Mycobacterium tuberculosis-specific CD4+ and CD8+ T cells differ in their capacity to recognize infected macrophages

Containment of Mycobacterium tuberculosis (Mtb) infection requires T cell recognition of infected macrophages. Mtb has evolved to tolerate, evade, and subvert host immunity. Despite a vigorous and sustained CD8+ T cell response during Mtb infection, CD8+ T cells make limited contribution to protection. Here, we ask whether the ability of Mtb-specific T cells to restrict Mtb growth is related to their capacity to recognize Mtb-infected macrophages. We derived CD8+ T cell lines that recognized the Mtb immunodominant epitope TB10.44−11 and compared them to CD4+ T cell lines that recognized Ag85b240-254 or ESAT63-17. While the CD4+ T cells recognized Mtb-infected macrophages and inhibited Mtb growth in vitro, the TB10.4-specific CD8+ T cells neither recognized Mtb-infected macrophages nor restricted Mtb growth. TB10.4-specific CD8+ T cells recognized macrophages infected with Listeria monocytogenes expressing TB10.4. However, over-expression of TB10.4 in Mtb did not confer recognition by TB10.4-specific CD8+ T cells. CD8+ T cells recognized macrophages pulsed with irradiated Mtb, indicating that macrophages can efficiently cross-present the TB10.4 protein and raising the possibility that viable bacilli might suppress cross-presentation. Importantly, polyclonal CD8+ T cells specific for Mtb antigens other than TB10.4 recognized Mtb-infected macrophages in a MHC-restricted manner. As TB10.4 elicits a dominant CD8+ T cell response that poorly recognizes Mtb-infected macrophages, we propose that TB10.4 acts as a decoy antigen. Moreover, it appears that this response overshadows subdominant CD8+ T cell response that can recognize Mtb-infected macrophages. The ability of Mtb to subvert the CD8+ T cell response may explain why CD8+ T cells make a disproportionately small contribution to host defense compared to CD4+ T cells. The selection of Mtb antigens for vaccines has focused on antigens that generate immunodominant responses. We propose that establishing whether vaccine-elicited, Mtb-specific T cells recognize Mtb-infected macrophages could be a useful criterion for preclinical vaccine development.

contribution to protection. Here, we ask whether the ability of Mtb-specific T cells to restrict Mtb 23 growth is related to their capacity to recognize Mtb-infected macrophages. 24 We derived CD8 + T cell lines that recognized the Mtb immunodominant epitope TB10. 44-25 11 and compared them to CD4 + T cell lines that recognized Ag85b240-254 or ESAT63-17. While the 26 CD4 + T cells recognized Mtb-infected macrophages and inhibited Mtb growth in vitro, the TB10.4-27 specific CD8 + T cells neither recognized Mtb-infected macrophages nor restricted Mtb growth. 28 TB10.4-specific CD8 + T cells recognized macrophages infected with Listeria monocytogenes 29 expressing TB10.4. However, over-expression of TB10.4 in Mtb did not confer recognition by 30 TB10.4-specific CD8 + T cells. Importantly, CD8 + T cells recognized macrophages pulsed with 31 irradiated Mtb, indicating that macrophages can efficiently cross-present the TB10.4 protein and 32 raising the possibility that viable bacilli might suppress cross-presentation. Importantly, polyclonal 33 CD8 + T cells specific for Mtb antigens other than TB10.4 recognized Mtb-infected macrophages 34 in a MHC-restricted manner. 35 As TB10.4 elicits a dominant CD8 + T cell response that poorly recognizes Mtb-infected 36 macrophages, we propose that TB10.4 acts as a decoy antigen. Moreover, it appears that this 37 response overshadows subdominant CD8 + T cell response that can recognize Mtb-infected 38 macrophages. The ability of Mtb to subvert the CD8 + T cell response may explain why CD8 + T 39 cells make a disproportionately small contribution to host defense compared to CD4 + T cells. The 40 Introduction 44 Unlike most disease-causing pathogens, Mycobacterium tuberculosis (Mtb), the cause of 45 tuberculosis (TB), persists in humans because of its highly evolved ability to evade and subvert 46 the host immunity [1]. Mtb subverts vesicular trafficking, prevents phagolysosome fusion, and 47 replicates in an intracellular niche within macrophages, allowing it to evade detection by humoral 48 immunity [2]. Mtb also delays the initiation and recruitment of T cell immunity to the lung, 49 promoting the establishment of a persistent infection [3]. Despite these challenges, T cell 50 immunity does occur and plays an essential role in controlling the infection in both mice and 51 humans [3][4][5]. With 10 million new TB cases annually, an effective vaccine would offer a cost-52 effective way to prevent TB and attenuate this persistent global pandemic. Given the importance 53 of T cells during host defense, strategies for TB vaccines largely aim at generating memory T 54 cells rather than neutralizing antibodies. Most subunit vaccines incorporate immunodominant Mtb 55 antigens, which elicit large T cell responses [6]. Several immunodominant antigens have been 56 identified in the murine TB model, including Ag85a, Ag85b, CFP-10, . T 57 cell responses to these antigens are also frequently detectable in Mtb-infected people, and these 58 highly prevalent responses represent the basis for TB immunodiagnostic tests [8]. By 59 incorporating these immunodominant antigens into vaccines, the expectation is that antigen-60 specific T cells will contain the infection before Mtb can establish a niche and evade host immunity 61 9 or 5 post-infection. Again, we did not observe any increase in Nur77 or CD69 expression (Fig 3e,  189 f). As a control for T cell health and function, we co-cultured TB10. 44-11-peptide-pulsed-, 190 uninfected-macrophages with the TB10Rg3 T cells and observed significant increases in their 191 Nur77 and CD69 expression (Fig 3). 192 Despite assessing recognition on multiple days, we considered whether the short assay 193 period (i.e. 2 hours) might not detect recognition of Mtb-infected macrophages by TB10Rg3 T 194 cells, especially if presentation of TB10.4 is inefficient or asynchronous. Therefore, we used IFNγ 195 production as a cumulative indicator of T cell activation during a 72-hour co-culture experiment. 196 Since cytokine-driven activation (e.g.,  can stimulate IFNγ production by T cells 197 independently of TCR signaling, we used MHC-matched (H-2 b ) or mismatched (H-2 k ) TGPM to 198 assess cognate recognition [26,[28][29][30]. As the infectious dose (MOI, multiplicity of infection) 199 increased, P25 T cells produced more IFNγ when co-cultured with MHC-matched, but not  mismatched, Mtb-infected TGPMs (Fig 3g). In contrast, TB10Rg3 T cells did not produce IFNγ 201 when co-cultured with Mtb-infected TGPMs (Fig 3h). As before, TB10Rg3 T cells produced IFNγ 202 when co-cultured with uninfected macrophages pulsed with the TB10.44-11 peptide (Fig 3h). 203 We next used the TB10.44-11-specific CD8 + T cell lines TB10RgP, TB10RgL and TB10RgR 204 to address whether TB10.44-11-specific CD8 + T cells other than TB10Rg3 can recognize Mtb-205 infected macrophages. While the TB10RgP, TB10RgL, and TB10RgR CD8 + T cell lines produced 206 IFNγ when cultured with uninfected macrophages pulsed with the TB10.44-11 peptide, none 207 produced IFNγ following a 72-hour co-culture with Mtb-infected macrophages (Fig 3i). These data 208 show that, regardless of the time point of T cell addition or the length of co-culture, P25 T cells, 209 but not TB10.44-11-specific CD8 + T cells, recognized Mtb-infected macrophages, based on their 210 increased Nur77 and CD69 expression as well as their IFNγ production. 211 212 TB10.4-specific CD8 + T cells do not recognize lung cells from Mtb-infected mice.

213
During in vivo infection, Mtb infects a variety of myeloid cells, and this diversity changes 214 10 over the course of the infection [31][32][33]. We considered that lung myeloid cells from Mtb-infected  215   mice are more physiologically relevant than TGPMs. Thus, we isolated MHC class II + lung cells  216 from Erdman-infected, RAG-1-deficient mice 4 weeks post-infection and tested their ability to 217 present Mtb antigens to TB10Rg3 T cells. We used RAG-1-deficient mice because of the 218 possibility that CD8 + T cells in the lungs of Mtb-infected, wild type mice may recognize and 219 eliminate any lung cells presenting the TB10.4 antigen. Since Mtb downregulates Ag85b 220 expression by 3 weeks post infection [11,12], we used an ESAT-6-specific CD4 + T cell line 221 derived from C7 transgenic mice, which we refer to as C7 T cells [10,34]. The immunodominant 222 antigen ESAT-6 retains high levels of expression throughout infection and elicits a dominant CD4 + 223 T cell response in C57BL/6 mice [11]. Due to the difficulty in obtaining large numbers of MHC 224 class II + cells from uninfected, RAG-1-deficient mice, we used TGPMs from age-matched, RAG-225 1-deficient mice as a source of uninfected, inflammatory macrophages. We stained C7 or 226 TB10Rg3 T cells with 5uM of the proliferation dye eFluor450 (eBioscience) before co-culturing 227 them with the lung myeloid cells. After 72 hours, we measured the T cell proliferation as a marker 228 of T cell recognition. C7 T cells proliferated extensively when co-cultured with the infected lung 229 myeloid cells but not when co-cultured with uninfected TGPMs (Fig 4a, b). In contrast, TB10Rg3 230 T cells did not proliferate when co-cultured with the lung myeloid cells (Fig 4c, d). To assess 231 whether TB10Rg3 T cells could proliferate if TB10.4 was present, we pulsed the lung APCs with 232 the TB10.44-11 peptide for 1 hour before adding the TB10Rg3 T cells. As predicted, TB10Rg3 T 233 cells proliferated after 72 hours of co-culture with peptide-pulsed, lung myeloid cells (Fig 4c, d). 234 We considered the possibility that Mtb in lung myeloid cells may not grow well in vitro, 235 leading to altered antigen abundance that could affect T cell recognition. To address this 236 possibility, we measured the bacterial burden in the lung myeloid cells. There was a 3-fold 237 increase in the bacterial numbers between the beginning (d1) and the end (d4) of the experiment, 238 indicating that the bacteria remained viable (Fig 4e) (Fig 5a-c). Bafilomycin, which inhibits vacuolar acidification and impairs the entry of the 259 DActA.TB10.4 strain into the cytosol, diminished the frequency of Nur77-expressing cells 260 (p<0.005) and Nur77 MFI (p<0.01) (Fig 5a, top induction for 24 hours, the esxG and esxH mRNA expression increased multiple folds (Fig 6a). 283 Prior to in vitro infection, we treated esxGH-OE.Mtb with or without tetracycline. The next day, 284 TGPMs were infected with induced or uninduced esxGH-OE.Mtb. P25 T cells produced similar 285 amounts of IFNγ when co-cultured with macrophages infected with either uninduced or induced 286 esxGH-OE.Mtb, which was expected since Ag85b expression should not be altered (Fig 6b). 287 Despite increasing the production of TB10.4 by Mtb, TB10Rg3 T cells still did not recognize Mtb-288 infected macrophages (Fig 6c). Although we cannot be certain that the induction of EsxGH leads 289 to an increased amount of antigen delivered to the antigen processing pathway, this result 290 13 suggests that antigen abundance is not limiting TB10. TGPMs did not achieve the same peak levels (Fig 6d, e; dotted lines). As expected, the regulation 306 of MHC class II was more sensitive to IFNγ. Uninfected TGPMs expressed low baseline levels of 307 MHC class II (Fig 6f, g; solid lines). IFNγ pretreatment resulted in a >100-fold increase in MHC 308 class II median fluorescence intensity (MFI) in the uninfected TGPMs, which peaked on day 3 309 with a >2000-fold increase over the baseline (Fig 6f, g; dotted lines). Mtb-infection alone did not 310 significantly affect MHC class II expression, and consistent with previous studies, Mtb significantly 311 impaired the induction of MHC class II by IFNγ pretreatment (Fig 6f, g). These data show that, in 312 our in vitro infection model where the TGPMs were unstimulated, Mtb infection did not inhibit class 313 I and II MHC expression. Importantly, the differences in MHC class I or class II expression by 314 14 infected macrophages. 316 317

318
T cells. 319 Next, we hypothesized that Mtb may interfere with MHC class I presentation of 320 mycobacterial antigens. Therefore, we tested the ability of the P25 and TB10Rg3 T cell lines to 321 recognize TGPMs cultured with γ-irradiated, nonviable Mtb. Activation of pattern recognition 322 receptors such as TLR2 and TLR4 by large amounts of dead bacteria might induce large amounts 323 of IL-12 and IL-18, resulting in cytokine-driven T cell activation. Taking this concern into 324 consideration, we used MHC-mismatched TGPMs as a control. We pulsed macrophages with a 325 dose titration of γ-irradiated Mtb, then added TB10Rg3 or P25 T cells, and measured IFNγ 326 secretion by the T cells after 72 hours. Both P25 and TB10Rg3 T cells produced high amounts of 327 IFNγ when cultured with MHC-matched (i.e., H-2 b ) but not with MHC-mismatched (i.e., H-2 k ) 328 TGPMs, and this response was dose dependent (Fig 6h,

334
Along with the previous finding that TB10.44-11-specific CD8 + T cells make up ~40% of total 335 lung CD8 + T cells during infection (S1 Figure) [19], our finding that TB10Rg3 T cells do not 336 recognize Mtb-infected macrophages suggests that TB10.4 may be a decoy antigen. This raises 337 the question whether the inability to recognize Mtb-infected macrophages is a general feature of 338 the CD8 + T cell response to Mtb, or if this is a unique feature of TB10.4-specific CD8 + T cells. 339 Therefore, we determined whether polyclonal CD8 + T cells from the lungs of infected mice could 340 15 recognize Mtb-infected macrophages. We carried out aerosol infection of C57BL/6 mice with 341 Erdman, and, 6-8 weeks post infection, we purified polyclonal CD4 + or CD8 + T cells from their 342 lungs and co-cultured them with Mtb-infected macrophages. After 72 hours of co-culture, 343 polyclonal CD4 + T cells produced high amounts of IFNγ in a MHC-restricted manner (Fig 7a). 344 Interestingly, polyclonal CD8 + T cells also produced IFNγ in a MHC-restricted manner when co-345 cultured with Mtb-infected macrophages (Fig 7b). These results indicate that other antigen-346 specific CD8 + T cells recognizing Mtb-infected macrophages do exist, and infected TGPMs can 347 present Mtb antigens to CD8 + T cells. However, based on the high abundance of TB10.4-specific 348 CD8 + T cells post infection (S1 Figure), the non-TB10.4-specific, Mtb-specific CD8 + T cells may 349 be dwarfed by the dominant TB10.4-specific CD8 + T cells. 350 To better assess whether the IFNγ production by polyclonal CD8 + T cells arose 351 predominantly from non-TB10.4-specific CD8 + T cells, we used the TB10.44-11-tetramer to 352 separate TB10.4-specific and non-TB10.4-specific, polyclonal CD8 + T cells from the lungs of 353 infected mice. After 72-hour co-culture with Mtb-infected macrophages, TB10.44-11-tetramer 354 negative CD8 + (non-TB10.4-specific CD8 + ) T cells produced significantly higher IFNγ compared 355 to that of uninfected control (p<0.005), and the production was MHC class I restricted (Fig 7c). In 356 contrast, TB10. 44-11-specific CD8 + T cells produced IFNγ in a non-MHC-restricted manner during 357 co-culture with both uninfected and Mtb-infected macrophages (Fig 7d). We cannot exclude the 358 possibility that the tetramer isolation might have inadvertently activated the TB10.44-11-specific 359 CD8 + T cells. Nevertheless, these data show that polyclonal, TB10.44-11-tetramer negative CD8 + 360 T cells recognized Mtb-infected macrophages, supporting the notion of a subdominant T cell 361 response that may be effective at detecting Mtb. 362

363
A complexity in defining T cell recognition is distinguishing cognate from non-cognate 364 recognition. T cell IFNγ production, a common readout for recognition, can be stimulated by IL-365 12 and IL-18, two cytokines secreted by Mtb-infected cells [26,[28][29][30]. Even cognate recognition 366 does not always signify recognition of infected cells. Uninfected macrophages and dendritic cells 367 (DCs) can acquire exosomes, soluble proteins, apoptotic vesicles or necrotic debris containing 368 non-viable bacilli or its antigens, and present these to T cells [13,[38][39][40]. This detour pathway 369 allows T cells to be activated by uninfected DCs [38,41]. Thus, T cell recognition of infected 370 macrophages, which is central to our fundamental paradigm of TB pathogenesis, remains poorly 371 defined. 372 Our study advances the understanding of T cell recognition of Mtb-infected cells. By 373 focusing on TCR-mediated recognition, our data show that T cells specific to immunodominant 374 antigens vary in their ability to recognize Mtb-infected macrophages. Despite being a persistent 375 and dominant population of CD8 + T cells in the lungs of Mtb-infected mice, TB10.44-11-specific 376 CD8 + T cells do not recognize Mtb-infected macrophages. While we primarily used TGPMs, which 377 have been used to model human macrophages [42,43], we also showed that TB10.4-specific 378 CD8 + T cells failed to recognize lung APCs from infected mice. Importantly, concurrent with our 379 analysis of CD8 + T cells, we systematically assessed recognition of Mtb-infected macrophages 380 by Ag85b-specific (i.e., P25) and ESAT-6-specific (i.e., C7) CD4 + T cells. Both recognized Mtb-381 infected macrophages and inhibited bacterial growth (here and [10]). Thus, under conditions that 382 activated Mtb-specific CD4 + T cells, no activation of TB10.4-specific CD8 + T cells occurred. This 383 finding has many implications, among which the most important is that not all Mtb-specific T cells 384 recognize Mtb-infected macrophages. 385 These results led us to re-examine the evidence that CD8 + T cells recognize infected cells. 386 In our evaluation of the literature, among the best evidence is: (1) direct ex vivo recognition of 387

Mtb-infected macrophages and DC by CD4 + and CD8 + T cells [44-47]; (2) murine T cells' cytolytic 388 activity (CTL) of MTb-infected cells [48-50]; (3) human CD8 + T cells that recognize Mtb-infected 389
DC [51][52][53]. However, these data have limitations. The murine studies never demonstrated 390 cognate recognition, and the frequencies were lower than expected. The human studies only used 391 DC and not macrophages and used a high MOI, raising concerns about death of infected cells 392 and presentation of nonviable antigen. Nevertheless, these studies support the idea that CD8 + T 393 cells recognize infected cells, but their frequency that recognize infected macrophages might be 394 lower than we previously expected. Such a finding might explain why CD8 + T cells make a 395 disproportionately small contribution to host defense, even though Mtb infection elicits a robust 396 CD8 + T cell response. 397 We investigated several mechanisms that might explain why TB10.4-specific CD8 + T cells Lindenstrom et al report that vaccination with TB10.4 (EsxH), which has a leucine at position 12 407 (i.e., IMYNYPAML), inefficiently generates TB10.4-specific CD8 + T cells [56]. However, 408 vaccination with TB10.3 (EsxR), a related antigen that also contains the same epitope followed 409 by a methionine (i.e., IMYNYPAMM), elicits TB10.4-specific CD8 + T cells. Based on that finding, 410 they conclude there is a processing defect that prevents the generation of the TB10.44-11 epitope 411 from the TB10.4 protein. However, they also find that TB10.4-specific CD8 + T cells elicited by 412 TB10.3 vaccination recognize splenocytes pulsed with the rTB10.4 proteins, showing that the full 413 18 length TB10.4 protein can be processed and presented. These data indicate that the lack of 414 vaccine-elicited TB10.4-specific CD8 + T cells is due to a problem with priming after vaccination 415 instead of an inability to process the IMYNYPAM epitope. Moreover, our data using TB10.4-416 expressing Listeria show that TGPMs can process the full length TB10.4 protein and present the 417 TB10.44-11 epitope. Therefore, we conclude that amino acid sequence of TB10.4 does not hinder 418 its processing. The Listeria experiments also show the potential importance of antigen location 419 and raise the possibility that sequestration of the TB10.4 antigen in the phagosome renders it 420 inaccessible to the MHC class I presentation pathway. 421 Low antigen abundance could also explain the lack of recognition. We have previously 422 argued that there is limited amount of TB10.4 antigen presentation in the lungs of infected mice, 423 leading to extreme bias in the TCR repertoire of the TB10.4-specific CD8 + T cell response and 424 defects in the memory-recall response in vivo [19,23]. To test the possibility of low antigen 425 abundance, we overexpressed EsxG and EsxH (TB10.4) together but did not see greater T cell 426 recognition of Mtb-infected macrophages, suggesting that abundance might not be the issue. 427 Unexpectedly, macrophages pulsed with γ-irradiated Mtb were recognized by TB10.4-428 specific CD8 + T cells, raising the possibility that live Mtb actively inhibits MHC class I presentation 429 of TB10.4. This is particularly interesting since the presentation of CFP10, another ESAT-6-like 430 protein, by human DCs to CD8 + T cells requires viable Mtb; DCs given heat-killed bacteria do not 431 present CFP10 to T cells [53]. While these data suggest that presentation requires active 432 secretion of CFP10 [57,58], the heat-killing process could have destroyed CFP10, or there might 433 not have been sufficient amounts of CFP10 available in the non-viable bacteria. In combination 434 with our data showing polyclonal CD8 + T cells recognize Mtb infected macrophages, these data 435 show that it is possible that certain antigens are presented by live Mtb while others are actively 436 prevented from being sampled by MHC class I. 437 Ag85b is an immunodominant antigen with an epitope recognized by CD4 + T cells in 438 C57BL/6 mice. In vivo data shows that Ag85b-specific CD4 + T cells can recognize Mtb-infected 439 19 cells early during infection; however, recognition decreases after infection is established [12, 14, 440 15, 59, 60]. The inability of Ag85b-specific CD4 + T cells to efficiently recognize Mtb-infected bone-441 marrow derived macrophages (BMDMs) or bone-marrow derived dendritic cells (BMDCs) stems 442 from a combination of reduced Ag85b expression by Mtb and because infected cells actively 443 export Ag85b into the extracellular milieu [12,13]. In our experiments, we found that P25 T cells 444 The TB10.44-11 epitope has been extensively used to characterize CD8 + T cell responses 455 in the mouse model of TB, and TB10.4-specific CD8 + T cell responses have also been 456 characterized in people with tuberculosis [19,21,23,56,[61][62][63]. The finding that TB10.4-specific 457 CD8 + T cells do not recognize infected macrophages was unexpected, particularly since TB10.4-458 specific CD8 + T cells persist in the lungs of infected mice and become more dominant with time 459 [18,19]. From these experiments, two questions warrant further investigation: 1) whether the 460 CD8 + T cells specific to other epitopes of TB10.4 also inefficiently recognize infected 461 macrophages, and 2) whether the species or the host genetic background influence recognition 462 of infected cells. 463 In retrospect, our findings may partially explain why eliciting TB10.4-specific CD8 + T cells 464 by vaccination fails to protect mice against Mtb infection [23,56]. While vaccination with 465 20 immunodominant antigens recognized by CD4 + T cells (e.g., Ag85b, ESAT-6) induce moderate 466 protection [64,65], we must consider the possibility that these antigens may not be the best 467 stimulators of protective immunity. Ag85b-specific CD4 + T cells have variable efficacy, in large 468 part due to its reduced expression by the bacterium as early as 3 weeks after infection [11,12]. 469 However, by their nature, the recruitment of memory T cell responses specific for 470 immunodominant antigens is only incrementally faster than the primary T cell response [10,23].

520
Retrogenic mice expressing TB10Rg3 TCR specific for the TB10.44-11 epitope were 521 generated as previously described [19]. The TB10Rg3 CD8 + T cells were isolated from these 522 mice, stimulated in vitro with irradiated splenocytes pulsed with the peptide TB10.44-11 in complete 523 media containing IL-2. P25 or C7 CD4 + T cells were isolated from transgenic P25 or C7 mice, 524 respectively [24,34]. The P25 and C7 cells were stimulated in vitro with irradiated splenocytes 525 pulsed with the Ag85b241-256 peptide or the ESAT-61-15, respectively, in complete media containing 526 IL-2 and anti-IL-4. After the initial stimulation, these T cells were split every two days for 3-4 527 divisions and rested for two to three weeks. After the initial stimulation, the cells were cultured in 528 complete media containing IL-2 and  Peptides 530 The following synthetic peptide epitopes were used as antigens: TB10.44-11 531 (IMYNYPAM); Ag85b241-256 (QDAYNAAGGHNAVFN); and ESAT-61-15 (MTEQQWNFAGIEAAA). 532 We also generated a negative control peptide predicted to not bind to H-2 K b : IMANAPAM. The 533 peptides were obtained from New England Peptides (Gardner, MA). 534 As positive controls assessing the function of macrophages to present antigen, uninfected 535 macrophages and, in certain experiments, infected macrophages were pulsed with the peptides 536 of interest. We pulsed macrophages by incubating 10uM of the peptides of interest with the 537 macrophages in supplemented complete RPMI 1640 media for 1 hour. After incubation, the cells 538 were washed 3 to 5 times with fresh supplemented complete RPMI 1640 media. The cells were 539 then resuspended in supplemented complete RPMI 1640 media for experiments. 540

541
H37Rv was grown as previously described [28]. Bacteria was grown to an OD600 of 0.6 -542 1.0, washed in RPMI, opsonized with TB coat (RPMI 1640, 1% heat-inactivated FBS, 2% human 543 serum, 0.05% Tween-80), washed again and filtered through a 5 micron filter to remove bacterial 544 clumps. The bacteria were counted using a Petroff-Hausser chamber. Infection was performed 545 as previously described [28]. The final multiplicity of infection (MOI), based on plating CFU, was 546 24 0.2-0.8 for all experiments. For CFU-based, bacterial growth inhibition assays, T cells were added 547 at a ratio of 5 T cells to each macrophage. Four replicate wells were used for each condition. Cell 548 cultures were lysed by adding 1/10 th volume of with 10% Triton X-100 in PBS (final concentration 549 of 1%), and CFUs were determined by plating in serial dilutions of the lysates on Middlebrook 550 7H10 plates. CFUs were enumerated after culture for 21 days at 37°C and 5% CO2. 551 NdeI and NotI-HF. The esxGH gene from H37Rv, along with 12 upstream nucleotides, was 594 inserted by HiFi Assembly following the plasmid-borne tetracycline-inducible promoter. All 595 enzymes used above were purchased from New England Biolabs. The resulting plasmid (pGB6) 596 was electroporated into Mtb H37Rv and integrated at the L5 site. RNA was purified from induced 597 26 and uninduced cultures using TRIzol (ThermoFisher) and chloroform extraction, followed by 598 purification on Zymo columns. cDNA was produced with Superscript IV (ThermoFisher), and 599 quantitative PCR was performed using the iTaq SYBR Green Supermix (Bio-Rad, Hercules, CA) 600 on an Applied Biosystems Viia 7 thermocycler. 601