Thomson et al 1

Modification of the cyclopropyl moiety of abacavir provides insight into the structure activity

relationship between HLA-B*57:01 binding and T-cell activation.

Short title: HLA-B*57:01-restricted T-cell responses to abacavir

Authors: Paul J Thomson,1 Patricia T Illing,2 John Farrell,1 Mohammad Alhaidari,1 Catherine C Bell,1

Neil Berry,1 Paul M O’Neill,1 Anthony W Purcell2, Kevin B Park,1 Dean J Naisbitt1

1MRC Centre for Drug Safety Science, Dept Molecular & Clinical Pharmacology, University of

Liverpool, Liverpool, UK; 2Infection and Immunity Program, Monash Biomedicine Discovery Institute

and Dept of Biochemistry and Molecular Biology, Monash University, Clayton 3800, Victoria,

Australia

*Correspondence Professor Dean J. Naisbitt (The University of Liverpool, Liverpool, England

[Telephone, 0044 151 7945346; e-mail, dnes@liv.ac.uk]).

Acknowledgements: The authors would like to thank the volunteers for agreeing to donate blood.

This work was funded by grants from the MRC (grant number MR/R009635/1). Core support was

received from the Medical Research Council Centre for Drug Safety Science (Grant MR/l006758).

AWP is supported by a Principal Research Fellowship from the Australian National Health and

Medical Research Council (NHMRC) and NHMRC Project Grant (APP1122099).

Conflict of Interest Statement: The authors declare no conflicts of interest.

Word count: 3497

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

Thomson et al 2

Graphical abstract

Alteration of the HLA-B*57:01

peptide binding repertoire

Synthesis of structural variants of abacavir

AbacavirNo Clashes Clashes

Abacavir analoguesNo Clashes

CD8+ T-cell activation

No CD8+ T-cell activation

No alteration of the HLA-B*57:01

peptide binding repertoire

23

24

25

26

Thomson et al 3

Abstract

Background: Abacavir is associated with hypersensitivity reactions in individuals positive for the

HLA-B*57:01 allele. The drug binds within the peptide-binding groove of HLA-B*57:01 altering

peptides displayed on the cell surface. Presentation of these HLA-abacavir-peptide complexes to T-

cells is hypothesised to trigger a CD8+ T-cell response underpinning the hypersensitivity. Thus, the

aim of this study was to explore the relationship between the structure of abacavir with HLA-

B*57:01 binding and CD8+ T-cell activation.

Methods: Seventeen abacavir analogues were synthesised and cytokine secretion from

abacavir/abacavir analogue-responsive CD8+ T-cell clones was measured using IFN-γ ELIspot. In silico

docking studies were undertaken to assess the predicted binding poses of the abacavir analogues

within the HLA-B*57:01 peptide binding groove. In parallel, the effect of selected abacavir analogues

on the repertoire of self-peptides presented by cellular HLA-B*57:01 was characterised using mass

spectrometry.

Results: Abacavir and ten analogues stimulated CD8+ T-cell IFN-γ release. Molecular docking of

analogues that retained antiviral activity demonstrated a relationship between predicted HLA-

B*57:01 binding orientations and the ability to induce a T-cell response. Analogues that stimulated

T-cells displayed a perturbation of the natural peptides displayed by HLA-B*57:01. The antigen-

specific CD8+ T-cell response was dependent on the enantiomeric form of abacavir at both

cyclopropyl and cyclopentyl regions.

Conclusion: Alteration of the chemical constitution of abacavir generates analogues that retain a

degree of pharmacological activity, but have variable ability to activate T-cells. Modelling and

immunopeptidome analysis delineate how drug HLA-B*57:01 binding and peptide display by antigen

presenting cells relates to the activation of CD8+ T-cells.

Keywords: Drug hypersensitivity, HLA, human, mass spectrometry, T-cells.

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

Thomson et al 4

Key messages: (1) Structural modelling aids prediction of the impact of abacavir modifications on

the activation of CD8+ T cells, but cannot be used in isolation; (2) analogues that stimulated CD8+ T-

cells displayed a perturbation of natural peptides displayed by HLA-B*57:01; (3) enantiomeric forms

of abacavir predicted to adopt different HLA-B*57:01 binding conformations display divergent CD8+

response profiles.

52

53

54

55

56

57

Thomson et al 5

Introduction.

Abacavir ((1S,4R)-4-(2-amino-6-(cyclopropylamino)-9H-purin-9-yl)-2-cyclopentene-1-methanol) is a

nucleoside reverse transcriptase inhibitor, used in the treatment of HIV. Despite its potent antiviral

activity, abacavir is associated with severe hypersensitivity reactions that occur in 4-8% of patients

receiving therapy (1). A relationship has been demonstrated between hypersensitivity and carriage

of the MHC class I allele HLA-B*57:01 (2–4). The presence of this “risk allele” has a 100% negative

predictive value in patients while yielding a positive predictive value of 48-55% (5,6). Prospective

genotyping for HLA-B*57:01 and exclusion of patients positive for the allele from abacavir therapy is

a widely applied, cost-effective means of eradicating hypersensitivity reactions (5,7,8).

Proteasomal breakdown of intracellular proteins generates the immunopeptidome that binds to

HLA class I molecules prior to display at the cell surface. Under normal circumstances, the

constitutive self-peptides displayed by HLA-B*57:01 are ignored by CD8+ T-cells; however,

introduction of novel peptides (e.g. virus-derived peptides) stimulates CD8+ T-cell proliferative

responses and the secretion of cytokines and cytolytic molecules that target cells/tissues bearing the

immunogenic peptides. Abacavir binds to HLA-B*57:01 in a non-covalent manner, occupying the C-F

pocket region of the peptide binding groove which usually accommodates the C-terminal end of the

bound peptide ligand, and leads to a dramatic alteration of the HLA-B*57:01 immunopeptidome (9–

11). The prevalence of peptides bearing aromatic amino acids such as tryptophan and phenylalanine

at the PΩ position (the C-terminus of the peptide) is diminished in the presence of abacavir, whilst

those terminating in smaller aliphatic amino acids such as valine, leucine and isoleucine increases

(9,10). As these newly presented HLA-drug-self-peptide complexes are not observed in the absence

of abacavir it has been proposed that they activate the CD8+ T-cells involved in abacavir

hypersensitivity (9,10).

58

59

60

61

62

63

64

65

66

67

68

69

70

71

72

73

74

75

76

77

78

79

80

81

82

Thomson et al 6

Abacavir-responsive CD8+ T-cell clones can be generated from the peripheral blood of healthy, drug

naïve, HLA-B*57:01+ donors (12–16). Chessman et al. identified that antigen processing was a

prerequisite for presentation of abacavir-associated antigens and the activation of CD8 + T-cells, while

Bell et al. and Adam et al. additionally demonstrated activation via a direct interaction of the drug

with cell surface HLA-B*57:01 (12,14,15). More recent studies suggest that abacavir preferentially

activates memory T-cells through cross-reactivity with peptides of viral origin (6,16). It has also

recently been shown that HLA-B*57:01 transgenic mice can be used as a model for abacavir

reactivity but require CD4+ T-cell depletion prior to the administration of the drug. This leads to

induction of skin-homing abacavir-specific CD8+ T-cells that induce skin inflammation (17).

The cyclopentyl and cyclopurine groups of abacavir sit in the D and E pockets of HLA-B*57:01,

respectively. However, the cyclopropyl group extends into the F-pocket altering its shape and

chemistry, accounting for the preferred accommodation of smaller amino acid moieties (10). The

abacavir metabolite carbovir, which lacks the cycloproyl moiety, does not activate CD8+ T-cell

responses (14) and abolition of abacavir-specific T-cell responses can be achieved via modifications

to the 6-amino cyclopropyl group (13). Fifteen abacavir analogues were previously synthesised, of

which two (isopropyl amino and isopropyl methyl amino) did not activate abacavir-specific CD8+ T-

cell clones. Molecular docking experiments predicted that both compounds would bind to HLA-

B*57:01 in a manner yielding unfavourable steric clashes between their functional groups and an

abacavir stabilised HLA-B*57:01 binding peptide (HSITYLLPV). This suggested an inability of the

compounds to facilitate presentation of this peptide in the same conformation (13). Expanding on

this concept, we have synthesized seventeen analogues with modifications to the 6-amino

cyclopropyl moiety as chemical probes to explore the molecular interaction between abacavir and

HLA-B*57:01, the display of abacavir-dependent HLA-B*57:01 binding peptides on the surface of

antigen presenting cells, and CD8+ T-cell activation. Our studies demonstrate that abacavir analogues

83

84

85

86

87

88

89

90

91

92

93

94

95

96

97

98

99

100

101

102

103

104

105

106

107

Thomson et al 7

have a distinct impact on the repertoire of peptides presented to T-cells on the surface of antigen

presenting cells, while cloning in the presence of analogue-stimulated antigen presenting cells had

the potential to generate T-cell clones with specificity outside the range of abacavir responsive T-cell

clones.

108

109

110

111

112

Thomson et al 8

Materials and methods.

A description of abacavir analogue synthesis and the assessment of antiviral activity is available as

supplementary material. Four HLA-B*57:01+ donors were selected to generate abacavir and

analogue responsive clones. Due to the journals restricted word count, text describing cell culture

methods and T-cell phenotypic and functional assessments are provided as supplementary material.

In silico modelling and mass spectrometric methods described in detail in the supplementary

material were used assess predicted abacavir (analogue) HLA-B*57:01 binding and the repertoire of

peptides displayed by HLA-B*57:01 in the presence of abacavir analogues, respectively. For T-cell

assays, the Mann-Whitney test was used to compare control and test values.

113

114

115

116

117

118

119

120

121

122

123

Thomson et al 9

Results

Abacavir-responsive T-cell clones display cross-reactivity to a subset of analogues.

From a total of 2279 clones tested across four donors, 101 proliferated in the presence of abacavir at

the first testing stage (Control, 1665 ± 188.5 cpm: abacavir (35µM), 6940 ± 729 cpm; P< 0.0001,

Figure 1a). All abacavir-responsive clones were phenotyped as CD8+. An abacavir-specific increase in

IFN-γ secretion was detected with all clones, when compared with controls (Figure 1b: 4

representative clones).

Six abacavir-responsive CD8+ T-cell clones from the four donors were incubated with abacavir and all

seventeen analogues, and IFN- γ secretion was measured. Figure 1c compares the individual

analogue response of the six clones with the mean level of abacavir-specific IFN-γ secretion using the

same six clones. Analogues A, C, D, E, H, J, K, O and P all stimulated IFN-γ secretion in abacavir-

responsive clones. High levels of IFN- γ were released from abacavir-responsive clones in the

presence of analogues A, C, D, J and P, especially analogue J (methoxy azetidine) which activated T-

cells to a higher degree than abacavir. Analogue D (azetidine) the original analogue upon which

others were constructed, demonstrated a dose-dependent increase in IFN-γ secretion in all abacavir

clones. Interestingly, analogue H (azetidine fluoro) activated T-cells at the higher concentration of

50µM, with no IFN-γ secretion observed at lower concentrations. In contrast, analogues B, F, G, I, L,

M, N, Q and analogue 15 (from our previous study (13)) were completely deficient in T-cell reactivity

with abacavir-responsive clones.

Antiviral activity.

It was possible to separate the antiviral and T-cell stimulatory properties of abacavir through the

synthesis of chemical analogues (Supplementary Table 4). Analogues D and H displayed good

124

125

126

127

128

129

130

131

132

133

134

135

136

137

138

139

140

141

142

143

144

145

146

147

Thomson et al 10

antiviral activity at concentrations similar to abacavir. Analogues A, B, C, G, K, M and O retained

antiviral activity (i.e., inhibited viral replication at 10-60 µM), while analogue J inhibited viral

replication at 86 µM. All remaining analogues were inactive.

Abacavir analogues are predicted to bind HLA-B*57:01 in a manner distinct from abacavir.

A molecular docking protocol was used to assess how abacavir and its analogues interact within the

peptide binding groove of HLA-B*57:01. The software replicated the crystal structure of the

predicted binding pose of abacavir within the F-pocket of HLA-B*57:01 with peptide HSITYLLPV (13).

The binding orientations of the abacavir analogues D, G, H, M, O, P and Q were predicted, with

abacavir’s predicted binding conformation superimposed as a comparator. The cyclopurine group of

abacavir forms hydrogen bonds with amino acid residues Asp 114, Ser 116, Asn 77 and Ile 124

(Figure 2b); however, a diverse range of binding interactions were observed with the abacavir

analogues.

First, analogues D, O and P, which activated abacavir-responsive clones (Figure 2a), were predicted

to bind HLA-B*57:01 in a similar manner to abacavir with their functional groups protruding into the

F pocket (Figure 2c i, iii and iv), and no clashes between the drug and amino acids of the antigen

binding cleft or pep HSITYLLPV were observed. While no clashes were also apparent for T-cell

stimulatory analogue H (Figure 2c ii), the guanosine portion of the molecule was inverted when

bound to HLA-B*57:01 indicating a predicted binding pose distinct to abacavir still yielding an ability

to induce T-cell activation.

The predicted binding poses of analogues devoid of abacavir T-cell reactivity (G, M and Q) indicated

the presence of unfavourable steric clashes between the functional groups of the respective

148

149

150

151

152

153

154

155

156

157

158

159

160

161

162

163

164

165

166

167

168

169

170

171

Thomson et al 11

analogues and amino acid residues present either within HLA-B*57:01 or residues of the predicted

HLA binding peptide HSITYLLPV (Figure 2d). This suggests that the analogues would either not bind

to HLA-B*57:01 or binding would occur in a manner which would require further conformational

changes within the peptide binding groove. Specifically, analogues M and Q induced steric clashes

with the amino acid residues of the binding peptide HSITYLLPV (Figure 2d i and iii) and this was not

observed with analogues that activated abacavir-specific T-cells. It is possible that these analogues

bind HLA-B*57:01 altering the conformation of co-presented peptides or favouring the binding of

different peptides when compared with abacavir, resulting in a loss of abacavir T-cell cross-reactivity.

Activation of abacavir-specific CD8+ T-cell clones is dependent upon drug chirality.

Analogues P (S-sec-butyl amino) and Q (R-sec-butyl amino) comprise the same structure, differing only

in the chirality of their functional group, yet exhibit different effects on T-cell activation (Figure 2a).

Analogue P stimulated IFN-γ release from the abacavir-responsive clones, whereas analogue Q did

not activate the clones (Figure 2a). This indicates that the enantiomeric state of a functional group is

a governing factor at the HLA peptide binding site. Indeed, when docked into HLA-B*57:01, the

predicted binding poses of analogues differ greatly. Analogue P is predicted to bind in a manner

conformationally similar to abacavir likely favouring the presentation of an array of altered self-

peptides similar to abacavir (Figure 2c iv), whilst analogue Q is predicted to bind in a more distinct

fashion (Figure 2d iii).

Given the enantiomer-specific activation of T-cell clones observed with analogues P and Q, we also

assessed whether chiral modification at the cyclopentyl group, alters HLA-B*57:01 binding and T-cell

activation. Abacavir-responsive clones were not stimulated to proliferate or secrete IFN-γ in the

presence of the alternative 1R,4S enantiomeric form of abacavir (Figure 3a and b). Incorporation of

172

173

174

175

176

177

178

179

180

181

182

183

184

185

186

187

188

189

190

191

192

193

194

195

Thomson et al 12

the 1R,4S enantiomer into our molecular docking protocol revealed a binding pose distinct from

abacavir (Figure 3c). This was characterised by a loss of interactions between the cyclopurine moiety

of the compound and amino acid residues Asp 114, Ser 116 and Ile 124. Furthermore, the

cyclopropyl moiety no longer protruded into the F-pocket. This would likely lead to a reduced

stability of the compound within HLA-B*57:01 and a loss of T-cell reactivity.

Abacavir analogues have distinct impacts on the HLA-B*57:01 immunopeptidome.

Analogues D (activates abacavir-responsive T-cells), H (activates abacavir-responsive T-cells at high

concentrations) and M (does not activate abacavir-responsive T-cells) were selected alongside

analogue 15 (from our previous study (13); does not activate abacavir-responsive T-cells) to explore

perturbation of the peptides displayed by HLA-B*57:01. Untreated and abacavir-treated cells were

used as controls. Peptides identified at a 5% FDR in the described dataset were used to define the

global peptide binding motif. HLA-B*57:01 peptide ligands were predominantly 9-11 amino acids in

length (Figure 5a i). 9mer ligands were biased towards Ser, Thr>Ala>Val at position 2 (P2) and

aromatic residues (Trp>Phe>Tyr) at the C-terminus (PΩ) whilst abacavir induced presentation of an

increased proportion of Ile (and Val) terminating peptides (Figure 4a ii, iii) (10). Similar changes were

induced for 10 and 11mer peptides (data not shown). This global alteration of the peptide-binding

motif was not recapitulated by either the cross-reactive or non-cross-reactive analogues.

To compare the precise peptide sequences presented across the conditions, peptides assigned with

a confidence above a 5% FDR in at least one data set were used to validate assignments below this

threshold in other data sets (Supplementary Table 3). Removal of peptides of the constitutive

repertoire (observed in untreated samples, or in our previous study (18) revealed 1065 potential

neo-epitopes induced by treatment with abacavir (778) or the analogues (analogue 15-277,

196

197

198

199

200

201

202

203

204

205

206

207

208

209

210

211

212

213

214

215

216

217

218

219

Thomson et al 13

analogue D-110, analogue H-83 and analogue M-150), with varying degrees of overlap (Figure 4d-g,

Supplementary table 3). Due to low peptide numbers, anchor site biases of 9-12mer peptides were

visualised collectively by aligning the first 3 (P1 to 3) and last 3 (PΩ-2 to PΩ) residues. Whilst anchor

residue enrichment in the filtered data set for analogue M (Figure 4g iii) resembled both the

unfiltered data set (Figure 4g iii) and the constitutive repertoire (Figure 4b), enrichment of PΩ Ile and

Met was observed in the filtered data sets for analogues D and H (Figure 4e and f iii). Most of these

Ile and Met terminating peptides were also present in the abacavir induced repertoire

(Supplementary Table 3), in which Ile, Leu and Met enrichment is evident at PΩ (Figure 4c). As for

abacavir, loss of Arg enrichment at the PΩ-2 secondary anchor site (19) on filtering for constitutive

ligands was also observed. These observations are consistent with the cross-recognition of

analogues D and H by abacavir primed T-cell clones, suggesting they may enable the formation of

similar HLA-drug-peptide complexes. The failure to observe a more global effect on the

immunopeptidome suggest they do so with reduced potency compared to abacavir. Strikingly, for

analogue 15 Ile enrichment was not as evident at PΩ in the filtered data set, instead enrichment of

Met and Ala was observed (Figure 4d iii). Only approximately half of these peptides were in the

overlap region with abacavir (Supplementary Table 3), suggesting analogue 15 may have a more

distinct impact on peptide binding and presentation than analogues D and H, explaining the lack of

cross-recognition of analogue 15 by the 6 abacavir T cell clones tested.

CD8+ T-cell clones can be generated to abacavir analogues, capable of cross-reacting with abacavir.

Given the distinct subset of analogue 15 induced HLA-B*57:01 ligands and its inability to activate the

abacavir-responsive T-cell clones tested, we sought to assess whether CD8+ T-cell clones could be

generated from two HLA-B*57:01+ donors after culturing their antigen presenting cells with the four

analogues (15 [does not activate abacavir clones, but displays distinct binding peptides], G [does not

activate abacavir clones], H [activates abacavir clones at high concentrations] and J [activates

220

221

222

223

224

225

226

227

228

229

230

231

232

233

234

235

236

237

238

239

240

241

242

243

244

Thomson et al 14

abacavir clones]) with PBMC. On initial testing, 24/514, and 95/282 analogue-responsive clones

were detected with analogue H and J, respectively (Supplementary figure 1; Figure 5a). Analogue H

and J dose-response studies revealed that clones secreted IFN-γ in a concentration-dependent

manner (Figure 5b). Analogue H- and J-responsive clones were also activated with abacavir (Figure

5c). All clones were phenotyped as CD8+.

Analogue 15-responsive CD8+ T-cell clones were generated in large numbers (99/401, numerous

producing a stimulation index of fifty or above) when PBMC from HLA-B*57:01+ donors were

cultured for 2 weeks the compound (Figure 5 a and b; Supplementary Figure 2). In contrast to

abacavir-responsive clones, these clones were stimulated to secrete IFN-γ in the presence of

analogue 15 and abacavir (Figure 5b and c).

None of the 568 clones generated from PBMC cultured with analogue G were activated with this

compound (Figure 5a).

245

246

247

248

249

250

251

252

253

254

255

256

257

258

259

260

Thomson et al 15

Discussion

X-ray crystal structures of HLA-B*57:01-abacavir-peptide complexes show that abacavir binds

directly to HLA-B*57:01, deep within the peptide binding cleft, underneath the C-terminal portion of

the peptide (9,10). In agreement with this, previous analyses of the immunopeptidomes of abacavir-

treated cells expressing HLA-B*57:01 revealed a change in the peptide repertoire (9–11), with a

decrease in the prevalence of peptides containing larger amino acids such as tryptophan and

phenylalanine at the PΩ position accompanied by an increase in peptides with smaller amino acids

such as valine, leucine and isoleucine at this location. Thus, the abacavir HLA-B*57:01 interaction

creates an altered chemical space for peptide binding and it has been proposed that the novel self-

peptide-abacavir-HLA-B*57:01 complexes are responsible for activating the T-cells involved in

abacavir hypersensitivity. Therefore, the objective of this study was to utilise synthetic analogues of

abacavir with structural substitutions in place of the 6-amino cyclopropyl moiety to probe the drug

HLA-B*57:01 binding interaction and to delineate how alteration of this functional group impacts

peptide presentation and T cell activation. The first fourteen analogues were synthesised around the

azetidine ring, a stable structure which allowed for the easy synthesis of further analogues with

additional functional groups. The final three analogues were structural variants of the amino group,

yielded by opening the azetidine ring structure. Analogues P and Q were of particular interest as

they allowed us to probe HLA binding and T-cell activation with different enantiomeric forms of the

same compound.

As described in our previous study (13), through structural modification several compounds that

retained a degree of pharmacological activity, but did not stimulate abacavir-responsive T-cells, were

identified (i.e., analogues B, G, and M; analogue H only activated clones at high concentrations).

Replacement of the cyclopropyl group with an azetidine ring per se yielded an analogue (D) that

displayed equipotent antiviral activity to abacavir, but this compound activated T-cells. However, the

261

262

263

264

265

266

267

268

269

270

271

272

273

274

275

276

277

278

279

280

281

282

283

284

285

Thomson et al 16

addition of branched carbon chains at the 3-position of the azetidine ring produced abacavir

analogues that did not activate cloned T-cells. This suggests the 3-position of the azetidine ring plays

a pivotal role in the formation of immunogenic drug HLA-B*57:01 peptide complexes. As there was

no simple chemical explanation for T-cell response profiles observed with the azetidine ring

containing analogues, molecular modelling and mass spectrometry-based assessment of HLA-

B*57:01 bound peptides were performed.

Analogue D activated abacavir-responsive T-cells and unsurprisingly was predicted to bind to HLA-

B*57:01 in a similar conformation to abacavir. Co-incubation of C1R-B*57:01 cells with analogue D

yielded an enrichment of C-terminal isoleucine in HLA-B*57:01 bound peptides, correlating with the

observed cross-reactivity with of abacavir-responsive T-cell clones. Abacavir-responsive clones were

also activated with analogue H, albeit only at high concentrations. The predicted binding pose of

analogue H within HLA-B*57:01 was distinct to abacavir and analogue D, with the guanosine portion

of the analogue inverted. Although C-terminal isoleucine enrichment was observed on HLA-B*57:01

binding peptides when C1R-B*57:01 cells were incubated with analogue H, a lower proportion of the

new peptide sequences overlapped with those observed with abacavir-treated cells. Unfavourable

binding interactions were predicted with the analogues G and M and the HLA-B*57:01 binding

peptide HSITYLLPV. Importantly, neither of these analogues activated abacavir-responsive clones.

Due to stock limitations, peptide repertoire studies were not performed with analogue G. However,

the peptides displayed by HLA-B*57:01 when C1R-B*57:01 were incubated with analogue M

mirrored the peptides constitutively expressed by HLA-B*57:01 suggesting no perturbation of the

immunopeptidome.

Analogue 15 from our previous study retained antiviral activity. It was predicted to bind in an

unfavourable conformation with HLA-B*57:01 and did not activate abacavir-responsive T-cells (13).

286

287

288

289

290

291

292

293

294

295

296

297

298

299

300

301

302

303

304

305

306

307

308

309

310

Thomson et al 17

Thus, the impact of analogue 15 on the HLA-B*57:01 peptide repertoire was studied. Minor

perturbation of the HLA-B*57:01 peptide repertoire displayed by C1R-B*57:01 cells was evident.

Unlike abacavir, which induces an increase in HLA-B*57:01 peptides terminating in isoleucine (9–11),

analogue 15 promoted an enrichment of alanine at the C-terminus. Given alanine terminating

peptides form a much smaller part of the observed abacavir-induced HLA-B*57:01 peptide

repertoire, it is not surprising that analogue 15 failed to activate abacavir-responsive T-cells.

However, the data raise the possibility that certain analogues contained within this and our previous

series of compounds (13) might activate alternative CD8+ T-cells to abacavir. To address this, PBMC

were cultured with 4 analogues (15 [does not activate abacavir clones, but displays distinct binding

peptides], G [does not activate abacavir clones], H [activates abacavir clones at high concentrations]

and J [activates abacavir clones]) prior to generation of T-cell clones and assessment of antigen

specificity. Clones responsive towards analogues 15, H and J were detected. Exposure of the clones

to antigen presenting cells and the analogue or abacavir promoted IFN-γ release. In contrast, the

clones generated from PBMC cultured with analogue G were not activated with analogue G or

abacavir. Collectively, these data show that although modelling experiments provide an effective

system to screen candidate compound HLA allele binding, it must be coupled with functional

analyses of compound-specific T-cell responses to predict immunogenicity.

The concept of enantiomeric-specific T-cell reactivity has seldom been explored in the context of

drug hypersensitivity. Thus, the different enantiomeric forms of butyl amino abacavir (analogues P

and Q) were synthesised. Enantiomers have identical chemical and physical properties, but are

mirror images that are non-superimposable. For this reason, enantiomeric forms of drugs may bind

differently to biological targets and display divergent pharmacological and toxicological properties.

Analogue P was predicted to bind to HLA-B*57:01 in an analogous manner to abacavir, while

analogue Q induced steric clashes with HLA-B*57:01 and the binding peptide. In agreement with the

311

312

313

314

315

316

317

318

319

320

321

322

323

324

325

326

327

328

329

330

331

332

333

334

335

Thomson et al 18

model, analogue P activated abacavir-responsive T-cells, while analogue Q did not. Abacavir also

exists in different enantiomeric forms at the cyclopentene ring. The pharmacological enantiomer has

a 1S, 4R confirmation. The alternative enantiomer, which is pharmacologically inactive, has a 1R, 4S

confirmation. This enantiomer displayed a predicted binding conformation distinct to abacavir

where the cyclopropyl group no longer protruded into the F-pocket and this led to a complete loss of

T-cell reactivity.

In conclusion, the coupling of modelling and analytical methods for functional analysis of compound-

specific human CD8+ T-cell responses has enhanced our understanding of how drugs interact with

HLA proteins to stimulate T-cells that participate in serious hypersensitivity reactions. The approach

offers an opportunity for safety scientists working in drug development to screen the potential

immunogenicity of second line compounds when HLA allele-restricted forms of drug hypersensitivity

are detected in the clinic.

336

337

338

339

340

341

342

343

344

345

346

347

348

349

350

351

352

353

354

355

Thomson et al 19

Author contributions: PJT, JF, CB and MA conducted the biological experiments. PI conducted the

peptide elution and immunopeptidome analyses. NB conducted the HLA-B*57:01 modelling

experiments. PON synthesized the abacavir analogues. DJN, BKP, AWP and PON designed the study

and supervised the project. PJT, PI, DJN and NB analysed. PJT, PI and DJN drafted the manuscript. All

authors critically reviewed the manuscript.

Abbreviations

Epstein-Barr Virus, EBV; reversed phase high performance liquid chromatography, HPLC; liquid

chromatography-tandem mass spectrometry, LC-MS/MS; stimulation index, SI; PBMC, peripheral

blood mononuclear cells.

356

357

358

359

360

361

362

363

364

365

366

Thomson et al 20

Figure legends

Figure 1. Activation of abacavir-responsive CD8+ T-cell clones with 6-amino cyclopropyl group

substituted analogues. a) Proliferative response and b) IFN-γ secretion of T-cell clones incubated

with abacavir (35µM). Data shown as mean ± SEM of all responsive clones vs control. * P<0.05, **

P<0.01, *** P<0.001. c) IFN-γ secretion from abacavir-responsive T-cell clones incubated with

analogues A-Q at concentrations of 0, 10, 20 and 50µM. Each analogue was tested against six

abacavir clones from four HLA-B*57:01+, healthy drug-naïve donors. Mean abacavir IFN-γ release

from the 6 clones (dashed line) was used as a comparator. Analogues K-Q were performed on an

independent occasion from A-N, accounting for the differences in mean abacavir value.

Figure 2. Direct comparison of the CD8+ T-cell activity of abacavir substituted analogues and the

binding orientations within the F-pocket of HLA-B*57:01. a) Representative ELIspot images from

wells containing abacavir and analogues D, G, H, M, O, P and Q at concentrations of 0, 10, 20 and

50µM. b) Crystal structure binding orientation of abacavir c) Crystal structure binding orientation of

analogues capable of activating abacavir-responsive T-cell clones i) D (azetidine), ii) H (azetidine-

fluoro), iii) O (isobutylamino) and iv) P (S-sec-butyl amino) within the F-pocket of HLA-B*57:01. d)

Crystal structure binding orientation of abacavir analogues with no cross-reactivity with abacavir

responsive T-cell clones i) G (azetidine-3-carbonitrile), ii) M (azetidine-trifluoromethyl), iv) Q (R- sec-

butyl amino) within the F-pocket of HLA-B*57:01. Crystal structure of HLA-B*57:01 (PDB:3UPR). Stick

representation of the peptide, (HSITYLLPV) shown in red. Key amino acid residues are shown as

yellow sticks. All non-polar hydrogen atoms removed. Key hydrogen bond interactions shown as

black dashes. Spheres used to illustrate the atomic radii of the atoms in the functional groups of

abacavir and substituted analogues, peptide (HSITYLLPV) and HLA-B*57:01 amino acids. Abacavir

structure superimposed over analogues shown as blue lines.

367

368

369

370

371

372

373

374

375

376

377

378

379

380

381

382

383

384

385

386

387

388

389

390

391

Thomson et al 21

Figure 3. Activation of abacavir-responsive CD8+ T-cell clones is enantiomer specific. a) Proliferation

of abacavir responsive T-cell clones cultured in the presence of abacavir and 1R, 4S abacavir

enantiomer at concentrations of 1, 5, 10, 50 and 100µM for 48 hours. b) Representative ELIspot

images from wells containing abacavir and 1R, 4S abacavir both at 50µM. c) Docking solution of

abacavir enantiomer (1R,4S) with abacavir superimposed, shown as blue lines in the F-pocket of

HLA-B*57:01. Stick representation of the peptide, (HSITYLLPV) shown in red. Amino acid protein

residues shown as yellow sticks, with key hydrogen bond interactions shown as black dashes. All

non-polar hydrogen atoms removed. Spheres used to illustrate the atomic radii of the atoms in the

functional group of the compounds.

Figure 4 a) Broad perturbation of C-terminal anchor preference is observed in the presence

abacavir but not analogues 15, D, H or M; however, potential neo-epitopes show a range of

overlap with abacavir induced ligands. a) Length (i) and primary anchor characteristics (ii. Position 2,

iii. C-terminal) of HLA-B*57:01 ligands isolated from CIR.B*57:01 grown in the absence of drug

treatment, or in the presence of 35μM abacavir (black) or analogues 15 (red), D (green), H (orange)

or M (blue). Analyses are based on non-redundant peptide identifications (by sequence,

modifications not considered) per data set made at a confidence greater than that for a 5% false

discovery rate (FDR) and filtered for ligands of endogenous HLA molecules of parental CIR cells.

Anchor residue preferences are shown for 9mers and are depicted as the proportion of peptides that

possess specific amino acids at position 2 (ii) and the C-terminus (iii). Data shown is the mean ± SD of

triplicate experiments for untreated and abacavir treated cells, duplicate experiments for analogue

15, and single experiments for the remaining analogues (due to availability of compounds). b)

iceLogos for P1 to P3 and PΩ-2 to PΩ of 9-12mer peptides in the constitutive repertoire of HLA-

B*57:01. c) iceLogos for 9-12mer HLA-B*57:01 ligands detected in the presence of abacavir in this

study either i) unfiltered or ii) filtered for constitutive ligands to identify potential neo-epitopes. d),

392

393

394

395

396

397

398

399

400

401

402

403

404

405

406

407

408

409

410

411

412

413

414

415

416

Thomson et al 22

e), f) and g) show i) icelogos for 9-12mer HLA-B*57:01 ligands detected in the presence of analogues

15, D, H and M, ii) Venn diagrams showing the numbers of potential neo-epitopes identified in both

abacavir and analogue treatments and iii. iceLogos for 9-12mer neo-epitopes identified. iceLogos

were generated using icelogo software utilising the human swiss-prot proteome as the reference

set. Letter height corresponds to % difference in frequency of the amino acid compared to presence

in the human proteome.

Figure 5. Generation of CD8+ T-cell clones to 6-amino substituted abacavir analogues. a) Antigen

specificity of T-cell clones generated to analogues G, H, 15 and J by way of cellular proliferation.

Clones yielding an SI<2 were considered positive. b) Representative ELIspot images from wells

containing clones from two HLA-B*57:01+ donors incubated in the presence of analogues H, 15 or J

(5-20µM). c) Representative ELIspot images from wells containing clones from two HLA-B*57:01+

donors incubated in the presence of abacavir and analogues H, 15 or J (35µM).

417

418

419

420

421

422

423

424

425

426

427

428

429

430

431

432

Thomson et al 23

References

1. Hetherington S, McGuirk S, Powell G, Cutrell A, Naderer O, Spreen B, et al. Hypersensitivity

reactions during therapy with the nucleoside reverse transcriptase inhibitor abacavir. Clin

Ther. 2001; 23:1603–14.

2. Mallal S, Nolan D, Witt C, Masel G, Martin AM, Moore C, et al. Association between presence

of HLA-B*5701, HLA-DR7, and HLA-DQ3 and hypersensitivity to HIV-1 reverse-transcriptase

inhibitor abacavir. Lancet. 2002; 359:727–32.

3. Hetherington S, Hughes AR, Mosteller M, Shortino D, Baker KL, Spreen W, et al. Genetic

variations in HLA-B region and hypersensitivity reactions to abacavir. Lancet. 2002; 359:1121–

2.

4. Martin AM, Nolan D, Gaudieri S, Almeida CA, Nolan R, James I, et al. Predisposition to

abacavir hypersensitivity conferred by HLA-B * 5701 and a haplotypic Hsp70-Hom variant.

Proc Natl Acad Sci U S A. 2004; 101:4180–5.

5. Mallal S, Phillips E, Carosi G, Molina J-M, Workman C, Tomažič J, et al. HLA-B*5701 Screening

for Hypersensitivity to Abacavir. N Engl J Med. 2008; 358:568–79.

6. Lucas A, Lucas M, Strhyn A, Keane NM, McKinnon E, Pavlos R, et al. Abacavir-reactive memory

T cells are present in drug naïve individuals. PLoS One. 2015; 10:e0117160.

7. Hughes DA, Vilar FJ, Ward CC, Alfirevic A, Park BK, Pirmohamed M. Cost-effectiveness analysis

of HLA B * 5701 genotyping in preventing abacavir hypersensitivity. Pharmacogenomics.

2004; 14:335–42.

8. Schackman BR, Scott C a, Walensky RP, Losina E, Freedberg K a, Sax PE. The cost-effectiveness

of HLA-B*5701 genetic screening to guide initial antiretroviral therapy for HIV. AIDS. 2008;

22:2025–33.

433

434

435

436

437

438

439

440

441

442

443

444

445

446

447

448

449

450

451

452

453

454

455

Thomson et al 24

9. Ostrov D a, Grant BJ, Pompeu Y a, Sidney J, Harndahl M, Southwood S, et al. Drug

hypersensitivity caused by alteration of the MHC-presented self-peptide repertoire. Proc Natl

Acad Sci U S A. 2012; 109:9959–64.

10. Illing PT, Vivian JP, Dudek NL, Kostenko L, Chen Z, Bharadwaj M, et al. Immune self-reactivity

triggered by drug-modified HLA-peptide repertoire. Nature. 2012; 486(7404):554–8.

11. Norcross MA, Luo S, Lu L, Boyne MT, Gomarteli M, Rennelsc AD, et al. Abacavir induces

loading of novel self-peptides into HLA-BM 57:01: an autoimmune model for HLA-associated

drug hypersensitivity. AIDS. 2012; 18:1199–216.

12. Bell CC, Faulkner L, Martinsson K, Farrell J, Al A, Tugwood J, et al. T-Cells from HLA-B*57:01+

Human Subjects Are Activated with Abacavir through Two Independent Pathways and Induce

Cell Death by Multiple Mechanisms. Chem Res Toxicol. 2013; 26:759–66.

13. Naisbitt DJ, Yang EL, Alhaidari M, Berry NG, Lawrenson AS, Farrell J, et al. Towards

depersonalized abacavir therapy : chemical modification eliminates HLA-B M 57 : 01-

restricted CD8 R T-cell activation. AIDS. 2015; 29:2385–95.

14. Chessman D, Kostenko L, Lethborg T, Purcell AW, Williamson N a, Chen Z, et al. Human

leukocyte antigen class I-restricted activation of CD8+ T cells provides the immunogenetic

basis of a systemic drug hypersensitivity. Immunity. 2008; 28:822–32.

15. Adam J, Eriksson KK, Schnyder B, Fontana S, Pichler WJ, Yerly D. Avidity determines T-cell

reactivity in abacavir hypersensitivity. Eur J Immunol. 2012; 42:1706–16.

16. Yerly D, Pompeu YA, Schutte RJ, Eriksson KK, Strhyn A, Bracey AW, et al. Structural Elements

Recognized by Abacavir-Induced T Cells. Int J Mol Sci. 2017; 18(1464):1–10.

17. Cardone M, Garcia K, Tilahun ME, Boyd LF, Gebreyohannes S, Yano M, et al. A transgenic

mouse model for HLA-B*57:01-linked abacavir drug tolerance and reactivity. J Clin Invest.

2018; 128:2819–32.

456

457

458

459

460

461

462

463

464

465

466

467

468

469

470

471

472

473

474

475

476

477

478

479

Thomson et al 25

18. Pymm P, Illing PT, Ramarathinam SH, O’Connor GM, Hughes VA, Hitchen C, et al. MHC-I

peptides get out of the groove and enable a novel mechanism of HIV-1 escape. Nat Struct

Mol Biol. 2017;24:387–94.

19. Illing PT, Pymm P, Croft NP, Hilton HG, Jojic V, Han AS, et al. HLA-B57 micropolymorphism

defines the sequence and conformational breadth of the immunopeptidome. Nat Commun.

2018;9(1).

480

481

482

483

484

485

486

Thomson et al 26

Supplementary methods

Culture media.

Cell culture medium for T-cells (R9) is composed of RPMI supplemented with 10% human AB serum,

HEPES (25mM), penicillin (1000 U/mL), streptomycin (0.1mg/mL), L-glutamine (2mM) and transferrin

(25µg/mL). Epstein-barr virus (EBV) -transformed B-lymphoblastoid cell lines were cultured in F1

medium composed of RPMI supplemented with 10% foetal bovine serum, HEPES (25mM), penicillin

(1000 U/mL), streptomycin (0.1mg/mL) and L-glutamine (2mM).

Synthesis of 6-amino substituted abacavir analogues.

The initial stage involved the synthesis of a stock of intermediate 1. From this chiral intermediate the

synthesis of seventeen target molecules (A-Q) was conducted (Supplementary Figure 1). For

analogues A-N, intermediate 1 (150.00 mg, 564.55 µmol, 1.00 eq), azetidine variants (2.00 eq) and N,

N-diisopropylethylamine (145.92 mg, 1.13 mmol, 2.00 eq) were taken up into a microwave tube in

isopropyl alcohol (2.00 mL). The sealed tube was heated at 70 °C for 2 hours under microwave. LC-

MS showed that the starting material was consumed completely. The mixture was concentrated in

vacuum to give crude product. The crude product was then purified by thin-layer chromatography.

Analogues O-Q were synthesized following the same procedure with the azetidine variants replaced

with derivatives of the amino group.

Donor characteristics and T-cell cloning.

Four HLA-B*57:01+ donors were selected from the Liverpool Centre for Drug Safety Science cell bank

containing peripheral blood mononuclear cells (PBMCs) from 1200 genotyped healthy donors.

Approval for the study was obtained from the Liverpool research ethics committee and informed

487

488

489

490

491

492

493

494

495

496

497

498

499

500

501

502

503

504

505

506

507

508

509

Thomson et al 27

consent was received from participants prior to inclusion in the study. PBMCs were incubated in the

presence of abacavir, or analogues G, H, 15 (from our previous study (1)) or J (35µM), in R9 medium

for a period of 14 days. On days 6 and 9 cells were fed with R9 medium containing recombinant

human IL-2 (100U/mL) to preserve the drug driven expansion of T-cells. On day 14, CD8+ T-cells were

positively selected using MultiSort kits (Miltenyi Biotec, Surrey UK) and T-cell clones were generated

via means of serial dilution and phytohaemagglutinin (PHA; 5µg/mL) stimulation. T-cells were fed

every 2 days with R9 medium containing IL-2 (100U/mL) and growing clones were transferred to a

new 96 well plate and expanded across 4 wells. Clones were restimulated and further expanded

every 14 days.

Drug-specific T-cell responses.

Specificity of CD8+ clones was measured by culturing T-cells with irradiated autologous EBV-

transformed B-cells (as antigen presenting cells; 1x104/well) ± abacavir (35µM). Following a 48 h

incubation, [3H] thymidine (0.5µCi) was added and cellular proliferation assessed 16 h later via

scintillation counting. CD phenotyping of clones was performed using BD FACSCanto II flow

cytometer.

IFN-γ ELIspot was used as a second readout of the drug-specific response to assess dose-dependent

T-cell activation and cross-reactivity. Drug-specific clones (5x104, 50µL) were added to ELIspot plates

with EBV transformed B-cells (1x104, 50µL) and abacavir (analogues) (10, 20, 50µM; 100µL) for a

period of 48 h. Spot forming units (cytokine-secreting T-cells) were visualised and quantified using an

AID ELIspot reader (Cadama Medical, Stourbridge, UK).

Modelling of abacavir (analogues) HLA-B*57:01 binding.

510

511

512

513

514

515

516

517

518

519

520

521

522

523

524

525

526

527

528

529

530

531

532

533

Thomson et al 28

HLA, HLA binding peptide (HSITYLLPV) and drug ligands (abacavir and analogues D, G, H, M, O, P and

Q) were prepared for docking in Spartan’08 (wavefunction inc. Irvine, California, USA: 1991-2009).

Native SMA ligand was exported from PDB 3UPR. For each analogue, the cyclopropyl group of

abacavir was replaced. Merk molecular force field minimization calculations were performed with all

atoms frozen except newly added substitutions. For docking studies GOLD 5.1 (CCDC Software

Limited, Cambridge, UK) was used to examine the predicted binding poses of the abacavir analogues

within the F-pocket of HLA-B*57:01, PDB code 3UPR (2). Figures of abacavir analogues predicted

binding conformations within HLA-B*57:01 were produced using PYMOL software version 2.5.

MHC purification and peptide elution.

C1R.B*57:01 are transfectants of C1R cells expressing HLA-B*57:01 under geneticin selection (3).

C1R.B*57:01 were grown in RF10 [RPMI 1640 (Life Technologies, USA) supplemented with 10%

foetal bovine serum (FBS; Sigma, St Louis, USA), 7.5 mM HEPES (MP Biomedicals, Germany), 100

U/mL Pen-Strep (benzyl-penicillin/streptomycin, Life Technologies, USA), 2 mM L-glutamine (MP

Biomedicals, Germany), 76μM β-mercaptoethanolamine (Sigma-Aldrich, USA) and 150μM non-

essential amino acids (LifeTechnologies, USA)]. HLA expression was maintained in long term culture

with 0.5mg/mL geneticin (G418; LifeTechnologies, USA).

C1R.B*57:01 cells were grown to high density in the presence or absence abacavir and analogues D

(activates abacavir-responsive T-cells), H (activates abacavir-responsive T-cells at high

concentrations), M (does not activate abacavir-responsive T-cells) and 15 (from our previous study

(1) does not activate abacavir-responsive T-cells) (35μM) for a minimum of 4 days, prior to washing

in PBS, pelleting and snap freezing in liquid nitrogen. Cell pellets of 4-5x108 cells were lysed by

mechanical and detergent based lysis, the lysates cleared by ultracentrifugation, and HLA class I

534

535

536

537

538

539

540

541

542

543

544

545

546

547

548

549

550

551

552

553

554

555

556

557

Thomson et al 29

complexes isolated by immunoaffinity purification using solid-phase bound pan class I antibody

W6/32 as described previously (4). Complexes were dissociated using 10 % acetic acid and

fractionated by Reversed Phase High Performance Liquid Chromatography (HPLC) on a 4.6mm

internal diameter x 100mm monolithic reversed-phase C18 HPLC column (Chromolith SpeedROD;

Merck Millipore) using an ÄKTAmicro HPLC (GE Healthcare) system. The peptide/MHC mixture was

loaded at 1mL/min onto the column in 98% Buffer A (0.1% Trifluoroacetic acid) and 2% Buffer B (80%

Acetonitrile, 0.1% trifluoroacetic acid), and bound material eluted by running a gradient of buffer B

at 2ml/min of 2-15% over 0.25 minutes, 15-30% over 4 minutes, 30-40% over 8 minutes, 40-45%

buffer B over 10 min, with collection of 500µL fractions. Peptide containing fractions were vacuum

concentrated, pooled into 9-12 pools, and reconstituted in 0.1% formic acid.

Mass spectrometric analysis.

Reconstituted fraction pools were analysed by liquid chromatography-tandem mass spectrometry

(LC-MS/MS) via a data dependent acquisition strategy using a NanoUltra cHiPLC system (Eksigent)

coupled to an SCIEX 5600+ TripleTOF mass spectrometer equipped with a Nanospray III ion source.

Samples were loaded onto a pre-equilibrated cHiPLC trap column (3µm, ChromXP C18CL, 120 Å, 0.5

mm x 200 µm), at 5µL/min in 0.1% formic acid, 2% acetonitrile, and separated over a cHiPLC column

(3µm, ChromXP C18CL, 120 Å, 15cm x 75µm) using a linear gradient of 2-35% Buffer B (80%

acetonitrile, 0.1% formic acid)/Buffer A (0.1% formic acid) over 75 minutes at a flow rate of

300nL/min. Data acquisition occurred with the following instrument parameters: ion spray voltage,

2,400 V; curtain gas, 30 l/min; ion source gas, 20 l/min; and interface heater temperature, 150 °C.

MS/MS switch criteria selected the top 20 ions meeting the following criteria per cycle: m/z >200

amu, charge state of +2 to +5, intensity >40 counts per second. After two selections for

fragmentation, ions were ignored for 30 seconds. For assignment, MS/MS spectra were searched

against the human proteome (UniProt/Swiss-Prot accessed November 2017) using ProteinPilot™

software (version 5.0, SCIEX), considering biological modifications and utilising a decoy database for

558

559

560

561

562

563

564

565

566

567

568

569

570

571

572

573

574

575

576

577

578

579

580

581

582

583

Thomson et al 30

false discovery rate (FDR) calculations. Identifications were filtered for peptides seen in HLA class I

immunoaffinity purifications from parental C1R (includes contaminants and peptide binders of

endogenous HLA-B*35:03 and HLA-C*04:01), class II purifications, common contaminants of MHC

pull downs observed in the lab, as well as peptides derived from the HLA proteins themselves

(Supplementary table 1).

HLA-B*57:01 immunopeptidome analysis.

To define the global peptide binding motif under different conditions, distinct peptides assigned by

ProteinPilot with a confidence above a local 5% FDR cut-off and delta mass < 0.05 were considered

(Supplementary table 2). Non-redundant sequences were used to calculate the prevalence of

peptides of each length, and of nine amino acid peptides possessing specific residues at the primary

anchor positions (P2 and PΩ). To compare peptide presentation across the conditions, and filter for

peptides of the constitutive HLA-B*57:01 repertoire, a combined list of peptides identified with a

confidence above a 5% FDR cut-off in at least one data set was used to interrogate all data sets

(Supplementary table 3). Where assignments were made in multiple data sets at a confidence above

the 5% FDR cut-off, the mean and SD of the retention time was calculated. If a peptide assignment

was made below the 5% FDR cut-off, it was considered valid if the retention time was within 5

minutes of the mean 5% FDR retention time. Assignments where the SD of the 5% FDR retention

time was greater than 2.5 were excluded. Abacavir and analogue treated data sets (non-redundant

by sequence) were filtered (by sequence, modifications not considered) of constitutive ligands

identified in the retention time validated untreated data sets as well as those described previously

(5). Sequence features of the retention time validated and filtered data sets were visualised using

Icelogo software (6) (percentage difference to human proteome using static reference method) at P1

to 3 and PΩ-2 to PΩ of 9-12mer peptide ligands.

584

585

586

587

588

589

590

591

592

593

594

595

596

597

598

599

600

601

602

603

604

605

606

607

608

Thomson et al 31

Antiviral activity.

The antiviral activity of the abacavir analogues was measured using cell viability assays carried out by

the Wuxi App Tech company in Shanghai. The antiviral activity was measured on the basis that HIV

kills MT-4 cells meaning cell survival in the presence of an analogue was a measure of antiviral

activity. The antiviral activity of the abacavir analogues was quantified as EC50 and classified into

four categories: good antiviral activity (>10µM), retained antiviral activity (10-60µM), little antiviral

activity (60-500µM) and no antiviral activity (>500µM).

Supplementary references

1. Naisbitt DJ, Yang EL, Alhaidari M, Berry NG, Lawrenson AS, Farrell J, et al. Towards

depersonalized abacavir therapy : chemical modification eliminates HLA-B M 57 : 01-

restricted CD8 R T-cell activation. AIDS. 2015; 29:2385-95.

2. Ostrov D a, Grant BJ, Pompeu Y a, Sidney J, Harndahl M, Southwood S, et al. Drug

hypersensitivity caused by alteration of the MHC-presented self-peptide repertoire. Proc Natl

Acad Sci U S A. 2012; 109:9959–64.

3. Chessman D, Kostenko L, Lethborg T, Purcell AW, Williamson N a, Chen Z, et al. Human

leukocyte antigen class I-restricted activation of CD8+ T cells provides the immunogenetic

basis of a systemic drug hypersensitivity. Immunity. 2008; 28:822–32.

4. Dudek NL, Tan CT, Gorasia DG, Croft NP, Illing PT, Purcell AW. Constitutive and inflammatory

immunopeptidome of pancreatic β-cells. Diabetes. 2012; 61:3018–25.

5. Pymm P, Illing PT, Ramarathinam SH, O’Connor GM, Hughes VA, Hitchen C, et al. MHC-I

peptides get out of the groove and enable a novel mechanism of HIV-1 escape. Nat Struct

Mol Biol. 2017; 24:387–94.

609

610

611

612

613

614

615

616

617

618

619

620

621

622

623

624

625

626

627

628

629

630

631

632

Thomson et al 32

6. Colaert N, Helsens K, Martens L, Vandekerckhove J, Gevaert K. Improved visualization of

protein consensus sequences by iceLogo. Nat Methods. 2009; 6:786–7.

Supplementary data

Supplementary table 1:

Peptide contaminants removed from analyses including those seen in similar HLA peptide analyses in

CIR parental cells, class II binders, HLA derived peptides, and peptides with the HLA-C*04:01 motif

(either P2 (F/Y) and P3 (D), or P2(F/Y) and PΩ (LFVM)) of endogenous CIR HLA-C.

Supplementary table 2:

Peptides identified above a 5% FDR cut-off by ProteinPilot™ 5.0 for individual LC-MS/MS data sets.

Supplementary table 3:

Comparison of peptides identified within the data sets using retention time validation of low

confidence assignments by comparison to assignments above 5% FDR cut-off. Identifications are

shown in the IDs tab. Comparison of the data sets by sequence is shown in the Sequences tab.

Supplementary Table 4. Antiviral activity of analogues A-Q (EC50) and ability to induce T-cell

activation in the presence of abacavir-responsive T-cell clones. *T-cell activity only observed at high

concentrations of the analogue.

633

634

635

636

637

638

639

640

641

642

643

644

645

646

647

648

649

650

651

652

Thomson et al 33

Supplementary Figure 1. Synthesis and structure of abacavir substituted analogues. a) Synthesis of

abacavir substituted analogues i) A-N ii) O-Q. b) Table comparing respective functional groups at the

6-amino cyclopropyl moiety of the abacavir analogues.

Supplementary Figure 2. Mean proliferative response from T-cell clones incubated in the presence

of a) analogue H. b) analogue 15 c) analogue J. Data shows as mean ± SEM of all responsive clones vs

control ± SEM (* P<0.05, ** P<0.01, *** P<0.001).

653

654

655

656

657

658