You are on page 1of 11

DOI: 10.1002/JLB.

MR0718-269R

REVIEW

Mechanisms of natural killer cell-mediated cellular cytotoxicity

Isabel Prager Carsten Watzl

Department for Immunology, Leibniz Research


Centre for Working Environment and Human Abstract
Factors at TU Dortmund (IfADo), Cellular cytotoxicity, the ability to kill other cells, is an important effector mechanism of the
Dortmund, Germany
immune system to combat viral infections and cancer. Cytotoxic T cells and natural killer (NK)
Correspondence
cells are the major mediators of this activity. Here, we summarize the cytotoxic mechanisms of
Carsten Watzl, Leibniz Research Centre for
Working Environment and Human Factors NK cells. NK cells can kill virally infected of transformed cells via the directed release of lytic
(IfADo) at TU Dortmund, Ardeystrasse 67, 44139 granules or by inducing death receptor-mediated apoptosis via the expression of Fas ligand or
Dortmund, Germany. TRAIL. The biogenesis of perforin and granzymes, the major components of lytic granules, is a
Email: watzl@ifado.de
highly regulated process to prevent damage during the synthesis of these cytotoxic molecules.
Additionally, NK cells have developed several strategies to protect themselves from the cytotoxic
activity of granular content upon degranulation. While granule-mediated apoptosis is a fast pro-
cess, death receptor-mediated cytotoxicity requires more time. Current data suggest that these
2 cytotoxic mechanisms are regulated during the serial killing activity of NK cells. As many modern
approaches of cancer immunotherapy rely on cellular cytotoxicity for their effectiveness, unravel-
ing these pathways will be important to further progress these therapeutic strategies.

KEYWORDS
apoptosis, death receptors, degranulation, granzyme, perforin

1 INTRODUCTION these cytotoxic cells against tumors and checkpoint inhibitor therapy
enhances the activity of cytotoxic T cells. These examples highlight the
Natural killer (NK) cells are cytotoxic innate lymphocytes that can importance of understanding the mechanisms of cellular cytotoxicity.
lyse cancerous or virally infected cells.1 Their activity is regulated In the following, we will summarize the current knowledge about the
by a variety of germline-encoded receptors that can mediate either cytotoxic mechanisms of NK cells.
activating or inhibitory signals and thereby elicit NK cell responses
while also ensuring self-tolerance. NK cells can mediate their cyto-
toxic activity via 2 distinct pathways. They can release cytotoxic gran- 2 PRODUCTION AND STORAGE
ules containing perforin and granzymes, or they can induce death OF GRANZYME AND PERFORIN
receptor-mediated apoptosis by expressing TRAIL and/or Fas ligand IN LYTIC GRANULES
(FasL) to engage TRAIL-R1/-R2 or CD95/Fas, respectively, on the sur-
face of diseased cells. NK cell or T cell-mediated cellular cytotoxicity One mechanism of cellular cytotoxicity is the targeted release of
are essential for many approaches of immunotherapy against cancer.2 lytic granules that contain granzymes and perforin. The synthesis of
Antibody-based therapies such as the anti-CD20 antibody rituximab these cytotoxic proteins is tightly regulated. Granzymes are a family
against non-Hodgkin’s lymphoma depend on antibody-dependent of closely related serine proteases that are expressed in cytotoxic
cellular cytotoxicity (ADCC) mediated by NK cells. Chimeric antigen T cells and NK cells.3 Most studies have investigated the biosynthesis
receptor T and NK cell therapies aim to directly target the activity of of granzyme B (Figure 1), which is expressed as a pre-pro-protein
(zymogen) containing an inhibitory dipeptide and an N-terminal
signal peptide that directs it to the endoplasmic reticulum (ER).4
Abbreviations: ADCC, antibody dependent cellular cytotoxicity; AIF, apoptosis-inducing After synthesis into the ER the signal peptide is cleaved off, resulting
factor; DISC, death-inducing signaling complex; EGF, epidermal growth factor; ER, in an inactive proenzyme. This proenzyme is then modified with a
endoplasmic reticulum; FasL, Fas ligand; M6P, mannose-6-phosphate; MACPF, membrane
attack complex/perforin; MTOC, microtubule organizing center; NK, natural killer; ROS,
mannose-6 phosphate (M6P) moiety by the N-acetylglucosamine-1-
reactive oxygen species; SB9, serpin B9. phosphodiester 𝛼-N-acetyl-glucosaminidase in the cis compartment of

Received: 7 February 2019 Revised: 22 March 2019 Accepted: 14 April 2019


J Leukoc Biol. 2019;105:1319–1329. www.jleukbio.org 2019
c Society for Leukocyte Biology 1319
1320 PRAGER AND WATZL

Granzyme B
cis trans
Mannose-6-Phosphate
Golgi Golgi
Inhibitory dipeptide
ER
Signal peptide
Serglycin

pro- lytic granule


granzyme
Nucleus

pre-pro- Active
granzyme granzyme
?
pH 5.5 cathepsin

F I G U R E 1 Biogenesis of granzymes. Most of the events described in this figure stem from studies investigating granzyme B. Granzymes are
synthesized into the ER as a pre-pro-protein. In the ER, the signal peptide is cleaved off, resulting in inactive progranzyme. In the cis-Golgi,
mannose-6-phosphate is added, providing the necessary signal for sorting into lytic granules. In the granules, mannose-6-phosphate may be
removed due to the acidic pH and the inhibitory dipeptide is cleaved by cathepsin C or cathepsin H. Active granzymes are stored in complex
with serglycin

the Golgi apparatus, which provides a sorting signal that is necessary sary for perforin protein stability and it regulates the export of perforin
to translocate the proenzyme to the secretory granules via the M6P- from the ER to the lytic granules.18 However, studies using endogenous
receptor.5 Inside the secretory granules, M6P may be removed due to instead of recombinant perforin showed that N-linked glycosylation of
the acidic environment of pH 5.56 ; however, it was also suggested that the C-terminal dodecapeptide inhibits its oligomerization and thereby
granzyme B may be able to retain its M6P modification.7 The inhibitory its pore-forming activity.19 This suggests a role of the glycosylated C-
dipeptide is then removed by the cysteine proteases cathepsin C or terminal dodecapeptide in preventing premature pore formation dur-
cathepsin H, which transform the granzyme into its active form.5,8,9 ing the biosynthesis of perforin. Additionally, fast and highly efficient
Granzymes are then stored in their active form inside the lytic gran- trafficking of perforin monomers from the ER to the Ca2+ -poor Golgi
ules. However, their activity inside the granules is inhibited by the apparatus may be key to protect cytotoxic lymphocytes from perforin
low pH and by associating with the chondroitin sulfate proteoglycan toxicity.16 This may be supported by calreticulin, a chaperone in the
serglycin,10 thereby preventing the cleavage of host cell proteins. As ER, which was also identified in lytic granules.20,21 Perforin contains
an alternative to this secretory pathway, some granzymes may also be the binding motif sequence KVFF (single amino acid code), which could
secreted via a constitutive nongranule biosynthesis pathway.11 mediate interaction with calreticulin and thereby limit its pore-forming
Perforin consists of 3 domains: The N-terminal membrane attack activity while in transit from the ER to the lytic granules.21–23 Also,
complex perforin like (MACPF)-dependent cytolysin pore-forming LAMP-1 (CD107a), which is widely used as a degranulation marker,
domain, a central epidermal growth factor (EGF) domain, and a C- has a critical role in trafficking of perforin to the lytic granules.24 Once
terminal membrane and Ca2+ -binding C2 domain.12 The pore-forming inside the granules, the C-terminus of perforin is cleaved by cathep-
activity of perforin is dependent on its polymerization and it is pH sin L25 and other proteases,19 enabling the activity of perforin once
and Ca2+ -dependent. Therefore, perforin is not active at the acidic pH it is secreted from the inhibitory acidic environment of the lytic gran-
of the lytic granules, even if Ca2+ is present.13,14 However, it is still ules. Additionally, there seems to be a second mechanism to transfer
unclear how cytotoxic lymphocytes evade the toxicity of perforin while newly synthesized perforin to the immunologic synapse independently
processing and trafficking it from the ER, containing a neutral environ- of secretory granules.26
ment and high Ca2+ -levels, to the lytic granules. One hypothesis is that
perforin is expressed as a zymogen similar to granzymes, carrying a
C-terminal dodecapeptide to block Ca2+ -binding to its C2 domain.15 3 THE ABC OF GRANZYMES
However, recombinant perforin with and without the dodecapeptide
shows the same pore-forming activity16 and the X-ray structure of per- The family of human granzymes consists of 5 members: granzyme A,
forin indicates that its C-terminus is diverted away from Ca2+ -binding granzyme B, granzyme H, granzyme K, and granzyme M.27 Together
region of the C2 domain,17 thereby challenging this hypothesis. Addi- with perforin and granulysin, they represent the major cytotoxic com-
tionally, Ca2+ binding to the C2 domain actually seems to be neces- ponents of secretory granules of NK cells and cytotoxic T cells. While
PRAGER AND WATZL 1321

much is known about the functions of granzyme A and B (reviewed in pase activation, mitochondrial disruption, or the generation of ROS,
Refs. 27–31), the other granzymes have received little attention until but instead induces a novel form of cell death termed “athetosis,”
a few years ago, although they were also discovered in the late 1980s which was dependent on the actin cytoskeleton.54 Therefore, several
and early 1990s. In the following section, we summarize the effect of aspects of granzyme-induced cell death are controversial.41 Granzyme
each granzyme individually. However, it should be noted that NK cells A also has species-specific effects. Human granzyme A was reported
release several different granzymes during their degranulation, so that to be poorly cytotoxic39 and may even be more involved in promoting
the killing of a target cell is likely a result of a combined action of proinflammatory immune responses,55 thereby challenging the above
different granzymes. described cell death inducing activity.

3.1 Granzyme B 3.3 Granzyme K


Granzyme B was initially named “cytotoxic T lymphocyte-associated Similar to granzyme A, granzyme K has a trypsin-like activity and
serine esterase 1” or granzyme 2. This serine protease has a wide cleaves substrates after basic residues Arg or Lys.56 It induces rapid
substrate specificity, but it preferentially cleaves after Asp residues. caspase-independent cell death with the production of ROS from mito-
One prominent substrate of granzyme B is caspase-3. Cleavage chondria, chromatin condensation, and apoptotic nuclear morphol-
of caspase-3 and of caspase-7 results in their activation and sub- ogy analogous to granzyme A.57–59 It also cleaves the nucleosome
sequently induces caspase-dependent cell death by cleavage of assembly protein SET to induce single-stranded DNA nicks.57,58 There-
different cellular substrates, including a number of structural and regu- fore, granzyme K seems to be a “backup” granzyme for the activ-
latory proteins in the nucleus, in the cytosol and at the cytoskeleton.32 ity of granzyme A, which is supported by the fact that granzyme
In addition to this, granzyme B can also induce cell death by cleav- A-deficient cytotoxic T cells can still mediate cell death via granzyme
age of the BH3-only protein Bid.33 The resulting truncated form, tBid, K.60 Granzyme K also shows some unique function by cleaving the
then relocates to the mitochondria where it interacts with the pro- pre-mRNA-binding protein heterogeneous ribonuclear protein K.61
apoptotic proteins Bax and/or Bak.34 This results in a disruption of However, granzyme K-deficient mice do not show impaired antiviral
the mitochondrial membrane integrity,35 the release of apoptotic fac- responses, suggesting that granzyme K does not play an essential role
tors such as smac/DIABLO, cytochrome c, apoptosis-inducing factor in this process.62
(AIF), and Omi/HtrA2,33 and subsequently to apoptosis of the cell.
This granzyme B-induced apoptosis can be inhibited by overexpres-
3.4 Granzyme M
sion of the antiapoptotic protein Bcl-2.35,36 Interestingly, human and
mouse granzyme B show differences in their substrate specificity.37–39 Granzyme M has an unusual enzyme specificity, preferring cleavage
While granzyme B from both species can cleave and activate caspase- after methionine or leucine.63 Together with perforin it rapidly induces
3 and caspase-7, only human granzyme B is very efficient in cleaving cell death that is independent of caspases, fails to trigger hallmark
Bid. Therefore, caspase inhibitors, Bid deletion or Bcl-2 overexpression mitochondrial changes, occurs in the absence of discernible DNA frag-
can have different effects on granzyme B-mediated cell death depend- mentation and cannot be inhibited by Bcl-2.64 One granzyme M sub-
ing on the species. This has created some controversies in the field in strate is the nucleolar phosphoprotein nucleophosmin, which induces
the past and it is important to note that other granzymes also show cell death upon cleavage.65 Additionally, granzyme M can cleave and
species-specific differences.39–41 inactivate Serpin B9 (SB9),66 an inhibitor of granzyme B, which likely
promotes granzyme B-induced apoptosis after degranulation. In con-
3.2 Granzyme A trast, Serpin B4 was found to be an inhibitor of granzyme M-induced
cell death.67 Granzyme M is specifically expressed in NK cells, CD56+
Granzyme A has a trypsin-like activity and cleaves substrates after
T cells and 𝛾𝛿-T cells with little to no detectable expression in CD4 or
basic residues Arg or Lys. It can induce a fast form of cell death
CD8 T cells.68
independently of caspases and without the release of apoptotic fac-
tors from mitochondria, creating large DNA fragments, which are not
3.5 Granzyme H
detectable by usual apoptosis assays.42–45 Within minutes, it activates
the production of reactive oxygen species (ROS) from the mitochon- Not much is known about the function of granzyme H, which shows
dria, cripples the mitochondrial electron transport and disrupts the homology to granzyme B, but which does not seem to have an ortholog
mitochondrial membrane potential.46 It targets the ER-associated SET in rodents. Despite its homology, granzyme H has a chymotrypsin-
complex, which translocates to the nucleus where SET is cleaved by like activity with a cleavage preference for hydrophobic amino acid
granzyme A, releasing 2 nucleases, which mediate DNA damage.47 residues (Phe or Tyr)69 and is predominantly expressed in NK cells.70
The exonuclease TREX1 is in the SET complex and acts in concert Granzyme H was shown to cleave the adenovirus DNA-binding pro-
with NM23–H1 to degrade DNA during granzyme A-mediated cell tein, thereby interfering with viral replication in infected cells and to
death.48 Other granzyme A substrates include HMG-2,49 lamins,50 his- inactivate the adenovirus 100K assembly protein, a major inhibitor
tone H1,51 Ku70,50,52 and the DNA damage sensor poly(adenosine of granzyme B.71 Additionally, it can cleave the cellular phosphopro-
5′ -diphosphate-ribose) polymerase-1.53 Conflicting reports describe tein La, which is necessary for hepatitis C virus replication,72 demon-
that cell death induced by mouse granzyme A does not require cas- strating interesting antiviral activities of this granzyme. Granzyme H
1322 PRAGER AND WATZL

A NK cell B
Granule Vesicle

Granzyme

Perforin FasL
TRAIL
TRAIL R FAS
FADD

Caspase
Granzyme –8 or –10

BID Caspase-
cleavage
Caspase-
Target cell independent
Mitochondrium

Apoptosis

F I G U R E 2 Two ways to kill. (A) Granule-mediated cytotoxicity is initiated by the targeted release of lytic granules toward a locally attached
target cell. Granzymes can then enter the target cell by perforin-pores in the plasma membrane (left) or by endocytosis and perforin-aided escape
from endosomes (right). Granzymes can then induce caspase activation, mitochondrial dysfunction, or caspase-independent apoptosis. (B) Death
receptor-mediated cytotoxicity is induced by surface expression of FasL or TRAIL, which can engage and activate their respective receptor. This
results in caspase activation, mitochondrial dysfunction, and apoptosis

can induce cell death with mitochondrial depolarization, ROS, DNA organizing center (MTOC) using dynein motor proteins and are then
degradation, and chromatin condensation, which is independent of polarized toward the synapse together with the MTOC.87,88 At the
caspase or Bid cleavage and does not involve cytochrome c release synapse, the granules fuse with the plasma membrane with the help
from mitochondria.73 of small GTPases of the Rab family, MUNC proteins and the formation
of a SNARE complex, resulting in the release of granular content into
the synaptic cleft (reviewed in Ref. 89). The synapse forms a gasket-
3.6 Granulysin
like structure with the help of LFA-1-mediated adhesion, which lim-
Granulysin was found as a gene to be induced 3–5 days after T cell its the diffusion of granular content and promotes granule-mediated
activation.74 It is expressed by cytotoxic T cells and NK cells and cytotoxicity.90
belongs to the saposin-like protein family.75 Granulysin is synthesized In the neutral pH of the extracellular space perforin can bind to
as a 15 kDa molecule, which is then processed to a 9 kDa form by the target cell membrane and in the presence of Ca2+ the monomers
proteolytic cleavage within the lysosomal compartment.76 The pro- aggregate and form pores in the membrane (Figure 2).91 These
cessed Granulysin is found in lytic granules, whereas the 15 kDa ring-shaped pores have a structure equivalent to a transmembrane
form can be constitutively secreted.77 Granulysin has cytolytic activ- channel with a pore lumen of 13–20 nm17 that may even coa-
ity against tumors and microbes, including Gram-positive and Gram- lesce into larger pores by recruiting additional perforin oligomers.92
negative bacteria, fungi/yeast and parasites.78–81 Like other members As granzyme B has a stokes radius of 5 nm, it should be able to
of the saposin-like protein family, it has pore-forming activity, which enter the target cell directly through the perforin pores,93 which
alters membrane permeability and results in osmotic lysis and mito- was demonstrated in recent studies.94,95 As an alternative mech-
chondrial damage.82,83 anism of granzyme entry into target cells, it was proposed that
granzymes are taken up by the target cell via endocytosis.96 This may
be facilitated by binding of granzyme B to the mannose 6-phosphate
4 NK CELL DEGRANULATION AND ENTRY receptor7,97 or by membrane repair processes induced by perforin-
OF GRANZYMES INTO THE TARGET CELL mediated membrane damage and Ca2+ influx.98 Granzymes would
then escape their endocytotic vesicles with the help of perforin.99
The killing of virally infected or malignant transformed cells by NK However, given the speed of perforin pore formation at the plasma
cells is a multistep process, which is initiated by integrin-mediated membrane that was shown to deliver granzymes within 80 s,95 it is
adhesion of the NK cell to the target cell84,85 and the formation of likely that the direct delivery of granzymes through perforin pores at
the immunologic synapse (reviewed in Ref. 86). Lytic granules are the plasma membrane is the dominant pathway for granzyme deliv-
transported along the microtubules in the direction of the microtubule ery. However, it cannot be excluded that a slower endocytic delivery
PRAGER AND WATZL 1323

of granzymes operates in parallel as a potential backup mechanism TRAIL, which can bind to different TRAIL receptors (reviewed in
to guarantee efficient delivery of granzymes into the cytosol of the Ref. 111). Cytotoxic lymphocytes can use all 3 receptor systems to kill
target cell. target cells. In the following, we will focus on the regulation of FasL and
TRAIL-mediated killing (Figure 2).

5 SELF-PROTECTION OF NK CELLS
7 FasL-MEDIATED CYTOTOXICITY
NK cells and cytotoxic T cells have various mechanisms to prevent
cellular damage during the synthesis and the storage of perforin and The FasL is a 40-kDa type II transmembrane protein belonging to the
granzymes. But what protects these cells once the content of the lytic TNF superfamily.112 It can be expressed on the cell surface of acti-
granules is released into the synaptic cleft? In other words, why does vated T cells and NK cells.113 After synthesis in the ER, it is transported
perforin not attack the NK cell membrane, delivering granzymes into via the Golgi apparatus to secretory lysosomes. For this trafficking,
the cytosol and thereby killing the NK cell? The delivery of perforin a proline-rich domain in the cytoplasmic tail of FasL is important.114
pores at the synapse appears to be unidirectional only toward the tar- Kinases associate with this domain, resulting in phosphorylation of
get cell95 and cytotoxic lymphocytes were shown to have some form of FasL, which is essential for its trafficking.115 In addition, ubiquitina-
resistance toward cell-mediated cytotoxicity.100 Several mechanisms tion of sequences adjacent to the proline-rich region is necessary for
have been discussed to account for this resistance. First, cytotoxic the correct sorting of FasL to secretory granules. As FasL is stored
T cells and NK cells express the granzyme B inhibitor SB9 in their in these secretory granules, surface expression is dependent on the
cytosol, which may protect them from misdirected granzyme B that degranulation of T cells and NK cells.116,117 However, there seem to
escapes the exocytotic pathway.101 SB9 contains a C-terminal reac- exist 2 distinct species of secretory granules in T cells and NK cells
tive center loop that represents a substrate for granzyme B.102 Cleav- that either contain perforin and granzymes or FasL118–120 indicating
age of this SB9 loop by granzyme B induces a rapid conformational that preformed FasL is stored in secretory granules that are distinct
change in the serpin, resulting in the formation of a stable serpin– from classical cytotoxic granules that contain perforin and granzymes.
proteinase complex, effectively inactivating granzyme B.103,104 As a Therefore, the signals necessary for the externalization and surface
second mechanism, it was shown that CD107a/LAMP-1 protects the expression of FasL may be different from the ones required for clas-
NK cell from degranulation-associated self-injury.105 CD107a is exter- sical degranulation.121 FasL brought to the surface by degranulation
nalized during the degranulation of lytic granules and thereby lines will quickly diffuse in the plasma membrane,122 but diffusion may be
the NK cell plasma membrane at the immunologic synapse. CD107a constrained within the synapse by LFA-1-mediated adhesion.90 Alter-
can inhibit the binding of perforin to the NK cell plasma membrane, natively, FasL may be transiently exposed to the surface by incom-
thereby preventing damage to the NK cell. A similar potential regu- plete granule fusion.122 Upon surface expression of FasL it can engage
lator of perforin-mediated lysis is the lipid packing. It was proposed and multimerized the CD95 receptor on a locally attached target cell
that certain phospholipids are more tightly packed in the membrane and thereby activate its apoptotic signaling cascade. Membrane-bound
of the effector cell, which can reduce the capacity of perforin to FasL can be cleaved by metalloproteinases123–125 and the resulting sol-
bind to or to intercalate with the lipid bilayer.106 As a final mech- uble FasL does not exert cytotoxicity126 and may in some cases even
anism, it was proposed that cathepsin B, which also reaches the inhibit apoptosis induced by membrane-bound FasL.126–129 Stimula-
plasma membrane of cytotoxic lymphocytes during degranulation, can tion of the CD95 receptor activates an intracellular signaling program,
cleave and inactivate perforin.107 Interestingly, tumor cells may use which starts with the assembly of the death-inducing signaling complex
the same mechanism of secreting cathepsin B at the synapse to escape (DISC), leading to the activation of caspases 8 and 10, subsequently
killing by cytotoxic T cells.108 However, in cathepsin B-deficient mice, resulting in a caspase cascade, the depolarization of the mitochon-
cytotoxic lymphocytes did not show enhanced apoptosis during the drial membrane potential and ultimately in the induction of apoptosis
killing of target cells,109 demonstrating that cathepsin B is not nec- (reviewed in Ref. 110).
essary for the self-protection of cytotoxic cells. However, it is likely
that the protection of cytotoxic cells from their own lytic machinery
is such an important aspect that there are several redundant mecha-
8 TRAIL-mediated cytotoxicity
nisms at play to guarantee the survival of the cell even if one of these
mechanisms should fail.
TRAIL is a type II transmembrane protein with homology to FasL
and TNF.130 It is expressed by cytotoxic cells such as T cells131 and
6 DEATH RECEPTOR-MEDIATED KILLING NK cells,132 but it can also be found on other cells such as mono-
cytes, macrophages and dendritic cells.132,133 While freshly isolated
In addition to the exocytosis of lytic granules, cytotoxic lymphocytes NK cells do not express detectable levels of TRAIL on their surface,
can also kill target cells by engaging so-called death receptors. There stimulation with IL-2, IL-15 or IL-12 can induce the surface expres-
are 3 different receptor/ligand systems that can mediate apoptosis sion of TRAIL.134,135 TRAIL-mediated cytotoxicity has been shown to
upon activation: TNF, binding to either TNF receptor-1 or -2, FasL be important for NK cell-mediated control of viral infections and can-
engaging the CD95 (APO-1/Fas) receptor (reviewed in Ref. 110), and cer, especially for liver NK cells.135–137 In Jurkat T cells TRAIL was
1324 PRAGER AND WATZL

shown to colocalize with CD107a/LAMP1 positive vesicles.138 How- ever, in type II cells, formation of the DISC is inefficient and only little
ever, the mechanism of how TRAIL traffics to the lysosomal compart- caspase 8 is activated, which is insufficient to process caspase 3, but
ment and if it is stored in the same vesicles as FasL is unknown. It is also sufficient to cleave the BH3-only protein Bid, resulting in the apopto-
unclear, how target cell exposure can induce the surface expression of genic activation of mitochondria and making CD95-induced apoptosis
TRAIL within the synapse. In case of cytokine-induced expression, it is in type II cells sensitive to Bcl-2 expression. Therefore, depending on
unlikely that TRAIL is concentrated at any specific part of the NK cell the cell type, FasL-induced apoptosis is a complex process that involves
surface. Therefore, the question arises how this surface expression can many regulated steps. In contrast, granzyme B can directly cleave and
contribute to directed target cell killing. In addition to the membrane- activate caspase 3 upon entry of the target cell and thereby induce
bound form, TRAIL can also be shed by the activity of an unknown cys- a very efficient and fast apoptotic program. The combination of dif-
teine protease.130 However, in contrast to soluble FasL, soluble TRAIL ferences in mobilization, necessary dose and apoptotic signaling cas-
can retain its apoptotic activity and potentially result in killing of cells cades are probably responsible for the differences in kinetics between
within a close proximity of TRAIL secreting NK cells. granule- and death receptor-mediated cytotoxicity.
Human TRAIL can bind to 4 distinct membrane receptors and 1
soluble receptor.111 TRAIL-R1 and TRAIL-R2 are the only receptors
that can induce apoptosis,139,140 whereas TRAIL-R3, TRAIL-R4 and 10 SERIAL KILLING
osteoprotegerin mostly induce NF-𝜅B signaling and may function as
decoy receptors to limit TRAIL binding to the pro-apoptotic TRAIL Soon after the discovery of cellular cytotoxicity, it was shown
receptors.139,141,142 Proapoptotic signaling of TRAIL-R1 and TRIAL- that individual cytotoxic T cells can kill multiple targets in a serial
R2 is very similar to CD95-mediated signaling, with the formation of fashion.151–153 More recent studies have shown a similar serial killing
a DISC and the activation of caspase-8.111 activity for NK cells.154–156 While the number of target cells killed by
an individual NK cell varies between studies, it became clear that only a
minority of NK cells is responsible for the majority of killing events.156
9 KINETICS OF GRANULE- AND DEATH However, it is currently unclear how to identify and how to isolate this
RECEPTOR-MEDIATED CYTOTOXICITY subpopulation of serial killing NK cells. It will be important to charac-
terize these cells, as they would be ideal effector cells for therapeutic
Cytotoxic T cells and NK cells can use the release of perforin and applications of NK cells. The killing of multiple targets by a single
granzymes from lytic granules and the surface expression of ligands NK cell is a strictly serial process, as the NK cell polarizes its lytic gran-
for death receptors to kill virally infected or transformed cells (Fig- ules with the help of the microtubules and the MTOC,157 facilitated
ure 2). When comparing the kinetics of these 2 cytotoxic mechanisms, by signals from the integrin LFA-1.117 However, MTOC polarization
several studies documented that death receptor-mediated killing is may not be strictly necessary for degranulation.158 This polarization is
slower.143–145 For example, while lytic granule-mediated killing can needed for efficient cytotoxicity and for preventing bystander killing.
be observed within minutes after target cell contact, death receptor- However, this also implies that the NK cell may only form 1 lytic
mediated killing by perforin-deficient cytotoxic cells can take 1–2 h.146 synapse with a target cell at a time. After a successful kill, the NK cell
There are several possible explanations for this kinetic difference. needs to establish a new contact with another target cell. For this, the
As perforin and granzymes were suggested to be in different gran- detachment from the killed target cell is necessary. Interestingly, also
ules compared to FasL or TRAIL, their timing of mobilization may this detachment is a regulated process,159 and it may be aided by the
differ. CD107a externalization as a marker for the release of lytic downregulation or shedding of activating receptors.160,161 Once a new
granules147 can be observed within minutes after NK cell activation. contact is established, the killing process can start all over again. How-
We only observed clearly detectable accumulation of FasL on the sur- ever, there is a limit to how many targets an individual NK cell can
face after about 30 min to 1 h after target cell contact (unpublished kill. The degranulation of as little as 2–4 granules was shown to be
data), whereas others observed low levels of FasL within 15–30 min.117 sufficient to cause cell death.148 However, NK cells release about 10%
Another explanation could be that the dose necessary to induce tar- of their total granules in a single killing event.148 Therefore, during
get cell death may differ between perforin/granzyme and FasL-induced the process of serial killing the NK cells lose the majority of their
killing. While the release of as little as 2–4 lytic granules was shown to preformed perforin and granzyme B.154 While this may be compen-
be sufficient to induce target cell death148 it may take more degranu- sated over time by de-novo synthesis, NK cells also lose activating
lation events to accumulate the amount of surface FasL necessary to receptors from their surface159–161 and signaling pathways may be
induce apoptosis. And finally, the kinetics of the 2 cell death pathways attenuated. These events limit the serial killing activity of NK cells.
could differ inside the target cell. Engagement of CD95 induces recep- We recently developed fluorescence localization reporters to mea-
tor micro-aggregates, followed by the assembly of the DISC and the sure granule and death receptor-mediated killing of tumor cells by
induction of larger receptor clusters.149 DISC formation can be seen NK cells.162 Using these reporters, we could show that NK cells use
within just 1 min after receptor triggering, while the formation of larger granule-mediated killing for their first serial killing events, while
clusters takes about 15 min. In addition, 2 different CD95 apoptosis they switch to death receptor-mediated killing once their perforin
pathways were described.150 In type I cells, caspase 8 is activated at and granzymes have been depleted (unpublished results). This sug-
the DISC in quantities sufficient to directly activate caspase 3. How- gests that NK cells alternate their cytotoxic mechanisms in a kinetic
PRAGER AND WATZL 1325

fashion during serial killing. Cellular cytotoxicity and serial killing are 7. Motyka B, Korbutt G, Pinkoski MJ, et al. Mannose 6-
crucial for the effectiveness of monoclonal antibody-based anticancer phosphate/insulin-like growth factor II receptor is a death receptor
for granzyme B during cytotoxic T cell-induced apoptosis. Cell.
therapeutics154,160 and likely contribute to the effectiveness of many
2000;103:491-500.
cellular immunotherapies against cancer. Therefore, it will be impor-
8. Smyth MJ, McGuire MJ, Thia KY. Expression of recombinant human
tant for future studies to identify subpopulations of NK cells with the granzyme B. A processing and activation role for dipeptidyl peptidase
best serial killing abilities and to find ways to enhance this activity. I. J Immunol. 1995;154:6299-6305.
9. Kummer JA, Kamp AM, Citarella F, Horrevoets AJ, Hack CE. Expres-
sion of human recombinant granzyme A zymogen and its activa-
11 CONCLUDING REMARKS
tion by the cysteine proteinase cathepsin C. J Biol Chem. 1996;271:
9281-9286.
Cellular cytotoxicity is a very important immune-effector mechanism.
10. Raja SM, Wang B, Dantuluri M, et al. Cytotoxic cell granule-
However, this deadly mechanism requires many regulatory steps to mediated apoptosis. Characterization of the macromolecular
ensure that cytotoxic cells do not damage themselves. Additionally, complex of granzyme B with serglycin. J Biol Chem. 2002;277:
cellular cytotoxicity has to be precisely targeted in order to prevent 49523-49530.
damage to bystanders. This creates a mechanism with many differ- 11. Isaaz S, Baetz K, Olsen K, Podack E, Griffiths GM. Serial killing by cyto-
ent layers of regulation and may explain why defects in the cyto- toxic T lymphocytes: t cell receptor triggers degranulation, re-filling
of the lytic granules and secretion of lytic proteins via a non-granule
toxic machinery not only cause defective immune responses against
pathway. Eur J Immunol. 1995;25:1071-1079.
pathogens but also result in immune-mediated pathologies themselves
12. Voskoboinik I, Smyth MJ, Trapani JA. Perforin-mediated target-
(reviewed in Ref. 163). In the future, it will be important to investi- cell death and immune homeostasis. Nat Rev Immunol. 2006;6:
gate which cytotoxic mechanism is used by specific subpopulations of 940-952.
cytotoxic cells and under which circumstances. Additionally, species- 13. Voskoboinik I, Thia MC, Fletcher J, et al. Calcium-dependent plasma
specific differences will have to be addressed in order to solve some membrane binding and cell lysis by perforin are mediated through its
controversies in the field. C2 domain: a critical role for aspartate residues 429, 435, 483, and
485 but not 491. J Biol Chem. 2005;280:8426-8434.

AUTHORSHIP 14. Praper T, Besenicar MP, Istinic H, et al. Human perforin permeabiliz-
ing activity, but not binding to lipid membranes, is affected by pH. Mol
I.P. and C.W. jointly prepared and wrote the manuscript. Immunol. 2010;47:2492-2504.
15. Uellner R, Zvelebil MJ, Hopkins J, et al. Perforin is activated by a pro-
teolytic cleavage during biosynthesis which reveals a phospholipid-
ACKNOWLEDGEMENTS
binding C2 domain. EMBO J. 1997;16:7287-7296.
We thank all members of our laboratory for fruitful discussions and 16. Brennan AJ, Chia J, Browne KA, et al. Protection from endogenous
input. perforin: glycans and the C terminus regulate exocytic trafficking in
cytotoxic lymphocytes. Immunity. 2011;34:879-892.
17. Law RH, Lukoyanova N, Voskoboinik I, et al. The structural basis
DISCLOSURES for membrane binding and pore formation by lymphocyte perforin.
Nature. 2010;468:447-451.
The authors declare no conflicts of interest.
18. Brennan AJ, Law RHP, Conroy PJ, et al. Perforin proteosta-
sis is regulated through its C2 domain: supra-physiological cell
ORCID death mediated by T431D-perforin. Cell Death Differ. 2018;25:
1517-1529.
Carsten Watzl https://orcid.org/0000-0001-5195-0995
19. House IG, House CM, Brennan AJ, et al. Regulation of perforin acti-
vation and pre-synaptic toxicity through C-terminal glycosylation.
REFERENCES EMBO Rep. 2017;18:1775-1785.
20. Dupuis M, Schaerer E, Krause KH, Tschopp J. The calcium-binding
1. Watzl C. How to trigger a killer: modulation of natural killer cell reac-
protein calreticulin is a major constituent of lytic granules in cytolytic
tivity on many levels. Adv Immunol. 2014;124:137-170.
T lymphocytes. J Exp Med. 1993;177:1-7.
2. Yang Y. Cancer immunotherapy: harnessing the immune system to
21. Fraser SA, Michalak M, Welch WH, Hudig D. Calreticulin, a compo-
battle cancer. J Clin Invest. 2015;125:3335-3337.
nent of the endoplasmic reticulum and of cytotoxic lymphocyte gran-
3. Ewen CL, Kane KP, Bleackley RC. A quarter century of granzymes. ules, regulates perforin-mediated lysis in the hemolytic model sys-
Cell Death Differ. 2012;19:28-35. tem. Biochem Cell Biol. 1998;76:881-887.
4. Jenne DE, Tschopp J. Granzymes, a family of serine proteases 22. Andrin C, Pinkoski MJ, Burns K, et al. Interaction between a Ca2+-
released from granules of cytolytic T lymphocytes upon T cell recep- binding protein calreticulin and perforin, a component of the cyto-
tor stimulation. Immunol Rev. 1988;103:53-71. toxic T-cell granules. Biochemistry. 1998;37:10386-10394.
5. Griffiths GM, Isaaz S. Granzymes A and B are targeted to the lytic 23. Fraser SA, Karimi R, Michalak M, Hudig D. Perforin lytic activity is
granules of lymphocytes by the mannose-6-phosphate receptor. J Cell controlled by calreticulin. J Immunol. 2000;164:4150-4155.
Biol. 1993;120:885-896.
24. Krzewski K, Gil-Krzewska A, Nguyen V, Peruzzi G, Coligan
6. Einstein R, Gabel CA. Cell- and ligand-specific dephosphorylation of JE. LAMP1/CD107a is required for efficient perforin deliv-
acid hydrolases: evidence that the mannose 6-phosphatase is con- ery to lytic granules and NK-cell cytotoxicity. Blood. 2013;121:
trolled by compartmentalization. J Cell Biol. 1991;112:81-94. 4672-4683.
1326 PRAGER AND WATZL

25. Konjar S, Sutton VR, Hoves S, et al. Human and mouse perforin are 46. Martinvalet D, Dykxhoorn DM, Ferrini R, Lieberman J. Granzyme
processed in part through cleavage by the lysosomal cysteine pro- A cleaves a mitochondrial complex I protein to initiate caspase-
teinase cathepsin L. Immunology. 2010;131:257-267. independent cell death. Cell. 2008;133:681-692.
26. Lesteberg K, Orange J, Makedonas G. Recycling endosomes in 47. Zhu P, Zhang D, Chowdhury D, et al. Granzyme A, which causes
human cytotoxic T lymphocytes constitute an auxiliary intracellu- single-stranded DNA damage, targets the double-strand break repair
lar trafficking pathway for newly synthesized perforin. Immunol Res. protein Ku70. EMBO Rep. 2006;7:431-437.
2017;65:1031-1045. 48. Fan Z, Beresford PJ, Oh DY, Zhang D, Lieberman J. Tumor suppressor
27. Chowdhury D, Lieberman J. Death by a thousand cuts: granzyme NM23-H1 is a granzyme A-activated DNase during CTL-mediated
pathways of programmed cell death. Annu Rev Immunol. apoptosis, and the nucleosome assembly protein SET is its inhibitor.
2008;26:389-420. Cell. 2003;112:659-672.
28. Smyth MJ, Kelly JM, Sutton VR, et al. Unlocking the secrets of cyto- 49. Fan Z, Beresford PJ, Zhang D, Lieberman J. HMG2 interacts with the
toxic granule proteins. J Leukoc Biol. 2001;70:18-29. nucleosome assembly protein SET and is a target of the cytotoxic T-
29. Cullen SP, Martin SJ. Mechanisms of granule-dependent killing. Cell lymphocyte protease granzyme A. Mol Cell Biol. 2002;22:2810-2820.
Death Differ. 2008;15:251-262. 50. Zhang D, Beresford PJ, Greenberg AH, Lieberman J. Granzymes A
30. Trapani JA. Granzymes: a family of lymphocyte granule serine pro- and B directly cleave lamins and disrupt the nuclear lamina dur-
teases. Genome Biol. 2001;2:REVIEWS3014. ing granule-mediated cytolysis. Proc Natl Acad Sci U S A. 2001;98:
5746-5751.
31. Lieberman J. The ABCs of granule-mediated cytotoxicity: new
weapons in the arsenal. Nat Rev Immunol. 2003;3:361-370. 51. Zhang D, Pasternack MS, Beresford PJ, Wagner L, Greenberg AH,
Lieberman J. Induction of rapid histone degradation by the cyto-
32. Fischer U, Janicke RU, K. Schulze-Osthoff. Many cuts to ruin: toxic T lymphocyte protease Granzyme A. J Biol Chem. 2001;276:
a comprehensive update of caspase substrates. Cell Death Differ. 3683-3690.
2003;10:76-100.
52. Beresford PJ, Zhang D, Oh DY, et al. Granzyme A activates
33. Sutton VR, Davis JE, Cancilla M, et al. Initiation of apoptosis by an endoplasmic reticulum-associated caspase-independent nucle-
granzyme B requires direct cleavageof bid, but not direct granzyme ase to induce single-stranded DNA nicks. J Biol Chem. 2001;276:
B-mediated caspase activation. J Exp Med. 2000;192:1403-1414. 43285-43293.
34. Heibein JA, Goping IS, Barry M, et al. Granzyme B-mediated 53. Zhu P, Martinvalet D, Chowdhury D, Zhang D, Schlesinger A, Lieber-
cytochrome c release is regulated by the Bcl-2 family members bid man J. The cytotoxic T lymphocyte protease granzyme A cleaves
and Bax. J Exp Med. 2000;192:1391-1402. and inactivates poly(adenosine 5’-diphosphate-ribose) polymerase-
35. Pinkoski MJ, Waterhouse NJ, Heibein JA, et al. Granzyme B-mediated 1. Blood. 2009;114:1205-1216.
apoptosis proceeds predominantly through a Bcl-2-inhibitable mito- 54. Susanto O, Stewart SE, Voskoboinik I, et al. Mouse granzyme A
chondrial pathway. J Biol Chem. 2001;276:12060-12067. induces a novel death with writhing morphology that is mechanisti-
36. Sutton VR, Vaux DL, Trapani JA. Bcl-2 prevents apoptosis induced by cally distinct from granzyme B-induced apoptosis. Cell Death Differ.
perforin and granzyme B, but not that mediated by whole cytotoxic 2013;20:1183-1193.
lymphocytes. J Immunol. 1997;158:5783-5790. 55. Metkar SS, Menaa C, Pardo J, et al. Human and mouse granzyme
37. Cullen SP, Adrain C, Luthi AU, Duriez PJ, Martin SJ. Human and A induce a proinflammatory cytokine response. Immunity.
murine granzyme B exhibit divergent substrate preferences. J Cell 2008;29:720-733.
Biol. 2007;176:435-444. 56. Wilharm E, Parry MA, Friebel R, et al. Generation of catalytically
38. Casciola-Rosen L, Garcia-Calvo M, Bull HG, et al. Mouse and human active granzyme K from Escherichia coli inclusion bodies and iden-
granzyme B have distinct tetrapeptide specificities and abilities to tification of efficient granzyme K inhibitors in human plasma. J Biol
recruit the bid pathway. J Biol Chem. 2007;282:4545-4552. Chem. 1999;274:27331-27337.
39. Kaiserman D, Bird CH, Sun J, et al. The major human and mouse 57. Zhao T, Zhang H, Guo Y, et al. Granzyme K cleaves the nucleosome
granzymes are structurally and functionally divergent. J Cell Biol. assembly protein SET to induce single-stranded DNA nicks of target
2006;175:619-630. cells. Cell Death Differ. 2007;14:489-499.
40. de Poot SA, Westgeest M, Hostetter DR, et al. Human and mouse 58. Guo Y, Chen J, Zhao T, Fan Z. Granzyme K degrades the redox/DNA
granzyme M display divergent and species-specific substrate speci- repair enzyme Ape1 to trigger oxidative stress of target cells leading
ficities. Biochem J. 2011;437:431-442. to cytotoxicity. Mol Immunol. 2008;45:2225-2235.
41. Susanto O, Trapani JA, Brasacchio D. Controversies in granzyme biol- 59. MacDonald G, Shi L, Vande Velde C, Lieberman J, Greenberg AH.
ogy. Tissue Antigens. 2012;80:477-487. Mitochondria-dependent and -independent regulation of Granzyme
42. Beresford PJ, Xia Z, Greenberg AH, Lieberman J. Granzyme A load- B-induced apoptosis. J Exp Med. 1999;189:131-144.
ing induces rapid cytolysis and a novel form of DNA damage indepen- 60. Shresta S, Goda P, Wesselschmidt R, Ley TJ. Residual cytotoxicity and
dently of caspase activation. Immunity. 1999;10:585-594. granzyme K expression in granzyme A-deficient cytotoxic lympho-
43. Martinvalet D, Zhu P, Lieberman J. Granzyme A induces caspase- cytes. J Biol Chem. 1997;272:20236-20244.
independent mitochondrial damage, a required first step for apopto- 61. Bovenschen N, Quadir R, van den Berg AL, et al. Granzyme K displays
sis. Immunity. 2005;22:355-370. highly restricted substrate specificity that only partially overlaps with
44. Shi L, Kraut RP, Aebersold R, Greenberg AH. A natural killer cell gran- granzyme A. J Biol Chem. 2009;284:3504-3512.
ule protein that induces DNA fragmentation and apoptosis. J Exp Med. 62. Joeckel LT, Allison CC, Pellegrini M, Bird CH, Bird PI. Granzyme
1992;175:553-566. K-deficient mice show no evidence of impaired antiviral immunity.
45. Shi L, Kam CM, Powers JC, Aebersold R, Greenberg AH. Purification Immunol Cell Biol. 2017;95:676-683.
of three cytotoxic lymphocyte granule serine proteases that induce 63. Smyth MJ, O’Connor MD, Kelly JM, Ganesvaran P, Thia KY, Trapani
apoptosis through distinct substrate and target cell interactions. J Exp JA. Expression of recombinant human Met-ase-1: a NK cell-specific
Med. 1992;176:1521-1529. granzyme. Biochem Biophys Res Commun. 1995;217:675-683.
PRAGER AND WATZL 1327

64. Kelly JM, WaterhousGranzyme NJ, et al. Granzyme M medi- 83. Anderson DH, Sawaya MR, Cascio D, et al. Granulysin crystal
ates a novel form of perforin-dependent cell death. J Biol Chem. structure and a structure-derived lytic mechanism. J Mol Biol.
2004;279:22236-22242. 2003;325:355-365.
65. Cullen SP, Afonina IS, Donadini R, et al. Nucleophosmin is cleaved 84. Hoffmann SC, Cohnen A, Ludwig T, Watzl C. 2B4 engagement medi-
and inactivated by the cytotoxic granule protease granzyme M during ates rapid LFA-1 and actin-dependent NK cell adhesion to tumor
natural killer cell-mediated killing. J Biol Chem. 2009;284:5137-5147. cells as measured by single cell force spectroscopy. J Immunol.
66. Mahrus S, Kisiel W, Craik CS. Granzyme M is a regulatory protease 2011;186:2757-2764.
that inactivates proteinase inhibitor 9, an endogenous inhibitor of 85. Bryceson YT, Ljunggren HG, Long EO. Minimal requirement for
granzyme B. J Biol Chem. 2004;279:54275-54282. induction of natural cytotoxicity and intersection of activation signals
67. de Koning PJ, Kummer JA, de Poot SA, et al. Intracellular serine pro- by inhibitory receptors. Blood. 2009;114:2657-2666.
tease inhibitor SERPINB4 inhibits granzyme M-induced cell death. 86. Mace EM, Dongre P, Hsu HT, et al. Cell biological steps and
PLoS One. 2011;6:e22645. checkpoints in accessing NK cell cytotoxicity. Immunol Cell Biol.
68. Sayers TJ, Brooks AD, Ward JM, et al. The restricted expression of 2014;92:245-255.
granzyme M in human lymphocytes. J Immunol. 2001;166:765-771. 87. Mentlik AN, Sanborn KB, Holzbaur EL, Orange JS. Rapid lytic gran-
69. Edwards KM, Kam CM, Powers JC, Trapani JA. The human cytotoxic ule convergence to the MTOC in natural killer cells is dependent
T cell granule serine protease granzyme H has chymotrypsin-like on dynein but not cytolytic commitment. Mol Biol Cell. 2010;21:
(chymase) activity and is taken up into cytoplasmic vesicles 2241-2256.
reminiscent of granzyme B-containing endosomes. J Biol Chem. 88. James AM, Hsu HT, Dongre P, et al. Rapid activation receptor-
1999;274:30468-30473. or IL-2-induced lytic granule convergence in human natural killer
70. Sedelies KA, Sayers TJ, Edwards KM, et al. Discordant regulation of cells requires Src, but not downstream signaling. Blood. 2013;121:
granzyme H and granzyme B expression in human lymphocytes. J Biol 2627-2637.
Chem. 2004;279:26581-26587. 89. de Saint Basile G, Menasche G, Fischer A. Molecular mechanisms
71. Andrade F, Fellows E, Jenne DE, Rosen A, Young CS. Granzyme H of biogenesis and exocytosis of cytotoxic granules. Nat Rev Immunol.
destroys the function of critical adenoviral proteins required for 2010;10:568-579.
viral DNA replication and granzyme B inhibition. EMBO J. 2007;26: 90. Liu D, Bryceson YT, Meckel T, Vasiliver-Shamis G, Dustin ML, Long
2148-2157. EO. Integrin-dependent organization and bidirectional vesicular traf-
72. Romero V, Fellows E, Jenne DE, Andrade F. Cleavage of La pro- fic at cytotoxic immune synapses. Immunity. 2009;31:99-109.
tein by granzyme H induces cytoplasmic translocation and interferes 91. Simone CB, Henkart P. Permeability changes induced in erthrocyte
with La-mediated HCV-IRES translational activity. Cell Death Differ. ghost targets by antibody-dependent cytotoxic effector cells: evi-
2009;16:340-348. dence for membrane pores. J Immunol. 1980;124:954-963.
73. Fellows E, Gil-Parrado S, Jenne DE, Kurschus FC. Natural killer 92. Leung C, Hodel AW, Brennan AJ, et al. Real-time visualization of per-
cell-derived human granzyme H induces an alternative, caspase- forin nanopore assembly. Nat Nanotechnol. 2017;12:467-473.
independent cell-death program. Blood. 2007;110:544-552. 93. Kurschus FC, Fellows E, Stegmann E, Jenne DE. Granzyme B
74. Jongstra J, Schall TJ, Dyer BJ, et al. The isolation and sequence delivery via perforin is restricted by size, but not by heparan
of a novel gene from a human functional T cell line. J Exp Med. sulfate-dependent endocytosis. Proc Natl Acad Sci U S A. 2008;105:
1987;165:601-614. 13799-13804.
75. Krensky AM, Clayberger C. Biology and clinical relevance of gran- 94. Lopez JA, Susanto O, Jenkins MR, et al. Perforin forms transient
ulysin. Tissue Antigens. 2009;73:193-198. pores on the target cell plasma membrane to facilitate rapid access
76. Pena SV, Hanson DA, Carr BA, Goralski TJ, Krensky AM. Process- of granzymes during killer cell attack. Blood. 2013;121:2659-2668.
ing, subcellular localization, and function of 519 (granulysin), a human 95. Lopez JA, Jenkins MR, Rudd-Schmidt JA, et al. Rapid and unidi-
late T cell activation molecule with homology to small, lytic, granule rectional perforin pore delivery at the cytotoxic immune synapse.
proteins. J Immunol. 1997;158:2680-2688. J Immunol. 2013;191:2328-2334.
77. Hanson DA, Kaspar AA, Poulain FR, Krensky AM. Biosynthesis 96. Bird CH, Sun J, Ung K, et al. Cationic sites on granzyme B contribute
of granulysin, a novel cytolytic molecule. Mol Immunol. 1999;36: to cytotoxicity by promoting its uptake into target cells. Mol Cell Biol.
413-422. 2005;25:7854-7867.
78. Stenger S, Hanson DA, Teitelbaum R, et al. An antimicrobial activity of 97. Froelich CJ, Orth K, Turbov J, et al. New paradigm for lympho-
cytolytic T cells mediated by granulysin. Science. 1998;282:121-125. cyte granule-mediated cytotoxicity. Target cells bind and inter-
79. Latinovic-Golic S, Walch M, Sundstrom H, Dumrese C, Groscurth nalize granzyme B, but an endosomolytic agent is necessary for
P, Ziegler U. Expression, processing and transcriptional regulation cytosolic delivery and subsequent apoptosis. J Biol Chem. 1996;271:
of granulysin in short-term activated human lymphocytes. BMC 29073-29079.
Immunol. 2007;8:9. 98. Keefe D, Shi L, Feske S, et al. Perforin triggers a plasma membrane-
80. Al-Wasaby S, de Miguel D, Aporta A, et al. In vivo potential of repair response that facilitates CTL induction of apoptosis. Immunity.
recombinant granulysin against human tumors. Oncoimmunology. 2005;23:249-262.
2015;4:e1036213. 99. Thiery J, Keefe D, Boulant S, et al. Perforin pores in the endosomal
81. Lu CC, Wu TS, Hsu YJ, et al. NK cells kill mycobacteria directly by membrane trigger the release of endocytosed granzyme B into the
releasing perforin and granulysin. J Leukoc Biol. 2014;96:1119-1129. cytosol of target cells. Nat Immunol. 2011;12:770-777.

82. Kaspar AA, Okada S, Kumar J, et al. A distinct pathway of 100. Blakely A, Gorman K, Ostergaard H, et al. Resistance of cloned
cell-mediated apoptosis initiated by granulysin. J Immunol. cytotoxic T lymphocytes to cell-mediated cytotoxicity. J Exp Med.
2001;167:350-356. 1987;166:1070-1083.
1328 PRAGER AND WATZL

101. Bird CH, Sutton VR, Sun J, et al. Selective regulation of apopto- 120. Lee J, Dieckmann NMG, Edgar JR, Griffiths GM, Siegel RM. Fas Lig-
sis: the cytotoxic lymphocyte serpin proteinase inhibitor 9 protects and localizes to intraluminal vesicles within NK cell cytolytic granules
against granzyme B-mediated apoptosis without perturbing the Fas and is enriched at the immune synapse. Immun Inflamm Dis. 2018;6:
cell death pathway. Mol Cell Biol. 1998;18:6387-6398. 312-321.
102. Sun J, Whisstock JC, Harriott P, et al. Importance of the P4’ residue 121. Lettau M, Armbrust F, Dohmen K, et al. Mechanistic peculiarities
in human granzyme B inhibitors and substrates revealed by scanning of activation-induced mobilization of cytotoxic effector proteins in
mutagenesis of the proteinase inhibitor 9 reactive center loop. J Biol human T cells. Int Immunol. 2018;30:215-228.
Chem. 2001;276:15177-15184. 122. Liu D, Martina JA, Wu XS, Hammer JA, 3rd, Long EO. Two modes
103. Sun J, Bird CH, Sutton V, et al. A cytosolic granzyme B inhibitor of lytic granule fusion during degranulation by natural killer cells.
related to the viral apoptotic regulator cytokine response modi- Immunol Cell Biol. 2011;89:728-738.
fier A is present in cytotoxic lymphocytes. J Biol Chem. 1996;271: 123. Tanaka M, Suda T, Takahashi T, Nagata S. Expression of the functional
27802-27809. soluble form of human fas ligand in activated lymphocytes. EMBO J.
104. Huntington JA, Read RJ, Carrell RW. Structure of a serpin-protease 1995;14:1129-1135.
complex shows inhibition by deformation. Nature. 2000;407: 124. Kayagaki N, Kawasaki A, Ebata T, et al. Metalloproteinase-mediated
923-926. release of human Fas ligand. J Exp Med. 1995;182:1777-1783.
105. Cohnen A, Chiang SC, Stojanovic A, et al. Surface CD107a/LAMP-1 125. Mariani SM, Matiba B, Baumler C, Krammer PH. Regulation of cell
protects natural killer cells from degranulation-associated damage. surface APO-1/Fas (CD95) ligand expression by metalloproteases.
Blood. 2013;122:1411-1418. Eur J Immunol. 1995;25:2303-2307.
106. Antia R, Schlegel RA, Williamson P. Binding of perforin to mem- 126. Schneider P, Holler N, Bodmer JL, et al. Conversion of membrane-
branes is sensitive to lipid spacing and not headgroup. Immunol Lett. bound Fas(CD95) ligand to its soluble form is associated with down-
1992;32:153-157. regulation of its proapoptotic activity and loss of liver toxicity. J Exp
107. Balaji KN, Schaschke N, Machleidt W, Catalfamo M, Henkart PA. Med. 1998;187:1205-1213.
Surface cathepsin B protects cytotoxic lymphocytes from self- 127. Suda T, Hashimoto H, Tanaka M, Ochi T, Nagata S. Membrane Fas lig-
destruction after degranulation. J Exp Med. 2002;196:493-503. and kills human peripheral blood T lymphocytes, and soluble Fas lig-
108. Khazen R, Muller S, Gaudenzio N, Espinosa E, Puissegur MP, and blocks the killing. J Exp Med. 1997;186:2045-2050.
Valitutti S. Melanoma cell lysosome secretory burst neutralizes 128. Sieg S, Smith D, Kaplan D. Differential activity of soluble versus cel-
the CTL-mediated cytotoxicity at the lytic synapse. Nat Commun. lular Fas ligand: regulation by an accessory molecule. Cell Immunol.
2016;7:10823. 1999;195:89-95.
109. Baran K, Ciccone A, Peters C, et al. Cytotoxic T lymphocytes from 129. Hohlbaum AM, Moe S. A. Marshak-Rothstein. Opposing
cathepsin B-deficient mice survive normally in vitro and in vivo effects of transmembrane and soluble Fas ligand expression
after encountering and killing target cells. J Biol Chem. 2006;281: on inflammation and tumor cell survival. J Exp Med. 2000;191:
30485-30491. 1209-1220.
110. Peter ME, Krammer PH. The CD95(APO-1/Fas) DISC and beyond. 130. Wiley SR, Schooley K, Smolak PJ, et al. Identification and character-
Cell Death Differ. 2003;10:26-35. ization of a new member of the TNF family that induces apoptosis.
111. von Karstedt S, Montinaro A, Walczak H. Exploring the TRAILs Immunity. 1995;3:673-682.
less travelled: tRAIL in cancer biology and therapy. Nat Rev Cancer. 131. Thomas WD, Hersey P. TNF-related apoptosis-inducing ligand
2017;17:352-366. (TRAIL) induces apoptosis in Fas ligand-resistant melanoma cells and
112. Bodmer JL, Schneider P, Tschopp J. The molecular architecture of the mediates CD4 T cell killing of target cells. J Immunol. 1998;161:
TNF superfamily. Trends Biochem Sci. 2002;27:19-26. 2195-2200.
113. Suda T, Okazaki T, Naito Y, et al. Expression of the Fas ligand in cells 132. Fanger NA, Maliszewski CR, Schooley K, Griffith TS. Human den-
of T cell lineage. J Immunol. 1995;154:3806-3813. dritic cells mediate cellular apoptosis via tumor necrosis factor-
114. Blott EJ, Bossi G, Clark R, Zvelebil M, Griffiths GM. Fas ligand is tar- related apoptosis-inducing ligand (TRAIL). J Exp Med. 1999;190:
geted to secretory lysosomes via a proline-rich domain in its cytoplas- 1155-1164.
mic tail. J Cell Sci. 2001;114:2405-2416. 133. Griffith TS, Wiley SR, Kubin MZ, Sedger LM, Maliszewski CR,
115. Zuccato E, Blott EJ, Holt O, et al. Sorting of Fas ligand to secretory Fanger NA. Monocyte-mediated tumoricidal activity via the tumor
lysosomes is regulated by mono-ubiquitylation and phosphorylation. necrosis factor-related cytokine, TRAIL. J Exp Med. 1999;189:
J Cell Sci. 2007;120:191-199. 1343-1354.

116. Bossi G, Griffiths GM. Degranulation plays an essential part in regu- 134. Kayagaki N, Yamaguchi N, Nakayama M, et al. Expression and func-
lating cell surface expression of Fas ligand in T cells and natural killer tion of TNF-related apoptosis-inducing ligand on murine activated
cells. Nat Med. 1999;5:90-96. NK cells. J Immunol. 1999;163:1906-1913.

117. Bryceson YT, March ME, Barber DF, Ljunggren HG, Long EO. 135. Smyth MJ, Cretney E, Takeda K, et al. Tumor necrosis factor-related
Cytolytic granule polarization and degranulation controlled by differ- apoptosis-inducing ligand (TRAIL) contributes to interferon gamma-
ent receptors in resting NK cells. J Exp Med. 2005;202:1001-1012. dependent natural killer cell protection from tumor metastasis. J Exp
Med. 2001;193:661-670.
118. Schmidt H, Gelhaus C, Nebendahl M, et al. Effector granules in human
T lymphocytes: proteomic evidence for two distinct species of cyto- 136. Takeda K, Hayakawa Y, Smyth MJ, et al. Involvement of tumor necro-
toxic effector vesicles. J Proteome Res. 2011;10:1603-1620. sis factor-related apoptosis-inducing ligand in surveillance of tumor
metastasis by liver natural killer cells. Nat Med. 2001;7:94-100.
119. Schmidt H, Gelhaus C, Nebendahl M, et al. 2-D DIGE analyses of
enriched secretory lysosomes reveal heterogeneous profiles of func- 137. Sato K, Hida S, Takayanagi H, et al. Antiviral response by natural killer
tionally relevant proteins in leukemic and activated human NK cells. cells through TRAIL gene induction by IFN-alpha/beta. Eur J Immunol.
Proteomics. 2008;8:2911-2925. 2001;31:3138-3146.
PRAGER AND WATZL 1329

138. Monleon I, Martinez-Lorenzo MJ, Monteagudo L, et al. Differen- 152. Berke G, Sullivan KA, Amos DB. Tumor immunity in vitro: destruc-
tial secretion of Fas ligand- or APO2 ligand/TNF-related apoptosis- tion of a mouse ascites tumor through a cycling pathway. Science.
inducing ligand-carrying microvesicles during activation-induced 1972;177:433-434.
death of human T cells. J Immunol. 2001;167:6736-6744. 153. Zagury D, Noelle Thierness JB, Feldman M, Behe G. Isolation and
139. Chaudhary PM, Eby M, Jasmin A, Bookwalter A, Murray J, Hood L. characterizationof individual functionally reactive cytotoxic T lym-
Death receptor 5, a new member of the TNFR family, and DR4 induce phocytes: conjugation, killing and recycling at the single cell leve. Eur
FADD-dependent apoptosis and activate the NF-kappaB pathway. J Immunol. 1975;5:818-822.
Immunity. 1997;7:821-830. 154. Bhat R, Watzl C. Serial killing of tumor cells by human natu-
140. Walczak H, Degli-Esposti MA, Johnson RS, et al. TRAIL-R2: a ral killer cells–enhancement by therapeutic antibodies. PLoS One.
novel apoptosis-mediating receptor for TRAIL. EMBO J. 1997;16: 2007;2:e326.
5386-5397. 155. Choi PJ, Mitchison TJ. Imaging burst kinetics and spatial coordination
141. Degli-Esposti MA, Smolak PJ, Walczak H, et al. Cloning and character- during serial killing by single natural killer cells. Proc Natl Acad Sci U S
ization of TRAIL-R3, a novel member of the emerging TRAIL receptor A. 2013;110:6488-6493.
family. J Exp Med. 1997;186:1165-1170. 156. Vanherberghen B, Olofsson PE, Forslund E, et al. Classification of
142. Degli-Esposti MA, Dougall WC, Smolak PJ, Waugh JY, Smith CA, human natural killer cells based on migration behavior and cytotoxic
Goodwin RG. The novel receptor TRAIL-R4 induces NF-kappaB and response. Blood. 2013;121:1326-1334.
protects against TRAIL-mediated apoptosis, yet retains an incom- 157. Hsu HT, Mace EM, Carisey AF, et al. NK cells converge lytic gran-
plete death domain. Immunity. 1997;7:813-820. ules to promote cytotoxicity and prevent bystander killing. J Cell Biol.
143. Hassin D, Garber OG, Meiraz A, Schiffenbauer YS, Berke G. Cyto- 2016;215:875-889.
toxic T lymphocyte perforin and Fas ligand working in concert even 158. Bertrand F, Muller S, Roh KH, Laurent C, Dupre L, Valitutti S. An initial
when Fas ligand lytic action is still not detectable. Immunology. and rapid step of lytic granule secretion precedes microtubule orga-
2011;133:190-196. nizing center polarization at the cytotoxic T lymphocyte/target cell
144. Lowin B, Mattman C, Hahne M, Tschopp J. Comparison of Fas(Apo- synapse. Proc Natl Acad Sci U S A. 2013;110:6073-6078.
1/CD95)- and perforin-mediated cytotoxicity in primary T lympho- 159. Netter P, Anft M, Watzl C. Termination of the activating NK cell
cytes. Int Immunol. 1996;8:57-63. immunological synapse is an active and regulated process. J Immunol.
145. Kagi D, Vignaux F, Ledermann B, et al. Fas and perforin path- 2017;199:2528-2535.
ways as major mechanisms of T cell-mediated cytotoxicity. Science. 160. Srpan K, Ambrose A, Karampatzakis A, et al. Shedding of CD16 disas-
1994;265:528-530. sembles the NK cell immune synapse and boosts serial engagement
146. Li J, Figueira SK, Vrazo AC, et al. Real-time detection of CTL function of target cells. J Cell Biol. 2018;217:3267-3283.
reveals distinct patterns of caspase activation mediated by Fas versus 161. Sandusky MM, Messmer B, Watzl C. Regulation of 2B4 (CD244)-
granzyme B. J Immunol. 2014;193:519-528. mediated NK cell activation by ligand-induced receptor modulation.
147. Alter G, Malenfant JM, Altfeld M. CD107a as a functional marker Eur J Immunol. 2006;36:3268-3276.
for the identification of natural killer cell activity. J Immunol Methods. 162. Liesche C, Sauer P, Prager I, et al. Single-fluorescent protein
2004;294:15-22. reporters allow parallel quantification of natural killer cell-mediated
148. Gwalani LA, Orange JS. Single degranulations in NK cells can mediate granzyme and caspase activities in single target cells. Front Immunol.
target cell killing. J Immunol. 2018;200:3231-3243. 2018;9:1840.
149. Algeciras-Schimnich A, Shen L, Barnhart BC, Murmann AE, Burkhardt 163. Meeths M, Chiang SC, Lofstedt A, et al. Pathophysiology and spec-
JK, Peter ME. Molecular ordering of the initial signaling events of trum of diseases caused by defects in lymphocyte cytotoxicity. Exp
CD95. Mol Cell Biol. 2002;22:207-220. Cell Res. 2014;325:10-17.
150. Scaffidi C, Schmitz I, Zha J, Korsmeyer SJ, Krammer PH, Peter ME.
Differential modulation of apoptosis sensitivity in CD95 type I and
type II cells. J Biol Chem. 1999;274:22532-22538.
How to cite this article: Prager I, Watzl C. Mechanisms of
natural killer cell-mediated cellular cytotoxicity. J Leukoc
151. Berke G, Sullivan KA, Amos B. Rejection of ascites tumor allo-
grafts. II. A pathway for cell-mediated tumor destruction in Biol. 2019;105:1319–1329. https://doi.org/10.1002/JLB.
vitro by peritoneal exudate lymphoid cells. J Exp Med. 1972;136: MR0718-269R
1594-1604.

You might also like