Vimentin deficiency in macrophages induces increased oxidative stress and vascular inflammation but attenuates atherosclerosis in mice

The aim was to clarify the role of vimentin, an intermediate filament protein abundantly expressed in activated macrophages and foam cells, in macrophages during atherogenesis. Global gene expression, lipid uptake, ROS, and inflammation were analyzed in bone-marrow derived macrophages from vimentin-deficient (Vim−/−) and wild-type (Vim+/+) mice. Atherosclerosis was induced in Ldlr−/− mice transplanted with Vim−/− and Vim+/+ bone marrow, and in Vim−/− and Vim+/+ mice injected with a PCSK9 gain-of-function virus. The mice were fed an atherogenic diet for 12–15 weeks. We observed impaired uptake of native LDL but increased uptake of oxLDL in Vim−/− macrophages. FACS analysis revealed increased surface expression of the scavenger receptor CD36 on Vim−/− macrophages. Vim−/− macrophages also displayed increased markers of oxidative stress, activity of the transcription factor NF-κB, secretion of proinflammatory cytokines and GLUT1-mediated glucose uptake. Vim−/− mice displayed decreased atherogenesis despite increased vascular inflammation and increased CD36 expression on macrophages in two mouse models of atherosclerosis. We demonstrate that vimentin has a strong suppressive effect on oxidative stress and that Vim−/− mice display increased vascular inflammation with increased CD36 expression on macrophages despite decreased subendothelial lipid accumulation. Thus, vimentin has a key role in regulating inflammation in macrophages during atherogenesis.


Results
Decreased Lipid Uptake of Native LDL but increased uptake of oxLDL in Vim −/− Macrophages. To elucidate the consequences of vimentin deficiency in macrophages, we first performed global gene expression profiling of bone marrow-derived macrophages from Vim −/− and wild-type mice. We carried out gene set analysis for Gene Ontology (GO) biological process terms and demonstrated that the gene sets with the highest number of significantly vimentin-regulated genes were associated with functions in the immune system process, regulation of NFκB transcription factor activity, defence response to virus, plasma membrane to endosome transport and glutathione metabolic process ( Supplementary Fig. S1). To improve the biological interpretation of these results, we also assessed the extent of up or downregulation of genes within the vimentin-regulated gene sets. We found that immune-related biological functions were associated with upregulated genes whereas glutathione metabolic process was associated with downregulated genes (Fig. 1).
The finding that the global gene expression analysis indicated that the plasma membrane to endosome transport was affected in Vim −/− macrophages was interesting since endocytosis of atherogenic lipoproteins is critical for atherogenesis. We therefore analyzed the uptake of LDL in bone marrow-derived macrophages from Vim −/− and wild-type mice and found decreased internalization of fluorescently labelled native LDL in Vim −/− macrophages ( Fig. 2A). Since the uptake of native LDL is mediated mainly by macropinocytosis in macrophages 24 , we then analyzed the uptake of fluorescent dextran as a measure of this endocytic process. We showed that dextran accumulation was significantly decreased in Vim −/− macrophages compared with wild-type macrophages (Fig. 2B). In contrast, the uptake of fluorescently labelled mildly oxidized LDL (oxLDL) was higher in Vim −/− macrophages than in wild-type macrophages (Fig. 2C). Together, these results show that vimentin deficiency induces a decreased uptake of native LDL, but increased uptake of oxLDL.
Increased Surface Expression of CD36 on Vim −/− Macrophages. We next used flow cytometry to analyse the surface expression of key regulators of lipid metabolism on bone marrow-derived macrophages from Vim −/− and wild-type mice. No differences were seen for low-density lipoprotein receptor (Ldlr), very low-density lipoprotein receptor (Ldlr), macrophage scavenger receptor 1 (MSR1), scavenger receptor B1 (SR-B1), ATP binding cassette A 1 or G1 (ABCA1, ABCG1) (Fig. 2D). However, the analysis showed an increased surface expression of the scavenger receptor CD36 on Vim −/− macrophages (Fig. 2D). Although CD36 mRNA expression did not differ between bone marrow-derived macrophages from Vim −/− and wild-type mice (Fig. 2E), increased CD36 protein levels in Vim −/− macrophages were verified by immunoblot (Fig. 2F). Notably, CD36 is one of the principal receptors for oxLDL and using CD36 knockdown with siRNA in wild-type macrophages we confirmed that mildly oxLDL is internalized via this receptor ( Supplementary Fig. S2B). Our results showed that the increase in oxLDL uptake correlated with increased surface expression of CD36 on Vim −/− macrophages.
Notably, CD36 binds also a number of other ligands including long-chain fatty acids 25 . We therefore incubated Vim −/− and wild-type macrophages with 180 μM oleic acid (OA) for 24 hours 25 . Quantification of the total Oil Red O-stained surface area showed significantly higher fatty acid uptake in Vim −/− macrophages than in wild-type macrophages (Fig. 2G), thus confirming that the CD36 was functional. These results show that vimentin deficiency promotes increased surface expression of functional CD36 on bone marrow-derived macrophages.
Increased Oxidative Stress and Glucose Uptake in Vim −/− Macrophages Associates with Increased Surface Expression of CD36. Lack of vimentin has been shown to augment production of ROS in macrophages 13 . We therefore confirmed these results by analysing levels of hydrogen peroxide and a byproduct of lipid peroxidation, thiobarbituric acid-reactive substances (TBARS) and found that these were significantly higher in Vim −/− macrophages than in wild-type macrophages (Fig. 2H,I). In line, the global gene expression analysis indicated downregulated genes linked to glutathione metabolism (i.e., protection against oxidative stress). We confirmed decreased glutathione reductase expression in bone marrow-derived Vim −/− macrophages using RT-PCR ( Supplementary Fig. S3). These results confirm that vimentin deficiency in macrophages leads to increased oxidative stress. ROS stimulates glucose uptake 26,27 . We therefore analyzed glucose cellular uptake and showed that glucose uptake was significantly higher in Vim −/− macrophages than in wild-type macrophages (Fig. 2J). Furthermore, immunoblot analysis of Vim −/− and wild-type macrophages showed increased levels of GLUT1, the primary rate-limiting glucose transporter on proinflammatory-polarized macrophages 28 in Vim −/− macrophages (Fig. 2K). We also showed that the increase in glucose uptake in Vim −/− macrophages compared with wild-type macrophages was abolished by knockdown of GLUT1 with two different siRNA (Fig. 2L). The GLUT1 protein expression levels after GLUT1 knockdown are shown in Supplementary Fig S4. Interestingly, glucose has been reported to regulate CD36 expression at the translational level in macrophages 29 , we therefore tested if CD36 protein levels could be further increased in Vim −/− macrophages by increasing the glucose concentration from 2.5 to 6 mg/ml (Fig. 2M), showing the importance of glucose availability for CD36 expression. No increased in CD36 protein expression was detected in wild-type macrophages by increasing the glucose concentration   Supplementary Fig. S5). These data indicate that vimentin deficiency induces increased expression of GLUT1 and increased glucose uptake, accompanied by increased expression of CD36 on macrophages.
Increased Inflammation in Vim −/− Macrophages is linked to CD36. Increased glucose uptake and metabolism through GLUT1 has been shown to induce inflammation in macrophages 28 . In line, the global gene expression analysis indicated increased NFκB transcription factor activity in Vim −/− macrophages ( Supplementary Fig. S1). We therefore investigated the inflammatory response of bone marrow-derived macrophages from Vim −/− and wild-type mice and confirmed higher activity of NFκB in Vim −/− macrophages than in wild-type macrophages (Fig. 3A,B). We also showed significantly higher secretion of proinflammatory cytokines in Vim −/− macrophages (Fig. 3C).
Since Vim −/− macrophages displayed increased surface expression of CD36, we tested if the proinflammatory response was, at least partly, mediated by CD36 25 . Results showed that knockdown of CD36 with siRNA in Vim −/− macrophages decreased NFκB activation (Fig. 3D,E) and the secretion of proinflammatory cytokines (Fig. 3F), whereas overexpression of CD36 in wild-type macrophages induced increased NFκB activation (Fig. 3G,H) and increased secretion of proinflammatory cytokines (Fig. 3I). The CD36 protein expression levels after CD36 knockdown and CD36 overexpression are shown in Supplementary Fig. S6. These results indicate that vimentin deficiency in macrophages in vitro is linked to CD36-mediated inflammation.

Decreased Atherogenesis Despite Increased Vascular Inflammation and Increased CD36 expression on macrophages in Vim −/− Mice. To investigate the role of vimentin deficiency in macrophages in
vivo, lethally irradiated low-density lipoprotein receptor-deficient (Ldlr −/− ) mice were transplanted with bone marrow cells from vimentin-deficient (Vim −/− ) mice or wild-type littermates and fed an atherogenic diet for 15 weeks. Body weight and plasma lipids did not differ between the two groups (Supplementary Tables S1 and S2).
Results showed a small but significant decrease in subendothelial lipid accumulation in the aortic root of Vim −/− mice (Fig. 4A,B). Immunohistochemical analysis showed similar numbers of macrophages in the aortic root in both groups of mice (Fig. 4C). In contrast, the number of CD4+ and CD8+ T cells was increased (Fig. 4D) and the number of smooth muscle cells was significantly reduced (Fig. 4E) in mice that received Vim −/− bone marrow, indicating a more inflammatory phenotype. The NFκB target gene intercellular adhesion molecule 1 (ICAM-1), but not vascular cell adhesion molecule 1 (VCAM-1), was increased significantly in lesions of mice that received Vim −/− bone marrow (Fig. 4F,G). In addition, the T cell-recruiting chemokine CXCL1 was increased in plasma from mice that received Vim −/− bone marrow (Fig. 4H), indicating an effect of Vim deficiency on systemic inflammation.
Irradiation of mice induces tissue damage and inflammation. In contrast, AAV viral infection does not elicit any tissue damage or immunologic response 30 . We therefore also injected Vim −/− and wild-type littermates with a single injection of a gain-of-function PCSK9 virus to induce hypercholesterolemia 31 . Body weight and plasma lipids did not differ between the two groups (Supplementary Table S3). In line with the results from the bone-marrow transplantation experiment, the subendothelial lipid accumulation was significantly decreased in the aortic root of Vim −/− mice ( Fig. 5A-C). In addition, the plaque size and CD68 macrophages were also significantly decreased in the aortic root of Vim −/− mice (Fig. 5E). Importantly, immunohistochemistry studies revealed increased expression of CD36 on macrophages in the aortic root of Vim −/− mice (Fig. 5D,F). We also showed significantly higher plasma levels of the proinflammatory cytokines INF-γ, IL-1β, TNF-α, CXCL-1 (Fig. 5G) in Vim −/− mice. These results support increased vascular and systemic inflammation in Vim −/− mice.
As vimentin deficiency impair endothelial function 21,32,33 , we tested if the decreased subendothelial lipid accumulation in Vim −/− mice was caused by compromised transcellular transport of LDL through the endothelium. Alexa fluor-594-labelled LDL was injected into Vim −/− and control mice, and the fluorescent signal of the labelled LDL in sections of the aortic root was quantified six hrs after injection. However, no difference was detected between Vim −/− mice and control mice, indicating that transport of LDL through the endothelium is not impaired in Vim −/− mice ( Supplementary Fig. S7).
Together, these data show that vimetin-deficient mice display modestly decreased atherogenesis despite marked vascular inflammation with increased CD36 expression on macrophages.

Discussion
Here we elucidated the role of vimentin in atherogenesis. Analysis of bone marrow-derived macrophages from Vim −/− mice showed impairment of extracellular lipoprotein uptake, increased markers of oxidative stress, increased GLUT1-mediated glucose uptake, increased CD36 expression, augmented secretion of proinflammatory cytokines, and activation of NFκB. We also linked CD36 to the inflammatory response and showed increased CD36 expression on macrophages in atherosclerotic lesions in vivo.
Atherogenesis is initiated by subendothelial accumulation of atherogenic lipoproteins 9 . The retained lipoproteins become modified (e.g., aggregated and oxidized), and elicit a series of biological responses that develop into an inflammatory response 8 . Thus, the vascular inflammation is caused by retained artery wall lipoproteins. Interestingly, we show that Vim −/− mice developed significantly less atherosclerosis despite increased vascular inflammation. glucose or mannose. Results are shown as mean ± SD, n = 18-20 for (A,H and I), n = 10−12 for (B,G and J), n = 4−8 for (C-F,K,L and M). *p < 0.05, **p < 0.01, ****p < 0.0001, Unpaired two tailed t-test in (A-D,F-K,M) vs Vim +/+ macrophages; One-way ANOVA followed by Sidak's multiple comparison test (CTR siRNA transfected Vim −/− macrophages vs CTR siRNA transfected Vim +/+ macrophages, and GLUT1 siRNA#1 and GLUT1 siRNA#2 respectively vs CTR siRNA transfected Vim −/− macrophages) in L. To clarify the underlying molecular mechanisms we tested if the decreased atherogenesis in Vim −/− mice was due to reduced transendothelial transport of LDL, as vimentin deficiency has been shown to impair endothelial function 21,32,33 . However, results indicated that transport of LDL through the endothelium was not impaired in Vim −/− mice. We also tested if Vim −/− macrophages displayed altered uptake of LDL, and showed reduced uptake of native LDL but increased uptake of oxLDL. Macrophage uptake of native LDL is mediated by macropinocytosis, a receptor independent endocytosis 34 . Once internalized, the LDL is sorted to late endosomes and lysosomes where the cholesterol-esters are degraded to free cholesterol, re-esterified and stored in lipid droplets forming foam cells 35 . Vimentin interacts with Ras-related protein (Rab)-7a and Rab9 [36][37][38] , two proteins that are important in the late endocytic pathway 39,40 . A lack of this interaction may explain the results from the global gene expression profiling indicating an impaired plasma membrane to endosome transport in Vim −/− macrophages. Interestingly, oxLDL is taken up differently by macrophages than native LDL, as binding of oxLDL to CD36 leads to endocytosis through a mechanism that is distinct from macropinocytosis 41 . This difference in endocytosis, may explain why uptake of native LDL is selectively impaired in Vim −/− macrophages.
The role of oxLDL in atherogenesis was recently elegantly demonstrated by Que X et al. who showed that antibodies recognizing the oxidized phosphocholine present on oxLDL, block the uptake of oxLDL to macrophages and greatly reduced atherosclerosis and inflammation 42 . Therefore, our results showing increased uptake of oxLDL in Vim −/− macrophages, despite modestly reduced atherosclerosis may seem counterintuitive. The underlying molecular mechanisms for this remain unclear but may be linked to an impaired transport of cholesterol to the site of re-esterification in Vim −/− macrophages 43 . It is also possible that the impaired uptake of native LDL in Vim −/− macrophages may, at least partly, contribute to the reduced atherosclerosis in vivo, as uptake of native LDL via macropinocytosis may contribute to foam cell formation during atherogenesis 34,35 .
Our results confirmed that vimentin deficiency in macrophages leads to increased oxidative stress 13,16 . Interestingly, we also showed increased GLUT1-mediated glucose uptake, in line with earlier studies showing that ROS augments glucose uptake 26,27 . Glucose is a critical component in the proinflammatory response of macrophages 28 , and our results indicate that the proinflammatory response was, at least partly, mediated by CD36, a central regulator of inflammasome activation 44 .
The regulation of CD36 is complex as the expression of CD36 is regulated at the transcriptional level by various cellular stimuli, including the transcription factor Nrf2 (nuclear factor-erythroid 2-related factor 2) 32, 34 . Under conditions of oxidative stress, Nrf2 is translocated to the nucleus where it initiates transcription of antioxidative genes including CD36 45 . CD36 expression has also been shown to be stimulated by metabolites of arachidonic acid through ROS production 46 , and a recent study showed that ROS mediates cholesterol crystals-induced CD36 expression and foam cell formation 47 . Thus, the increased production of ROS in Vim −/− macrophages could potentially mediate increased expression of CD36 through several mechanisms. However, CD36 mRNA expression was not altered by vimentin deficiency, suggesting that posttranslational regulation might be most important in determining the increased CD36 protein levels observed in Vim −/− macrophages. Notably, glucose augments surface expression of CD36 at the level of translation in macrophages 29 . The role of CD36 in atherogenesis has been controversial 48 . Febbraio et al. crossed a Cd36-null strain with the atherogenic ApoE-null strain and found a 76% decrease in aortic tree lesion area (Western diet) when compared with controls 49 . Febbraio et al. also reported that mice transplanted with Cd36-null bone-marrow were profoundly protected against atherosclerosis, and that re-introduction of macrophages with CD36 induced a twofold increase in the atherosclerotic lesion area 50 . In line, recent studies support a major role for CD36 in atherosclerotic lesion development in vivo 44,48,[51][52][53][54] . However, Moore et al. presented different results, ApoE/Cd36 double-null mice that were fed a high-fat diet had a modest reduction only or even an increase in some atherosclerotic lesions compared with ApoE-null mice 55 . The reason for these conflicting results is unclear 56,57 .
In conclusion, vimentin is highly abundant in activated macrophages and foam cells but its role during atherogenesis has been unknown. Here we demonstrate that vimentin has a strong suppressive effect on ROS and that vimentin deficient mice display increased vascular inflammation with increased CD36 expression on macrophages despite decreased subendothelial lipid accumulation. These findings demonstrate that vimentin has a key role in regulating the inflammation in macrophages during atherogenesis.

Methods
Cell Culture. Bone marrow cells were isolated from Vim −/− and wild-type littermate mice. Femur was cleaned from muscle and tissue and bone marrow was flushed out using DMEM containing 2% heat inactivated fetal calf serum (FCS) as described 58 . See extended Method section on-line. Isolated marrow was washed with PBS containing 10 mM EDTA, red blood cells were lysed in 2% acetic acid, washed once more in PBS EDTA and plated in high-glucose DMEM supplemented with 10% FCS, 1% HEPES, 1% glutamine, 1% gentamicin, 0.01% β-mercaptoethanol, and 10% whole supernatant of cell line CMG14-12 as a source of mouse M-CSF 58 . Experiments were performed on differentiated macrophages 7-10 days after plating. Experiments were performed at least twice with macrophages prepared on different days.

Gene expression analysis.
Total RNA was prepared as previously described 59 from bone marrow-derived macrophages from Vim −/− and wild-type littermate mice and gene expression was measured using the Affymetrix Mouse Genome 430 2.0 Array. Raw probe intensity values were background corrected, normalized with quantile normalization, transformed to the log2 scale, and summarized into probe sets using the Robust Multichip Analysis algorithm 60 . Pair-wise comparison of the gene expression in Vim −/− and wild-type macrophages was performed using the Piano R package 61 (Supplementary Fig. S8). Raw data have been deposited to the GEO database: http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?token=kbqfcegmvhyntkx&acc=GSE63653. mRNA Expression in Bone Marrow Macrophages. Total RNA was extracted using RNeasy Kit (QIAGEN) and cDNA was synthesized using the high capacity cDNA Reverse Transcription Kit (Applied Biosystems). mRNA expression of genes of interest was analyzed with TaqMan real-time PCR in an ABI Prism 7900 HT Detection System (Applied Biosystems).
In Vitro Analysis of Lipoprotein Uptake and Lipid Accumulation. LDL was isolated from human plasma by sequential ultracentrifugation 62 . Mildly oxidized LDL (oxLDL) was prepared by oxidation of LDL with 5 µM CuSO4 for 8 h at 37 °C 63 . LDL and oxLDL were labelled with 1,1′-dioctadecyl-1-3,3,3′,3′-tetramethylindocarbocyanin (Dil) 64 . Macrophages were incubated in medium without serum with 10 µg/ml of Dil-labelled LDL or Dil-labelled oxLDL for 3 h before fixation. The incubation with Dil-labelled oxLDL was done in the presence of 180 µM oleic acid 65 . Micrographs were captured by fluorescence scanner microscopy and lipoprotein uptake was determined by measuring the intracellular fluorescent area per cell 66 . Bone marrow-derived macrophages were incubated for 16 h with 180 μM oleic acid. The total Oil Red O surface area was quantified as described 67 . Macrophage lipids were extracted 68 and quantified using a combination of HPLC and mass spectrometry according to previous work 69 . The use of human plasma was approved by the Regional ethical review board in Gothenburg, informed consent was obtained for the use of plasma, and the study was performed conform the declaration of Helsinki. After the initial surface-staining step, cells were fixed with paraformaldehyde, permeabilized in FACS buffer with 0.5% saponin and then stained for intracellular CD68 using rat-anti-mouse-PE antibodies (AbD Serotec). 10,000 cells were collected for each staining using a FACSCanto II equipped with the Diva 6:2 software (BD Bioscience) and were analyzed using the FlowJo software (Tree Star).
In Vitro Analysis of Cytokine Secretion. Cytokine levels in the media were analyzed with a SECTOR Imager 2400 reader (Meso Scale Discovery).

Analysis of TBARS and H 2 O 2 .
Thiobarbituric acid-reactive substances (TBARS) were determined as described 70 . Fluorescence was measured at 553 nm with 515 nm excitation. Levels of hydrogen peroxide equivalents (H 2 O 2eq ) were analyzed in cultured bone marrow-derived macrophages 70 . The assay is based on the oxidation of ferrous ions to ferric ions by hydrogen peroxide at acidic pH (OXIS International).

Lipid extraction, lipid class fractionation and lipid analysis using mass spectrometer. Lipids
were analyzed according to previous work 71  Induction of Atherosclerosis in Mice using Virus-mediated Overexpression of Mutant PCSK9. Atherosclerosis was induced in female Vim −/− or wild type littermates controls by intravenous injection of adeno-associated viruses containing gain of function mutant (D377Y) of mouse PCSK9 (1.5 × 10 11 vector genomes/mouse) 31 . One day after virus injection of mouse PCSK9, mice were given western atherogenic diet for 12 weeks 31 . Plasma cholesterol and triglyceride levels were collected after 4 h fasting and analyzed using Infinity kits (Sigma).

Analysis of Aortae.
The aortic roots were embedded in OCT Tissue-Tec medium, frozen in dry ice and isopentane, cut into 10-µm-thick cross sections starting from the commissures of the aortic cups upwards, and stained with 0.5% Oil Red O 74 . For bone marrow transplantation model Oil Red O stained area in the plaque (plaque lipids) and vessel area (excluded the lumen) were quantified at 200, 400 and 600 µM distance from the aortic valve cups for each mouse using KS-400 software (Zeiss). The ratio of plaque lipids/vessel area was calculated for each level and thereafter area under the curve was calculated for each mouse. The results are presented as area under the curve for each experimental group. Immunohistochemistry was done with antibodies against α-actin, Mac-2, CD4/CD8, ICAM and VCAM 73 on cross sections of aortic sinuses. For PCSK9-induced model of atherosclerosis sections at 200-240 µm distance from the three aortic valve cups were stained with 0.5% Oil Red O. Images were acquired using Metasystem automated slide scanner (MetaSystems, Germany) equipped with SpectraSplit TM filter system for extended multicolour imaging and a Carl Zeiss AxioImager.Z2 microscope. The Oil Red O fluorescent signal was detected using the Texas red filter (excitation BP 530-585; emission LP 615) 75 . The Oil Red O stained area in the plaque (plaque lipids) and vessel area (excluded the lumen) were quantified using Visiopharm software program version 5.3.0.1562 (Denmark). The lipid accumulation in the plaque was estimated as percent of vessel area.

Analysis of CD36 on Macrophages in Atherosclerotic Lesions.
At sacrifice, the aortic roots were embedded in OCT Tissue-Tec medium, frozen in dry ice and isopentane, sectioned consecutively (10 µM sections) starting from the commissures of the aortic cups upwards. Sections at 200-240 µm distance from the three aortic valve cups were fixed with 2% formaldehyde, blocked with 1% BSA and Fc block (BD Bioscience), and stained with rat anti-CD68-Alexa fluor 594 antibodies (Biolegend) and goat anti-CD36 (R&D systems) antibodies followed by donkey anti-goat Alexa fluor 488 conjugated secondary antibodies. Images were acquired using Metasystem automated slide scanner (MetaSystems, Germany) equipped with SpectraSplit TM filter system for extended multicolour imaging and a Carl Zeiss AxioImager.Z2 microscope. The total macrophage area and the area of CD36 positive macrophages in the atherosclerotic lesions were quantified using Visiopharm software program version 5.3.0.1562 (Denmark). The CD36 positive macrophages were estimated as percentage of total macrophage area.
Blood Analysis. Blood was obtained after a 4 h fast the day before the mice were sacrificed. Cholesterol and triglycerides were measured on a Konelab 20 autoanalyzer (Thermo, Vantaa, Finland). Plasma cytokines were analyzed with a SECTOR Imager 2400 reader (Meso Scale Discovery, Gaithersburg, MD).

Statistical analysis.
Data are shown as means ± SD. Measurements were compared with the two-tailed t-test, Mann-Whitney rank sum test or one-way ANOVA with Sidak's or Dunnett's multiple comparison tests.

Data Availability
The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.