within as-yet-unreleased granules (fig. S2C).
To determine whether some of the membra-
nous material within the intercellular space
might also have been derived from the target

cell, we examined the surface topology of the

postsynaptic target cell. We noted multiple

tubular and bud-like protrusions of the target

cell membrane that extended into the synaptic

space; thus, at least some of the membrane

structures observed were still in continuity

with the target cell (Fig. 2C and movie S6, TS

0:58 to 1:11). ESCRT proteins have been shown

SCIENCEscience.org 22 APRIL 2022¥VOL 376 ISSUE 6591 381

FSC

SSC

VPS4a

E228Q

VPS4a WT

0 ng/mL Perforin 160 ng/mL Perforin 1280 ng/mL Perforin

E

APC

Cell Count

0 ng/mL [GZMB]

+160 ng/mL [PRF]

10,000 ng/mL [GZMB]

+160 ng/mL [PRF]

*Y axis normalized to mode

VPS4a WT

VPS4a E228Q

G

(^001234)
200
400
600
800
Time (hours)

Dead cells/mm

2
N4
A
F
C
TCR:pMHC affinity

ns
Chmp4b
actin
D
TCR:pMHC affinity

• 
  
  ns
  VPS4a
  (anti-HA)
  actin
  Control Chmp4b KO
  T 0h
  T 4h
  B
  Scale 150 um
  0.6 ± 2.4
  173.0 ± 27.7
  10.1 ± 3.4
  573.4± 46.2
  Images from T4 peptide condition
  TCR:pMHC affinity

  ---

  
  

  ns ns ns **
  ns ns ns
  
  H
  (^001234)
  200
  400
  600
  800
  Time (hours)
  T4

  
  

  Dead cells/mm

  
  

  2
  (^001234)
  200
  400
  600
  800
  Time (hours)
  Q4H7

  
  

  Dead cells/mm

  
  

  2
  (^001234)
  200
  400
  600
  800
  Time (hours)
  G4

  
  

  Dead cells/mm

  
  

  2
  Control
  Chmp4b KO
  (^001234)
  200
  400
  600
  800
  Time (hours)
  unpulsed

  
  

  Dead cells/mm

  
  

  2
  37
  50
  37
  50
  50
  37
  ControlChmp4b KO VP
  S4
  WT
  VPS4 E228Q
  0 80 160 320 640 1280
  100
  80
  60
  40
  20
  0
  Perforin viability titration
  [mPerforin]
  ng/mL/10^6 cells
  % Live Cells
  % AnnexinV-APC+ counts
  100
  80
  60
  40
  20
  0
  5,000 GZMB0 PRF^0 2,500 5,000 10,000
  [mGZMB]
  ng/mL/10^6 cells
  VPS4a WT
  VPS4a E228Q
  VPS4a WT
  VPS4a E228Q
  Fig. 4. ESCRT inhibition enhances susceptibility of cancer cells to CTL
  killing and recombinant lytic proteins.(A) Representative time-lapse data of
  killing of peptide-pulsed Chmp4b knockout (KO) or control B16-F10 cells by OT-I
  CTLs. Affinity of the pulsed peptide to OT-I TCR decreases from left to right.
  Error bars indicate SDs. (B) Images extracted from T4 medium-affinity peptide
  condition show software-detected caspase 3/7+ events in control and Chmp4b
  KO conditions. (CandD) Data representing the 4-hour time point of assays
  measuring OT-I T cells killing either Chmp4b KO (C) or VPS4 dominant-negative
  (D) target cells with matched controls. Error bars indicate SDs of data. Data are
  representative of at least three independent experimental replicates. pMHC,
  peptide-MHC; HA, hemagglutinin. (EandF) Determination of sublytic dose of Prf.
  B16-F10 cells expressing VPS4a (WT or E228Q) were exposed to increasing
  concentrations of Prf. Cell viability was determined by morphological gating (E).
  FSC, forward scatter; SSC, side scatter. (GandH) B16-F10 cells expressing
  VPS4a (WT or E228Q) were exposed to a sublytic dose of Prf in combination with
  increasing concentrations of recombinant GZMB (rGZMB). Cell death was
  determined by Annexin VÐallophycocyanin (APC) staining (G). Controls include a
  condition with no perforin and 5000 ng/ml rGZMB and sublytic perforin with no
  rGZMB. Graphs in (F) and (H) represent the means of three experiments, and
  error bars indicate SDs. Statistical significance was determined by multiple
  unpairedttests with alpha = 0.05. ns, not significant; P< 0.05; P< 0.01;
  P< 0.001.
  RESEARCH | RESEARCH ARTICLES