Science - USA (2019-02-15)

SCIENCE sciencemag.org

GRAPHIC: V. ALTOUNIAN/

SCIENCE

chemical experiments ( 9 ) and thus present

an excellent snapshot of how a substrate can

be bound in the active site just before cleav-

age occurs. It is possible, though, that other

substrates or native substrates without the

forced cysteine cross-link show somewhat

different structures.

A comparison of the substrate-bound

structures to the known structure of g-secre-

tase without substrate and to the structure

of the substrates without protease yields ex-

citing and partly unexpected insights into

the molecular steps that are likely to have

happened before the substrate arrived in the

active site. Imagine nine transmembrane he-

lices of PS that are linked by loops and form

a well-ordered, compact structure within

the membrane. Suddenly, they need to make

space to integrate the substrate TMD, which

requires substantial structural rearrange-

ments in both g-secretase and its substrate.

This is in line with the induced-fit model of

enzymology and may be one of the reasons

why g-secretase has slow enzymatic kinet-

ics ( 12 ). Specifically, to fit below the nicas-

trin lid, the substrate ducks its head, which

corresponds to a turn of the short amino-

terminal substrate extracellular domain to-

ward the membrane (see the figure). Then

TMD2 and TMD6 of PS are likely to open

up because they are an obvious lateral entry

point of the substrate TMD into g-secretase,

although other entry points also appear to

be possible ( 9 , 13 ). Subsequently the still

“ducked head” may allow the substrate to

pass below the loop connecting TMD2 with

TMD1 of PS. Next, the amino terminus of the

substrate TMD seems to tilt outward, which

brings the scissile peptide bond at the car-

boxyl terminus, that is, the «-cleavage site,

in close proximity to the two catalytic as-

partate residues in PS. This is accompanied

by the arguably most remarkable structural

change seen in both structures, namely the

partial unfolding of the substrate TMD helix

around the «-cleavage site and its formation

of a b-pleated sheet. This requires substan-

tial helix flexibility ( 14 ) and makes the scis-

sile peptide bond accessible to cleavage. At

present, the exact order of the above steps

is unknown, but mutational studies suggest

that formation of the b-pleated sheet is key

to correctly position the substrate TMD in

the active site ( 2 , 3 ).

The new insights into substrate binding

by g-secretase have major implications be-

yond understanding the molecular opera-

tion mode of g-secretase. Now, it is possible

to better understand the molecular basis of

dominantly inherited forms of AD, which

are caused by point mutations at more than

100 different amino acid residues in PS1 or

PS2 and partly around the g-secretase cleav-

age site in APP ( 15 ). Mapping of the muta-

tions onto the new structures suggests that

many of them directly affect the interaction

between APP and PS1, resulting in altered

cleavage efficiency and higher levels of the

pathogenic, AD-linked APP cleavage product

Ab42 ( 2 , 9 ). Final proof will come from future

cryo-EM structures of mutated APP or PS1.

Interestingly, g-secretase subunits other than

PS did not undergo major structural rear-

rangements upon substrate binding, suggest-

ing that this is the reason why AD-causing

inherited mutations have not been found in

those subunits.

Another important implication is rational

drug development of GSIs that preferentially

block g cleavage of APP over the cleavage of

other substrates. This could now become pos-

sible and may revive the use of g-secretase as

an AD drug target. Similarly, Notch-specific

GSIs may reduce side ef-

fects in cancer treatment.

Yet, the structures with

Notch and APP are more sim-

ilar than different, making

rational drug development

a longer-term goal. Poten-

tially, the exosites, where

Notch and APP initially bind

to g-secretase ( 9 ), are more

different from each other and

thus may offer additional op-

portunities for developing

substrate-specific GSIs.

A future challenge is to

understand how substrates

with substantially longer

amino termini than the in-

vestigated APP and Notch

fragments—such as B cell

maturation antigen ( 6 ), the

AD-relevant C99 fragment

of APP, or artificial Notch

cleavage constructs with

long ectodomains ( 8 )—bind

to g-secretase and pass be-

low the loop in PS. Another

major task is to solve the

structure of additional substrate-bound in-

tramembrane proteases, such as the rhom-

boids [which are distant PS homologs and

members of the signal peptide peptidase-

like (SPPL) family] and the site-2 proteases

(S2Ps). Together, these structures will un-

cover the mechanistic secrets of how targeted

proteolysis within the lipid bilayer controls

cell signaling in health and disease. j

REFERENCES

1. T. E. Golde et al., Biochim. Biophys. Acta 1828 , 2898
   (2013).


2. R. Zhou et al., Science 363 , eaaw0930 (2019).


3. G. Ya n g et al., Nature 565 , 192 (2019).


4. D. Edbauer et al., Nat. Cell Biol. 5 , 486 (2003).


5. M. S. Wolfe et al., Nature 398 , 513 (1999).


6. S. A. Laurent et al., Nat. Commun. 6 , 7333 (2015).


7. S. F. Lichtenthaler et al., EMBO J. 37 , e99456 (2018).


8. D. M. Bolduc et al., Proc. Natl. Acad. Sci. U.S.A. 113 , E509
   (2016).


9. A. Fukumori, H. Steiner, EMBO J. 35 , 1628 (2016).


10. M. Takami et al., J. Neurosci. 29 , 13042 (2009).


11. X. C. Bai et al., Nature 525 , 212 (2015).


12. F. Kamp et al., Biophys. J. 108 , 1229 (2015).


13. M. Hitzenberger et al., ACS Chem. Neurosci. 10.1021/
    acschemneuro.8b00725 (2019).


14. D. Langosch, H. Steiner, Biol. Chem. 398 , 441 (2017).


15. H. Steiner et al., Cell Stress 2 , 292 (2018).
    10.1126/science.aaw5547

PS APH-1 PEN-2 Nicastrin Substrate

1 A substrate’s long ectodomain

is cleaved by diferent proteases.

2 Substrate transfer from the exosites to the

active site requires the substrate to enter PS,

likely between transmembrane domains 2 and 6.

3 The substrate’s transmembrane helix tilts

and partially unfolds at its C terminus, where it

forms a b-pleated sheet, required for substrate

cleavage at the « site.

Cytosol

« site

(^3276)
3
2
N
C
6
6a
7
Active
sites
The short remaining
ectodomain moves down.
The substrate’s N-terminal
stub moves below the loop.
b 1 b 2
15 FEBRUARY 2019 • VOL 363 ISSUE 6428 691
Substrate binding by g-secretase is a multistep process
The four g-secretase subunits PS, PEN-2, APH-1, and nicastrin form a horseshoe-like structure in the
membrane, with the nicastrin ectodomain forming a lid on top of the horseshoe structure. Insights from
Zhou et al. and Yang et al. clarify the mechanism of substrate binding by g-secretase.
Published by AAAS
on February 14, 2019^
http://science.sciencemag.org/
Downloaded from