key: cord-0934678-3pv22nbs authors: Zhang, J.; Koolmeister, C.; Han, J.; Filograna, R.; Hanke, L.; Adori, M.; Sheward, D. J.; Teifel, S.; Liu, Y.; Harris, R. A.; Murrell, B.; Mcinerney, G.; Aoun, M.; Backdahl, L.; Holmdahl, R.; Pekalski, M.; Wedell, A.; Engvall, M.; Wredenberg, A.; Karlsson Hedestam, G. B.; Castro Dopico, X.; Rorbach, J. title: Antigen receptor stimulation drives selection against pathogenic mtDNA variants that dysregulate lymphocyte responses date: 2021-10-07 journal: nan DOI: 10.1101/2021.10.05.21264464 sha: 0be73fc38641e7a82476918cf3a77993d70805e0 doc_id: 934678 cord_uid: 3pv22nbs Pathogenic mitochondrial (mt)DNA molecules can exhibit heteroplasmy in single cells and tissues and cause a range of clinical phenotypes, although their contribution to immunity is poorly understood. Here, in mice carrying heteroplasmic C5024T in mt-tRNA Ala - that impairs oxidative phosphorylation - we found a reduced mutation burden in peripheral T and B memory lymphocyte subsets, compared to their naive counterparts. Furthermore, selection diluting the mutation was induced in vitro by triggering T and B cell antigen receptors. While C5024T dysregulated naive CD8+ T cell respiration and metabolic remodeling post-activation, these phenotypes were partially ameliorated by selection. Analogous to mice, peripheral blood memory T and B lymphocyte subsets from human MELAS (Mitochondrial Encephelomyopathy with Lactic Acidoses and Stroke-like episodes) patients - carrying heteroplasmic A3243G in mt-tRNA Leu - displayed a reduced mutation burden, compared to naive cells. In both humans and mice, mtDNA selection was observed in IgG+ antigen-specific B cells after SARS-CoV-2 Spike vaccination, illustrating an ongoing process in vivo. Taken together, these data illustrate purifying selection of pathogenic mtDNA variants during the oxidative phosphorylation checkpoints of the naive-memory lymphocyte transition. *Shared senior and corresponding authors: Pathogenic mitochondrial (mt)DNA molecules can exhibit heteroplasmy in single cells 32 and cause a range of clinical phenotypes, although their contribution to immunity is 33 poorly understood. Here, in mice carrying heteroplasmic C5024T in mt-tRNA Ala -that 34 impairs oxidative phosphorylation -we found a reduced mutation burden in peripheral 35 T and B memory lymphocyte subsets, compared to their naïve counterparts. 36 Furthermore, selection diluting the mutation was induced in vitro by triggering T and B 37 cell antigen receptors. While C5024T dysregulated naïve CD8 + T cell respiration and 38 metabolic remodeling post-activation, these phenotypes were partially ameliorated by 39 selection. Analogous to mice, peripheral blood memory T and B lymphocyte subsets 40 from human MELAS (Mitochondrial Encephalomyopathy with Lactic Acidosis and 41 Stroke-like episodes) patients -carrying heteroplasmic A3243G in mt-tRNA Leu -42 displayed a reduced mutation burden, compared to naïve cells. In both humans and 43 mice, mtDNA selection was observed in IgG + antigen-specific B cells after SARS-CoV-44 2 Spike vaccination, illustrating an on-going process in vivo. Taken together, these 45 data illustrate purifying selection of pathogenic mtDNA variants during the oxidative 46 phosphorylation checkpoints of the naïve-memory lymphocyte transition. 47 48 49 Key words: immunometabolism; mitochondrial diseases; mtDNA mutation; OXPHOS; 50 purifying selection; immunological memory 51 52 53 Introduction 4 disease phenotypes in C5024T (mt-tRNA Ala ) mice [28] [29] [30] . Despite the clinical importance 126 of such mutations, how such pathogenic polyploidy affects the mammalian immune 127 system is generally unknown, partly because until recently, models such as C5024T 128 mice were unavailable due to technical limitations of engineering the mitochondrial 129 genome. In this study, we investigated the effects on immune cells of the C5024T mutation in 132 murine tRNA Ala , that impairs OXPHOS and recapitulates different aspects of human 133 mitochondrial diseases, such as cardiomyopathy and altered organ mass ratios on the 134 murine C57BL/6N background 29,31,32 . In these animals, mtDNA mutation burden 135 reduction -the percentage by which a tissue had reduced the pathogenic burden 136 compared to baseline -is more pronounced in highly proliferative tissues, such as 137 colonic epithelium and blood 31,33 . A deleterious mtDNA molecule harboring a 3.1-kb 138 deletion in a heteroplasmic C. elegans strain 34,35 also showed increased selection in 139 the intestine, compared to the body wall muscle. It thus follows that different tissues 140 and cell types have variable tolerable mutation thresholds, beyond which a phenotype 141 manifests 31 . Although it is known that proliferation plays an important role in purifying 142 selection, the precise drivers (cues, contexts) and molecular mechanisms of selection 143 in single cells from different tissues remain unknown. Such knowledge in the context 144 of the immune system, could potentially open new avenues for therapeutic targeting 145 of pathogenic mtDNA molecules and processes. 146 147 In this respect, we sought to extend our observations in C5024T mice to MELAS 148 syndrome patients who carry different levels of A3243G in another mt-tRNA gene 149 encoding mt-tRNA Leu(UUR) . MELAS patients may exhibit severe multi-systemic clinical 150 phenotypes when the heteroplasmy burden is above a given threshold. For example, 151 MELAS patients included in this study -with a baseline mutation burden >90% -152 suffered from hypertonia, neurological disorders, stroke-like episodes, 153 cardiomyopathy and retinitis pigmentosa, amongst other phenotypes that could have 154 an underlying immunological component [36] [37] [38] . In a recent study published during the 155 course of our research, total T and B cells from three MELAS patients were reported 156 to have a lower A3243G mutation burden than in myeloid lineages (which we 5 towards low mtDNA burden clones. Using C5024T mice and MELAS patient samples, 170 we demonstrate that T and B lymphocyte populations select against pathogenic 171 mtDNA mutations in response to antigen receptor stimulation both in vitro and in vivo. 172 This purifying selection illustrates the OXPHOS checkpoints towards immunological 173 memory and reveals molecular phenotypes of heteroplasmic mitochondrial disease. 174 175 . CC-BY 4.0 International license It is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted October 7, 2021. ; https://doi.org/10.1101/2021.10.05.21264464 doi: medRxiv preprint 7 change in C5024T frequency, relative to heteroplasmy in the ear (sampled at weaning, 220 baseline), as previously described 29 . Previous studies have shown that ear 221 heteroplasmy at the time of weaning is comparable to heteroplasmy in other tissues 222 in aged mice that do not select against the variant 31 . 223 224 We found selection against the pathogenic variant across hematopoietic tissues was 225 age-associated, with cells from adult animals having lower mutation burdens than did 226 the same population isolated from young animals (Fig. 1C) . These results closely 227 recapitulate mitochondrial disease patients' heteroplasmy tissue distribution and 228 decreased burden with ageing (for MELAS patients with pathogenic A3243G) 20,25,43-229 46 . 230 231 In agreement with a recent study of three MELAS patients (carrying A3243G) 39 , we 232 determined myeloid lineages from C5204T mice to have a higher mutant mtDNA 233 burden than did lymphoid lineages, concurring with their reduced lifespans and 234 proliferation. In C5024T monocytes (CD3 -CD19 -CD11b + Ly6G -Ly6C hi ), dendritic cells 235 (CD3 -CD19 -CD11b + CD11c + MHC-II + ) and neutrophils (CD3 -CD19 -CD11b + Ly6G + ) 236 sorted from young and adult mice spleens, myeloid selection was also associated with 237 age (Fig 1D and S2A) , coinciding with an in silico model that predicted selection of 238 A3243G in haemopoietic stem cells 47 . In this respect, age-associated selection of 239 lower mutation burdens was also observed in purified naïve CD4 + and CD8 + T cells 240 (Fig. 1E ). Together, these data illustrated that circulating lymphocyte precursors and stem cells 243 in the thymus and bone marrow also undergo selection against the mutation, in 244 addition to what might be induced in peripheral immune cells, in agreement with 245 studies showing increased oxidative metabolism in proliferating hematopoietic stem 246 cells (HSCs) 48 , and bursts of high glucose uptake in the common lymphoid progenitor 247 and thymocytes 49 . Differential selection of C5024T between naïve and memory T and B cell subsets 250 251 As appropriate OXPHOS and mitochondrial functionality is central to lymphocyte 252 activation and the maintenance of T and B cell memory, we hypothesized that mutant 253 mtDNA levels in lymphocytes would delineate by naïve and memory phenotype, with 254 antigen-experienced cells having a reduced pathogenic burden. To test this possibility, 255 we isolated T and B cell subsets of interest by Fluorescence-Activated Cell Sorting 256 (FACS) directly ex vivo ( Fig. 1F and S2B ). As selection became more pronounced with 257 age, and memory cell numbers increased over time, we primarily analyzed the 258 heteroplasmy burden in lymphocytes sorted from adult (10-month-old) animal spleens, 259 with the use of younger (2-month-old) animals being highlighted in the figures. Within 260 the sorted populations from an individual mouse, percentage heteroplasmy reduction 261 for each lineage was calculated relative to ear, as described above. Ear heteroplasmy 262 levels for individual animals are presented in Figure S3 . is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint 8 264 Within the splenic CD4 + T cell compartment of adult mice (n=17), we found effector 265 phenotype cells (Tem: CD3 + CD4 + CD44 hi CD62L lo ) to have a ~6% lower mutation 266 burden than their naïve (Tn: CD3 + CD4 + CD44 lo CD62L + ) counterparts ( Fig. 1G and 267 S3A), although selection was more pronounced in some animals than others. 268 Furthermore, CD3 + CD4 + CD25 + FOXP3 + T-regulatory cells sorted from six animals 269 (three young and three adult) had reduced their mutation burden comparably to 270 effector CD4 + T cells in adults, in-keeping with their maintenance via autoantigens 50 271 and restraining of past effector T cell responses ( Fig. 1H and S3B) . Indeed, FOXP3 + 272 Tregs sorted from these animals were CD44 + , although cell surface levels were lower 273 than those on effector/memory non-Tregs (Fig. S3B) . A similar pattern was observed in splenic CD8 + T cells, with effector (Tem: 276 CD3 + CD8α + CD44 hi CD62L lo ) and central memory (Tcm: subsets of adult animals (n=14) having ~5% and ~6% lower mtDNA mutation burdens, 278 respectively, than did naïve (Tn: CD3 + CD8α + CD44 lo CD62L hi ) cells ( Fig. 1I -J and S3C). Burden reduction was not statistically different between Tem and Tcm subsets. We next sought to determine whether the same applied to B cells and first isolated 282 fetal-origin B-1a cells (CD3 -CD19 hi B220 lo CD5 + CD43 + IgM + ) from the peritoneal cavity, 283 as these cells -which spontaneously produce natural IgM -are known to depend 284 upon peripheral self-renewal and to have a high capacity for OXPHOS as they protect 285 serosal surfaces 51,52 . Indeed, with found B-1a cells isolated from the peritoneal cavity 286 to have a higher OXPHOS capacity than did naïve B-2 cells (CD3 -287 CD19 + B220 + IgM hi IgD hi ) from the same site (Fig. S3D ). In agreement with their fetal 288 ontogeny and elevated maximal OCR, B-1a cells had selected more (~15% reduction) 289 than naïve B-2 cells (~11% reduction) by 2-months-of-age (Fig. S3E ). However, 290 peritoneal naïve B-2 cells selected more than did γδ-T cells (CD3 + γδ-TCR + ) isolated 291 from cavity (which reduced mutation burden ~8% on average), perhaps indicative of 292 the innate-like features of IL-17-producing γδ-T cells that predominate body cavity 293 responses 53,54 (Fig. S3E ). We confirmed these B-1a findings in adult animals, and in 294 one mouse, peritoneal B-1a cells had a striking >20% mutation burden reduction 295 difference to age-selected naïve splenic B-2 cells, which themselves showed a striking 296 40% burden reduction relative to ear (Fig. 1K ). Amongst splenic B cells, class-switched IgG + B-2 cells (CD3 -CD19 + B220 + IgG + IgD -299 IgM -, n=16) had a reduced mutation burden compared to un-switched naïve cells 300 (CD3 -CD19 + B220 + IgM hi IgD hi IgG -). (Fig. 1L and S3F ). Class-switched IgG + B cells 301 generally showed the strongest naïve vs. memory selection in the subsets studied, 302 averaging an 8.5% difference in burden reduction (Fig. 1L ). As observed in T cells, 303 several animals showed more striking (>15%) differences between their naïve and 304 IgG + memory B-2 or B-1a cells, while in others, differences were more subtle, 305 illustrating how selection pressures vary between individuals, perhaps due to 306 environmental effects provoking the immune system. 307 . CC-BY 4.0 International license It is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted October 7, 2021. Taken together, these results demonstrate that the generation/maintenance of 309 immunological memory in vivo is associated with purifying selection against a 310 pathogenic mtDNA variant, strongly suggesting the C5024T mutation impairs adaptive 311 immune cell function. In vitro stimulation of T and B cells -but not BMDMs -from C5024T mice induces 314 rapid selection 315 316 As the T and B memory pools had selected against C5024T in vivo, we next sought to 317 determine whether we could induce this process through antigen receptor stimulation 318 in vitro, which would represent an OXPHOS checkpoint of immune memory. 319 320 To this end, we stimulated naïve CD4 + and CD8 + T cells via the T cell receptor (TCR) 321 complex (anti-CD3/CD28) and cultured them for 5 days in the presence of IL-2 ( Fig. 322 2A). By staining naïve CD4 + and CD8 + T cells with a proliferation tracing dye and 323 FACS-isolating cells from different divisions, we observed heteroplasmy burden 324 reduction (purifying selection in the culture) to occur strongly from the third-fourth 325 division onwards ( Fig. 2B-C) , suggesting that T cells with higher mutation burdens can 326 undergo a few rounds of division under these conditions. 327 328 Compared to ex vivo naïve CD4 + and CD8 + T cells collected at day 0, live cells FACS-329 isolated from the latest division from either lineage (after 5 days of culture) had 330 significantly lower C5024T burdens ( Fig. 2D -E). In these cultures, CD8 + T cells 331 selected against the variant more strongly than CD4 + cells, approximating ~17% and 332 ~9%, respectively, in agreement with their increased divisions ( Fig. 2B -C) and higher 333 reported dependence on OXPHOS than CD4 + cells 9,55 . This sharp decrease in 334 heteroplasmy between ex vivo naïve and the most-proliferated CD8 + T cells was 335 confirmed by Sanger sequencing (Fig. S4 ). Selection against the variant was thus 336 more pronounced in vitro (~17% reduction in stimulated CD8 + T cells) than was 337 observed in vivo (~5% difference at 10 months-of-age), in agreement with the 338 supraphysiological polyclonal activation conditions of conventional protocols. 339 Furthermore, we found that culturing C5024T CD4 + T cells in the presence of 340 galactose (which forces the cells to rely on OXPHOS) exacerbated selection in both 341 un-divided and divided cell populations (Fig. 2F ). To further explore the heteroplasmy-proliferation relationship, we analyzed clonal 344 expansion of C5024T T cells. As reported for other tissues in these mice and 345 mitochondrial disease patients 19,21,23,24,31 , within the peripheral CD8 + T cell pool, we 346 found single naïve cells to have a wide range distribution of mutation burden (Fig. 2G ). 347 After 8 days of culture-post TCR stimulation, we found larger CD8 + clonal populations 348 (>200 cells) to have lower C5024T burdens (~25% reduction vs. ear) than smaller 349 ones (<200 cells) (~2% reduction) (Fig. 2H ). These results show that even high 350 mutation burden cells -e.g., >75% -can proliferate, although they do so with slower 351 . CC-BY 4.0 International license It is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint selection at the cellular level in BMDMs. This further supported the notion that most of 370 the myeloid burden reduction with age occurs in the self-renewing precursors of 371 circulating cells. As the degree of heteroplasmy remained similar to ear baseline levels in macrophages 374 polarized under these conditions, it holds that they may suffer functional deficits from 375 the mutation. In this respect, the expression of MHC class II (IA/IE) was found to be 376 significantly lower at the cell surface in C5024T M0, M1 and M2 cells, compared to 377 WT cells (Fig. 2L ). This observation with flow cytometry was mirrored at the mRNA 378 level for the class II IA gene but did not reach statistical significance in M1-polarized 379 cells (Fig. 2L ). Furthermore, using the Seahorse MitoStress test that characterizes the 380 extracellular metabolic flux, we found C5024T M2 macrophages -which required 381 higher levels of OXPHOS than M1 cells (Fig. S5D ) -to have reduced mitochondrial 382 respiration compared to WT cells, and mutant cultures were typified by a high 383 extracellular acidification rate ( Fig. 2M is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted October 7, 2021. 11 maximal oxygen consumption rate (OCR) and ATP-coupled OCR (Fig. 3A) , compared 396 to WT CD8 + T cells, illustrating that the mutation impairs the population response to 397 TCR triggering. However, after 8 days of culture post-TCR activation and a 6-hour re-398 stimulation with CD3/CD28, these differences were absent, with both C5024T and WT 399 cells responding similarly (Fig. 3B) . These results suggest that the T cell population in 400 vitro partially restores more optimal OXPHOS through purifying selection after rapid 401 proliferation. To further determine how naïve CD8 + T cell cultures were metabolically dysregulated 404 by C5024T, mutant and WT cells were (unstimulated or) stimulated via the TCR for 6 405 and 96 hours (4 days) and isolated RNA was subjected to transcriptomic analysis 406 using the Nanostring nCounter mouse metabolic pathways panel that analyses the 407 expression of 786 genes across 34 pathways central to cellular metabolism 56 . 408 409 Validating our stimulation conditions, within 6 hours of stimulation, the expression of 410 genes important for CD8 + T cell effector functions, such as Il2ra (CD25), Tbx21 (T-411 bet), Irf4, Gzmb (Granzyme B) and Bcl2l1, were strongly upregulated, compared to 412 time 0 (Fig. S6A ). Metabolically and at the same timepoint, the expression of serine 413 hydroxymethyltransferase (Shmt1) was strongly induced, confirming a role for one-414 carbon metabolism in early CD8 + T cell activation 57 , while highly-upregulated 415 spermidine synthase (Srm) and ornithine decarboxylase (Odc1) illustrated the 416 important role recently reported for polyamine metabolism during T cell activation 58 . Gene module analysis (extracting pathway-level information from a group of genes 419 using the first principal component of their expression data 59 ) was used to reveal the 420 overall metabolic remodeling of early T cell activation ( Fig. S6B is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted October 7, 2021. ; 12 ( Fig. S6D ), relative to WT. In contrast, genes involved in epigenetic remodeling were 440 less expressed in C5024T than WT cells. These results highlight a specific pattern of 441 dysregulated metabolic reprogramming of naïve C5024T CD8 + T cells after TCR 442 activation. The upregulation of nuclear-encoded metabolic and inflammatory genes 443 (compared to WT cells) perhaps occurs as a compensatory mechanism to improve 444 mitochondrial function in T cells with more severely impaired OXPHOS, as suggested 445 by similar results from fibroblasts from a MELAS (A3243G) patient 62 . Such a 446 compensatory increase could itself dysregulate effector T cell responses. Consistent with the in vitro selection and Seahorse data ( suggested that purifying selection in culture partially relieved some of the genetic 451 dysregulation present in the pool immediately after TCR triggering (Fig. S6B ). 452 Nevertheless, to determine whether functional deficits persisted in the CD8 + 453 population after proliferation and selection, as suggested by reduced expression of 454 epigenetic remodeling and elevated cytokine signaling genes in C5024T T cells after 455 6 h activation, we re-stimulated 5 day-activated cultures with PMA/Ionomycin. Six 456 hours after re-stimulation, we found a greater proportion of C5024T CD8 + T cells from 457 the latest division to produce IFN-γ, compared to those from WT animals (Fig. 3D ), 458 indicating that dysregulated pro-inflammatory responses persist in the T cell 459 population after selection. Furthermore, we found circulating IFN-γ levels to be 460 increased in C5024T mice (Fig. 3E ). Increased IFN-γ production has previously been 461 associated with mitochondrial insufficiency and may represent an OXPHOS-glycolysis 462 imbalance in the system that could have pathophysiological consequences 14,63 . These data indicate that C5024T dysregulates the naïve CD8 + T cell response to 465 antigen-receptor stimulation at the transcriptional, protein and cellular levels, and that 466 purifying selection in cultured C5024T cells correlated with partial resolution of 467 phenotypic differences compared to WT cells. C5024T mice mount a class-switched neutralizing antibody response after SARS-470 CoV-2 Spike protein subunit vaccination 471 472 To determine whether we could observe a reduction in pathogenic mtDNA burden in 473 response to a specific antigen, and to test whether C5024T mice had an intact T cell-474 independent B cell response, we inoculated mice with a SARS-CoV-2 Spike (S) 475 glycoprotein trimer vaccine that we have previously studied in mice and other 476 species 64-66 ( Fig. 4A -B). All but one C5024T (from n=10) and one WT animal (from n=9) developed S-specific 479 IgG responses by five weeks after the priming dose (2 μg S trimers in AddaVax by is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted October 7, 2021. ; 13 was only detectable after the boost dose in all animals (Fig. 4C ), and at similar titers 484 in WT and C5024T mice when the AUCs of serial dilutions were analyzed, although 485 these trended towards being reduced in C5024T mice ( Fig. 4C and S7A). Despite 486 having similar binding titers at these timepoints, we found that the capacity of 487 vaccinated mice sera to neutralize virus infection in vitro was lower for C5024T mice 488 [mean ID50 735 vs. 420 serum dilution] 5 weeks after the boost ( Fig. 4D and S7B ), 489 suggesting that the quality of the antibody response is negatively affected by the 490 mutation. Thus, although all C5024T mice mounted a neutralizing response after 491 vaccination, further molecular studies are needed to determine whether antibody 492 repertoires are adversely affected by the mutation, e.g., whether vaccine-induced 493 C5024T (vs. WT) antibody lineages show lower levels of somatic hypermutation 494 (SHM) due to a less efficient germinal center response. This might be apparent at the 495 level of virus neutralization, but not at the level of S or RBD binding, as affinity is less 496 of a determinant for the latter. Indeed, no C5024T mice generated ID50 neutralizing 497 titers above a 1:600 serum dilution, while 5/10 WT mice did (Fig. 4D ). Post-vaccination differences were also observed in the frequencies of T and B cells 500 between the C5024T and WT mice. C5024T mice had an increased B cell frequency 501 and reduced CD8 + T cell frequency in draining (pooled inguinal) lymph nodes 5 weeks 502 after the boost, compared to unvaccinated and vaccinated WT animals (Fig. S7C ). 503 However, the frequency of CD4 + T cells was not altered. Absolute lymph node cell 504 counts were not significantly different between the mutant and control mice, 505 suggesting that the S vaccine (which mostly engaged B cells and not T cells) altered 506 normal lymphocyte proportions in situ in C5024T mice. 507 508 Our vaccination results, together with our in vitro CD8 + data, highlight the need to 509 assess vaccine efficacy and infectious disease susceptibility in the C5024T mouse 510 model, despite that this model can mount a neutralizing, class-switched response. We 511 note, however, that in this experiment, young animals (2-month-old) were 512 inoculated 67,68 . Five weeks-post boost and using a fluorescently-labelled S probe we generated, we 517 FACS-isolated IgG + S-specific B cells from the draining (inguinal) lymph nodes of 518 vaccinated C5024T mice (n=6) to determine their mutation burden (Fig. 4E ). 519 Compared to S-negative naïve B-2 cells (CD3 -CD19 + B220 + IgM hi IgD hi IgG -S -), class-520 switched S-reactive cells (CD3 -CD19 + B220 + IgM hi IgD hi IgG -S + ) had a lower mutation is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted October 7, 2021. ; MELAS patients. 528 529 Given our observations in the heteroplasmic mouse model, we sought to determine 530 whether human memory T and B cells carrying heteroplasmic mtDNA mutations 531 exhibited the same purifying selection observed in mice. For this purpose, we used 532 FACS to examine different peripheral blood immune cell types from two MELAS 533 patients carrying the A3243G mutation in mt-tRNA Leu (Fig. 5A -B and S8-9). Patient 534 clinical characteristics and measured parameters are detailed in Table 1 . In MELAS Patient A (male, 16-20 years-of-age), who had a whole blood mutation 537 burden of 50% by age 16-20 (compared to a 95% A3243G burden detected by skeletal 538 muscle biopsy at the same timepoint), we found sorted myeloid lineages to have 539 higher mutant mtDNA burdens than did T and B cells (Fig. 5C ), in agreement with a 540 previous study 39 . In this patient, freshly isolated monocytes (CD3 -CD20 -CD56 -541 NK cells (CD3 -CD20 -CD56 + ) reduced their A3243G burden by approximately 40%, 543 while total PBMCs reduced their burden by ~50%. Within T cell subsets from Patient A, both naïve CD4 + and CD8 + cells had a higher 546 mutation burden (~60% reduction) than their respective effector, central memory and 547 TEMRA subsets (each approaching an ~80% reduction) (Fig. 5C ). Here, CD8 + effector 548 memory cells showed a striking ~30% greater burden reduction than their naïve 549 counterparts (Fig. 5C ). Within the B cells, we also observed naïve (CD3 -CD20 + CD27 -550 IgM hi ) B cells to have a higher mutation burden (-48%) than CD27 + memory B cells, 551 with the latter showing a 63% reduction, and were therefore subjected to weaker 552 selection than were memory T cells in this patient. 553 554 In non-relative Patient B (female, 51-55 years-of-age with 90% heteroplasmy burden 555 in skeletal muscle, 25% in blood), the same patterns were observed. However, as 556 Patient B was considerably older than Patient A, selection was more pronounced (Fig. 557 5D). Monocytes, DCs and NK cells from Patient B showed a ~70% burden reduction 558 at sampling (compared to 40% in Patient A), while PBMCs approximated an 80% 559 reduction. Despite the strong selection across the hematopoietic system in this patient, naïve T 562 and B cells still had a clearly higher A3243G burden than their memory counterparts. 563 For example, Tregs (CD3 + CD4 + CD25 hi CD127 lo CD45RA -) in this patient showed a 564 striking 95% reduction, while naïve CD4 + T cells reduced their burden by 81%. It is 565 noteworthy that mitochondrial complex III is essential for Treg-mediated 566 suppression 69 . CD27 + memory B cells also displayed greater selection than did naïve 567 B cells (84 vs. 76% reduction) (Fig. 5D ), and as in Patient A, B cells from the female 568 patient also selected against the A3243G variant less strongly than did T cells, 569 perhaps because thymic involution in humans more strongly constricts the peripheral 570 . CC-BY 4.0 International license It is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted October 7, 2021. ; 15 lymphocyte pool than do B cell tolerance mechanisms and output from the bone 571 marrow. 572 573 In non-relative female Patient C (female, 51-55 years-of-age), a similar pattern was 574 observed (Fig. 5E ). However, this patient had a much lower baseline level of mutation 575 (65% in skeletal muscle) compared to the other patients and differences were more 576 subtle, in line with a less severe clinical history (Table 1) . Indeed, the level of mutation 577 in total PBMCs in this patient was 9%, indicating this individual had already undergone 578 strong selection across the hematopoietic system by the time of sampling. As in the 579 other two patients, myeloid lineages selected against the variant less than did total 580 PBMCs, while CD4 + and CD8 + memory T cell subsets again had a reduced level of 581 mutation compared naïve cells. CD4 + T central memory cells in this patient had a 582 heteroplasmy level of just 3%, illustrating very low levels of pathogenic burden in this 583 lineage. We did not, however, observe a difference between naïve and memory B cells 584 in Patient C, perhaps because a ~10% mutation burden in B cells approaches levels 585 at which the variant is not pathogenic; in agreement with B cells selecting against the 586 variant less strongly than T cells in the patients with higher mutation burdens. 587 Together, these results further highlight the particular need to study the antigen-588 specific T cell response in MELAS patients. 589 590 Finally, at the time of sample collection, five weeks had passed since Patient B had 591 received her first dose of the AstraZeneca COVID-19 adenovirus vaccine. Using 592 ELISA, we found the patient had detectable IgG reactivity to both S and the RBD (Fig. 593 S10), although this was at the mid-low end of the titer range evident in mild (non-594 hospitalized) adult infections classified using the same assay 70 , in agreement with the 595 patient awaiting her boost dose. Furthermore, using our fluorescent Spike probe, we 596 sorted IgG + S-specific CD27 + memory B cells from Patient B, and found them to also 597 exhibit marked selection (~10% difference in absolute A3243G frequency) with respect 598 to the naïve population ( Fig. 5F -G). This illustrates that the selection process was 599 ongoing and was likely induced by vaccination in this case, in agreement with our data 600 from mice. Patient B was not knowingly previously infected with SARS-CoV-2. 601 602 As for C5024T, the human A3243G mutation in mt-tRNA Leu is, therefore, also 603 subjected to purifying selection at the naïve-memory lymphocyte transition, further 604 emphasizing the need to study the human and mouse mutations with regards to 605 mitochondrial disease patient immune responses. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted October 7, 2021. ; https://doi.org/10.1101/2021.10.05.21264464 doi: medRxiv preprint As reported by other studies, selection against pathogenic mtDNA variants that 615 dysregulate OXPHOS takes place in hematopoietic stem cells and highly proliferative 616 tissues in an age-dependent manner 31,47 . In addition, selection was observed to take 617 place more-so in lymphocytes than in myeloid cells in samples from three MELAS 618 patients 39 , a phenotype here validated and extended to the C5024T mouse model. 619 These results demonstrate the high metabolic demands of adaptive immunity, which 620 purges the pathogenic mtDNA variants over time, as OXPHOS insufficiency takes its 621 toll on both immature and mature lymphocytes. However, the (i) specific triggers or 622 checkpoints of such purifying selection in stem cells, thymocytes or the peripheral 623 lymphocyte pool; (ii) the relative contributions of these (e.g., central tolerance signaling 624 vs. peripheral antigen encounter); (iii) how these mutations impact on T and B cell 625 responses, have not been previously studied. 626 627 Our pyrosequencing data from ex vivo mouse and human immune cell subsets 628 revealed that the peripheral lymphocyte naïve-memory transition was accompanied 629 by purifying selection against the pathogenic C5024T and A3243G variants. 630 Therefore, after a lymphocyte has undergone V(D)J recombination and enters 631 circulation, further genetic selection, this time in mtDNA, occurs during the 632 generation/maintenance of adaptive immune memory (and presumably in tandem with 633 somatic hypermutation in B cells). Certain memory T and B cells, like hematopoietic 634 stem cells, are predicted to undergo self-renewal for the lifetime of the organism 71 and 635 require considerable metabolic remodeling to achieve optimal function and maintain 636 clonal longevity. For example, antigen-experienced B cells are known to require 637 elevated OXPHOS and proliferation to achieve high-affinity antibody maturation in the 638 GC 16 . This may contribute to the slightly reduced neutralizing titers we observed in 639 C5024T mice (vs. WT mice) in response to Spike vaccination, as fewer high affinity 640 anti-RBD clonal lineages with higher levels of somatic hypermutation would be 641 generated, compared to in WT mice, as cells with higher burdens remain in earlier 642 divisions or die as the metabolic demands of the GC response are not met. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted October 7, 2021. ; https://doi.org/10.1101/2021.10.05.21264464 doi: medRxiv preprint adulthood and old age 67 . Therefore, in such heteroplasmic models, peripheral self-657 renewal of ever fewer clonal T and B lineages is predicted to maintain immunity in 658 such patients, as cells with higher mutation burdens fail to fully differentiate and a 659 smaller proportion of the activated pool goes on to form robust memory cells. One 660 consequence could be increased infectious disease susceptibility due to reduced 661 antibody/TCR diversity, especially with ageing, and epidemiological studies of 662 mitochondrial diseases, for long associated with abnormal immune responses 72 , need 663 updating. With regards to immunological phenotypes associated with the C5024T mutation, our 666 in vitro transcriptional and cellular data together indicate that the mutation 667 dysregulates macrophage and T and B cell responses at transcriptional, protein and 668 cellular levels, e.g., by dampening epigenetic regulation of gene expression 58 and 669 increasing IFN-γ production by activated CD8 + T cells 14,63 , while MHC class II cell 670 surface expression and OXPHOS capacity were also significantly reduced in BMDMs. For example, as myeloid cells had a higher tolerance for the mutation, it will be 672 important to decipher how deficits in innate cells contribute to any differences in 673 disease susceptibility in C5024T mice and MELAS patients, e.g., by delaying the 674 initiation of the adaptive immune response. Importantly, we show that selection is readily inducible in vitro by triggering lymphocyte 677 division via the TCR or BCR, providing further evidence that antigenic activation is a 678 major mediator of purifying selection in vivo, while facilitating the study of mtDNA 679 selection mechanisms in primary cells. Indeed, as selection was observed in vaccine-680 induced B cells approximately 5 weeks after immunization, selection takes place 681 shortly after antigen encounter, although homeostatic proliferation and tonic signaling 682 in the periphery likely contribute to ongoing burden reduction in newly formed memory 683 cells over longer periods. Although our studies do not focus on the mechanistic 684 understanding of the selection process itself, several interesting features can be 685 observed that can be important for further analyses. For example, autophagy 686 transcripts were elevated in naïve CD8 + T cells and downregulated after TCR 687 triggering (Fig. S6C) , while selection occurred strongly from division 4 onwards, 688 suggesting that this pathway is not determinative in the selection process. In addition, 689 we noticed upregulated hypoxia pathway genes in activated C5024T CD8 + T cells 690 (compared to WT), while a recent study has reported that oxygen tension contributes 691 to tissue-specific mutation loads 73 . Thus, it is likely that the selection process is 692 influenced by the metabolic environmental conditions in which activated immune cells 693 work, helping to improve their function. It follows that studies of anti-tumor immunity in 694 C5024T mice and mitochondrial disease patients are warranted. Together, our data reveal an immunophenotype of human mitochondrial disease and 697 support that C5024T mice are a translationally valuable model to study the metabolic 698 requirements of adaptive immunity. Using such models, researchers can further 699 . CC-BY 4.0 International license It is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted October 7, 2021. ; into lithium heparin-coated tubes and processed for PBMC isolation within one hour 739 of venesection. PBMCs were then isolated by density gradient centrifugation over 740 Lymphoprep (StemCell Technologies). Briefly, whole blood was diluted 1:1 in D-PBS 741 with 2% FBS (Cytiva HyClone) and layered onto density gradient medium before 742 centrifugation at 800 g for 20 min without brake. An aliquot of plasma was removed 743 prior to gradient centrifugation and stored at -80 o C for serological assays before the 744 mononuclear cell layer was collected using Pasteur pipettes. PBMCs were then 745 . CC-BY 4.0 International license It is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint Monocytes: CD3 -CD19 -CD11b + Ly6G -Ly6C hi 775 Neutrophils: CD3 -CD19 -CD11b + Ly6G + 776 Dendritic cells: CD4 + T effector: CD8 + T central memory: CD3 + CD8 + CD44 hi CD62L + 784 γδ-T cells: CD19 -CD3 + γδ-TCR + 785 B-1a: CD19 hi B220 lo CD5 + CD43 + IgM + 786 Naïve B-2: CD3 -CD19 + B220 + IgD hi IgM hi IgG -787 Memory B-2: CD3 -CD19 + B220 + IgD -IgM -IgG + 788 Antigen-specific B-2: CD3 -CD19 + B220 + IgM -IgG + Spike + 789 . CC-BY 4.0 International license It is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The following panels were used to classify human cells: 812 Monocytes: CD45 + CD3 -CD20 -CD56 -CD14 + CD16 + HLA-DR + 813 Dendritic cells: CD45 + CD3 -CD20 -CD56 -CD14 -CD16 + HLA-DR + 814 NK cells: CD45 + CD3 -CD20 -CD56 + 815 CD4 + T naïve: CD45 + CD3 + CD20 -CD4 + CD45RA + CCR7 + 816 CD4 + T effector memory: CD45 + CD3 + CD20 -CD4 + CD45RA -CD27 low CCR7 low 817 CD4 + T central memory: CD45 + CD3 + CD20 -CD4 + CD45RA -CD27 hi CCR7 hi 818 CD4 + Tregs: CD45 + CD3 + CD20 -CD4 + CD45RA -CD127 low CD25 + 819 CD8 + T naïve: CD45 + CD3 + CD20 -CD8 + CD45RA + CCR7 + 820 CD8 + T effector memory:CD45 + CD3 + CD20 -CD8 + CD45RA -CD27 low CCR7 low 821 CD8 + T central memory: Memory B cells: CD45 + CD3-CD20+CD27+IgM low 825 SARS-CoV-2 Spike-specific B cells: CD45 + CD3 -CD20 + CD27 + IgM -IgG + Spike + 826 827 In Naïve CD4 + or CD8 + T cells and resting B cells were enriched from splenocytes 830 isolated from 2-month-old mice, as described above. Isolated naïve T cells were 831 stained with CFSE and 250,000 cells per well were seeded on 96-well U-bottom plates 832 with/without a 1:1 (cell:bead) ratio of Mouse T-Activator CD3/CD28 beads (Invitrogen) 833 . CC-BY 4.0 International license It is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted October 7, 2021. ; 23 in complete RPMI-1640 (with 10% FBS, 2.05 mM L-glutamine (Sigma), 1% 834 penicillin/streptomycin (Sigma), 100 U/ml IL-2 (Peprotech) and 55 μM 2-ME (Gibco)). 835 For some experiments, naïve CD4 + T cells were also stimulated in complete RPMI-836 1640 media with no glucose supplemented with 10 mM galactose. Polyclonal T cell 837 stimulation of whole spleen suspensions was done using plate-bound anti-CD3 (5 838 μg/ml, clone 145-2C11, BD Biosciences) and anti-CD28 (2 μg/ml, clone 37.51, 839 Biolegend) in 24-well plates at a density of 2x10 6 cells/well. In re-stimulation 840 experiments, PMA/Ionomycin was added for 1 h before protein transport was inhibited 841 for 4 h (1X Cell stimulation/Inhibitor Cocktail, eBioScience). Samples were then 842 washed and stained for intracellular antigens before analysis by flow cytometry. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted October 7, 2021. ; https://doi.org/10.1101/2021.10.05.21264464 doi: medRxiv preprint 25 sortase A recognition site (LPETG) and a 6xHis tag and expressed in 293F cells, as 922 described above. RBD-His was purified from filtered supernatant on His-Pur Ni-NTA 923 resin (ThermoFisher Scientific), followed by size-exclusion chromatography on a 924 Superdex 200. Fluorescent spike was generated by first attaching 925 dibenzocyclooctyine-N-hydroxysuccinimidyl ester (DBCO-NHS, Sigma) to the spike 926 trimer in a 3:1 molar ratio before attaching AbberiorStar-635P-azide (Abberior GMBH) 927 by strain-promoted azide-alkyne cycloaddition. The final product was purified from 928 unreacted DBCO and fluorophore using a PD-10 desalting column (Cytiva HyClone). SARS-CoV-2 spike vaccination 931 932 Ten WT and 10 C5024T male animals of 2-months-of-age were used; one C5024T 933 animal had to be sacrificed before the end of the experiment (week 8) because of 934 infection. Immunogens were aliquoted and diluted immediately before administration. 935 Animals were injected sub-cutaneously over the inguinal lymph nodes with 100 μl of 2 936 μg Spike trimers in PBS plus AddaVax (Invivogen) at day 0 (prime dose) and after 5 937 weeks (boost dose). Two WT and two C5024T mice were injected with sterile PBS for 938 use as un-vaccinated controls. Tail vein blood was collected 5 weeks after the prime 939 dose (pre-boost) and 5 weeks after the boost dose (at week 10, endpoint). After blood 940 clotting at room temperature, sera were separated by centrifugation (10 min, 6,000 g) 941 and stored at −80°C until use. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted October 7, 2021. For qPCR of mouse BMDMs, RNA was obtained from Trizol samples are previously 1013 described. 500 ng RNA was reverse transcribed using the High-capacity cDNA 1014 Reverse Transcription Kits (ThermoFisher Scientific) in a 20 μl reaction. 20 μl cDNA 1015 was diluted to 100 μl and 1 μl cDNA was used for the real-time quantitative PCR by 1016 using the Platinum SYBR Green qPCR SuperMix-UDG (ThermoFisher Scientific) and 1017 PCR machine QuantStudio 6 Flex (ThermoFisher Scientific). Primers (5'-3') used: 1018 Arg1 F: CTCCAAGCCAAAGTCCTTAGAG; R: AGGAGCTGTCATTAGGGACATC. 1019 Il1b F: GCAACTGTTCCTGAACTCAACT; R: ATCTTTTGGGGTCCGTCAACT. Statistical analyses 1025 1026 Statistical analyses were carried out in Prism 9 (GraphPad) and results are 1027 represented as mean ± SEM. Comparisons between naïve and memory cells from one 1028 individual were calculated using paired, two-tailed Student's t-tests. Unpaired, two-1029 tailed Student's t-tests were used for group comparisons. Analysis of more than two 1030 types of immune cell from one individual were calculated using paired one-way 1031 ANOVA with Bonferroni correction. Bonferroni-corrected analyses were used for gene 1032 expression data. Please contact the authors about material or data, available upon request. Acknowledgments 1037 1038 Our gratitude is for the MELAS patients and animals that contributed to the study. The is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted October 7, 2021. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted October 7, 2021. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted October 7, 2021. ; Figure 1 -Purifying selection against C5024T in memory T and B lymphocytes A. Schematic representation of the murine C5024T mutation in mt-tRNA Alanine. An electropherogram showing C/T heteroplasmy in purified naïve CD8 + T cells is inset. B. Pyrosequencing results (raw C5024T heteroplasmy percentage in total mtDNA) from different anatomical compartments of a representative, adult (10-month-old) C5024T mouse. LN: lymph node. C. C5024T burden reduction -always the % heteroplasmy reduction relative to ear clip at weaning -is shown for selected hematopoietic tissues isolated from young (2month-old) and adult (10-month-old) mice. D. Percentage C5024T burden reduction is shown for selected myeloid (red bars) and lymphoid (blue bars) lineages FACS-isolated from young and adult mice. DCs: dendritic cells. E. Percentage C5024T burden reduction is shown for purified naïve CD4 + and CD8 + T cells from young and adult animals. F. Gates used to FACS-isolate key naïve and memory T and B lymphocyte subsets from C5024T mice, cell surface markers are defined in the text and M&M. Tn: naïve T cells; Tem: effector memory T cells; Tcm: central memory T cells. All subsets were isolated from the spleen, apart from B-1a cells, which were isolated from the peritoneum. G. Paired t-test of percentage heteroplasmy reduction for CD4 + effector memory vs. naïve in adult mice (n=17). The mean of differences in percentage heteroplasmy reduction between both populations is shown for each animal on the right-hand axis. H. Paired t-test of percentage heteroplasmy reduction for CD4 + T regulatory vs. naïve; shown in red for three young animals (n=6 total). I. Paired t-test of percentage heteroplasmy reduction for CD8 + T effector memory vs. naïve in adult mice (n=15). J. Paired t-test of percentage heteroplasmy reduction for CD8 + T central memory vs. naïve in adult mice (n=14). K. Paired t-test of percentage heteroplasmy reduction for B-1a vs. naïve B-2; shown in red for five young animals. B-1a cells from young animals were compared to peritoneal naïve B-2 cells, otherwise splenic naïve B-2 cells were used (n=11). L. Paired t-test of percentage heteroplasmy reduction for B-2 IgG + memory vs. naïve B-2 (n=16). . CC-BY 4.0 International license It is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted October 7, 2021. ; carried out in 96-well U-bottom plates with 250,000 cells/well and a 1:1 bead:cell ratio in the presence of 100 U/ml IL-2. B. A cell trace violet (CTV; proliferation dye) pseudocolor dot-plot from ex vivostimulated naïve CD4 + T cells from a representative mouse after 5 days is shown. Raw heteroplasmy frequency is plotted for cells in each division (0-5) sequenced after FACS isolation (n=6). C. A cell trace violet (CTV; proliferation dye) pseudocolor dot-plot from ex vivo stimulated naïve CD8 + T cells from a representative mouse after 5 days is shown. Raw heteroplasmy frequency is plotted for cells in each division (0-7) sequenced after FACS isolation (n=6). D. Paired t-test of C5024T heteroplasmy burden reduction in ex vivo naïve CD4 + T cells vs. stimulated cells sorted from the last division after 5 days stimulation (n=13). E. Paired t-test of C5024T heteroplasmy burden reduction in ex vivo naïve CD8 + T cells vs. stimulated cells sorted from the last division after 5 days stimulation (n=11). F. Heteroplasmy burden reduction is shown for purified unstimulated (naïve ex vivo) or stimulated naïve CD4 + T cells cultured in galactose (Gal) or glucose (Glu) media with IL-2 for 5 days. After 5 days, FACS was used to isolate cells from division (DV) 0 or the last division (n=3). G. Raw heteroplasmy percentage distribution in single CD8 + T cell clones from a single young mouse, compared to the peripheral CD8 + pool and the ear baseline value. Data are representative of multiple animals (n=5). H. Clonally expanded single CD8 + T cells after 7 days of culture post-stimulation were classified as belonging to large (>200 cell/well) or small (<200 cell/well) clonal populations. Total DNA from each well was then subjected to pyrosequencing. The percentage heteroplasmy reduction relative to ear is plotted for each clonal population and the naïve CD8 + T cell pool after activation. I. Schematic of conditions used to stimulate murine B cells in vitro. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted October 7, 2021. ; Figure 3 -C5024T dysregulates naïve CD8 + T cell activation A. Seahorse Mito Stress Test of ex vivo naïve CD8 + T cells isolated from young mice (n=6 C5024T, n=5 WT) and stimulated for 6 h with anti-CD3/CD28 beads and IL-2 before analysis. Data were normalized according to protein amount/well. B. Seahorse Mito Stress Test after re-stimulation CD8 + T cells isolated from young mice (n=5 C5024T, n=4 WT). After 8 days of culture with activation at time 0, T cells were counted and re-stimulated for 6 h (anti-CD3/CD28 + IL-2) before Seahorse. C. Nanostring metabolic gene module analysis of mRNA isolated from CD8 + T cells from young C5024T (n=3) and WT (n=4) mice stimulated for 6 h with anti-CD3/CD28 + IL-2. Hierarchical clustering of principle component differences in normalized gene expression per module is shown. D. Proportion of IFN-γ + cells in the latest division, assessed by intracellular flow cytometry after PMA/Ionomycin re-stimulation of CD8 + T cells previously expanded for 8 days with anti-CD3/CD28 and IL-2. C5024T (n=8) and WT (n=6). E. Serum IFN-γ levels in C5024T (n=3 adult, n=8 young) and WT (n=3 adult, n=8 young) animals measured by electrochemiluminescence. . CC-BY 4.0 International license It is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted October 7, 2021. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted October 7, 2021. ; https://doi.org/10.1101/2021.10.05.21264464 doi: medRxiv preprint Figure 4 -SARS-CoV-2 Spike trimer vaccination of C5024T mice A. Schematic of the vaccination timeline and readouts. n=10 C5024T and n=10 ageand sex-matched WT animals (2-months-of-age) were immunized. One C5024T mouse was sacrificed during the experiment (week 8) due to developing a wound. B. Cryo-EM structure of the SARS-CoV-2 Spike protein from Wrapp et al. 75 , showing the smaller, neutralization-sensitive receptor-binding domain (RBD) in green. C. Anti-S and anti-RBD IgG 5 weeks post-prime (pre-boost) and 5 weeks post-boost. Repeat analysis of unvaccinated controls (n=2 WT, n=2 C5024T) are shown as open circles. D. ID50 neutralizing titers for vaccinated mice sera. C5024T (n=9) shown in red, WT (n=9) shown in blue. Unvaccinated controls (n=4) are shown at the assay limit of detection. One WT datapoint was excluded from group analysis for being an outlier (>20-fold above the group mean: ID50 = 20,059). E. Flow cytometry strategy for isolation of vaccine-induced class-switched antigenspecific IgG + B cells from the draining lymph node of vaccinated C5024T mice (n=6). Spike-binding cells in unvaccinated did not class-switch to IgG, and therefore, likely represent naïve B cells reacting with the probe in vitro. Sorting gates are shown in bold outline. F. Heteroplasmy burden reduction (vs. ear) for sorted B cell subsets from vaccinated C5024T mice (n=6). Naïve: CD3 -CD19 + B220 + IgD hi IgM hi IgG -; Memory (Ag-specific): CD3 -CD19 + B220 + IgD -IgM -IgG + S + ; Memory (non-Ag-specific): CD3 -CD19 + B220 + IgD -IgM -IgG + Spike -. Ear heteroplasmy at weaning for each animal is shown in red brackets. G. Paired t-test of percentage heteroplasmy burden reduction for antigen-specific IgG + cells vs. naïve cells. H. Frequency of S-specific B cells from C5024T (n=10) and WT (n=9) mice plotted as the frequency of Ag-specific from total B-2 cells. . CC-BY 4.0 International license It is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted October 7, 2021. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted October 7, 2021. ; https://doi.org/10.1101/2021.10.05.21264464 doi: medRxiv preprint Figure 5 -Purifying selection of A3243G in memory T and B lymphocytes from human MELAS patients A. Schematic representation of the human A3243G mutation in mt-tRNA Leu . B. FACS strategy for isolation of key human T lymphocyte populations from MELAS patients. Key T cell subsets are shown, the sorting strategy for all lymphoid and myeloid lineages is shown in the supplementary figures. Subset markers for FACS are defined in the main text and Materials and Methods. C. A3243G heteroplasmy burden reduction vs. skeletal muscle (sm) is shown for Patient A (male, 16-20 years-of-age). Skeletal muscle heteroplasmy level in patient: 95%. D. A3243G heteroplasmy burden reduction vs. skeletal muscle (sm) is shown for Patient B (female, 51-55 years-of-age). Skeletal muscle heteroplasmy level in patient: 90%. E. A3243G heteroplasmy burden reduction vs. skeletal muscle (sm) is shown for Patient C (female, 51-55 years-of-age). Skeletal muscle heteroplasmy level in patient: 65%. F. FACS-isolation of antigen-specific memory B cells from Patient B (5 weeks after the first dose of an adenovirus COVID-19 vaccine). Antigen-specific cells were defined as: CD3 -CD20 + CD27 + IgG + Spike + and compared to naïve (CD3 -CD20 + CD27 -IgM hi IgG -S -) cells. The pseudocolor plot of the CD3 -CD20 + CD27 + B cell population is shown, along with the sorting gate. G. A3243G heteroplasmy frequency (percentage) in naïve and SARS-CoV-2 S-specific B cells from MELAS Patient B. . CC-BY 4.0 International license It is made available under a perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted October 7, 2021. ; https://doi.org/10.1101/2021.10.05.21264464 doi: medRxiv preprint The dynamics of mitochondrial DNA 1076 heteroplasmy: implications for human health and disease Mitochondrial respiratory-chain diseases Myocardial Function of the ESC Purifying Selection against Pathogenic Mitochondrial DNA 1173 in Human T Cells Chronic inflammation in the etiology of disease across the life span Human memory T cells: 1177 Generation, compartmentalization and homeostasis CD28 is 1180 an inducible T cell surface antigen that transduces a proliferative signal in 1181 CD3+ mature thymocytes Segregation and Manifestations of the mtDNA tRNALys A-G 1184 8344 Mutation of Myoclonus Epilepsy and Ragged-Red Fibers Heteroplasmy levels of a mitochondrial gene mutation 1188 associated with diabetes mellitus decrease in leucocyte DNA upon aging Decrease of 3243 1191 A→G mtDNA mutation from blood in MELAS syndrome: A longitudinal study Varying loads of the mitochondrial DNA A3243G mutation in 1194 different tissues: Implications for diagnosis Selection against 1197 Pathogenic mtDNA Mutations in a Stem Cell Population Leads to the Loss of 1198 the 3243A→G Mutation in Blood Metabolic regulation of 1200 stem cell function in tissue homeostasis and organismal ageing The requirement for pyruvate dehydrogenase in leukemogenesis 1203 depends on cell lineage B1a B 1207 cells require autophagy for metabolic homeostasis and self-renewal Murine CD27(−) Vγ6(+) γδ T cells producing IL-17A promote 1212 ovarian cancer growth via mobilization of protumor small peritoneal 1213 macrophages Interleukin-17-1215 Producing γδ T Cells Selectively Expand in Response to Pathogen Products 1216 and Environmental Signals Cutting Edge: CD4 and CD8 T Cells Are Intrinsically 1218 Different in Their Proliferative Responses Metabolic reprogramming toward oxidative phosphorylation 1220 identifies a therapeutic target for mantle cell lymphoma Mitochondrial Biogenesis and Proteome Remodeling 1223 Promote One-Carbon Metabolism for T Cell Activation Polyamine metabolism is a central determinant of helper 1226 T cell lineage fidelity Pathway level analysis of gene expression using singular 1228 value decomposition T cell apoptosis by kynurenines Modulating 1232 T Cell Responses via Autophagy: The Intrinsic Influence Controlling the 1233 Function of Both Antigen-Presenting Cells and T Cells Quantitative Variation in m.3243A > G Mutation Produce 1236 Interferon gamma (IFN-γ) disrupts energy expenditure and 1238 metabolic homeostasis by suppressing SIRT1 transcription SARS-CoV-2 protein subunit vaccination of mice and 1241 rhesus macaques elicits potent and durable neutralizing antibody responses Driving potent neutralization of a SARS-CoV-2 Variant of 1246 Concern with a heterotypic boost Thymic output in aged mice Mitochondrial complex III is essential for regulatory T cell 1253 suppressive function Memory T and memory B cells share a transcriptional 1258 program of self-renewal with long-term hematopoietic stem cells The emerging role of immune 1261 dysfunction in mitochondrial diseases as a paradigm for understanding 1262 immunometabolism Oxygen tension modulates the mitochondrial genetic 1264 bottleneck and influences the segregation of a heteroplasmic mtDNA variant in 1265 vitro Diversification Occurs before, during, and after the First T Cell Division Cryo-EM structure of the 2019-nCoV spike in the prefusion 1270 conformation 2656-2668 (2003) .