Mass-tag labeling reveals site-specific and endogenouslevels of protein S-fatty acylationAvital Perchera,1, Srinivasan Ramakrishnana,1, Emmanuelle Thinona, Xiaoqiu Yuana, Jacob S. Younta,b,2,and Howard C. Hanga,2

aLaboratory of Chemical Biology and Microbial Pathogenesis, The Rockefeller University, New York, NY 10065; and bDepartment of Microbial Infectionand Immunity, Center for Microbial Interface Biology, The Ohio State University, Columbus, OH 43210

Edited by Pietro De Camilli, Yale University and Howard Hughes Medical Institute, New Haven, CT, and approved March 1, 2016 (received for review February11, 2016)

Fatty acylation of cysteine residues provides spatial and temporalcontrol of protein function in cells and regulates important biologicalpathways in eukaryotes. Although recent methods have improved thedetection and proteomic analysis of cysteine fatty (S-fatty) acylatedproteins, understanding how specific sites and quantitative levels ofthis posttranslational modification modulate cellular pathways are stillchallenging. To analyze the endogenous levels of protein S-fattyacylation in cells, we developed amass-tag labeling method based onhydroxylamine-sensitivity of thioesters and selective maleimide-mod-ification of cysteines, termed acyl-PEG exchange (APE). We demon-strate that APE enables sensitive detection of protein S-acylationlevels and is broadly applicable to different classes of S-palmitoylatedmembrane proteins. Using APE, we show that endogenous interferon-induced transmembrane protein 3 is S-fatty acylated on three cysteineresidues and site-specific modification of highly conserved cysteinesare crucial for the antiviral activity of this IFN-stimulated immuneeffector. APE therefore provides a general and sensitive methodfor analyzing the endogenous levels of protein S-fatty acylationand should facilitate quantitative studies of this regulated anddynamic lipid modification in biological systems.

fatty-acylation | palmitoylation | PEGylation | influenza virus | IFITM3

Protein S-fatty acylation describes the covalent attachment oflong-chain fatty acids to cysteine (Cys) residues through a thio-

ester bond, which alters the hydrophobicity of diverse proteins andregulates their stability, trafficking, and activity in eukaryotic cells(Fig. 1A) (1, 2). Cys residues are predominately acylated with pal-mitic acid (S-palmitoylation), but can also be modified with longerchain and unsaturated fatty acids, thus more generally described asS-fatty acylation (1, 3, 4). The fatty-acylation of Cys residues isregulated by the DHHC-family of protein acyltransferases (5, 6) anddifferent classes of thioesterases (7, 8) that are associated with avariety of important physiological pathways and diseases (1). De-termining the precise levels of protein S-fatty acylation is thereforecrucial for understanding how this dynamic lipid modification isregulated and quantitatively controls specific cellular pathways.Recent methods to detect and enrich S-fatty acylated proteins

have facilitated the characterization of key regulatory mecha-nisms and discovery of new S-fatty acylated proteins (1, 2). Forexample, alkyne-modified fatty acid chemical reporters haveenabled the fluorescent detection and enrichment of metaboli-cally labeled proteins using bioorthogonal ligation methods (Fig.S1A) (9, 10). Alternatively, exploitation of thioester sensitivity tohydroxylamine (NH2OH) has enabled selective capture andanalysis of S-acylated proteins by acyl-biotin exchange (ABE)(Fig. S1B) (11, 12) or acyl-resin capture (acyl-RAC) (Fig. S1C)(13). However, all of these methods do not readily reveal thefraction of unmodified versus S-fatty acylated proteins or thenumber of sites of S-acylation, which are both crucial for un-derstanding how quantitative differences in S-fatty acylationcontrol protein function and associated cellular phenotypes.To evaluate endogenous levels of S-fatty acylation, we de-

veloped a mass-tag labeling method termed acyl-PEG exchange(APE) that exploits the NH2OH-sensitivity of thioesters and

selective reactivity of Cys residues for site-specific alkylation withmaleimide-functionalized polyethylene glycol reagents (Fig. 1B).APE induces a mobility-shift on S-acylated proteins that can bereadily monitored by Western blot of target proteins and circum-vents the need for metabolic labeling or affinity enrichment ofproteins. We demonstrate that APE can reveal endogenous levels,stoichiometry and specific sites of S-acylation on diverse S-palmi-toylated proteins. Using APE, we characterized the endogenouslevels of interferon-induced transmembrane protein 3 (IFITM3)S-fatty acylation and show that site-specific modifications of highlyconserved Cys residues are crucial for the antiviral activity of thiskey IFN-induced immune effector. APE provides a sensitive andreadily accessible method for evaluating endogenous S-fatty acyl-ation levels and should facilitate the quantitative analysis of thisdynamic lipid modification in diverse cell types and tissues.

ResultsAPE Enables Direct Visualization of S-Acylated Protein Isoforms. Toestablish the utility of APE, we first analyzed HRas, a well-characterized S-palmitoylated protein. HRas contains one far-nesylation site at Cys-186 and two S-palmitoylation sites atCys-181 and 184, which regulate the trafficking and activity of thissmall GTPase at the plasma membrane and internal membranecompartments (14). To explore mass-shift detection by APE,HEK293T cells were transfected with N-terminally HA-taggedHRas (HA-HRas) and lysed with SDS-containing buffer. Lysateswere reacted with Tris(2-carboxyethyl)phosphine (TCEP) andN-ethylmaleimide (NEM) to reduce and block free Cys residues.Proteins were then treated with NH2OH to cleave thioesters,alkylated with methoxy-PEG-maleimide (mPEG-Mal), and analyzed

Significance

Cysteine fatty acylation (S-fatty acylation) regulates the stability,trafficking, and activity of proteins in eukaryotes. Currentmethods to study S-fatty acylation rely on the metabolic incor-poration of fatty acid analogs or selective chemical labeling andaffinity purification, which limits the detection of endogenousS-fatty acylated protein isoforms and levels. Here we demon-strate that the biochemical exchange of acyl groups on cysteineswith defined mass-tags enables the direct visualization of en-dogenous S-fatty acylated protein levels. The application of thismass-tag labeling method revealed that site-specific S-fatty ac-ylation of interferon-induced transmembrane protein 3 controlsthe antiviral activity of this IFN-induced effector.

Author contributions: A.P., S.R., E.T., J.S.Y., and H.C.H. designed research; A.P., S.R., E.T., X.Y.,and J.S.Y. performed research; A.P., S.R., E.T., X.Y., J.S.Y., and H.C.H. analyzed data; and A.P.and H.C.H. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1A.P. and S.R. contributed equally to this work.2To whom correspondence may be addressed. Email: yount.37@osu.edu or hhang@rockefeller.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1602244113/-/DCSupplemental.

4302–4307 | PNAS | April 19, 2016 | vol. 113 | no. 16 www.pnas.org/cgi/doi/10.1073/pnas.1602244113

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 11

, 202

0

by Western blot (Fig. 2A). Reaction of cell lysates with 5 kDamPEG-Mal alone revealed three slower migrating HRas species,consistent with three unmodified Cys residues at positions 51, 80, and118 on HRas that were blocked by pretreatment with NEM (lanes1–3, Fig. 2A). Cleavage of thioesters with NH2OH followedby mPEG-Mal alkylation resulted in unmodified polypeptide at25 kDa (apo) and two slower migrating polypeptides, whichcorresponded to mono-PEGylated (*) and di-PEGylated (**)proteins (lane 4, Fig. 2A). Reaction of cell lysates with 10 kDamPEG-Mal induced a mass shift approximately twice that observedfor the 5 kDa reagent, consistent with site-specific PEGylation(lane 5, Fig. 2A). The NH2OH-dependent PEGylation-inducedmass shift correlates with the two predicted sites of S-palmi-toylation on HRas (14). In parallel, we also analyzed endogenouscalnexin (CANX), a type I integral membrane protein with twoS-palmitoylation sites (15, 16), which yielded mono- and di-PEGylatedpolypeptides, suggesting CANX is fully S-acylated under theseconditions (Fig. 2A). To optimize APE detection of S-acylatedproteins, we tested varying concentrations of NH2OH andmPEG-Mal for selective thioester cleavage and Cys alkylation ofproteins, respectively. We determined that 0.75 M NH2OHtreatment for 1 h and 1 mM mPEG-Mal labeling for 2 h wasoptimal for APE detection of transfected HA-HRas and endog-enous CANX, which provides a useful protein loading control forAPE (Fig. S2 A and B). Increasing concentrations of mPEG-Mal canresult in nonspecific alkylation, as seen for CANX in samples nottreated with NH2OH (Fig. S2C). Rab7, a prenylated GTPase that isnot S-palmitoylated (17), was also analyzed and did not exhibitNH2OH-dependent PEGylation (Fig. S2B), despite the presence ofmultiple Cys residues, confirming the specificity of the APEprotocol for S-acylated proteins. To further characterize ourprotocol for APE, we evaluated the effect of Triton X-100 andEDTA in the lysis and reaction buffers. Although additional TritonX-100 was not necessary to further solubilize proteins in SDS-containing lysis buffer, EDTA was required for complete NH2OH-dependent PEGylation of CANX, HRas, and IFITM3 (Fig. S2 Dand E). The addition of EDTA was particularly crucial during theNH2OH-treatment step of cell lysates, which likely chelates metalsinvolved in oxidation of the newly liberated Cys residues that wouldinterfere with subsequent PEGylation (Fig. S2 D and E). Using this

optimized APE protocol, a pan-Ras antibody readily detected theendogenous levels of mono- and di-S-fatty acylated Ras in HeLa cells(Fig. 2B), showing that the APE can reveal S-fatty acylation levels ofother endogenously expressed proteins in addition to abundantproteins such as CANX.We next compared our APE protocol with other protein

S-acylation detection methods. In comparison with ABE (11, 12) oracyl-RAC (13) (Fig. S1 B and C), APE enables the direct visual-ization of the different S-acylated isoforms of endogenouslyexpressed CANX and Ras. APE also circumvents the need foraffinity enrichment of proteins required for ABE and acyl-RAC,which can yield incomplete recovery of membrane-associatedproteins and confound estimates of S-acylated and unmodifiedprotein levels (Fig. S3A). With respect to the PEG-switch assay(PSA) (18), the reported conditions showed incomplete NH2OH-dependent PEGylation of CANX, Ras, and IFITM3 (lanes 2 and 8,Fig. S3B). In the PSA protocol, 2 mM mPEG-Mal was added to-gether with 200 mM NH2OH, which could react with the mal-eimide group and affect the efficiency of PEGylation. We thereforeseparated the NH2OH treatment and mPEG-Mal labeling steps byremoving excess NH2OH with chloroform-methanol precipitation.Separating the NH2OH-mediated deacylation and maleimide al-kylation reactions restored the PEGylation efficiency to levelssimilar to our APE conditions (lane 6, Fig. S3B). Coincubation ofNH2OH and mPEG-Mal in our APE protocol also significantlyreduced the efficiency of PEGylation (lane 4, Fig. S3B), confirmingthe need to separate these steps for optimal mass shift analysis. Tofurther determine the specificity and utility of our APE protocol,we analyzed several different classes of S-palmitoyated proteins inparallel with CANX, which provides an important S-fatty acylatedprotein loading control that typically yields ∼1:1 ratio of mono- anddi-S-acylated polypeptides with our optimized APE conditions. Inaddition to dually lipidated cell signaling factors such as HRas,APE can readily reveal the S-fatty acylation isoforms and levels ofother small GTPases (IRGM1) (19), multiple-pass membraneproteins (CD9) (20), as well as IFITM3 (21, 22) (Fig. 2C). Theanalysis of reported S-palmitoylation–deficient Cys mutants forthese proteins confirms their sites of modification and shows thelevels of S-fatty acylated isoforms when expressed in HEK293Tcells. These results suggest our optimized NH2OH-mediated acyl-PEGylation exchange protocol, which separates Cys deacylationand PEGylation steps, provides a robust and general method toevaluate protein S-acylation levels.APE shows S-acylated protein isoforms and their relative levels

but does not reveal whether specific proteins are directly fatty acid-modified or the dynamics of modifications at specific sites. Indeed,other S-acylated proteins in mammalian cells (e.g., enzymes, in-volved in ubiquitin and ubiquitin-like protein modification) (23),also contain thioester intermediates that could be subject toNH2OH-mediated PEGylation. To characterize protein S-acylationlevels relative to fatty acid incorporation, we directly comparedHRas S-fatty acylation by APE (Fig. 1B) and metabolic labeling(Fig. S1A). HEK293T cells were transfected with HA-HRas con-structs, pulse-labeled with an alkyne-palmitic acid chemical reporter(alk-16), lysed, and analyzed by APE or subjected to bioorthogonalreaction with azide-rhodamine and in-gel fluorescence detection(Fig. 3). APE analysis of HA-HRas showed three α-HA reactivepolypeptides corresponding to ∼0.35 unmodified (apo), 0.51 mono-PEGylated (*) and 0.14 di-PEGylated (**) species (Fig. 3A). Thesingle Cys mutants (C181S, C184S) revealed ∼0.20 unmodified and0.80 mono-PEGylated species, and as expected, the double Cysmutant (C181S, C184S) revealed only unmodified HA-HRas at25 kDa (Fig. 3A). These results show that mutation of individualS-fatty acylation sites still afforded significant levels, ∼80%, of mono-S-acylated HRas (Fig. 3A). In contrast, metabolic labeling withalk-16 and in-gel fluorescence detection showed at least 50%decrease in labeling of single Cys mutants of HA-HRas (Fig. 3B).These results suggest that mono-S-fatty acylated HA-HRas isrelatively stable, whereas dual S-fatty acylation is more dynami-cally regulated, which is consistent with previous studies ofHRas fatty acylation, localization, and activity (14). This direct

Fig. 1. Mass-tag analysis of protein S-fatty acylation. (A) S-fatty acylationcontrols the trafficking, stability, and function of membrane proteins andare regulated by DHHC-protein acyltransferases (DHHC-PATs), and esterases.(B) With APE, cell lysates are reduced with TCEP, and then free cysteineresidues are capped with NEM. S-fatty acid groups are removed by NH2OH,and the exposed cysteines are reacted with mPEG-Mal. Proteins are sepa-rated by SDS/PAGE and analyzed by Western blot, enabling the detection ofboth unmodified and S-fatty acylated proteins.

Percher et al. PNAS | April 19, 2016 | vol. 113 | no. 16 | 4303

BIOCH

EMISTR

Y

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 11

, 202

0

comparison highlights the utility of monitoring protein S-fattyacylation with both APE and metabolic labeling, which reveals therelative levels of unmodified and S-fatty acylated protein isoformsas well as site-specific differences in dynamic S-fatty acylation.

IFITM3 Antiviral Activity Is Controlled by Specific S-Fatty AcylationSites and Levels. We next focused on sites and levels of IFITM3S-fatty acylation, which was discovered by our laboratory from alk-16metabolic labeling and proteomic studies (21, 22). Our initial studiesdemonstrated that the mutation of all three Cys residues (71, 72, and105) to alanine (Ala) in murine IFITM3 abrogated S-fatty acylationand decreased antiviral activity (21, 22). However, the endogenouslevels of IFITM3 S-fatty acylation and specific sites of modificationhave not been characterized directly. APE analysis of endogenouslyexpressed IFITM3 in IFN/LPS-stimulated murine NIH 3T3 fibro-blasts and Raw264.7 macrophages revealed three slower migratingPEGylated polypeptides that correspond to the mono, di-, and tri-S-fatty acylated isoforms of mIFITM3 (Fig. 4A and Fig. S4A). SimilarAPE results were also observed for human IFITM3 in IFN-stimu-lated A549 lung epithelial cells (Fig. S5A), demonstrating that themajority, if not all, of endogenously expressed IFITM3 in IFN-stimulated cells is S-fatty acylated on at least one Cys residue. Toevaluate the contribution of individual Cys residues on IFITM3S-fatty acylation, we generated single and double Cys mutants andanalyzed their levels of S-fatty acylation by APE and metaboliclabeling. APE analysis of the wild-type and the individual Cys toAla HA-mIFITM3 mutants transfected into unstimulated murineNIH 3T3 fibroblasts showed that mutation of Cys71 and 105 to Alaeliminated the tri-S-acylated fraction and decreased the levels ofthe mono- and di-S-acylated mIFITM3 (Fig. 4B and Fig. S4B). In

contrast, the C72A mutant or multiple Cys mutants abrogated thetri- and di-S-acylated IFITM3 and even lost the mono-S-acylatedfraction for the C72,105A mutant (Fig. 4B). Similar APE resultswere also observed with human IFITM3 constructs expressedin HEK293T cells (Fig. S5B). Metabolic labeling of these HA-mIFITM3 Cys mutants with alk-16 and in-gel fluorescence de-tection also revealed C72 is the major site of S-fatty acylation inNIH 3T3 fibroblasts (Fig. 4C) and HEK293T cells (Fig. S4C). C72is highly conserved among IFITM isoforms (24, 25) and is also akey site of S-fatty acylation for human IFITM1, -2, and -3 (Fig.S5C). The comparison of APE and alk-16 labeling suggests that,although the C72A IFITM3 mutant is still significantly mono-S-fatty acylated (Fig. 4B and Figs. S4B and S5B), the majority offatty acid metabolic labeling is abrogated (Fig. 4C and Figs. S4Cand S5C). These results demonstrate that, similar to HRas (Fig.3), mono-S-fatty acylation of IFITM3 is relatively stable but dualS-fatty acylation is more dynamically regulated.To determine the functional significance of site-specific IFITM3

S-fatty acylation, we evaluated the cellular localization and anti-viral activity of these IFITM3 Cys mutants. Consistent with pre-vious overexpression and immunofluorescence studies (21, 26, 27),the majority of wild-type HA-mIFITM3 constructs colocalizedwith lysosomes marked by LAMP1-GFP in NIH 3T3 fibroblasts(Fig. S6A). A minor fraction of HA-mIFITM3 also colocalizedwith GFP-Rab5 and Rab7-labeled endocytic vesicles (Fig. S6 Cand B). Despite their differences in protein S-fatty acylation levels(Fig. 4), mIFITM3 Cys mutants showed similar cellular distribu-tion in NIH 3T3 fibroblasts by quantitative immunofluorescenceanalysis (Fig. S6 A–D). Similar cellular distribution was alsoobserved for the wild-type and HA-hIFITM3 Cys mutants in

Fig. 2. APE enables robust detection of protein S-fatty acylation levels. (A) HEK293T transfected with HA-HRas were lysed and total cell lysates were sub-jected to APE with NEM (25 mM), NH-2OH (0.75 M), and mPEG-Mal (1 mM), and compared with negative controls. Samples were analyzed by Western blotusing anti-HA and anti-CANX antibodies. The number of PEGylation events are indicated by asterisks (*). Apo refers to non-PEGylated protein. (B) HeLa cellswere subjected to APE and endogenous Ras palmitoylation measured by Western blot with a pan-Ras antibody. (C) HEK293T cells expressing HA-tagged WTor palmitoylation-deficient (P△) constructs (HRas C181,184S; IRGM1 C371, -373, -374, -375A; mIFITM3 C71, -72, -105A; CD9 C9, -78, -79, -87, -218, -219A) ofknown S-palmitoylated proteins were analyzed with the APE. Representative of multiple experiments for all constructs.

4304 | www.pnas.org/cgi/doi/10.1073/pnas.1602244113 Percher et al.

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 11

, 202

0

HEK293T cells (Fig. S7). Nonetheless, these IFITM3 Cys mutantsexhibited site-specific effects on antiviral activity against influenzavirus infection (Fig. 4D and Fig. S8). To evaluate the site-specificeffects of S-fatty acylation on IFITM3 antiviral activity, unstimu-lated NIH 3T3 cells were transfected with the HA-mIFITM3constructs, then infected with influenza virus (PR8) and the per-centage of infected cells was quantified by anti-influenza nuclearprotein (NP) staining using flow cytometry, as previously described(21, 22). Our results show that the C71A mutation does not sig-nificantly impair HA-mIFITM3 antiviral activity comparedwith wild-type, whereas C72A and C105A mutants show ∼50%decrease in antiviral activity (Fig. 4D and Fig. S8A). The doubleCys mutants of mIFITM3 also exhibited diminished antiviral

activity, with the C71,72A mutant showing loss of antiviral activitycomparable to the S-fatty acylation-deficient C71,72,105A mutant(Fig. 4D and Fig. S8A). Similar antiviral activity for the murine andhuman IFITM3 Cys mutants were observed in HEK293T cells (Fig.S8 B and C). These results demonstrate that, whereas site-specificdifferences in S-fatty acylation do not significantly affect IFITM3cellular distribution upon overexpression, the lower levels of S-fattyacylation at specific Cys residues decrease IFITM3 antiviral activity.

Discussion and ConclusionThe functional roles of protein S-fatty acylation in eukaryotes areexpanding from proteomic studies and characterization of regula-tory enzymes (1, 2). To understand the biological significance ofprotein S-fatty acylation in diverse biological pathways and disease,robust methods are needed for quantifying S-fatty acylated proteinisoforms and levels. Inspired by mass shift-based assays for evalu-ating membrane topology of proteins (28), oxidation of Cys resi-dues (29), and protein glycosylation (30, 31), we demonstrate thatthe NH2OH-mediated APE provides a sensitive and readily ac-cessible method for determining the S-acylation levels of specificproteins that is not possible by previously reported methods usingmetabolic labeling or affinity enrichment (Fig. 1B). APE enablesthe direct visualization of overexpressed and endogenous levels ofS-acylated protein isoforms by Western blot analysis and is broadlyapplicable to different classes of S-fatty acylated proteins (Fig. 2).The combination of APE and metabolic labeling provides com-plementary methods to visualize different isoforms as well as thelevels and dynamics of S-fatty acylation (Fig. 3).Using APE we demonstrate that endogenous IFITM3 is mostly

S-fatty acylated in IFN-stimulated cells and that S-fatty acylation ofspecific Cys residues and levels are crucial for IFITM3 antiviral ac-tivity against influenza virus. The bioinformatic analysis of IFITMisoforms in diverse vertebrates shows that these three S-fatty acyl-ated Cys residues (C71, -72, and -105 in mIFITM3) are highlyconserved within this family of membrane proteins (Fig. S9) (24, 25).Notably, our APE analysis and fatty acid metabolic labeling show, toour knowledge for the first time, that the majority of endogenousIFITM3 is S-fatty acylated in IFN-stimulated cells and Cys72 is themost prominent S-fatty acylation site (Fig. 4). Although previousstudies have suggested that mutation of these conserved Cys residuesmay partially affect the localization or trafficking of IFITMs inmammalian cells (21, 22, 27, 32), our quantitative immunofluores-cence analysis suggest that site-specific S-fatty acylation may regulatemore subtle functions of IFITM3 beyond lysosomal targeting (Figs.S7 and S8), such as protein–protein interactions (33, 34) or mem-brane tethering, but which awaits more detailed biochemical orbiophysical studies. Nonetheless, these differences in site-specificS-fatty acylation have significant effects on IFITM3 antiviral activity.In murine and human IFITM3, even though Cys71 is highly con-served, this residue is dispensable for dual S-fatty acylation and an-tiviral activity (Fig. 4). In contrast, Cys72 in IFITM3 is the mostprominent site of S-fatty acylation and results in the loss of antivi-ral activity when mutated to Ala (Fig. 4 and Figs. S5 and S6). TheS-fatty acylation and antiviral activity of Cys105 may depend on theIFITM isoform, expression levels and cell-type, as the mutation ofthis Cys residue decreases murine IFITM3 antiviral activity but is notrequired for human IFITM3 activity (Fig. 4 and Figs. S5 and S6).Our results are consistent with previous IFITM mutagenesis andantiviral activity studies (21, 22, 27, 32, 35), but demonstrates thatantiviral activity is directly correlated to S-fatty acylation levels andthat dually S-fatty acylated IFITM3 at Cys72 and Cys105 is likely themost active isoform in mammalian cells (Fig. S9). In summary, wedemonstrate that NH2OH-mediated APE provides a sensitive andreadily accessible method to measure S-fatty acylation levels on di-verse proteins in different cell types that should greatly facilitate theanalysis of protein S-fatty acylation in physiology and disease.

Materials and MethodsFor APE, cell samples were lysed with 4% (wt/vol) SDS (Fischer) in TEA buffer[pH 7.3, 50 mM triethanolamine (TEA), 150 mM NaCl] containing 1× proteaseinhibitor mixture (Roche), 5 mM PMSF (Sigma), 5 mM EDTA (Fischer), and 1,500

Fig. 3. APE reveals site-specific S-fatty acylation levels. (A) HEK293T cellstransfected with HA-HRas wild-type or cysteine mutants were labeled for 2 hwith 50 μM alk-16 in DMEM containing charcoal-dextran–treated FBS.Samples were lysed, subjected to APE, separated by SDS/PAGE, and analyzedby Western blot. NT refers to nontreated control. The number of PEGylationevents are indicated by asterisks (*). (B) Samples from the same cell lysate asin A were immunoprecipitated with anti-HA agarose-beads, and reactedwith azide-rhodamine (az-rho) by Cu(I)-catalyzed azide-alkyne cycloaddition(CuAAC). Whole-cell lysate (WCL) was reacted with az-rho by CuAAC andincluded for comparison. Samples were separated by SDS/PAGE and ana-lyzed by in-gel fluorescence and Western blot.

Percher et al. PNAS | April 19, 2016 | vol. 113 | no. 16 | 4305

BIOCH

EMISTR

Y

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 11

, 202

0

units/mL benzonase (EMD). The protein concentration of the cell lysate was thenmeasured using a BCA assay (Thermo) and adjusted to 2 mg/mL with lysis buffer.Typically, 200 μg of total protein in 92.5 μL of lysis buffer was treatedwith 5 μL of200 mM neutralized TCEP (Thermo) for final concentration of 10 mM TCEP for30 min with nutation. NEM (Sigma), 2.5 μL from freshly made 1 M stock inethanol, was added for a final concentration of 25 mM and incubated for 2 h atroom temperature. Reductive alkylation of the proteins was then terminated bymethanol-chloroform-H2O precipitation (4:1.5:3) with sequential addition ofmethanol (400 μL), chloroform (150 μL), and distilled H2O (300 μL) (all prechilledon ice). The reactions were then mixed by inversion and centrifuged (Centrifuge5417R, Eppendorf) at 20,000 × g for 5 min at 4 °C. To pellet the precipitatedproteins, the aqueous layer was removed, 1 mL of prechilled MeOH was added,the Eppendorf tube inverted several times and centrifuged at 20,000 × g for3 min at 4 °C. The supernatant was then decanted, and the protein pellet washedonce more with 800 μL of prechilled MeOH, centrifuged again, and dried using aspeed-vacuum (Centrivap Concentrator, Labconco) To ensure complete removalof NEM from the protein pellets, the samples were resuspended with 100 μL ofTEA buffer containing 4% SDS, warmed to 37 °C for 10 min, briefly (∼5 s) soni-cated (Ultrasonic Cleaner, VWR), and subjected to two additional rounds ofmethanol-chloroform-H2O precipitations, as described above.

For hydroxylamine (NH2OH) cleavage and mPEG-maleimide alkylation,the protein pellet was resuspended in 30 μL TEA buffer containing 4% SDS,4 mM EDTA and treated with 90 μL of 1 M neutralized NH2OH (J. T. Baker)dissolved in TEA buffer pH 7.3, containing 0.2% Triton X-100 (Fisher) toobtain a final concentration of 0.75 M NH2OH. Protease inhibitor mixture orPMSF should be omitted, as these reagents can interfere with the NH2OHreactivity. Control samples not treated with NH2OH were diluted in 90 μLTEA buffer with 0.2% Triton X-100. Samples were incubated at room tem-perature for 1 h with nutation. The samples were then subjected to meth-anol-chloroform-H2O precipitation, as described above, and resuspendedin 30 μL TEA buffer containing 4% SDS, 4 mM EDTA, warmed to 37 °C for10 min, and briefly (∼5 s) sonicated and treated with 90 μL TEA buffer with0.2% Triton X-100 and 1.33 mM mPEG-Mal (5 or 10 kDa; Sigma) for a finalconcentration of 1 mM mPEG-Mal. Samples were incubated for 2 h at roomtemperature with nutation before a final methanol-chloroform-H2O pre-cipitation. Dried protein pellets were resuspended in 50 μL 1× Laemmlibuffer (Bio-Rad) and then heated for 5 min at 95 °C. Typically, 15 μL of thesample was loaded in 4–20% Criterion-TGX Stain Free polyacrylamide gels(Bio-Rad), separated by SDS/PAGE, and analyzed byWestern blot. For Westernblots, primary antibodies used were anticalnexin (1:2,000 ab22595; Abcam),

Fig. 4. Murine IFITM3 S-fatty acylation levels, sites of modification and antiviral activity. (A) NIH 3T3 or Raw264.7 macrophages were activated with 500 ng/mLLPS, 100 U/mL IFN-γ for 16 h, subjected to APE, and analyzed by Western blot. (B) NIH 3T3 cells transfected with WT HA-mIFITM3, or Cys to Ala mutant constructswere subjected to APE and analyzed by Western blot. The number of PEGylation events are indicated by asterisks (*). Analysis of whole-cell lysate (WCL) indicateslevels of protein expression without APE. (C) NIH 3T3 cells transfected with HA-mIFITM3 constructs were metabolically labeled for 2 h with 50 μM alk-16. Celllysates prepared with 1% Brij 97 were subjected to immunoprecipitation with anti-HA agarose-beads, reacted with az-rho by CuAAC, separated by SDS/PAGE, andvisualized by fluorescence gel scanning. Comparable protein loading was confirmed by anti-HA Western blotting. (D) NIH 3T3 cells were transfected with HA-mIFITM3 constructs, followed by infection with PR8 influenza virus at a multiplicity of infection of 5 for 18 h. Cells were fixed, permeabilized with 0.1% Triton X-100, and stained with anti-influenza NP antibody and analyzed by flow cytometry. Graph shows anti-influenza NP+ cells for each condition; average of triplicate.Error bar represents SEM, n = 3.

4306 | www.pnas.org/cgi/doi/10.1073/pnas.1602244113 Percher et al.

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 11

, 202

0

anti-pan Ras (1:500; Ras10, Millipore), anti-mouse IFITM3 (1:1,000; ab15592,Abcam), anti-FLAG (1:1,000; F1804, Sigma), anti-HA (1:1,000; ab9134, Abcam),and HRP-conjugated anti-HA (3F10, Roche). Secondary antibodies used wereHRP-conjugated goat anti-rabbit (DC03L, Calbiochem), and goat anti-mouse(ab97023, Abcam). Protein detection was performed with ECL detection re-agent (GE Healthcare) on a Bio-Rad ChemiDoc MP Imaging System.

Other materials and methods are provided in SI Materials and Methods.

ACKNOWLEDGMENTS. This work was supported in part by a NationalScience Foundation graduate fellowship (to A.P.); a Marie Sklodowska-Curieactions for a postdoctoral fellowship (to E.T.); National Institutes of Health-National Institute of Allergy and Infectious Diseases Grants R00AI095348 andR56AI114826 (to J.S.Y.); Starr Cancer Consortium Grant I7-A717 (to H.C.H.);and National Institutes of Health-National Institute of General MedicalSciences Grant R01GM087544 (to H.C.H.). X.Y. is a graduate student in theTri-Institutional Program in Chemical Biology.

1. Chamberlain LH, Shipston MJ (2015) The physiology of protein S-acylation. Physiol Rev

95(2):341–376.2. Peng T, Thinon E, Hang HC (2016) Proteomic analysis of fatty-acylated proteins. Curr

Opin Chem Biol 30:77–86.3. Liang X, et al. (2001) Heterogeneous fatty acylation of Src family kinases with poly-

unsaturated fatty acids regulates raft localization and signal transduction. J Biol

Chem 276(33):30987–30994.4. Brett K, et al. (2014) Site-specific S-acylation of influenza virus hemagglutinin: the

location of the acylation site relative to the membrane border is the decisive factor

for attachment of stearate. J Biol Chem 289(50):34978–34989.5. Fukata Y, Fukata M (2010) Protein palmitoylation in neuronal development and

synaptic plasticity. Nat Rev Neurosci 11(3):161–175.6. Linder ME, Deschenes RJ (2007) Palmitoylation: Policing protein stability and traffic.

Nat Rev Mol Cell Biol 8(1):74–84.7. Lin DT, Conibear E (2015) Enzymatic protein depalmitoylation by acyl protein thio-

esterases. Biochem Soc Trans 43(2):193–198.8. Lin DT, Conibear E (2015) ABHD17 proteins are novel protein depalmitoylases that

regulate N-Ras palmitate turnover and subcellular localization. eLife 4:4.9. Hang HC, Wilson JP, Charron G (2011) Bioorthogonal chemical reporters for analyzing

protein lipidation and lipid trafficking. Acc Chem Res 44(9):699–708.10. Charron G, et al. (2009) Robust fluorescent detection of protein fatty-acylation with

chemical reporters. J Am Chem Soc 131(13):4967–4975.11. Kang R, et al. (2008) Neural palmitoyl-proteomics reveals dynamic synaptic palmi-

toylation. Nature 456(7224):904–909.12. Wan J, Roth AF, Bailey AO, Davis NG (2007) Palmitoylated proteins: Purification and

identification. Nat Protoc 2(7):1573–1584.13. Forrester MT, et al. (2011) Site-specific analysis of protein S-acylation by resin-assisted

capture. J Lipid Res 52(2):393–398.14. Rocks O, et al. (2005) An acylation cycle regulates localization and activity of palmi-

toylated Ras isoforms. Science 307(5716):1746–1752.15. Lakkaraju AK, et al. (2012) Palmitoylated calnexin is a key component of the ribo-

some-translocon complex. EMBO J 31(7):1823–1835.16. Lynes EM, et al. (2012) Palmitoylated TMX and calnexin target to the mitochondria-

associated membrane. EMBO J 31(2):457–470.17. Thomä NH, Niculae A, Goody RS, Alexandrov K (2001) Double prenylation by

RabGGTase can proceed without dissociation of the mono-prenylated intermediate.

J Biol Chem 276(52):48631–48636.18. Howie J, et al. (2014) Substrate recognition by the cell surface palmitoyl transferase

DHHC5. Proc Natl Acad Sci USA 111(49):17534–17539.

19. Henry SC, Schmidt EA, Fessler MB, Taylor GA (2014) Palmitoylation of the immunityrelated GTPase, Irgm1: Impact on membrane localization and ability to promotemitochondrial fission. PLoS One 9(4):e95021.

20. Charrin S, et al. (2002) Differential stability of tetraspanin/tetraspanin interactions:Role of palmitoylation. FEBS Lett 516(1-3):139–144.

21. Yount JS, Karssemeijer RA, Hang HC (2012) S-palmitoylation and ubiquitination dif-ferentially regulate interferon-induced transmembrane protein 3 (IFITM3)-mediatedresistance to influenza virus. J Biol Chem 287(23):19631–19641.

22. Yount JS, et al. (2010) Palmitoylome profiling reveals S-palmitoylation-dependentantiviral activity of IFITM3. Nat Chem Biol 6(8):610–614.

23. Kerscher O, Felberbaum R, Hochstrasser M (2006) Modification of proteins by ubiq-uitin and ubiquitin-like proteins. Annu Rev Cell Dev Biol 22:159–180.

24. Zhang Z, Liu J, Li M, Yang H, Zhang C (2012) Evolutionary dynamics of the interferon-induced transmembrane gene family in vertebrates. PLoS One 7(11):e49265.

25. Hickford D, Frankenberg S, Shaw G, Renfree MB (2012) Evolution of vertebrate in-terferon inducible transmembrane proteins. BMC Genomics 13:155.

26. Feeley EM, et al. (2011) IFITM3 inhibits influenza A virus infection by preventing cy-tosolic entry. PLoS Pathog 7(10):e1002337.

27. John SP, et al. (2013) The CD225 domain of IFITM3 is required for both IFITM proteinassociation and inhibition of influenza A virus and dengue virus replication. J Virol87(14):7837–7852.

28. Lu J, Deutsch C (2001) Pegylation: A method for assessing topological accessibilities inKv1.3. Biochemistry 40(44):13288–13301.

29. Burgoyne JR, Oviosu O, Eaton P (2013) The PEG-switch assay: A fast semi-quantitativemethod to determine protein reversible cysteine oxidation. J Pharmacol ToxicolMethods 68(3):297–301.

30. Rexach JE, et al. (2010) Quantification of O-glycosylation stoichiometry and dynamicsusing resolvable mass tags. Nat Chem Biol 6(9):645–651.

31. Ortiz-Meoz RF, Merbl Y, Kirschner MW, Walker S (2014) Microarray discovery of newOGT substrates: The medulloblastoma oncogene OTX2 is O-GlcNAcylated. J Am ChemSoc 136(13):4845–4848.

32. Narayana SK, et al. (2015) The interferon-induced transmembrane proteins, IFITM1,IFITM2, and IFITM3 inhibit hepatitis C virus entry. J Biol Chem 290(43):25946–25959.

33. Peng T, Hang HC (2015) Bifunctional fatty acid chemical reporter for analyzingS-palmitoylated membrane protein-protein interactions in mammalian cells. J AmChem Soc 137(2):556–559.

34. Tsukamoto T, et al. (2013) Role of S-palmitoylation on IFITM5 for the interaction withFKBP11 in osteoblast cells. PLoS One 8(9):e75831.

35. Hach JC, McMichael T, Chesarino NM, Yount JS (2013) Palmitoylation on conservedand nonconserved cysteines of murine IFITM1 regulates its stability and anti-influenzaA virus activity. J Virol 87(17):9923–9927.

Percher et al. PNAS | April 19, 2016 | vol. 113 | no. 16 | 4307

BIOCH

EMISTR

Y

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 11

, 202

0