HeLa cells were labeled using growth media containing heavy or light forms of arginine and lysine. (A) Cells grown with the light label were subjected to the γ-secretase inhibitor DAPT, which leads to accumulation of the APP CTF compared to the heavy-labeled control cells (lanes 1 and 2). Equal numbers of cells from the Light+DAPT and Heavy+DMSO conditions were combined and fractionated into cytosolic and membrane proteins (lanes 3 and 4).
(B) One hundred micrograms of the combined Light+DAPT and Heavy+DMSO membrane fraction were separated by SDS-PAGE. The gel was divided into ten horizontal slices and each subjected to trypsinization and LC-MS/MS. In quantitative comparisons, substrates would be expected to have equal relative abundance of full-length proteins (upper spectrum) of the Light+DAPT and Heavy+DMSO conditions; whereas truncated fragments normally cleaved by γ-secretase (CTFs) would show an increased abundance of light-labeled peptides in the presence of DAPT (lower spectrum).
(C) Portion of the APP primary sequence, with a tryptic peptide identified from the higher–molecular weight (MW) portion of the gel and unchanged in relative abundance indicated in blue. Red indicates tryptic peptides found in the lower-MW portion of the gel that were more abundant in the Light+DAPT condition. The transmembrane sequence is highlighted in yellow, and arrowheads indicate the β-, α-, γ-40, and γ-42 secretase cleavage sites (from left to right). APP-770 amino acid numbering is indicated on the left.
(D) List of all substrates identified through the proteomics screen, with the number of higher-MW, equal-abundance peptides in the left column and the number of lower MW peptides that accumulate from DAPT treatment in the right column. Note that all unique HLA peptides identified were combined in the table due to high sequence redundancy among the different HLA isoforms.

(A) Fibroblasts derived from Presenilin1/2 null mice were transduced to stably express a control empty vector, a catalytically inactive PS1-D257A construct, or the wild-type PS1 construct. PS1-D257A and wild-type PS1 enabled the formation of the γ-secretase complex, as indicated by the maturation of nicastrin (mNct) from incompletely glycosylated immature nicastrin (iNct, top panel). The PS1-NTF and a small amount of PS1-holoprotein were observed with wild-type PS1 expression, as expected, but cells expressing PS1-D257A were unable to convert PS1-holo into PS1-NTF (middle panel). Endogenous APP CTFs (bottom panel) indicate that proteolytically active γ-secretase was present only in the wild-type PS1 condition.
(B–E) In empty vector, PS1-D257A, and wild-type PS1 cells, FLAG-tagged candidate substrates were stably expressed to validate their processing by γ-secretase: (B) vasorin (Vasn), (C) dystroglycan (DG), (D) Delta/Notch-like EGF-related receptor (DNER), and (E) natriuretic peptide receptor-C (NPR-C). Two type I transmembrane proteins found in the screen not to be processed by γ-secretase—(F) integrin β-1 (Itgβ1) and (G) natriuretic peptide receptor-A (NPR-A)—were similarly analyzed and validated, with the predicted size of a theoretical CTF indicated with a vertical line and asterisk.
(H) Quantitative changes in substrate CTF levels normalized to the empty vector control. For comparisons between PS1-D257A and PS1 conditions, *p < 0.05, **p < 0.01, ***p < 0.001.

HEK cells stably expressing the indicated constructs were treated for 16 h with DAPT (+) or DMSO vehicle alone (–) (left panels), or for 6–16 h with DAPT alone or in combination with the β-secretase inhibitor C3, the metalloprotease inhibitor GM6001, the phorbol ester PMA, and the proteasome inhibitor epoxomicin (right panels). The constructs expressed are (A) vasorin, (B) DG, (D) DNER, (E) NPR-C, (F) DSG2, (G) PLXDC2, (H) Itgβ1, and (I) NPR-A.
(C) DG was also expressed in HeLa cells, revealing just a single CTF that again accumulated with DAPT treatment. The asterisk in (G) indicates a PLXDC2 fragment that accumulated with simultaneous epoxomicin and DAPT treatment (data not shown). A quantitative analysis of these experiments is shown in Figure S1.

The lumenal, transmembrane, and/or cytoplasmic domains of integrin (a nonsubstrate; red) and vasorin (a γ-secretase substrate; blue) were interchanged to determine how proteins are targeted for γ-secretase processing. HEK cells expressing the constructs were treated with DAPT or vehicle alone for 16 h. In the case of Itgβ1 CTF25, (E) PS-DKO fibroblast cells were also analyzed as in Figure 2.
(A and B) The full-length ectodomains of integrin (Int) and vasorin (Vas) were exchanged, yielding a CTF regulated by γ-secretase only from the Vas/Int chimera.
(C–K) Recombinant CTFs were generated by fusing the protein's signal peptide to the lumenal domain either 25 (CTF25) or 13 (CTF13) residues N-terminal to the transmembrane domain. Integrin CTF constructs with an ectodomain of 25 residues (D and E) or 13 residues (F) were not processed by γ-secretase. This was also true of the Itgβ1-LZ CTF13 construct that has a leucine zipper at its C-terminus to facilitate homo-dimerization (K).
(C, I, and J) Chimeric CTFs containing an integrin transmembrane and/or cytoplasmic domain were not processed by γ-secretase.
(G and H) Vasorin CTF13 and chimeric constructs containing both the vasorin transmembrane and cytoplasmic domains (Int/Vas CTF13) are processed by γ-secretase to produce ICDs (lower panel, darker exposure), and CTFs accumulate in the presence of DAPT. Note that in (H), Int/Vas CTF13 chimeras were generated using either the vasorin signal peptide (Int/Vas^SP CTF13, left panel) or the integrin signal peptide (Int/Vas CTF13, right panel), producing similar results.
(L) The fold accumulation of CTFs in response to DAPT versus DMSO control; *p < 0.05, ***p < 0.001.

Regions of NPR-C (green) and NPR-A (yellow) were recombined to generate chimeric CTFs. (A–C) The NPR-A CTF13 (A) and other NPR constructs containing the full-length NPR-A cytoplasmic domain, whether fused to the NPR-C lumenal and transmembrane domains (NPR-C/A^cyto CTF13) (B) or fused to the end of NPR-C (NPR-C+A CTF13) (C), were not regulated by γ-secretase.
(D–F) The levels of NPR-C CTF13 (D) and NPR-A/C chimeras containing the lumenal domain of NPR-A (NPR-A/C CTF13) (E) or the lumenal and transmembrane domains of NPR-A (NPR-A/C^cyto CTF13) (F) were significantly elevated in response to DAPT, indicating their processing by γ-secretase.
(G) NPR-A ΔKG CTF13, which lacks the characterized NPR-A cytoplasmic domains, was also elevated by γ-secretase inhibition. By Western blot, this construct appears as two distinct bands.
(H) Vasn+NPR-A CTF13 levels were increased modestly with DAPT treatment, but relative accumulation of the CTF was far less than vasorin CTF13 (Figure 4G).
(I) The fold accumulation of CTFs arising from DAPT treatment versus DMSO control; *p < 0.05, **p < 0.01, ***p < 0.001.

(A) HEK cells stably expressing the indicated FLAG-tagged, full-length or CTF constructs or else untransfected control were immunoprecipitated for the FLAG epitope to probe for association with the γ-secretase complex. Immunoprecipitated samples were probed by Western blot for nicastrin (top panel), PS1 (middle panel), and the FLAG-tagged protein (bottom panel). Whole-cell lysate from control cells was run in lane 1 for comparison.
(B) Amino-acid sequences of the γ-secretase substrates identified in our screen and two nonsubstrates (Itgβ1 and NPR-A; bottom), with the putative α-secretase (ectodomain) and γ-secretase cleavage sites indicated at the top. Sequences are aligned along their transmembrane domain, and amino acids showing similarity down a column are highlighted according to side chain polarity: acidic (red), basic (blue), polar neutral (grey), and nonpolar (green).