Roseltide rT7 is a disulfide-rich, anionic, and cell-penetrating peptide that inhibits proteasomal degradation

Disulfide-rich plant peptides with molecular masses of 2–6 kDa represent an expanding class of peptidyl-type natural products with diverse functions. They are structurally compact, hyperstable, and underexplored as cell-penetrating agents that inhibit intracellular functions. Here, we report the discovery of an anionic, 34-residue peptide, the disulfide-rich roseltide rT7 from Hibiscus sabdariffa (of the Malvaceae family) that penetrates cells and inhibits their proteasomal activities. Combined proteomics and NMR spectroscopy revealed that roseltide rT7 is a cystine-knotted, six-cysteine hevein-like cysteine-rich peptide. A pair-wise comparison indicated that roseltide rT7 is >100-fold more stable against protease degradation than its S-alkylated analog. Confocal microscopy studies and cell-based assays disclosed that after roseltide rT7 penetrates cells, it causes accumulation of ubiquitinated proteins, inhibits human 20S proteasomes, reduces tumor necrosis factor-induced IκBα degradation, and decreases expression levels of intercellular adhesion molecule-1. Structure-activity studies revealed that roseltide rT7 uses a canonical substrate-binding mechanism for proteasomal inhibition enabled by an IIML motif embedded in its proline-rich and exceptionally long intercysteine loop 4. Taken together, our results provide mechanistic insights into a novel disulfide-rich, anionic, and cell-penetrating peptide, representing a potential lead for further development as a proteasomal inhibitor in anti-cancer or anti-inflammatory therapies.

Hibiscus sabdariffa, a medicinal plant that is native to Central and West Africa, is highly popular in Southeast Asia (see Fig. 1A). It belongs to the Malvaceae family and is known by the common name, roselle. The plant is used in the treatment of degenerative and inflammatory diseases, such as hypertension and cancer (1).
Previously, we reported the discovery of a family of six-cysteine hevein-like peptides (6C-HLPs) 3 from H. sabdariffa, designated roseltides (rT1-rT8) (2). Of the eight members identified in this family, roseltide rT1 and rT7 are the two most abundant roseltides found in the aqueous extract of the calyces. Roseltide rT1 was first identified as a human neutrophil elastase inhibitor (2) and recently as a cell-penetrating mitochondriatargeting peptide that modulates cellular bioenergetics via interaction with ATP synthase (3). However, roseltide rT7 has not yet been characterized.
Sequence comparison of roseltides rT1 and rT7 showed striking differences in the amino acid sequences and lengths of their intercysteine loop 4 (see Fig. 1B). Loop 4 of roseltide rT7 contains a 12-residue sequence, IIMLPTWPPRYV, which is proline-rich, hydrophobic, cationic, and eight amino acid residues longer than the corresponding loop 4 of rT1. A sequence search of roseltide rT7 loop 4 revealed that the hydrophobic IIML motif shares high sequence identity to epoxomicin and carmaphycin, two naturally occurring proteasome inhibitors of the ␣Ј,␤Ј-epoxyketone peptide family (see Fig. 1C) (4 -6). In addition, the presence of three proline residues and one bulky hydrophobic residue in loop 4 suggests that roseltide rT7 could penetrate cells and inhibit proteasomal activities.
In this study, we report the identification of roseltide rT7 as a hyperstable cell-penetrating proteasome inhibitor. Plants produce diversified protease inhibitors that impact nearly all biological processes for their defense and development (21,22). However, no plant-derived cell-penetrating cysteine-rich peptide (CRP) inhibitor has been reported to target chymotryspinlike proteasomes. In this regard, the cysteine-and proline-rich roseltide rT7 represents a "first-in-class" cell-penetrating proteasomal inhibitor with a high preference for chymotrypsinlike activities.

Characterization and synthesis of roseltide rT7
Roseltide rT7, the second most abundant CRP in H. sabdariffa aqueous calyces extract (Fig. 1B), has an m/z value of 3696 (2,3). Fig. S1 shows purified roseltide rT7 from the calyces extract using our previously established purification protocol for roseltide rT1. The primary amino acid sequence of roseltide rT7, CVSSGIVDACSECCEPDKCIIMLPTWPPRYVCSV, was confirmed using de novo peptide sequencing (Figs. S2-S4). Fig. 2 (A and B) shows the NMR structure of roseltide rT7 using the distance restraints obtained from 1 H-1 H 2D NOESY. All 1 H spin-spin systems of rT7 were identified, except prolines 16, 24, 27, and 28 (Table S1). We observed roseltide rT7 has at least three conformations in solution. This is due to several proline residues present in the peptide sequence. The major solution structure of rT7 was determined based on a total of 220 NMR-derived distance restraints. Fig. 2A shows the NMR ensemble of the 17 lowest-energy rT7 structures. The root-mean-square deviation (RMSD) value of the 17 best structures for residues Cys 1 -Glu 15 , Cys 19 -Leu 23 , and Arg 29 -Ser 33 was 0.81 Ϯ 0.24 Å, and that for all heavy atoms was 1.33 Ϯ 0.33 Å ( Table 1). The structure of rT7 is well-defined by a number of medium and long-range NOEs, which consists of two short extended anti-parallel ␤-strands (␤1, Ser 4 -Val 7 ; and ␤2, Try 30 -Ser 33 ). The loop region (Pro 24 -Pro 28 ) cannot be well-defined because there are three proline residues located within this loop. The solution structures of roseltide rT7 showed that the three disulfide bonds (Cys I-IV, II-V, and III-VI) adopted a cystine-knot fold similar to our previously published knottin family peptides of the same cysteine spacing pattern such as roseltide rT1 (2), bleogen pB1 (23), and alstotide as1 (24). Compared with roseltide rT1, roseltide rT7 has an unusually long intercysteine loop 4 in which the three prolines promote turn formation (Fig. 2, A and B).
To characterize the biological activity of roseltide rT7, we prepared its synthetic version via a stepwise solid-phase method using Fmoc chemistry (Fig. 3). After the removal of protecting groups and cleavage from the solid support by TFA, the linear precursor of roseltide rT7 was oxidatively folded using a combination of redox reagents consisting of cysteamine and cystamine in 0.1 M ammonium bicarbonate (pH 8) at 4°C for 1 h before being subjected to RP-HPLC purification. The purified synthetic and natural roseltide rT7 were indistinguishable as demonstrated by a co-eluted single peak using a RP-HPLC and 2D NOESY spectra overlay (Figs. S5-S7). Using similar methods, we also prepared the N-terminally tetramethylrhodamine (TAMRA)-labeled roseltide rT7 (TAMRA-rT7) as a chemical probe for biochemical and bioimaging studies. Synthetic roseltide rT7 was used throughout this study. Figure 1. A, the calyx of H. sabdariffa. B, sequence comparison between roseltide rT1 and rT7 as described previously (2). C, sequence comparison between the IIML motif of roseltide rT7, epoxomicin, and carmaphycin.

Disulfide-rich roseltide rT7 displays high proteolytic and structural stability
Cysteine-rich peptide cross-linked by multiple disulfides are known for their high stability against proteolytic degradation (2,3,23,25,26). Figs. 4 (A-D) shows that roseltide rT7 is stable against trypsin, Pronase, neutrophil elastase, and pepsin. It is Ͼ100-fold more stable than the S-alkylated form of roseltide rT7, an analog with all disulfide bridges reduced and S-alkylated. Using far-UV CD spectroscopy, we showed that the CD spectrum of roseltide rT7 was not altered at temperature up to 90°C, suggesting that roseltide rT7 is highly resistant to heat denaturation (Fig. S8).

Roseltide rT7 is cell penetrating
A few cysteine-rich peptides, including roseltide rT1, alstotide as1, kalata B1, and MCoT-1, are known to be cell membrane-permeable (3,24,(27)(28)(29)(30). However, these peptides are generally cationic as compared with the anionic roseltide rT7. To determine cellular roseltide rT7 uptake, flow cytometry and live-cell confocal microscopy were used. Fig. 5A shows the live-cell images of A549 cells after a 2-h incubation with 1 M TAMRA-rT7. The confocal images revealed that TAMRA-rT7 is internalized, is distributed throughout the cell, and enters the nucleus. To provide a quantitative measurement for TAMRA-rT7 cellular uptake in A549 cells, we used flow cytometry, which showed an increase in fluorescence intensity of the cell population after incubation with 1 M TAMRA-rT7 and reached a plateau by 2 h (Fig. 5B).
To determine whether the cellular uptake is mediated by endocytosis, TAMRA-rT7 was incubated with A549 cells at 4°C for 1 h. Fig. 5C shows that the cellular uptake of TAMRA-rT7 at 4°C was substantially reduced as compared with those at 37°C. Additionally, cellular uptake of TAMRA-rT7 was reduced by endocytosis inhibitors, such as dynasore and ethyl-isopropyl amiloride (Fig. S9). Roseltide rT7 did not cause significant membranolytic effects in A549 up to 100 M as measured by lactate dehydrogenase leakage assay (Fig. S10). These results suggested that the cellular uptake of TAMRA-rT7, at least in part, involves energy-dependent endocytosis.

Roseltide rT7 as a human 20S proteasome inhibitor
To show that roseltide rT7 is a human proteasome inhibitor, we compared roseltide rT7 with a commercially available covalent proteasome inhibitor, MG132, and confirmed that they both promoted the accumulation of ubiquitinated proteins using Western blotting in A549 cells (Fig. 6A). Furthermore, using a commercially available purified human 20S proteasome, we showed that roseltide rT7 inhibited chymotrypsinlike human 20S proteasome activities with an IC 50 of 3.3 M, whereas the commercially available covalent proteasome inhibitor, MG132, has an IC 50 of 0.13 M (Fig. 6B). In contrast, the negative control, roseltide rT1, did not inhibit chymotryspinlike human 20S proteasome activities at concentrations up to 10 M (Fig. S11). Roseltide rT7 also weakly inhibited caspaselike 20S proteasome activity with a 20% inhibition at 10 M (Fig. S12) but not trypsin-like human 20S proteasome activity at concentrations up to 10 M (Fig. S13). Using proteasome activity probe, Me4BodipyFL-Ahx3Leu3VS, we further showed that roseltide rT7 inhibited human 20S proteasome (Fig. 6C). Using MALDI-TOF MS, we showed that roseltide rT7 is stable against human 20S proteasome-mediated cleavage (Fig. 6D).

Table 1 Structural statistics for the final 17 conformers of rT7
The statistics were selected from 100 calculated conformers according to overall energy.
To provide mechanistic insights into the proteasome inhibitory effects of roseltide rT7, several substrate-like pentapeptide analogs with the Ile-Ile-Met-Leu-Xaa sequence, in which Leu-Xaa is the scissile bond, were chemically synthesized. Our results are summarized in Fig. 6F. IIMLA, IIMLK, and IIMLP caused an inhibition of chymotrypsin-like activities with IC 50 values of 101, 129, and 140 M, respectively. Because Pro is not tolerated at the P1 position of the scissile bond, IIMPA and IIMPK were synthesized as the negative controls. Using HPLC and LC-MS analysis of pentapeptide analogs, we further showed that Ile-Ile-Met-Leu-Xaa is a proteasome substrate recognition sequence ( Fig. 6F and Figs. S14 -S17).
IB␣ is a proteasome substrate and an inhibitor for NF-B activation (31,32). Previous reports have shown that tumor necrosis factor (TNF) induced NF-B activity and inflammatory markers by promoting the proteasomal degradation of IB␣ (33)(34)(35). One of the well-studied NF-B target gene is intercellular adhesion molecule (ICAM)-1 (36 -38). Fig. 7A shows that pretreatment with roseltide rT7 or MG132 reduced TNF-mediated down-regulation of IB␣ expression in A549 cells. Furthermore, to determine the effects of roseltide rT7 on TNF-mediated inflammation, we examined its effect on TNF-induced ICAM-1 expression in A549 cells. Fig. 7 (B and C) shows that TNF caused down-regulation of mRNA and protein expression of ICAM-1. Pretreatment with roseltide rT7 or MG132 attenuated TNF-induced ICAM-1 mRNA and protein expressions.

Discussion
The present study identified the 34-residue roseltide rT7 as a cell-penetrating 20S-proteasome inhibitor. To the best of our

Disulfide-rich proteasome inhibitor
knowledge, this is the first report of a cysteine-rich peptide inhibitor against the chymotryspin-like human 20S proteasome. A previous report described another plant-derived cysteine-rich peptide, MCTI-1 (Momordica charantia trypsin inhibitor), which appeared to cross-react against the trypsinlike human 20S proteasome (39). However, roseltide rT7 does not share sequence homology with MCTI-1.
Roseltide rT7 belongs to the family of the non-chitin-binding 6C-HLPs (2). 6C-HLPs have an evolutionally conserved cysteine motif in which a tandemly connecting CC at the Cys III and IV positions forms a cystine-knot disulfide connectivity (Cys I-IV, II-V, and III-VI) (2,3,23,26). Structurally stabilized by its cystine-knot scaffold, roseltide rT7 displays exceptional stability against heat denaturation and proteolytic degradation, a desirable feature that is highly sought after by pharmaceutical industries for biologics. In contrast, removing all three disulfide constraints, represented by the S-alkylated rT7, reduces its proteolytic stability by Ͼ100-fold.
Contrary to the general perception that cell-penetrating peptides need to be cationic (40 -42), this study by using live-cell confocal microscopy and flow cytometry showed that the anionic roseltide rT7 internalizes into living cells partly through an energy-dependent endocytosis mechanism. The cell-penetrating effect of roseltide rT7 is likely contributed by two major factors. First, the cystine scaffold is responsible for an "insideout" structural feature of a CRP (25). This phenomenon is caused by the dense cystine core, which forces the hydrophobic side chains to point outwards. As such, within the same molecule, CRPs can display both hydrophilic and hydrophobic surface patches. The NMR structure of roseltide rT7 revealed the presence of this "inside-out" feature with all the side chains pointing outwards. Because of this, roseltide rT7 possesses both hydrophilic and hydrophobic surface properties (Fig. S18).
Second, roseltide rT7 contains four prolines within its 34 residues, three of which are embedded in loop 4 and account for 25% of the residues in this loop. The hydrophobic surface region of roseltide rT7 located at the Pro-rich region together with the Arg residue at loop 4 creates an exposed cationic hydrophobic surface that could be important for driving its cell penetration. This finding is similar to the observations from our and others previous studies on bactenecin-7 and pyrrhocoricin into which cationic Pro-rich peptides can penetrate cells (43,44).
The cross-linked disulfide bonds of 6C-HLPs resulted in a topological resemblance to a four-leaf clover consisting of four intercysteine loops (2,3,23). In 6C-HLP, these loops are evolvable and generally differ from each other in sequence and size. Importantly, they are often functional sites contributing to their multiple biological activities (25,45,46). This feature has also been extensively exploited for the design of hyperstable and multitargeting peptide biologics for extracellular and intracellular targets (47)(48)(49)(50)(51)(52)(53)(54). Roseltide rT7 has a long intercysteine loop 4 with an amino acid sequence of IIMLPTWPPRYV that is absent in roseltide rT1. The IIML motif in roseltide rT7 showed sequence similarity to the tetrapeptide guiding sequence of the ␣Ј,␤Ј-epoxyketone-type proteasome inhibitors epoxomicin and carmaphycin. We showed that roseltide rT7 inhibits chymotrypsin-like human 20S proteasome activities, but roseltide rT1, missing the IIML motif, does not display any inhibition. Our results suggest that loop 4 of roseltide rT7 is the reactivesite loop for its effects against human 20S proteasome.

Disulfide-rich proteasome inhibitor
Epoxomicin and carmaphycin are the ␣Ј,␤Ј-epoxyketonetype covalent 20S proteasome inhibitors derived from microorganisms (5,6,55). They have two important features of a tetrapeptide guiding sequence and a C-terminal epoxyketone group (4, 20, 56 -58). The tetrapeptide guiding sequence determines its binding specificity through the formation of an antiparallel ␤ sheet. Previous studies have shown that the P1 and P3 positions in the ␤5 proteasome subunit are important for its binding specificity (20, 56, 57, 59 -61). The P1 position favors Leu, whereas the P3 position prefers a hydrophobic amino acid residue (61). In contrast, the P2 position does not participate in the interaction of the binding pocket, and the P4 position accepts a wide range of incoming groups with preference for hydrophobic moieties (20, 62). The C-terminal epoxyketone group serves chymotrypsin-like proteasome activities using a Proteasome-Glo TM chymotrypsin-using assay kit. The proteasome cleavage of P1Ј-P1 site was determined using RP-HPLC and LC-MS analysis. P(X) refers to N-terminal to the substrate cleavage site. P(X)Ј refers to C-terminal to the substrate cleavage site.

Disulfide-rich proteasome inhibitor
as a "warhead" that links covalently to the catalytic threonine residue of the proteasome subunit (20). Because the IIML motif of roseltide rT7 is homologous to the tetrapeptide-guiding sequence, we envisioned that it could bind to the ␤5 proteasome subunit. Structure-activity studies using synthetic pentapeptides with the Ile-Ile-Met-Leu-Xaa sequence, in which Leu-Xaa is the scissile bond, supported the interaction of the IIML motif with the proteasome in a substrate-like manner, consistent with previous reports (5, 6, 63).
The biological consequences of proteasome inhibition are numerous. Studies on proteasome targeting have led to new promising treatments for a variety of diseases including multiple myeloma, cancer, rheumatoid arthritis, and other inflammatory diseases (64 -66). In particular, three protea-some inhibitors have been granted fast-track Food and Drug Administration approval for treating multiple myeloma. In our study, roseltide rT7 accumulated polyubiquitinated proteins, which is indicative of its intracellular targeting proteasome inhibitory activities. A classical biological effect of proteasome inhibition is anti-inflammation, which results from the accumulation of IB␣ that is a natural inhibitor for NF-B activation (31,(33)(34)(35). TNF is a known pro-inflammatory cytokine that promotes IB␣ phosphorylation to initiate its ubiquitination for proteasomal degradation. We proposed that roseltide rT7, being a cell-penetrating proteasome inhibitor, inhibits TNF-mediated IB␣ proteasomal degradation, which in turn suppresses nuclear translocation of NF-B subunits and its activation for protein synthesis  (Fig. 7D). Activation of NF-B has been reported to up-regulate ICAM-1, a cell-surface adhesion molecule that promotes the recruitment of immune cells to trigger inflammatory response (36 -38). Using ICAM-1, a TNF-inducible cell adhesion molecule as a marker, we showed that roseltide rT7 accumulated IB␣ and suppressed the up-regulated expressions of both ICAM-1 mRNA and protein in TNF-treated A549 cells, similar to MG132 and other proteasome inhibitors (67).

Disulfide-rich proteasome inhibitor
In conclusion, this study describes the discovery of roseltide rT7 as a novel cell-penetrating hyperstable proteasome inhibitor. Collectively, our results suggested that the 12-residue hydrophobic and proline-rich loop 4 of roseltide rT7 is responsible for its functional properties. Also, our studies of the Ile-Ile-Met-Leu-Xaa sequence motif provided mechanistic insights into its inhibitory activities. Most importantly, roseltide rT7 represents a conceptual lead for the development of other protease inhibitors through epitope grafting of substrate sequence into structurally stable scaffolds.

Materials
All chemicals and solvents were purchased from Sigma-Aldrich and Thermo Fisher Scientific unless specified otherwise.

Extraction and purification of roseltide rT7
Dried calyces (1 kg) of H. sabdariffa were extracted with water and centrifuged at 9000 rpm for 10 min at 4°C (Beckman Coulter). The supernatant was filtered, and the filtrate was loaded onto a C18 flash column (Grace Davison) and eluted with 60% ethanol, 0.01% TFA. The eluted fractions were loaded onto an SP-Sepharose resin column (GE Healthcare) and eluted with 1 M NaCl (pH 3.0) followed by ultrafiltration (ViVaflow 200; 2000 molecular weight cutoff hydrostat). Further purification was done by RP-HPLC (Shimadzu, Japan). A linear gradient of mobile phase A (0.05% TFA in H 2 O) and mobile phase B (0.05% TFA in acetonitrile (ACN)) was used with the C18 column (250 ϫ 22 mm, 5 m, 300 Å; Grace Davison). MALDI-TOF MS was used to identify the presence of roseltide rT7 in the eluted fractions.

S-Reduction and S-Alkylation
Purified roseltide rT7 (1 mg/ml) was S-reduced by 10 mM DTT in ammonium bicarbonate buffer (25 mM) pH 8 at 37°C for 30 min followed by S-alkylation with 60 mM of iodoacetamide at 37°C for 45 min. MALDI-TOF MS was used to confirm the mass shift after S-reduction and S-alkylation.

De novo peptide sequencing
S-Alkylated roseltide rT7 (1 mg/ml) was digested with Lys-C, Arg-C, or Glu-C in 5:1 (w/w) ratio in ammonium bicarbonate buffer (25 mM), pH 8, at 37°C for 10 min. The digested peptide fragments were then analyzed by MALDI-TOF MS followed by MS/MS (AB SCIEX 4800 MALDI-TOF/TOF). De novo peptide sequencing was performed using the b and y ions.

Chemical synthesis of linear pentapeptide analogs
The linear pentapeptide with the amino acid sequences IIMLP, IIMLA, IIMLK, IIMPA, and IIMPK were synthesized by Fmoc-based solid-phase peptide synthesis on rink-amide Pro-Tide resin (LL) using an automatic microwave-assisted peptide synthesizer. The linear precursor peptide was cleaved in a mixture consisting of 92.5% TFA, 2.5% H 2 O, 2.5% 1,2-ethanedithiol, and 2.5% triisopropylsilane at room temperature for 2 h and precipitated using diethyl ether. The pentapeptide was purified by preparative HPLC and identified using an Orbitrap Elite mass spectrometer (Thermo Scientific Inc.) coupled with a Dionex UltiMate 3000 UHPLC system (Thermo Scientific Inc.) and RP-HPLC (Table S2 and Figs. S19 -S28).

NMR spectroscopy and structural determination
All NMR experiments were conducted at 25°C on a BRUKER Avance 800 NMR spectrometer with a cryogenic probe. The concentrations of natural and synthetic roseltide rT7 were ϳ1 mM and contained 5% D 2 O and 95% H 2 O. For 1 H-1 H 2D TOCSY and NOESY, the mixing times were 80 and 200 ms, respectively. The spectral width was 12 ppm for both dimensions. The NMR spectra were processed using NMRPipe software (68). All data analyses were performed using Sparky software based on the 2D NOESY and TOCSY experiment (69). The proton chemical shift assignments for each amino acid residue were achieved by 2D 1 H-1 H TOCSY and 1 H-1 H NOESY, whereas the proton-proton distances restraints were obtained from 2D 1 H-1 H NOESY based on the intensities of the NOE cross-peaks.
The solution structures of roseltides were calculated using CNSsolve 1.3 software (70). The proton-proton distance Disulfide-rich proteasome inhibitor restraints and three disulfide bond restraints were employed in a standard simulated annealing protocol. The distance restraints were divided into three classes based on the intensities of NOE cross-peaks: 1) strong: 1.8 Ͻ d Ͻ 2.9 Å; 2) medium: 1.8 Ͻ d Ͻ 3.5 Å; and 3) weak: 1.8 A Ͻ d Ͻ 5 Å. A total of 100 structures were calculated, and the 17 lowest energy structures were chosen for data statistics and presentation. The structure was verified using the PROCHECK program and presented using Chimera version 1.6.2 (71). The PDB code for this deposition is 6KLM. The BMRB code for this deposition is 36273.

Cell cultures
Human alveolar basal epithelial adenocarcinoma A549 cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum and 100 units/ml of penicillin and streptomycin.

Cellular uptake analyses by flow cytometry
To study cellular TAMRA-rT7 uptake using flow cytometry, the cells were incubated with 1 M TAMRA-rT7 in serum-free medium at 37°C. Following incubation, the cells were harvested and collected by centrifugation at 500 ϫ g for 5 min. To quench extracellular fluorescence, the cells were treated with 150 g/ml of trypan blue, and the samples were analyzed by flow cytometry. A total of 10,000 cells was analyzed using a BD LSRFortessa TM X-20 flow cytometer. For temperature-dependent uptake studies, A549 cells were incubated at 4°C for 30 min prior to incubation with TAMRA-rT7 for 1 h at 4°C. For endocytosis inhibitor studies, A549 cells were pretreated with endocytosis inhibitors, including dynasore (50 M), ethylisopropylamiloride (50 M), and nystatin (50 g/ml) for 30 min followed by incubation with TAMRA-rT7 for 1 h at 37°C.

Confocal microscopy analysis
To examine the intracellular TARMA-rT7 distribution, the cells were seeded on an 8-well chamber slide (Ibidi). Prior to incubation with TARMA-rT7, the cells were stained with Hoechst 333241. TARMA-rT7 was incubated on cells in phenol red-free and serum-free medium at 37°C. The slides were washed gently with PBS three times, and the medium was replaced prior to imaging. The slides were observed using a Zeiss LSM 710 confocal microscope.

Human 20S proteasome assay using proteasome activity probe
Me4BodipyFL-Ahx3Leu3VS (BostonBiochem) was used to probe proteasome activity. Briefly, purified human 20S protea-some (100 pM) was incubated with MG132 or roseltide rT7 for 1 h at 37°C. 500 nM Me4BodipyFL-Ahx3Leu3VS was then added for 30 min at 37°C. The resultant mixture was mixed with 6ϫ loading dye with 2-mercaptoethanol and heated for 10 min at 85°C. The samples were resolved using a 12% SDS-PAGE at 100 V constant for 120 min. In-gel fluorescence detection was performed using the ChemiDoc MP system (Bio-Rad).

Western blotting analysis
A549 cells treated with roseltide rT7 or MG132 were harvested and lysed in CelLytic TM M lysis buffer supplemented with both protease and phosphatase inhibitor cocktails on ice with frequent agitation for 30 min. The cell homogenates were centrifuged at 12,000 rpm for 30 min at 4°C, and the supernatants were collected. Protein concentrations were determined using bicinchoninic acid (BCA) reagent. Total protein (30 g) mixed with 6ϫ loading dye with 2-mercaptoethanol was heated for 10 min at 85°C. The resultant mixture was centrifuged at 12,000 ϫ g for 1 min and resolved using 10% SDS-PAGE at 100 V constant for 120 min. Blot transfer was performed onto a polyvinylidene difluoride membrane (GE Healthcare) at 250 mA for 120 min on ice. The blot was blocked with 5% BSA in TBS and Tween 20 (TBST) before being incubated overnight at 4°C with a mouse anti-ubiquitin (P4D1) (1:200 in 5% BSA-TBST; Santa Cruz Biotechnology), rabbit anti-IB␣ (1:500 in 5% TBST; GeneTex), mouse anti-glyceraldehyde-3-phosphate dehydrogenase (1:10,000 in 5% BSA-TBST; Avivas Systems Biology), or mouse anti-B-actin antibodies (1:10,000 in 5% TBST; Merck). After overnight incubation, the membrane was washed with TBST at room temperature three times for 10 min each. The blot was then incubated with secondary mouse or rabbit anti-horseradish peroxidase (1:10,000 in 5% BSA-TBST; Cell Signaling Technology) for 1 h at room temperature. The blot was washed five times for 10 min each with TBST at room temperature before addition of chemiluminescence substrate (Advansta) and exposure on X-ray film (Fujifilm, Japan).

Flow cytometric analyses for ICAM-1 expressions
Surface expressions of ICAM-1 were determined by flow cytometry. Briefly, A549 cells were exposed to TNF (50 ng/ml) (Peprotech) for 6 h with or without 2 h roseltide rT7 or MG132 pretreatment. The cells were harvested and collected by centrifugation at 500 ϫ g for 5 min. The cell pellet was stained with anti-ICAM-1-PE (ebioscience) in serum-containing medium for 30 min on ice. The pellet was then washed three times, and the samples were subjected to flow cytometric analyses. Ten thousand cells were analyzed using the BD LSRFortessa TM X-20 flow cytometer. The results were analyzed by FlowJo vX.0.7 (FlowJo) and presented as mean fluorescence intensity.

Peptide stability assay
Purified roseltide rT7 (0.1 M) and S-alkylated rT7 (iodoacetamido-) were tested in the following stability studies with respective buffer conditions: pepsin stability, Pepsin (Roche Applied Science) of 50:1 (w/v) ratio in 0.2 M HCl at 37°C; Pronase stability, Pronase (0.2 mg/ml; Roche Applied Science) in PBS at 37°C; neutrophil elastase stability, human neutrophil elastase (0.05 mg/ml; Molecular Innovations) in PBS at 37°C; trypsin stability, trypsin (0.2 mg/ml; Sigma-Aldrich) in PBS at 37°C; and analysis for stability assays, all samples collected at various time points from the stability assays were analyzed by RP-HPLC with a linear gradient of mobile phase A (0.05% TFA/ H 2 O) and mobile phase B (0.05% TFA/ACN) on aeris peptide XB-C18 column (Phenomenex). The resulting peaks were collected and identified by MALDI-TOF MS. The results were expressed as percentage of initial concentration using the peak area of the HPLC profile.

Circular dichroism spectroscopy
Far-UV CD spectra were recorded using a Chirascan TM CD spectrometer (Applied Photophysics, Leatherhead, UK). CD Spectra were acquired between 190 and 280 nm from 20 to 90°C using a 10-mm path-length quartz cuvette, a 1-nm spectral bandwidth, and a 0.5-nm step size.

20S proteasome cleavage of pentapeptide
20S proteasome cleavage of pentapeptide was performed in a 50-l reaction mixture containing 10 mM HEPES (pH 7.6), human 20S proteasome (5 nM), and pentapeptide (50 M). Each reaction was performed at 37°C. The samples were separated using RP-HPLC. The identity of the HPLC peak was analyzed using an Orbitrap Elite mass spectrometer (Thermo Scientific Inc.) coupled with a Dionex UltiMate 3000 UHPLC system (Thermo Scientific Inc.).

Statistical analyses
Statistical comparisons were performed using GraphPad version 6.0d. The data were analyzed by one-way analysis of variance followed by Newman-Keuls post hoc test. The data are expressed as the means Ϯ S.E. p Ͻ 0.05 was considered statistically significant.