion binding affinity of EF1. (a) Far‐UV CD spectra of EF1 in the absence and presence of Ca²⁺ at pH 8.5, 7.8 and 6.0 (20°C). (b) Relative change in MRE at 222 nm as a function of increasing Ca²⁺ at pH 8.5, 7.8 and 6.0 (20°C). (c) Far‐UV CD spectra of EF1 in the absence and presence of Ca²⁺ at 4, 20 and 37°C (pH 7.8). (d) Relative change in MRE at 222 nm as a function of increasing Ca²⁺ at 4, 20 and 37°C (pH 7.8). (e) Far‐UV CD spectra of EF1 in the absence and presence of Ca²⁺, Mn²⁺ and Mg²⁺ (pH 7.8, 20°C). (f) The relative change in MRE at 222 nm as a function of increasing Ca²⁺, Mg²⁺ and Mn²⁺ (pH 7.8, 20°C). (g) Intrinsic Tyr fluorescence spectra of EF1_Y668ins in the absence and presence of Ca²⁺, Mg²⁺ and Mn²⁺ (pH 7.8, 20°C). (h) Relative change in peak intrinsic Tyr fluorescence intensity as a function of increasing Ca²⁺, Mg²⁺ and Mn²⁺ (pH 7.8, 20°C). In all Far‐UV CD and intrinsic Tyr fluorescence experiments, 30 and 74 μM protein was used, respectively. All binding curves were fitted to a one‐site binding model that accounts for protein concentration. In B, D, F and H, representative binding curves are shown. Spectra are means ± SEM of n = 3 separate protein expression samples

TABLE 2.

Summary of divalent cation binding affinities for LETM1 EF‐hand containing constructs

LETM1 construct	divalent cation	pH	temperature (°C)	probe	salt	aKd (mM)	bmean Kd (mM)
EF1 WT	Ca2+	6.0	20	CD	NaCl	5.63 ± 1.17	4.44 ± 1.1
						3.49 ± 0.32
						4.20 ± 0.61
EF1 WT	Ca2+	6.0	20	ANS fluorescence	NaCl	42.92 ± 6.72	61.39 ± 16.5
						74.74 ± 10.69
						66.52 ± 6.12
EF1 WT	Ca2+	7.8	4	CD	NaCl	8.99 ± 1.20	9.52 ± 0.5
						9.68 ± 1.38
						9.89 ± 1.48
EF1 WT	Ca2+	7.8	20	CD	NaCl	2.67 ± 0.49	2.46 ± 0.2
						2.39 ± 0.41
						2.32 ± 0.34
EF1 WT	Ca2+	7.8	20	ANS fluorescence	NaCl	12.7 ± 0.89	13.24 ± 0.9
						14.26 ± 1.54
						12.75 ± 1.20
EF1 WT	Ca2+	7.8	37	CD	NaCl	1.45 ± 0.16	1.37 ± 0.1
						1.33 ± 0.16
						1.34 ± 0.17
EF1 WT	Ca2+	7.8	20	CD	KCl	1.84 ± 0.22	1.79 ± 0.2
						1.57 ± 0.12
						1.95 ± 0.19
EF1 WT	Ca2+	8.5	20	CD	NaCl	2.55 ± 0.34	2.38 ± 0.2
						2.25 ± 0.28
						2.35 ± 0.26
EF1 WT	Ca2+	8.5	20	ANS fluorescence	NaCl	14.36 ± 1.52	14.75 ± 0.8
						15.65 ± 2.32
						14.25 ± 1.98
EF1 WT	Mg2+	7.8	20	CD	NaCl	22.78 ± 2.39	23.50 ± 2.1
						25.88 ± 2.05
						21.84 ± 2.18
EF1 WT	Mg2+	7.8	20	ANS fluorescence	NaCl	69.87 ± 5.46	59.41 ± 10.6
						59.66 ± 3.37
						48.70 ± 2.79
EF1 WT	Mn2+	7.8	20	CD	NaCl	1.04 ± 0.04	0.91 ± 0.1
						0.83 ± 0.04
						0.86 ± 0.07
EF1Y668ins	Ca2+	7.8	20	Tyr fluorescence	NaCl	4.54 ± 1.17	5.70 ± 1.1
						5.87 ± 1.44
						6.69 ± 1.80
EF1Y668ins	Mg2+	7.8	20	Tyr fluorescence	NaCl	933 ± 6965	397 ± 470
						199 ± 213
						59 ± 105
EF1Y668ins	Mn2+	7.8	20	Tyr fluorescence	NaCl	35.08 ± 93.68	16.22 ± 16.7
						3.14 ± 0.67
						10.44 ± 5.33
EF1Asp687Glu	Ca2+	7.8	20	CD	NaCl	0.55 ± 0.03	0.48 ± 0.1
						0.44 ± 0.02
						0.44 ± 0.02
EF2‐EF1 WT	Ca2+	7.8	4	CD	NaCl	15.51 ± 3.36	14.58 ± 2.0
						12.25 ± 4.87
						15.99 ± 8.41

^aErrors (±) are standard error of the non‐linear regression fits.

^bErrors (±) are standard deviation of the n=3 K_d measurements.

Given the weak apparent binding affinity of the LETM1 EF1 to Ca²⁺, we next tested the binding affinity to magnesium (Mg²⁺) and manganese (Mn²⁺) ions, which are also present in mitochondria. Using far‐UV CD, EF1 showed an increase in α‐helical content in the presence of Mn²⁺, similar to Ca²⁺. Further, at pH 7.8 and 20°C, the apparent Mn²⁺ binding K_d was ~1 mM, somewhat lower than the K_d estimated for Ca²⁺ (Figure 3(e), (f); Table 2). In contrast, Mg²⁺ did not appreciably change the far‐UV CD spectrum of EF1, and a systematic Mg²⁺ titration revealed an estimated K_d of ~24 mM (Figure 3(e), (f); Table 2). Finally, given the possibility that LETM1 is involved in mediating K⁺ homeostasis, we replaced the sodium chloride (NaCl) in our buffer with potassium chloride (KCl), and repeated the Ca²⁺ titration. The KCl containing buffer slightly increased the weak Ca²⁺ binding affinity compared to NaCl at pH 7.8 and 20°C (i.e., K_d = 1.8 versus 2.5 mM) (Figure S1A, S1B; Table 2).

Since the human LETM1 EF1 contains no aromatic residues, we next engineered a Lys667_Leu668insTyr insert mutant (EF1_Y668ins) to enable intrinsic fluorescence measurements. The Tyr was inserted in the entering helix of EF1, guided by the similar insertion of a Phe in the entering helix of the wild‐type Drosophila LETM1 EF‐hand (Figure 1(b)). Tyr was selected due to the structural similarity to Phe but higher quantum fluorescence yield. EF1_Y668ins exhibited a similar far‐UV CD spectrum and increase in α‐helicity as observed with EF1 when incubated with Ca²⁺ (Figure S2). Using an excitation wavelength of 276 nm, EF1_Y668ins exhibited an intrinsic fluorescence emission maximum of 305 nm. Addition of Ca²⁺ increased the fluorescence intensity and slightly blue shifted the peak maximum by ~2 nm (Figure 3(g)). The increased fluorescence is consistent with exposure of the Tyr to a more polar environment, assuming no local charge interactions. A plot of the change in fluorescence at 303 nm, representing the wavelength showing the greatest change in intensity in the free and bound states, as a function of increasing Ca²⁺ indicated saturable binding with an apparent K_d of ~6 mM (Figure 3(h); Table 2). Interestingly, the EF1_Y668ins appeared to show a more cooperative binding curve (i.e., sigmoidal) compared to our CD evaluations. Similar fluorescence titrations were performed using Mg²⁺ and Mn²⁺. As expected, Mg²⁺ did not appreciably alter the fluorescence emission spectrum of EF1_Y668ins. Unexpectedly, Mn²⁺ also minimally effected the spectrum, despite the more robust change observed in secondary structure (Figure 3(e)). The estimated K_d values extracted from the fluorescence data for the Mn²⁺ and Mg²⁺ were > 16 mM (Table 2). Thus, while Mn²⁺ appears to promote α‐helicity in EF1, the arrangement of the helices (i.e., tertiary structure) is minimally impacted by this cation.

Collectively, these data suggest that both secondary and tertiary structural changes in EF1 are mediated by selective binding of Ca²⁺ over either Mn²⁺ or Mg²⁺; moreover, pH and temperature significantly affect the ~mM binding affinity of EF1.

2.3. Ca²⁺ binding increases solvent‐exposed hydrophobicity of the LETM1 EF1

Having observed that Ca²⁺ binding to EF1_Y668ins potentially causes conformational changes that increase the solvent exposure of the entering helix, we next assessed the more global solvent exposed hydrophobicity of EF1 in response to Ca²⁺ using 1‐anilinonaphthalene‐8‐sulfonate (ANS). ANS binds to solvent exposed hydrophobic regions on proteins resulting in a blue shifted fluorescence emission maximum wavelength with increased intensity. ²⁷ In the absence of Ca²⁺, EF1 at pH 6.0 showed markedly higher levels of ANS binding and fluorescence compared to pH 7.8 and 8.5, suggesting more solvent exposed hydrophobicity (Figure 4(a)‐(c)). Addition of Ca²⁺ resulted in a systematic increase in fluorescence concomitant with a blue shift in the wavelength (~5 nm) of the peak maximum (475 nm) at all pH values tested. At 110 mM Ca²⁺, EF1 showed similar ANS fluorescence intensities at pH 6.0, 7.8 and 8.5 (Figure 4(a)‐(d)). Nevertheless, plots of the change in ANS fluorescence versus Ca²⁺ implied a weaker Ca²⁺ binding for pH 6.0 compared to pH 7.8 and 8.5. Note that the K_d extracted in this manner reflects both Ca²⁺ and ANS binding events and thus, cannot be directly compared with our CD and intrinsic fluorescence measurements. Qualitatively, however, the tighter apparent Ca²⁺ binding observed at pH 7.8 and 8.5 compared to pH 6.0 using the ANS assessment is consistent with the rank order observed by CD. Unexpectedly Mg²⁺ also increased fluorescence; however, the change in ANS fluorescence was markedly lower compared to the addition of Ca²⁺ (Figure S3A, S3B).

FIGURE 4.

ANS binding to EF1. Extrinsic ANS fluorescence emission spectra acquired in the presence of EF1 as a function of increasing Ca²⁺ (0–110 mM) at (a) pH 8.5, (b) pH 7.8 and (c) pH 6.0. (d) Comparison of the mean peak ANS fluorescence intensities in the absence (red) and presence (blue) of 110 mM Ca²⁺ for each pH condition. Different lowercase letters indicate means that are significantly different at p < .05, following a two‐way ANOVA and Tukey's post‐hoc test. In A‐C, insets show representative plots of peak ANS fluorescence intensities (475 nm) versus Ca²⁺ fit to a one site binding model that accounts for protein concentration. Titrations were performed using 30 μM of protein at 20°C. Spectra are means ± SEM of n = 3 separate protein expression samples

Collectively, the extrinsic ANS fluorescence data indicate that Ca²⁺ binding causes conformational changes in EF1 that increase solvent exposed hydrophobicity, typical of other Ca²⁺ binding EF‐hands and qualitatively show that Ca²⁺ binding is tighter at alkaline pH.

2.4. The LETM1 EF1 undergoes Ca²⁺‐dependent higher order oligomerization

Given that Ca²⁺ binding EF‐hand motifs are typically found in pairs, stabilized by inter‐strand hydrogen bonding between short β‐strands located within the loop of each motif, ²⁸ we next assessed whether EF1 may homo‐dimerize as a pairing mechanism. We evaluated the molecular weight of EF1 prepared at pH 7.8 and 6.0 by size exclusion chromatography with in‐line multi angle light scattering (SEC‐MALS). In the absence of Ca²⁺ and at pH 7.8, EF1 exists primarily as a monomer with an estimated molecular weight of 7.9 ± 0.3 kDa, similar to the theoretical EF1 monomer molecular weight of 6.7 kDa (Figure 5(a); Table 3). At pH 6.0, the stoichiometry of Ca²⁺‐free EF1 shifted toward dimer with a molecular weight of 10.3 ± 0.1 kDa (Figure 5(b); Table 3). The Ca²⁺‐free molecular weights at both pH 7.8 and 6.0 did not show a strong protein concentration dependence in the 4–8 mg/ml range (Table 3). Remarkably, addition of Ca²⁺ caused LETM1 EF1 to oligomerize. At pH 6.0, these homo‐oligomers ranged in molecular weight from trimer (i.e., 20.4 ± 0.3 kDa) to tetramer (i.e., 26.2 ± 0.1 kDa), at injection concentrations between 4–8 mg/ml, respectively (Figure 5(c); Table 3). At pH 7.8, the molecular weights ranged from trimer (i.e., 20.8 ± 0.6 kDa) to pentamer (i.e., 32.8 ± 0.7 kDa), at injection concentrations between 4–8 mg/ml, respectively (Figure 5(d); Table 3).

FIGURE 5.

Quaternary structure of EF1. SEC with in‐line MALS data of EF1 acquired in the absence of Ca²⁺ at (a) pH 7.8 and (b) pH 6.0. SEC with in‐line MALS data of EF1 acquired in the presence of 30 mM Ca²⁺ at (c) pH 6.0 and (d) pH 7.8. In A‐D, 100 μl EF1 was injected onto an S200 10/300 GL column at 8.0 (green), 6.0 (blue) and 4.0 (red) mg/ml, and data were collected at ~10°C. Insets show Coomassie‐blue stained SDS‐PAGE gels of the elution fractions from the 8.0 mg ml⁻¹ injections

TABLE 3.

Summary of LETM1 EF1 self‐association stoichiometry determined by SEC‐MALS

		‐Ca2+					+Ca2+
pH	concentration (mg·ml‐1)	molecular weight (kDa)	amean molecular weight (kDa)	bstoichiometry	pH	concentration (mg·ml‐1)	molecular weight (kDa)	bstoichiometry
6.0	8.0	10.5 ± 0.3	10.3 ± 0.2	1.5	6.0	8.0	26.2 ± 0.1	3.9
	6.0	10.0 ± 0.3				6.0	21.6 ± 0.1	3.2
	4.0	10.4 ± 0.6				4.0	20.4 ± 0.3	3.0
7.8	8.0	8.5 ± 0.5	7.9 ± 0.3	1.2	7.8	8.0	32.8 ± 0.7	4.9
	6.0	7.9 ± 0.1				6.0	27.1 ± 0.3	4.0
	4.0	7.4 ± 0.4				4.0	20.8 ± 0.6	3.1

^aErrors (±) are standard error of the means from n=3 separate SEC‐MALS experiments.

^bStoichiometry calculated as SEC‐MALS molecular weight divided by the theoretical weight of the EF1 monomer (i.e., 6.7 kDa).

Collectively, the SEC‐MALS data demonstrate that, in isolation and absence of Ca²⁺, EF1 has a low tendency to oligomerize; however, Ca²⁺ binding induces a robust protein concentration‐dependent higher order oligomerization.

2.5. The LETM1 EF1 coordinates Ca²⁺ in the canonical loop

Given the apparently low Ca²⁺ binding affinity of EF1 in isolation, we tested whether a second region within LETM1, previously suggested to contain an EF‐hand loop, ³ , ¹³ although non‐canonical, could cooperatively enhance Ca²⁺ binding of EF1. Cooperative Ca²⁺ binding at multiple EF‐hand motifs is a hallmark of calmodulin and other archetypal Ca²⁺ sensors. ²⁹ First, we expressed and purified this EF2, corresponding to LETM1 residues 537–634. At 4, 20 and 37°C and pH 7.8, this EF2 construct showed a far‐UV CD spectrum with a low level of negative ellipticity at ~222 nm and a strongly negative minimum at ~201 nm, indicating nominal α‐helicity mixed with primarily random coil. Addition of Ca²⁺ had no effect on either the far‐UV CD or intrinsic Tyr fluorescence emission spectra (Figure S4A, S4B). Next, we created a tandem construct encompassing both EF1 and EF2 (i.e., residues 537–699) (Figure 1(b)). In the absence of Ca²⁺, the tandem EF2‐EF1 exhibited a far‐UV CD spectrum with minima at ~204 and 222 nm; moreover, addition of Ca²⁺ caused the minima to move to ~207 and 222 nm with enhanced negative ellipticity at the latter wavelength (Figure S5A). Monitoring change in ellipticity at 222 nm and 4°C as a function of Ca²⁺ revealed a K_d of ~15 mM, indicating no Ca²⁺ binding cooperativity and consistent with the lack of Ca²⁺ sensitivity observed for isolated EF2 (Figure S5B). The weak affinity is in‐line with the ~10 mM K_d observed for isolated EF1 at 4°C (Table 2). Cold unfolding was not observed in the EF2 isolate but was present in our tandem EF2‐EF1 construct (Figure S6A, S6B).

Having found no evidence for cooperative Ca²⁺ binding in the EF2‐EF1 context, we next engineered an Asp687Glu mutant and probed whether the Asp at this position 12 of the EF1 binding loop underlies the relatively weak Ca²⁺ binding affinity. Archetypal Ca²⁺ binding motifs typically have a Glu at position 12 because the longer side chain permits two‐oxygen ligand coordination for Ca²⁺, in contrast to the shorter Asp that may only provide one‐oxygen ligand due to a rotation and stretch required to reach Ca²⁺. ²⁸ The wild‐type EF1 and Asp687Glu EF1 showed similar far‐UV CD spectra in the absence and presence of Ca²⁺, indicating that the mutation does not perturb the Ca²⁺‐mediated secondary structural changes of the motif (Figure 6(a)). Nevertheless, the change in CD signal began to saturate at ~5 mM CaCl₂ in a titration with Asp687Glu EF1, in contrast to the ~10–15 mM observed with wild‐type EF1. Fitting a one‐site binding model to the binding curve revealed a K_d of 0.5 ± 0.1 mM, ~4‐fold lower than our measurements for wild‐type EF1 (Figure 6(b); Table 2).

FIGURE 6.

Mechanism of Ca²⁺ coordination by EF1. (a) Far‐UV CD spectra of the Asp687Glu mutant (blue) and wild‐type EF1 (red) constructs in the absence (dashed line) and presence (solid line) of 30 mM Ca²⁺ (pH 7.8, 20°C). (b) Relative change in MRE at 222 nm as a function of increasing Ca²⁺ for Asp687Glu (blue) and wild‐type (red) EF1 samples (pH 7.8, 20°C). Inset shows the titrations with a logarithmic Ca²⁺ concentration scale. (c) ¹H‐¹⁵N HSQC of EF1 in the absence (red) and presence of 30 mM Ca²⁺ (blue) (pH 7.8, 35°C). The unique downfield shifted Gly681 ¹H(¹⁵N) amide of Ca²⁺‐loaded EF1 is indicated. Data in B are representative curves fit to a one‐site binding model that accounts for protein concentration. Far‐UV CD spectra are means ± SEM of n = 3 separate protein expression samples

Given the weak Ca²⁺ binding affinity of EF1 mediated by the Asp at position 12 of the loop, we next used solution nuclear magnetic resonance (NMR) spectroscopy to ascertain whether the EF1 loop encircles the Ca²⁺ ion in a canonical manner. Canonically, the Gly at position 6 of the loop permits a near 90° turn enabling all seven Ca²⁺ binding ligands to take up coordinating positions around the Ca²⁺ ion, resulting in a very distinct Gly amide N(H) chemical environment. We prepared uniformly ¹⁵N‐labeled EF1 and acquired ¹H‐¹⁵N heteronuclear single quantum coherence (HSQC) spectra in the absence and presence of Ca²⁺. Indeed, we observed a unique, Ca²⁺‐dependent and downfield‐shifted ¹⁵N(¹H) peak at ~10.6 ppm in the ¹H dimension and 112.5 ppm in the ¹⁵N dimension, consistent with canonical positioning of Gly681 relative to Ca²⁺ within the EF1 loop (Figure 6(c)). We also note that Ca²⁺‐loaded EF1 exhibited ¹H(¹⁵N) peak dispersion between ~10.6–7.0 ppm (excluding side chains), in contrast to Ca²⁺‐free EF1, showing most ¹H(¹⁵N) peaks clustered in the ~8.5–7.3 ppm region in the ¹H dimension; moreover, the differences in peak dispersion highlights the enhanced protein folding induced by Ca²⁺ binding to EF1.

Collectively, the LETM1 EF‐hand construct variants, EF1 mutant and structural assessments indicate that EF1 does not cooperatively bind Ca²⁺ with a previously proposed EF2, and the weak Ca²⁺ binding affinity of EF1 is, at least in part, due to a non‐ideal Asp at position 12 despite a canonical‐like turn adopted by the EF1 loop when bound to Ca²⁺.

3. DISCUSSION

Previous work showed deletion of the EF1 loop (i.e., residues 676–687) within LETM1 decreased mitochondrial Ca²⁺ flux, emphasizing that EF1 plays an important functional role in LETM1 Ca²⁺/H⁺ antiporter activity. ²¹ Nevertheless, there remains a major knowledge gap concerning the precise mechanistic role of LETM1 EF1, which is the only sequence identifiable EF‐hand among vertebrate LETM1 homologues. Therefore, here we isolated, biophysically, biochemically and structurally characterized EF1 from human LETM1. We found that this EF‐hand motif binds Ca²⁺ with low affinity and not only shows a robust structural and stability sensitivity in response to binding but also pH and temperature. Remarkably, low pH increases while low temperature decreases folding and stability of EF1 in the absence of Ca²⁺. Additionally, we discovered that EF1 exhibits enhanced solvent‐exposed hydrophobicity upon Ca²⁺ binding, which may underlie a strong Ca²⁺‐dependent oligomerization. Finally, we found that, while Ca²⁺ binding is weak due to a non‐ideal Asp in position 12 of the binding loop, the loop adopts a canonical‐like conformation when coordinating Ca²⁺.

Canonically, a common consequence of Ca²⁺ binding to an EF‐hand motif is the conformational change of the entering and exiting helices that sandwich Ca²⁺ in the chelating loop. These helices shift from a parallel (closed) to a perpendicular (open) arrangement, as the exiting helix is kicked out to accommodate Ca²⁺ in the loop. ³⁰ The open helical arrangement exposes a hydrophobic cleft, which mediates protein–protein interactions with various targets. Consistent with this canonical mechanism, human LETM1 EF1 exhibits an enhanced solvent‐accessible hydrophobicity when bound to Ca²⁺, as revealed by our ANS binding and intrinsic Tyr fluorescence analyses (Figures 3, 4). Further, using SEC‐MALS, we found Ca²⁺ promotes self‐association of EF1, suggesting a hydrophobicity‐mediated assembly (Figure 5). EF1 remains monomeric in the absence of Ca²⁺; however, in the presence of Ca²⁺ EF1 forms dimers up to pentamers (Table 3). Penta‐EF‐hand family proteins, which contain an odd number of EF‐hands, homodimerize via interactions between unpaired EF‐hands. ³¹ As LETM1 is reported to contain only one or two transmembrane domains per polypeptide chain (Figure 1), oligomerization is likely necessary to create an ion permeation path(s) for Ca²⁺/H⁺ antiporter function. The bacterial Ca²⁺/H⁺ antiporter (ChaA) forms a homotrimer requiring six transmembrane helices, two from each monomer to form the Ca²⁺/H⁺ pathway, and the Ca²⁺/Na⁺ exchanger family, including the mitochondrial Na⁺/Li⁺/Ca²⁺ exchanger (NCLX), functions with ten transmembrane helices. ³² , ³³ Indeed, previous studies have shown LETM1 self‐associates into a hexamer to reconstitute Ca²⁺/H⁺ antiporter function. ⁵ , ⁶ Further, LETM1 co‐precipitates with BCS1 homolog, ubiquinol‐cytochrome c reductase complex chaperone (BSC1L), a mitochondrial chaperone suggested to promote oligomerization of the LETM1 complex in cellulo. ³⁴ Thus, EF1 may act as an intrinsic Ca²⁺‐dependent driver of oligomerization necessary for LETM1 function and/or may be involved in heteromeric interactions.

Nevertheless, we also discovered several unique properties of the human LETM1 EF1. First, the Ca²⁺ binding affinity is relatively low with K_d values in the mM range (Figure 3; Table 2). Work on an isolated Troponin‐C EF‐hand motif showed that Ca²⁺‐binding to one motif can promote dimerization and folding of a second Ca²⁺‐free motif before Ca²⁺ binding to the second motif, with the two Ca²⁺ binding sites exhibiting ~300‐fold differing affinities. ³⁵ Here, we found minimal protein concentration‐dependent changes in apparent Ca²⁺ affinities, and thus fitting protein concentration‐dependent binding data to complex three and four‐state models failed to converge on multiple dissociation constants; moreover, the midpoints of Ca²⁺ saturation (Figure 3) are in‐line with the K_d values extracted using our simple two state model (Table 2), collectively suggesting our low Ca²⁺ binding affinity estimates for LETM1 EF1 are reliable.

To gain insight into possible mechanisms underlying the low affinity, we searched the Protein Data Bank for EF‐hand structures with similar sequences to LETM1 EF1. Pond snail caltubin (6VAN.pdb) contains an EF‐hand loop (EF4) with 58% identity and 92% similarity to the human LETM1 EF1 loop (i.e., residues 676–687). The caltubin EF4 loop structure shows pentagonal bipyramidal coordination to Sr²⁺; however, rather than bridging a metal‐coordinating water molecule via the ninth (−X) loop position, the hydroxyl oxygen of Asp at position 5 (Z) bridges the water (Figure S7A, S7B). Thus, Asp at position 5 (Z) provides two oxygen ligands for the coordination: the carbonyl oxygen (Oδ1) directly and the hydroxyl oxygen (Oδ2) indirectly to the Sr²⁺ via a water bridge. Position 9 (−X) of caltubin EF4 does not appear to be involved in metal ion coordination, which is unusual for canonical Ca²⁺ coordinating EF‐hands. ²⁸ , ³⁰ While caltubin contains a Glu at position 12 (−Z), human LETM1 EF1 contains an Asp at this position (Figure S7C). Position 12 (−Z) typically exists as a Glu in Ca²⁺ binding EF‐hands, which confers an appropriate reach relative to the backbone for direct two side‐chain oxygen coordination of Ca²⁺. ²⁸ Given the caltubin EF4 loop structure and sequence deviations from LETM1 EF1, the low Ca²⁺ binding affinity of LETM1 EF1 may be due to a combination of an Asp at position 12 (−Z), acting as a monodentate ligand rather than bidentate, and no coordination by Asn at position 9 (−X) of the loop, resulting in only one of the final three typically Ca²⁺ liganding atoms engaging the metal in LETM1. Our Asp687Glu caused a ~ 4‐fold increase in binding affinity, consistent with the Asp at position 12 contributing to the weak affinity. Despite these non‐idealities for Ca²⁺ binding, LETM1 EF1 much more preferentially binds Ca²⁺ over Mg²⁺, and while Mn²⁺ binding was detected in the same concentration range as Ca²⁺, the EF1 structural responsiveness is not fully recapitulated by Mn²⁺.

Alignment of the LETM1 EF1 loop with other weak Ca²⁺ binding EF‐hand loops (i.e., STIMs and the voltage‐gated sodium channel, Na_V1.5), caltubin and tight Ca²⁺ binding calmodulin ³⁶ , ³⁷ , ³⁸ , ³⁹ (and see references below), reinforces the difficulty in predicting whether an EF‐hand motif binds Ca²⁺ and affinity based on sequence, in part due to metal ion coordination that involves main chain atoms (Figure S7C). A second putative EF‐hand has been reported in LETM1 without experimental evidence, centered within predicted CC3 (Figure 1) ³ , ¹³ ; however, position 6 of this EF2, which is typically a highly invariant Gly necessary to mediate the sharp turn around the ion, exists as a Leu in EF2. Even though EF2 lacks this Gly, Asp residues are present at positions 1 (X) and 5 (Z), and a Glu occupies position 12 (−Z), as are typically found in canonical EF‐hand loops (Figure S7C). Yet, we observed no evidence for Ca²⁺ binding in an EF2 isolate or when expressed in an EF2‐EF1 tandem construct (Figure S4, S5). In contrast, our solution NMR confirmed the canonical Gly positioning adopted by EF1 when bound to Ca²⁺ (Figure 6).

Previous Na_v1.5 EF‐hand work showed that interactions with other non‐EF‐hand domains may influence affinity, as extension of the Na_V1.5 construct to include ~55–60 residues, C‐terminal to the EF‐hand domain, increased the Ca²⁺ binding affinity (decreased K_d) from ~4–8 mM to ~1–15 μM. ⁴⁰ , ⁴¹ We note, however, some studies have been unable to detect Ca²⁺ binding to the atypical Na_V1.5 EF‐hands using similar extended constructs by fluorescence ⁴² or within crystal structures. ⁴³ , ⁴⁴ EF‐hand Ca²⁺ binding affinities evolve to suit the location and function of the protein in which they are integrated. ³⁰ , ⁴⁵ Global matrix Ca²⁺ has been suggested to reach as high as ~500 μM, ⁴⁶ , ⁴⁷ insufficient to fully saturate EF1; however, Ca²⁺ binding saturation of EF1 may occur close to a Ca²⁺ entry pore, where local concentrations can reach mM levels. ⁴⁸ , ⁴⁹ Further, CC‐ and/or RBD‐mediated LETM1 assembly may result in pairing of EF1 or other stabilizing interactions that enhance in situ Ca²⁺ binding affinity, as observed with the extended Na_V1.5 constructs. Indeed, full‐length mouse LETM1 has been suggested to exist as a hexamer, ⁵ and human LETM1 migrates as oligomers on native gels. ³⁴ , ⁵⁰ , ⁵¹ LETM1 contains four putative CC domains (Figure 1) that could mediate this oligomerization under low Ca²⁺ conditions, and a recent study has shown that three conserved residues (382–384) within the RBD also play a role in LETM1 assembly. ²²

Second, Ca²⁺ binding caused a robust increase in the α‐helicity of EF1 (Figure 2). This Ca²⁺‐induced folding was supported by an increased apparent T_0.5 compared to the apo state (Figure 2; Table 1) and by our ¹H‐¹⁵N HSQC spectra showing ¹H(¹⁵N) peaks collapsed in the central region but well dispersed in the absence and presence of Ca²⁺, respectively (Figure 6). While unique among the EF‐hand superfamily, stromal interaction molecules (STIMs) also contain a single canonical EF‐hand that undergoes Ca²⁺‐induced folding. ⁵² Remarkably, STIMs contain a second non‐canonical EF‐hand that pairs with the canonical, enhancing the overall structural stability. It is also noteworthy that the STIM1 canonical EF‐hand in isolation exhibits a similar affinity as when expressed together with the non‐canonical motif. ⁵³ , ⁵⁴ Hence, we do not rule out the possibility of a second, non‐canonical EF‐hand motif within LETM1 structurally stabilizing EF1. High‐resolution structural data on larger or full‐length LETM1 constructs is needed to unequivocally determine whether a second EF‐hand stabilizes EF1, as revealed for the STIMs.

Third, we found that LETM1 EF1 is highly sensitive to pH, exhibiting increased α‐helical structure and ANS binding at pH 6.0 compared to 7.8 and 8.5 in the absence of Ca²⁺ (Figure 2, 4). The sensitivity in this pH range can most likely be attributable to changes in the protonation states of His and Cys residues. Since EF1 contains no Cys but one His within the entering helix, we posit that the EF1 His662 is involved in pH sensing; moreover, His662 within EF1 could be sensitive to local H⁺ concentration fluctuations in respiring mitochondria. ⁵⁵ Indeed, a His residue was found to play an important pH sensing role in the pH‐gated K⁺ channel of streptomyces A (KcsA), where mutation abrogates the effects of pH on the channel, resulting in constitutive activation. ⁵⁶ Consistent with a pH sensing role for EF1, a low‐resolution transmission electron microscopy model of self‐oligomerized LETM1 has revealed a central cavity conformation that can be modulated by pH. ⁵

Finally, LETM1 EF1 exhibits an uncommon cold denaturation at temperatures above 0°C. Remarkably, Ca²⁺‐free EF1 begins to unfold below 37°C, exhibiting primarily random coil at 4°C, as assessed by far‐UV CD spectroscopy. This cold‐unfolding phenomenon was observed at all pH values and was attenuated upon Ca²⁺ binding (Figure 2). Consistently, decreasing the temperature to 4°C also decreased the Ca²⁺ binding affinity ~4‐fold (Figure 3; Table 2). The cold denaturation may be due to temperature‐dependent hydration of hydrophobic residues that becomes more thermodynamically favourable as temperature decreases. ⁵⁷ , ⁵⁸ , ⁵⁹

Overall, the present study has not only demonstrated that canonical Ca²⁺ binding to human LETM1 EF1 leads to increased hydrophobic exposure and oligomerization but also exposed several unique properties for this EF‐hand motif. Specifically, we have discovered that Ca²⁺ binding, pH and temperature are robustly coupled with folding of the motif at physiologically relevant temperature and pH ranges. Collectively, these characteristics suggest that EF1 may serve as a transducer of multiple environmental stimuli into regulation of LETM1 Ca²⁺/H⁺ function through a common folding/unfolding mechanism. This tripartite sensitivity is straight‐forwardly observed in the far‐UV CD spectra of EF1 (Figure 7(a)) and allows the conceptualization of how these stimuli may non‐mutual exclusively regulate LETM1 (Figure 7(b)). In this model, temperature, pH and Ca²⁺ affect the structure and conformation of EF1, driving homomeric and/or heteromeric protein–protein interactions that regulate LETM1 function. We speculate that the Ca²⁺/H⁺ antiporter function itself could drive positive and negative feedback signals via EF1 through local changes in Ca²⁺ and H⁺ concentrations, promoting a remarkable autoregulation of LETM1 function. Precise LETM1 regulation is essential given the importance of Ca²⁺ and H⁺ concentrations to mitochondrial bioenergetics, apoptosis and autophagy and is in‐line with the multiple regulatory inputs observed for other critical Ca²⁺ handling proteins found in the mitochondria. ⁶⁰ , ⁶¹

FIGURE 7.

EF1 sensing in LETM1 function. (a) Far‐UV CD spectra demonstrating the robust tripartite structural sensitivity of EF1 to Ca²⁺, temperature and pH. (b) Conceptual model integrating non‐mutually exclusive temperature, pH and Ca²⁺ sensing of EF1 with solvent accessible hydrophobicity and homomeric and heteromeric protein–protein interactions, which regulate LETM1 function. Both local Ca²⁺ and H⁺ concentrations may feedback modulate EF1 conformation, thus establishing a LETM1 autoregulation

4. MATERIALS AND METHODS

4.1. Protein expression and purification

Constructs encompassing human LETM1 (NCBI accession NP_036450.1) residues 643–699 (EF1), residues 537–634 (EF2) and residues 537–699 (EF2‐EF1) were subcloned into pET‐28a vectors (Novagen) using NheI and XhoI restriction sites. Asp687Glu and Lys667_Leu668insTyr mutants of the pET‐28a‐EF1 construct were engineered using PCR‐mediated mutagenesis and confirmed by DNA sequencing of the open reading frame.

Protein expression in BL21 (DE3) Escherichia coli cells cultured in Luria broth (LB) was induced with 0.4 mM isopropyl β‐d‐1‐thiogalactopyranoside (IPTG) for 4 hours at 37°C. Proteins were purified under denaturing conditions using nickel‐nitrilotriacetic acid agarose beads as per the manufacturer guidelines (HisPur; Thermo Fisher Scientific). The lysis buffer was 30 mM Tris, 6 M guanidine‐HCl, pH 8.0, the wash buffer was 20 mM Tris, 150 mM NaCl, 6 M Urea, pH 8.0 and the elution buffer was 20 mM Tris, 150 mM NaCl, 300 mM imidazole, 6 M Urea, pH 8.0 for the metal affinity chromatography. The chaotrope was removed by dialysis in 20 mM Tris, 150 mM NaCl, pH 7.8, 4°C using a 3,500 Da molecular weight cutoff membrane (Thermo Fisher Scientific). The N‐terminal hexahistidine tag was cleaved with ~2 units of thrombin (Sigma) per 1 mg of protein. The final purification steps were anion exchange chromatography using a QFF ANX column (GE Healthcare) and dialysis into experimental buffers. The anion exchange chromatography was performed using 20 mM Tris, pH 7.8 and a 0–1 M NaCl gradient over 60 column volumes. Protein concentrations were estimated using the bichinchoninic acid assay (Pierce). Experimental buffers were 20 mM Tris, 150 NaCl, pH 8.5 or 7.8, and 20 mM bis‐Tris, 150 mM NaCl, pH 6.0.

Uniformly ¹⁵N‐labeled EF1 was expressed in BL21 (DE3) E. coli cultured in M9 minimal medium with ¹⁵N‐NH₄Cl (Sigma) as the sole nitrogen source. Purification was performed as per the LB‐expressed protein.

4.2. Far‐UV CD spectroscopy

Far‐UV CD spectra were obtained using a Jasco J‐810 CD spectrometer with electronic Peltier temperature regulator (Jasco, Inc.). Each spectrum was an average of 3 accumulations, recorded at 4, 20 and 37°C using a 1 mm pathlength quartz cuvette in 1 nm increments, 8 s averaging time and 1 nm bandwidth. All spectra were corrected for buffer contributions. After acquiring divalent cation free spectra, CaCl₂, MgCl₂ and MnCl₂ were added to the same samples to ascertain the structural changes.

Thermal melts were recorded using a 1 mm pathlength quartz cuvette by monitoring the change in CD signal at 222 nm from 20–95°C. A scan rate of 1°C min⁻¹, 1 nm bandwidth and 8 s averaging time was used for these thermal measurements. Thermodynamic stability parameters were extracted from the normalized Ca²⁺‐free thermal melts using a two‐state native (N) to unfolded (U) model, as previously described. ⁶² , ⁶³ The midpoint of unfolding (T_m), specific heat capacity of unfolding (ΔC_p) and enthalpy of unfolding (ΔH) were individually fit for each individual thermal melt at each solution pH. These parameters were used to calculate the Gibbs free energy of unfolding (ΔG), using the Gibbs‐Helmholtz equation. ⁶² , ⁶³ Due to a lack of well‐defined native and unfolded baselines, thermodynamic fitting could not be reliably performed for the Ca²⁺‐supplemented samples. Thus, the temperature where the fractional change in ellipticity between was 0.5 (T_0.5) was taken as an indicator of stability.

4.3. Intrinsic and extrinsic fluorescence

All fluorescence spectroscopy experiments were performed using a Cary Eclipse spectrofluorimeter (Varian, Inc.) and a 1 cm pathlength quartz cuvette. Intrinsic Tyr fluorescence emission spectra were acquired between 290 and 360 nm using a 276 nm excitation wavelength. Data was acquired at a scan rate of 120 nm min⁻¹ with excitation and emission slit widths of 5 and 10 nm, respectively.

Extrinsic ANS fluorescence emission spectra were recorded between 400 and 600 nm using a 372 nm excitation wavelength. Data were acquired using a scan rate of 120 nm min⁻¹ with excitation and emission slit widths of 10 and 20 nm, respectively. ANS (Sigma) at 0.05 mM was incubated with protein for 10 min in the dark at ambient temperature prior to data acquisition.

4.4. Cation binding affinities

Intrinsic Tyr and extrinsic ANS fluorescence emission spectra were acquired 20°C after sequential additions of CaCl₂ up to a final concentration of 80 mM (intrinsic) and 110 mM (extrinsic), and the intensity at the wavelength showing the maximum difference in fluorescence emission intensity compared to the 0 CaCl₂ spectrum was plotted versus total CaCl₂ concentration to construct the binding curves. A similar approach was used in evaluations of MgCl₂ and MnCl₂ binding.

Far UV‐CD spectra were acquired at 4, 20 and 37°C after sequential additions of CaCl₂, MgCl₂ or MnCl₂ up to a final concentration of 80 mM. The ellipticity at 222 nm was plotted versus total divalent cation concentration to construct the binding curves.

The equilibrium dissociation constants (K_d) were estimated from the intrinsic fluorescence, extrinsic fluorescence and CD ellipticity binding curves using a one‐site binding model that accounts for protein concentration, fit to the data by non‐linear regression.

4.5. Size exclusion chromatography with multi‐angle light scattering

SEC‐MALS was performed with a Superdex Increase S200 10/300 GL column (GE Healthcare) connected in‐line with a sixteen‐angle Dawn Heleos II light‐scattering instrument and Optilab TrEX differential refractometer (Wyatt Technologies). Flow through the SEC‐MALS system was controlled by an AKTA Pure FPLC (GE Healthcare) housed at ~10°C. Molecular weight was calculated using the ASTRA software (Wyatt Technologies) based on the Zimm plot analysis and using a protein refractive index increment (dn dc⁻¹) = 0.185 L g⁻¹.

4.6. Solution NMR spectroscopy

NMR experiments were performed on a 600 MHz Varian/Inova NMR spectrometer equipped with a triple resonance HCN probe. ¹H‐¹⁵N heteronuclear single quantum coherence (HSQC) spectra were recorded using 32 transients, 64 increments in the nitrogen dimension, 8,000 Hz ¹H sweep width and 1,800 Hz ¹⁵N sweep width. Approximately 300 μM uniformly ¹⁵N‐labeled EF1 was solubilized in 20 mM Tris, 150 mM NaCl, pH 7.8, with or without 30 mM CaCl₂ and 10 mM 3‐[(3‐cholamidopropyl)dimethylammonio]‐2‐hydroxy‐1‐propanesulfonate (CHAPS) in the buffer optimized CaCl₂ loaded sample. Sixty μM 4,4‐dimethyl‐4‐silapentane‐1‐sulfonic acid and 10% (v/v) D₂O were added to all NMR samples for referencing.

AUTHOR CONTRIBUTIONS

Qi‐Tong Lin: Conceptualization; formal analysis; investigation; methodology; writing‐original draft. Rachel Lee: Investigation; methodology. Allen Feng: Investigation; methodology. Michael Kim: Investigation; methodology. Peter Stathopulos: Conceptualization; data curation; formal analysis; funding acquisition; project administration; resources; supervision; validation; writing‐original draft; writing‐review and editing.

Supporting information

Supplementary Fig. S1 Effect of KCl on the EF1 conformational response and binding to Ca²⁺. (A) Far‐UV CD spectra of EF1 in the absence (dashed lines) and presence (solid lines) of 30 mM Ca²⁺ (pH 7.8, 20°C). Spectra acquired in 150 mM NaCl and 150 mM KCl are shown in red and black, respectively. (B) Relative change in MRE at 222 nm as a function of increasing Ca²⁺ (pH 7.8, 20°C). Representative binding curves acquired in the presence of 150 mM NaCl and 150 mM KCl are shown in red and black, respectively. Binding curves were fit to a one‐site binding model that accounts for protein concentration (solid lines). Spectra in A are means ± SEM of n = 3 separate protein expression samples, acquired using 30 μM protein.

Supplementary Figure S2. Secondary structural responsiveness of EF1_Y668ins to Ca²⁺. Far‐UV CD spectra are shown in the absence (dashed lines) and presence (solid lines) of 30 mM Ca²⁺ for EF1_Y668ins (blue) and wild‐type EF1 (red) (pH 7.8, 20°C). Spectra are means ± SEM of n = 3 separate protein expression samples, acquired using 30 μM protein.

Supplementary Figure S3. ANS binding to EF1 in the presence of Mg²⁺. (A) Extrinsic ANS fluorescence emission spectra acquired in the presence of EF1 with 10 mM Ca²⁺ (solid blue lines) and 10 mM Mg²⁺ (solid red lines) (pH 7.8, 20°C). (B) Extrinsic ANS fluorescence emission spectra acquired in the presence of EF1 with 30 mM Ca²⁺ (solid blue lines) and 30 mM Mg²⁺ (solid red lines) (pH 7.8, 20°C). In A and B, spectra acquired in the presence of EF1 without Ca²⁺ (dashed blue lines) and Mg²⁺ (dashed red lines) are shown for reference. Buffer only ANS spectra with 110 mM Ca²⁺ (dashed‐dotted blue lines) and 110 mM Mg²⁺ (dashed‐dotted red lines) are also shown. Spectra are means ± SEM of n = 3 separate protein expression samples, acquired using 30 μM protein.

Supplementary Figure S4. Structural response of EF2 to Ca²⁺. (A) Far‐UV CD spectra of EF2 in the absence (red) and presence (blue) of 30 mM Ca²⁺ (pH 7.8, 20°C). (B) Intrinsic Tyr fluorescence of EF2 in the absence (red) and presence (blue) of 30 mM Ca²⁺ (pH 7.8, 20°C). In A and B, data are representative of n = 2 experiments from two separate protein expressions. Far‐UV CD and intrinsic Tyr fluorescence experiments were performed with 30 μM and 74 μM protein, respectively.

Supplementary Figure S5. Secondary structural responsiveness and affinity of EF2‐EF1 to Ca²⁺. (A) Far‐UV CD spectra of EF2‐EF1 in the absence (red) and presence (blue) of 80 mM Ca²⁺ (pH 7.8, 4°C). (B) Relative change in MRE at 222 nm as a function of increasing Ca²⁺ (pH 7.8, 4°C). Binding curve was fitted to a one‐site binding model that accounts for protein concentration. All Far‐UV CD experiments were performed with 27 μM protein. In A and B, data are means ± SEM of n = 3 samples from two separate protein expressions.

Supplementary Figure S6. Thermal melt profiles of EF2 and EF2‐EF1 protein constructs. (A) Fractional change in CD ellipticity at 222 nm as a function of temperature for EF2 in the absence (red) and presence (blue) of 30 mM Ca²⁺. (B) Fractional change in ellipticity at 222 nm as a function of temperature for EF2‐EF1 in the absence (red) and presence (blue) of 30 mM Ca²⁺. In A and B, thermal melts were acquired using 42 μM and 27 μM protein for EF2 and EF2‐EF1, respectively, and data are representative of n = 2 experiments from two separate protein expressions.

Supplementary Figure S7. LETM1 EF1 loop homology. (A) Cartoon backbone chain representation of strontium (Sr²⁺)‐bound Lymnaea stagnalis (great pond snail) caltubin EF‐hand motif 3 and 4 [EF3 (grey)/EF4 (light blue)] structures (residues 109–179; 6VAN_A.pdb). An NCBI blastp search of the Protein Data Bank reveals that caltubin EF4 shares the most sequence homology to the human LETM1 EF1 loop, among structures elucidated to date. (B) Stick representation of the caltubin EF4 loop residues facilitating approximately pentagonal bipyramidal coordination of Sr²⁺. Ligand residues labeled with red text indicate sequence identity with LETM1 EF1. *Note that S169 (−X) is a non‐coordinating residue in caltubin EF4, atypical for canonical EF‐hands. (C) Sequence alignment of human LETM1 EF1 and caltubin EF4 with other weak Ca²⁺ binding EF‐hand loops (i.e., STIM1, STIM2 and Na_V1.5). The tight Ca²⁺ binding calmodulin EF4 (bold) and purported LETM1 EF2 (bottom) loop sequences are shown for reference. Residue range numbering are shown at left and right of the loop sequences according to NCBI accession numbers for the human LETM1 (NP_036450.1), human calmodulin (CaM; AAD45181.1), human STIM1 (AFZ76986.1), human STIM2 (NP_065911.3) and human Na_V1.5 (NP_932173.1). Snail caltubin residue numbering is as per the 6VAN_A coordinate file. Caltubin EF4 and LETM1 EF1 loop residues are indicated as fully conserved (*; red text in the caltubin sequence), highly conserved (:) or partially conserved (.). The estimated Ca²⁺ binding K_d values were taken from ^apresent study, ^breference,^{60 c}reference,^{61 d}references ^52,53,62,63, ^ereference,^{46 f}references.^40‐44 n.d., not detected; cEF, canonical EF‐hand (EF1); ncEF, non‐canonical EF‐hand (EF2). The structures shown in A and B, were rendered using The PyMOL Molecular Graphics System (Schrodinger LLC, Version 2.5).

Lin Q‐T, Lee R, Feng AL, Kim MS, Stathopulos PB. The leucine zipper EF‐hand containing transmembrane protein‐1 EF‐hand is a tripartite calcium, temperature, and pH sensor. Protein Science. 2021;30:855–872. 10.1002/pro.4042