A Dynamical Motif Comprising the Interactions Between Antigens And

Home , Immunopathology

Downloaded by guest on September 25, 2021 ct infection acute or infections persistent clear to routes immunopathology. interactions present limit the may modulate motif that the clear- Interventions in spontaneous C. and the hepatitis in outcomes, response of tuning immune ance rheostats innate as the of of act role ability to the the cells pathology, advan- (NK) lethal killer evolutionary preventing natural the in size, exhaustion inoculum in of tage viral variation the con- the of with including number outcomes the a observations, explains understand experimental also to It founding framework infection. of conceptual outcomes a variable observed. presents outcomes motif the compris- of motif The all the recapitulates that interactions these show ing we however, function. modeling, effector stimulation, mathematical turn, compromising Using antigen exhaustion, in T-cell which Sustained the CD8 cells, cause determining cells. can T predominantly infected CD8 as kill activate it can can posit antigen and Viral cells outcomes. antigens T between CD8 interactions dynami- essential and simple the a comprising identify motif we cal Here, different cytokines). (virus, and factors cells, different of immune number interac- some, large complex a from the between of in tions consequence mortality a the as cause cleared that arise is outcomes view and different rapidly conventional The others, others. in yet be in in outcomes immunopathology infection can 2019) different 5, persistent February very They review cause for in (received individuals. 2019 23, culminate July different approved infections and CA, viral Jolla, La Some Institute, Research Scripps Oldstone, A. B. Michael by Edited www.pnas.org/cgi/doi/10.1073/pnas.1902178116 model mathematical and a infections Baral Subhasish viral underlie of may outcomes cells the T CD8 and antigens interactions between the comprising motif dynamical A h iu 7.Mr eety aua ilr(K el aebeen have cells (NK) killer natural cleared recently, rapidly control persis- More have the (7). in would virus otherwise to resulted the that critical deficiency mice of be T-cell infection to CD8 Soon tent identified 6). infections: survival, (5, were LCMV virus to cells the of led of T titers inocula CD8 high large of after, with persistence the infections with and albeit death; to leading immunopathology, severe caused infections inocula cleared; intermediate were with inoculum virus Infec- small (5). a infection of with LCMV effect initiated of tions the outcome showed the ago on determine y size that 25 inoculum viral over factors study the seminal A understand outcomes. better the to used consequently, been and have outcomes these of many recapitulate infection given a what with and infection individuals, outcome? can different the in Why determines outcomes (4). different others have in virus not some but in respiratory immunopathology children example, severe causes for in of infection vary; extent virus infection also syncytial The can (3). persistent infection adults causes after in pathology self-limiting (HBV) typically virus is but B neonates Hepatitis (HCV), 2). infected chronically virus (1, become rest C the whereas hepatitis spontaneously, tion mortal- With host instance, immunopathology. and for to persistence, due long-term ity infection, acute after V eateto hmclEgneig ninIsiueo cec,Bnaoe501,India; 560012, Bangalore Science, of Institute Indian Engineering, Chemical of Department uiemdl flmhctccoimnnii iu (LCMV) virus choriomeningitis lymphocytic of models Murine c etefrBoytm cec n niern,Ida nttt fSine aglr 602 India 560012, Bangalore Science, of Institute Indian Engineering, and Science Biosystems for Centre iul.Trefeunl bevdotoe r clearance are indi- outcomes different observed in frequently outcomes Three different viduals. have can infection iral | 5 fteidvdasifce la h infec- the clear infected individuals the of ∼25% hoi infection chronic a | utmAntia Rustom , bistability | immunopathology b n aedaM Dixit M. Narendra and , | a,c,1 ulse nieAgs 4 2019. 14, August online Published 1073/pnas.1902178116/-/DCSupplemental. y at online information supporting contains article This 1 oeiaie ies . C BY-NC-ND) (CC 4.0 NoDerivatives License distributed under is article access open This Submission.y Direct PNAS a is article This interest.y of conflict data; no declare analyzed authors N.M.D. paper.y The and the and wrote R.A., R.A., N.M.D. S.B., and S.B., R.A., research; tools; S.B., and performed reagents/analytic S.B. new research; contributed N.M.D. designed S.B. contributions: Author CD8 the of T-cell of strengths CD8 strength the relative cause with together The can influences 20–22). competing duration these (5, anti- extended exhaustion of an via level for dysfunction high present (18, a cells is when infected gen however, the kill stimulation, can Sustained that response 19). histo- a triggers major CD8 and activates I cells cells T class infected Antigen on by cells. molecules presented complex T compatibility peptides CD8 outcomes. viral on the via influences recapitulate stimulation competing to may adequate exert interactions and Antigens latter necessary the both that follows be suggest reasoning that Our to observations infections. central as viral from it of posit outcomes and the cells T defining CD8 and antigens of interactions is it determin- and in reached. (14–17), is which infections involved determines outcome factors other these be between and interplay to the this how unclear seems of outcomes thus the factors (13) ing infection. of genetics LCMV of host plethora outcomes recently, the A influence most to factors argued and interferon been (8), 12), I have cells type (11, 10), T signaling (9, regulatory dysfunction (IFN) T-cell studies, CD8 other to rapid In contributing to LCMV pathology (6). of of death levels dose low depletion host high with virus a cell the with of NK infection persistence outcomes: from of outcome tuning the rheostats change as can act to argued owo orsodnemyb drse.Eal [email protected] Email: addressed. be may correspondence whom To rpssinterventions. and The proposes outcomes, observations, the experimental interactions. confounding of several understanding essential explains conceptual a these presents motif modulating outcomes the influence by anti- factors viral other indirectly The between cells. T interactions CD8 and essential gens motif few, dynamical a a numer- by comprising determined we are of contrast, outcomes striking the interplay In that argue outcomes. complex the determines a factors ous that been have suggesting outcomes the identified, influence chal- inoculum that viral genetics, what longstanding host including and Identifying a factors, size Many been damage. immunology. in has tissue lenge or outcomes to chronic, these due turn determines death spontaneously, cleared host be individ- cause different could in They outcomes uals. different have infections Viral Significance ee eietf yaia oi opiigteessential the comprising motif dynamical a identify we Here, b eateto ilg,EoyUiest,Alna A30322; GA Atlanta, University, Emory Biology, of Department PNAS | uut2,2019 27, August . y raieCmosAttribution-NonCommercial- Commons Creative | o.116 vol. www.pnas.org/lookup/suppl/doi:10. | o 35 no. | 17393–17398

IMMUNOLOGY AND INFLAMMATION T-cell response seem to deﬁne the outcomes realized. While a AB100 strong CD8 T-cell response is critical to viral clearance (7), per- Infected CD8 T ) sistent infections are typically associated with a dysfunctional cells cells max -2 CD8 T-cell response (7, 18, 23). Furthermore, T-cell exhaus- 10 Clearance

tion has been proposed to be an altered differentiation program I E Saddle point + 10-4 to combat chronic infection while preventing immunopathol- Persistence ogy (24–26). The exhausted state is maintained by inhibitory CD8 T cells (E/I molecules, such as PD-1 (10), and can be partly reversed by + 10-6 10-6 10-4 10-2 100 blocking the inhibitory molecules or lowering antigen burden Infected cells (I/I ) max (9, 10, 27), which in turn, can alter the disease state from per- CD E sistence to clearance (28). We hypothesized, therefore, that the 0 10 100 20 ) dynamical motif due to these underlying interactions governs the ) max outcomes. Furthermore, we suggest that the other factors that max 15 -2 influence the outcomes do so by modulating the interactions in 10 10-2 the motif. 10 -4 10 10-4 Results 5 CD8 T cells (E/I Infected cells (I/I The Dynamical Motif and Its Mathematical Model. To test the -6 -6 10 10 Cytokine pathology (A.U.) 0 hypothesis, we constructed the dynamical motif comprising the 01020 01020 01020 competing influences of antigen on CD8 T cells and the sup- Days post infection Days post infection Days post infection pressive influence of CD8 T cells on antigen (Fig. 1A). We F With exhaustion Without exhaustion ) developed the following minimal mathematical model to analyze ) max the motif: max

dI I (infected cells) = k1I 1 − − k2IE dt I max . [1] CD8 T cells (E/I dE IE IE CD8 T cells (E/I (CD8 response) = k3 − k4 dt kp + I ke + I

Fig. 1. Dynamical motif underlying the outcomes of viral infections. (A) We modeled the growth of infected cells I (the source of anti- Schematic of the motif depicting the essential interactions between infected genic stimulation) using a logistic term with a per capita prolifer- cells, I, and CD8 T cells, E. (Note that arrows and blunt arrows imply posi- ation rate k1 and a carrying capacity Imax. Infected cells are killed tive and negative regulation, respectively.) (B) Phase portrait of the system, by activated CD8 T cells, E, with the second-order rate con- where regions leading to the 2 stable steady states (filled red dots), clearance and persistence, are separated by a basin boundary (red line). The stant k2. Following previous models (29, 30), we described T-cell activation and exhaustion by infected cells using Hill functions unstable saddle (open red dot) underlies immunopathology. Also shown are trajectories with increasing initial antigen load (light to dark blue) and with maximal per capita rates k3 and k4 and scaling constants kp increasing initial effector pool size (light to dark pink) and the correspond- and ke , respectively. To ensure that activation happens at lower ing time course of (C) infected cells, (D) effector pool size, and (E) pathology. antigen levels than exhaustion and that the latter dominates at (F) Heat maps of peak pathology for different initial conditions on the phase high antigen levels, we let k3 < k4 and kp < ke (30). Thus, as portrait with and without exhaustion. Parameters values are in SI Appendix, antigen levels rise, they first activate CD8 T cells, whereas con- Table S1. tinued increase of antigen levels triggers CD8 T-cell exhaustion. Our key findings are robust to alternative formulations, where exhaustion is triggered by cumulative rather than instantaneous to one outcome from those that lead to the other. Infections antigen stimulation (24) (SI Appendix, Text S1 and Fig. S1). We initiated in the region above the basin boundary, which marks show below that the motif recapitulates the observed outcomes higher than a threshold CD8 T cells for a given antigen load, of viral infections. lead to viral clearance (Fig. 1 C and D). In contrast, infections initiated in the region below the basin boundary lead to Viral Clearance and Persistence as Steady States of the Motif. To persistence. describe the long-term outcomes of infection, we solved the model equations (Eq. 1) for its steady states. The model admits 3 Immunopathology as a Dynamical Outcome of the Motif. steady states. The steady states can be derived analytically when Immunopathology can result from a variety of processes asso- the exhaustion rate is far from saturation (ke I ) and can be ciated with the destruction of virus-infected cells or the release approximated using the mass action term k4IE. The resulting of cytokines resulting in a “cytokine storm.” Immunopathology steady states are (i) I = 0 and E ≥ 0, (ii) I = Imax and E = 0, is high when the infection spreads to a sizeable proportion k3 k1 1 k3 and (iii) I = − kp and E = (1 − ( − kp )). Using lin- k4 k2 Imax k4 of target cells, and there is a large population of CD8 T ear stability analysis, we found the first 2 steady states to be cells that kill these cells, directly producing tissue damage or stable, and the third to be an unstable saddle (SI Appendix, indirectly causing pathology due to cytokine release. We thus Text S2). The bistability existed across wide parameter ranges modeled immunopathology P with the equation dP/dt = αIE − (SI Appendix, Fig. S2). A phase portrait shows the 2 stable dc P, letting P increase in proportion to the rate of contact states separated by a basin boundary on which the saddle lies between I and E and recover with a natural first-order decay. (Fig. 1B). The first steady state has no antigen and thus, marks We found immunopathology to be maximum with intermediate the clearance of infection. The second, in contrast, has maxi- initial antigen load and CD8 T-cell counts (Fig. 1E). When mal antigen and no effector cells, which indicate suppression the initial antigen load is small, CD8 T cells eliminate antigen of the immune response and persistent infection. The basin before it can spread substantially. When the load is large, boundary (red line in Fig. 1B) separates conditions that lead CD8 T cells become exhausted before they can kill a sizeable

17394 | www.pnas.org/cgi/doi/10.1073/pnas.1902178116 Baral et al. Downloaded by guest on September 25, 2021 Downloaded by guest on September 25, 2021 neto,weesalrepplto e ocernei the in inter- With clearance 2B). to (Fig. (31) 13 led mutations clone escape population immune LCMV large of by absence persistent introduced a prevent small infection, whereas not of A infection, did start (6). the transfer, infected at adoptive persistently population T-cell became CD8 remaining the infection, and With 13 2A). clone to (Fig. with led (5) dose infected and infection, high infection, mice with the Docile of infection clone intermediate-dose 100% whereas or in Sizes. clearance, Docile persistence Pool to LCMV Effector led low-dose Initial 13 with and Inoculum infection Viral experiments, the of Influence outcomes. of the roles influence the to describe known can factors it key whether next examined We observed. persistence. viral host is enables however, and price, The viremia survival. high with associated 1F pathology (Fig. curtails sizes inoculum pool viral effector with and increased monotonically now pathology that aa tal. et Baral exhaus- of absence (k the tion in calculations this our test repeated To we argument, (24–26). immunopathology prevent the to adaptation near ary wide conditions 1F (Fig. a that pathology Spanning found most the states. we yielded boundary steady conditions, basin stable initial the of of range and devi- one antigen before toward both immunopathology triggering of trajecto- ating and buildup long levels a traverses T-cell causing it states. CD8 boundary, boundary, steady the the stable along to the ries close of starts one it into When settles quickly it boundary, yoieptooy aaee ausaetesm si i.1. Fig. in as same peak the (C and are (31). cells values sizes Parameter state-infected pool pathology. steady effector cytokine of (B) predictions and model different 6) Corresponding (5, with D) sizes outcomes inoculum of LCMV initial observations (A) Experimental sizes. pool effector 2. Fig. boundary basin the tissue to close 1 sizeable (Fig. commences infection in when results results which cells, T CD8 damage. continuously is antigen CD8 by and significant the spreads eliminated antigen trigger time, the can Thus, same that exhaustion. extents infection the T-cell to the At accumulate clearing not significantly. of does spreads capable it the not loads, is before antigen population intermediate T-cell At CD8 cells. infected of proportion Steady state CD Persistently infected (%) AB 100 h oi hsrcptltsteotoe fvrlinfections viral of outcomes the recapitulates thus motif The D -elehuto a enage ob nevolution- an be to argued been has exhaustion T-cell CD8 rmadnmclssesvepit immunopathology viewpoint, systems dynamical a From 20 40 60 80

infected cells (I/I ) 0 max 0.5 10 0 1 4 Initial infectedcells(I/I eedneo ucmso neto nvrliouu n initial and inoculum viral on infection of outcomes of Dependence B o eimHigh Medium Low -6 0 = Docile Docile Clone13 Clone13 and Initial viral load Initial CD8 Initial T cells viral load Initial .I otatt h aiu enaoe efound we above, seen maximum the to contrast In ). 10 .We h neto omne wyfo the from away commences infection the When E). -4 0 eaepritnl netd() whereas (5), infected persistently became ∼50% 10 -2 max 10 0 iddet immunopathology, to due died ∼20% 0 5 10 15 20 25 ) 0 0 20 40 60 80 100 Peak cytokine

pathology (A.U.) Death from pathology (%) exhaustion that demonstrating ), Steady state Persistently infected (%) 100 20 40 60 80

infected cells (I/I ) 0 max 0.5 10 1 0 0 iecerdthe cleared mice ∼50% Initial CD8Tcells(E/I o eimHigh Medium Low -6 10 -4 10 Clone13 ). -2 max 10 ) 0 5 10 15 20 25 0 0 20 40 60 80 100

and Peak cytokine In pathology (A.U.) Death from pathology (%) ncls hc otosvrlrpiaini netdclsand cells infected in replication viral of state controls antiviral expression which an the molecule, create cells, triggers collectively signaling in and that immune genes infection hundred innate viral several key after secreted a is IFN, triggered. is instance, response to T-cell For begins CD8 and the before infection control after viremic Response. exert early mounted Immune typically Innate is response the of tune Influence (6): to motif rheostats dynamical as the function in response outcome. cells the effector 3 NK the limit (Fig. how they influence explains minimal and a vations had cells C With NK 3D). (Fig. infection, depletion 3C). cell low-dose NK (Fig. after doses depleted and high doses were with intermediate increased cells with NK cells decreased when immunopathology NK Thus, boundary when basin closer boundary the moved infection to the high-dose the from Conversely, NK depleted. removed were with became boundary basin but the cells to intermediate- close the commenced chosen, infection parameters effect of dose the an killing For significant less outcomes.) direct have the to (The on seemed depletion. cells NK cell by NK cells NK infected with by response relieved effector was the of cells suppression the steady because the was 3C of This (Fig. attraction clearance of representing basin state the expanding boundary, basin 3 (Fig. and infection persis- with high-dose trend to with the led reversed immunopathology depletion heightened dose cell intermediate-dose NK high survival, cells, host and and NK tence immunopathology to normal led with infection whereas that, found h oi yltigN el ohkl netdclsadsuppress and 3B). cells (Fig. infected response kill T-cell both in CD8 cells effects the NK these letting incorporated by typi- We motif cells help. the T T-cell CD4 CD8 the on of compromise relies activation can cally because (6), which response cells, T-cell T CD8 CD4 eliminate cytolytically low-dose with inconsequential seem infection. cells NK 3A). (Fig. sur- infection and 100% dose in resulted intermediate cells with NK vival depleting 3A, contrast, (Fig. stark infection In persistent Inset). with but survived administered inocula mice high-dose 100% whereas immunopathology, infection to 13 due clone died response, LCMV intermediate-dose cell to NK subjected normal mice a of With (6). follows as outcomes Cells. NK of Influence motif. the of embellishments suitable interhost clearance and between effects split stochastic per- to even variations. due was the possibly damage in persistence, cells. tissue resulting and T the critical, CD8 infection, not of Docile haps transfer LCMV adoptive from the with increased more With 100% mortality found to host pre- without we infection, these ∼20% 13 boundary, with clone basin Consistent with the 1E). dictions, effec- (Fig. along higher immunopathology sizes associated to severe the moved pool conditions of cell initial criticality tor the the As and mortal- damage. severity Host tissue the 1). on (Fig. depends above predictions ity This our sizes. from pool understood effector initial is or inocula, maxi- inocula was intermediate fixed Immunopathology with 2D). mum With (Fig. small sizes with 2C). pool resulted effector persistence (Fig. initial and sizes large with for pool resulted inocula clearance viral effector large initial with persistence fixed and small with to clearance due died infected mice the of (31). all immunopathology sizes, pool effector mediate ovn h eutn qain ( equations resulting the Solving also can they However, cells. virus-infected kill can cells NK constructed we factors, other of influence the describe To predicted It observations. these recapitulated model Our and .O h hs otat eltn Kclssitdthe shifted cells NK depleting portrait, phase the On D). .Ormtftu eaiuae xeietlobser- experimental recapitulates thus motif Our D). PNAS Kclshv enfudt nunethe influence to found been have cells NK | uut2,2019 27, August 0 otlt ihhigh-dose with mortality ∼60% ,we S3), Text Appendix , SI and | o.116 vol. i.S3). Fig. Appendix, SI h naeimmune innate The | o 35 no. | ∼25% 17395 C

IMMUNOLOGY AND INFLAMMATION AB further that the rise of viremia during the acute phase was bi- 100 100 NK phasic with a sharp initial rise followed by a slower, more grad- 100 100 ual rise to the peak viremia. The arrest of the sharp initial rise 80 80 50 50 was due to the innate immune response. The decline after peak 60 60 viremia was attributable to the CD8 T-cell response (Fig. 5B, 0 0 40 40 Inset). The CD8 T-cell response was potentiated by the strong I E innate immune response, which suppressed viremia and limited 20 20 + exhaustion.

Persistently infected (%) 0 0

Death from pathology (%) Interestingly, the biphasic rise of viremia after infection has + Low Medium High been seen in chimpanzees infected with HCV (38). Of the C D 10 animals challenged, 6 had persistent infection, whereas 4 ) cleared the infection after the acute phase. We reasoned that -2 3 max 10 the earlier the transition from the ﬁrst- to the second-phase 10-2

-4 2 rise of viremia, the stronger the innate immune response must 10 be. The resulting peak viremia, which sets the antigen load 10-4 10-2 100 1 for the CD8 T cells, would be correspondingly lower, facili- 10-4 tating cure. From previous ﬁts to the viral load data (37), we CD8 T cells (E/I estimated the transition times and peak viral loads in each ani-

-4 -2 0 Cytokine pathology (A.U.) 0 10 10 10 0 5 10 15 20 Infected cells (I/I ) mal. We found, remarkably, that the 6 animals that suffered max Days post infection persistent infection had significantly delayed transition times Fig. 3. The influence of NK cells. (A) Outcomes observed experimentally in (mean ∼12 vs. ∼7 d; P = 0.035) and higher peak viral loads mice infected with different doses of LCMV clone 13 with (Inset) and with- (mean ∼6.3 vs. ∼5.8 log10 copies per milliliter; P = 0.038) out NK cells (6). (B) Schematic of the motif embellished to incorporate the than the 4 that cleared the infection (Fig. 5C). Our hypothesis influence of NK cells. (C) Projections of trajectories corresponding to low- of the innate immune response modulating the outcomes by low- (gray; Inset), medium- (black), and high-dose (blue) infection with (solid) ering the antigen load with which the CD8 T cells must contend and without (dashed) NK cells. Projections of the basin boundaries are in is thus consistent with the data. red. Three-dimensional versions are in SI Appendix, Fig. S3. (D) The corresponding time evolution of cytokine pathology. Parameter values are in SI Discussion Appendix, Table S1. The quest for an understanding of the outcomes of viral infections has led, over the years, to the identification of an array of prevents the infection of uninfected cells (32). The early innate factors, including the size of the viral inoculum (5), the innate immune response may be viewed thus as a way to suppress and NK cell responses (6, 34), and host genetic variations (13), viremia before the CD8 T-cell response is activated. A strong that can potentially influence the outcomes. Mathematical mod- innate immune response would result in a low antigen load els have been developed to explain the outcomes under specific at the onset of the CD8 T-cell response, which based on our circumstances, such as protection after vaccination (24), post- understanding of the motif above, would facilitate the clear- treatment control of HIV infection (30), and posttreatment cure ance of infection. To test this hypothesis, we constructed a of HCV infection (39). Here, we argue that underlying the phenomenological description of the innate immune response, effects of these factors is an essential dynamical motif (com- where the innate immune response is triggered by antigen and prising the key interactions between antigens and CD8 T cells) in turn, suppresses the growth of antigen (Fig. 4A). A key that governs the outcomes. Using minimal mathematical models difference from the CD8 T-cell response is the absence of prolif- of the motif, we recapitulated the different outcomes observed. erative amplification, the latter a hallmark of adaptive immune Broadly, when the CD8 T-cell response is strong, the infection is responses. cleared, whereas when the antigen induces significant CD8 T-cell Solving the resulting model equations (SI Appendix, Text S4), exhaustion, long-term persistence results. When the 2 effects are we found that a strong innate immune response resulted in comparable, substantial spread of infection and associated CD8 clearance and a weak response in persistence (Fig. 4B, Inset). T-cell killing ensue, causing tissue damage and pathology. Fur- Akin to the NK cell response, a strong innate immune response thermore, using suitable embellishments of the motif, we showed expanded the basin of attraction of the steady state repre- that other factors could influence the outcomes by modulating senting clearance (Fig. 4B and SI Appendix, Fig. S4). Unlike the NK cell response, however, which altered the CD8 T-cell response, the innate immune response modulated the outcomes by suppressing the antigen load and tilting the balance in favor AB X 100 of effector cells. We considered a realization of the latter ) 100 phenomenon next. max -2 10 10-2 10-4 Spontaneous Clearance of HCV Infection. Spontaneous clearance 01020 -4 of HCV infection has been attributed to strong IFN responses I E 10 D.P.I (33, 34). It is also known to require strong CD8 T-cell responses

+ CD8 T cells (E/I (35). To test whether spontaneous clearance is a manifestation 10-6 -6 -4 -2 0 of the inﬂuence of the innate immune response described above, 10 10 10 10 Infected cells (I/I ) we constructed a model of HCV dynamics that included the max innate immune response. The model combined standard models of HCV kinetics (36, 37) with our motif above (Fig. 5A). Fig. 4. Inﬂuence of the innate immune response. (A) Schematic of the motif with the embellishment representing the innate immune response, . SI Appendix X We solved the resulting equations ( , Text S5) using (B) Projections of trajectories and time evolution of pathogen load (Inset) parameter values representative of HCV infection. We found with strong (dashed) and weak (solid) innate immune responses. D.P.I., days that, with a weak innate immune response, the infection became post-infection. Red lines represent the projections of the corresponding persistent, whereas with a strong innate immune response, the basin boundaries. Three-dimensional versions are in SI Appendix, Fig. S4. infection was cleared after an acute phase (Fig. 5B). We observed Parameters values are in SI Appendix, Table S1.

17396 | www.pnas.org/cgi/doi/10.1073/pnas.1902178116 Baral et al. Downloaded by guest on September 25, 2021 Downloaded by guest on September 25, 2021 aa tal. et Baral ml ecnaeo I-netdidvdastetderywith early treated individuals a HIV-infected in seen of been sug- percentage also also has small as control load Posttreatment (39). antigen earlier in gested 44). reduction sufficient of 43, underly- expectation a (39, our after mechanisms with clearance detectable investigated however, The being consistent, with is (42). are phenomenon The individuals cure treatment clear- posttreatment of the many this end to ing in the lead infection at to virus HCV recently found direct-acting of been the per- ance with has from treatment outcome agents Indeed, the antiviral alter clearance. could to which sistence cancer levels, antigen during lower inhibitors checkpoint with (41). treatments seen to antigen immune-related specific broad be the viral pathologies from to distinct significant likely thus, the and is clearance. tissues of pathology Reinvigorating infected to The presence note. pathology. it the increase cautionary may drive in a persis- to response sounds toward push immune also boundary larger study basin much the Our a from requiring away likely tence, is veered infection the have Furthermore, higher. exhaus- to be of to level expected clearance. the is however, to basin tion progressed, persistence has the from infection across the outcome After infection the alter the and push boundary could under strength circumstances effector in these increase dynam- small a more a From viewpoint, be exhaustion. systems prevent, T may ical not if inhibitors CD8 reverse, checkpoint antigen-specific to able immune of readily stage, pools this small At and cells. checkpoint with loads commence immune (40). typically antigen using Infections cancers small to infection. in and advantage actively early an infections inhibitors being predicts persistent is study strategy treating Our The for exhaustion. explored and T-cell receptors CD8 inhibitory block reverse to inhibitors checkpoint immune limit or infections persistent clear the to in immunopathology. routes interactions implication the present An manipulate of may emerges. can motif understanding that thus interventions conceptual infections that A viral is of motif. outcomes the the in interactions the persis- or (black) clearance One-tailed infection. in (Right P the and resulting of viremia (red) HCV tence of with rise infected second-phase chimpanzees to first- of the course in Time given are (B) S2 used dynamics. Parameters responses. viral immune innate including (Inset and motif load the viral of Schematic (A) 5. Fig. B

values. Viral load, copies/mL (C . 10 sa lentv taey niia rg ol eue to used be could drugs antiviral strategy, alternative an As use to is today interest great of strategy intervention One 10 10 10 -2 1 4 7 xeietlosrain 3,3)o (Left of 38) (37, observations Experimental ) 204060800 naeimn epneadteotoe fHVinfection. HCV of outcomes the and response immune Innate

Days postinfection Effector population 0.2 0.4 0.6 0 A 050 T fetrpo iefrsrn bak n ek(red) weak and (black) strong for size pool effector ) + V C t et iheulvrac yielded variance equal with tests

Transition time, days X I 10 15 20 25 5 ct Chronic Acute p=0.035 rniintm from time transition ) ekvrlla in load viral peak ) E + + Table Appendix, SI Peak viral load, copies/mL 10 10 10 E 5 6 7 ct Chronic Acute p=0.038 nemdaeiouamyotmlysiuaeC4Tcells, T CD4 infection. stimulate of optimally control poor may and sub- inocula in CD4 responses resulting Intermediate cases, their T-cell delayed, both trigger or CD8 weakly In compromised may optimal (6). be sizes killing sizes could specu- inoculum cell help large T-cell NK We small whereas and/or that (14). (55) cells, exhaustion context, T chimpanzees size latter CD4 inoculum of stimulate the on under- infection in outcome also HBV may the late, and of with influence 7) dependence to seen (6, inverse found infection the motif been LCMV our lie of have in outcomes interactions They the the outcomes. block- Sim- influence modulate IFN (11–13). may infections cells and by persistent T relieved is of CD4 be immune- here, clearance ilarly, can achieve and considered and to antiviral not infection ade For in both effect, variations. influence late latter have above dominant the The to the describe effects. known underlie suppressive to is may necessary IFN that embellish- such instance, are factors further factors, motif other responses, key of our immune of innate of roles instances and the ments some cells addressed in NK have pathology as we to responses While contribute cyto- T-cell (54). Virus-induced also viruses. CD8 influenza may influenza and pathicity 1918 dictated other responses, with (16) the immune outcomes innate titers the of with together viral lethality latter responses the IFN High be (53), the Poor to (17). in study. seem and virus implicated our which clearance in been for considered outcomes, have responsible those 2 infec- to factors Other admit similar the infec- (52). may HIV mortality, persistence influenza, or host virus to (51), like measles leads cleared with tions, extreme, invariably always as the which nearly such In tion, realized, is (13). be which the virus may infection, in same outcome differences single the the a with in Inter- the implicated infections. realized may of been other quantified, outcomes strengths have be with variations the to observed host in remain frequencies as which the (33). motif, well explain infection our as HCV in factors of interactions other clearance fre- in spontaneous the Variations explain the IFN to of of argued strength been quency the has of frequen- individuals distribution across the A responses occur. describe they not which does with cies it observed, outcomes different feedback. double-negative including motif involved in thus more embedded infections a pleiotropy viral paradoxical of of manifestation Outcomes which a 50). represent by cells, (6, responses cells NK effector T of compromise CD4 also role destroying and the para- cells in infected Furthermore, kill also can (49). evident is also networks pleiotropy is biological which doxical of many latter in infected the coupled loop, present kill feedback pleiotropy can double-negative paradoxical cells a represents with T the thus that CD8 motif is stimulus; Our motif cells. the our of suppress feature can additional target sup- An and cells. activate T both paradoxical CD8 can press antigen recognize where We motif, argued our 47). in been pleiotropy (46, has processes robustness and differentiation pleiotropy concentrations death in paradoxical cell the cytokine homeostatic Such maintain as the help cells. well to instance, T as includ- proliferation CD8 For the of systems, (46–48). induce been biological circuits can have interleukin-2 in cell target, prevalent immune its ing suppress designs and as activate identified both can stimulus large in result can and pool again control. exhaustion cell compromising causing latent reactivation, after small large viremia a of A surges from (30). happens pool is reignition cell when T hypothesis latent control CD8 enables exhausted One and reinvigorates posttreatment. which cells viremia, infection lowers cells ART responses infected reignite that immune latently host can and of ART and escape because which HCV HIV, by than formed involved however, more interpretation, The is (45). (ART) therapy antiretroviral u td a iiain.Wieorsuyrcptltsthe recapitulates study our While limitations. has study Our a where elements, paradoxical studies, elegant of series a In PNAS | uut2,2019 27, August | o.116 vol. | o 35 no. | 17397

IMMUNOLOGY AND INFLAMMATION leading to potent CD8 T-cell responses and rapid viral clearance, clear persistent infections by manipulating the interactions in consistent with observations (14). Finally, whereas our model our motif. lets CD8 T-cell exhaustion be fully reversible, recent studies ACKNOWLEDGMENTS. The study was supported by NIH Grants U19 have identiﬁed subsets of exhausted CD8 T cells that cannot AI117891 (to R.A.) and R01AI110720 (to R.A.) and Wellcome Trust/DBT be reversed, especially with checkpoint inhibitors (40). These (Department of Biotechnology) India Alliance Senior Fellowship IA/S/14/ subsets must be considered in devising accurate strategies to 1/501307 (to N.M.D.).

1. J. Micallef, J. Kaldor, G. Dore, Spontaneous viral clearance following acute hepatitis 28. D. L. Barber et al., Restoring function in exhausted CD8 T cells during chronic viral C infection: A systematic review of longitudinal studies. J. Viral Hepatitis 13, 34–41 infection. Nature 439, 682–687 (2006). (2006). 29. S. Bonhoeffer, M. Rembiszewski, G. M. Ortiz, D. F. Nixon, Risks and benefits of struc- 2. B. Hajarizadeh, J. Grebely, G. J. Dore, Epidemiology and natural history of HCV tured antiretroviral drug therapy interruptions in HIV-1 infection. AIDS 14, 2313–2322 infection. Nat. Rev. Gastroenterol. Hepatol. 10, 553–562 (2013). (2000). 3. D. Ganem, A. M. Prince, Hepatitis B virus infection—natural history and clinical 30. J. M. Conway, A. S. Perelson, Post-treatment control of HIV infection. Proc. Natl. Acad. consequences. N. Engl. J. Med. 350, 1118–1129 (2004). Sci. U.S.A. 6, 5467–5472 (2015). 4. H. Nair et al., Global burden of acute lower respiratory infections due to respiratory 31. J. N. Blattman, E. J. Wherry, S. J. Ha, R. G. van der Most, R. Ahmed, Impact of epitope syncytial virus in young children: A systematic review and meta-analysis. Lancet 375, escape on PD-1 expression and CD8 T-cell exhaustion during chronic infection. J. Virol. 1545–1555 (2010). 83, 4386–4394 (2009). 5. D. Moskophidis, F. Lechner, H. Pircher, R. M. Zinkernagel, Virus persistence in acutely 32. O. Haller, G. Kochs, F. Weber, The interferon response circuit: Induction and infected immunocompetent mice by exhaustion of antiviral cytotoxic effector T cells. suppression by pathogenic viruses. Virology 344, 119–130 (2006). Nature 362, 758–761 (1993). 33. V. Venugopal, P. Padmanabhan, R. Raja, N. M. Dixit, Modelling how responsiveness 6. S. N. Waggoner, M. Cornberg, L. K. Selin, R. M. Welsh, Natural killer cells act as to interferon improves interferon-free treatment of hepatitis C virus infection. PLoS rheostats modulating antiviral T cells. Nature 481, 394–398 (2012). Comput. Biol. 14, e1006335 (2018). 7. M. Matloubian, R. J. Concepcion, R. Ahmed, CD4+ T cells are required to sustain 34. R. Raja, S. Baral, N. M. Dixit, Interferon at the cellular, individual, and population level CD8+ cytotoxic T-cell responses during chronic viral infection. J. Virol. 68, 8056–8063 in hepatitis C virus infection: Its role in the interferon-free treatment era. Immunol. (1994). Rev. 285, 55–71 (2018). 8. J. M. Lund, L. Hsing, T. T. Pham, A. Y. Rudensky, Coordination of early protective 35. S. F. Abdelwahab, Cellular immune response to hepatitis-C-virus in subjects without immunity to viral infection by regulatory T cells. Science 320, 1220–1224 (2008). viremia or seroconversion: Is it important? Infect. Agent Cancer 11, 23 (2016). 9. S. S. Tay et al., Antigen expression level threshold tunes the fate of CD8 T cells during 36. A. U. Neumann et al., Hepatitis C viral dynamics in vivo and the antiviral efficacy of primary hepatic immune responses. Proc. Natl. Acad. Sci. U.S.A. 111, E2540–E2549 interferon-α therapy. Science 282, 103–107 (1998). (2014). 37. H. Dahari et al., Mathematical modeling of primary hepatitis C infection: Noncytolytic 10. E. J. Wherry, M. Kurachi, Molecular and cellular insights into T cell exhaustion. Nat. clearance and early blockage of virion production. Gastroenterology 128, 1056–1066 Rev. Immunol. 15, 486–499 (2015). (2005). 11. J. R. Teijaro et al., Persistent LCMV infection is controlled by blockade of type I 38. M. E. Major et al., Hepatitis C virus kinetics and host responses associated with disease interferon signaling. Science 340, 207–211 (2013). and outcome of infection in chimpanzees. Hepatology 39, 1709–1720 (2004). 12. E. B. Wilson et al., Blockade of chronic type I interferon signaling to control persistent 39. S. Baral, R. Roy, N. M. Dixit, Modeling how reversal of immune exhaustion elicits cure LCMV infection. Science 340, 202–207 (2013). of chronic hepatitis C after the end of treatment with direct-acting antiviral agents. 13. M. B. A. Oldstone et al., Lymphocytic choriomeningitis virus clone 13 infection causes Immunol. Cell Biol. 96, 969–980 (2018). either persistence or acute death dependent on IFN-1, cytotoxic T lymphocytes (CTLs), 40. M. Hashimoto et al., CD8 T cell exhaustion in chronic infection and cancer: and host genetics. Proc. Natl. Acad. Sci. U.S.A. 115, E7814–E7823 (2018). Opportunities for interventions. Ann. Rev. Med. 69, 301–318 (2018). 14. S. Asabe et al., The size of the viral inoculum contributes to the outcome of hepatitis 41. M. A. Postow, R. Sidlow, M. D. Hellmann, Immune-related adverse events associated B virus infection. J. Virol. 83, 9652–9662 (2009). with immune checkpoint blockade. N. Engl. J. Med. 378, 158–168 (2018). 15. S. Duan, P. G. Thomas, Balancing immune protection and immune pathology by CD8+ 42. A. Kohli et al., Virological response after 6 week triple-drug regimens for hepatitis C: T-cell responses to influenza infection. Front. Immunol. 7, 25 (2016). A proof-of-concept phase 2a cohort study. Lancet 385, 1107–1113 (2015). 16. M. D. de Jong et al., Fatal outcome of human influenza a (H5N1) is associated with 43. A. Goyal et al., Modeling HCV cure after an ultra-short duration of therapy with direct high viral load and hypercytokinemia. Nat. Med. 12, 1203–1207 (2006). acting agents. Antivir. Res. 144, 281–285 (2017). 17. D. Kobasa et al., Aberrant innate immune response in lethal infection of macaques 44. T. H. T. Nguyen et al., The paradox of highly effective sofosbuvir-based combination with the 1918 influenza virus. Nature 445, 319–323 (2007). therapy despite slow viral decline: Can we still rely on viral kinetics? Sci. Rep. 7, 10233 18. R. M. Zinkernagel, P. C. Doherty, MHC-restricted cytotoxic T cells: Studies on the (2017). biological role of polymorphic major transplantation antigens determining T-cell 45. A. Saez-Ciri´ on´ et al., Post-treatment HIV-1 controllers with a long-term virological restriction-specificity, function, and responsiveness. Adv. Immunol. 27, 51–177 (1979). remission after the interruption of early initiated antiretroviral therapy anrs visconti 19. A. Crawford, E. J. Wherry, The diversity of costimulatory and inhibitory receptor study. PLoS Pathog. 9, e1003211 (2013). pathways and the regulation of antiviral T cell responses. Curr. Opin. Immunol. 21, 46. Y. Hart, Y. E. Antebi, A. E. Mayo, N. Friedman, U. Alon, Design principles of cell circuits 179–186 (2009). with paradoxical components. Proc. Natl. Acad. Sci. U.S.A. 109, 8346–8351 (2012). 20. A. J. Zajac et al., Viral immune evasion due to persistence of activated T cells without 47. Y. Hart et al., Paradoxical signaling by a secreted molecule leads to homeostasis of effector function. J. Exp. Med. 188, 2205–2213 (1998). cell levels. Cell 158, 1022–1032 (2014). 21. E. J. Wherry, J. N. Blattman, K. Murali-Krishna, R. van der Most, R. Ahmed, Viral per- 48. E. D. Sontag, A dynamic model of immune responses to antigen presentation predicts sistence alters CD8 T-cell immunodominance and tissue distribution and results in different regions of tumor or pathogen elimination. Cell Syst. 4, 231–241 (2017). distinct stages of functional impairment. J. Virol. 77, 4911–4927 (2003). 49. P. Padmanabhan, U. Garaigorta, N. M. Dixit, Emergent properties of the interferon- 22. E. J. Wherry, T cell exhaustion. Nat. Immunol. 131, 492–499 (2011). signalling network may underlie the success of hepatitis C treatment. Nat. Commun. 23. R. Ahmed, A. Salmi, L. D. Butler, J. M. Chiller, M. B. Oldstone, Selection of genetic 5, 3872 (2014). variants of lymphocytic choriomeningitis virus in spleens of persistently infected mice: 50. C. Kes¸mir, R. J. De Boer, Clonal exhaustion as a result of immune deviation. Bull. Role in suppression of cytotoxic T lymphocyte response and viral persistence. J. Exp. Math. Biol. 65, 359–374 (2003). Med. 160, 521–540 (1984). 51. B. Laksono, R. de Vries, S. McQuaid, W. Duprex, R. de Swart, Measles virus host 24. P. L. F. Johnson et al., Vaccination alters the balance between protective immu- invasion and pathogenesis. Viruses 8, 210 (2016). nity, exhaustion, escape, and death in chronic infections. J. Virol. 85, 5565–5570 52. C. A. Sabin, J. D. Lundgren, The natural history of HIV infection. Curr. Opin. HIV AIDS (2011). 8, 311–317 (2013). 25. M. Cornberg et al., Clonal exhaustion as a mechanism to protect against severe 53. K. A. Pawelek et al., Modeling within-host dynamics of influenza virus infection immunopathology and death from an overwhelming CD8 T cell response. Front. including immune responses. PLoS Comput. Biol. 8, e1002588 (2012). Immunol. 4, 1–8 (2013). 54. T. Daidoji et al., H5N1 avian influenza virus induces apoptotic cell death in 26. D. E. Speiser et al., T cell differentiation in chronic infection and cancer: Functional mammalian airway epithelial cells. J. Virol. 82, 11294–11307 (2008). adaptation or exhaustion? Nat. Rev. Immunol. 14, 768–774 (2014). 55. B. Raziorrouh et al., Inhibitory phenotype of HBV-specific CD4+ T-cells is characterized 27. D. T. Utzschneider et al., T cells maintain an exhausted phenotype after antigen by high PD-1 expression but absent coregulation of multiple inhibitory molecules. withdrawal and population reexpansion. Nat. Immunol. 14, 603–610 (2013). PLoS One 9, e105703 (2014).

17398 | www.pnas.org/cgi/doi/10.1073/pnas.1902178116 Baral et al. Downloaded by guest on September 25, 2021