Redox Enzymes P4HB and PDIA3 Interact with STIM1 to Fine-Tune Its Calcium Sensitivity and Activation

Sensing the lowering of endoplasmic reticulum (ER) calcium (Ca2+), STIM1 mediates a ubiquitous Ca2+ influx process called the store-operated Ca2+ entry (SOCE). Dysregulated STIM1 function or abnormal SOCE is strongly associated with autoimmune disorders, atherosclerosis, and various forms of cancers. Therefore, uncovering the molecular intricacies of post-translational modifications, such as oxidation, on STIM1 function is of paramount importance. In a recent proteomic screening, we identified three protein disulfide isomerases (PDIs)—Prolyl 4-hydroxylase subunit beta (P4HB), protein disulfide-isomerase A3 (PDIA3), and thioredoxin domain-containing protein 5 (TXNDC5)—as the ER-luminal interactors of STIM1. Here, we demonstrated that these PDIs dynamically associate with STIM1 and STIM2. The mutation of the two conserved cysteine residues of STIM1 (STIM1-2CA) decreased its Ca2+ affinity both in cellulo and in situ. Knockdown of PDIA3 or P4HB increased the Ca2+ affinity of wild-type STIM1 while showing no impact on the STIM1-2CA mutant, indicating that PDIA3 and P4HB regulate STIM1’s Ca2+ affinity by acting on ER-luminal cysteine residues. This modulation of STIM1’s Ca2+ sensitivity was further confirmed by Ca2+ imaging experiments, which showed that knockdown of these two PDIs does not affect STIM1-mediated SOCE upon full store depletion but leads to enhanced SOCE amplitudes upon partial store depletion. Thus, P4HB and PDIA3 dynamically modulate STIM1 activation by fine-tuning its Ca2+ binding affinity, adjusting the level of activated STIM1 in response to physiological cues. The coordination between STIM1-mediated Ca2+ signaling and redox responses reported herein may have implications for cell physiology and pathology.

STIM1 function can be modulated by a wide range of post-translational modifications (PTMs), such as phosphorylation [14], acylation [15,16], glycosylation [17,18], and oxidation by reactive oxygen species (ROS) [19,20].Yet enzymes for such modifications are less well defined.Utilizing proximity labeling and imaging, we recently systemically screened and unveiled the pivotal ER-luminal interactome of STIM1 [21].The six key ER-luminal proteins that could dynamically interact with both STIM1 and STIM2 can be divided into two groups: one related to redox reactions and the other to N-glycan processing.Earlier in vitro studies have shown that N-glycosylation decreases STIM1 Ca 2+ -binding affinity [18].Our findings further revealed that STIM-interacting Glucosidase II (consists of PRKCSH and GANAB) is the enzyme responsible for reducing STIM1's Ca 2+ affinity via N-glycosylation [21].
In the present study, we first characterized the dynamic interactions of three oxidoreductases, namely TXNDC5, P4HB, and PDIA3, with STIM proteins.Subsequently, employing in situ Ca 2+ -titration of STIM1 via a highly dynamic FRET reporter for STIM1 activation, we unveiled that PDIA3 and P4HB could regulate STIM1's Ca 2+ affinity by acting on the ER-luminal conserved cysteine residues.Although this modulation of STIM1's Ca 2+ sensitivity does not impact its full capacity to mediate SOCE, it is likely to influence the initiation of SOCE triggered by various physiological stimuli.The elucidated molecular mechanisms herein have significantly enriched our understanding of redox modulation on Ca 2+ signaling.

STIM Proteins Dynamically Interact with Three Protein Disulfide Isomerases (PDIs) (P4HB, TXNDC5, and PDIA3)
To assess the interactions between STIM proteins and the three PDI proteins (P4HB, PDIA3, and TXNDC5), we employed confocal imaging to study their dynamic co-localization, as a recent report demonstrated that direct protein-protein interactions could be identified by their dynamic co-localizations [34].We employed two strategies to redistribute STIM proteins: triggering STIM1's transition from a uniform to puncta ER distribution through ER-Ca 2+ depletion [3] and reducing STIM2's constitutive puncta via intracellular acidification [35].Following these treatments, we evaluated the subcellular localization of these PDI proteins, both when expressed alone and when co-expressed with STIM proteins.Proteins exhibiting subcellular distribution changes solely when co-expressed with STIM were recognized as dynamic STIM interactors.
The colocalization between STIM1 and an ER-localizing mNeonGreen∆N5 (ER-mNG∆N5) [21] served as a negative control in our experimental framework.When expressed alone, ER-mNG∆N5's distribution remained unaltered before and after store depletion with 2.5 µM Ca 2+ ionophore ionomycin (IONO) (Figure 1A).In cells co-expressing ER-mNG∆N5 together with STIM1, IONO induced the formation of STIM1 puncta, while the distribution of co-expressed ER-mNG∆N5 remained unaltered (Figure 1B), indicating no association between ER-mNG∆N5 and STIM1.P4HB also showed no changes in subcellular distribution upon store depletion when expressed alone (Figure 1C).However, when co-expressed with STIM1, P4HB transitioned from an even ER-like distribution to a punctate distribution after IONO treatments, showing clear colocalization with STIM1 (Figure 1D).Similarly, the other two PDIs, TXNDC5 (Figure S1A,B) and PDIA3 (Figure S1C,D), also exhibited STIM1-dependent redistribution following store depletion with IONO.We further performed Pearson correlation coefficient analysis to quantify the extent of co-localization between co-expressed proteins.The results reveal that the co-localization between ER-mNG∆N5 and STIM1 was significantly decreased upon ER depletion (Figure 1E, leftmost bar chart), while the extent of colocalization between STIM1 and all three PDIs significantly increased after ER-Ca 2+ depletion (Figure 1E, three bar charts on the right).These results collectively demonstrate dynamic co-localizations between STIM1 and the three PDIs.
We next examined the dynamic co-localization between STIM2 and these three PDIs with confocal microscopy, using ER-mNG∆N5 as a negative control (Figures S1E-H and S2A-D).When expressed alone, ER-mNG∆N5, P4HB, TXNDC5, and PDIA3 did not exhibit redistribution following intracellular acidification (Figures S1E,G and S2A,C).In cells co-expressing ER-mNG∆N5 with STIM2, intracellular acidification greatly diminished constitutive STIM2 puncta but had no effect on the distribution of ER-mNG∆N5 (Figure S1F), suggesting no association between ER-mNG∆N5 and STIM2, with their colocalization significantly increasing upon acidification (Bar charts in Figure S1F).This phenomenon could be explained by the acidification-induced dynamic redistribution of STIM2 molecules.ER-mNG∆N5 remains uniformly distributed throughout the ER, regardless of ER luminal pH.However, under resting conditions, constitutively active STIM2 proteins are primarily localized at the ER-PM junctions, resulting in minimal colocalization with ER-mNG∆N5 within the bulk of the ER.Upon acidification, these constitutive punctate STIM2 transition to an auto-inhibitory state and distribute uniformly throughout the ER, thereby increasing their colocalization with ER-mNG∆N5 within the bulk of the ER.In contrast to ER-mNG∆N5, P4HB (Figure S1H), TXNDC5 (Figure S2B) and PDIA3 (Figure S2D) also exhibited STIM2dependent transition in cellular distribution, with their colocalization significantly diminishing upon acidification (Bar charts in Figures S1H and S2B,D).Together, these results similarly reveal dynamic co-localizations between the three PDIs and STIM2.
To quantify the associations between STIM1 and P4HB, PDIA3, or TXNDC5 at the nanometer scale, we evaluated the basal FRET signals between these PDI proteins and co-expressed STIM molecules, with the non-STIM-interacting ER-mNG△N5 serving as a negative control.Despite exhibiting good colocalization with both STIM1 and STIM2 (Figure 1E, leftmost panel; Figure S1F, bar chart), ER-mNG△N5 showed minimal basal FRET signals with STIM proteins (Top panels in Figures 1F and S2E).In contrast, the resting FRET signals between P4HB and STIM molecules were significantly higher (Bottom panels in Figures 1F and S2E).Similarly, the basal FRET signals between STIM proteins and PDIA3, or TXNDC5 were also significantly higher than negative control (Figures 1G and S2F).These results thus strongly suggest that P4HB, TXNDC5, or PDIA3 may physically associate with both STIM1 and STIM2 (Figures 1F-G and S2E,F).
PDIA3 has been shown to modulate STIM1-mediated SOCE [25].Interestingly, among the three STIM-interacting PDIs, PDIAs showed the lowest basal FRET signal with STIM proteins (Figures 1G and S2F).This suggests that the other protein disulfide isomerases may likely affect SOCE, possibly via their redox modifications on STIM proteins.
PDIA3 has been shown to modulate STIM1-mediated SOCE [25].Interestingly, among the three STIM-interacting PDIs, PDIAs showed the lowest basal FRET signal with STIM proteins (Figures 1G and S2F).This suggests that the other protein disulfide isomerases may likely affect SOCE, possibly via their redox modifications on STIM proteins.

STIM Mutants Lacking Disulfide-Bond-Forming Ability Exhibited Lower Ca 2+ -Binding Affinity
To investigate whether redox modifications of two conserved cysteine residues (2C) in the ER-luminal region of STIM molecules (STIM1-C49-C56 or STIM2-C140-C147) affect their function, we utilized a FRET assay recently developed by us to examine the effects of mutating the 2C residues into Alanine (2CA) on their ability to bind Ca 2+ [36].In the assay, we used engineered, PM-localized STIM constructs to expose the luminal Ca 2+binding EF-SAM domain of STIM to the extracellular space (PM-SC1111, Figure 2A), allowing precise manipulation of Ca 2+ levels in the vicinity of the extracellular EF-SAM [36].We previously demonstrated their successful localization to the PM and their ability to

STIM Mutants Lacking Disulfide-Bond-Forming Ability Exhibited Lower Ca 2+ -Binding Affinity
To investigate whether redox modifications of two conserved cysteine residues (2C) in the ER-luminal region of STIM molecules (STIM1-C49-C56 or STIM2-C140-C147) affect their function, we utilized a FRET assay recently developed by us to examine the effects of mutating the 2C residues into Alanine (2CA) on their ability to bind Ca 2+ [36].In the assay, we used engineered, PM-localized STIM constructs to expose the luminal Ca 2+ -binding EF-SAM domain of STIM to the extracellular space (PM-SC1111, Figure 2A), allowing precise manipulation of Ca 2+ levels in the vicinity of the extracellular EF-SAM [36].We previously demonstrated their successful localization to the PM and their ability to sense fluctuations in extracellular Ca 2+ levels [36].Extracellular Ca 2+ -induced FRET responses mediated by cytosolic PM-SC and YFP-SOARL (STIM1 343-491 ), a longer version of SOAR, could be used to faithfully deduce the in cellulo Ca 2+ -binding affinities of these STIM constructs.To avoid artifacts induced by endogenous STIM1 or STIM2 molecules and the filling status of the ER Ca 2+ stores, the FRET experiments were performed in HeLa STIM1 and STIM2 double knockout (SK) cells.
tions (PTMs) of STIM within the ER lumen.These findings are consistent with those previously reported by our laboratory [36].The measured Kd value of STIM11-310-2CA (0.78 ± 0.01 mM) was significantly higher than wild-type (WT) STIM11-310 (Figure 2B).Meanwhile, our Western blot analysis results indicate that C49-C56 residues in STIM1 form disulfide bonds (Figure S3A).These results suggest that the breaking of disulfide bonds decreases STIM1's Ca 2+ affinity.This aligns with a previous in vitro report showing that the STIM1-23-213 fragment with Cys49Ser-Cys56Ser mutation exhibited a higher Kd value compared with WT STIM1-23-213 [31].The in-cell Ca 2+ titration results show that the PM-SC1111-C49A-C56A (PM-SC1111-2CA) mutant exhibited lower Ca 2+ -binding ability (Figure 2A).We subsequently measured the in situ K d value of STIM1 using an improved FRET tool with significantly larger dynamics.In this new tool, STIM1 1-310 -ECFP△C11 (SC1111-ECFP△C11) and mNG△N5-SOAR1L serve as readouts for STIM1 activation [21].The results show that the in situ Ca 2+ affinity of STIM1 is 0.69 ± 0.02 mM, with a Hill number of 2.9 ± 0.1.This Hill coefficient is similar to our previous observation [36], indicating a significant positive synergistic effect during the activation of truncated STIM1 dimers, while the Hill number measured with a full-length STIM1 is significantly higher (Hill n = 9.7) [37], likely reflecting STIM1's oligomerization facilitated by PM binding via the C-terminal K-rich region.By lacking the K-rich region, our tool is suitable for dissecting the initial stages of STIM1 activation.Given that both the in cellulo data and the in situ results were obtained using the same FRET readout in the same type of cells, the observed differences in Ca 2+ binding clearly suggest the presence of potential additional modulators or post-translational modifications (PTMs) of STIM within the ER lumen.These findings are consistent with those previously reported by our laboratory [36].The measured K d value of STIM1 1-310 -2CA (0.78 ± 0.01 mM) was significantly higher than wild-type (WT) STIM1 1-310 (Figure 2B).Meanwhile, our Western blot analysis results indicate that C49-C56 residues in STIM1 form disulfide bonds (Figure S3A).These results suggest that the breaking of disulfide bonds decreases STIM1's Ca 2+ affinity.This aligns with a previous in vitro report showing that the STIM1-23-213 fragment with Cys49Ser-Cys56Ser mutation exhibited a higher K d value compared with WT STIM1-23-213 [31].
We then investigated the effects of the 2CA mutation on STIM2's K d value for Ca 2+ using similar in cellulo and in situ assays.Since the binding of STIM2 SOAR with its CC1 region is weaker compared with those of STIM1, we utilized STIM1 chimeric constructs in which the luminal Ca 2+ -binding EF-SAM region was swapped with that of STIM2 (SC2211) to better report the activation of STIM2.Unfortunately, the 2CA mutant of the PM version of SC2211 (PM-SC2211-C140A-C147A, or PM-SC2211-2CA) showed impaired plasma membrane localization and Ca 2+ responses, limiting further investigation (Figure 2C).We then compared the in situ Ca 2+ affinity of STIM2 EF-SAM and its corresponding 2CA mutant (SC2211-C140A-C147A, or SC2211-2CA) (Figure 2D).Similar to in cellulo observation, SC2211-2CA exhibited greatly diminished Ca 2+ responses, hindering accurate estimation of its Ca 2+ affinity (Figure 2D, red trace, middle left panel).Nevertheless, SC2211-2CA exhibited a notably lower basal FRET signal with SOAR1L compared with the wild type, indicating a reduced Ca 2+ affinity (Figure 2D, bar chart).
Of note, the in situ K d value of STIM2 is 1.61 ± 0.01 mM (Figure 2D, rightmost panel), much higher than previous in situ measurements [38,39].Since our high signal-to-noise assay utilized direct manipulation of ER Ca 2+ levels, avoiding artifacts from inaccurate estimation using ER Ca 2+ indicators with insufficiently low affinity, the value reported herein likely provides a more accurate estimation of STIM2's affinity.The basal ER Ca 2+ level in HEK293 cells, estimated using our newly developed highly sensitive ER Ca 2+ indicator [40], TuNer-s, is 1.44 mM.Therefore, based on calculations using the Hill equation, approximately 75% of STIM2 molecules exist in Ca 2+ -free, active state, elucidating its well-documented constitutively active nature [3].Furthermore, the observed Ca 2+ -binding ability of STIM2 is considerably lower than that of STIM1 (Figure 2D, rightmost panel, and Figure 2B), which is consistent with previous findings from our group and others [36,38,39].
Collectively, these in cellulo and in situ results clearly demonstrate that 2CA mutations reduce Ca 2+ affinities of STIM proteins.STIM2-2CA mutants exhibited diminished dynamics, indicating that redox modifications within its luminal region may impair its activation, thereby hindering further dissection.Ca 2+ imaging results show that 2CA mutation has no significant effect on STIM2-mediated constitutive Ca 2+ entry, indicated by GEM-GECO1 [41] (STIM2: 7.5 ± 0.30, STIM2-2CA: 7.3 ± 0.35), indicating that redox modifications may have minimal impact on STIM2-mediated Ca 2+ responses.Consequently, our attention focused on STIM1 for the remaining study.

Knocking down P4HB or PDIA3 Reduce the Ca 2+ Affinity of STIM1
We next set out to explore the impact of the three STIM-interacting PDIs on STIM1's Ca 2+ -binding behavior.We first examined the effects of P4HB, PDIA3, or TXNDC5 overexpression on Ca 2+ -induced FRET responses between STIM1 1-310 (SC1111) and SOAR1L.We performed Western blot analysis to quantify and compare the levels of overexpressed PDIA3, P4HB, or TXNDC5 proteins relative to their endogenous counterparts (Figure S3B).Our results demonstrate that the overexpression of PDIA3, P4HB, or TXNDC5 did not significantly change the levels of their respective endogenous counterparts (Figure S3C).Furthermore, the total levels of these aforementioned proteins in overexpressing cells were markedly higher than those in blank control cells (Figure S3D).Interestingly, the overexpression of P4HB, PDIA3 (Figure 3A), or TXNDC5 (Figure S4A) did not significantly modify the K d values of STIM1.It is likely that endogenous levels of these proteins are sufficient to make necessary redox modifications on STIM1.We thus proceeded to investigate the effects of knocking down these proteins using CasRx technology [42].The efficiency of the knockdown was assessed with quantitative RT-PCR, revealing a significant decrease in their mRNA levels (Figure S4B).This was further confirmed by Western blot analysis, demonstrating a notable reduction in the expression of these proteins (Figure S4C,D).Interestingly, knocking down TXNDC5 did not affect the K d value of STIM1 (Figure S4E).This might be attributed to the less sufficient knockdown efficiency of TXNDC5 (Figure S4B-D, red) or the existence of alternative regulatory processes that compensate for TXNDC5 function within cells.In contrast, in cells with PDIA3 or P4HB knockdown, the K d values of STIM1 significantly increased compared with those in control cells (Control: 0.69 ± 0.01 mM; CasRx-PDIA3: 0.78 ± 0.01 mM; CasRx-P4HB: 0.78 ± 0.01 mM) (Figure 3B).These findings highlight the critical roles of both PDIA3 and P4HB in regulating STIM1's Ca 2+ affinity.
cells were markedly higher than those in blank control cells (Figure S3D).Interestingly, the overexpression of P4HB, PDIA3 (Figure 3A), or TXNDC5 (Figure S4A) did not significantly modify the Kd values of STIM1.It is likely that endogenous levels of these proteins are sufficient to make necessary redox modifications on STIM1.We thus proceeded to investigate the effects of knocking down these proteins using CasRx technology [42].The efficiency of the knockdown was assessed with quantitative RT-PCR, revealing a significant decrease in their mRNA levels (Figure S4B).This was further confirmed by Western blot analysis, demonstrating a notable reduction in the expression of these proteins (Fig- ures S4C,D).Interestingly, knocking down TXNDC5 did not affect the Kd value of STIM1 (Figure S4E).This might be attributed to the less sufficient knockdown efficiency of TXNDC5 (Figure S4B-D, red) or the existence of alternative regulatory processes that compensate for TXNDC5 function within cells.In contrast, in cells with PDIA3 or P4HB knockdown, the Kd values of STIM1 significantly increased compared with those in control cells (Control: 0.69 ± 0.01 mM; CasRx-PDIA3: 0.78 ± 0.01 mM; CasRx-P4HB: 0.78 ± 0.01 mM) (Figure 3B).These findings highlight the critical roles of both PDIA3 and P4HB in regulating STIM1's Ca 2+ affinity.To assess whether these regulatory effects are related to redox reactions on STIM1-C49-C56 residues, we examined the impact of knocking down PDIA3 or P4HB on the Kd values of STIM1-2CA mutants lacking the ability to form disulfide bonds.The results show that knocking down PDIA3 or P4HB did not alter the Kd value of STIM1-2CA (Figure 4A), indicating that PDIA3 or P4HB alter the Ca 2+ -binding behavior of STIM1 by their actions on STIM1-C49-C56 residues.Furthermore, the resting FRET signals between To assess whether these regulatory effects are related to redox reactions on STIM1-C49-C56 residues, we examined the impact of knocking down PDIA3 or P4HB on the K d values of STIM1-2CA mutants lacking the ability to form disulfide bonds.The results show that knocking down PDIA3 or P4HB did not alter the K d value of STIM1-2CA (Figure 4A), indicating that PDIA3 or P4HB alter the Ca 2+ -binding behavior of STIM1 by their actions on STIM1-C49-C56 residues.Furthermore, the resting FRET signals between STIM1-2CA and PDIA3 or P4HB were significantly reduced compared with those detected with wildtype STIM1 (Figure 4B,C), indicating that these two conserved residues are crucial for the interactions of PDIA3 or P4HB with STIM1.Interestingly, the 2CA mutation showed no significant effect on the basal FRET efficiency between STIM1 and TXNDC5 (STIM1 + TXNDC5: 0.64 ± 0.05; STIM1-2CA + TXNDC5: 0.61 ± 0.12; ns, not significant, Student's t-test).Thus, C49-C56 residues of STIM1 appear nonessential for its binding with TXNDC5, indicating a different interaction mode between TXNDC5 and STIM1.Consistent with prior findings [25], knockdown of PDIA3 or P4HB did not significantly alter the disulfide bond formation of STIM1.This suggests the presence of additional redox enzymes compensating for the function of PDIA3 or P4HB (Figure S4F).These findings collectively suggest that the interaction of PDIA3 and P4HB with the cysteine residues of STIM1 is required to regulate its Ca 2+ sensitivity.
tation showed no significant effect on the basal FRET efficiency between STIM1 and TXNDC5 (STIM1 + TXNDC5: 0.64 ± 0.05; STIM1-2CA + TXNDC5: 0.61 ± 0.12; ns, not significant, Student's t-test).Thus, C49-C56 residues of STIM1 appear nonessential for its binding with TXNDC5, indicating a different interaction mode between TXNDC5 and STIM1.Consistent with prior findings [25], knockdown of PDIA3 or P4HB did not significantly alter the disulfide bond formation of STIM1.This suggests the presence of additional redox enzymes compensating for the function of PDIA3 or P4HB (Figure S4F).These findings collectively suggest that the interaction of PDIA3 and P4HB with the cysteine residues of STIM1 is required to regulate its Ca 2+ sensitivity.

The Knockdown of P4HB or PDIA3 Promotes STIM1 Activation and SOCE
We further investigated the functional consequences of these interactions by firstly examining the effects of PDIA3 or P4HB on STIM1-mediated SOCE responses.HEK293 cells stably expressing a cytosolic Ca 2+ indicator, GEM-GECO1 [41] (GEM-GECO1 cells),

The Knockdown of P4HB or PDIA3 Promotes STIM1 Activation and SOCE
We further investigated the functional consequences of these interactions by firstly examining the effects of PDIA3 or P4HB on STIM1-mediated SOCE responses.HEK293 cells stably expressing a cytosolic Ca 2+ indicator, GEM-GECO1 [41] (GEM-GECO1 cells), were transiently transfected with PDIA3-mScarlet or P4HB-mScarlet.After passive depletion of the ER Ca 2+ store with 1 µM thapsigargin (TG), an inhibitor of ER Ca 2+ pump, 1 mM extracellular Ca 2+ was added to an extracellular bath to allow Ca 2+ influxes via SOCE.Compared with blank controls, GEM-GECO1 cells expressing PDIA3, P4HB, or TXNDC5 exhibited GEM-GECO1 responses of similar amplitudes (Figure 5A), indicating no significant alteration in SOCE responses.Similarly, knocking down PDIA3, P4HB, or TXNDC5 still did not affect the amplitudes of SOCE indicated by GEM-GECO1 responses (Figure 5B).These results demonstrate that PDIA3, P4HB, or TXNDC5 does not affect SOCE signals activated by maximal ER Ca 2+ store depletion.Unlike a previous report that found PDIA3 inhibits SOCE in mouse embryonic fibroblasts (MEFs) [25], our results align with reports in HEK293 cells, where the STIM1-C49S-C56S mutant mediates SOCE with amplitudes similar to WT STIM1 [27,30].It appears that the PDIA3's effect on SOCE may be cell-type-specific.Nevertheless, our findings are consistent with their ability to modify STIM1's K d value for Ca 2+ .Thus, upon complete store depletion with TG, all STIM1 molecules, regardless of whether their affinities are altered by PDIs, will be all activated, resulting in SOCE with similar amplitudes in HEK 293 cells.
pared with blank controls, GEM-GECO1 cells expressing PDIA3, P4HB, or TXNDC5 exhibited GEM-GECO1 responses of similar amplitudes (Figure 5A), indicating no significant alteration in SOCE responses.Similarly, knocking down PDIA3, P4HB, or TXNDC5 still did not affect the amplitudes of SOCE indicated by GEM-GECO1 responses (Figure 5B).These results demonstrate that PDIA3, P4HB, or TXNDC5 does not affect SOCE signals activated by maximal ER Ca 2+ store depletion.Unlike a previous report that found PDIA3 inhibits SOCE in mouse embryonic fibroblasts (MEFs) [25], our results align with reports in HEK293 cells, where the STIM1-C49S-C56S mutant mediates SOCE with amplitudes similar to WT STIM1 [27,30].It appears that the PDIA3's effect on SOCE may be celltype-specific.Nevertheless, our findings are consistent with their ability to modify STIM1's Kd value for Ca 2+ .Thus, upon complete store depletion with TG, all STIM1 molecules, regardless of whether their affinities are altered by PDIs, will be all activated, resulting in SOCE with similar amplitudes in HEK 293 cells.Subsequently, we investigated the effects of knocking down (KD) these PDIs on STIM1-dependent signaling triggered by submaximal stimulation mimicking physiological conditions.We envisioned that PDIA3 or P4HB KD would reduce STIM1's K d value for Ca 2+ , making them more sensitive to partial ER Ca 2+ store depletion.We first assessed the formation of STIM1 puncta, indicative of STIM1 activation [3], triggered by submaximal activation of muscarinic acetylcholinergic receptors with 10 µM carbachol (CCh).Following knockdown of TXNDC5, there was no significant difference in the proportion of cells forming STIM1 puncta in response to 10 µM CCh compared with blank controls.Interestingly, a larger proportion of cells with PDIA3 or P4HB KD exhibited STIM1 puncta following CCh stimulation compared with blank control cells (Figure 5C), suggesting heightened STIM1 activation in PDIA3 or P4HB KD cells but not in TXNDC5 KD cells.We next examined the effects of PDIs KD on STIM1-mediated SOCE induced by 10 µM CCh.However, SOCE amplitudes triggered by partial store depletion with 10 µM CCh were quite small.Although within GEM-GECO1's linear range (Kd~340 nM) [41], the response was too small for accurate quantification.Therefore, we used a higher Ca 2+ concentration (3 mM) along with a more sensitive Ca 2+ indicator TurNm [40] to ensure enhanced detection of SOCE responses (Figure S4G).The knockdown of TXNDC5 didn't affect the magnitude of 10 µM CCh-induced SOCE (Figure 5D), which aligns with the effect on the K d value of STIM1 observed upon knocking down TXNDC5.Interestingly, the results reveal a significantly larger CCh-induced SOCE in cells with PDIA3 or P4HB KD (Figure 5D).These findings collectively indicate that the knockdown of PDIA3 or P4HB renders STIM1 more readily activated, potentially resulting in increased SOCE in response to physiological stimuli.

Gene Knockdown by CasRx
Knockdown of PDIA3, P4HB, or TXNDC5 in HeLa STIM1 and STIM2 double knockout (SK) cells was achieved via the CasRx system [42].Cells were transfected with corresponding CasRx plasmids via electroporation, and the transfected cells were used for further experiments after 48 h of transfection.The efficiency of CasRx transfection was confirmed by qPCR and Western blot analysis.

Total RNA Isolation and Quantitative Real-Time Polymerase Chain Reaction (qPCR) Analysis
Total RNA was extracted from cells using TRIzol, Waltham, MA, USA, and reverse transcription was performed with PrimeScript™ RT Master Mix (Takara, cat.no.RR036A, Kusatsu, Japan) following the manufacturer's instructions.The cDNA product was used as a template for qPCR run and mixed with primers (Table 1) and SYBR Green PCR Master Mix (GenStar Biosolutions, cat.no.A314, Beijing, China).qPCR reaction was performed on QuantStudio™ 6 Flex Real-Time PCR System (Applied Biosystems, Foster City, CA, USA).Relative mRNA levels were calculated using the Comparative Ct (△△CT) method.Gene expression levels were normalized to those of human GAPDH [45].[36,46].Transfections were performed by electroporation using the Gene Pulser Xcell system (Bio-Rad, Hercules, CA, USA) in 4 mm cuvettes and OPTI-MEM medium.For HEK293 cells, a voltage step pulse (180 V, 25 ms, in 0.4 mL of the medium) was used; for HeLa cells, an exponential pulse (260 V, 525 µF, in 0.5 mL medium) was used.Transfected cells were seeded on round coverslips, first cultured in serum-free OPTI-MEM for 40 min, then in regular DMEM medium containing 10% FBS and 1% P/S for 24 h.
To establish stable cells stably expressing GEM-GECO1 [41], the Ca 2+ indicator GEM-GECO1 was transfected into HEK293 cells.After selection with 2 µg/mL puromycin for 5-7 days, the cells were then diluted to single clones and expanded in culture.Healthy clones with high expression and normal Ca 2+ responses were selected for usage.

Förster Resonance Energy Transfer (FRET) Imaging and Measurements of Ca 2+ Affinity of STIM1
The above-mentioned ZEISS observer Z1 system for Ca 2+ imaging was used for the FRET imaging, with calibrations and offline analysis performed as previously described [49][50][51].In this study, three pairs of fluorescent proteins, CFP and YFP, ECFP△C11 and mNG△N5, and mTurquoise2 and mNG△N5, were used.Fluorescence (F) of CFP/ECFP△C11/mTurquoise2 (438 ± 12 nm Ex /483 ± 16 nm Em ), YFP/mTurquoise2 (510 ± 5 nm Ex /542 ± 13.5 nm Em ), and FRET raw (438 ± 12 nm Ex /542 ± 13.5 nm Em ) was captured every 10s [36].The related parameters for the CFP-YFP FRET pair were exactly the same as before [51], ECFP△C11 and mNG△N5 and mTurquoise2 and mNG△N5-mediated FRET were recalibrated and calculated as FRET c = FRET raw − F d /D d × F ECFP△C11 − F a /D a × F mNG△N5 .In this for- mula, F d /D d represents the measured bleed-through of ECFP△C11 into the FRET filter (0.73), and F a /D a represents the measured bleed-through of mNG△N5 through the FRET filter (0.36).For mTurquoise2-mNG△N5 FRET measurements, bleed-through values were 0.25 for mNG△N5 and 0.65 for mTurquoise2.Normalized FRET was obtained by normalizing FRET c against F ECFP△C11 or F mTurquoise2 to avoid differences in expression levels [51].Fluorescence readings from regions of interest were exported from the SlideBook6.0.23 software and processed with Matlab 2023b to calculate the system-independent apparent FRET efficiency, FRET c /F ECFP△C11 or FRETc/F mTurquoise2 .Representative traces from at least three independent experiments performed on 25-40 cells were shown as mean ± SEM.
In situ or in cellulo Ca 2+ titration of STIM was performed using HeLa SK cells.For in cellulo measurements, cells transiently co-expressing YFP-SOAR1L with either PM-SC1111-CFP or PM-SC2211-CFP.The FRET signals of cells bathed in Ca 2+ imaging solutions containing different concentrations of free Ca 2+ were collected to obtain dose-response curves.Similarly, in situ measurements were conducted in cells transiently co-expressing mNG△N5-SOAR1L with either STIM1 1-310 -ECFP△C11 (SC1111-ECFP△C11) or variants.These measurements utilized a solution containing 10 mM NaCl, 140 mM KCl, 1 mM MgCl 2 , 20 mM HEPES, 0.025 mM digitonin, 0.01 mM ionomycin, and 1 mM EGTA (pH 7.4).During measurements, cells were permeabilized with the above solution containing different free Ca 2+ concentrations, ranging from zero to 2 mM Ca 2+ or up to 8 mM, to obtain corresponding Ca 2+ responses.Ca 2+ affinities of the STIM fragments or variants were then calculated by fitting the FRET-Ca 2+ relationship to the Hill equation using Prism 9.5.1 software.All experiments were carried out at room temperature.Traces shown were representative of at least three independent repeats, with 30-60 single cells analyzed per repeat.

Confocal Microscopy
Images were taken using a ZEISS LSM880 system equipped with 63 × oil objective (NA 1.4) and controlled by ZEN 2.1 software.CFP or mTurquoise2, YFP or mNeonGreen, and mScarlet were excited by 405, 488, and 543 nm laser, respectively, and detected at 420-500 nm, 470-540 nm, and 590-690 nm.The thickness of the optical slice is 1 µm.The acquired images were analyzed using Image J 1.54f software (NIH) [36].All experiments were repeated at least three times, and the representative data were shown.

Statistical Analysis
All quantitative data are presented as means ± SEM of at least three independent biological repeats.Analysis of statistical significance was performed using unpaired Student's t-test and paired Student's t-test with GraphPad Prism 9.5.1 software, with a p-value < 0.05 considered statistically significant.

Conclusions
We identified P4HB and PDIA3, two luminal oxidoreductases within the ER dynamically associating with STIM1.This association potentially orchestrates redox modifications on STIM1's two conserved cysteine residues (STIM1-C49-C56).Consequently, these two PDIs may fine-tune STIM1's sensitivity to Ca 2+ ions, thereby regulating its responsiveness to physiological cues and the subsequent generation of SOCE.This intricate mechanism likely plays a pivotal role in coordinating intracellular Ca 2+ signaling and redox responses.Further exploration will undoubtedly shed light on the precise mechanisms by which these oxidoreductases influence intracellular Ca 2+ signaling, as well as their specific roles in cell physiology and pathology.

Author
Contributions: Y.W. and X.Z.supervised and coordinated the study; Y.W. and Y.D. designed the experiments; Y.D. designed and generated most plasmid constructs, with some help from J.L.; Y.D. performed most live cell Ca 2+ imaging, FRET, and confocal experiments; F.W., S.Z. and R.H. performed some Ca 2+ imaging experiments and confocal experiments; P.L. performed some FRET experiments; J.L. performed the FRET calibration experiment and confocal experiments of TXNDC5, P4HB, and STIM1; Y.D. performed qPCR measurements; W.L. provided technical support for imaging and qPCR equipment; Y.D. analyzed data with input from the other authors; Y.W. and Y.D. wrote and revised the manuscript with inputs from all the other authors.All authors have read and agreed to the published version of the manuscript.Funding: This work was supported by the Ministry of Science and Technology of China (2019YFA0802104 to Y.W.), the National Natural Science Foundation of China (92254301 and 91954205 to Y.W.).Institutional Review Board Statement: Not applicable.Informed Consent Statement: Not applicable.