Immunosuppression by NMDA-Receptor Antagonists is Mediated ...

MCB Accepts, published online ahead of print on 16 December 2013 Mol. Cell. Biol. doi:10.1128/MCB.01273-13 Copyright © 2013, American Society for Microbiology. All Rights Reserved.

1

Immunosuppression by NMDA-Receptor Antagonists is Mediated Through

2

Inhibition of Kv1.3 and KCa3.1 Channels in T cells

3 Sascha Kahlfußa,, Narasimhulu Simmaa,, Judith Mankiewicza,, Tanima Boseb, Theresa Lowinusa,

5

Stefan Klein-Hesslingc, Rolf Sprengeld, Burkhart Schravena,, Martin Heineb,# and Ursula

6

Bommhardta, #

7

Institute of Molecular and Clinical Immunology, Otto-von-Guericke-University Magdeburg,

8

Magdeburg, Germanya; Leibniz Institute for Neurobiology, Magdeburg, Germanyb; Institute of

9

Pathology, University of Würzburg, Würzburg, Germanyc; Max-Planck-Institute for Medical

10

Research, Heidelberg, Germanyd

11 12

Running Head: NMDAR antagonists inhibit T-cell function

13

#

14

[email protected]

15

S.K, N.S., J.M. and T.B. contributed equally to this work.

Address correspondence to Martin Heine, [email protected]; Ursula Bommhardt,

16 17

Introduction, Results, Discussion word count: 3571

18

Abstract word count: 197

19

Material and methods word count: 2152

20 1

Downloaded from http://mcb.asm.org/ on November 8, 2017 by guest

4

ABSTRACT

22

N-methyl-D-aspartate receptors (NMDARs) are ligand-gated ion channels that play an important

23

role in neuronal development, plasticity, and excitotoxicity. NMDAR antagonists are

24

neuroprotective in animal models of neuronal diseases, and the NMDAR open-channel blocker

25

memantine is used to treat Alzheimer’s disease. In view of the clinical application of these

26

pharmaceuticals and the reported expression of NMDARs in immune cells, we analyzed the

27

drug’s effects on T-cell function. NMDAR antagonists inhibited antigen-specific T-cell

28

proliferation and cytotoxicity of T cells and the migration of the cells towards chemokines. These

29

activities correlated with a reduction in TCR-induced Ca2+-mobilization and nuclear localization

30

of NFATc1, and they attenuated the activation of Erk1/2 and Akt. In the presence of antagonists,

31

Th1 effector cells produced less IL-2 and IFN-γ, whereas Th2 cells produced more IL-10 and IL-

32

13. However, in NMDAR knock-out mice the presumptive expression of functional NMDARs in

33

wild-type T cells was inconclusive. Instead, inhibition of NMDAR antagonists on the

34

conductivity of Kv1.3 and KCa3.1 potassium channels was found. Hence, NMDAR antagonists

35

are potent immunosuppressants with therapeutic potential in the treatment of immune diseases,

36

but their effects on T cells have to be considered in that Kv1.3 and KCa3.1 channels are their

37

major effectors.

38 39 40 41 42 43 44 2


21

45

INTRODUCTION

46

N-methyl-D-aspartate

47

isoxazolepropionic acid receptors (AMPARs) are the main ionotropic glutamate receptors

48

involved in glutamatergic neurotransmission in the central nervous system. Their functions in

49

synaptic transmission and plasticity, long term potentiation/depression, and excitotoxicity are

50

well established (1). Heterotetrameric NMDARs consist of the obligatory GluN1 subunit and two

51

homodimeric or heterodimeric subunits formed by GluN2A-D, GluN3 or GluN4 (2). Activation

52

of NMDARs requires the binding of glutamate or aspartate, the co-agonists glycine or D-serine,

53

and membrane depolarization. The NMDAR opening kinetic depends on the subunit composition

54

and has profound consequences for downstream signaling pathways. NMDARs can sense

55

different activation patterns and trigger specific intracellular signaling cascades through the

56

induction of intracellular Ca2+-changes at small domains below the neuronal plasma membrane.

57

Activation of protein kinase C members, the mitogen activated protein kinase (MAPK) Erk1/2

58

and phosphatidylinositol 3-kinase (PI3-K)-Akt pathways culminate in the induction of

59

transcription factors that orchestrate specific gene-expression programs guiding neuronal

60

homeostasis, death or plasticity (3). The location and composition of NMDARs in the neuronal

61

membrane are fundamental for the initiation of those intracellular signaling events (4). NMDAR

62

activity is effectively blocked by ifenprodil, a non-competitive antagonist that binds to the

63

GluN2B subunits of NMDARs, and by the non-competitive open-channel blockers MK801 and

64

memantine (5). These pharmaceuticals have been neuroprotective in animal models of stroke,

65

epilepsy, and experimental autoimmune encephalomyelitis, and memantine is used to treat

66

Alzheimer’s disease (6). NMDARs themselves can be targets of immune attack as in anti-NMDA

67

receptor encephalitis, which is caused by autoantibodies directed against the GluN1 subunit of

68

NMDARs (7).

receptors

(NMDARs)

and

α-amino-3-hydroxy-5-methyl-4-


3

In recent years, evidence has emerged that immune cells, including dendritic cells (DCs),

70

release glutamate and can be regulated by glutamate present in the blood stream, peripheral

71

organs, and central nervous system (8, 9). NMDARs, AMPARs (GluA3-subunit), and

72

metabotropic glutamate receptors (group 1 mGluRs) were found expressed in human peripheral

73

blood lymphocytes and Jurkat T cells, and modulated their function (10-14). For murine

74

CD4+CD8+ double-positive thymocytes in contact with antigen-presenting DCs, inhibition of

75

NMDARs regulated TCR-induced Ca2+-flux and, thereby, the apoptosis of double-positive cells

76

(9).

77

For a beneficial therapeutic application of NMDAR antagonists it is important to

78

understand how they influence T-cell function and, thereby, the adaptive immune response. Here,

79

we show profound inhibition on CD4+ and CD8+ T cell effector function by NMDAR

80

antagonists. The inhibition correlated with reduced activation of major TCR-induced signaling

81

pathways, including Ca2+-mobilization and activation of Erk1/2, Akt, and NFATc1. Consistent

82

with published results (9), we detected mRNA expression and positive immunoreactivity for

83

NMDAR subunits in thymocytes and peripheral T cells. However, GluN1 protein expression was

84

not evident in wild-type (wt) thymocytes compared with control thymocytes from GluN1 knock-

85

out (ko) mice. Assuming the strong influence of NMDAR antagonist pharmacology on Ca2+-

86

mediated signaling involves off-target effects, we demonstrate that NMDAR antagonists inhibit

87

the activity of Kv1.3 and KCa3.1 potassium channels. Hence, NMDAR antagonists, which are

88

potent immune modulators/suppressors, seem to act via their inhibitory effects on Kv1.3 and

89

KCa3.1 channels.

90 91

MATERIALS AND METHODS

4


69

Mice. C57/BL6 wild-type (wt), BALB/c, OT2 TCR transgenic (tg) (15), OT1 TCR tg (16) and

93

NFATc1-EGFP mice (17) on C57/BL6 background , at the age of 6-10 weeks were used. GluN1

94

ko mice, generated by crossing GluN1flx/flx with Cre deleter mice (18, 19), both on C57/BL6

95

background, and littermate mice were used within hours after birth. All animal experimentation

96

was conducted in compliance with institutional guidelines.

97 98

Antibodies, flow cytometry and Th cell differentiation. Antibodies (Abs) for cell isolation,

99

cell stimulation, and flow cytometry were obtained from BD Bioscience (San Jose, CA, USA):

100

CD4-FITC (GK1.5), CD8-APC/PE (53-6.7), CD25-PE (7D4), TCR-FITC (H57-597), CD69-PE

101

(H1.2F3), CD44-FITC (IM7), CD3 (145.2C11), CD28 (37.51); BioLegend (London, UK): CD4-

102

APC (GK1.5); eBioscience (San Diego, CA, USA): CD3 (145.2C11), CD28 (37.51), CD127-PE

103

(A7R34), IL-2-PE (JES6-5H4), IFNγ-PE (XMG1.2), IL-4-PE (11B11), IL-10-PE (JES5-16E3),

104

IL-13-PE (eBio13A), B220-FITC/PE (RA3-6B2), IgG2a-PE/FITC (eBR2a); Dianova (Hamburg,

105

Germany): PE-conjugated F(ab’)2 fragment donkey anti-rabbit IgG (H+L), IgG2b (eBMG2b),

106

IgG2a (pk); Alomone Laboratories (Jerusalem, Israel): rabbit anti-mouse GluN1 (AGC-001),

107

GluN2A (AGC-002) and GluN2B (AGC-003); Synaptic Systems (Göttingen, Germany): mouse

108

anti-mouse GluN1 (M68); Cell Signaling (Beverly, MA, USA): pErk1/2 (Thr202/Tyr204), pS6

109

(S240/244), pAkt-Alexa488 (S473) and IgG2a-Alexa488. Intracellular staining of NMDAR subunits,

110

cytokines or other signaling proteins was performed with the FoxP3 staining kit from

111

eBioscience after surface labeling with CD4, CD8 or B220 Abs. Primary Abs were detected with

112

PE-conjugated F(ab’)2 fragment donkey anti-rabbit IgG (H+L). Specificity of GluN1, GluN2A

113

and GluN2B Abs was controlled with subunit-specific peptides (Alomone Labs), which were

114

incubated with Abs for 30 min at room temperature in FoxP3 staining buffer. Ab-peptide

115

conjugates were added to the cells, followed by staining with PE-conjugated F(ab’)2 fragment 5


92

donkey anti-rabbit IgG (H+L) (data now shown). For measurement of intracellular cytokine

117

production, OT2 TCR tg (OT2) CD4+ T cells were stimulated with OVA-peptide-loaded (amino

118

acids 323-339, 10 µg/ml, a gift of M. Gunzer) bone marrow-derived DCs (pOVA-DCs) in cell

119

ratios as indicated. For evaluation of Th1 polarization, anti-mouse IL-4 (2 µg/ml, 11B11) and for

120

skewing towards Th2 cells anti-mouse IL-12 (3 µg/ml, C18.2), anti-mouse IFNγ (5 µg/ml, AN-

121

18, all from Biolegend) and rIL-4 (20 U/ml, eBioscience) were added to cell cultures on day 0.

122

On day 5, cells were re-stimulated with PMA and ionomycin (P/IO, 100 ng/ml each, Calbiochem,

123

Darmstadt, Germany) for 4 h in the presence of Brefeldin A (2 µg/ml, Calbiochem) and analyzed

124

for the expression of intracellular cytokines with flow cytometry using FACSFortessa and

125

CellQuest Pro software (BD Bioscience).

126 127

Generation of bone marrow derived DCs (BMDCs). Bone marrow (BM) cells were suspended

128

in RPMI 1640 medium (Biochrom, Berlin, Germany) reconstituted with 1% non-essential amino

129

acids (GIBCO; Invitrogen, Carlsbad, CA, USA ), 5% FCS (PAN Biotech, Aidenbach, Germany),

130

1% L-glutamine (GIBCO), 0.1% gentamycine (Carl Roth, Karlsruhe, Germany), 0.1% 2-

131

mercaptoethanol (GIBCO), IL-4 (48 ng/ml) and GM-CSF (10 ng/ml) from hybridoma

132

supernatant (a gift of M. Gunzer). 3x106 BM cells/5 ml BMDC medium were cultured for 7 days.

133

At day 3, 2 ml medium were replaced by fresh BMDC medium, at day 6 total medium was

134

replaced and cells were stimulated with LPS (20 ng/ml; Sigma-Aldrich, Steinheim, Germany) for

135

24 h. DCs, used at day 9 or 10, were re-stimulated with LPS 24 h before the start of experiments.

136

Maturation of BMDCs was verified at day 7 by staining cells with Abs against MHC-II, CD11c,

137

CD80 and CD86 (BD Bioscience).

138

6


116

Cell isolation and proliferation assay. CD4+ or CD8+ T cells were isolated from pooled lymph

140

nodes by negative selection using a cocktail of biotinylated Abs: NK1.1 (PK136), CD8α (53-6.7),

141

CD4 (GK1.5), I-A/I-E (2G9), CD45R/B220 (RA3-6B2), and Ter-119 (all from BD Bioscience)

142

and streptavidin magnetic beads (Miltenyi Biotec, Bergisch Gladbach, Germany). Purity of CD4+

143

or CD8+ T cells was routinely above 90%. Mature BMDCs (MHC-II+, CD11c+, CD80+, CD86+)

144

were pulsed with OVA-peptide (aa 323-339, 10 µg/ml or aa 257-264, SIINFEKL, 5 µg/ml,

145

AnaSpec, Fremont, CA, USA) for 2 h and cultured with OT2 CD4+ or OT1 TCR tg (OT1) CD8+

146

T cells, respectively. CD4+ T cells (0.5-1x105) were stimulated with plate-bound CD3 Abs (3 or

147

10 μg/ml) or CD3+CD28 Abs (3 and 5 µg/ml). For mixed lymphocyte reactions, 1x105 CD4+ T

148

cells from C57/BL6 mice were co-cultured with irradiated (3 Gy) splenocytes from BALB/c mice

149

at 1:3 to 1:5 ratio for 5 days. Cells were cultured in the presence or absence of ifenprodil, MK801

150

or memantine (diluted in ddH20 or PBS, Tocris, Bristol, Great Britain) in concentrations as

151

indicated in complete RPMI 1640 medium/10% FCS. Proliferation was measured with 3[H]-

152

Thymidine incorporation (0.2 µCi/well, MP Biomedicals, Heidelberg, Germany) for 8-16 h at 24

153

or 48 h, or after 5 days. For analysis of cell-cycle progression, OT1 or OT2 T cells were labeled

154

with CFSE (5 µM, Invitrogen) and cultured with pOVA-loaded DCs for the indicated days, then

155

analyzed with flow cytometry.

156 157

Measurement of apoptosis. Apoptosis was determined with the Apoptosis detection kit 1 from

158

BD Pharmingen. 1x106 OT2 CD4+ T cells were left untreated or were activated with pOVA (aa

159

323-339)-loaded BMDCs in a DC-T cell ratio of 1:10 in the presence or absence of NMDAR

160

inhibitors for the indicated time points. After harvest, cells were stained with Annexin V-FITC

161

and propidiumiodide (PI) according to manufacturer’s protocol and analyzed with flow

162

cytometry. The percentage of vital cells was determined by gating on AnnexinV-PI- cells. 7


139

163 CTL assay. Lymph node cells from OT1 TCR tg mice were stimulated with rIL-2 (20 ng/ml,

165

Biolegend) and pOVA (1 µg/ml SIINFEKL) in complete RPMI 1640/10% FCS medium for 72 h.

166

Cells were then expanded in the presence of rIL-2 (20 ng/ml) for 1-2 days and killing assays with

167

pOVA-loaded (5 µg/ml SIINFEKL) RMA-S target cells were performed in triplicates in 96-well

168

round-bottom plates. When indicated, CTLs were pre-treated with ifenprodil (10-30 µM) for 20

169

min at 37°C. CTL assays were performed for 4 h at 37°C, cells were stained with CD8-APC and

170

Annexin V-FITC, and the percentage of apoptotic AnnexinV+ RMA-S cells was determined with

171

flow cytometry. The relative killing efficiency was calculated by relating the percentage of

172

apoptotic target cells to the ratio of CTL to target cells.

173 174

RNA isolation and PCR. RNA from brain, thymocytes and peripheral T cells was isolated with

175

TRIzol reagent (Life Technologies, Darmstadt, Germany) and reverse transcribed with a First-

176

Strand cDNA Synthesis Kit (Thermo Scientific, Karlsruhe, Germany). Oligonucleotides for RT-

177

PCR and PCR, obtained from Apara Bioscience GmbH, Denzlingen, Germany, were: Fwd-β-

178

actin: 5’-CCAGGTCATCACTATTGGCAAGGA-3’; Rev-β-actin: 5’-GAGCAGTAATCTCCT

179

TCTGCATCC; Fwd-GAPDH: 5’-CAAGGTCATCCATGACAACTTTG; Rev-GAPDH: 3'

180

GTCCACCACCCTGTTGCTGTAG; Fwd-GluN1C1: 5`-TGTGTCCCTGTCCATACTCAAG-3`,

181

Rev-GluN1C1: 5`-GTCGGGCTCTGCTCTACCACTC-3’; Fwd GluN2A: 5´-GGAGAAGG

182

GTACTCCAGCGCTGAA-3´; Rev GluN2A: 5`-AGTCTGTGAGGAGATAAAAT CCAGC-3’;

183

Fwd-GluN2B: 5´-GCAAGCTTCTGTCATGCTCAACATC-3´, Rev-GluN2B: 5`-GCTCTGCA

184

GCTTCTTCAGCTGATTC-3´ (20). Wt and floxed GluN1 alleles and GluN1 excision were

185

analyzed with PCR using DNA isolated from tail, thymocytes or brain; primers were:

8


164

186

GluN1flox/wt: Fwd-5’-CTGGGACTCAGCTGTGCTGG-3’; Rev-5’-AGGGGAGGCAACACTGT

187

GGAC-3’; GluN1 excision: Fwd-5’-GAGAAAGACATGGGGCATTATCC-3’.

188 Western blot analysis

190

Isolated CD4+ T cells (5x106) were stimulated with plate-bound CD3 Abs (10 µg/ml for short-

191

term and 3 µg/ml for long-term stimulation) or CD3+CD28 Abs (3 and 5 µg/ml) in the presence

192

or absence of ifenprodil (30-50 µM) for the indicated time points. Cells were lysed to obtain total

193

or cytoplasmic and nuclear protein extracts as described (21). Protein lysate obtained from T cells

194

(5-15 µg), thymocytes (20 µg), or brain (5 µg) was subjected to 8-10% SDS-PAGE. Separated

195

proteins were transferred onto nitrocellulose membrane and blocked with 5% non-fat milk

196

powder in TBST. The expression/activation of signaling proteins was analyzed with primary Abs

197

specific for pSrc (Y416), pPLCγ1 (Y783), pErk1/2 (Thr202/Tyr204), pAkt (Ser473, DE9), pS6

198

(S240/244), pmTOR (S2448), pGSK3β (S9) (all from Cell Signaling Technology, Frankfurt,

199

Germany), NFATc1 (7A6, Alexis Biochemicals, Lörrach, Germany), GluN1 (Synaptic Systems),

200

β-actin (AC 40, Sigma-Aldrich) and Lamin B (Santa Cruz, Biotechnology, Santa Cruz, CA,

201

USA) followed by HRP-coupled mouse anti-rabbit, goat anti-mouse or donkey anti-goat

202

secondary Abs (Jackson ImmunoResearch Laboratories, Dianova) and detection with the ECL

203

system (Thermo Scientific Pierce, Rockford, IL, USA). The immune reactive bands were scanned

204

and quantified with Kodak software.

205 206

Measurement of Ca2+-flux. Lymph node cells from wt mice were stained with 4 µM Indo-1 AM

207

(Invitrogen, Molecular Probes) for 45 min at 37 °C. After being washed, the cells were stained

208

for CD8 and B220 or CD4 and B220 surface expression for 15 min, washed and re-suspended in

209

Hank`s buffer (Biochrom) supplemented with 1 mM CaCl2. CD3-biotin Abs (145.2C11, 10 9


189

µg/ml) plus streptavidin (25 µg/ml, Dianova) were added to induce Ca2+-flux. The NMDAR

211

antagonist ifenprodil (10 or 30 µM) was added for 5 min followed by CD3 Ab and streptavidin

212

treatment. Towards the end of each measurement, ionomycin (2 μM, Calbiochem) was added as a

213

positive control for cell reactivity. Ca2+-flux was measured on a LSRII flow cytometer (BD

214

Biosciences), data files were transferred to FlowJo V3.6.1 (Tree Star, Ashland, OR, USA), mean

215

Ca2+-flux was determined for unlabelled CD4+ or CD8+ T cells and data were further processed

216

with IgorPro5.04B software (WaveMetrics Inc., Portland, OR, USA). For each graph, ΔCa2+-flux

217

was defined as the difference between the maximum and minimum value of Ca2+-intensity.

218 219

Migration assay. Splenocytes (4x106), untreated or pre-incubated with ifenprodil (30 and 50

220

µM) for 30 min in D-MEM medium (Biochrom) supplemented with 0.1% bovine serum albumin

221

(BSA) and 10 mM Hepes pH 7.4, were transferred to transwell chambers (6.5 mm diameter, 3.0

222

µm pore size, Corning Costar, Tewksburry, MA, USA) coated with fibronectin (6.5 µg/ml, Roche

223

Diagnostics, Basle, Switzerland). Cells were allowed to migrate towards SDF1α (100 ng/ml) or

224

CCL21 (300 ng/ml, both from PeproTech, Hamburg, Germany) for 150 min at 37°C with

225

migration in the absence of chemokine serving as control. Migration was stopped by the addition

226

of 0.1 M EDTA. Migrated cells were stained with CD4 and CD8 Abs and measured for 30 s at a

227

FACSFortessa. Relative migration was calculated by dividing the number of cells migrated in the

228

presence of chemokine (set as 1.0) by the number of cells migrated in the absence of chemokine.

229 230

Confocal microscopy. CD4+ T cells from OT2 TCR tg mice were incubated for 10 min with

231

LPS-matured and pOVA-loaded (10 µg/ml, aa 323-339) BMDCs in RPMI1640/10% FCS

232

medium at a DC-T cell ratio of 1:3. Obtained DC-T cell pairs or freshly isolated thymocytes were

233

transferred onto poly-L-lysine coated slides (Marienfeld, Königshofen, Germany). Cells were 10


210

fixed with 2% paraformaldehyde for 15 min at 37°C, permeabilized with 0.1% Triton X100 for

235

10 min, blocked with 1% BSA (Carl Roth GmbH) for 30 min at room temperature, and incubated

236

with Abs against GluN1 (Synaptic Systems) and GluN2B (Alomone Labs) subunits or with

237

isotype control Abs, overnight at 4°C or for 2 h at room temperature. Thereafter, cells were

238

stained with goat anti-rabbit F(ab’)2-Alexa488 (Invitrogen) or donkey anti-mouse-Alexa488 Ab

239

(Jackson ImmunoResearch Laboratories, West Grove, PA, USA) for 1 h at room temperature in

240

the presence of phalloidin-Cy5 (Dyomics, Jena, Germany). Confocal settings were the same for

241

NMDAR-specific, isotype and control stainings with NMDAR subunit-specific peptides. Images

242

show the maximum intensity projection of 3 planes (5 planes in supplementary FIG 3) from a z-

243

stack. Image acquisition was performed with a Leica TCS SP5 or STED confocal microscope

244

(Leica, Houston, TX, USA) and data were processed with ImageJ software.

245 246

Electrophysiology. All experiments were carried out in the whole-cell configuration of the

247

patch-clamp technique with an EPC10 amplifier and PatchMaster v. 2.11 (HEKA Electronic,

248

Lambrecht, Germany) at room temperature (20-24°C) and CD4+ T cells activated with

249

CD3+CD28 Abs (3+5 μg/ml) for 48 h (22) or EL-4 lymphoma cells. Patch pipettes from

250

borosilicate glass used for recordings had a resistance between 3-5 MΩ. Kv1.3 and KCa3.1

251

currents were recorded with an external solution of the following compositions (in mM): 160

252

NaCl, 4.5 KCl, 5 HEPES, 1 MgCl2, 2 CaCl2 or 160 Na-Aspartate, 4.5 KCl, 2 CaCl2, 1 MgCl2, 10

253

HEPES, respectively. The pipette solution was (in mM): 162 KF, 11 EGTA, 10 HEPES, 1 CaCl2,

254

2 MgCl2 (23) or 145 K-aspartate, 8.5 CaCl2, 2 MgCl2, 10 EGTA, 10 HEPES (24, 25),

255

respectively, adjusted to pH7.4 for external solution and to pH7.2 for internal solution;

256

osmolarity was set to 300-340 mOsM for both solutions. Kv1.3 currents were measured with

257

depolarizing voltage steps up to +60 mV from a holding potential of -80 mV every 30 s. KCa3.1 11


234

currents were elicited by a 200 ms voltage ramp from -120 to +40 mV from a holding potential of

259

-80 mV every 15 s. Sampling rate was 50 kHz for Kv1.3 and 20 kHz for KCa3.1. Antagonists

260

(ifenprodil, memantine, MK801, ketamine and APV; Tocris) at a constant inhibitor concentration

261

were added during the recording. Analysis of transient currents was performed in HEKA

262

FitMaster v2x53. Amplitude values were transferred to GraphPad Prism 5.0 for analyzing the

263

dose-response curve and Hill slope.

264 265

Statistical analysis. Statistical differences in cell number, percentage, mean fluorescence

266

intensity and proliferation was analysed by use of Student’s t test or, in the case of more than two

267

groups, by one-way ANOVA followed by a post-test (Dunnett's test) to analyse individual

268

differences. The p value range of significance is indicated as *p