Resource

Functional Interplay between CaspaseCleavage and Phosphorylation Sculptsthe Apoptotic ProteomeMelissa M. Dix,1,3 Gabriel M. Simon,1,3 Chu Wang,1 Eric Okerberg,2 Matthew P. Patricelli,2 and Benjamin F. Cravatt1,*1The Skaggs Institute for Chemical Biology and Department of Chemical Physiology, The Scripps Research Institute,

10550 N. Torrey Pines Road, La Jolla, CA 92037, USA2ActivX Biosciences, La Jolla, CA 92307, USA3These authors contributed equally to this work*Correspondence: cravatt@scripps.edu

http://dx.doi.org/10.1016/j.cell.2012.05.040

SUMMARY

Caspase proteases are principal mediators ofapoptosis, where they cleave hundreds of proteins.Phosphorylation also plays an important role inapoptosis, although the extent to which proteolyticand phosphorylation pathways crosstalk duringprogrammed cell death remains poorly understood.Using a quantitative proteomic platform that inte-grates phosphorylation sites into the topographicalmaps of proteins, we identify a cohort of over500 apoptosis-specific phosphorylation events andshow that they are enriched on cleaved proteinsand clustered around sites of caspase proteolysis.We find that caspase cleavage can expose new sitesfor phosphorylation, and, conversely, that phosphor-ylation at the +3 position of cleavage sites candirectly promote substrate proteolysis by caspase-8.This study provides a global portrait of the apoptoticphosphoproteome, revealing heretofore unrecog-nized forms of functional crosstalk between phos-phorylation and caspase proteolytic pathways thatlead to enhanced rates of protein cleavage and theunveiling of new sites for phosphorylation.

INTRODUCTION

Proteolysis and phosphorylation are two of the most pervasive

forms of protein posttranslational modification, playing essential

roles in the majority of (patho)physiological processes, including

tissue development, cancer, and cell death (Kurokawa and Korn-

bluth, 2009; Lopez-Otın and Hunter, 2010). Apoptosis, or pro-

grammed cell death, is orchestrated by a family of cysteine

proteases called caspases, which cleave their protein substrates

after aspartic acid residues (Crawford andWells, 2011; Fuentes-

Prior and Salvesen, 2004; Thornberry and Lazebnik, 1998).

Recent advances in global protease substrate identification

technologies have generated a large inventory of proteins that

are cleaved by caspases during apoptosis, demonstrating

426 Cell 150, 426–440, July 20, 2012 ª2012 Elsevier Inc.

that as much as 5% of the proteome is subject to caspase-

mediated proteolysis (Arntzen and Thiede, 2011; Crawford and

Wells, 2011).

Protein kinases are prominently represented among caspase

substrates and, in some cases, cleavage activates these kinases

so that they can perform important functions in apoptosis (Kuro-

kawa and Kornbluth, 2009). Caspase-mediated activation of

Rho-associated kinase 1 (ROCK1), for instance, promotes the

characteristic membrane blebbing associated with apoptosis

(Coleman et al., 2001). Kinases can also be inactivated by

caspase-mediated cleavage to block their activity during

apoptosis (Kurokawa and Kornbluth, 2009). The crosstalk

between caspases and kinases also includes the phosphoryla-

tion of caspases to either enhance or suppress their activity

(Kurokawa and Kornbluth, 2009). Likewise, the phosphorylation

of some caspase substrates, notably BID phosphorylation on

Thr59 (which is the P2 residue of the caspase-8 cleavage site)

blocks caspase cleavage (Degli Esposti et al., 2003). These find-

ings suggest that caspase and kinase pathways interact in

intricate ways to influence the balance between cell survival

and death (Janes et al., 2005). Nonetheless, whether a more

global relationship between proteolysis and phosphorylation

exists in apoptosis has not been investigated.

We recently introduced a proteomic method termed

PROTOMAP (short for Protein Topography and Migration Anal-

ysis Platform) that can be used to characterize proteolytic events

in cells by detecting shifts in protein migration through a combi-

nation of SDS-PAGE and mass spectrometry (MS)-based

proteomics (Dix et al., 2008). Using this approach, we identified

over 250 cleaved proteins in apoptotic cells, including 170

proteins that were not previously known to be cleaved by cas-

pases. In the current study, we sought to create an advanced,

quantitative version of PROTOMAP that enables simultaneous

analysis of proteolytic and phosphorylation processes in cells,

such that phosphorylation sites could be directly integrated

into the topographical maps of cleaved proteins during

apoptosis. We applied this method to study the intrinsic

apoptotic cascade in Jurkat T cells, resulting in the identification

of more than 700 cleaved proteins and 5,000 sites of phosphor-

ylation. The integration of these global data sets revealed that

phosphorylation events are enriched on cleaved proteins and

A

B C

In-gel

digestion

Phosphopeptide

Enrichment

(IMAC)SDS-PAGE

High

MW

Low

MW

Peptide

Identification

(MS/MS)

Peptide

quantitation

(MS)

qPeptograph

N-termHigh

MW

Low

MW

C-term

+

+

Combine

SILAC

Ratios

Spectral

counts

0 -55

Ctrl Apop.

SILAC “light” SILAC “heavy”

P P P

P P P

P P P

P P P

P P P

P P P

90 435309

PROTOMAP

(2008)

qP-PROTOMAP

(2012)

STS (2 hrs) STS (4 hrs)

776, 24%389, 12%

2133, 64%

Apoptosis-specific Control-specific Static

581, 16%

2721, 75%

327, 9%

(log2)

Figure 1. Quantitative Profiling of Phosphorylation and Proteolytic Pathways in Apoptosis by qP-PROTOMAP

(A) General features of qP-PROTOMAP method as described in the main text. Peptides are colored red and blue to represent signals detected in healthy control

(light) and apoptotic cells (heavy), respectively.

(B) Number of cleaved proteins detected using the original PROTOMAPmethod (Dix et al., 2008) versus qP-PROTOMAP as described in this study. See Table S1

for peptographs of cleaved proteins identified by qP-PROTOMAP.

(C) Distribution of phosphorylation events identified in control and apoptotic cells by qP-PROTOMAP. Phosphorylation events were designated ‘‘control specific’’

or ‘‘apoptosis specific’’ if they showed light/heavy SILAC ratios of >2 or <0.5, respectively (corresponding to log2 values of 1 or �1). Phosphorylation events

displaying light/heavy ratios between these values were designated as ‘‘static.’’

See also Table S2.

are clustered around sites of caspase cleavage. We further iden-

tified a cohort of previously unreported phosphorylation sites

that were specific to apoptotic cells, suggesting the existence

of a cell-death-related phosphorylation network. We show using

activity-based proteomic methods that at least a part of this

network is driven by caspase-mediated activation of DNA-

dependent protein kinase (DNA-PK) at early stages during the

time course of apoptosis. Finally, we interrogated the functional

relationship between proteolysis and phosphorylation, uncover-

ing multiple forms of crosstalk that include the caspase process-

ing of proteins to expose new sites for phosphorylation and the

phosphorylation of proteins at the +3 (P3) position of caspase

recognition sequences to dramatically enhance proteolysis by

caspase-8.

RESULTS

Quantitative Proteomic Analysis of Phosphorylation andProteolysis by qP-PROTOMAPThe proteomic measurement of dynamic posttranslational

modifications, like phosphorylation, requires quantification of

individual peptides, and we therefore sought to combine

PROTOMAP with stable isotopic labeling methods (SILAC; Ong

et al., 2002) for this purpose. We also needed to incorporate a

phosphopeptide enrichment step without sacrificing the protein

size and topography information provided by the SDS-PAGE

step of the original PROTOMAPmethod. The workflow for the re-

sulting quantitative phospho-PROTOMAP (or qP-PROTOMAP)

platform was therefore as follows (Figure 1A): Control and

Cell 150, 426–440, July 20, 2012 ª2012 Elsevier Inc. 427

apoptotic cells were grown in media containing isotopically light

and heavy amino acids, respectively. Equal quantities of each

cell proteome were then combined and separated by SDS-

PAGE. Next, as in the original PROTOMAP method, gel lanes

were sliced into 22 evenly spaced bands that were digested

in-gel with trypsin to extract peptides. Phosphopeptides were

then enriched via immobilized metal-affinity chromatography

(IMAC) and subjected to reverse-phase liquid chromatography

and MS analysis on an LTQ-Velos Orbitrap. Flow-through from

the IMAC step (containing unphosphorylated peptides) was

also analyzed, and the combined SILAC ratios of unphosphory-

lated and phosphorylated peptides were integrated into quanti-

tative peptographs to provide a complete picture of protein

phosphorylation and proteolysis.

Peptographs for qP-PROTOMAP experiments display de-

tected peptides from left-to-right based on their position in

the primary sequence of their proteins, and from top-to-bottom

depending on the gel band in which they were detected (the

vertical dimension thus represents molecular weight) (Fig-

ure 1A). Phosphopeptides are marked by a circle. For the

purposes of quantitation, each peptide is assigned a color on

a continuum from red to blue reflecting the light/heavy ratio:

peptides exhibiting no-change (1:1 ratio) are displayed in

purple; control- and apoptosis-specific peptides are shown in

red and blue, respectively. To facilitate visual interpretation of

these quantified peptide data, a box plot is provided in the

middle panel (to the right of the peptograph) that displays the

distribution of ratios found in each band. Spectral-count infor-

mation is displayed in a third panel to enable estimation of

the relative abundance of each protein isoform. Most impor-

tantly, because qP-PROTOMAP integrates phosphorylation

sites into the topographical maps of cleaved proteins, the

approach can determine the precise protein isoforms that

possess individual phosphorylation events. Thus, we are able

to identify phosphorylation events that may occur exclusively

on full-length proteins or, alternatively, on fragments of these

proteins generated during apoptosis. We present an example

peptograph and more details on its interpretation in Figure S1

(available online).

For our assessment of crosstalk between proteolysis and

phosphorylation, we induced the intrinsic apoptotic pathway in

Jurkat T cells with staurosporine (‘‘STS’’). Although it might

initially seem counterintuitive to use a broad-spectrum kinase

inhibitor like STS to study phosphorylation events in apoptosis,

we hypothesized that this drug, through the inhibition of many

kinases, might simplify the phosphoproteome to facilitate the

characterization of phosphorylation events that were important

for programmed cell death. Indeed, we, and others, have shown

that STS induces a highly efficient apoptotic cascade in Jurkat

T cells that is essentially complete by �6 hr (Dix et al., 2008;

Na et al., 1996; Stolzenberg et al., 2000), and we therefore

expected that any kinase pathways relevant for this rapid cell

death process would necessarily be insensitive to this drug.

We therefore analyzed STS-treated Jurkat cells at an ‘‘early’’

(2 hr) and ‘‘late’’ (4 hr) stage of apoptosis by qP-PROTOMAP.

In total, 4,521 proteins were detected across both time points,

and 5,034 sites of phosphorylation were quantified on serine,

threonine, or tyrosine residues from 1,624 of the proteins (36%

428 Cell 150, 426–440, July 20, 2012 ª2012 Elsevier Inc.

of all proteins detected). Peptographs were generated for each

protein at both time points, enabling rapid visual interpretation

of their (1) cleavage status, (2) cleavagemagnitude, and (3) phos-

phorylation status on individual protein isoforms in a time-

dependent manner.

Quantitative Analysis of Cleaved Proteins in ApoptoticCells by qP-PROTOMAPWe initially evaluated the performance of qP-PROTOMAP as

a global method for characterizing cleaved proteins, which

were expected to: (1) possess multiple peptides in the parental

protein band that substantially deviate from the 1:1 SILAC ratio

observed for uncleaved proteins, and/or (2) display persistent

protein fragments selectively in apoptotic cells. We developed

abundance thresholds for quantifying the cleavage state of

proteins, and, out of 2,867 proteins that met these thresholds,

744 of them (26%) showed strong evidence of cleavage

(Table S1; see the Extended Experimental Procedures for

details). A majority of the cleaved proteins identified in our orig-

inal PROTOMAP study were also observed to be cleaved by

qP-PROTOMAP (77%, Figure 1B), andwe detected 349 proteins

that had not been previously described in the literature as cas-

pase substrates (Table S1). Considerably more proteins were

found to be cleaved than has been previously reported, both in

absolute number and as a percent of the proteome. This can be

ascribed to the higher accuracy and sensitivity that is achievable

using SILAC quantitation, enabling high-confidence assess-

ments of lower-abundance proteins. As reported previously

(Dix et al., 2008), the majority of cleaved proteins (67%) dis-

played one or more persistent fragments (Table S1).

Quantitative Analysis of the Phosphoproteomein Apoptotic Cells by qP-PROTOMAPWe next assessed the performance of qP-PROTOMAP as

a global method for characterizing phosphorylation events in

apoptotic cells. Phosphorylation events that showed >2-fold

SILAC ratios in control or apoptotic cellswere defined as ‘‘control

specific’’ or ‘‘apoptosis specific,’’ respectively. We should note

the potential for these phosphopeptide SILAC ratios to be influ-

enced by the cleavage of proteins. For instance, reductions in

protein abundance during apoptosis could indirectly cause

a loss of phosphopeptide signals. Conversely, the stochastic

nature of peptide detection in individual data-dependent MS

runs could result in the identification of a static phosphorylation

event exclusively on one isoform of a protein. We attempted to

address at least some of the complexities by performing

numerous replicates for our phosphoproteomic experiments

(see Extended Experimental Procedures for details), which

yielded rapidly diminishing returns for unique phosphorylation-

site identification with each replicate (Figure S2).

A global analysis of SILAC ratios for the 5,060 phosphoryla-

tion events identified in our combined qP-PROTOMAP data

sets led to several important discoveries. First, a majority of

phosphorylation events (>85%) either showed no change or

were elevated in control cells (Figure 1C). This is not a surprising

result given that we induced apoptosis with the broad-spectrum

kinase inhibitor STS. Even with this mode of inducing apoptosis,

however, we still identified 531 phosphorylation sites that were

apoptosis specific (Figure 1C; Table S2). Striking examples

included the pSer347 and pSer882 events in the retinoblastoma

protein (RB1) (Figure 2A, features ‘‘1’’ and ‘‘2,’’ respectively),

a well-studied tumor suppressor that is known to be cleaved

at two distinct caspase sites following the induction of

apoptosis (Fattman et al., 2001). We identified 18 additional

sites of phosphorylation on RB1, virtually all of which were

control specific (Figures 2A and 2B). Most of these control-

specific phosphorylation events occurred at well-characterized

sites identified in numerous (>10) independent studies as deter-

mined by searches of the PhosphoSite database (Hornbeck

et al., 2011) (Figure 2B); interestingly, however, the apoptosis-

specific pSer347 and pSer882 events had only been reported

rarely in previous phosphoproteomic analysis (%3 times each;

Figure 2B) and, in the case of pSer347, only in Jurkat cells

treated with pervanadate, a known proapoptotic stimulus

(Hehner et al., 1999). These initial data suggested that apoptosis

might activate a special phosphorylation network that is distinct

in content from other cellular processes. We more systemati-

cally assessed this possibility by comparing static/control-

specific and apoptosis-specific phosphorylation events, as

estimated by the absence of annotation of these events in

the PhosphoSite database. Strikingly, close to half of the

apoptosis-specific phosphorylation sites were previously unre-

ported in the literature, whereas less than 15% of the static/

control-specific phosphorylation sites fell into this category (Fig-

ure 2C). Apoptosis-specific phosphorylation events were also

underrepresented among phosphorylations that were frequently

detected in the literature (R5 citations) (Figure 2D). We

conclude from these data that apoptosis leads to the activation

of a specific set of kinases (and/or inactivation of phosphatases)

to create a rare pool of phosphorylation events that are not

observed in healthy cells.

A closer examination of the apoptosis-specific phosphoryla-

tion sites on RB1 uncovered another provocative feature—

both of these events are proximal to known sites of caspase

cleavage (green lines in Figure 2A) that generate persistent frag-

ments detectable by qP-PROTOMAP and western blotting

(Figure 2A). The pSer347 event, for instance, occurs just two

residues upstream (the P3 position) from the scissile aspartate

(Fattman et al., 2001). The pSer882 event is located four residues

upstream of a known scissile aspartate (at the P5 position,

Fattman et al., 2001) and was identified by qP-PROTOMAP on

a half-tryptic peptide ending at this residue, indicating that the

phosphorylation event resides on a caspase-cleaved fragment

of RB1 (in this case, the cleavage event is not expected to

produce a shift in gel migration of RB1 because cleavage occurs

near the C terminus of the protein). These observations led us to

wonder whether such cleavage-site-proximal phosphorylation

events were unique to RB1, or whether they might represent

a more general phenomenon that occurs during apoptosis.

Systems-wide Crosstalk between Proteolysis andPhosphorylation during ApoptosisTo evaluate how phosphorylation events that occur during

apoptosismight globally intersect with caspase-mediated prote-

olysis, we compiled all of the known sites of caspase cleavage,

including 75 sites that were identified in the current study (Table

S3), to give 679 total sites on 566 distinct proteins. Four hundred

and thirteen of these proteins were detected in our analysis of

Jurkat T cells, and we aligned their sequences such that they

were all anchored around the scissile P1 aspartate residue. We

then searched for phosphorylation events in our data sets that

were located 200 residues up- or downstream of the P1 resi-

dues, resulting in the discovery of �675 such phosphorylation

events on 210 proteins. These phosphorylation sites were strik-

ingly clustered in the region immediately surrounding the scissile

aspartate, in particular from the P6 to P60 residues (shaded

region in Figure 3A; also see Table 1). This clustering was evident

not only for apoptosis-specific phosphorylation events, but also

for static and control-specific phosphorylation events (Fig-

ure S2). We furthermore found that known caspase substrates

were more likely to be phosphorylated in Jurkat T cells than

were uncleaved proteins (Figure 3B).

We next asked whether the phosphorylation events that

occurred in apoptotic cells were catalyzed by a specific set

of kinases. To discover kinases that might be activated

during apoptosis, we employed a functional proteomic platform

termed KiNativ. The KiNativ technology uses active site-

directed chemical probes containing biotin-conjugated elec-

trophilic analogs of ATP or ADP for covalent capture of

ATP-binding proteins from proteomes. Conserved lysines in

kinase active sites react with the probes and are then enriched

and quantified with streptavidin chromatography and targeted

MS analysis, respectively (Patricelli et al., 2011; see Extended

Experimental Procedures for details). The majority of kinases

showed reduced KiNativ signals in apoptotic cells (Figure 3C

and Table S4), likely reflecting inhibition by STS. However,

a handful of kinases showed stronger KiNativ signals in STS-

treated cells, the most dramatic of which was DNA-dependent

protein kinase (DNA-PK) (Figure 3C). DNA-PK is known to pref-

erentially phosphorylate serines and threonines that are located

before glutamine residues on proteins ([S/T]-Q motif; Kim et al.,

1999). Consistent with the activation of DNA-PK during

apoptosis, a motif-x analysis (Schwartz and Gygi, 2005) re-

vealed that S-Q phosphorylations were the most overrepre-

sented motifs among the apoptosis-specific phosphorylation

events in our data sets (Figure 3D and Table S4). No such

enrichment of S-Q motifs was observed for static or control-

specific phosphorylation events (Table S4). These proteomic

data were confirmed by western blotting using an antibody

that recognizes p[S/T]-Q motifs, which showed a time-depen-

dent increase in p[S/T]-Q-immunoreactive proteins in apoptotic

cells compared to control cells that peaked at 2 hr post-STS

treatment (Figure 3E).

The [S/T]-Q substrate motif is utilized by other kinases, most

notably ATM and ATR, which, along with DNA-PK, are important

regulators of genome stability and the DNA-damage response

(Kim et al., 1999). No change in ATM or ATR activity was seen

in our KiNativ data (Table S4), but this finding does not rule

out a contribution of these kinases to phosphorylation events

in apoptosis. We more directly tested for this possibility by treat-

ing Jurkat T cells with selective inhibitors of DNA-PK (NU-7441

and NU-7026), ATM (KU5633), or ATM/ATR (CGK733) for 1 hr

prior to induction of apoptosis. Western blotting revealed a

near-complete block of p[S/T]-Q events upon treatment with

Cell 150, 426–440, July 20, 2012 ª2012 Elsevier Inc. 429

250kDa

150kDa

100kDa

75kDa

50kDa

37kDa

25kDa

15kDa

TLQTDSIDSFETQR

...EGS*DEAD

*

TLQTDS*IDSFETQR

A

LRFDIEGSDEADGSK LRFDIEGS*DEAD

TLQTDS*IDSFETQR

B

C D

1111 11

22

RB1

SILAC Ratio Spectralcounts

0e+0

02e

+06

4e+0

66e

+06

050

0010

000

1500

020

000

050

000

1500

0025

0000

050

0015

000

2500

0

22 22

ApoptoticControl ApoApoApoApoApoApoApoApoApoApoApoApoApoApoApoApoApoApoApoApApoptoptoptoptoptoptoptoptoptoptoptoptoptoptoptoptoptoptoptoptoptoticticticticticticticticticticccticticticticcticticictCConConConConConConConConConConConConConConConConConCoConConControtrotrorotrotrotrotrotrorotrorotrttrtrrtrotrol

# of

refe

renc

esin

Pho

spho

Site

DB

Static or control-specific

Apoptosis-specific

Nov

el s

ites

of p

hosp

hory

latio

n (%

)

Site

s of

pho

spho

ryla

tion

(%)

0%10%20%30%40%50%60%

# of references in PhosphoSite0 1 2 3 4 >4

Apoptosis-specificStatic or control-specific

2 hrs post-STS4 hrs post-STS

SIL

AC

ratio

50

-5

0%

10%

20%

30%

40%

Contro

l

Apopto

tic

0

60

120

180

S37

T140

S16

3S

230

S24

9T2

52Y

321

Y32

5S

347

S35

0T3

53T3

56S

360

T373

S56

7T5

83S

588

T601

Y60

6S

608

S61

2S

618

S62

4T6

25Y

651

Y65

9S

773

T774

T778

S78

0S

788

Y79

0S

794

S79

5Y

805

S80

7S

811

Y81

3S

816

T821

T823

T826

S83

4S

838

T841

S84

2S

855

S88

2

Figure 2. Identification of a Cohort of Apoptosis-Specific Phosphorylation Events

(A) Western blot and quantitative peptograph for RB1 at 4 hr post-STS treatment. Two apoptosis-specific phosphorylation events were identified at Ser347 (1)

and Ser882 (2). To the right of the peptograph, representative MS1 chromatographs for phosphorylated and unphosphorylated versions of these peptides are

shown in which light (control) and heavy (apoptotic) signals are colored red and blue, respectively. Green lines mark known sites of caspase cleavage (Fattman

et al., 2001) that likely generate the observed N-terminal and internal fragments visible on the peptograph and, in the case of the internal fragment, by western

blotting. Asterisk denotes a nonspecific band on the blot.

(B) Shown are the 39 phosphorylation sites on RB1 listed in the PhosphoSite database, along with the number of literature references for each site. Note that the

two apoptosis-specific phosphorylation events (blue) have been rarely reported in previous studies, in contrast to control-specific phosphorylation events (red),

which were, in general, detected in many previous studies. Below each phosphorylation site, we show the corresponding SILAC ratios. Sites shown in black were

not detected in our study.

430 Cell 150, 426–440, July 20, 2012 ª2012 Elsevier Inc.

DNA-PK inhibitors, whereas inhibitors of ATM and/or ATR were

without effect (Figure 3F). We also generated two Jurkat T cell

lines with stable shRNA-mediated knockdowns of DNA-PK (Fig-

ure S2) and found that these cells showed substantially blunted

[S/T]-Q phosphorylation following induction of apoptosis (Fig-

ure 3F). Finally, we performed a qP-PROTOMAP study of

apoptotic Jurkat cells pretreated with NU-7441, which resulted

in a two-fold or greater reduction in the majority of p[S/T]-Q

events (�60%), with other non-p[S/T]-Q events being minimally

affected (Figure S2). Interestingly, of the p[S/T]-Q events reduced

by NU-7441 treatment, over 80% were apoptosis specific (Fig-

ure S2 and Table S4), which, when combined with our immuno-

blotting results (Figure 3F), indicate that DNA-PK is responsible

for a large fraction of the p[S/T]-Q events observed in apoptotic

cells.

We next investigated how DNA-PK might be activated during

apoptosis. We found that, early in the apoptotic cascade, DNA-

PK relocated from the nucleus to the cytoplasm, where, interest-

ingly, the enzyme was cleaved to generate a stable �150 kDa

C-terminal fragment that contains the kinase domain (Figures

3G and 3H). The appearance of a cleaved form of DNA-PK in

the cytoplasm directly correlated with the increased p[S/T]-Q

immunoreactive proteins (Figure 3E) and the enhanced KiNativ

signals for this kinase observed 2 hr after induction of apoptosis

(Figure 3H, top). Pretreatment of cells with the caspase inhibitor

Z-VAD-FMK blocked STS-induced cleavage of DNA-PK and p

[S/T]-Q events (Figure S2). Previous studies have also reported

the caspase-mediated cleavage of DNA-PK in apoptotic cells

(Casciola-Rosen et al., 1995), but have mostly interpreted this

proteolytic event to inactivate DNA-PK. The assays used in

such studies, however, typically measured DNA-dependent

DNA-PK activity with a peptide substrate (Allalunis-Turner

et al., 1995; Han et al., 1996; Song et al., 1996). Our data support

an alternative model wherein caspase cleavage releases

DNA-PK from genomic DNA to generate an active, truncated

form of the enzyme that traverses into the cytoplasm to catalyze

a large number of apoptosis-specific phosphorylation events.

Caspase Cleavage Can Expose Phosphorylation SitesPrevious studies that have examined the functional effects of

phosphorylation on caspase cleavage with individual protein

substrates in vitro have mostly uncovered instances where

phosphorylation blocks caspase cleavage (Duncan et al., 2010;

Kurokawa and Kornbluth, 2009; Tozser et al., 2003), leading to

a model where phosphorylation serves to ‘‘protect’’ proteins

from proteolytic processing. Many of the apoptosis-specific

phosphorylation events identified in our study, however, did

not appear to conform to this scenario because they were

located on half-tryptic peptides ending in C-terminal aspartates,

the hallmark of caspase cleavage. One such example is SF3B2,

which contains an apoptosis-specific phosphorylation event at

Ser861 that is located at the P2 position adjacent to a site of cas-

(C) Approximately 40% of apoptosis-specific phosphorylation events have not be

only �13% of static or control-specific phosphorylation events fall into this cate

(D) Histogram showing the number of literature references found in the Phospho

phosphorylation events. Note that apoptosis-specific events are overrepresented

specific events are overrepresented in the group with many previous identificatio

pase cleavage (Figures 4A and 4B). To ascertain whether this

phosphorylation event occurs before or after caspase-mediated

proteolysis, we used a targeted MS approach with isotopically

labeled peptides to measure the four possible forms of the

SF3B2 peptide: (1) uncleaved/unphosphorylated, (2) cleaved/

unphosphorylated, (3) uncleaved/phosphorylated, and (4)

cleaved/phosphorylated. These experiments provided two key

lines of evidence supporting that phosphorylation of Ser861

occurs after caspase-mediated proteolysis. First, the cleaved/

unphosphorylated peptide appeared at an earlier time point

than the cleaved/phosphorylated peptide (Figure 4C). Second,

the uncleaved/phosphorylated peptide was not detected at any

time point, suggesting that the full-length (parental) form of

SF3B2 is not phosphorylated at Ser861. These predictions

were also supported by in vitro substrate assays,wherewe found

that the unphosphorylated, but not phosphorylated peptide

served as a substrate for caspases (Figure 4D and Figure S3).

A broader search of our qP-PROTOMAP data set identified

several additional apoptosis-specific phosphorylation events

that were found exclusively on half-tryptic, aspartate-terminating

peptides (Table 1, Figure S3 and Table S5). One candidate was

a previously unreported phosphorylation event found at the P4

position (S*QTD) on an N-terminal fragment of HCLS1 (Fig-

ure 4E). Similar to what was observed for pSer861 in SF3B2,

phosphorylation of Ser112 completely blocked caspase-3

cleavage of the HCLS1 peptide (Figure 4F and Figure S3). Cas-

pase-8 cleavage was also significantly reduced by phosphoryla-

tion of Ser112, although residual hydrolytic activity was detected

(Figure 4F). These data, taken together, are consistent with

phosphorylation of Ser112 occurring after caspase-mediated

cleavage of HCLS1.

Phosphorylation Can Promote Caspase CleavageWe next wondered whether phosphorylation might also, in

certain instances, directly promote (rather than block) cas-

pase-mediated proteolysis. We accordingly searched our qP-

PROTOMAP data sets for apoptosis-specific phosphorylation

events that were located on the parental forms of cleaved

proteins in close proximity to sites of caspase-mediated proteol-

ysis. A compelling example was found in the protein KHSRP

(Figure 5A), where we observe an apoptosis-specific phosphor-

ylation event at Thr100 at 2 hr post-STS treatment on the

parental 74 kDa form of the protein (band 6/7), and at 4 hr on

a half tryptic, aspartate (Asp103)-terminating peptide in an

N-terminal �15 kDa fragment (band 21/22) (Figures 5A and 5B,

respectively). pThr100 was thus located at the P3 position

relative to the Asp103 caspase cleavage site. We also detected

the unphosphorylated version of the half-tryptic, Asp103-

terminal peptide in the KHSRP fragment.

To determine the relative kinetics of Thr100 phosphorylation

versus caspase-mediated proteolysis at Asp103 in KHSRP, we

performed a targeted MS analysis using isotopically labeled

en reported previously based on PhosphoSite database searches. In contrast,

gory.

Site database for apoptosis-specific (blue) and static or control-specific (red)

in the previously unreported group (zero references), whereas static or control-

ns (>4 references).

Cell 150, 426–440, July 20, 2012 ª2012 Elsevier Inc. 431

A B

C

D E F

G

H

432 Cell 150, 426–440, July 20, 2012 ª2012 Elsevier Inc.

peptides following the protocol outlined above for the SF3B2

protein. In band 6, where the parental form of KHSRP migrates,

we detected the uncleaved/phosphorylated form of the peptide,

which was strongly increased over the first 2 hr following STS

treatment and then decreased thereafter (Figure 5C). In contrast,

the cleaved forms of the peptide in bands 21/22 did not appear

until 2.5 hr and continued to accumulate throughout the

remainder of the time course (Figure 5C). These data indicate

that phosphorylation at Thr100 precedes proteolysis by a

substantial time window during the apoptotic cascade. We

should note that the vast majority of the cleaved peptide was

found in the unphosphorylated form, with only trace levels of

the cleaved/phosphorylated peptide being detected throughout

the time course. Nonetheless, we were intrigued by the comple-

mentary time courses for phosphorylation versus proteolysis, as

well as the similar stoichiometries of the uncleaved/phosphory-

lated and cleaved/unphosphorylated peptides, both of which

peaked at �10% of the total quantity of uncleaved/unphos-

phorylated peptide (Figure S4). These data correlate well with

the low overall magnitude of cleavage for KHSRP (see pepto-

graphs in Figures 5A and 5B) and suggest further that phosphor-

ylation and proteolysis may have a quantitative relationship

wherein phosphorylation at Thr100 promotes caspase proteol-

ysis at Asp103. In this model, the lack of accumulation of the

cleaved/phosphorylated peptide could be explained by rapid

dephosphorylation of pThr100 following caspase cleavage.

We tested whether phosphorylation at Thr100 directly affects

caspase cleavage at Asp103 using in vitro peptide substrate

assays. Caspase-3 hydrolyzed the phosphorylated and unphos-

phorylated KHSRP peptides at equivalent rates (Figure 5D and

Figure S4); caspase-8, however, exhibited a dramatic increase

in hydrolytic activity (>20-fold) for the phosphorylated form of

the peptide (Figure 5D and Figure S4). The increased hydrolytic

activity of caspase-8 could be completely blocked by preincu-

bation with the inhibitor Z-VAD-fmk (Figure S4). These findings

intrigued us because it is known that caspase-8, but not

caspase-3, displays a strong preference for glutamic acid, which

is an approximate isostere of phosphorylated serine/threonine

residues (Pearlman et al., 2011), at the P3 position (Chereau

et al., 2003; Fuentes-Prior and Salvesen, 2004). These data sug-

gested that caspase-8 may have evolved a special capacity to

accommodate and even prefer phosphorylated residues at the

Figure 3. Global Crosstalk between Phosphorylation and Proteolytic P

(A) Phosphorylation events are found near sites of caspase cleavage in apoptotic

cleavage were aligned around their scissile aspartate residues (P1) and the num

enrichment of phosphorylation surrounding scissile aspartate residues is shaded

(B) Caspase-cleaved proteins are more likely to be phosphorylated than uncleav

(C) Kinase activity profiles in Jurkat T cells as measured by KiNativ analysis at 2 h

Table S4 for full KiNativ data sets generated at 1, 2, and 4 hr post-STS treatmen

(D) The most highly overrepresented motifs for apoptosis-specific phosphorylati

(E) Anti-p[S/T]-Q western blot showing increased phosphorylation of [S/T]-Q mo

(F) Anti-p[S/T]-Q blot demonstrating that DNA-PK, but not ATM or ATR kinases,

(G) Quantitative peptograph showing proteolysis of DNA-PK at 2 hr post-STS

catalytic domain.

(H) Confirmation by western blotting that, upon induction of apoptosis, DNA-PK is

from the nucleus (Nuc) to the cytoplasm (Cyto). The cleavage kinetics for DNA-

(bar graph) or p[S/T]-Q immunoreactivity (E).

Error bars represent the SEM. See also Figure S2.

P3 position. To further explore this concept, we modeled the

interaction of phosphorylated and unphosphorylated KHSRP

peptides in the active sites of caspase-3 and caspase-8 (Fig-

ure 5E and Figure S4). These models predict a clear interaction

between the pThr100 of the KHSRP substrate and an arginine

residue (Arg177) in caspase-8 that is not found in caspase-3 (Fig-

ure 5E and Figure S4). Arg177 has also been found to interact

with the P3 glutamic acid residue of inhibitors in caspase-8

cocrystal structures (Blanchard et al., 2000; Ekici et al., 2006).

Inspired by this discovery, we searched our qP-PROTOMAP

data for additional examples of apoptosis-specific Ser/Thr phos-

phorylation events occurring at the P3 position of known cas-

pase cleavage sites (Table 1). We have already briefly discussed

another such example - the apoptosis-specific pSer347 in RB1,

which is located at the P3 position adjacent to the Asp350

cleavage site. Utilizing synthetic RB1 peptides, we again found

that the phosphorylated peptide served as a much better

caspase-8 substrate compared to the unphosphorylated variant

(Figure 5F). In this case, caspase-3 also showed improved

activity for the phosphorylated peptide, but exhibited a less

dramatic increase than caspase-8 (Figure 5F). Finally, we

noticed that caspase-3 itself possesses an apoptosis-specific

phosphorylation event, pSer26 (detected at 2 hr post-STS treat-

ment), that is located at the P3 position relative to the known

caspase-cleavage site Asp29 between the prodomain and the

large catalytic subunit (Figure 5G). Cleavage at this site is

thought to occur primarily by autocatalytic processing; however,

there is some evidence that caspase-8 also proteolyzes this site

(Rank et al., 2001). As we found for KHSRP and RB1, caspase-8

displayed markedly greater hydrolytic activity for the phosphor-

ylated versus unphosphorylated caspase-3 peptide (Figure 5H).

These results, taken together, indicate that phosphorylation

can promote the caspase cleavage of proteins during apoptosis

primarily through a mechanism involving the P3 position of

caspase proteolytic sites, which, upon phosphorylation, dramat-

ically increases substrate hydrolysis by caspase-8.

DISCUSSION

The potential for crosstalk between phosphorylation and proteo-

lytic pathways in apoptosis and other cell biological processes

has long been recognized (Kurokawa and Kornbluth, 2009;

athways in Apoptosis

cells. 210 proteins observed in our data that contain known sites of caspase

ber of phosphorylation sites detected ±200 residues are shown. The region of

.

ed proteins.

r post-STS treatment relative to kinase activities measured in control cells (see

t).

on events as determined by the motif-x algorithm (also see Table S4).

tifs following induction of apoptosis.

is responsible for the apoptosis-related increase in p[S/T]-Q events.

treatment. The C-terminal persistent fragment contains the PI-3-kinase-like

cleaved and a C-terminal fragment containing the kinase domain translocates

PK match closely the kinetics of DNA-PK activation as measured by KiNativ

Cell 150, 426–440, July 20, 2012 ª2012 Elsevier Inc. 433

Table 1. Apoptosis-Specific Phosphorylation Sites Found within Six Residues (P6–P60) of Caspase Cleavage Sites

IPI Number Symbol Phosphosite Sequence Position

IPI00007423 ANP32B 158 S*DAEVDjGVDEEE P6

IPI00032064 PALM2-AKAP2 698 T*QEELDjSGLDEL P6

IPI00794135 SPTBN1 1454 S*TDEVDjSKRLTV P6

IPI00004363 STK39 387 S*DDEMDjEKSEEG P6

IPI00438229 TRIM28 683 S*LDGADjSTGVVA P6

IPI00465428 VPS13C 1400 S*QDVHDjSKNTLT P6

IPI00026156 HCLS1 111 HS*SQTDjAAKGFG P5

IPI00003168 PRPSAP2 219 ES*DLVDjGRHSPP P5

IPI00302829 RB1 882 GS*DEADjGSKHLP P5

IPI00026156 HCLS1 112 HSS*QTDjAAKGFG P4

IPI00376199 IRF2BP2 492 PAS*LPDjSSLATS P4

IPI00029822 SMARCA4 699 DVS*EVDjARHIIE P4

IPI00292140 CASP3 26 GSES*MDjSGISLD P3

IPI00855957 KHSRP 100 NNST*PDjFGFGGQ P3

IPI00604620 NCL 591 LKES*FDjGSVRAR P3

IPI00026940 NUP50 246 TEDT*PDjKKMEVA P3

IPI00302829 RB1 347 QTDS*IDjSFETQR P3

IPI00745092 SPTAN1 1484 KGDS*LDjSVEALI P3

IPI00178440 EEF1B2 153 WDDET*DjMAKLEE P2

IPI00395014 RSRC1 237 EAIES*DjSFVQQT P2

IPI00221106 SF3B2 861 KEDFS*DjMVAEHA P2

IPI00329528 VPRBP 1421 DDDDT*DjDLDELD P2

IPI00304171 H2AFY 173 KAASADjS*TTEGT P10

IPI00745092 SPTAN1 1484 TEDKGDjS*LDSVE P10

IPI00219913 USP14 228 SVKETDjS*SSASA P10

IPI00604620 NCL 595 LKESFDjGS*VRAR P20

IPI00397904 NUP93 159 GEDALDjFT*QESE P20

IPI00219913 USP14 229 SVKETDjSS*SASA P20

IPI00166394 ARMC10 89 PEDLTDjGSY*DDV P30

IPI00747447 EIF3B 164 PEDFVDjDVS*EEE P30

IPI00102670 FNBP1 522 DRESPDjGSY*TEE P30

IPI00028065 NCK1 91 KPSVPDjSAS*PAD P30

IPI00221106 SF3B2 302 EEMETDjARS*SLG P30

IPI00219913 USP14 230 SVKETDjSSS*ASA P30

IPI00455210 CHD4 367 EVTAVDjGYET*DH P40

IPI00000856 PLEKHC1 351 EVDEVDjAALS*DL P40

IPI00604620 NCL 206 DEDDDDjDEEDDS* P60

IPI00163505 RBM39 337 VTERTDjASSASS* P60

See Table S1 for corresponding peptographs of the proteins.

Lopez-Otın and Hunter, 2010); however, investigating such

network interactions at a global level has proven technically chal-

lenging due to the lack of proteomic technologies that can coor-

dinately profile protein phosphorylation and proteolysis in cells.

qP-PROTOMAP addresses this problem by quantifying phos-

phorylation events in proteomes and incorporating these modifi-

cations into the topographical maps of proteins such that their

relationship to proteolytic processing can be directly inferred.

Using this approach, we have uncovered several ways that

434 Cell 150, 426–440, July 20, 2012 ª2012 Elsevier Inc.

phosphorylation and proteolytic pathways intersect in apoptotic

cells. This crosstalk is evident on a global level by the enrichment

of phosphorylation events on proteolyzed proteins at locations

that are in close proximity to caspase cleavage sites. From a

functional perspective, we show that caspase cleavage can

unveil new sites for phosphorylation on proteins and, conversely,

apoptosis-specific phosphorylation events at the P3 position of

caspase recognition sites can promote the cleavage of proteins

(Figure 6). Caspase cleavage can also activate kinases, like

S*QQEEEEMETDS*QQEEEEMETDAR

Cleaved/unphosphorylated

A

C D

B

AQVEKEDFS*DBand 3:Band 13:

SF3B2 (4 hrs)

SILAC Ratio Spectralcounts

250kDa

150kDa

100kDa

75kDa

50kDa

37kDa

25kDa

15kDa

SF3B2 (2 hrs)

Band 3: SILAC Ratio SpectralcountsAQVEKEDFS*D

Band 3:

pSer861pSer289

F

HCLS1

SILAC Ratio Spectralcounts

E

250kDa

150kDa

100kDa

75kDa

50kDa

37kDa

25kDa

15kDa

AEVEKHSS*QTDBand 20:

pSer861 SF3B2 peptide

Ser861 SF3B2 peptide

pSer112 HCLS1 peptideSer112 HCLS1 peptide

*

***

***

Abso

lute

pro

duct

mas

s io

n in

tens

ity(a

rbitr

ary

units

)

0

10

20

30

40

0 1 2 3 4Time (hours)

Endo

geno

us p

eptid

e (fm

ols)

Cleaved/phosphorylated

Uncleaved/unphosphorylated

CASP3 CASP80

50

100

150

pSer289

pSer861

k cat

/Km

(M-1

·s-1

)

CASP3 CASP80

40

80

120

Figure 4. Caspase Cleavage Exposes New Sites for Phosphorylation

(A and B) Quantitative peptographs showing SF3B2 at 2 hr (A) and 4 hr (B) post-STS treatment. A C-terminal apoptosis-specific phosphorylation event at Ser861

occurs at the P2 position relative to the caspase cleavage site at Asp862. An additional apoptosis-specific phosphorylation event is observed at Ser289 on the

parental form of SF3B2 (band 3) at 2 hr, which is 10 residues from another site of caspase cleavage (Asp299, see band 13 in B).

(C) MS-based quantitation showing that the cleaved/unphosphorylated (Ser861) SF3B2 peptide is generated prior to the cleaved/phosphorylated (pSer861)

peptide during apoptosis. Quantified peptides: uncleaved/unphosphorylated - EQQAQVEKEDFSDMVAEHAAK, uncleaved/phosphorylated - EQQAQVEKEDFS*

DMVAEHAAK (endogenous form not detected), cleaved/unphosphorylated - EQQAQVEKEDFSD, cleaved/phosphorylated - EQQAQVEKEDFS*D.

Cell 150, 426–440, July 20, 2012 ª2012 Elsevier Inc. 435

DNA-PK, that contribute to the creation of a network of phos-

phorylation events that are specific to apoptotic cells (Figure 6).

This network is enriched in phosphorylation events that lack liter-

ature precedent, further supporting their potentially special rela-

tionship to the cell death process.

Although we do not yet understand precisely how caspase

cleavage promotes the phosphorylation of proteins, it is possible

that the kinases responsible for these phosphorylation events

cannot gain access to their substrates due to steric hindrance.

Caspase cleavage at a proximal location along the protein back-

bone could then relieve this steric blockade to expose sites for

phosphorylation (Figure 6). Alternatively, there may be kinases

that selectively phosphorylate proteins near their N or C termini,

although we are not aware of any specific kinases that have been

reported to show this substrate preference. Finally, it is possible

that cleavage promotes the redistribution of kinases like DNA-PK

to distinct subcellular compartments where they phosphorylate

new sets of substrates.

Phosphorylation events that promote proteolysis were found

to occur at the P3 position relative to caspase cleavage sites,

where they dramatically enhanced substrate hydrolysis by

caspase-8. This finding is unexpected and important because

phosphorylation events within caspase consensus motifs

(P4–P10 residues) have, in the past, been found to hinder cas-

pase cleavage (Kurokawa and Kornbluth, 2009). Our results

are, however, consistent with previous structural work on cas-

pases, which have shown that caspase-8, as well as caspase-9,

possess a unique arginine residue not found in other caspases

that enhances binding to substrates with acidic residues in the

P3 position (Blanchard et al., 2000; Chereau et al., 2003;

Fuentes-Prior and Salvesen, 2004). This feature has historically

been interpreted to explain the preference that caspase-8

displays for substrates with a P3 glutamic acid residue

(Fuentes-Prior and Salvesen, 2004), but our data suggest

another level of biological significance, namely, that caspase-8

may have evolved to recognize a set of substrates selectively

in their phosphorylated state. We should mention, however,

that so far, we have only assessed the impact of P3-phosphory-

lation on a handful of caspase substrates, and it is therefore not

yet clear whether P3-phosphorylation will serve as a general

or substrate-selective mechanism to enhance proteolysis by

caspase-8.

The intricate, systems-level interactions between kinase

and caspase networks uncovered by qP-PROTOMAP analysis

of apoptotic cells sets the stage for several important lines

of future research. First, we have only examined one cell line

(Jurkat T cells) and its response to a single apoptotic stimulus

(STS). Although the rapid and near-complete apoptotic pro-

(D) In vitro peptide substrate assays (1 mM peptide substrate) demonstrating

Peptide substrates: EQQAQVEKEDFSDMVAEHAAK and EQQAQVEKEDFS*DMV

(E) Quantitative peptograph of HCLS1 showing an apoptosis-specific phosphory

at Asp115.

(F) In vitro peptide substrate assays (1 mM peptide substrate) demonstrating t

and hinders proteolysis by caspase-8. Peptide substrates: SAVGHEYVAEV

SAVGHEYVAEVEKHSSQTD and SAVGHEYVAEVEKHSS*QTD.

See also Figure S3. For (D) and (F) error bars represent the SEM.

436 Cell 150, 426–440, July 20, 2012 ª2012 Elsevier Inc.

gression observed in STS-treated Jurkat cells has made it

a preferred model for cell biological and proteomic investiga-

tions of programmed cell death (Dix et al., 2008; Mahrus

et al., 2008; Short et al., 2007), and a recent study has shown

that different apoptotic stimuli (STS versus TRAIL) cause similar

overall patterns of protein cleavage in cells (Agard et al., 2012),

assessing the broader significance of our findings would

certainly benefit from qP-PROTOMAP studies of additional

cell types and with distinct apoptotic stimuli. Second, we do

not yet fully understand which kinases are responsible for the

phosphorylation events observed specifically in apoptotic cells.

Although our results indicate that DNA-PK makes a substantial

contribution to this apoptosis-specific phosphorylation net-

work, many of its constituent phosphorylation events do not

conform to the [S/T]-Q motif preferred by DNA-PK (Figure S2),

pointing to the potential activation of other kinases (or inactiva-

tion of phosphatases) during apoptosis. Our functional proteo-

mic data suggest candidates like AKT1 and 2, MAPK14, and

BRAF for future investigation (Figure 3C). Disrupting such

kinases could reveal the functional contribution that they (and

their cognate substrates) make to apoptosis, as has been

shown previously for DNA-PK (Bharti et al., 1998; Chakravarthy

et al., 1999; Chen et al., 2005a, 2005b; Wang et al., 2000).

Finally, there are other potential forms of crosstalk between

phosphorylation and proteolytic pathways that may have

eluded detection in our study. Phosphorylation events that,

for instance, block caspase cleavage would not have been

easily identified because the resulting phosphoprotein would

not be detected as a cleaved product. Future studies that

compare the apoptotic process under different cellular condi-

tions may reveal context-dependent changes in protein

cleavage that are due to such ‘‘protective’’ phosphorylation

events. In fact, others have speculated, for instance, that

cancer cells displaying resistance to apoptosis may possess

specific kinase networks that mark proteins with phosphoryla-

tion events that protect against caspase cleavage (Ahmed

et al., 2002). The qP-PROTOMAP method described herein

represents a versatile proteomic platform for addressing such

questions through its ability to generate global, quantitative,

and integrated profiles of phosphorylation and proteolytic path-

ways in biological systems.

EXPERIMENTAL PROCEDURES

Cell Culture and Induction of Apoptosis

Jurkat cells were grown under standard conditions and seeded to a density of

13 106 cells/ml prior to induction of apoptosis. Staurosporine (1 mM final) was

added and the cells were incubated for 2 or 4 hr at 37�C prior to lysis. See

Extended Experimental Procedures for more detail.

that phosphorylation of SF3B2 at Ser881 prevents cleavage by caspases.

AEHAAK. Peptide products: EQQAQVEKEDFSD and EQQAQVEKEDFS*D.

lation event at Ser112 occurring at the P4 position of a caspase cleavage site

hat phosphorylation of HCLS1 at Ser112 prevents proteolysis by caspase-3

EKHSSQTDAAK and SAVGHEYVAEVEKHSS*QTDAAK. Peptide products:

...NST*PD

A

C D

FE

G

B

Band 21:

KHSRP (4 hrs)

SILAC Ratio Spectral

counts...NST*PDFGFGGQKR

250kDa

150kDa

100kDa

75kDa

50kDa

37kDa

25kDa

15kDa

KHSRP (2 hrs)

Band 6: SILAC Ratio Spectral

counts

H

IIHGSES*MDSGISLDNSYK

250kDa

150kDa

100kDa

75kDa

50kDa

37kDa

25kDa

15kDa

CASP3 (2 hrs)

Band 22: SILAC Ratio Spectral

counts

CASP3 CASP8

pThr100 KHSRP peptide

Thr100 KHSRP peptide

pSer882 RB1 peptide

Ser882 RB1 peptide

pSer26 caspase-3 peptide

Ser26 caspase-3 peptide

CASP3 CASP8

***

***

***

***

**

CASP3 CASP8

0

10

20

30

0 1 2 3 4Time (hours)

En

do

ge

no

us

pe

pti

de

(fm

ols

)

Cleaved/unphosphorylated

Cleaved/phosphorylated

Uncleaved/phosphorylated

CASP8 w/ ST*PD CASP8 w/ STPD

Arg177 Arg177

Arg341 Arg341

0

50

100

150

200

250

kcat/K

m (

M-1

·s-1

)k

cat/K

m (

M-1

·s-1

)k

cat/K

m (

M-1

·s-1

)

0

500

1000

1500

2000

2500

0

100

200

300

400

0

50

100

150

Figure 5. Phosphorylation at the P3 Position of Caspase Cleavage Sites Promotes Caspase-8-Mediated Proteolysis

(A and B) Quantitative peptographs showing an apoptosis-specific phosphorylation event at Thr100 on the parental form of KHSRP (A, band 6/7) at 2 hr, and on

a half-tryptic, aspartate (Asp103)-terminating peptide of a stable fragment of this protein at 4 hr (B, band 21). Note that this half-tryptic peptide is shown in gray

because it lacks an isotopically labeled amino acid.

Cell 150, 426–440, July 20, 2012 ª2012 Elsevier Inc. 437

Inactive Kinase

Active Kinase

Caspase 3

ESMD|SG

SG

Mature

Caspase 3

Caspase 8

SG

ee

ESMD|SESMD|S

P

Substrate

XS/TXD|XX

Phosphorylated

Substrate

XS/TXD|XX

Phosphorylated

Fragment

XS/TXDXS/TXD

P

XS/TXD|XXS/TXD|X

P

P

Pho

F

1

2

2

3

3

4

1

P

S/TXXD|XX

Substrate

S/TXXD

Cleaved

fragment

S/TXXD

Phosphorylated

fragment

Figure 6. Mechanisms of Crosstalk

between Phosphorylation and Proteolytic

Pathways in Apoptosis

Several distinct forms of crosstalk between

caspase and kinase pathways were uncovered by

qP-PROTOMAP: (1) kinases (such as DNA-PK)

can be cleaved and activated by caspases; (2)

caspase cleavage can expose previously oc-

cluded residues that are then phosphorylated by

kinases; (3) phosphorylation at the P3 position

relative to scissile aspartates promotes proteolysis

of proteins by caspase-8. This type of proteolysis-

promoting P3 phosphorylation was also found on

caspase-3 itself (4).

Sample Preparation, SDS-PAGE, and Mass Spectrometry

400 mg of cytosolic protein (200 mg + 200 mg of light and heavy protein) was

separated via a 10% SDS-PAGE gel and cut into 22 0.5 cm bands. Peptides

were extracted via in-gel trypsin digestion and subjected to immobilized

metal affinity chromatography (IMAC). IMAC eluate (enriched in phosphopep-

tides) or flow-through (for unphosphorylated peptides) was loaded onto

(C) MS-based quantitation showing a rapid increase in the uncleaved/phosphorylated (pThr100) KHSRP pep

which is �1 hr prior to the appearance of the cleaved forms of this peptide. Note that the uncleaved/unphos

higher levels than the other peptides and was therefore not shown in the figure for the sake of clarity (se

uncleaved/unphosphorylated - IGGDAATTVNNSTPDFGFGGQK, uncleaved/phosphorylated - IGGDAATTVN

IGGDAATTVNNSTPD, cleaved/phosphorylated - IGGDAATTVNNST*PD.

(D) In vitro peptide substrate assays (1 mMpeptide substrate) demonstrating that phosphorylation at Thr100 of

substrates: IGGDAATTVNNSTPDFGFGGQK and IGGDAATTVNNST*PDFGFGGQK. Peptide products: IGGD

(E) Structure of caspase-8 (PDB: 1QTN) with the tetrapeptide ST*PD modeled into the active site. See the Ex

additional details.

(F) In vitro peptide substrate assays (1 mM peptide substrate) demonstrating that phosphorylation at Ser882

a lesser extent, by caspase-3. Peptide substrates: TLQTDSIDSFETQR and TLQTDS*IDSFETQR. Peptide pr

(G) Quantitative peptograph showing caspase-3 at 2 hr post-STS treatment, revealing an apoptosis-specifi

position relative to the known caspase cleavage site at Asp28.

(H) In vitro peptide substrate assays (1 mMpeptide substrate) demonstrating that phosphorylation at Ser26 pro

a lesser extent, by caspase-3. Peptide substrates: IIHGSESMDSGISLDNSYK and IIHGSES*MDSGISLDNSYK

See also Figure S4. For (D), (F), and (H), error bars represent the SEM.

438 Cell 150, 426–440, July 20, 2012 ª2012 Elsevier Inc.

a 100 mm (inner diameter) fused silica capillary

column containing C18 resin, and eluted directly

into an LTQ-Velos Orbitrap mass spectrometer

operated in data-dependent scanning mode,

with one full MS scan in the Orbitrap (60,000 reso-

lution) followed by ten MS2 scans in the ion trap.

See Extended Experimental Procedures for more

detail.

Data Analysis

MS2 data were searched using the ProLuCID algo-

rithm (Xu et al., 2006) with a reverse-concate-

nated, nonredundant variant of the human IPI

database. Peptides from each gel-band were

grouped and filtered using DTASelect (Tabb

et al., 2002), SILAC ratios were obtained with

Cimage (Weerapana et al., 2010), and these data

were assembled into quantitative peptographs

using custom software. The ‘‘phosphorylation

site data set’’ released on November 3, 2011

from the PhosphoSite database (Hornbeck et al.,

2011) was used for assessments of phosphosites.

The CASBAH database (Luthi and Martin, 2007)

was downloaded on December 19, 2011 and

used for assessments of caspase-substrates. See Extended Experimental

Procedures for more detail.

Caspase Activity Assays with Synthetic Peptide Substrates

Recombinant human caspases were diluted into buffer containing substrate

peptide and an internal standard, and incubated for 1–2 hr at 37�C. Assays

tide (yellow line) from 0 to 2 hr post-STS treatment,

phorylated KHSRP peptide was found at ten times

e Figure S4 for these data). Quantified peptides:

NST*PDFGFGGQK, cleaved/unphosphorylated -

KHSRP enhances cleavage by caspase-8. Peptide

AATTVNNSTPD and IGGDAATTVNNST*PD.

tended Experimental Procedures and Figure S4 for

of RB1 promotes cleavage by caspase-8 and, to

oducts: TLQTDSID and TLQTDS*ID.

c phosphorylation event at Ser26, which is the P3

motes cleavage of caspase-3 by caspase-8 and, to

. Peptide products: IIHGSESMDand IIHGSES*MD.

were quenched by acidification and subjected to ZipTip purification before MS

analysis. Absolute quantities of product were calculated by comparison to

synthetic standards. Assays were performed under nonsaturating substrate

concentrations and resulted in less than 20% turnover of the substrate. See

Extended Experimental Procedures for more detail.

SUPPLEMENTAL INFORMATION

Supplemental Information includes Extended Experimental Procedures, four

figures, and five tables and can be found with this article online at http://

dx.doi.org/10.1016/j.cell.2012.05.040.

ACKNOWLEDGMENTS

This work was supported by the NIH (CA087660), a Predoctoral Fellowship

from the California Breast Cancer Foundation (M.M.D.), the ARCS Foundation

(M.M.D.), a Koshland Graduate Fellowship in Enzyme Biochemistry (G.M.S.),

and the Skaggs Institute for Chemical Biology. We are grateful to Brian

Grabiner and David Sabatini for their generous gift of lentiviral vectors. This

work was supported in part by a grant from Activx Biosciences.

Received: January 3, 2012

Revised: March 26, 2012

Accepted: May 16, 2012

Published: July 19, 2012

REFERENCES

Agard, N.J., Mahrus, S., Trinidad, J.C., Lynn, A., Burlingame, A.L., and Wells,

J.A. (2012). Global kinetic analysis of proteolysis via quantitative targeted

proteomics. Proc. Natl. Acad. Sci. USA 109, 1913–1918.

Ahmed, K., Gerber, D.A., and Cochet, C. (2002). Joining the cell survival squad:

an emerging role for protein kinase CK2. Trends Cell Biol. 12, 226–230.

Allalunis-Turner, M.J., Lintott, L.G., Barron, G.M., Day, R.S., III, and Lees-

Miller, S.P. (1995). Lack of correlation between DNA-dependent protein kinase

activity and tumor cell radiosensitivity. Cancer Res. 55, 5200–5202.

Arntzen, M.O., and Thiede, B. (2011). ApoptoProteomics: An integrated data-

base for analysis of proteomics data obtained from apoptotic cells. Mol. Cell.

Proteomics 11, M111.010447.

Bharti, A., Kraeft, S.K., Gounder, M., Pandey, P., Jin, S., Yuan, Z.M., Lees-

Miller, S.P., Weichselbaum, R., Weaver, D., Chen, L.B., et al. (1998). Inactiva-

tion of DNA-dependent protein kinase by protein kinase Cdelta: implications

for apoptosis. Mol. Cell. Biol. 18, 6719–6728.

Blanchard, H., Donepudi, M., Tschopp, M., Kodandapani, L., Wu, J.C., and

Grutter, M.G. (2000). Caspase-8 specificity probed at subsite S(4): crystal

structure of the caspase-8-Z-DEVD-cho complex. J. Mol. Biol. 302, 9–16.

Casciola-Rosen, L.A., Anhalt, G.J., and Rosen, A. (1995). DNA-dependent

protein kinase is one of a subset of autoantigens specifically cleaved early

during apoptosis. J. Exp. Med. 182, 1625–1634.

Chakravarthy, B.R., Walker, T., Rasquinha, I., Hill, I.E., and MacManus, J.P.

(1999). Activation of DNA-dependent protein kinase may play a role in

apoptosis of human neuroblastoma cells. J. Neurochem. 72, 933–942.

Chen, G.G., Sin, F.L., Leung, B.C., Ng, H.K., and Poon, W.S. (2005a). Differen-

tial role of hydrogen peroxide and staurosporine in induction of cell death in

glioblastoma cells lacking DNA-dependent protein kinase. Apoptosis 10,

185–192.

Chen, G.G., Sin, F.L., Leung, B.C., Ng, H.K., and Poon, W.S. (2005b). Glioblas-

toma cells deficient in DNA-dependent protein kinase are resistant to cell

death. J. Cell. Physiol. 203, 127–132.

Chereau, D., Kodandapani, L., Tomaselli, K.J., Spada, A.P., and Wu, J.C.

(2003). Structural and functional analysis of caspase active sites. Biochemistry

42, 4151–4160.

Coleman, M.L., Sahai, E.A., Yeo, M., Bosch, M., Dewar, A., and Olson, M.F.

(2001). Membrane blebbing during apoptosis results from caspase-mediated

activation of ROCK I. Nat. Cell Biol. 3, 339–345.

Crawford, E.D., and Wells, J.A. (2011). Caspase substrates and cellular

remodeling. Annu. Rev. Biochem. 80, 1055–1087.

Degli Esposti, M., Ferry, G., Masdehors, P., Boutin, J.A., Hickman, J.A., and

Dive, C. (2003). Post-translational modification of Bid has differential effects

on its susceptibility to cleavage by caspase 8 or caspase 3. J. Biol. Chem.

278, 15749–15757.

Dix, M.M., Simon, G.M., and Cravatt, B.F. (2008). Global mapping of

the topography and magnitude of proteolytic events in apoptosis. Cell 134,

679–691.

Duncan, J.S., Turowec, J.P., Vilk, G., Li, S.S., Gloor, G.B., and Litchfield, D.W.

(2010). Regulation of cell proliferation and survival: convergence of protein

kinases and caspases. Biochim. Biophys. Acta 1804, 505–510.

Ekici, O.D., Li, Z.Z., Campbell, A.J., James, K.E., Asgian, J.L., Mikolajczyk, J.,

Salvesen, G.S., Ganesan, R., Jelakovic, S., Grutter, M.G., and Powers, J.C.

(2006). Design, synthesis, and evaluation of aza-peptide Michael acceptors

as selective and potent inhibitors of caspases-2, -3, -6, -7, -8, -9, and -10.

J. Med. Chem. 49, 5728–5749.

Fattman, C.L., Delach, S.M., Dou, Q.P., and Johnson, D.E. (2001). Sequential

two-step cleavage of the retinoblastoma protein by caspase-3/-7 during

etoposide-induced apoptosis. Oncogene 20, 2918–2926.

Fuentes-Prior, P., and Salvesen, G.S. (2004). The protein structures that shape

caspase activity, specificity, activation and inhibition. Biochem. J. 384,

201–232.

Han, Z., Malik, N., Carter, T., Reeves, W.H., Wyche, J.H., and Hendrickson,

E.A. (1996). DNA-dependent protein kinase is a target for a CPP32-like

apoptotic protease. J. Biol. Chem. 271, 25035–25040.

Hehner, S.P., Hofmann, T.G., Droge,W., and Schmitz, M.L. (1999). Inhibition of

tyrosine phosphatases induces apoptosis independent from the CD95

system. Cell Death Differ. 6, 833–841.

Hornbeck, P.V., Kornhauser, J.M., Tkachev, S., Zhang, B., Skrzypek, E.,

Murray, B., Latham, V., and Sullivan, M. (2011). PhosphoSitePlus: a compre-

hensive resource for investigating the structure and function of experimentally

determined post-translational modifications in man and mouse. Nucleic Acids

Res. 40, D261–D270.

Kim, S.T., Lim, D.S., Canman, C.E., and Kastan, M.B. (1999). Substrate spec-

ificities and identification of putative substrates of ATM kinase family

members. J. Biol. Chem. 274, 37538–37543.

Janes, K.A., Albeck, J.G., Gaudet, S., Sorger, P.K., Lauffenburger, D.A., and

Yaffe, M.B. (2005). A systems model of signaling identifies a molecular basis

set for cytokine-induced apoptosis. Science 310, 1646–1653.

Kurokawa, M., and Kornbluth, S. (2009). Caspases and kinases in a death grip.

Cell 138, 838–854.

Lopez-Otın, C., and Hunter, T. (2010). The regulatory crosstalk between

kinases and proteases in cancer. Nat. Rev. Cancer 10, 278–292.

Luthi, A.U., and Martin, S.J. (2007). The CASBAH: a searchable database of

caspase substrates. Cell Death Differ. 14, 641–650.

Mahrus, S., Trinidad, J.C., Barkan, D.T., Sali, A., Burlingame, A.L., and Wells,

J.A. (2008). Global sequencing of proteolytic cleavage sites in apoptosis by

specific labeling of protein N termini. Cell 134, 866–876.

Na, S., Chuang, T.H., Cunningham, A., Turi, T.G., Hanke, J.H., Bokoch, G.M.,

and Danley, D.E. (1996). D4-GDI, a substrate of CPP32, is proteolyzed during

Fas-induced apoptosis. J. Biol. Chem. 271, 11209–11213.

Ong, S.E., Blagoev, B., Kratchmarova, I., Kristensen, D.B., Steen, H., Pandey,

A., and Mann, M. (2002). Stable isotope labeling by amino acids in cell culture,

SILAC, as a simple and accurate approach to expression proteomics. Mol.

Cell. Proteomics 1, 376–386.

Patricelli, M.P., Nomanbhoy, T.K., Wu, J., Brown, H., Zhou, D., Zhang, J.,

Jagannathan, S., Aban, A., Okerberg, E., Herring, C., et al. (2011). In situ kinase

profiling reveals functionally relevant properties of native kinases. Chem. Biol.

18, 699–710.

Cell 150, 426–440, July 20, 2012 ª2012 Elsevier Inc. 439

Pearlman, S.M., Serber, Z., and Ferrell, J.E., Jr. (2011). A mechanism for the

evolution of phosphorylation sites. Cell 147, 934–946.

Rank, K.B., Mildner, A.M., Leone, J.W., Koeplinger, K.A., Chou, K.C., Tomas-

selli, A.G., Heinrikson, R.L., and Sharma, S.K. (2001). [W206R]-procaspase 3:

an inactivatable substrate for caspase 8. Protein Expr. Purif. 22, 258–266.

Schwartz, D., and Gygi, S.P. (2005). An iterative statistical approach to the

identification of protein phosphorylation motifs from large-scale data sets.

Nat. Biotechnol. 23, 1391–1398.

Short, D.M., Heron, I.D., Birse-Archbold, J.L., Kerr, L.E., Sharkey, J., and

McCulloch, J. (2007). Apoptosis induced by staurosporine alters chaperone

and endoplasmic reticulum proteins: Identification by quantitative proteomics.

Proteomics 7, 3085–3096.

Song, Q., Lees-Miller, S.P., Kumar, S., Zhang, Z., Chan, D.W., Smith, G.C.,

Jackson, S.P., Alnemri, E.S., Litwack, G., Khanna, K.K., and Lavin, M.F.

(1996). DNA-dependent protein kinase catalytic subunit: a target for an ICE-

like protease in apoptosis. EMBO J. 15, 3238–3246.

Stolzenberg, I., Wulf, S., Mannherz, H.G., and Paddenberg, R. (2000). Different

sublines of Jurkat cells respond with varying susceptibility of internucleosomal

DNA degradation to different mediators of apoptosis. Cell Tissue Res. 301,

273–282.

440 Cell 150, 426–440, July 20, 2012 ª2012 Elsevier Inc.

Tabb, D.L., McDonald, W.H., and Yates, J.R., III. (2002). DTASelect and

Contrast: tools for assembling and comparing protein identifications from

shotgun proteomics. J. Proteome Res. 1, 21–26.

Thornberry, N.A., and Lazebnik, Y. (1998). Caspases: enemies within. Science

281, 1312–1316.

Tozser, J., Bagossi, P., Zahuczky, G., Specht, S.I., Majerova, E., and

Copeland, T.D. (2003). Effect of caspase cleavage-site phosphorylation on

proteolysis. Biochem. J. 372, 137–143.

Wang, S., Guo, M., Ouyang, H., Li, X., Cordon-Cardo, C., Kurimasa, A., Chen,

D.J., Fuks, Z., Ling, C.C., and Li, G.C. (2000). The catalytic subunit of DNA-

dependent protein kinase selectively regulates p53-dependent apoptosis

but not cell-cycle arrest. Proc. Natl. Acad. Sci. USA 97, 1584–1588.

Weerapana, E., Wang, C., Simon, G.M., Richter, F., Khare, S., Dillon, M.B.,

Bachovchin, D.A., Mowen, K., Baker, D., and Cravatt, B.F. (2010). Quantitative

reactivity profiling predicts functional cysteines in proteomes. Nature 468,

790–795.

Xu, T., Venable, J.D., Park, S.K., Cociorva, D., Lu, B., Liao, L.,Wohlschlegel, J.,

Hewel, J., and Yates, J.R., III. (2006). ProLuCID, a Fast and Sensitive Tandem

Mass Spectra-based Protein Identification Program. Mol. Cell. Proteomics 5,

S174.

Supplemental Information

EXTENDED EXPERIMENTAL PROCEDURES

Cell Culture and Induction of ApoptosisJurkat cells were grown at 37�C under 5% CO2 in RPMI 1640 media supplemented with 10% fetal calf serum (FCS) and 2 mM

glutamine. For metabolic labeling (SILAC), cells were maintained in RPMI media containing 2 mM glutamine and light or heavy

arginine and lysine (Sigma) were supplemented at a concentration of 100 mg/ml. Cells were passaged six times in heavymedia before

testing for full incorporation of the heavy amino acids. Prior to induction of apoptosis, Jurkat cells were seeded to a density of 13 106

cells/ml in RPMI 1640 media containing 10% FCS and 2 mM glutamine. DMSO or staurosporine (1 mM final concentration, Sigma)

was added and the cells were incubated for 1, 2, or 4 hr at 37�C under 5% CO2. For kinase inhibition experiments, 1 3 106 cells/ml

were pretreated with inhibitor for 1 hr prior to addition of staurosporine. Kinase inhibitors [CGK 733 (10 mM final), KU55933 (10 mM

final), NU-7441 (1 mM final), and NU-7026 (10 mM final)] were purchased from Tocris, with the exception of CGK 733, which was

purchased from Calbiochem. For caspase inhibition experiments, 1 3 106 cells were pre-treated with 60 mM z-VAD-FMK (Roche)

or DMSO for 1 hr prior to STS addition.

Preparation of Cell LysatesSoluble and particulate fractions were prepared as previously described (Jessani et al., 2002). Briefly, cells were washed 3 times

in cold PBS and resuspended in 200 ml of PBS containing protease inhibitors (complete EDTA-free protease inhibitor cocktail,

Roche), phosphatase inhibitors (PHOSTOP, Roche), and z-VAD-FMK (Roche). Cells were then sonicated to lyse and centrifuged

at 100,000 x g for 45 min. The supernatant was collected as the soluble fraction. For experiments containing cytosolic and nuclear

fractions, samples were prepared according to manufacturers instructions (NE-PER Nuclear and Cytosolic Extraction Kit, Pierce).

Sample Preparation, SDS-PAGE, and IMAC EnrichmentFor phosphopeptide enrichment, 200 mg of each control (light) and apoptotic (heavy) soluble fraction were combined and separated

via a 10% SDS-PAGE gel for 850 V hr. The gel was washed in water and manually excised into twenty-two 0.5 cm bands. Bands

corresponding to the migration of molecular-weight markers were noted and this information was used to estimate the molecular

weights of proteins migrating in each band. Bands were subjected to in-gel trypsin digestion as previously described (Rosenfeld

et al., 1992) with minor modifications. Briefly, bands were washed in 100 mM ammonium bicarbonate and proteins were reduced

in 10 mM tris(2-carboxyethyl)phosphine (TCEP) at 37�C for 0.5 hr and then alkylated with 55 mM iodoacetamide in the dark for

0.5 hr. The bandswere then dehydrated bywashing in 50:50 acentonitrile:100mMammonium bicarbonate, followed by 100%aceto-

nitrile. Gel bands were then dried and resuspended in 40 ml of trypsin at 10 ng/ml. Upon re-swelling of the gel bands, 25 mM ammo-

nium bicarbonate was added to a final volume of 200 ml and the gel bands were placed at 37�C overnight. Supernatants containing

peptides were removed, and the gel bandswere further extracted with 5% formic acid and acetonitrile and dried down via speed vac.

For phosphopeptide enrichment, the dried peptides were resuspended in 350 ml of IMAC binding buffer (250 mM acetic acid, 30%

acetonitrile). 30 ml of equilibrated IMAC slurry (PHOS-Select, Sigma) was added to each band. Samples were then placed on a rotator

at room temperature for 1.5 hr. The peptide/slurry mixture was then washed twice with 800 ml binding buffer followed by one wash

with 300 ml water. Peptide elution was accomplished with 300 ml of 400 mM NH4OH and then dried in a speed vac and stored at

�80�C prior to use. The 4 hr data set was derived from five separate biological replicates and consisted of three IMAC flow-through

(nonphosphopeptide) samples and six IMAC eluate (enriched for phosphopeptides) samples. The 2 hr data set was derived from two

separate biological replicates and consisted of one IMAC flow-through sample and two IMAC eluate samples.

Mass Spectrometric AnalysisPhosphopeptides and unenriched peptideswere analyzed separately via LC-MS/MS in the sameway: peptides fromeach bandwere

resuspended in 10 ml buffer A (95% H2O, 5% acetonitrile, 0.1% formic acid) and loaded via autosampler onto a 100 mm (inner

diameter) fused silica capillary column with a 5 mm tip that was packed with 10 cm of C18 resin (aqua 5mm, Phenomenex).

LC-MS/MS analysis was performed on an LTQ-Velos Orbitrap mass spectrometer (ThermoFisher) coupled to an Agilent 1200 series

HPLC. Peptides were eluted from the column using a 2 hr gradient of 5%–100% buffer B (5% H2O, 95% acetonitrile, 0.1% formic

acid). The flow rate through the column was 0.25 ml/min and the spray voltage was 1.7 kV. The mass spectrometer was operated

in data-dependant scanning mode, with one full MS scan (400-1,800 m/z) occurring in the Orbitrap (60,000 resolution) followed

by ten MS2 scans of the nth most abundant ions with dynamic exclusion enabled (20 s duration).

Data Processing, Analysis, and DepositionRaw mass spectrometry data were stored as RAW files generated by XCalibur version 2.1.0.1139 running on a Thermo Scientific

LTQ-Velos Orbitrap mass spectrometer. RAW files were converted to MS2 format (McDonald et al., 2004) using RAW-Xtract version

1.8 and these MS/MS data were searched using ProLuCID (Xu et al., 2006). ProLuCID searches were performed using a reverse-

concatenated nonredundant variant of the human IPI database version 3.33. Cysteine residues were required to be carboxyamido-

methylated (+57.02146 Da) and up to three differential phosphorylation marks (+79.9663 Da) were permitted on serine, threonine, or

tyrosine residues in each peptide. Peptides were required to have at least one tryptic terminus. ProLuCID data from each gel band

were quality-filtered and sorted with DTASelect version 2.0.25, which performs linear discriminant analyses within each charge- and

Cell 150, 426–440, July 20, 2012 ª2012 Elsevier Inc. S1

modification-state to achieve a peptide false-positive rate below 1% (Tabb et al., 2002). Actual peptide false-positive rates (as

defined in Elias and Gygi, 2007) were below 1% at this stage (0.9% and 0.7% in the 2- and 4 hr data sets, respectively). RAW

data were also converted to mzXML format (Pedrioli et al., 2004) using ReAdW version 4.3.1 and SILAC quantitation was performed

using an in-house software package called Cimage (described in Weerapana et al., 2010). Cimage was run using a 10 ppm mass-

window and requiring an R2 correlation value for light/heavy co-elution of 0.8 or greater. Peptides that were found exclusively in

healthy or apoptotic cells (only light or heavy peak observed) were considered candidate singlet peptides and were subject to

additional filters to remove singlet peaks with insufficient abundance or excessive background noise. At this stage the peptide false

positive rates were 0.6% and 0.3% in the 2- and 4 hr data sets, respectively, indicating that incorporation of MS1-based SILAC infor-

mation improves accuracy of peptide identifications. Peptides were then assembled into quantitative peptographs using in-house

software. This process involved additional unbiased peptograph-level noise-filtering steps such as requirement that at least two

distinct peptides be observed in a given band across all replicates. No reverse (decoy) peptides remained in either data set following

application of these filters. All peptographs detected in both data sets can be viewed at: http://www.scripps.edu/chemphys/cravatt/

protomap/. SILAC ratios and chromatographs from every peptide can also be found at this website. Additionally, a complete descrip-

tion including quality scores for peptide spectral matches (XCorr and deltaCN values, Eng et al., 1994) of the 17,396 phosphopep-

tides detected in this study can be found in Table S2. Finally, all raw data generated in this study was deposited in the Proteome

Commons repository (Smith et al., 2011), and can be accessed at https://proteomecommons.org using the following hashes: BFKUG

bPTPJBlQIchJClUgSsF7nQVaMfsf2beMkuBiKBts4J6SworZyln/U92D9R1ZJHHHohg024CtAyl+ptQpieu/HkAAAAAAAA09A== and

Bz6bOoFiA2IIo7acYv8jXfB6u2rvxcpUh8qdKxaVYVkkiC3tTNsGOvsfkSD0MxBH42G54G4g5j6/gQsXzp/Kvq9ff3UAAAAAAACa1Q==

for the 2- and 4 hr data sets, respectively. Assessments of phosphosites were derived from the ‘‘phosphorylation site data set’’

released on November 3, 2011, from the PhosphoSitePlus database (Hornbeck et al., 2011), except for Figure 2C and Table S5,

which were updated with the most-recent data available on the PhosphoSite website on April 18, 2012. The CASBAH database,

which catalogs the known substrates of apoptotic proteolysis (Luthi and Martin, 2007) was downloaded on December 19, 2011,

and used for assessments of caspase-substrates.

Identification of Cleaved ProteinsCleaved proteins were identified on the basis of the distribution of peptide-ratios in each band. Only proteins and fragments of

sufficient abundance were considered: two spectral counts from at least two distinct peptide sequences were required in a given

band and eight spectral counts from at least four distinct peptides were required for each protein. The distribution of peptide ratios

in each band were organized into quartiles and, if the ratios in the upper three quartiles were more than 3-fold elevated in the control-

cells, the bandwas flagged as control specific, indicative of a parental degradation event. If the ratios in the lower three quartiles were

more than 3-fold elevated in the apoptotic cells, then the band was flagged as apoptosis specific, indicative of a cleaved fragment.

Proteins that did not display apoptosis-specific fragments were further categorized as partially cleaved (where some bands were

control specific and others were not) versus completely cleaved (where all bands showed control-specific ratios). Entries were

flagged as candidate cleaved proteins if: (1) they displayed apoptosis-specific fragments at either time point, or (2) they showed

complete parental cleavage at the 4 hr time point. Of the 2,867 proteins that met the abundance thresholds, 837 proteins were

flagged as potential substrates of apoptotic proteolysis using this algorithm. This list was thenmanually pruned to remove ambiguous

or mis-categorized protein entries resulting in a final high-confidence list of 744 proteins that are cleaved or degraded during

apoptosis (Table S1).

Classification of Phosphorylation Sites and Motif-x AnalysisSILAC ratios for all peptides containing a given phosphorylated residue were extracted from the 2- and 4 hr data sets. Those sites

with peptides displaying SILAC ratios that were all at least 2-fold enriched in apoptotic cells were deemed apoptosis specific. The

remaining phosphosites had SILAC ratios that were either unchanged (‘‘static,’’ either displaying less than 2-fold change in either

direction or displaying both control-specific and apoptosis-specific ratios indicative of a static phosphorylation event on a cleaved

protein) or control specific (at least 2-fold enriched in control-cells). For the DNA-PK inhibitor experiments, phosphorylation sites

were categorized using the same algorithm described above, except that sites displaying two-fold or greater reduction upon

treatment with NU-7441 were designated ‘‘suppressed by NU-7441’’ and all other sites were classified as ‘‘insensitive’’ (Table

S4). For motif analysis, sequences surrounding each phosphosite (+/� 9 residues, referred to as ‘‘sequons’’) were extracted from

our 2- and 4 hr data sets and analyzed with the motif-x algorithm (Schwartz and Gygi, 2005) using either serine or threonine as

the central residue. Apoptosis-specific sequons were submitted in one batch and control-specific and static sequons were

submitted separately (Figure 3D and Table S4). The human proteome (ipi.HUMAN.fasta) was used as a background database

and the significance and number-of-occurrences and significance thresholds were set at 20 and 0.000001, respectively.

Western BlottingEither the soluble or the cytoplasmic and nuclear fractions of both control and staurosporine-treated Jurkat cells were analyzed via

western blotting using standard methods. Blots were probed with antibodies for caspase-3, PARP1, DNA-PK, pS/T-Q, actin, lamin,

and RB1 (Cell Signaling Technology numbers 9662, 9542, 4602, 2851, 4970, 2032, and 9313, respectively). The GRAP2 antibody was

from R&D Systems (AF4640).

S2 Cell 150, 426–440, July 20, 2012 ª2012 Elsevier Inc.

Lentiviral KnockdownDNA-PK shRNApLKO.1 lentiviral contructs were purchased fromOpenBiosystems. Short hairpin-plasmid DNA, alongwith envelope

protein (psPAX2) and coat protein (CMV-VSVG) vectors were co-transfected into HEK293T cells. The virus-containing media con-

taining was collected and filtered. Polybrene was added to the filtered media to a final concentration of 10 mg/ml. Varying amounts

of virus-containing media were then used to infect Jurkat cells. Two days postinfection, Jurkat cells were resuspended in selection

media containing 1 mg/ml puromycin. 7 days postselection, cells were collected and nuclear fractions were prepared. DNA-PK

knockdown efficiency was measured by western blot (Figure S1). Sequences for the clones that gave the best knockdowns were

CCAGTGAAAGTCTGAATCATT and GCAGCCTTATTACAAAGACAT.

Peptide Quantification by LC-MS1x106 cells/ml of Jurkat cells were treated with 1 mM staurosporine and collected every 30 min for 4 hr. The cells were then washed

three times with PBS and resuspended in PBS containing protease inhibitors, phosphatase inhibitors, and z-VAD-FMK. The soluble

fraction was isolated via high-speed centrifugation (see preparation of cell lysates) and then 200 mgwas subjected to SDS-PAGE. Gel

bands corresponding to relevantmolecular weights were excised and digested in-gel with trypsin (as described above). Isotoptically-

labeled peptides (incorporating heavy lysine and arginine residues where present) were then added to tryptic digests prior to analysis

via LC-MS/MS. When quantification of phosphopeptides was needed, an IMAC enrichment step was performed prior to LC-MS/MS

analysis. LC-MS/MS analyses utilized targeted fragmentation by targeting the mass-to-charge ratios of relevant peptides for MS2

fragmentation. Peptides masses were then extracted, and a diagnostic MS2 ion was selected for quantitation via ‘‘pseudo-

MRM.’’ This quantification method is referred to as pseudo-MRM because, unlike true MRM (multiple reaction monitoring, typically

performed on a triple-quadrupole mass spectrometer) all of the fragment ion masses are measured in the ion-trap, rather than

isolating a single daughter ion for quantification. Quantitation is then performed at the software level, after-the-fact, by measuring

peaks comprised of a ‘‘transition’’ from parent ion to one of several diagnostic daughter ions (a similar approach is described in detail

in Scherl et al., 2008).

KiNativ Profiling of Active KinasesJurkat cells were plated at 13 106 cells/ml and treated with 1 mM STS for 1, 2, or 4 hr. Cells were then washed, pelleted, lysed in cell

lysis buffer (25 mM Tris pH 7.6, 150 mM NaCl, 1% CHAPS, 1% Tergitol NP-40 type, 1% v/v phosphatase inhibitor cocktail II [EMD/

Calbiochem, #524625]), and sonicated. Lysates were filtered and probe reactions were performed at room temperature with a final

probe concentration of 5 mM. Samples were labeled with both Biotin-Hex-Acyl-ATP and Biotin-Hex-Acyl-ADP probes, as previously

described (Patricelli et al., 2007). All reactions were performed in duplicate. Labeled lysates were denatured and reduced, alkylated,

and digested with trypsin as described above. Desthiobiotinylated peptides were captured using high-capacity streptavidin resin

(Thermo Scientific), and analyzed by mass spectrometry as previously described. The full results of these KiNativ profiling experi-

ments can be found in Table S4.

Substrate AssaysSynthetic peptides were purchased from Thermo (HeavyPeptide AQUA standards) and diluted in assay buffer containing 50 mM

HEPES, 100 mM NaCl, 0.1% CHAPS, 1 mM EDTA, 10% glycerol, and 10 mM DTT to a final reaction volume of 10 ml. Recombinant

human caspase-8 or caspase-3 (carrier free, R&D Systems) was diluted to 100 nM into assay buffer containing substrate peptide

as well as an internal standard peptide that does not serve as a caspase substrate. Samples were incubated for varying lengths

of time (120 min for KHSRP substrate peptides and 60 min for SF3B2, Caspase 3, RB1, and HCLS1) at 37�C. Assays were then

quenched by acidification with formic acid (0.5%final) and then subjected to ZipTip purification (Millipore) beforemass spectrometric

analysis on an LTQ-Velos Orbitrapmass spectrometer using a gradient that consisted of a 30min loading phase followed by a 30min

gradient from 5% to 100% B (95% acetonitrile, 5% water, 0.1% formic acid). Ionization efficiencies of the product peptides relative

to the internal standard were calculated using synthetic standards. These values were then used to calculate absolute rates of

product formation. All assays were performed under nonsaturating substrate concentrations for a period of time that resulted in

less than 20% turnover of the substrate (typically 1 mM substrate concentration for 60-120 min). Peptide substrates used for the

assays are listed here, with the phosphorylated residue shown in bold. SF3B2: substrate - EQQAQVEKEDFSDMVAEHAAK, product -

EQQAQVEKEDFSD. HCLS1: substrate - SAVGHEYVAEVEKHSSQTDAAK, product – SAVGHEYVAEVEKHSSQTD. RB1: substrate –

TLQTDSIDSFETQR, product – TLQTDSID. KHSRP: substrate – IGGDAATTVNNSTPDFGFGGQK, product – IGGDAATTVNNSTPD.

Caspase-3: substrate – IIHGSESMDSGISLDNSYK, product – IIHGSESMD. These peptide substrates were the only ones tested

for the studies presented in this manuscript (i.e., no additional peptide substrate data were omitted from inclusion in the manuscript).

Structural ModelingModels of caspase-substrate peptide interaction were generated using the flexible peptide docking program (Raveh et al., 2010)

released in the Rosetta 3.0 software suite (Leaver-Fay et al., 2011). Crystal structures of caspase-8 (PDB: 1QTN) and caspase-3

(PDB: 1PAU) boundwith tetrapeptide analogs were used as the starting template for modeling. First, the unphosphorylated substrate

peptide sequences (STPD for KHSRP or ESMD for caspase-3) was grafted onto the peptide backbone in the crystal structure at the

catalytic cysteine. Second, the rigid-body orientation between the catalytic cysteine in the caspase and the P1 aspartic acid in the

Cell 150, 426–440, July 20, 2012 ª2012 Elsevier Inc. S3

peptide substratewas fixed tomimic the acyl-enzyme intermediate, as observed in the crystal structures. The rest of enzyme-peptide

interaction across the binding interface was locally refined usingMonte-Carlo energyminimization (MCM) by varying torsion angles in

all protein sidechains as well in the peptide backbone and sidechains. Out of the 500 structural models generated, the one ranked by

the best total energy was selected as the final model. Based on the models with unphosphorylated peptides, the phosphorylated

peptides (ST*PD or ES*MD) were generated by replacing the P3 serine or threonine residue with a phosphorylated variant that

has an additional sidechain torsion angle allowing the position of the phosphate group to be energetically optimized. The same

MCM-based flexible peptide docking method was applied to refine the local interaction across the caspase-peptide interface and

after 50 structures were generated, the best-rankedmodel by total energy was subjected to a further round of gradient-based energy

minimization to optimize the electrostatic interaction between the phosphorylated residue and its neighboring residues.

SUPPLEMENTAL REFERENCES

Dix, M.M., Simon, G.M., and Cravatt, B.F. (2008). Global mapping of the topography and magnitude of proteolytic events in apoptosis. Cell 134, 679–691.

Elias, J.E., and Gygi, S.P. (2007). Target-decoy search strategy for increased confidence in large-scale protein identifications by mass spectrometry.

Nat. Methods 4, 207–214.

Eng, J.K., McCormack, A.L., and Yates Iii, J.R. (1994). An approach to correlate tandem mass spectral data of peptides with amino acid sequences in a protein

database. J. Am. Soc. Mass Spectrom. 5, 976–989.

Hornbeck, P.V., Kornhauser, J.M., Tkachev, S., Zhang, B., Skrzypek, E., Murray, B., Latham, V., and Sullivan, M. (2011). PhosphoSitePlus: a comprehensive

resource for investigating the structure and function of experimentally determined post-translational modifications in man and mouse. Nucleic Acids Res.

Jessani, N., Liu, Y., Humphrey, M., and Cravatt, B.F. (2002). Enzyme activity profiles of the secreted and membrane proteome that depict cancer cell invasive-

ness. Proc. Natl. Acad. Sci. USA 99, 10335–10340.

Leaver-Fay, A., Tyka, M., Lewis, S.M., Lange, O.F., Thompson, J., Jacak, R., Kaufman, K., Renfrew, P.D., Smith, C.A., Sheffler, W., et al. (2011). ROSETTA3: an

object-oriented software suite for the simulation and design of macromolecules. Methods Enzymol. 487, 545–574.

Luthi, A.U., and Martin, S.J. (2007). The CASBAH: a searchable database of caspase substrates. Cell Death Differ. 14, 641–650.

McDonald, W.H., Tabb, D.L., Sadygov, R.G., MacCoss, M.J., Venable, J., Graumann, J., Johnson, J.R., Cociorva, D., and Yates, J.R., III (2004). MS1, MS2, and

SQT-three unified, compact, and easily parsed file formats for the storage of shotgun proteomic spectra and identifications. Rapid Commun.Mass Spectrom. 18,

2162–2168.

Patricelli, M.P., Szardenings, A.K., Liyanage, M., Nomanbhoy, T.K., Wu, M., Weissig, H., Aban, A., Chun, D., Tanner, S., and Kozarich, J.W. (2007). Functional

interrogation of the kinome using nucleotide acyl phosphates. Biochemistry 46, 350–358.

Patricelli, M.P., Nomanbhoy, T.K., Wu, J., Brown, H., Zhou, D., Zhang, J., Jagannathan, S., Aban, A., Okerberg, E., Herring, C., et al. (2011). In situ kinase profiling

reveals functionally relevant properties of native kinases. Chem. Biol. 18, 699–710.

Pedrioli, P.G., Eng, J.K., Hubley, R., Vogelzang, M., Deutsch, E.W., Raught, B., Pratt, B., Nilsson, E., Angeletti, R.H., Apweiler, R., et al. (2004). A common open

representation of mass spectrometry data and its application to proteomics research. Nat. Biotechnol. 22, 1459–1466.

Raveh, B., London, N., and Schueler-Furman, O. (2010). Sub-angstrom modeling of complexes between flexible peptides and globular proteins. Proteins 78,

2029–2040.

Rosenfeld, J., Capdevielle, J., Guillemot, J.C., and Ferrara, P. (1992). In-gel digestion of proteins for internal sequence analysis after one- or two-dimensional gel

electrophoresis. Anal. Biochem. 203, 173–179.

Scherl, A., Shaffer, S.A., Taylor, G.K., Kulasekara, H.D., Miller, S.I., and Goodlett, D.R. (2008). Genome-specific gas-phase fractionation strategy for improved

shotgun proteomic profiling of proteotypic peptides. Anal. Chem. 80, 1182–1191.

Schwartz, D., and Gygi, S.P. (2005). An iterative statistical approach to the identification of protein phosphorylation motifs from large-scale data sets. Nat. Bio-

technol. 23, 1391–1398.

Smith, B.E., Hill, J.A., Gjukich, M.A., and Andrews, P.C. (2011). Tranche distributed repository and ProteomeCommons.org. Methods Mol. Biol. 696, 123–145.

Tabb, D.L., McDonald, W.H., and Yates, J.R., III (2002). DTASelect and Contrast: tools for assembling and comparing protein identifications from shotgun

proteomics. J. Proteome Res. 1, 21–26.

Weerapana, E., Wang, C., Simon, G.M., Richter, F., Khare, S., Dillon, M.B., Bachovchin, D.A., Mowen, K., Baker, D., and Cravatt, B.F. (2010). Quantitative reac-

tivity profiling predicts functional cysteines in proteomes. Nature 468, 790–795.

Xu, T., Venable, J.D., Park, S.K., Cociorva, D., Lu, B., Liao, L.,Wohlschlegel, J., Hewel, J., and Yates, J.R., III (2006). ProLuCID, a Fast and Sensitive TandemMass

Spectra-based Protein Identification Program. Mol. Cell. Proteomics 5, S174.

S4 Cell 150, 426–440, July 20, 2012 ª2012 Elsevier Inc.

A B

C

Figure S1. Exemplary Quantitative Peptograph, Related to Figure 1

(A) Western blot and peptograph for a representative cleaved and phosphorylated protein GRAP2, where apoptosis-specific (blue) and control-specific (red)

phosphorylated peptides are marked by numbers 1 and 2, respectively. Control-specific (red) and apoptosis-specific (blue) unphosphorylated peptides are also

marked on the parental (3) and fragment (4) bands of GRAP2, respectively. Note that the western blot confirmed reductions in the parental form of GRAP2 and the

appearance of at least one of persistent fragment in apoptotic cells. An additional persistent fragment was detected by qP-PROTOMAP that was not observed by

western blotting. As we have noted previously (Dix et al., 2008), PROTOMAP methods can detect fragments of proteins that are difficult to visualize by western

blotting methods, which are reliant on antibody reagents that may recognize only a small number of epitopes on proteins.

(B) Representative MS1 chromatographs for peptides 1–4. Asterisks in the MS1 traces designate the times at which MS2 spectra were acquired. Asterisks in the

listed peptide sequences designate phosphorylated residues.

(C) Description of the box-and-whisker plots used in the central panels of quantitative peptographs.

Cell 150, 426–440, July 20, 2012 ª2012 Elsevier Inc. S5

A B

C D

FE

G

Figure S2. Crosstalk between Phosphorylation and Proteolysis during Apoptosis, Related to Figure 3

(A) Detection of new sites of phosphorylation occurred at a diminishing rate as additional data were acquired. The orange line represents the number of unique

sites of phosphorylation identified for a particular number of phosphopeptides (green line).

(B) The diminishing rate of new phosphopeptide discovery with increasing replicate experiments was observed for both static/control-specific (red line) and

apoptosis-specific (blue line) phosphorylation events.

(C) Apoptosis-specific (cyan) and static/control-specific (orange) phosphorylation events are both enriched in regions surrounding the scissile aspartate residues

of known sites of caspase cleavage.

(D) Lentiviral knockdown of DNA-PK. See the Extended Experimental Procedures for details.

(E) Left: The increase in p[S/T]-Q phosphorylation observed upon treatment of Jurkat cells with STS is blocked by pre-treatment with the caspase inhibitor Z-VAD-

fmk treatment also blocked DNA-PK cleavage (right).

(F) Treatment of Jurkat cells with the DNA-PK inhibitor NU-7441 prior to the induction of apoptosis results in a two-fold or greater reduction in the majority of p[S/

T]-Q phosphorylation events. In contrast, all other non-p[S/T]-Q events were largely unaffected by NU-7441 treatment. The vast majority of NU-7441-sensitive p

[S/T]-Q sites (>80%) were apoptosis specific (pie-chart).

(G) Representative examples of apoptosis-specific (i, iii, iv) and static (ii) phosphorylation events that are (i, iii) or are not (ii, iv) blocked byNU-7441-treatment. Note

that examples i and ii represent S-Q and non-S-Q phosphorylation sites on the same protein, respectively, demonstrating the specificity of DNA-PK and the

selectivity of the NU-7441 inhibitor. The chromatographs on the left represent the 2 hr apoptotic data set, with control-cells in red and apoptotic (STS) cells in blue.

In the chromatographs on the right side, the blue trace again represents 2 hr apoptotic data, and the green line represents cells treated with both STS and NU-

7441 for 2 hr.

S6 Cell 150, 426–440, July 20, 2012 ª2012 Elsevier Inc.

0

50

100

150

200

250

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5

0

5

10

15

20

25

30

35

40

0 1 2 3 4 5 6 7 8 9

Ab

solu

te m

ass

ion

inte

nsi

ty(a

rbit

rary

un

its)

fmo

l/(m

in*p

mo

l e

nzym

e)

Caspase 3

pSer861 SF3B2 peptide

Ser861 SF3B2 peptide

Caspase 3

pSer112 HCLS1 peptide

Ser112 HCLS1 peptide

A

C

B

0%

10%

20%

30%

40%

50%

60%

0 1 2 3 4 > 4

Number of references in PhosphoSiteDB

Ph

osp

ho

Sit

es (

%)

Control-specific or static

Apoptosis-specific

Apoptosis-specific sites on half-tryptic peptides

+/- 6 residues from scissile aspartate

Figure S3. In Vitro Substrate Assays Showing Linearity of Product Formation over the Tested Range of Substrate Concentrations, Related to

Figure 4

(A) SF3B2 peptide substrates: EQQAQVEKEDFSDMVAEHAAK and EQQAQVEKEDFS*DMVAEHAAK. SF3B2 product peptides: EQQAQVEKEDFSD and EQ-

QAQVEKEDFS*D. Note that the phosphorylated peptide substrate is not turned over at any concentration.

(B) HCLS1 peptide substrates: SAVGHEYVAEVEKHSSQTDAAK and SAVGHEYVAEVEKHSS*QTDAAK. HCLS1 product peptides: SAVGHEYVAEVEKHSSQTD

and SAVGHEYVAEVEKHSS*QTD. Note that the phosphorylated peptide is not turned over at any concentration by caspase-3, although caspase-8 displays

moderate activity with this substrate (see Figure 4).

(C) Phosphorylation events occurring within six amino acids of scissile aspartate residues in apoptotic proteomes are overrepresented in previously unreported

phosphorylation sites.

See also Table S5.

Cell 150, 426–440, July 20, 2012 ª2012 Elsevier Inc. S7

O

N

O

OO

N

OO

N

S

ON

O

OS

NN

N

O

N

NN

ON

N

NN

P OO

O

OO

NO

N

NN

N

Trp348 Arg341

Ser339

Gly237

His236Arg207

Ser205

Gly122His121

Cys163

Gln161

Phe250

Gln208

Trp214

Arg207Arg64

Cys279

Gln277

Arg179Arg341

Arg341 Arg177

CASP8 w/ ES*MD CASP3 w/ ES*MD CASP3 w/ ESMDCASP8 w/ ESMD

Arg177

Arg177

Arg341

O

N

O

OO

N

O O

N

S

ON

O

OS

N

NN

ON

N

NN

P OO

O

OO

NO

NN

N

O

O

O

N

N

Arg207 Arg207

CASP3 w/ STPDCASP8 w/ STPD

Arg177 Arg177Arg341 Arg341 Arg207 Arg207

A

B

C

G

D E F

fmol

/(m

in*p

mol

enz

yme)

0

10

20

30

40

50

60

70

80

0

0

2

4

6

3 6 9

Caspase 3

pThr100 KHSRP peptide

Thr100 KHSRP peptide

fmol

/(m

in*p

mol

enz

yme)

Caspase 8

pThr100 KHSRP peptide

Thr100 KHSRP peptide

0

20

40

60

80

100

120

140

160

180

0 3 6 9

fmol

/(m

in*p

mol

enz

yme)

100nM Caspase 3

100nM Caspase 3+

z-VAD-fmk

pThr100 KHSRPpeptide

Thr100 KHSRPpeptide

fmol

/(m

in*p

mol

enz

yme)

0

2

4

6

8

10

12

14

100nM Caspase 8

100nM Caspase 8+

z-VAD-fmk

*** *** *** **

pThr100 KHSRPpeptide

Thr100 KHSRPpeptide

0

100

200

300

400

500

600

2 hours 4 hours

En

do

gen

ou

s p

ep

tid

e (

fmo

ls)

Uncleaved/unphosphorylatedUncleaved/phosphorylatedCleaved/unphosphorylatedCleaved/phosphorylated

Figure S4. Phosphorylation at the P3 Position Relative to the Scissile Aspartate Enhances Substrate Hydrolysis by Caspase-8, Related to

Figure 5

(A and B) Phosphorylated and unphosphorylated tetrapeptide substrates representing the caspase-3 sequence containing pSer26 (ES*MD, (A) or the KHSRP

sequence containing pThr100 (ST*PD, (B) were modeled into the active sites of caspase-8 or caspase-3 (PDB: 1QTN and 1PAU, respectively, see Extended

Experimental Procedures for details). Hydrogen bonding interactions with the P3 residues are shown as dashed yellow lines. The lower panels in (A) show

schematic representations of the interactions with the phosphorylated substrates. Hydrogen bonding interactions (<4A) are shown as dashed lines. Notably,

Arg177 in caspase-8 interacts with the phosphorylated, but not unphosphorylated substrates, and caspase-3 does not contain a homologous cationic residue.

Note that the left panels of (B) are identical to the ones shown in Figure 5E, and are reproduced here for clarity.

(C–F) Representative in vitro substrate assays with a KHSRP peptide containing the Thr100 residue showing linear response of product formation by caspase-3

and �8 over the tested range of substrate concentrations. Caspase-3 and caspase-8 proteolytic activity were completely blocked by pre-incubation with the

caspase inhibitor z-VAD-FMK (D) and (F), 1 mM peptide substrate). Peptide substrates: IGGDAATTVNNSTPDFGFGGQK and IGGDAATTVNNST*PDFGFGGQK.

Product peptides: IGGHAATTVNNSTPD and IGGHAATTVNNST*PD. Similar assays were conducted for RB1 and caspase-3 substrate peptides with both

enzymes, but are not shown due to space constraints. (G) Quantitation of endogenous KHSRP peptides shows that the absolute amounts of uncleaved/

phosphorylated peptide and cleaved/unphosphorylated peptide were similar at their respective peak accumulation values (2 and 4 hr time points, respectively).

Peptide sequences are described in Figure 5.

Data in (D) and (F) are presented as means ± SEM. **p < 0.01, ***p < 0.001.

S8 Cell 150, 426–440, July 20, 2012 ª2012 Elsevier Inc.