1

The CD8+ memory stem T cell (TSCM) subset is associated with improved 1

prognosis in chronic HIV-1 infection 2

3

4

Susan P. Ribeiro1,2, Jeffrey M. Milush3, Edecio Cunha-Neto1,2,4, Esper G. Kallas1,2, 5

Jorge Kalil1,2,4,5, Ma Somsouk6, Peter W. Hunt7, Steven G. Deeks7, Douglas F. Nixon8, 6

Devi SenGupta3# 7

8

1. Laboratory of Clinical Immunology and Allergy-LIM60/ University of Sao Paulo 9

School of Medicine, São Paulo, Brazil 10

2. Institute of Investigation in Immunology – iii-INCT, São Paulo, Brazil 11

3. Division of Experimental Medicine, Department of Medicine, University of 12

California San Francisco, San Francisco, California, USA. 13

4. Laboratory of Immunology, Heart Institute, University of Sao Paulo School of 14

Medicine 15

5. Butantan Institute, Butantã, São Paulo - SP, Brazil 16

6. Division of Gastroenterology, Department of Medicine, University of California 17

San Francisco, San Francisco, CA, USA. 18

7. HIV/AIDS Division, Department of Medicine, San Francisco General Hospital, 19

University of California, San Francisco, CA, USA. 20

8. Department of Microbiology, Immunology & Tropical Medicine, The George 21

Washington University, Washington, D.C., USA 22

23

Running title: CD8+ TSCM in chronic HIV-1 infection 24

JVI Accepts, published online ahead of print on 24 September 2014J. Virol. doi:10.1128/JVI.01948-14Copyright © 2014, American Society for Microbiology. All Rights Reserved.

2

25

#Address correspondence to Devi SenGupta, devi.sengupta@ucsf.edu 26

Abstract word count: 249 27

Text word count: 3401 28

3

Abstract 29

Memory stem T cells (TSCM) constitute a long-lived, self-renewing lymphocyte 30

population essential for the maintenance functional immunity. The hallmarks of HIV-1 31

pathogenesis are CD4+ T cell depletion and abnormal cellular activation. We 32

investigated the impact of HIV-1 infection on the TSCM compartment, as well as any 33

protective role these cells may have in disease progression, by characterizing this 34

subset in a cohort of 113 subjects with variable degrees of viral control on and off 35

highly active antiretroviral therapy (HAART). We observed that the frequency of CD8+ 36

TSCM is decreased in all individuals with chronic, untreated HIV-1 infection, and that 37

HAART has a restorative effect on this subset. In contrast, natural controllers of HIV-1 38

had the highest absolute number of CD4+ TSCM cells among all of the infected groups. 39

The frequency of CD4+ TSCM predicted higher CD8+ TSCM frequencies, consistent with a 40

role for the CD4+ subset in helping to maintain CD8+ memory T cells. In addition, TSCM 41

appeared to be progenitors for effector T cells (TEM), as these two compartments were 42

inversely correlated. Increased frequencies of CD8+ TSCM predicted lower viral loads, 43

higher CD4+ counts and less CD8+ T cell activation. Finally, we found that TSCM 44

express the mucosal homing integrin α4β7 and can be identified in the gastrointestinal-45

associated lymphoid tissue (GALT). The frequency of mucosal CD4+ TSCM was 46

inversely correlated with that in the blood, potentially reflecting the ability of these self-47

renewing cells to migrate to a crucial site of ongoing viral replication and CD4+ T cell 48

depletion. 49

Importance: HIV-1 infection leads to profound impairment of the immune system. TSCM 50

constitute a recently identified lymphocyte subset with stem-cell like qualities including 51

4

the ability to generate other memory T cell subtypes, and are therefore likely to play an 52

important role in controlling viral infection. We investigated the relationship between the 53

size of the CD8+ TSCM compartment and HIV-1 disease progression in a cohort of 54

chronically infected individuals. Our results suggest that HAART restores a normal 55

frequency of CD8+ TSCM and that the natural preservation of this subset in the setting of 56

untreated HIV-1 infection is associated with improved viral control and immunity. 57

Therefore, the CD8+ TSCM population may represent a correlate of protection in chronic 58

HIV-1 infection that is directly relevant to the design of T cell-based vaccines, adoptive 59

immunotherapy approaches or the pharmacologic induction of TSCM. 60

61

Keywords: Memory stem T cell, HIV-1 infection, memory T cells, immune activation, 62

HIV-1 controllers, GALT 63

64

5

Introduction 65

Human immunodeficiency virus type 1 (HIV-1) causes a slow and progressive disease 66

characterized by a gradual decline in CD4+ T helper cells. In addition to the loss of 67

CD4+ cells, there is widespread degeneration of the immune system resulting from 68

immune activation caused in part by gut barrier disruption (1, 2). After infection, cellular 69

activation and continued viral replication drive the immune system to exhaustion, 70

reflected by an increase in cellular markers of aging and senescence. Non-human 71

primate studies suggest that the dynamics of memory T cell subsets also play a role in 72

the pathogenesis of the disease (3). During untreated infection, HIV-1 preferentially 73

infects activated effector memory T cells (TEM) (4). Initiation of highly active 74

antiretroviral therapy (HAART) is accompanied by a reduced frequency of T cells 75

harboring active HIV-1 replication (5), although viral DNA is still detected at low levels 76

in resting memory T cells. T cell survival and homeostatic proliferation are two major 77

mechanisms implicated in the maintenance of the memory T cell pool. 78

Conventionally, memory T cells have been divided into central memory (TCM) and 79

effector memory (TEM) subsets, and these cells home to secondary lymphoid or 80

peripheral tissues, respectively (6). More recently, a new subset of memory T cells with 81

stem cell–like properties (TSCM) has been identified (7, 8). These cells are the least 82

differentiated of all distinct memory populations, expressing multiple naive markers as 83

well as the memory antigen CD95. Functionally, TSCM cells can generate multiple 84

memory T cell populations, and they possess an enhanced capacity for self-renewal (7). 85

In addition, they are endowed with superior immune reconstitution potential in 86

immunodeficient hosts and can mediate antitumor immunity in a humanized mouse 87

6

model (7). Given the crucial immune properties of TSCM, we evaluated the presence of 88

these cells in HIV-1 infected individuals on and off HAART, and their relationship with 89

viral load, CD4+ count and activation of the immune system. We observed that HAART 90

had a restorative effect on both the frequency and absolute number of cells in the CD8+ 91

TSCM compartment. This could represent an important benefit of treatment, as these 92

cells are thought to be the progenitors of long-term T cell memory in the setting of viral 93

infection (9). We found that the TSCM population was also present in gastrointestinal-94

associated lymphoid tissue (GALT), but at a lower frequency than in the peripheral 95

blood. Furthermore, the frequency of circulating CD8+ TSCM cells in untreated 96

individuals was inversely correlated with T cell activation and viral load, suggesting 97

they may contribute to protection from disease progression in chronic HIV-1 infection. 98

99

Materials and Methods 100

Study Subjects 101

Samples of cryopreserved peripheral blood mononuclear cells (PBMCs) were selected 102

from participants in the San Francisco-based HIV-1-infected SCOPE cohort. Samples 103

from HIV-1-seronegative controls were obtained from 20 donors to the Stanford blood 104

bank. The study was approved by the local Institutional Review Board (University of 105

California San Francisco Committee on Human Research) and individuals gave written 106

informed consent. PBMC samples were obtained from the following categories of HIV-107

1-infected individuals: 29 untreated virologic “controllers” (viral load <2000 HIV-1 108

copies/mL), 27 HAART-suppressed patients (viral load <50-75 HIV-1 copies/mL) and 109

28 untreated “virologic non-controllers” (viral load >10,000 copies/mL). All had CD4+ T 110

7

cell counts >250 cells/mm3. A fourth group of untreated HIV-1-infected patients was 111

also tested. These 29 individuals were defined as “immunologic progressors”, with an 112

HIV-1 viral load >10,000 copies/mL and CD4+ T cell counts <250 cells/mm3. All 113

patients had been diagnosed with HIV-1 at least one year prior to inclusion in this study. 114

See Table 1 for baseline subject characteristics. 115

GALT samples 116

Gut-associated lymphoid tissue (GALT) from rectosigmoid biopsies and paired fresh 117

PBMC samples were obtained from two HIV-1-infected viremic subjects, five HIV-1-118

infected controllers, and three HIV-1-seronegative subjects. Rectal mononuclear cells 119

(RMCs) were isolated from biopsies by collagenase type II digestion at 37°C for one 120

hour, followed by blunt dissection, passage through a 70-μm cell strainer, and washing 121

with R10 media and FACS buffer (PBS with 0.5% bovine serum albumin, 2 mM EDTA) 122

prior to staining with antibodies for flow cytometry analysis. 123

Antibodies and flow cytometry 124

Cryopreserved PBMC were thawed in RPMI 1640 with 10% FBS and washed in FACS 125

buffer. When available, one million fresh lymphocytes from disrupted GALT biopsies 126

and paired PBMC were stained. Phenotypic staining was performed on 106 cells by 127

incubation with a viability marker (AmCyan live-dead kit from Invitrogen) and with 128

antibodies conjugated to CD3, CD4, CD8, CD45RA, CCR7, CD27, CD95, and α4β7 129

(BD Biosciences, San Diego, CA) for 30 minutes on ice. Subsequently, cells were 130

washed, fixed with paraformaldehyde 4% for 5 minutes, washed and acquired with an 131

LSR-II flow cytometer (Becton Dickinson). 132

Statistical Analysis 133

8

Statistical analysis was performed using GraphPad Prism statistical software, version 134

6b (GraphPad Software, San Diego, CA). Non-parametric Kruskal-Wallis and Mann-135

Whitney U tests were used for group comparisons. Dunn’s post-hoc test (which 136

incorporates the Bonferroni adjustment to correct for multiple comparisons) was used 137

for between-group analyses. The Spearman rank test with linear regression was used 138

for correlation analyses. P-values less than 0.05 were considered significant. The 139

Benjamini-Hochberg test was applied to correct for multiple comparisons, with a q-140

value (false discovery rate) threshold of less than 0.05. 141

142

Results 143

Comparison of TSCM distribution in HIV-1-infected subjects 144

In this cross-sectional study of chronically HIV-1 infected subjects we studied 145

the following groups: (1) controllers who naturally suppress HIV-1 with a viral 146

load<2000 copies/ml in the absence of treatment, (2) non-controllers who are untreated 147

and have a viral load>10,000 copies/ml but maintain CD4+ counts >250 cells/mm3, (3) 148

immunological progressors who are also untreated with viral loads>10,000 copies/ml, 149

and have progressed to immunodeficiency with CD4+ counts <250 cells/mm3 and (4) 150

HAART-suppressed individuals with viral loads<75 copies/ml on treatment (see Table 151

1). A cohort of age-matched healthy control subjects was also included. There were no 152

significant differences in the median ages of the five groups (including HIV-1-infected 153

and uninfected subjects) or duration of diagnosis among the HIV-1-infected groups 154

(p>0.05 for all pairwise comparisons). 155

9

We evaluated the relative frequency of the TSCM population within the CD4+ and 156

CD8+ T cell compartments in each of subject groups. The gating strategy to define this 157

subset is shown in Fig. 1. Briefly, singlet cells were defined followed by gating on 158

lymphocytes and live cells. Among the live cells, CD3+ T lymphocytes were identified, 159

followed by the definition of CD4+ and CD8+ subpopulations. Subsequently, the 160

expression of CD45RA/CCR7 was analyzed in the CD4+ and CD8+ T lymphocytes and 161

further gated based on CD27/CD95 expression. Central memory T cells (TCM) are 162

CD45RA-CCR7+CD27+, effector memory T cells (TEM) are CD45RA-CCR7-CD27-, naive 163

T cells are CD45RA+CCR7+CD27+CD95- and TSCM are CD45RA+CCR7+CD27+CD95+. 164

Consistent with previous reports (10), comparison of conventional memory T cell 165

subsets among the healthy controls and HIV-1-infected subjects revealed a skewing 166

towards a terminally differentiated TEM phenotype in the infected groups, particularly in 167

the individuals with the most advanced disease (immunologic progressors) (Fig. 2A 168

and B). Confirming the findings of Gattinoni et al. (7), we observed that the TSCM 169

represent a relatively small percentage of all circulating CD4+ and CD8+ T lymphocytes. 170

In our cohort, healthy uninfected controls had a median of 1.2% CD4+ and 1.5% CD8+ 171

TSCM. Comparing the frequency of this subset among all HIV-1-infected groups we did 172

not observe any difference in the CD4+ T cell compartment (Fig. 2A). However, in the 173

CD8+ T cell compartment, there was a decrease in the frequency of this subpopulation 174

in the untreated groups when compared to the HAART-suppressed group (p<0.001, 175

p<0.0001, and p<0.001 for HAART-suppressed vs. controllers, immunologic 176

progressors, and non-controllers, respectively; Fig. 2B) and healthy control group 177

(p<0.001, p<0.0001, and p<0.001 for healthy controls vs. controllers, immunologic 178

10

progressors, and non-controllers, respectively; Fig. 2B). Controllers had the greatest 179

absolute numbers of CD4+ TSCM among the untreated groups (p<0.0001 and p<0.05, 180

for controllers vs. immunologic progressors and non-controllers, respectively; Fig. 2C), 181

and had a trend towards greater CD4+ TSCM numbers than the HAART-suppressed 182

group. The HAART-suppressed subjects had higher numbers of CD8+ TSCM than the 183

untreated subjects (p<0.01, p<0.0001, p<0.05, for HAART-suppressed vs. controllers, 184

immunologic progressors, and non-controllers respectively; Fig. 2D). Of note, the 185

duration of antiretroviral treatment did not affect TSCM frequencies or absolute numbers 186

(not shown). Among all subjects, the levels of CD4+ and CD8+ TSCM cells, measured by 187

either frequency (r=0.51, p<0.0001) or absolute number (r=0.56, p<0.0001) were 188

positively correlated, indicating that within an individual, the maintenance of these two 189

subsets may be linked (Figs. 2E and 2F respectively). The correlation within each 190

group as well as across all subjects is shown within Fig. 2. 191

TSCM and memory T cell maintenance 192

TSCM have been reported to be important in the homeostasis of the immune 193

system due to their ability to rapidly generate other memory subtypes. In all infected 194

untreated subjects we observed that the frequency of CD4+ TSCM was positively 195

correlated with the frequency of CD4+ central memory (TCM) (r=0.20, p=0.03; Fig. 3A) 196

and inversely correlated with the frequency of CD4+ effector memory (TEM) (r=-0.63, 197

p<0.0001; Fig. 3B), potentially indicating that the TSCM subset supports the TCM 198

population, which may then subsequently differentiate into the peripheral tissue-homing 199

TEM. The same pattern was observed in the CD8+ TSCM compartment (r=0.55, p<0.0001 200

for CD8+ TSCM vs. TCM, Fig. 3C; r=-0.47, p<0.0001 for CD8+ TSCM vs. TEM, Fig. 3D). The 201

11

relationships between TSCM and conventional memory T cell subsets were consistent 202

within each of the subject groups, except for the CD4+ TSCM vs. CD4+ TCM correlation, 203

which was driven primarily by the non-controllers. 204

TSCM, clinical outcome and immune activation 205

Next we investigated whether the TSCM subset was correlated with markers of 206

disease progression in our untreated cohort. The HAART group was excluded given 207

the known effect of antiretroviral treatment on viral replication and immune activation. 208

Across all three untreated groups (controllers, non-controllers and immunologic 209

progressors) we found that the frequency of CD8+ TSCM was inversely correlated with 210

HIV-1 RNA levels (r=-0.25, p=0.02; Fig. 4F) and positively correlated with CD4+ T cell 211

count (r=0.24, p=0.03; Fig. 4N). The positive correlation with CD4+ count was largely 212

driven by the controller group (r=0.54, p=0.003), while the inverse relationship between 213

TSCM and viral load was found within all three groups. This relationship was also 214

observed for the CD8+ naïve (Fig. 4E and 4M) and TCM (Fig. 4G and 4O) subsets, 215

whereas the frequency of TEM was correlated with higher HIV-1 viral loads (Figs. 4D, 216

4H) and lower CD4+ T cell counts (Figs. 4L and 4P). CD4+ TSCM were not correlated 217

with HIV-1 RNA levels (r=-0.10, p=0.37; Fig. 4B) or CD4+ T cell count (r=0.05, p=064; 218

Fig. 4J). 219

We then sought to determine the relationship between TSCM and the level of 220

CD4+ and CD8+ T cell activation based on the expression of CD38 and HLA-DR. As 221

expected, the immunologic progressors had higher frequencies of activated CD4+ and 222

CD8+ T cells than the HAART-suppressed and controller groups and similar levels 223

compared to the non-controller group (data not shown). The frequency of the CD8+ 224

12

TSCM subpopulation was inversely correlated with the frequency of CD8+CD38+ T cells 225

across all subjects (r=-0.25, p<0.05; Fig. 5F) as well as within each group. The same 226

inverse correlation was observed when considering CD8+ Tnaive and TCM cells (p<0.01 227

and p<0.0001, respectively; Figs. 5E and 5G). In contrast, the presence of TEM cells 228

was positively correlated with %CD8+CD38+ T cells (Figs. 5D and 5H), which is similar 229

to the findings in another recent study (11). The only significant relationship between 230

CD38 and HLA-DR co-expressing populations and memory subsets was between the 231

frequency of CD8+ TCM and %CD4+CD38+HLA-DR+ (r=-0.30, p=0.006; not shown). 232

233

TSCM in gastrointestinal-associated lymphoid tissue (GALT) 234

In a smaller cohort of HIV-infected and uninfected individuals, we interrogated 235

the TSCM subset in the GALT, as this is a primary site of HIV-1 replication and CD4+ 236

depletion. To our knowledge, this cell population has not previously been studied in the 237

mucosal compartment of humans. Consistent with their naïve T cell-like phenotype, we 238

found that both CD4+ and CD8+ TSCM are more prevalent in peripheral blood than in 239

GALT (Fig. 6A). However, the frequency of CD4+ TSCM cells in GALT was inversely 240

associated with peripheral CD4+ TSCM (r=-0.93, p<0.01), suggesting the potential 241

migration of this population to the gut mucosa (Fig. 6B). Interestingly, in the total cohort 242

of all healthy and infected subjects, about 34% of CD4+ and 30% of peripheral CD8+ 243

TSCM expressed the gut homing marker α4β7, close to twice that of circulating TEM, 244

which more frequently localize to peripheral tissues (not shown). 245

246

247

13

Discussion 248

We report here the distribution of the multi-potent memory stem T cell (TSCM) 249

subset in chronically HIV-1 infected subjects with variable disease control. Furthermore, 250

we describe for the first time the relationship of TSCM with key parameters linked to HIV-251

1 pathogenesis. In summary, we observed that untreated HIV-1-infected subjects had 252

lower frequencies of circulating CD8+ TSCM cells than individuals on antiretroviral 253

treatment. This subset was directly correlated with CD4+ T cell count and inversely 254

correlated with viral load in the untreated groups. Furthermore, higher levels of the 255

CD8+ TSCM subpopulations were associated with lower levels of CD8+ T cell activation, 256

suggesting an association with a more functional immune system. TSCM were also 257

present in gastrointestinal lymphoid tissue, but to a lesser extent than in the peripheral 258

circulation, and the frequency of CD4+ TSCM in these two sites was inversely correlated. 259

Although our study design precludes the demonstration of causality, our data are 260

broadly consistent with a model in which preservation of the CD8+ TSCM subset 261

contributes to improved control of HIV-1, which in turn contributes to maintenance of 262

immune function and lower levels of T cell activation. Our data also suggest that 263

antiretroviral treatment is associated with a frequency of these cells closer to that of 264

healthy individuals, which may explain in part the beneficial effect therapy has on 265

immune recovery and function. 266

It has been reported that one mechanism for maintenance of long-term T cell 267

memory derives from the unique homeostatic properties of TSCM cells (9). The superior 268

persistence of TSCM cells following antigen loss suggests that they are the main 269

precursors of T cell memory in the post-antigen phase. In this context, we observed 270

14

that CD4+ and CD8+ TSCM cells were positively correlated with the TCM and inversely 271

correlated with the TEM cell compartments, suggesting that the differentiation of TSCM 272

leads to a shift in the proportion of these cell populations as a possible mechanism to 273

re-establish the homeostasis of the immune system. 274

The role of CD8+ T cells in virus control and consequent maintenance of CD4+ T 275

cell counts is well described (12, 13). Herein, we observed that CD8+ TSCM cells were 276

inversely correlated with viral load and positively correlated with CD4+ count, indicating 277

that a high percentage of this subset is associated with a good prognosis. Of note, the 278

naïve and TCM subpopulations were also associated with less advanced disease in this 279

cohort. As previous studies have shown that TSCM can generate TCM and are 280

preferentially maintained compared to TCM following escape from cognate antigen in a 281

model of chronic SIV infection (9), HIV-1-specific CD8+ TSCM may be an optimal target 282

for therapeutic vaccine induction. In contrast, the CD4+ TSCM population was not 283

correlated with any measured disease parameters. Interestingly, the CD4+ TSCM subset 284

was recently described as being permissive to HIV-1 infection in vitro (14), and also 285

disproportionately contributing to a long lasting latent reservoir (15). However, we 286

found that virologic controllers had preserved levels of CD4+ TSCM. A recent study (16) 287

reported that another rare subset of HIV-infected individuals known as viremic non-288

progressors (VNPs) also maintain CD4+ TSCM counts, which is associated with 289

decreased HIV infection of these cells. The mechanism of TSCM resistance to infection 290

in VNPs and potentially the classic controllers in our study remains to be determined. 291

It is well known that HIV-1 infection is associated with systemic immune 292

activation, in part through bacterial translocation after injury of the epithelial barrier of 293

15

the gut, as well as other mechanisms (17). This activation leads to immune exhaustion, 294

increased non-AIDS related comorbidities such as cardiovascular disease, and 295

eventually progression to AIDS (18). Here we found that in untreated subjects, the 296

CD8+ TSCM subset was correlated with lower frequencies of activated CD8+ T cells. The 297

mechanism for a possible protective effect on immune activation by CD8+ TSCM is not 298

known, but may be linked to the maintenance of gut barrier function or through the 299

provision of a stable precursor pool of antigen-specific memory T cells. 300

In a recent study of healthy nonhuman primates, Lugli et al. (9) showed that 301

TSCM cells have a tropism for secondary lymphoid tissues, with a distribution most 302

similar to naive T (TN) cells, while central and effector memory T cells predominate in 303

mucosal surfaces. In the present work, we observed a median of 0.8 and 0.4% of CD4+ 304

TSCM and CD8+ TSCM cells, respectively, in GALT samples obtained from infected and 305

healthy subjects. This corroborates the findings of Lugli et al (9), as we found a median 306

of 3.2% of CD4+ and 2% of CD8+ TSCM in the peripheral blood. We also observed an 307

inverse correlation between the CD4+ TSCM subsets in PBMC and GALT, indicating a 308

possible migration of these cells to the gut, the main site of HIV-1 infection and 309

replication. In support of this, we found that there are high frequencies of CD4+ TSCM 310

expressing the gut-homing marker α4β7 in the circulation. Further longitudinal and 311

cross-sectional studies of TSCM in human and animal models within the context of viral 312

infection will be useful in defining the function of this cell type distributed across various 313

tissues. 314

Our study has a number of limitations. With a cross-sectional design, we are 315

unable to demonstrate causality. While we suggest here that preservation of CD8+ 316

16

TSCM cells is causally associated with improved immune function and virus control and 317

that the mechanism may be related to homeostatic maintenance of other memory T cell 318

subsets, it is also possible that lack of disease progression (defined by stable CD4+ T 319

cell counts and/or low viral load) contributes to higher levels of TSCM cells. Longitudinal 320

studies in humans should help untangle these associations, although the only definitive 321

study to demonstrate the causal role of TSCM cells in disease control would be to 322

therapeutically increase their number in a controlled manner. It is also possible that the 323

function rather than the number of TSCM might prove to be the most important 324

prognostic characteristic. Although we lacked sufficient cells to characterize the HIV-1-325

specific TSCM cell population, experiments in a non-human primate model of SIV 326

infection have demonstrated that SIV-specific TSCM preferentially survive after antigen 327

elimination compared to other memory subsets, and are fully functional even in chronic 328

infection (9). Therefore, HIV-1-specific CD8+ TSCM cells can presumably directly 329

contribute to HIV-1 control and should be investigated in future studies. 330

331

Conclusions 332

In summary, this study describes the distribution of stem cell-like memory T cells 333

in HIV-1-infected humans and identifies CD8+ TSCM as a correlate of protection from 334

disease progression. Our findings suggest an important role for TSCM in supporting 335

durable immunity in vivo, and are therefore directly relevant to the design of T cell-336

based vaccines, adoptive immunotherapy approaches or the pharmacologic induction 337

of TSCM. 338

339

17

List of abbreviations 340

TSCM: Stem memory T cell 341

TEM: Effector memory T cell 342

TCM: Central memory T cell 343

HAART: highly active antiretroviral therapy 344

GALT: gut-associated lymphoid tissue 345

SIV: simian immunodeficiency virus 346

347

Competing interests 348

The authors do not have a commercial or other association that might pose a conflict of 349

interest. 350

351

Acknowledgments 352

This work was supported by the Delaney AIDS Research Enterprise (DARE; AI096109), 353

NIAID (K24 AI069994), the UCSF/Gladstone Institute of Virology & Immunology CFAR 354

(P30 AI027763), the UCSF Clinical and Translational Research Institute Clinical 355

Research Center (UL1 RR024131), the Center for AIDS Prevention Studies (P30 356

MH62246), and the CFAR Network of Integrated Systems (R24 AI067039). MS was 357

supported by NIH NCI K23 CA157929. DS was supported by NIH NIAID K08 A120071. 358

This research was supported by the Brazilian Council for Scientific and Technological 359

Development (CNPq) and the São Paulo State Research Funding Agency (FAPESP). 360

Fundação de Amparo a Pesquisa do Estado de São Paulo (2010/05845-0/EGK/DFN), 361

CNPq/CAPES 056/2012 (DFN). ECN and JK are recipients of productivity awards from 362

18

the Brazilian Council for Scientific and Technological Development (CNPq). The 363

funders had no role in study design, data collection and analysis, decision to publish, or 364

preparation of the manuscript. 365

We would also like to thank Miguel de Mulder, Henri-Alexandre Michaud and André 366

Raposo for helpful comments and assistance with a subset of the GALT experiments. 367

368

19

References 369

1. Grelli S, D'Ettore G, Lauria F, Montella F, Di Traglia L, D'Agostini C, Lichtner M, 370

Vullo V, Favalli C, Vella S, Macchi B, Mastino A. 2003. CD4+ lymphocyte increases 371

in HIV patients during potent antiretroviral therapy are dependent on inhibition of CD8+ 372

cell apoptosis. Annals of the New York Academy of Sciences 1010:560-564. 373

2. Hunt PW. 2012. HIV and inflammation: mechanisms and consequences. Curr 374

HIV/AIDS Rep 9:139-147. 375

3. Picker LJ, Hagen SI, Lum R, Reed-Inderbitzin EF, Daly LM, Sylwester AW, Walker 376

JM, Siess DC, Piatak M, Jr., Wang C, Allison DB, Maino VC, Lifson JD, Kodama T, 377

Axthelm MK. 2004. Insufficient production and tissue delivery of CD4+ memory T cells 378

in rapidly progressive simian immunodeficiency virus infection. J Exp Med 200:1299-379

1314. 380

4. Douek DC, Brenchley JM, Betts MR, Ambrozak DR, Hill BJ, Okamoto Y, Casazza 381

JP, Kuruppu J, Kunstman K, Wolinsky S, Grossman Z, Dybul M, Oxenius A, Price 382

DA, Connors M, Koup RA. 2002. HIV preferentially infects HIV-specific CD4+ T cells. 383

Nature 417:95-98. 384

5. Koelsch KK, Liu L, Haubrich R, May S, Havlir D, Gunthard HF, Ignacio CC, 385

Campos-Soto P, Little SJ, Shafer R, Robbins GK, D'Aquila RT, Kawano Y, Young 386

K, Dao P, Spina CA, Richman DD, Wong JK. 2008. Dynamics of total, linear 387

nonintegrated, and integrated HIV-1 DNA in vivo and in vitro. J Infect Dis 197:411-419. 388

6. Sallusto F, Lenig D, Forster R, Lipp M, Lanzavecchia A. 1999. Two subsets of 389

memory T lymphocytes with distinct homing potentials and effector functions. Nature 390

401:708-712. 391

7. Gattinoni L, Lugli E, Ji Y, Pos Z, Paulos CM, Quigley MF, Almeida JR, Gostick E, 392

Yu Z, Carpenito C, Wang E, Douek DC, Price DA, June CH, Marincola FM, 393

20

Roederer M, Restifo NP. 2011. A human memory T cell subset with stem cell-like 394

properties. Nat Med 17:1290-1297. 395

8. Allen TM, Altfeld M, Yu XG, O'Sullivan KM, Lichterfeld M, Le Gall S, John M, 396

Mothe BR, Lee PK, Kalife ET, Cohen DE, Freedberg KA, Strick DA, Johnston MN, 397

Sette A, Rosenberg ES, Mallal SA, Goulder PJ, Brander C, Walker BD. 2004. 398

Selection, transmission, and reversion of an antigen-processing cytotoxic T-lymphocyte 399

escape mutation in human immunodeficiency virus type 1 infection. J Virol 78:7069-400

7078. 401

9. Lugli E, Dominguez MH, Gattinoni L, Chattopadhyay PK, Bolton DL, Song K, Klatt 402

NR, Brenchley JM, Vaccari M, Gostick E, Price DA, Waldmann TA, Restifo NP, 403

Franchini G, Roederer M. 2013. Superior T memory stem cell persistence supports 404

long-lived T cell memory. J Clin Invest 123:594-599. 405

10. Emu B, Moretto WJ, Hoh R, Krone M, Martin JN, Nixon DF, Deeks SG, McCune JM. 406

2014. Composition and function of T cell subpopulations are slow to change despite 407

effective antiretroviral treatment of HIV disease. PLoS One 9:e85613. 408

11. Ghiglione Y, Falivene J, Ruiz MJ, Laufer N, Socias ME, Cahn P, Giavedoni L, Sued 409

O, Gherardi MM, Salomon H, Turk G. 2014. Early Skewed Distribution of Total and 410

HIV-Specific CD8+ T-Cell Memory Phenotypes during Primary HIV Infection Is Related 411

to Reduced Antiviral Activity and Faster Disease Progression. PLoS One 9:e104235. 412

12. Walker BD, Chakrabarti S, Moss B, Paradis TJ, Flynn T, Durno AG, Blumberg RS, 413

Kaplan JC, Hirsch MS, Schooley RT. 1987. HIV-specific cytotoxic T lymphocytes in 414

seropositive individuals. Nature 328:345-348. 415

13. Betts MR, Nason MC, West SM, De Rosa SC, Migueles SA, Abraham J, Lederman 416

MM, Benito JM, Goepfert PA, Connors M, Roederer M, Koup RA. 2006. HIV 417

nonprogressors preferentially maintain highly functional HIV-specific CD8+ T cells. 418

Blood 107:4781-4789. 419

21

14. Tabler CO, Lucera MB, Haqqani AA, McDonald DJ, Migueles SA, Connors M, 420

Tilton JC. 2014. CD4+ Memory Stem Cells Are Infected by HIV-1 in a Manner 421

Regulated in Part by SAMHD1 Expression. J Virol 88:4976-4986. 422

15. Buzon MJ, Sun H, Li C, Shaw A, Seiss K, Ouyang Z, Martin-Gayo E, Leng J, 423

Henrich TJ, Li JZ, Pereyra F, Zurakowski R, Walker BD, Rosenberg ES, Yu XG, 424

Lichterfeld M. 2014. HIV-1 persistence in CD4+ T cells with stem cell-like properties. 425

Nat Med 20:139-142. 426

16. Klatt NR, Bosinger SE, Peck M, Richert-Spuhler LE, Heigele A, Gile JP, Patel N, 427

Taaffe J, Julg B, Camerini D, Torti C, Martin JN, Deeks SG, Sinclair E, Hecht FM, 428

Lederman MM, Paiardini M, Kirchhoff F, Brenchley JM, Hunt PW, Silvestri G. 2014. 429

Limited HIV Infection of Central Memory and Stem Cell Memory CD4+ T Cells Is 430

Associated with Lack of Progression in Viremic Individuals. PLoS Pathog 10:e1004345. 431

17. Klatt NR, Funderburg NT, Brenchley JM. 2013. Microbial translocation, immune 432

activation, and HIV disease. Trends in microbiology 21:6-13. 433

18. Giorgi JV, Hultin LE, McKeating JA, Johnson TD, Owens B, Jacobson LP, Shih R, 434

Lewis J, Wiley DJ, Phair JP, Wolinsky SM, Detels R. 1999. Shorter survival in 435

advanced human immunodeficiency virus type 1 infection is more closely associated 436

with T lymphocyte activation than with plasma virus burden or virus chemokine 437

coreceptor usage. J Infect Dis 179:859-870. 438

439

440

441

22

Table 1. Clinical data for HIV-1-infected subjects. 442

CD4+ count

(cells/mm3)

Median (IQR)

HIV-1 Viral load

(copies/ml)

Median (IQR)

Age

(years)

Median (IQR)

HAART-suppressed 674 (534-981) 50 (0-75) 45 (38-54)

Controllers 813 (585-1260) 96 (11-489) 47.5 (39-54)

Non-controllers 576 (446-730) 35,650 (24,586-52675) 43 (34-49)

Immunologic

Progressors

228 (175-296) 71,585 (29,925-191,067) 41.5 (36-49)

IQR= interquartile range 443

444

23

Figure legends 445

Figure 1. Gating strategy for TSCM. Lymphocytes were stained and acquired as 446

described in Methods. Cells were gated as follows: (1) FSC-A x FSC-H for singlets, (2) 447

FSC-A x SSC-A for lymphocytes, (3) live-dead (AmCyan negative) for live cells, (4) 448

CD3+ for T lymphocytes, (5) CD4+ or CD8+, (6) CD45RA and CCR7 for conventional 449

memory subsets and (7) CD27 and CD95 for TSCM vs. naïve T cells. Central memory T 450

cells (TCM): CD45RA-CCR7+CD27+, effector memory T cells (TEM): CD45RA-CCR7-451

CD27-, naïve T cells: CD45RA+CCR7+CD27+CD95- and TSCM: 452

CD45RA+CCR7+CD27+CD95+. 453

454

Figure 2. TSCM prevalence in HIV-1-infected subjects. (A) Percentages of CD4+ and 455

CD8+ (B) naïve and memory T cell subsets across uninfected and infected subject 456

groups. (C) Absolute numbers (cells/mm3) of CD4+ and CD8+ (D) memory T cells 457

across infected subject groups. Non-parametric Kruskal-Wallis and Dunn’s post-hoc 458

tests for multiple comparisons were used. Correlation Spearman rank test between the 459

percentages (E) as well absolute numbers (F) of CD4+ and CD8+ TSCM are depicted. 460

*p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. Healthy controls (HC); HAART-461

suppressed (HS ); controllers (C *); immunologic progressors (IP Δ); non-controllers 462

(NC ). Individual r and p-values are shown for each subject group within the legends 463

in (E) and (F). 464

465

Figure 3. TSCM are the progenitors for TEM cells and help the maintenance of TCM 466

cells. Correlation Spearman rank tests between frequencies of CD4+ TSCM and CD4+ 467

24

central memory cells (TCM) (A) and CD4+ effector memory cells (TEM) (B); between 468

frequencies of CD8+ TSCM and CD8+ central memory cells (TCM) (C) and CD8+ effector 469

memory cells (TEM) (D); are shown for untreated subjects. Controllers (C *); 470

immunologic progressors (IP Δ); non-controllers (NC ). 471

472

Figure 4. CD8+ TSCM cells are correlated with low HIV-1 viral load (VL) and high 473

CD4+ T cell count. Correlation Spearman rank tests between the percentages of CD4+ 474

(A-D) and CD8+ (E-H) memory T cell subsets and HIV-1 (log10) viral load, and between 475

of CD4+ (I-L) and CD8+ (M-P) memory T cell subsets and CD4+ count. Untreated 476

subjects are depicted. Controllers (C *); immunologic progressors (IP Δ); non-477

controllers (NC ). 478

479

Figure 5. CD8+ TSCM cells are inversely correlated with T cell activation in 480

untreated subjects. Correlation Spearman rank tests between the frequencies of 481

CD4+ (A-D) and CD8+ (E-H) memory T cell subsets and %CD8+CD38+. Untreated 482

subjects are depicted. Controllers (C *); immunologic progressors (IP Δ); non-483

controllers (NC ). 484

485

Figure 6. Prevalence of TSCM cells in peripheral blood and GALT. (A) Percentage of 486

CD4+ TSCM and CD8+ TSCM in peripheral blood (PBMC) and GALT. (B) Correlation 487

Spearman rank test between percentage of CD4+ TSCM cells in GALT and PBMC. Data 488

from untreated HIV-1-infected and healthy subjects are shown. 489