Phosphorylation of Serine 225 in Hepatitis C Virus NS5A Regulates Protein-Protein Interactions

Previously, we reported that phosphorylation of serine 225 within LCSI of NS5A played a role in the regulation of JFH-1 genome replication (24). Mutation of this residue to alanine (S225A; phosphoablatant) resulted in a 10-fold reduction in genome replication and altered subcellular distribution of NS5A, whereas the phosphomimetic mutation (S225D) had no phenotype. To understand the mechanism behind this phenotype, we sought to identify cellular proteins that interacted with NS5A in an S225 phosphorylation-dependent fashion. For this, we exploited the One-Strep tag (OST) affinity purification strategy that we had previously used to identify sites of phosphorylation within NS5A (19). The OST is a peptide that structurally resembles biotin and binds to recombinant streptavidin (Strep-Tactin). S225A and S225D mutations were cloned into pSGR-Neo-JFH-1-5A-OST, which contained the OST cloned into a well-tolerated insertion site near the C terminus of NS5A domain III (25) (Fig. 1a and b). These subgenomic replicons (SGRs) were used to establish stable Huh7 cell lines expressing either the wild-type or the two mutant SGRs. The phenotype of these mutants was confirmed by Western blotting and fluorescence microscopy (Fig. 1c and d). As expected (24), the S225A mutation resulted in a reduction in hyperphosphorylation (Fig. 1c) and in a distribution of the protein that was restricted to the perinuclear region (Fig. 1d). We had previously demonstrated that this phenotype was not a consequence of the reduced level of RNA replication exhibited by the S225A mutant (24). We further confirmed this by expressing either wild-type or S225A mutant NS5A in the context of the NS3-NS5B polyprotein from a T7 RNA polymerase-driven construct (pTM, a kind gift from Volker Lohmann). Following transfection of these plasmids into Huh7-Lunet T7 cells (stably expressing T7 RNA polymerase, also a kind gift from Volker Lohmann), the restricted distribution of NS5A S225A was recapitulated (Fig. 1d), confirming that the phenotype was not dependent on genome replication or the level of NS5A expression. We also applied a superresolution approach (Fig. 1e and f), and this revealed an additional S225A phenotype: discrete clusters of NS5A localizations were equivalent to the diffraction limited puncta observed by wide-field microscopy. Clusters were observed distributed throughout the cytoplasm apart from the S225A mutant, which was more condensed and perinuclear. Consistent with our previous findings (24), larger NS5A clusters were observed for the S225A mutant than for the wild type and the S225D mutant. Taken together, these data demonstrate that S225 phosphorylation regulated not only the distribution of the replication complexes but also their architecture. These observations gave further impetus for the need to understand the molecular mechanism behind this phenotype.

To identify cellular candidates that are potentially involved in the S225A phenotype, we next performed affinity purification from cytoplasmic lysates of Strep-tagged wild-type, S225A, and S225D NS5A and analyzed the bound fractions by mass spectrometry. A large number of known NS5A binding proteins were identified by this procedure. Wild-type NS5A and S225D mutant NS5A showed no difference in their binding to associated proteins (see Data Set S1 in the supplemental material). As we were interested in the role of S225 phosphorylation, we mined the data set for those cellular proteins that bound well to wild-type or S225D NS5A but exhibited a reduced level of interaction with S225A mutant NS5A (Fig. 2). We focused on three of these differentially enriched proteins. Bin1 (also known as amphiphysin II) is a BAR domain-containing protein involved in generating membrane curvature and was previously characterized as an NS5A interactor (26–28). VAP-A (also known as hVAP-33) is a vesicle-associated protein previously shown to participate in the formation of HCV genome replication complexes and to bind to NS5A (29–32). As well as these previously characterized NS5A partners, we also focused on the nucleosome assembly protein 1-like proteins 1 and 4 (NAP1L1 and NAP1L4). NAP1L1 has been shown to interact with HCV core and NS5A (33, 34) as well as Kaposi's sarcoma herpesvirus (KSHV) LANA (35) and HIV-1 Tat (36). Although its roles in the nucleus associated with chromatin structure and gene expression are well defined, NAP1L1 is predominantly located in the cytoplasm, where its function(s) remains to be elucidated. We chose these proteins for further analysis because they exhibited the greatest difference between wild-type and S225A NS5A (Fig. 2b) and thus represented the best candidates to explain the mechanism underpinning the S225A phenotype. As a control, we also examined VAP-B, which was enriched as an NS5A interactor but did not exhibit S225 phosphorylation dependence.

We next sought to validate the proteomic analysis/mass spectrometry results by a specific immunoprecipitation approach. To this end, lysates from stable Huh7 cell lines harboring either the wild-type or S225A/S225D mutant SGRs, or control parental cells, were immunoprecipitated with antibodies to NAP1L1, Bin1, VAP-A, or VAP-B. The immunoprecipitates were probed with an anti-NS5A antiserum by Western blotting (Fig. 3a to d). The adjacent graphs show quantification of 3 independent assays. Consistent with the proteomic analysis, wild-type and S225D mutant NS5A bound equally to all four cellular proteins; in contrast, S225A NS5A bound weakly, or not at all, to NAP1L1, Bin1, and VAP-A. In addition, binding to VAP-B was confirmed to be less dependent on S225 phosphorylation, as there was only a 50% reduction in the level of S225A NS5A bound to VAP-B. Input levels of NS5A and the four cellular proteins were confirmed by Western blotting of lysates and quantification (Fig. 3e, graph). As seen previously, levels of S225A NS5A were reduced by approximately 50%; however, this did not account for the reduction in levels of NS5A bound to NAP1L1, Bin1, or VAP-A. As the reduction in expression of S225A correlated with the observed reduction in the amount of NS5A coimmunoprecipitated with VAP-B, we confirm that the NS5A–VAP-B interaction was not dependent on S225 phosphorylation.

In order to further validate the interactions between NS5A and either NAP1L1 or Bin1, we applied two imaging techniques to examine their colocalization and interaction in Huh7 cells stably harboring SGRs. To complement conventional coimmunofluorescence, which can only determine colocalization, we used the proximity ligation assay (PLA), which allows the detection of direct protein-protein interactions in intact cells (37) PLA involves the use of oligonucleotides attached to secondary antibodies which guide the formation of circular DNA strands when bound in close proximity (approximately 5 to 30 nm). These DNA circles then template localized rolling-circle amplification (RCA), allowing individual interacting pairs of proteins to be visualized and enumerated in fixed samples (37, 38). We were able to apply the PLA technique only for NAP1L1 and Bin1, due to a lack of suitable antibody pairs for VAP-A and VAP-B; thus, we also restricted the coimmunofluorescence analysis to these two proteins.

Both NAP1L1 (Fig. 4a) and Bin1 (Fig. 4b) were distributed diffusely throughout the cytoplasm; in the context of both wild-type and S225D NS5A SGR, there was extensive colocalization with NS5A which was particularly noticeable close to the nucleus and not so apparent at the periphery of the cell (quantification shown in lower part of Fig. 4). S225A NS5A showed a different picture: NS5A and either NAP1L1 or Bin1 seemed to occupy mutually exclusive areas within the cytoplasm, and indeed, quantification showed no significant colocalization. As shown in Fig. 5, we observed strong fluorescence signals (red punctae) for both NS5A/NAP1L1 (Fig. 5a) and NS5A/Bin1 (Fig. 5b) in the cytoplasm of Huh7 cells harboring either wild-type or S225D SGR. In contrast, no PLA signal was observed for either NAP1L1 or Bin1 in cells harboring the S225A mutant SGR. These data confirm that NS5A interacts with both NAP1L1 and Bin1 in the cytoplasm of SGR-harboring cells in an S225 phosphorylation-dependent fashion.

We previously demonstrated that the S225A mutation in the context of either an SGR or infectious virus resulted in a 1-log reduction in genome replication (19, 24). Because this mutation also disrupted binding of NS5A to NAP1L1, Bin1, VAP-A, and, to a lesser extent, VAP-B, we hypothesized that these cellular proteins might play a role in genome replication. To test this, we adopted an siRNA approach to ablate expression of each of these cellular proteins in Huh7 cells stably harboring a wild-type JFH-1 SGR. Cells were transfected with siRNA pools targeting different sites of NAP1L1, Bin1, VAP-A, or VAP-B or with a control nontargeting siRNA and harvested at 72 hours posttransfection (hpt), and protein expression levels were determined by Western blot analysis. As shown in Fig. 6a, ablation of NAP1L1 and Bin1 expression was efficient and resulted in a concomitant reduction in NS5A protein levels (Fig. 6b). The siRNAs for VAP-A and VAP-B were less efficient; however, VAP-A ablation did reduce NS5A expression, although not as effectively as NAP1L1 or Bin1. In contrast, we observed no significant effect of the VAP-B ablation on NS5A levels, possibly because, as reported previously, the role of VAP-B in HCV genome replication is mediated via interactions with both NS5B and NS5A (39). As NS5A expression levels are an indirect measure of genome replication, we also directly assessed the levels of SGR RNA in siRNA-transfected cells by real-time quantitative PCR (qRT-PCR) (Fig. 6c). Consistent with the effects of NAP1L1, Bin1, and VAP-A ablation, we observed reductions of between 100- and 1,000-fold in HCV-specific RNA levels compared to the levels in cells transfected with the control siRNA. Interestingly, despite the fact that VAP-B ablation had no effect on NS5A levels (Fig. 6b), it also significantly reduced HCV RNA levels, although not as effectively as the other 3 targets. Lastly, we confirmed that the ablation of NAP1L1, Bin1, VAP-A, or VAP-B expression had no effect on levels of S225A NS5A (Fig. 6d). These data confirm that the S225 phosphorylation-dependent interacting proteins NAP1L1, Bin1, and VAP-A (and, to a lesser extent, VAP-B) are involved in HCV genome replication.

As we had shown previously (Fig. 1 and reference 24), the S225A mutant NS5A exhibited a perinuclear restricted distribution. We therefore considered that the loss of interaction with cellular factors might be, at least in part, responsible for this restricted distribution. To test this hypothesis, we again used siRNA to ablate expression of NAP1L1, Bin1, VAP-A, or VAP-B in Huh7 cells stably harboring a wild-type SGR. Cells transfected with siRNA were analyzed by immunofluorescence with antibodies to both NS5A and the four cellular proteins. As shown in Fig. 7a, NAP1L1 ablation led to an overall reduction in NS5A expression which was restricted to a perinuclear distribution compared with the negative-control siRNA-transfected (scrambled) or untransfected (mock) cells. To provide quantitative confirmation of these observations, we determined NS5A spatial distribution data for 12 randomly selected cells (Fig. 7a, graph). A similar pattern was observed when Bin1 (Fig. 7b) or VAP-A (Fig. 8a) expression was ablated. In contrast, ablation of VAP-B expression had only a modest effect on the distribution of NS5A which did not reach significance upon quantitative analysis (Fig. 8b). Lastly, we sought to confirm that the effects of NAP1L1, Bin1, or VAP-A ablation on NS5A distribution were not due to a reduction in levels of expression, as an indirect consequence of the inhibition of genome replication. To test this, Huh7-Lunet T7 cells were transfected with pTM-NS3-5B followed by NAP1L1, Bin1, or VAP-A siRNAs. As shown in Fig. 9, the restricted distribution of NS5A was also observed using this expression system, confirming that the phenotype was not dependent on genome replication or the level of NS5A expression. These data demonstrate that S225 phosphorylation is required for the interaction of NS5A with a number of cellular proteins and that these interactions control both the distribution of NS5A and HCV genome replication.

A DNA construct of the neomycin phosphotransferase (Neo) containing subgenomic replicon pSGR-Neo-JFH-1-5A-OST, in which the One-Strep tag (OST) was introduced into the C terminus of NS5A domain III (25), was used in this study (Fig. 1a and b). Previously, the OST has been shown to have no effect on virus genome replication (25). Wild-type pSGR-Neo(RsrII^del)-JFH-1-5A-OST was first constructed by abolishing an RsrII site within the Neo gene using a Q5 site-directed mutagenesis kit (New England BioLabs [NEB]). Then the desired NS5A mutants from SGR-luc-JFH-1 (24) were cloned into pSGR-Neo(RsrII^del)-JFH-1-5A-OST via flanking Nsil/RsrII, and correct insertion was confirmed by sequencing.

Huh7 cells were maintained in Dulbecco's modified Eagle's medium (DMEM; Sigma-Aldrich) supplemented with 10% fetal bovine serum (FBS), 100 IU of penicillin/ml, 100 μg of streptomycin/ml, and 1% nonessential amino acids (NEAA) in a humidified incubator at 37°C with 5% CO₂. Huh7 cells carrying a subgenomic JFH-1 replicon (SGR-Neo-JFH-1) were maintained in the same medium supplemented with 300 μg/ml of G418 (BioPioneer). Huh7-Lunet T7 cells (expressing T7 RNA polymerase, a kind gift from Volker Lohmann) were maintained in the same medium supplemented with 5 μg/ml of Zeocin (Life Technologies).

The preparation of in vitro transcripts and electroporations for pSGR-Neo-JFH-1-5A-OST-derived constructs were conducted as described previously (25). In brief, 2 × 10⁶ Huh7 cells in diethyl pyrocarbonate (DEPC)–phosphate-buffered saline (PBS) were electroporated with 3 μg of in vitro RNA transcripts using a square-wave protocol at 260 V for 25 ms. Subsequently, cells were resuspended in complete DMEM and seeded at a culture area of 1 × 10⁴ cells/cm² into 10-cm dishes. Forty-eight hours postelectroporation, the cells were selected with 300 μg/ml of G418. After 2 weeks, emerging colonies were pooled and kept under continuous G418 selection for 1 to 2 weeks in order to establish stable SGR-harboring cell lines. The maintenance of the S225 mutations was verified by RT-PCR and sequencing.

To detect proteins bound to OST-tagged NS5A by affinity purification and mass spectrometry, cell pellets from Huh7 cells stably harboring NS5A wild-type (WT) and S225A and S225D mutant SGRs were prepared by snap-freezing cells in liquid nitrogen. Cell pellets were lysed in TAP lysis buffer (50 mM Tris-HCl [pH 7.5], 100 mM NaCl, 5% [vol/vol] glycerol, 0.2% [vol/vol] Nonidet P-40, 1.5 mM MgCl₂) in the presence of protease inhibitor cocktail (EDTA free, cOmplete; Roche), phosphatase inhibitor (PhosSTOP; Roche), and 750 U of Benzonase (Core Facility, Max Planck Institute of Biochemistry [MPI-B]) for 30 min on ice. After incubation on ice, cell lysates were sonicated using a Bioruptor with 15 alternating 30-s on/off cycles and clarified by centrifugation at 16,000 × g. Subsequently, Strep-Tactin Sepharose beads (Iba) were incubated with 2 mg of protein of clarified lysate for 60 min at 4°C and washed once with TAP lysis buffer and three times with TAP wash buffer (50 mM Tris-HCl [pH 7.5], 100 mM NaCl, 5% [vol/vol] glycerol, 1.5 mM MgCl₂) lacking Nonidet P-40 to remove residual detergent. Sample preparation and LC-MS/MS analysis were performed as described previously (49). Briefly, four independent affinity purifications were performed for each bait, samples were sequentially digested with LysC (Wako Chemicals, USA) and trypsin (Promega), acidified with 0.1% trifluoroacetic acid (TFA), desalted with C18 stage tips, and analyzed by liquid chromatography coupled to mass spectrometry on an Orbitrap XL instrument (Thermo Fisher Scientific). Mass spectrometry raw files were processed with MaxQuant software version 1.5.5.1 (50) using the built-in Andromeda engine to search against human and JFH-1 proteomes containing forward and reverse sequences. Additionally, the label-free quantification (LFQ) (51) algorithm and Match Between Runs option were used. Perseus software version 1.5.5.1 was used to further process the data. In this manner, only proteins identified on the basis of at least two peptides and a minimum of three quantification events in at least one experimental group were considered. LFQ protein intensity values were log transformed and missing values filled by imputation. Significantly enriched proteins were determined by Welch's t test with permutation-based false-discovery rate (FDR) statistics, performing 250 permutations. The FDR threshold was set to 0.01 and S0 parameter was set to 0.1 to separate background from specifically enriched proteins. Results were plotted using R (https://www.R-project.org ) and visually adapted using Adobe Illustrator.

Cells were washed twice in PBS, lysed in 1× Glasgow lysis buffer (GLB; 1% [vol/vol] Triton X-100, 120 mM KCl, 30 mM NaCl, 5 mM MgCl₂, 10% [vol/vol] glycerol, and 10 mM PIPES-NaOH, [pH 7.2], with protease and phosphatase inhibitors) and harvested by centrifugation (2,800 × g, 10 min, and 4°C) before determination and normalization of protein concentration by bicinchoninic acid (BCA) assay (Pierce). Following separation by SDS-PAGE, proteins were transferred to a polyvinylidene difluoride (PVDF) membrane and blocked in 50% (vol/vol) Odyssey blocking (OB) buffer (LI-COR) in Tris-buffered saline (TBS). The membrane was incubated with primary antibodies overnight at 4°C, followed by secondary antibodies for 2 h at room temperature, both prepared in 25% OB buffer. Primary antibodies used were anti-NS5A (sheep, prepared in-house) at 1:4,000 (52), anti-NAP1L1 (rabbit; Santa Cruz) at 1:350, anti-Bin1 (rabbit; Generon) at 1:300, anti-VAP-A (rabbit; Generon) at 1:1,000, anti-VAP-B (rabbit; Generon) at 1:1,000, and anti-α-actin (mouse; Sigma) at 1:10,000. Secondary antibodies were anti-rabbit, anti-sheep (800 nm), or anti-mouse (700 nm) antibodies, used at 1:10,000 prior to imaging using a LI-COR Odyssey Sa infrared imaging system. Quantification of Western blots was carried out using Image Studio v3.1 (LI-COR) using a background subtraction method.

Cells were washed with PBS before fixation for 15 min at room temperature in 4% (wt/vol) paraformaldehyde (PFA); cells were subsequently permeabilized in 0.1% (vol/vol) Triton X-100–PBS and blocked with PBS-Tween (PBS-T) and 5% (wt/vol) bovine serum albumin (BSA) before immunostaining for anti-NS5A (mouse monoclonal, 1:1,000) and either anti-NAP1L1 (rabbit monoclonal, 1:100) or anti-Bin1 (rabbit monoclonal, 1:50) overnight at 4°C. Coverslips were washed 3 times for 5 min in PBS-T buffer under gentle shaking and incubated with proximity ligation assay (PLA) probes Duolink In Situ PLA Probe Anti-Mouse PLUS (DUO92001; Sigma-Aldrich) and Duolink In Situ PLA Probe Anti-Rabbit MINUS (DUO92005) for 2 h at 37°C. For PLA, all incubations were performed in a preheated humidity chamber and according the manufacturer's recommendations using a Duolink In Situ Detection Reagents Red kit (DUO92008). Coverslips were washed 3 times for 5 min in PBS-T buffer under gentle shaking and incubated with a DNA ligase previously diluted in ligation buffer for 30 min at 37°C. Coverslips were washed 3 times for 5 min in PBS-T buffer under gentle shaking and incubated with a DNA polymerase previously diluted in amplification buffer for 90 min at 37°C. Finally, coverslips were washed for 20 min under gentle shaking and then washed for 2 min with PBS and air dried. Coverslips were mounted with Duolink In Situ mounting medium with 4′,6-diamidino-2-phenylindole (DAPI), and fluorescence was visualized with a Zeiss LSM880 upright microscope.

SGR-harboring Huh7 cells or Huh7-Lunet T7 cells were transfected with 10 nM pooled siRNA (Santa Cruz) or 10 nM AllStars negative-control siRNA (Qiagen) using Lipofectamine 2000 (Invitrogen) according to the manufacturer's protocol. sc-75871 targets NAP1L1, sc-29804 targets Bin1, sc-61768 targets VAP-A, and sc-61770 targets VAP-B. For Huh7-Lunet T7 cell experiments, the cells were transfected with pTM-NS3-5B plasmid at 48 h posttransfection (hpt). Cells were incubated in DMEM supplemented with 5% fetal calf serum (FCS) and harvested at 72 hpt. Proteins and total RNA were isolated with TRIzol (Life Technologies), and subsequent Western blotting and real-time quantitative PCR (qRT-PCR) were performed.

To quantify the number of HCV genomes, total cell RNA was extracted using TRIzol reagent by following the manufacturer's instructions (Invitrogen). Total extracted cellular RNA (100 ng) was analyzed using a one-step qRT-PCR TaqMan-based kit (Eurogentec), with primers and probe designed against the 5′ untranslated region (UTR) as described previously (19, 53).

Cells were washed with PBS before fixation for 20 min in 4% (wt/vol) PFA; cells were subsequently permeabilized in 0.1% (vol/vol) Triton X-100 and PBS and blocked with PBS-T and 5% (wt/vol) BSA before being immunostained with primary antibody as described above. Various fluorescently conjugated secondary antibodies were used at 1:500 (Life Technology). Nuclei were counterstained with DAPI. Confocal microscopy images were acquired on a Zeiss LSM880 upright microscope with Airyscan; postacquisition analysis was conducted using Zen software (Zen version 2015 black edition 2.3; Zeiss) or Fiji (version 1.49) software (54).

The direct stochastic optical reconstruction microscopy (dSTORM) system described previously (55) was modified with a cylindrical lens (f = 150 mm, where f is focal length; Thorlabs). Round glass 25-mm-diameter coverslips (Warner Instruments) were cleaned in a 1:1:5 solution of NH₃ (aq), H₂O₂, and H₂O at 80°C for 16 h. Cells stably harboring NS5A wild-type, S225D, and S225A mutant subgenomic replicons were seeded onto cleaned coverslips at 1 × 10⁵ cells per well in six-well plates. Coverslips were fixed in 2% (wt/vol) PFA in normal medium for 10 min, permeabilized in 0.2% Triton X-100 in PBS for 10 min, and blocked in 1% (vol/vol) normal donkey serum (Sigma-Aldrich) for 1 h. Mouse monoclonal anti-NS5A antibody was directly labeled at an approximately 1:1 ratio using the carboxylic acid succinimidyl ester of the photoswitchable dye Alexa Fluor 647 (Life Technologies) in PBS containing 125 mM NaHCO₃ in the dark for 30 min. Unincorporated dye was removed by size exclusion using Zeba microspin desalting columns (Thermo Fisher Scientific). Immunostaining was conducted with directly labeled antibody (1:2,000) for 1 h. Coverslips were then treated with 0.01% poly-l-lysine (Sigma-Aldrich) for 10 min and incubated with a suspension of 150-nm gold nanoparticles (Sigma-Aldrich).

The image acquisition and processing software was used as described previously (56). Labels were stochastically activated with 642-nm laser excitation under wide-field illumination in the presence of fluorescence quenching buffer (glucose oxidase [10 U], catalase [50 U], 12.5 mg ml⁻¹ of d-glucose, 1 mM 2-mercaptoethylamine in PBS [pH 8.0]). Data sets consisted of 11,000 image frames at a frame rate of 20 Hz. Localized emission events were binned into histograms for display and correction of image distortion by the cylindrical lens. Image smoothing was conducted in R using kernel density estimation to reflect x-y-z localization precisions measured from fiducial markers, and z-stacks were visualized in Fiji. Clustering analysis was conducted in Python using density-based spatial clustering of applications with noise (DBSCAN) (57) on the localization coordinates extracted from the palm3d software (MinPts = 30; ε = 150 nm). Cluster sizes were determined from the mean Euclidean distance between all localizations in identified clusters. Image analysis in R and Python used custom scripts (available on request), and statistical tests were conducted in GraphPad Prism using a one-way analysis of variance (ANOVA) with Tukey's multiple comparisons.

For quantification of NS5A spatial arrangement, images were acquired with the same acquisition parameters, but with variable gain to ensure correct exposure. The spatial coordinates of NS5A were determined using the FindFoci function of the GDSC plugin for Fiji, with the nuclear envelope being manually outlined (utilizing the DAPI staining as a reference) and coordinates generated by Fiji. The distance from each NS5A to the nuclear envelope was then determined using trigonometry. NS5A spatial distribution data were generated for 12 randomly selected cells for each replicon variant and data combined into a box-and-whisker plot.

For colocalization analysis, Manders' overlap coefficient was calculated using ImageJ software with Just Another Co-localization Plugin (JACoP) (National Institutes of Health). Coefficient M1 reflects the fraction of the anti-NS5A signal that overlaps either the anti-NAP1L1 or anti-Bin1 signal. Coefficient M2 reflects the fraction of either the anti-NAP1L1 or anti-Bin1 signal that overlaps the anti-NS5A signal. Coefficient values range from 0 to 1, corresponding to nonoverlapping images and 100% colocalization images, respectively. Colocalization calculations were performed on >10 cells from at least two independent experiments.

Data sets were analyzed using Student's t test assuming a two-tailed, unequal variance to determine statistical difference from the wild type (WT) (n = 3 or greater throughout).