The amphiregulin/EGFR axis has limited contribution in controlling autoimmune diabetes

Conventional immunosuppressive functions of CD4+Foxp3+ regulatory T cells (Tregs) in type 1 diabetes (T1D) pathogenesis have been well described, but whether Tregs have additional non-immunological functions supporting tissue homeostasis in pancreatic islets is unknown. Within the last decade novel tissue repair functions have been ascribed to Tregs. One function is production of the epidermal growth factor receptor (EGFR) ligand, amphiregulin, which promotes tissue repair in response to inflammatory or mechanical tissue injury. However, whether such pathways are engaged during autoimmune diabetes and promote tissue repair is undetermined. Previously, we observed that upregulation of amphiregulin at the transcriptional level was associated with functional Treg populations in the non-obese diabetic (NOD) mouse model of T1D. From this we postulated that amphiregulin promoted islet tissue repair and slowed the progression of diabetes in NOD mice. Here, we report that islet-infiltrating Tregs have increased capacity to produce amphiregulin, and that both Tregs and beta cells express EGFR. Moreover, we show that amphiregulin can directly modulate mediators of endoplasmic reticulum stress in beta cells. Despite this, NOD amphiregulin deficient mice showed no acceleration of spontaneous autoimmune diabetes. Taken together, the data suggest that the ability for amphiregulin to affect the progression of autoimmune diabetes is limited.

a susceptibility gene for diabetic neuropathy and diabetes associated cardiac dysfunction 12,13 .We hypothesized, however, that amphiregulin could have a direct role in beta cell survival, based on the observations that beta cells require EGFR signaling for normal development and postnatal proliferation 14 , and that constitutive activation of EGFR on beta cells protects against streptozotocin induced diabetes 15 .
Here, we report that islet-infiltrating Tregs upregulate amphiregulin in the NOD mouse model of autoimmune diabetes.Moreover, we show that amphiregulin can have a direct effect on beta cells by reducing ER stress in ex vivo cultured islets.However, the deletion of amphiregulin in NOD mice does not change spontaneous autoimmune diabetes development, suggesting that any protective effects induced by amphiregulin are not sufficient to maintain or recover beta cell mass.Therefore, our data suggest that the amphiregulin/EGFR axis is not always effective in controlling inflammatory tissue damage.While potentially promising, the use of amphiregulin as a therapeutic target presents several challenges that will need to be addressed through further study.

Islet-infiltrating Tregs upregulate amphiregulin
We had previously observed that islet-infiltrating CD4 + Foxp3 + Tregs upregulate amphiregulin at the transcriptional level 8,9 .To confirm that amphiregulin was indeed expressed by islet-infiltrating Tregs at the protein level, we used flow cytometric analysis of T cells isolated directly from infiltrated islets or peripheral lymphoid organs and stimulated in vitro.Islet, pancreatic-draining lymph node, and non-draining lymph node T cells from prediabetic 14-week-old female NOD mice were analyzed for amphiregulin expression.First, our data show that CD4 + Foxp3 + Tregs expressed amphiregulin at a greater frequency compared to Foxp3 -CD4 T effectors (Teffs) and CD8 Teffs regardless of the tissue site (Fig. 1a,b).Secondly, among Tregs, the highest frequency of amphiregulin expression was observed within the pancreatic islets (~40%+) (Fig. 1b), suggesting that tissue resident Tregs have an enhanced capacity to express amphiregulin.
To determine whether Tregs actively produce amphiregulin in vivo, we treated NOD mice with brefeldin A to block amphiregulin secretion, and then analyzed intracellular amphiregulin directly ex vivo.Introduction of systemic brefeldin A in mice blocks cytokine secretion, and is an effective approach to detect in vivo cytokine production through subsequent flow cytometric analysis 16 .In parallel, we sought to determine if amphiregulin expression by Tregs changed during disease progression by analyzing both younger and older mice.While overall CD4 T cell frequency and number increased in older mice (15-17-week-old) compared to younger mice (6-week-old) (Fig. 1c), Foxp3+ frequency remained consistent, although Foxp3+ cell number did increase (Fig. 1d), indicating an expected increase in the number of immune cells infiltrating into the pancreatic islets.Frequency of amphiregulin positive cells within the Foxp3 + islet population was unchanged overtime (~10%+) (Fig. 1e).Moreover, Foxp3 + Tregs maintained amphiregulin production at diabetes onset (Fig. 1f), suggesting that islet-resident Tregs retain the capacity to produce amphiregulin throughout the course of disease.However, older mice had a greater frequency of amphiregulin positive Tregs in both draining and non-draining lymph nodes compared to younger mice, suggesting systemic activation of regulatory T cells at later stages of disease (Fig. 1e).These data show that Foxp3 + Tregs continue to be major producers of amphiregulin throughout disease progression.
Prior models of injury in which Treg derived amphiregulin was essential to tissue recovery showed coexpression of amphiregulin and the IL-33 receptor, ST2 3 , consistent with the ability for the IL-33/ST2 pathway to induce amphiregulin expression in Tregs 17,18 .To determine if islet Tregs similarly expressed ST2, we analyzed ST2 expression on islet CD4 T cells by flow cytometry.In congruence with previous reports 19 , a higher frequency of Tregs expressed ST2 compared to Teffs (Fig. 1g).Taken together, these data show that islet-resident Tregs express the canonical markers of tissue repair Tregs, amphiregulin and ST2.

Beta cells express amphiregulin receptor EGFR
Once we defined Tregs as a major source of amphiregulin within the islets, we analyzed ex vivo expression of the amphiregulin receptor, EGFR, to determine the potential cellular targets of amphiregulin.We observed a relatively high level of EGFR expression on insulin positive beta cells (~73% EGFR+), and beta cell EGFR levels remained consistent between 6-week and 15-17-week-old NOD mice (Fig. 2a,b).To determine whether inflammatory signals during T1D modify levels of EGFR expression on beta cells, we measured EGFR levels on beta cells from NOD.scid mice, which lack T and B cells and do not develop autoimmunity.As there was no difference in EGFR expression between NOD and NOD.scid mice (Fig. 2b), we concluded that beta cell EGFR expression is stable regardless of inflammation, which suggested that beta cells could respond to amphiregulin throughout disease.
In addition to beta cells, a proportion of both CD4 + effector and CD4 + Foxp3 + regulatory T cells within the islets expressed EGFR on their cell surface, although Tregs expressed significantly higher levels (~8% compared to ~22%) (Fig. 2c).Therefore, it is possible that Tregs may utilize amphiregulin in a paracrine or autocrine fashion during autoimmunity, as well as secrete amphiregulin to promote beta cell tissue repair 20 .Furthermore, Treg expression of EGFR did not change over the course of disease, as Tregs from 15 to 17-week-old NOD mice maintained EGFR expression comparable to that of 6-week-old mice (Fig. 2d), thus retaining an ability to sense amphiregulin regardless of age or disease status.
Although levels of EGFR expressed on beta cells did not differ between normal and inflammatory conditions, previous studies suggested that EGFR expression on Tregs could be induced in response to inflammation 20 .To determine if EGFR expression is induced on Tregs in response to autoimmune stress, we compared EGFR levels on Tregs in the islets to the levels expressed by Tregs in the non-draining lymph nodes (Fig. 2e).Tregs in the islets indeed showed higher levels of EGFR expression compared to the non-draining lymph nodes, indicating

Amphiregulin modulates BiP, a mediator of the UPR and ER stress in beta cells
The high demand for insulin production in beta cells results in transient endoplasmic reticulum (ER) stress in response to glucose stimulation 21 .However, under chronic ER stress conditions, such as those found during diabetes, the unfolded protein response (UPR) may lead to loss of beta cell function, and further unresolved ER stress can lead to beta cell death 22,23 .Importantly, amphiregulin was previously shown to be upregulated during beta cell recovery phase after transient hyperglycemia induced by deletion of the UPR sensor IRE1α 24 , suggesting a role for amphiregulin in modulating beta cell mass in response to ER stress.Thus, we asked if amphiregulin was directly capable of reducing ER stress in beta cells.Using thapsigargin, a SERCA pump inhibitor, we induced ER stress in cultured NOD.scid islets, and measured transcription of genes commonly associated with the three arms of the UPR (Fig. 3).Insulin expression served as a control for similar beta cell distribution among the treatment groups.All of the assessed ER stress genes were upregulated in response to thapsigargin treatment, as expected.ER stressed islets were then treated with recombinant amphiregulin protein to assess the potential for amphiregulin to modulate thapsigargin induced ER stress.Early downstream targets of the UPR (sXBP1, ATF4, and ATF6) were unchanged with the addition of amphiregulin, although sXBP1 and ATF6 did trend towards decreased transcription levels.In addition, transcription of the pro-apoptotic factor, CHOP, was unaltered by amphiregulin.However, transcription levels of the chaperone binding immunoglobin protein (BiP) (Grp78) were significantly downregulated with amphiregulin treatment, showing that amphiregulin can directly act upon beta cells.While this evidence suggests that amphiregulin cannot completely prevent or reduce ER stress in ex vivo cultured islets, amphiregulin may still be able to moderate certain aspects of the UPR.As BiP is situated upstream of ER stress transcriptional changes, the return to basal Grp78 levels in the amphiregulin treated group may indicate the return of beta cells to normal, transient ER stress.

Lack of systemic inflammation in amphiregulin deficient NOD mice
To determine the physiological importance of amphiregulin in the development and progression of autoimmune diabetes, we backcrossed the amphiregulin mutation onto the NOD genetic background to create amphiregulin deficient NOD mice.Areg Mcub(−/−) is a point mutation in the amphiregulin locus in a donor splice site of exon 1, causing the formation of a premature stop codon 25 .Loss of amphiregulin production was confirmed by ELISA using enriched and ex vivo stimulated CD4 T cells (Fig. 4a).To determine if this mutation within the amphiregulin gene led to global immune dysregulation on the susceptible NOD background, markers of T cell  Areg +/+ ) littermate controls.There were no differences in the level of activation of peripheral CD4 T effector cells based on the expression of CD44, CD69, or Ki67 (Fig. 4b), and CD4 + Foxp3 + Treg frequencies in amphiregulin deficient mice were similar to wild-type control mice (Fig. 4c).Tregs from both groups also showed a similar level of activation based on CD44, CD69, and Ki67; however, there was a slight, but significant increase in CD25 expression in amphiregulin deficient mice (Fig. 4d).Interestingly, there was a trend towards decreased EGFR levels on both Teffs and Tregs of NOD.Areg −/− mice, possibly as a result of reduced positive feedback in response to amphiregulin binding (Fig. 4e).Overall, these observations did not point to systemic immune dysregulation or loss of Tregs in amphiregulin deficient animals.

Amphiregulin expression does not impact development of autoimmunity
To fully determine whether loss of amphiregulin in NOD mice leads to reduced beta cell survival and accelerated diabetes, we monitored female NOD.Areg −/− mice and wild-type littermate controls for the development of spontaneous autoimmune diabetes (Fig. 5a).Surprisingly, we detected no difference in diabetes incidence or progression between the two groups.The similarity in diabetes development between NOD.Areg +/+ and NOD.Areg −/− mice indicates that amphiregulin does not have a role in autoimmune diabetes disease progression.Finally, to determine if the limited role of amphiregulin during NOD spontaneous diabetes development was organ specific, and if amphiregulin could have a functional role in other types of autoimmunity, we induced experimental autoimmune encephalomyelitis (EAE), a well-established model of multiple sclerosis, in 7-weekold female C57BL/6.Areg −/− and littermate wild-type control mice.EAE was induced with a MOG 35-55 peptide emulsion and pertussis toxin as described previously 26 .Similar to the results obtained from our NOD diabetes model, amphiregulin seemed to have no impact on the development of EAE (Fig. 5b).This was in congruence with others who have suggested that amphiregulin is dispensable for protective Treg function during EAE 27 .

Discussion
Regulatory T cells that infiltrate or reside within tissues are often uniquely adapted to be effective within their environment.In some situations, Tregs with repair functions exhibit a unique repair transcriptional signature, indicating a specialized subpopulation or lineage 28,29 .Based on single-cell transcriptomics, tissue-resident www.nature.com/scientificreports/Tregs differentiate from a lymphoid tissue phenotype into subpopulations that have conserved transcriptional programs 30 .Some of these programs involve distinctive functions such as maintenance of tissue homeostasis and tissue repair, and include the EGFR ligand amphiregulin.Here we show that pancreatic Tregs can express amphiregulin in addition to other known immunosuppressive molecules 1 .Furthermore, islet-resident Tregs express ST2, the receptor for the alarmin cytokine IL-33.The expression of ST2 and amphiregulin on isletinfiltrating Tregs implies that these cells can respond to tissue damage signals as part of a repair program, in accordance with previous findings 3,17,18 .Importantly, we show that beta cells express high levels of the amphiregulin receptor, EGFR, suggesting that beta cells can respond to amphiregulin.In fact, treatment with recombinant amphiregulin during thapsigargin induced ER stress in cultured islets significantly reduced Grp78 (BiP) transcription, a regulator of the unfolded protein response pathway.All of these observations point to amphiregulin as a potentially important mechanism employed by islet-infiltrating Tregs to curb beta cell destruction.Surprisingly, however, amphiregulin deficiency in NOD mice did not significantly alter autoimmune diabetes incidence or progression.The lack of impact of amphiregulin deficiency on autoimmune diabetes led us to question whether amphiregulin also fails to protect in other autoimmune models.Using C57BL/6 amphiregulin deficient mice, we saw that loss of amphiregulin did not impact the severity of EAE induced by MOG 35-55 peptide immunization.This observation is consistent with a recent study showing that amphiregulin knock-down in Treg adoptive therapy does not impede Treg ability to suppress EAE 27 .The protective role of amphiregulin seems to be disease and/or context dependent, as amphiregulin is effective in controlling immune responses during autoimmune colitis 20 and pristane-induced lupus nephritis 31 .
The effectiveness of amphiregulin in tissue repair may also be dependent on the target cell.In mice, adult beta cells show a relatively low proliferative capacity 32 .This low regenerative potential may limit the ability for amphiregulin to repair damaged beta cells at a rate that can compete with the damage caused by autoimmune assault during T1D.Furthermore, in certain previously examined disease models, the target of amphiregulin during injury are stem-like cells that undergo induced differentiation into tissue cells to help support tissue repair.For example, amphiregulin expressing Tregs that accumulate in muscle fibers following acute injury prompt the differentiation of muscle satellite cells 5 .Such stem-like pre-cursor cells for beta cells have not been identified.In contrast to acute muscle injury, chronic infection with the parasite Toxoplasma gondii, increases amphiregulin expression in muscle Tregs, but does not prevent tissue damage 33 .In the case of T. gondii, Tregs appear to increase the ratio of IM/M1 (inflammatory monocytes/pro-inflammatory macrophages) to M2 macrophages (pro-regenerative macrophages) preventing skeletal muscle fiber regeneration.Therefore, amphiregulin has limited function under certain inflammatory conditions.
As the field looks to innovative ways of preventing, attenuating, or reversing T1D disease progression, several methods have become promising.One of these approaches is to attenuate the autoimmune attack on beta cells by reducing the number of activated autoreactive T effectors.Teplizumab-the first FDA approved immunotherapy to delay T1D-is a monoclonal antibody against CD3 that showed increased frequencies of unresponsive CD8 T cells in treated individuals 34 .In addition, current clinical trials are using expanded autologous Tregs re-infused into patients to suppress autoimmune responses 35 .However, a second and complementary strategy in T1D therapies is directed towards improving beta cell survival and restoring beta cell mass and function, either through the protection of beta cells against autoimmune attack, or through beta cell regeneration.Whether cellular Treg therapy could be harnessed to enhance beta cell survival in addition to inducing autoimmune suppression is unknown, but is an active area of investigation.Apart from endogenous Treg functions, cells could be enhanced to achieve additional functional competency.For example, Tregs could be re-engineered in order to increase antigen specificity either through exogenous CAR or TCR 36 .Treg expression of amphiregulin can be driven by alarmin cytokines (i.e.IL-33 and IL-18) that are produced by injured tissues 3,4,37 , and tissue-resident Tregs are poised to respond to damage since they exhibit high expression of the IL-33 receptor, ST2 38 .Indeed, a recent study has overexpressed ST2 on human Tregs to successfully induce TCR-independent reparative mechanisms, in particular increasing the production of amphiregulin 18 .Furthermore, treatment of T1D patient PBMCs in vitro with IL-33 increased Treg frequencies and FOXP3 expression 39 , consistent with previous reports indicating an ability of IL-33 to enhance Treg induction 40 .Thus, with these new technologies emerging, it is of great interest to understand whether the IL-33/amphiregulin/EGFR axis can act directly or indirectly to retain beta cell function during T1D, and whether amphiregulin should be further pursued as a therapeutic target for T1D.
During autoimmune diseases such as T1D, many tissue resident Tregs are antigen specific 41 and express markers associated with enhanced immunosuppressive functions 42 ; however, it is unclear whether pancreatic Tregs also have a non-immunological purpose within the islets.Our findings provide evidence for a population of islet-infiltrating Tregs that express amphiregulin and ST2; markers associated with tissue repair Tregs.However, while amphiregulin can mitigate beta cell ER stress, analysis of NOD amphiregulin deficient mice suggests that amphiregulin does not overtly impact autoimmune diabetes incidence.Why the amphiregulin/EGFR pathway is not effective in supporting beta cell function is unclear, but could point to another reason for the inability of Tregs to effectively control anti-beta cell autoimmunity.As our study did not investigate the outcomes of other tissues often damaged by hyperglycemia (kidneys, blood vessels, heart), it is possible amphiregulin could have an effect in improving other comorbidities commonly associated with T1D 13 .
Lastly, it is important to consider potential limitations of our study before ruling out amphiregulin as an effective pro-beta cell factor.The first being that the effects of amphiregulin during autoimmunity were studied using mice that have a global deletion of amphiregulin.The amphiregulin mutation may have resulted in activation of compensatory mechanisms, such as upregulation of other EGFR ligands.Alternatively, Tregs are known to compensate for lost regulatory functions via upregulation of alternative suppressive mechanisms 43 .Secondly, although islet Tregs produce amphiregulin and beta cells express EGFR, we did not investigate whether EGFR is actively signaling in beta cells.It is possible that post-translational processing of amphiregulin, or downstream EGFR signaling is changed during T1D leading to sub-optimal amphiregulin/EGFR axis activation.Additional studies will be necessary to determine whether there are deficiencies in beta cell response to amphiregulin during autoimmune diabetes and inflammatory tissue damage.

Mice
All mice were housed and used according to IACUC approval at Baylor College of Medicine Houston, TX and The University of Utah Salt Lake City, UT.B6.Cg-Areg Mcub Rhbdf2 cub /J were obtained from The Jackson Laboratory (Strain #003628) and crossed with C57BL/6 mice to remove the Rhbdf2 cub /J mutation.B6.Areg Mcub (i.e.Areg −/− ) mice were bred heterozygously and littermates were used for experiments.The B6.Areg Mcub strain was backcrossed to NOD mice for 10 generations to produce the NOD.Areg −/− mice.NOD.Areg −/− mice were bred heterozygously and littermates were used for experiments.All experiments were performed in accordance with relevant guidelines and regulations and reported following recommendations in the ARRIVE guidelines.

Islet isolation
Single islet isolation was performed as previously described 44 .Briefly, pancreata were perfused via the bile duct with Collagenase 4 (Worthington Biochemical), incubated at 37° for 30 min, and islets were picked with a p10 pipette under a dissecting scope.Islets were dissociated into single cell suspension with enzyme free dissociation buffer (Gibco) for 15 min at 37° with vortexing every 5 min.Dissociated islets were washed with HBSS before proceeding to analysis.

Flow cytometry
Cell phenotype analysis Isolated cells were stained with antibodies for surface markers at room temperature for 15 min.Intracellular staining protocol was modified from eBioscience Foxp3 staining kit.Cells were fixed in 2% methanol free paraformaldehyde for 30 min at room temperature.Intracellular staining was done in Foxp3 permeabilization buffer (eBioscience) overnight at 4°.Cells were analyzed by flow cytometry on a BD Biosciences LSRFortessa.Data was analyzed using FlowJo 10 analysis software (BD Biosciences).www.nature.com/scientificreports/

In vivo cytokine quantification
In vivo cytokine production was evaluated as described previously 16 .Briefly, Brefeldin A (BFA) powder was dissolved in DMSO (20 mg/mL) and stored at − 20 °C.Immediately before use, concentrated BFA stock was diluted to a working concentration (0.5 mg/mL) in PBS and 500 uL was injected via tail vein.Six hours post injection, mice were euthanized, and islets and lymph nodes were analyzed by flow cytometry on Cytek Aurora (Cytek Biosciences).

In vitro T cell activation
Islets from 14-week-old female NOD mice were isolated and stimulated at 37° in RPMI supplemented with 10% FBS for 5 h with PMA/Ionomycin (1 ug/mL), Brefeldin A (5 ug/mL), and Monensin (2 uM).Intracellular amphiregulin was detected using biotinylated amphiregulin antibody (R&D Systems BAF989) followed by conjugated streptavidin for 30 min on ice.

Ex vivo islet ER stress
Pancreatic islets were isolated from NOD.scid mice as described above.Whole islets were then cultured in RPMI supplemented with 10% Fetal Bovine Serum (FBS) for 7 h.Additional stimulation conditions included recombinant mouse Areg (100 ng/mL) (Recombinant mouse Amphiregulin, Carrier-free, Biolegend), Thapsigargin (5 uM) (Enzo Life Sciences), or mAreg plus Thapsigargin.After 7 h, total RNA was extracted following the manufacturer's protocol from Qiagen RNeasy kit.First strand cDNA synthesis was performed using the manufacturer's protocol from High-Capacity Reverse Transcriptase kit (Applied Biosciences).The following targets were analyzed by qPCR (Insulin, Grp78, sXBP1, CHOP, ATF4, ATF6, GAPDH) using a QuantStudio6 (Applied Biosystems).Primers for each target are listed in Table 1.

Amphiregulin ELISA
Splenocytes from NOD.Areg +/+ , NOD.Areg +/− , and NOD.Areg −/− mice were enriched for CD4 + cells using the AutoMACS Pro Separator (Miltenyi Biotec) possel program and stimulated for 24 h in standard RPMI media supplemented with 10% FBS and with the addition of PMA and Ionomycin (1 ug/mL).Supernatant was collected and used in the mouse amphiregulin duoset ELISA (R&D Systems) following the manufacturer's protocol.Absorbance was measured on a microplate reader (BioTek Synergy H1) at 450 nm and 540 nm.Readings at 540 nm were subtracted from readings at 450 nm for wavelength correction, and concentrations were calculated using GraphPad Prism 9.0 Sigmoidal Curve based on recombinant amphiregulin standard curve.

Diabetes monitoring
Female NOD.Areg +/+ and NOD.Areg −/− mice were monitored weekly for overt diabetes using urinalysis (Diastix, Ascensia Diabetes Care).If urinalysis was positive, blood glucose readings using a glucometer and glucometer Table 1.Primer sequences of targets used in qPCR.www.nature.com/scientificreports/strips (Freestyle Lite) was performed.Mice were determined to be diabetic with one reading > 400 mg/dL or two consecutive readings of > 300 mg/dL.

Experimental autoimmune encephalomyelitis (EAE)
EAE was induced as previously described 26 .Briefly, 4 mg/mL Complete Freund's Adjuvant was used to make a peptide emulsion of MOG 35-55 at a 1:1 ratio the day prior to injecting mice.On day zero, 7-week-old female C57BL/6.Areg +/+ and Areg −/− mice were immunized with 100 ul MOG 35-55 peptide/CFA emulsion s.q.into each flank (total 200 ul), and injected with 200 ul of 1 ug/ml Bordetella pertussis toxin diluted in PBS i.p.On day two, each mouse was injected with an additional 200 ul of 1 ug/ml Bordetella pertussis toxin.Beginning on day seven post first injection, mice were scored for motor symptoms daily using the following scale.
Score 0: No obvious physical motor differences are observed.Score 1: Complete flaccidity of the tail or hind limb weakness (not both).Score 2: Both limp tail and hind limb weakness or partial paralysis.Score 3: Total hind limb paralysis.The mouse can no longer use hind limbs to maintain rump posture or walk.Score 4: Hind limb paralysis and front limb weakness/paralysis.With the total loss of movement in hind limbs, the mouse drags itself only on its forelimbs.Mice appear alert and feeding, but do not move around the cage.Score 5: Moribund.Mice are not feeding, not alert, and close to death.

Figure 3 .
Figure 3. Amphiregulin reduces beta cell ER stress.qPCR analysis of cultured islets treated with thapsigargin.Islets from four NOD.scid mice were pooled and either cultured in 10% FBS RPMI with or without 100 ng/mL recombinant mouse Areg and/or 5 mM Thapsigargin for 7 h.Expression relative to Gapdh.One-way ANOVA with Tukey multiple comparison.N = 3, Data shown as Mean with SEM.