FADD and caspase-8 control the outcome ofautophagic signaling in proliferating T cellsBryan D. Bella,b, Sabrina Leverriera,b, Brian M. Weista,b, Ryan H. Newtona,b, Adrian F. Arechigaa,b,1, Keith A. Luhrsa,b,2,Naomi S. Morrissettea, and Craig M. Walsha,b,3

aDepartment of Molecular Biology and Biochemistry and bCenter for Immunology, University of California, Irvine, Irvine, CA 92697-3900

Communicated by Anthony A. James, University of California, Irvine, CA, September 8, 2008 (received for review May 26, 2008)

Fas-associated death domain protein (FADD) and caspase-8 (casp8)are vital intermediaries in apoptotic signaling induced by tumornecrosis factor family ligands. Paradoxically, lymphocytes lackingFADD or casp8 fail to undergo normal clonal expansion followingantigen receptor cross-linking and succumb to caspase-indepen-dent cell death upon activation. Here we show that T cells lackingFADD or casp8 activity are subject to hyperactive autophagicsignaling and subvert a cellular survival mechanism into a potentdeath process. T cell autophagy, enhanced by mitogenic signaling,recruits casp8 through interaction with FADD:Atg5-Atg12 com-plexes. Inhibition of autophagic signaling with 3-methyladenine,dominant-negative Vps34, or Atg7 shRNA rescued T cells express-ing a dominant-negative FADD protein. The necroptosis inhibitorNec-1, which blocks receptor interacting protein kinase 1 (RIPkinase 1), also completely rescued T cells lacking FADD or casp8activity. Thus, while autophagy is necessary for rapid T cell prolif-eration, our findings suggest that FADD and casp8 form a feedbackloop to limit autophagy and prevent this salvage pathway frominducing RIPK1-dependent necroptotic cell death. Thus, linkage ofFADD and casp8 to autophagic signaling intermediates is essentialfor rapid T cell clonal expansion and may normally serve topromote caspase-dependent apoptosis under hyperautophagicconditions, thereby averting necrosis and inflammation in vivo.

apoptosis � autophagy � caspases � FADD � necroptosis

L igation of TNF-receptor family molecules leads to the assemblyof oligomeric structures termed death-inducing signaling com-

plexes (DISCs) (1). Comprised of the adaptor FADD, the cysteineprotease casp8, and the casp8-like molecule c-FLIP, DISC assemblyis essential for induction of apoptosis following CD95 ligation (2).Mice lacking casp8, FADD, or c-FLIP paradoxically fail to developand die after roughly ten days of gestation in utero, demonstratingprofound defects in cardiac and hematopoietic development (3–6).In the immune system, while DISC signaling is likely necessary tocontrol lymphoid homeostasis and prevent autoreactivity, the func-tions of these DISC proteins are complicated by their enigmaticroles during T cell proliferation and survival (7). Indeed, T cellsexpressing a dominantly interfering form of FADD (FADDdd) orthose lacking FADD, casp8, or c-FLIP fail to efficiently proliferatefollowing antigen receptor stimulation (4, 8–13). These defectswere not rescued by provision of interleukin-2 (IL-2), a potent T cellmitogenic cytokine, and the proliferative defects observed in thesemutant T cells were not due to defective IL-2 production (4, 14, 15).Curiously, the most profound requirement for FADD and casp8exists in CD8� T cells (14, 16), potentially as a consequence of theirrapid proliferative capacity (17). B lymphocytes also require FADDand casp8 for proliferation, but only to certain mitogens (18, 19).While anti-IgM cross-linking induced normal proliferation, TLR3and TLR4 agonists failed to induce expansion of FADD- andcasp8-deficient B cell populations.

A second cellular catabolic process termed autophagy pro-motes cell survival, although certain conditions lead to a form ofcell death termed type-II death (20). Autophagy involves theformation of autophagosomes, double-membrane vacuoles thatsurround intracellular proteins and organelles and fuse with

lysosomes to degrade the contents (21). Autophagosome for-mation proceeds through a ubiquitin-like conjugation processinvolving numerous autophagy-related factors termed Atg pro-teins, including Atg5, which is conjugated to Atg12, whileAtg8/LC3 (light chain associated protein 3) is lipidated to formLC3-II, a hallmark of mature autophagosomes (22). Autophagyand apoptosis are both responsible for the removal of damagedcells and organelles. Beyond this ability to remove unwantedcellular constituents, autophagy can act as a recycling system toproduce metabolic substrates from macromolecular structures intimes of nutrient or growth factor restriction (23).

Autophagy and apoptosis have been found to be linked underspecific circumstances (24). Likely since they both control theoutcome of cellular stress, it would be expected that there wouldexist several modes of crosstalk between apoptosis and autoph-agy. Consistent with this, the Bcl-2-binding factor Beclin-1/Atg6has been found to contribute both to regulation of apoptosis andautophagy (20). Also consistent with this is the finding that bothcasp8 and FADD modulate autophagic signaling (25–27). Re-cently, is has been reported that autophagy is both induced byand necessary for proper T cell proliferation following antigenreceptor ligation (28, 29). Given that both autophagy and DISCproteins are essential for mitogenic responses in T cells, wecharacterized the link between these pathways in FADDdd andcasp8�/� T cells. We also sought to determine whether adisruption in the linkage of these pathways might contribute tothe defects observed in T cells lacking FADD or casp8 activity.As shown here, FADD and casp8 are essential for tempering theautophagic response in primary T cells and murine embryonicfibroblasts.

ResultsSince previous work established that FADD and casp8 bothimpact autophagic signaling (25–27), we hypothesized thatFADDdd T cells might display autophagic defects. Consistentwith this, we recently reported defects in regulation of S6 kinasein FADDdd transgenic T cells (16), a key intermediate inautophagy regulation (30). Further, bromodeoxyuridine (BrdU)incorporation assays revealed diminished entry of FADDddtransgenic and casp8�/� T cells into S-phase and enhancedproportions of cells bearing subdiploid DNA content, a defectparticularly acute in FADDdd-expressing CD8� T cells bearingan ovalbumin (OVA)-specific OT-I TCR transgene (14, 15). Toinvestigate the morphology of activated FADDdd T cells, naïve

Author contributions: B.D.B., S.L., B.M.W., R.H.N., and C.M.W. designed research; B.D.B.,S.L., B.M.W., R.H.N., A.F.A., and K.A.L. performed research; N.S.M. contributed new re-agents/analytic tools; B.D.B., S.L., B.M.W., R.H.N., A.F.A., K.A.L., N.S.M., and C.M.W. ana-lyzed data; and B.D.B. and C.M.W. wrote the paper.

The authors declare no conflict of interest.

1Present address: Benaroya Research Institute at Virginia Mason, Seattle, WA 98101.

2Present address: Allergan, Irvine, CA 92612.

3To whom correspondence should be addressed. Email: cwalsh@uci.edu.

This article contains supporting information online at www.pnas.org/cgi/content/full/0808597105/DCSupplemental.

© 2008 by The National Academy of Sciences of the USA

www.pnas.org�cgi�doi�10.1073�pnas.0808597105 PNAS � October 28, 2008 � vol. 105 � no. 43 � 16677–16682

IMM

UN

OLO

GY

Dow

nloa

ded

by g

uest

on

July

1, 2

020

and 3-day activated OT1 and OT1-FADDdd (OT1-FD) T cellswere subjected to transmission electron microscopy (TEM)analysis (Fig. 1A, I–V). Activated OT1-FD T cells containednumerous large, double-membrane cytoplasmic vacuoles char-acteristic of autophagosomes (arrowheads) and a high level ofvacuolation (asterisks; Fig. 1 A, IV and V). These autophago-somes/autolysosomes enclosed other intracellular organelles(Fig. 1 A, V), and were mostly absent from naïve subsets and onlymodestly present in activated OT1 T cells [Fig. 1B and supportinginformation (SI) Fig. S1a]. A hallmark of autophagy, the cleav-age and lipidation of Atg8/LC3 allows for its deposition intoautophagosomal membranes. We observed a large increase inLC3-II, the autophagosome associated isoform, in OT1-FD Tcells activated for 3 days (Fig. 1C). To further expand theseresults, the level of autophagy was assessed in CD4� T cells.FADDdd mice were crossed onto the OTII background (31), andLC3 lipidation was assessed (Fig. 1D). While OTII-FD T cellspossessed increased levels of LC3-II compared with wild-type

(WT) OTII cells, the highest level of LC3 lipidation occurred inOT1-FD cells, consistent with the more profound cell cycle andsurvival defects in FADDdd CD8� T cells compared withFADDdd CD4� T cells (14). Similarly, when WT and FADDddT cells were activated in the presence or absence of Th1 and Th2skewing media, the presence of the transgene increased the levelof LC3-II and decreased the survival (Fig. S1 b and c). Nodecrease in survival was apparent in resting FADDdd T cells, nordid we observe differences in survival of FADDdd T cellsfollowing culture under low-serum conditions (Fig. S2 a and b).Since 48-h culture in low serum induced only modest levels ofLC3-II, our data suggests that these T cells succumb to a formof death promoted by dysregulated autophagy only after induc-tion of proliferation (Fig. S3 a and b). Treatment with thelysosomal protease inhibitors E64d and pepstatin A (pepA) forthe last four hours of culture in WT and FADDdd T cellsenhanced the overall detection of LC3, showing that this increasewas not a consequence of defective fusion and subsequentdegradation of autophagosomes within lysosomes (32) (Fig. S4).These data demonstrate that FADD represses autophagic in-duction in activated T cells.

To characterize the generality of this FADD signaling axis, weanalyzed the extent of LC3 processing in WT and FADD�/�

murine embryonic fibroblasts (MEFs) cultured for 24 h undernormal or low-serum conditions, the latter of which inducesautophagy (24). While WT MEFs responded to low serum byincreasing the expression of LC3-II, FADD�/� MEFs had ele-vated levels under both conditions, and this was diminished byretroviral introduction of shRNA to Atg7 (Fig. 1E). Reconsti-tution of FADD�/� MEFs with full-length FADD using retro-viral transduction decreased basal autophagy as measured byLC3-II processing after culture in media with 10% serum (Fig.1F). Consistent with the T cell results, treatment with E64d andpepA led to enhanced LC3-II detection in both subsets of MEFs(Fig. 1G). Using GFP-tagged LC3, we observed a much greaterproportion of punctate GFP-LC3 staining under both normaland autophagic (rapamycin-treated) conditions in FADD�/�

compared with WT MEFs (Fig. S5 a and b), suggesting that thelack of FADD in MEFs led to a similar dysregulation inautophagic signaling as observed in OT1-FD, OTII-FD, andFADDdd T cells. Although not as profound as for FADDdd Tcells, FADD�/� MEFs also displayed a reduced proliferativecapacity and enhanced cell death, especially following low-serumculture (Fig. S6 A and B). Similarly, FADD- and casp8-, but notRIPK1-deficient Jurkat clones (33–35), displayed enhancedlevels of LC3 processing compared to parental clones (Fig. S7).We noted only weak processing of LC3 in WT Jurkat clones inseveral experiments, even under low serum conditions, perhapsdue to hyperactive mTOR signaling that results in these PTEN-deficient cells. Even so, since we only observed the presence ofLC3-II in Jurkat T cells lacking casp8 or FADD, our results areconsistent with the hypothesis that both FADD and casp8 act inconcert to limit general macroautophagy in a number of celltypes.

At the apex of the autophagic signaling cascade, a complex ofproteins, including Atg6/Beclin-1, p150, and Vps34, is essentialfor autophagosome formation. To assess the role of hyperauto-phagy in the phenotype of FADDdd T cells, CFSE-labeled OT1and OT1-FD T cells were activated in the presence and absenceof 0.8 mM 3-MA, an inhibitor of Vps34 and autophagy (36). Lowdoses of 3-MA dramatically rescued cycling OT1-FD T cellaccumulation, as measured by CFSE dilution and live cellrecovery, while higher doses completely blocked both OT1 andOT1-FD T cell proliferation (Fig. 2A, Fig. S8 (Upper), and datanot shown), consistent with the reported findings that T cellsrequire limited autophagy for productive expansion and survival(29). Increasing doses of 3-MA had a modest effect on WT OT1proliferation but provided a near complete rescue of cell cycle

A B

C

D

E

I II

IVIII

VLC3-I

LC3-II

Tubulin

Naïve 24 hr 72 hr

OTIOTI-FD

++

++

++

15

49

10 10 0.50.5

Tubulin

%serum

Atg7

WT FADD-/-

scrambleshAtg7

++

++

++

++

15

49

82

LC3-ILC3-II

OTI-FD

OTI15

10

5

0Naïve 72 hr

*

auto

phag

asom

es/c

ell

15

49

LC3-ILC3-II

Tubulin

OTIOTI-F

D

OTIIOTII-

FD

LC3-I

LC3-II

Tubulin%serum

Empty FADD

10 100.5 0.5

FADD-/-

Virus

F WTFADD-/-

10 1010 10+ +- -E64d

PepA + +- -

%serum

LC3-I

LC3-II

Tubulin

+ ++ +G

Fig. 1. Increased autophagy in cells lacking FADD function. (A) Naïve (I andII), SIINFEKL activated (III–V), OT1 (I and III), and OT1-FD (II, IV, and V) cells wereanalyzed by TEM; representative micrographs presented (Scale bar: 0.5 �m).(B) Number of autophagosomes per cell section (*P value: 0.005). (C and D)Immunoblot showing extent of LC3 processing in naïve and activated (C), or3-day activated (D) OT1, OTII, OT1-FD, and OTII-FD cells [Molecular weight(arrow heads) in kDa]. (E) Immunoblot of WT and FAD�/� MEFs infected witheither scrambled or Atg7 shRNA with 10% or 0.5% serum. (F) Immunoblot ofFADD�/� MEFs infected with MIT-empty or MIT-FADD, followed by growthwith 10% or 0.5% serum. (G) Immunoblot of WT and FADD�/� MEFs grownwith 10% serum, with or without E64d and pepA for the last 4 h.

16678 � www.pnas.org�cgi�doi�10.1073�pnas.0808597105 Bell et al.

Dow

nloa

ded

by g

uest

on

July

1, 2

020

progression and survival of OT1-FD T cells (Fig. 2B). 3-MA alsoblocked the hyperexpression of LC3-II seen in OT1-FD T cellsactivated for 3d (Fig. 2C). To ensure that the rescue of OT1-FDT cells by 3-MA was not due to a potential off-target effect, weused retroviral transduction of a dominant-negative form ofVps34 (Vps34-KD) to block autophagic signaling (37). At 2-, 4-,and 6-d postinfection, cultures were harvested and live T cellswere assayed for Thy1.1 expression, a marker for the proportionof cells expressing the MiT retrovirus (38). Whereas infectionwith Vps34-KD or empty retroviruses led to only modest changesin the proportion of Thy1.1� OTI T cells over time, the propor-tion of Thy1.1� OTI-FD T cells expressing Vps34-KD increaseddramatically over the course of the assay (Fig. 2D). The meanfluorescence of Thy1.1 on recovered cells was selectively en-riched and the level of LC3-II was diminished in OT1-FD T cellsby Vps34-KD (Fig. 2E and F). These results demonstrate thatVps34-KD provides a selective advantage to proliferatingOTI-FD T cells. As an additional approach, we used RNAinterference to knock down expression of Atg7, an essentialautophagy-related gene that was hyperexpressed in OT1-FD Tcells, effectively reducing LC3-II to near WT levels (Fig. 2G).Using retrovirally encoded shRNA to knock down Atg7 expres-sion, we observed an increase in the recovery of proliferatingFADDdd T cells bearing shRNA-Atg7 relative to a scrambledretroviral control after culture in puromycin to enrich forretroviral transductants (Fig. 2H). Taken together, these datashow for the first time that blockade of autophagic signaling

rescues the survival and proliferative defects observed inFADDdd T cells.

The FADD death domain forms a complex with Atg5 topromote cell death, although the means by which this may occurremains uncertain (26). To more carefully classify the membersof this complex, we made use of a highly specific tandem affinitypurification (TAP) approach to characterize the in vivo bindingpartners of FADD in FADD�/� MEFs. For binding-partnerrecovery, a TAP tag (39) encompassing an HA epitope, IgG-binding domain, TEV cleavage site, and calmodulin-bindingdomain was fused to full-length FADD (Fig. S8 (Lower)).Infection of FADD�/� MEFs with TAP-FADD retrovirus andgrowth in 10% serum, a condition in which there is detectablelevels of basal autophagy in MEFs, led to expression of TAP-FADD equal to endogenous FADD in WT MEFs (Fig. 3A).Following TAP purification (39), TAP-FADD containing com-plexes were analyzed by Western blotting, revealing the expectedbinding between FADD and Atg5 (Fig. 3B). We found thatTAP-FADD also interacted with an Atg5:Atg12 covalently-linked complex, Atg16L, a protein bound to Atg5:12 on preau-tophagosomal membranes, and with casp8 and RIPK1, but notwith Atg7, LC3, or tubulin. Demonstrating that Atg5:FADDcomplexes specifically contained these DISC molecules (1),anti-HA immunoprecipitation of HA-Atg5 retrovirus infectedWT MEFs yielded FADD, casp8, and RIPK1 (Fig. 3C). Thesefindings suggest that FADD is involved in tethering casp8 andRIPK1 to autophagosomal membranes, where cellular

Fig. 2. Inhibition of autophagic signaling rescues FADDdd T cells. (A) Acti-vated CFSE-labeled CD8� T cell proliferation for 3 days, �0.8 mM 3-methy-ladenine. (B) Three-day activated T cells pulsed with BrdU for last hour ofculture. (C) Immunoblot of activated T cells at 3 days, �3-MA. (D–F) ActivatedT cells infected with Vps34-KD or empty retrovirus. (D) Fold change in Thy1.1expression compared to 2 days postinfection. (E and F) Thy1.1 expression andimmunoblot of Thy1.1� CD8� cells at 3 days postinfection. (G and H) Immu-noblot and live cell recovery of WT and FADDdd (FD) T cells infected withshAtg7 followed by growth in the presence of 10 �g/ml puromycin. (H)Expressed as a ratio of live shAtg7 cells over live empty-vector control cells (Allerror bars: �SD).

AC

B

Casp8

RIPK1

LC3-I

LC3-II

Atg5-12 (HA)

Atg5 (HA)

Tubulin

FADD

WCL IP

Mit-empty

Mit-HA-Atg5

+ +

+ +

15

49

64

26

82

37

49

HA

FADD

Tubulin

TAP-FADD +--

FADD-/-WT

26

49

49

26

49

WCL Post

Casp8

RIPK1

LC3-I

LC3-IIAtg5-12

Atg5

Atg7

HA

Tubulin

*

**

64

82

15

49

37

82

49

37

82Atg16L

DEVDIETD

30

10

20

Rat

io 3

6hr/

naïv

e

40

substrate

50FD

WTWT +3MA

FD +3MAWT +anti-Fas

48hr0hr

zVAD

LC3-ILC3-II

Tubulin

+--

15

49

D

E

Fig. 3. FADD, casp8, and RIPK1 form a complex with Atg5-Atg12/Atg16L. (A)Immunoblot to show expression level of TAP-FADD construct in MEFs. (B)FADD�/� MEFs grown in 10% serum were infected with Mit-TAP-HA-FADD for24 h, followed by cell lysis and TAP purification. Whole cell lysate (WCL) andpost-TAP-FADD purification immunoblot shown. Anti-HA antibody detectsTAP-HA-FADD (**) and TAP-HA-FADD post-TEV cleavage (*). (C) WT MEFsgrown in 10% serum were infected with Mit-empty or Mit-HA-Atg5 retrovi-ruses, followed by cell lysis. Immunoblot of WCL and anti-HA immunoprecipi-tated fractions. (D) IETDase (casp8) and DEVDase (casp3) activities in naïve,36-h activated, or 36-h activated � last 6 h with anti-FAS plate-bound antibodyWT and FADDdd T cells, �3-methyladenine. Activities expressed as ratios ofactivated versus naïve cells (All error bars: �SD). (E) Immunoblot of naïve and2-day activated OT1 T cells plus or minus 50 �M zVAD-FMK.

Bell et al. PNAS � October 28, 2008 � vol. 105 � no. 43 � 16679

IMM

UN

OLO

GY

Dow

nloa

ded

by g

uest

on

July

1, 2

020

Atg5:Atg12/Atg16L complexes are found, leading to the induc-tion of casp8 activity, a hypothesis supported by observation ofcolocalization of FADD and GFP-LC3 punctae (40). Consistentwith previous observations that caspases are enzymatically activein non-apoptotic T cells following mitogen stimulation (41, 42),we observed that activation of WT T cells with anti-CD3 plusanti-CD28 for 36 h led to a significant increase in IETDase(casp8), but not DEVDase (casp3) activity. Much of thisIETDase activity was blocked by 3-MA (Fig. 3D). Activationof OT1 T cells for 3 d in the presence of the pan-caspaseinhibitor zVAD-FMK also led to increased levels of LC3-II(Fig. 3E), suggesting that casp8 activity represses autophagicsignaling. These results are consistent with the hypothesis thatautophagic signaling induces an interaction betweenAtg5:Atg12 and FADD that promotes the activation of casp8in mitogenically stimulated live T cells (41, 42).

Since these results demonstrate the assembly of a RIPK1-containing, DISC-like structure via Atg5, we hypothesized thathyperautophagic cells lacking casp8 activity may die through aRIPK1-dependent/caspase-independent ‘‘necroptotic’’ pathway,similar to that observed when such cells are stimulated withdeath ligands (2). Consistent with this, OT1-FD T cell cycling(Fig. 4A) and proliferation (Fig. 4B) were rescued by thepresence of the RIPK1-specific necroptosis inhibitor Nec-1 (43,44), while the proportion of nonviable Annexin-V� cells wasreduced to WT levels (Fig. 4C). Nec-1 also dose-dependentlyrestored live cell recovery (Fig. 4D), demonstrating thatblockade of necroptotic signaling was sufficient to restore bothsurvival and proliferation to OT1-FD T cells. As with 3-MA,Nec-1 reduced LC3 processing in OT1-FD T cells to WT levels(Fig. 4E), suggesting that RIPK1-dependent necroptotic sig-naling, or perhaps necroptosis itself, promotes autophagy.These results support the view that casp8 activity leads to theproteolytic inactivation of RIPK1, thereby preventing theinduction of necroptosis by this DISC-associated serine/threonine kinase (45).

Given the potential that the failure to appropriately inducecasp8 activity in OT1-FD T cells might lead to hyperactiveRIPK1 signaling, we characterized the response of T cellsbearing a T cell-specific deletion of casp8. Casp8FL/FL mice (18)were bred onto a CD4-Cre background to promote deletion ofcasp8 during the double positive stage of thymocyte develop-ment, and the resulting T cells lack casp8 expression in periph-eral CD4� and CD8� T cell subsets. T cells lacking casp8 beara similarly diminished proliferative capacity as observed inFADDdd T cells (11), with the greatest defects noted in theCD8� subset (16). Supporting the hypothesis that the hyper-autophagic phenotype of FADDdd T cells is indeed due to aninability to activate casp8, casp8�/� CD8� T cells activated for3 days possessed high levels of LC3-II, and this was reduced withaddition of Nec-1 (Fig. 5A). While casp8�/� CD8� T cells failedto accumulate following mitogenic stimulation as expected (11),addition of Nec-1 to cultures restored normal proliferation andaccumulation (Fig. 5B and C). As observed for FADDdd T cells,Nec-1 also restored cell cycling and survival to casp8�/� T cells,while knockdown of Atg7 led to increased live T cell recovery inthe absence of casp8 (Figs. 5 D and E). Taken together, thesedata demonstrate that both FADD and casp8 act, likely inconcert, to prevent casp8-independent cell death via the inhibi-tion of autophagy, and suggest that RIPK1 activity may beproteolytically targeted by casp8 in this signaling paradigm.

We have found that autophagic signaling promotes the for-mation of a FADD, casp8, and RIPK1 containing complex inlive, clonally expanding T cells and MEFS, and that FADD andcasp8 are actively involved in the prevention of cell deaththrough the attenuation of autophagic signaling. This viewdiffers from previous reports suggesting that the interactionbetween FADD and Atg5 leads solely to cell death (26, 27).

Recent work demonstrates that T cells require autophagy duringclonal expansion (29), presumably due to acute bioenergeticstress or damaged organelles encountered during high-rate celldivision (21). Here, we demonstrate that this process must belimited to prevent cell death. Although FADDdd-expressingTh1, Th2, and CD8� T cells all displayed enhanced LC3-II

Fig. 4. Nec-1 restores OT1-FD T cell proliferation and survival throughinhibition of RIPK1 signaling. (A) Three-day activated CD8� T cells grown inincreasing doses of Nec-1 and pulsed with BrdU for last hr of culture. (B)CFSE-labeled CD8� T cell proliferation, �10 �M Nec-1. (C) AnnexinV stainingon CD8� T cells from B. (D) Live CD8� cell recovery from cells in B (triplicatecultures: �SD; P values: *, 0.017; **, 0.0025; ***, 0.005). (E) Immunoblot of3-day activated CD8� T cells, �10 �M Nec-1.

16680 � www.pnas.org�cgi�doi�10.1073�pnas.0808597105 Bell et al.

Dow

nloa

ded

by g

uest

on

July

1, 2

020

processing compared with WT T cells, we previously establisheda much greater defect in FADDdd and casp8�/� CD8� vs. CD4�

T cells (14, 16). We surmise that autophagic signaling impactsCD8� T cells more than CD4� cells due to the enhancedproliferative capacity and/or metabolic activity of the former.Consistent with this, greater defects were observed in CD8� vs.CD4� subsets in Atg5�/� mice (29).

Casp8 has been found to inhibit autophagy (25), likely throughdirect cleavage of RIPK1 (46). In this scenario, T cells induceautophagy in response to energetic demands, resulting in for-mation of a DISC-like complex including Atg5–12/Atg16L,FADD, casp8, and RIPK1. Active casp8 then cleaves RIPK1,thus constituting a negative feedback loop to limit autophagicinduction (Fig. 5F). Since Nec-1 was shown to block necroptosisand autophagy induced directly by RIPK1 activity (45), wehypothesize that RIPK1 may influence autophagic signaling

either directly, or perhaps indirectly as a response to necroptoticstress. Indeed, Nec-1 rescued the hyperactive autophagy andrestored the cell cycle profile and survival capacity of activelydividing FADDdd and casp8�/� T cells. It is important toemphasize that autophagic signaling is required in rapidly pro-liferating T cells (29), although the basis for this remains to bedescribed. Given this, we chose to interfere with autophagicsignaling using RNA interference, dominant-negative Vps34 and3-MA, approaches unlikely to interfere with ALL macroauto-phagy. Indeed, under culture with higher levels of 3-MA, weobserved reduced survival of proliferating WT T cells. Takentogether, these results suggest that limited autophagy in prolif-erating T cells is beneficial, whereas unrestricted autophagy thatoccurs in the absence of FADD or casp8 signaling promotes Tcell death.

Our results also imply that the interplay between autophagyand cell death is context dependent. We have found thatblockade of hyperautophagy prevents cell death in proliferatingT cells lacking FADD or casp8, whereas the inverse is true forTNF-receptor signaling (43), likely due to the direct recruitmentand activation of RIPK1 in the latter case by the DISC. Sincec-FLIP is similarly required for efficient T cell proliferation (12,13), this DISC protein likely modulates the activation of casp8during this process. Differential expression of c-FLIP during thecourse of an immune response may control the outcome ofautophagy-induced casp8 activity, likely contributing to T cellhomeostasis. While this hypothesis remains to be tested, ourfindings support the concept that FADD and casp8 modulateautophagic signaling in a manner vital to T cell clonal expansion.

MethodsMice. 1017-FADDdd transgenic mice (Tg(Lck-FADD)1Hed) were bred to ho-mozygosity and maintained on a C57BL/6J background. These were crossedwith OT-1 (Tg(TcraTcrb)1100Mjb) or OT-II (Tg(TcraTcrb)425Cbn) and used forindicated experiments. CD4-Cre mice (Tg(CD4-cre)1Cwi) were crossed withmice containing exon 3 of caspase-8 flanked with LoxP sites and maintainedon a C57BL/6J background. Age-matched littermates were used as controls.Mice were bred and maintained in accordance with the institutional animaluse and care committee at the University of California Irvine vivarium.

T cell Activation, FACs, Immunoblots, and shRNA. OT1 and OT-II splenocyteswere red blood cell lysed and plated in 24 well dishes at 3 � 106 per well with1 �M OVA257–263 or 1 �M OVA323–339, respectively. Non-Ova T cell experiments,24-well dishes coated with 1 �g anti-CD3, plus 200 ng/ml soluble anti-CD28.For BrdU assays, 10 �M BrdU was added for the last hour of culture aspreviously described (16) and gated on CD8� T cells. T cells were lysed andimmunoblotted as previously described (16). MEFs were infected with shRNAvirus for 2 days, followed by 2 days growth in 8 �g/ml puromycin, and replatedin media indicated for 24 h. T cells activated for 24 h were infected with shRNAfor 48 h and grown in 10 �g/ml puromycin. shDRAK2, a protein not expressedin MEFs, was used as a non-specific control. Empty vector was used as a controlin FADDdd T cells. A scrambled control from Open Biosystems was used forcasp8-/- T cells. See SI Methods for more details.

ACKNOWLEDGMENTS. The authors thank Angela Wandinger-Ness andAimee Edinger for supplying Vps34-KD and GFP-LC3 constructs, respectively,Stephen Hedrick for providing Casp8FL/FL mice, Tak Mak for supplying FADD�/�

MEFs, and Dr. Wandy Beatty for assistance with TEM studies. We appreciatethe comments from Drs. David Fruman and Aimee Edinger and from membersof the Walsh lab, and the technical assistance of Cindy Cheung and HuyNguyen. This work was supported by grants from the National Institutes ofHealth (T32CA09054 to B.D.B. and K.A.L., T32AI60573 to R.H.N., R01AI50506and R01AI63419 to C.M.W.) and from the Arthritis National Research Foun-dation (C.M.W.).

1. Kischkel F, et al. (1995) Cytotoxicity-dependent APO-1 Fas/CD95-associated proteins forma death-inducing signaling complex DISC with the receptor. EMBO J 14:5579–5588.

2. Walsh CM, Luhrs KA, Arechiga AF (2003) The ‘‘fuzzy logic’’ of the death-inducingsignaling complex in lymphocytes. J Clin Immunol 23:333–353.

3. Varfolomeev EE, et al. (1998) Targeted disruption of the mouse Caspase 8 gene ablatescell death induction by the TNF receptors, Fas/Apo1, and DR3 and is lethal prenatally.Immunity 9:267–276.

4. Zhang J, Cado D, Chen A, Kabra N, Winoto A (1998) Fas-mediated apoptosis andactivation-induced T cell proliferation are defective in mice lacking FADD/Mort1.Nature 392:296–300.

5. Yeh WC, et al. (1998) FADD: essential for embryo development and signaling fromsome, but not all, inducers of apoptosis. Science 279:1954–1958.

6. Yeh WC, et al. (2000) Requirement for Casper (c-FLIP) in regulation of death receptor-induced apoptosis and embryonic development. Immunity 12:633–642.

Fig. 5. Nec-1 restores cell proliferation and survival in T cells lacking casp8through inhibition of RIPK1 signaling. (A) Immunoblot of 3-day activatedCD8� T cells �10 �M Nec-1. (B) Live CD8� T cell recovery from 3-day activatedWT, casp8�/�, and casp8�/� cells (triplicate cultures: �SD). (C) CFSE-labeledCD8� T cell proliferation, �10 �M Nec-1. (D) Three-day activated CD8� T cells�Nec-1 and pulsed with BrdU for last h of culture. (E) Live cell recovery of WTand casp8�/� (C8) T cells infected with shAtg7 or scrambled control followedby growth in the presence of 10 �g/ml puromycin. Expressed as a ratio of liveT cells infected with shAtg7 vs. a scrambled hairpin (All error bars: �SD). (F)Proposed mechanism of FADD, casp8, and RIPK1 negative feedback on auto-phagic signaling.

Bell et al. PNAS � October 28, 2008 � vol. 105 � no. 43 � 16681

IMM

UN

OLO

GY

Dow

nloa

ded

by g

uest

on

July

1, 2

020

7. Siegel RM (2006) Caspases at the crossroads of immune cell life and death. Nat RevImmunol 6:308–317.

8. Newton K, Harris A, Bath M, Smith K, Strasser A (1998) A dominant interfering mutantof FADD/MORT1 enhances deletion of autoreactive thymocytes and inhibits prolifer-ation of mature T lymphocytes. EMBO J 17:706–718.

9. Walsh C, Wen B, Chinnaiyan A, O’Rourke K, Dixit V, Hedrick S (1998) A role for FADDin T cell activation and development. Immunity 8:439–449.

10. Zornig M Hueber AO, Evan G (1998) p53-dependent impairment of T cell proliferationin FADD dominant-negative transgenic mice. Curr Biol 8:467–470.

11. Salmena L, et al. (2003) Essential role for caspase 8 in T cell homeostasis and Tcell-mediated immunity. Genes Dev 17:883–895.

12. Zhang N, He YW (2005) An essential role for c-FLIP in the efficient development ofmature T lymphocytes. J Exp Med 202:395–404.

13. Chau H, et al. (2005) Cellular FLICE-inhibitory protein is required for T cell survival andcycling. J Exp Med 202:405–413.

14. Beisner DR, Chu IH, Arechiga AF, Hedrick SM, Walsh CM (2003) The requirements forfas-associated death domain signaling in mature T cell activation and survival. J Im-munol 171:247–256.

15. Arechiga AF, et al. (2005) Cutting Edge: FADD Is Not Required for Antigen Receptor-Mediated NF-{kappa}B Activation. J Immunol 175:7800–7804.

16. Arechiga AF, et al. (2007) A Fas-Associated Death Domain Protein/Caspase-8-SignalingAxis Promotes S-Phase Entry and Maintains S6 Kinase Activity in T Cells Responding toIL-2. J Immunol 179:5291–5300.

17. van Stipdonk MJ, Lemmens EE, Schoenberger SP (2001) Naive CTLs require a single briefperiod of antigenic stimulation for clonal expansion and differentiation. Nat Immunol2:423–429.

18. Beisner DR, Ch’en IL, Kolla RV, Hoffmann A, Hedrick SM (2005) Cutting edge: innateimmunity conferred by B cells is regulated by caspase-8. J Immunol 175:3469–3473.

19. Imtiyaz HZ, et al.(2006) The Fas-associated death domain protein is required in apo-ptosis and TLR-induced proliferative responses in B cells. J Immunol 176:6852–6861.

20. Edinger AL, Thompson CB (2004) Death by design: apoptosis, necrosis and autophagy.Curr Opin Cell Biol 16:663–669.

21. Lum JJ, DeBerardinis RJ, Thompson CB 2005. Autophagy in metazoans: cell survival inthe land of plenty. Nat Rev Mol Cell Biol 6:439–448.

22. Tanida I, Ueno T, Kominami E (2004) LC3 conjugation system in mammalian autophagy.Int J Biochem Cell Biol 36:2503–2518.

23. Lum JJ, et al. (2005) Growth factor regulation of autophagy and cell survival in theabsence of apoptosis. Cell 120:237–248.

24. Levine B, Yuan J (2005) Autophagy in cell death: an innocent convict? J Clin Invest115:2679–2688.

25. Yu L, et al. (2004). Regulation of an ATG7-beclin 1 Program of Autophagic Cell Deathby Caspase-8. Science 304:1500–1502.

26. Pyo JO, et al. (2005) Essential roles of Atg5 and FADD in autophagic cell death:dissection of autophagic cell death into vacuole formation and cell death. J Biol Chem280:20722–20729.

27. Thorburn J, et al. (2005) Selective inactivation of a Fas-associated death domain protein(FADD)-dependent apoptosis and autophagy pathway in immortal epithelial cells. MolBiol Cell 16:1189–1199.

28. Li C, et al. (2006) Autophagy is induced in CD4� T cells and important for the growthfactor-withdrawal cell death. J Immunol 177:5163–5168.

29. Pua HH, Dzhagalov I, Chuck M, Mizushima N, He YW (2007) A critical role for theautophagy gene Atg5 in T cell survival and proliferation. J Exp Med 204:25–31.

30. Klionsky DJ (2004) Cell biology: regulated self-cannibalism. Nature 431:31–32.31. Barnden MJ, Allison J, Heath WR, Carbone FR (1998) Defective TCR expression in

transgenic mice constructed using cDNA-based alpha- and beta-chain genes under thecontrol of heterologous regulatory elements. Immunol Cell Biol 76:34–40.

32. Tanida I, Minematsu-Ikeguchi N, Ueno T, Kominami E (2005) Lysosomal turnover, butnot a cellular level, of endogenous LC3 is a marker for autophagy. Autophagy 1:84–91.

33. Ting AT, Pimentel-Muinos FX, Seed B (1996) RIP mediates tumor necrosis factorreceptor 1 activation of NF-kappaB but not Fas/APO-1-initiated apoptosis. EMBO J15:6189–6196.

34. Juo P, Kuo CJ, Yuan J, Blenis J (1998) Essential requirement for caspase-8/FLICE in theinitiation of the Fas-induced apoptotic cascade. Curr Biol 8:1001–1008.

35. Juo P, et al. (1999) FADD is required for multiple signaling events downstream of thereceptor Fas. Cell Growth Differ 10:797–804.

36. Kihara A, Noda T, Ishihara N, Ohsumi Y (2001) Two distinct Vps34 phosphatidylinositol3-kinase complexes function in autophagy and carboxypeptidase Y sorting in Saccha-romyces cerevisiae. J Cell Biol 152:519–530.

37. Stein MP, et al. (2005) Interaction and functional analyses of human VPS34/p150phosphatidylinositol 3-kinase complex with Rab7. Methods Enzymol 403:628–649.

38. Hildeman DA, et al. (2002) Activated T cell death in vivo mediated by proapoptotic bcl-2family member bim. Immunity 16:759–767.

39. Puig O, et al. (2001) The tandem affinity purification (TAP) method: a general proce-dure of protein complex purification. Methods 24:218–229.

40. Towns R, Guo C, Shangguan Y, Hong S, Wiley JW (2008) Type 2 diabetes with neurop-athy: autoantibody stimulation of autophagy via Fas. Neuroreport 19:265–269.

41. Kennedy NJ, Kataoka T, Tschopp J, Budd RC (1999) Caspase activation is required for Tcell proliferation. J Exp Med 190:1891–1896.

42. Alam A, Cohen LY, Aouad S, Sekaly RP (1999) Early activation of caspases during Tlymphocyte stimulation results in selective substrate cleavage in nonapoptotic cells. JExp Med 190:1879–1890.

43. Degterev A, et al. (2005) Chemical inhibitor of nonapoptotic cell death with thera-peutic potential for ischemic brain injury. Nat Chem Biol 1:112–119.

44. Degterev A, et al. (2008) Identification of RIP1 kinase as a specific cellular target ofnecrostatins. Nat Chem Biol 4:313–321.

45. Holler N, et al. (2000) Fas triggers an alternative, caspase-8-independent cell deathpathway using the kinase RIP as effector molecule. Nat Immunol 1:489–495.

46. Lin Y, Devin A, Rodriguez Y, Liu ZG (1999) Cleavage of the death domain kinase RIP bycaspase-8 prompts TNF-induced apoptosis. Genes Dev 13:2514–2526.

16682 � www.pnas.org�cgi�doi�10.1073�pnas.0808597105 Bell et al.

Dow

nloa

ded

by g

uest

on

July

1, 2

020