Phosphorylation of Ubc9 by Cdk1 Enhances SUMOylation Activity

Increasing evidence has pointed to an important role of SUMOylation in cell cycle regulation, especially for M phase. In the current studies, we have obtained evidence through in vitro studies that the master M phase regulator CDK1/cyclin B kinase phosphorylates the SUMOylation machinery component Ubc9, leading to its enhanced SUMOylation activity. First, we show that CDK1/cyclin B, but not many other cell cycle kinases such as CDK2/cyclin E, ERK1, ERK2, PKA and JNK2/SAPK1, specifically enhances SUMOylation activity. Second, CDK1/cyclin B phosphorylates the SUMOylation machinery component Ubc9, but not SAE1/SAE2 or SUMO1. Third, CDK1/cyclin B-phosphorylated Ubc9 exhibits increased SUMOylation activity and elevated accumulation of the Ubc9-SUMO1 thioester conjugate. Fourth, CDK1/cyclin B enhances SUMOylation activity through phosphorylation of Ubc9 at serine 71. These studies demonstrate for the first time that the cell cycle-specific kinase CDK1/cyclin B phosphorylates a SUMOylation machinery component to increase its overall SUMOylation activity, suggesting that SUMOylation is part of the cell cycle program orchestrated by CDK1 through Ubc9.

Studies of a temperature-sensitive mutant in budding yeast have demonstrated that Ubc9 is required for progression through mitosis [4]. Ubc9 (Hus5) mutants in fission yeast also display defects during chromosome segregation and reduced cellular growth [5,6]. Loss of Ubc9 in mouse embryos causes chromosome mis-segregation and loss of nuclear integrity [6]. Studies in Zebra fish have also suggested an in vivo requirement of Ubc9 for G2/M transition and/or progression through mitosis during vertebrate organogenesis [7].
Studies of other SUMOylation machinery components have suggested a similar role of SUMOylation in cell cycle progression through mitosis. In budding yeast, temperature sensitive Smt3 (budding yeast SUMO) mutants show defects in chromosome segregation [8]. In fission yeast, pmt3D cells also show defects in mitotic chromosome structure or segregation errors [9]. These defects include high frequency loss of mini-chromosomes and a cut (cell untimely torn) phenotype. It has also been demonstrated that SUMO-1 targets RanGAP1 to kinetochores and mitotic spindles [10]. In addition to Ubc9 and SUMO, SUMO proteases have also been shown to play an important role in mitosis as budding yeast ulp1 mutants show cell cycle delays at the G 2 /M boundary and elevated chromosome mis-segregation [11]. In addition to the genetic analysis, biochemical studies have also demonstrated that key chromosome structural components (i.e. condensin and cohesin complexes, and DNA topoisomerase II) are mitotic SUMOylation targets [2,[12][13][14][15]. Condensin and cohesin contain Structural Maintenance of Chromosomes (SMC) proteins [16]. SMC proteins such as Smc1p, Smc3p, Smc5p have been identified as SUMOylation targets in budding yeast via proteomic screens [12,14]. topoisomerase II has been shown to be a major target of SUMOylation in both budding yeast and vertebrates [2,13,15]. SUMOylation of topoisomerase II has been shown to play a key role in centrosome cohesion and chromosome segregation [2,13]. Expression of the SUMO-null form of Aurora B also results in abnormal chromosome segregation and cytokinesis failure [3].
Cyclin-dependent kinases (Cdks) are master regulators for cell cycle progression [17]. Cdks associate with distinct cyclins to activate its phosphorylation activity. At least 4 interphase Cdks (Cdk2, Cdk3, Cdk4 and Cdk6) are required to drive cells through interphase, while CDK1 functions primarily to drive mitosis in mammalian cells. Cyclin B is known to associate with CDK1 to drive G2/M transition and progression through mitosis, as evidenced by its involvement in chromosomal condensation, nuclear envelope breakdown, centrosome separation and mitotic exit [18][19][20][21]. The central role of CDK1/cyclin B in regulating mitosis has prompted us to investigate a possible link between CDK1/cyclin B and SUMOylation.
In the current studies, we have studied the possible interplay between the master mitotic kinase CDK1/cyclin B and SUMOylation. We show that CDK1/cyclin B, but not many other cell cycle kinases, enhances sumoylation activity. In addition, CDK1/ cyclin B specifically phosphorylates Ubc9 at serine 71, but not other SUMO machinery components such as SAE1/SAE2 and SUMO1. Most importantly, CDK1/cyclin B-mediated phosphorylation of Ubc9 leads to elevated SUMOylation activity and stimulates the formation of the Ubc9-SUMO1 thioester conjugate. In the aggregate, these results suggest that Ubc9 phosphorylation and hence elevated SUMOylation may be part of the cell cycle regulatory program orchestrated by CDK1/cyclin B.

CDK1/cyclin B Specifically Phosphorylates the SUMOylation Machinery Component Ubc9
The stimulatory activity of CDK1/cyclin B on in vitro SUMOylation could possibly be related to its kinase activity. To test this possibility, the CDK1/cyclin B-mediated phosphorylation of the SUMO machinery components (SAE1/SAE2, SUMO1 and/or Ubc9) was investigated in vitro in the presence of [c-32 P] ATP (see Methods and Materials). As shown in Figure 2A, CDK1/cyclin B was shown to cause an observable, but relatively weak, increase in the radio-labeling of His 6 -SAE1 and His 6 -SAE2 ( Figure 2A). However, the radio-labeling exhibited a low background in the absence of CDK1/cyclin B. CDK1/cyclin B did not cause any detectable phosphorylation of His 6 -SUMO1 in the same CDK1/cyclin B concentration range ( Figure 2B). By contrast, CDK1/cyclin B caused a concentration-dependent increase of phosphorylated His 6 -Ubc9 as evidenced by the appearance of a radio-labeled band (revealed by autoradiography) with the same mobility as His 6 -Ubc9 band (as revealed by the corresponding coomassie-stained band) ( Figure 2C). On the other hand, SUMOylatable substrate GST-hTOP1 (110-125) AA was unaffected by the stimulatory effect of CDK1/cyclin B ( Figure 2D). Histone H1, a known CDKs target protein, was strongly phosphorylated by CDK1/cyclin B, illustrating the normal activity of CDK1/cyclin B ( Figure S2; Text S1). Collectively, these observations suggest that Ubc9 is the only SUMO machinery component to be significantly phosphorylated by CDK1/cyclin B.

CDK1/cyclin B-phosphorylated Ubc9 Exhibits Elevated SUMOylation Activity
To prove that CDK1/cyclin B-mediated phosphorylation of Ubc9 is responsible for the observed stimulatory effect of CDK1/ cyclin B on SUMOylation in vitro, His 6 -Ubc9 was pretreated with GST-CDK1/cyclin B in the presence or absence of ATP. After pretreatment, the reaction mixtures were passed through glutathione sepharose three times to remove GST-tagged CDK1/cyclin B ( Figure 3A). The flow-through fraction 3 (FT3) of phosphorylated and non-phosphorylated His 6 -Ubc9 were subjected to in vitro SUMOylation assay. As shown in Figure 3B, phosphorylated His 6 -Ubc9 exhibited elevated (about 2-fold) SUMOylation activity as compared to non-phosphorylated His 6 -Ubc9 ( Figure 3B; upper panel). The loading amount of both forms of Ubc9s were relatively equal ( Figure 3B; lower panel). These results suggest that the SUMOylation-stimulatory activity of CDK1/cyclin B is at least in part mediated through its phosphorylation of Ubc9.

CDK1/cyclin B upregulates SUMOylation by Increasing
Thioester Bond Formation between Phosphorylated Ubc9 and SUMO1 but not between SAE1/SAE2 and SUMO1 SUMO1 conjugates to E1 activating enzyme (SAE1/SAE2) through thioester bond linkage; whereas, SUMO1 conjugates Ubc9 through thioester and isopeptide bond linkage in the SUMOylation pathway. To examine whether CDK1/cyclin B stimulates SU-MOylation through SUMO1 activation of SAE1/SAE2 (in thioester linkage), the SAE1/His 6 -SAE2 and His 6 -SUMO1 were mixed together with increased concentration of CDK1/cyclin B. Result showed that increasing concentration of CDK1/cyclin B did not alter the thioester bond formation between His 6 -SUMO1 and SAE1/His 6 -SAE2 ( Figure 4A), suggesting that the activation of SUMO1 by SAE1/SAE2 is irrelevant to CDK1/cyclin B treatment. To examine whether CDK1/cyclin B increases SUMOylation through conjugational activity of phosphorylated Ubc9 (in thioester linkage). His 6 -Ubc9, SAE1/His 6 -SAE2 and His 6 -SUMO1 were mixed together with increased concentration of GST-CDK1/cyclin B in the presence or absence of 200 mM DTT. In the non-reduced condition, three different molecular species containing SUMO conjugations in the form of isopeptide and thioester bond were observed. The summation of thioester and isopeptide bond conjugates was greater than the isopeptide bond conjugates as shown in Figure 4C. Moreover, the intensity of the summation of thioester and isopeptide bond conjugates was strongly enhanced in a concentration-dependent manner by increased concentration of CDK1/cyclin B ( Figure 4B). However, due to isopeptide bond formation was not affected by CDK1/cyclin B as shown in Figure 4C, therefore, we conclude that thioester bond conjugates was enhanced by CDK1/cyclin B. In the reducing condition, an observation of at least two different molecular species containing either one or two SUMO conjugations in the form of isopeptide bond were observed. This observation agrees with the report that Ubc9 contains multiple SUMO conjugation sites [24,25]. Moreover, the isopeptide bond conjugations were not enhanced by the increased concentration of CDK1/cyclin B, suggesting that CDK1/ cyclin B does not affect the isopeptide bond formation between phosphorylated His 6 -Ubc9 and His 6 -SUMO1 ( Figure 4C). Altogether, results strongly suggest that CDK1/cyclin B affects Ubc9 function by phosphorylating Ubc9, therefore facilitates and increases thioester bond formation between phosphorylated Ubc9 and SUMO1, ultimately upregulates SUMOylation activity of substrate protein. Additionally, CDK1/cyclin B neither phosphorylates SAE1/SAE2 nor enhances thioester bond formation between SUMO1 and SAE1/SAE2 enzyme.

CDK1/cyclin B-phosphorylated Ubc9 Stimulates Multi-SUMOylation of Full-length Human Topoisomerase I
Full-length human topoisomerase I (hTOP1), a cell cycleregulated protein, contains multiple SUMOylation sites at its Nterminus [25,22,26]. Multi-SUMOylation of hTOP1 at its Nterminus occurs both in vivo and in vitro [25,22,26]. hTOP1 also contains multiple phosphorylation sites at its N-terminus including the Ser-112 phosphorylation site that is phosphorylated by CDK1/cyclin B [23]. We therefore sought to determine whether CDK1/cyclin B stimulates multi-SUMOylation of hTOP1 in vitro and whether CDK1/cyclin B-phosphorylated Ubc9 is responsible for the stimulation. The substrate used in the in vitro system is hTOP1. As shown in Figure 5A, CDK1/cyclin B stimulated the formation of multiple high molecular weight species of hTOP1, representing hTOP1 conjugates with one, two and three conjugated SUMO1, suggesting that CDK1/cyclin B stimulates multi-SUMOylation of hTOP1 ( Figure 5A). We have also investigated the effect of CDK1/cyclin B-phosphorylated Ubc9 on multi-SUMOylation of hTOP1 using GST-hTOP1  as substrate. As shown in left panel of Figure 5B, unphosphorylated Ubc9 stimulated multi-SUMOylation of GST-hTOP1 1-200 as compared to the control (no Ubc9). Importantly, multi-SUMOy- lation of GST-hTOP1  was further stimulated when Ubc9 (E2) was replaced with CDK1/cyclin B-phosphorylated Ubc9 (P-E2) ( Figure 5B; left panel). Higher molecular weight SUMO1modified species of GST-hTOP1  were observed together with a concomitant decrease in the level of unmodified GST-hTOP1   (Figure 5B; left panel). Because the intensity of Ubc9 and phosphorylated Ubc9 after immunoblotting with anti-Ubc9 antibody is almost the same, which is consistent with Biorad protein assay, thus we use Ubc9 and phosphorylated Ubc9 for SUMOylation assay at a final concentration of 6 mM ( Figure 5B; right panel). These results suggest that CDK1/cyclin Bstimulated multi-SUMOylation of hTOP1 is also likely to be mediated through phosphorylated Ubc9, which provides additional evidence for the possible physiological relevance of CDK1/ cyclin B-mediated phosphorylation of Ubc9 in stimulation of SUMOylation. In order to identify the phosphorylation sites of Ubc9, in vitro phosphorylated as well as non-phosphorylated Ubc9 was first tryptic digested and phosphopeptides were enriched by immobilized metal affinity chromatography (IMAC). Enriched phosphopeptides were then subjected to liquid chromatography coupled with tandem mass spectrometry (LC-MS/MS). As shown in Figure 6, the MS/MS analysis revealed a phosphorylation site at serine 71 in a peptide corresponding to residues MLFKDDYPSpSPPK of Ubc9, whereas no phosphorylation was detected at serine 70 ( Figure 6). The detection of b 9 ions, with no neutral loss, indicates that the site of phosphorylation is Serine 71. Phosphorylation at 71 is supported through the detection of y 4 -98 and b 10 -98 ions. In addition the phosphate fingerprint of the neutral loss (owing to H 3 PO 4 ) from the parent ion is also a positive signal of phosphorylation. In comparison to the negative control, no phosphorylated amino acid residue was detected ( Figure S3). Overall, these results suggest that Cdk1/ cyclin B mediates the phosphorylation of Ubc9 at serine 71.

CDK1/cyclin B Enhances SUMOylation Activity through Wild Type Ubc9 but not Mutant Ubc9
To determine whether serine 71 plays a role in regulating SUMOylation of Ubc9, serine 70 and serine 71 were mutated to alanine (S70A/S71A), followed by assaying SUMOylation in vitro. As predicted, the SUMOylation activity driven by wild-type Ubc9 was enhanced in the presence of CDK1/cyclin B (Figure 7). By contrast, SUMOylation activity driven by S70A/S71A mutant Ubc9 was not affected by the addition of CDK1/cyclin B, suggesting that CDK1/cyclin B regulates SUMOylation activity through phosphorylating serine 71 of Ubc9.

Discussion
The crosstalk between cell cycle and SUMOylation has been widely studied. Our demonstration that CDK1/cyclin B stimulates SUMOylation in vitro could indicate a potential molecular link for the crosstalk. CDK1/cyclin B could stimulate SUMOylation through phosphorylation of SUMO target proteins, SUMO machinery components, or a combination of both. In our studies, CDK1/cyclin B was shown to phosphorylate Ubc9 but not SUMO1 or SAE1/SAE2 (Figure 2). Our results agree in part with the findings that SUMO machinery components can be phosphorylated and SUMOylated in vivo [24,27]. More importantly, phosphorylation of Ubc9 by CDK1/cyclin B, as demonstrated in our studies, is a novel finding which could have broad implications in cell cycle regulation. Specifically, phosphorylation   and ATP at 37uC for 30 min, 60 min and 120 min. The resulting SUMOylation reaction mixtures were run on a 8% SDS-PAGE, followed by immunoblotting with anti-topoisomerase I antibody. The amount of phosphorylated Ubc9 (P-E2) and Ubc9 (E2) for SUMOylation assay in It is more difficult to determine whether CDK1/cyclin B can also stimulate SUMOylation through its phosphorylation of the SUMO substrate. Hackbarth et al suggested that human topoisomerase I contains an in vivo phosphorylation site (Ser-112) regulated by CDK1/cyclin B [23]. In our studies, we have confirmed that CDK1/cyclin B indeed phosphorylates GST-hTOP1 (110-125) but not GST-hTOP1 (110-125) AA mutant (Ser 111 and Ser 112 converted to Ala 111 and Ala 112). However, CDK1/cyclin B stimulates SUMOylation of both GST-hTOP1 (110-125) and GST-hTOP1 (110-125) AA mutant to the same extent. These results strongly suggest that the SUMOylationstimulatory effect of CDK1/cyclin B is due to CDK1/cyclin Bmediated phosphorylation of the SUMOylation machinery (i.e. Ubc9) rather than that of the SUMO substrate (i.e. GST-hTOP1 (110)(111)(112)(113)(114)(115)(116)(117)(118)(119)(120)(121)(122)(123)(124)(125) ) in this in vitro system. We have also performed in vitro SUMOylation assay using a more physiologically relevant substrate, GST-hTOP1  , which contains the entire Nterminus of hTOP1. The N-terminus of hTOP1, which contains multiple CDK1/cyclin B phosphorylation sites, is known to be multi-SUMOylated in vivo [22,26]. Indeed, CDK1/cyclin B stimulates multi-SUMOylation of GST-hTOP1  in our in vitro system. More importantly, CDK1/cyclin B-phosphorylated Ubc9 is much more efficient than unphosphorylated Ubc9 in stimulating multi-SUMOylation of GST-hTOP1  . These results suggest the SUMOylation-stimulatory effect of CDK1/ cyclin B is primarily due to its phosphorylation of the SUMOylation machinery (i.e. Ubc9). However, we cannot rule out the possibility that CDK1/cyclin B may also stimulate SUMOylation through its phosphorylation of GST-hTOP1  .
The function of Ubc9 is to conjugate SUMO1 received from SAE1/SAE2. The conjugation requires formation of thioester bond. According to Knipscheer et al., Ubc9 is autoSUMOylated thereby leads to target discrimination in the following SUMO modification, suggesting that Ubc9 itself serves as a substrate and forms isopeptide bond with SUMO on specific lysine [24]. This raises a question whether CDK1/cyclin B enhances thioester bond conjugational activity of Ubc9 with SUMO1 or enhances isopeptide bond linkage of SUMO1 with Ubc9 or substrate protein. Our study showed that the isopeptide bond linkage between CDK1/cyclin B phosphorylated Ubc9 and SUMO1 was  unchanged in reduced condition ( Figure 4C). Conversely, the summation of thioester and isopeptide bond formation between CDK1/cyclin B phosphorylated Ubc9 and SUMO1 was increased in a concentration-dependent manner in non-reduced condition ( Figure 4B). Because isopeptide bond formation was unaffected by CDK1/cyclin B illustrated in reduced condition, thus thioester bond formation was the only covalent bond to be increased by CDK1/cyclin B. As the amount of thioester bond formation reflects the conjugational capacity of Ubc9, our in vitro results clearly indicate that the SUMOylation activity in cell cycle is upregulated when Ubc9 is phosphorylated by CDK1/ cyclin B, in addition, the upregulation of SUMOylation activity is directed by increased conjugational activity of phosphorylated Ubc9.
The crosstalk between phosphorylation and SUMOylation has been widely studied in recent years. Notably, many SUMO substrates contain a highly conserved motif having a SUMO consensus site and a proline-directed phosphorylation site (YKxExxSP) leading to a concept of phosphorylation-dependent SUMO modification (PDSM) [28]. Thus, phosphorylation has widely been accepted to serve as a regulator for SUMOylation at the substrate level. On the other hand, recent studies have pointed to the possibility that the SUMOylation machinery components may be subject to post-translational modifications. For example, SUMO1 itself has been shown to be phosphorylated in vivo by using mass spectrometric approach and Ubc9 is itself SUMOmodified which results in alteration of SUMO target specificity [24,27]. Our current studies have demonstrated that CDK1/ cyclin B can phosphorylate the SUMOylation machinery component Ubc9, adding new evidence for regulation of SUMOylation at the level of SUMOylation machinery.
While our studies implicate Ubc9 phosphorylation and hence SUMOylation to be part of the mitosis regulatory program orchestrated by CDK1/cyclin B, SUMOylation has also been implicated in the regulation other cell cycle phases. For example, SUMO modification of retinoblastoma (Rb) leads to the release of repressive activity of Rb on the E2F transcription factor, resulting in G1/S transition [29]. p73, a member of the p53 family, is SUMO-modified via binding to PIAS-1 during the S phase [30,31]. Interestingly, CDK1, by partnering with other cyclins, is also known to regulate other cell cycle phases [32,17]. Therefore, it seems plausible that CDK1, by partnering with different cyclins, could also phosphorylate Ubc9 to enhance its SUMOylation activity that may contribute cell cycle progression at other phases of the cell cycle. Clearly, further studies are necessary to establish the role of Ubc9 phosphorylation in SUMOylation regulation during the cell cycle.

In vitro Phosphorylation Assays
The phosphorylation assay was performed in 20 ml of reaction mixtures. SAE1/His 6 -SAE2 (1 mg) was incubated with various concentrations (3 nM, 6 nM and 12 nM) of CDK1/cyclin B in the presence of [c-32 P] ATP (0.05 mCi/ml) in Protein kinase buffer provided by supplier (New England Biolabs). CDK1/cyclin B (3 nM, 6 nM and 12 nM) were also incubated with [c-32 P] ATP (0.05 mCi/ml) in protein kinase buffer in the presence or absence of His 6 -SUMO1 (1 mg), His 6 -Ubc9 (1 mg) or GST-hTOP1 (110-125) AA (1 mg). All reaction mixtures were incubated at 30uC for 30 min and boiled in SDS sample buffer for 10 min. Samples were then analyzed by SDS-PAGE followed by Coomassie blue staining and autoradiography.

Isolation of Phosphorylated Ubc9
GST-CDK1/cyclin B from Cell Signaling Technology (35 nM) was incubated with His 6 -Ubc9 (10 mg) in a 200 ml phosphorylation reaction mixture containing CDK1/cyclin B kinase buffer in the presence or absence of ATP (5 mM) at 30uC for 30 min. The reaction mixture was passed through Glutathione sepharose 4B (200 ml bed volume) to remove the GST-tagged CDK1/cyclin B from the mixture. For checking the integrity of phosphorylated Ubc9, the flow-through was analyzed by SDS-PAGE and immunoblotted with goat anti-GST (GE Healthcare) and goat anti-Ubc9 antibodies (Santa Cruz).
To examine whether thioester conjugational activity of Ubc9 is affected by CDK1/cyclin B, reaction mixture containing His 6 -Ubc9 (1 mg), His 6 -SUMO1 (1 mg) and SAE1/His 6 -SAE2 (100 ng) were incubated with increased concentration (1.2 nM, 3 nM and 6 nM) of CDK1/cyclin B in a total reaction volume of 20 ml at 37uC for 60 min. The reactions were prepared in two sets, one set of the reaction samples were reduced with DTT (200 mM) and the other set was remained in non-reduced at 40uC for overnight. The intramolecular noncovalent bonds of both reaction samples were heat-denatured for 10 min with urea (8 M) and sample buffer without ß-mercaptoethanol or DTT. The samples were resolved on SDS-PAGE and immunoblotted with goat anti-Ubc9 antibody (Santa Cruz).

In-solution Digestion and Phosphopeptide Enrichment
To identify the phosphorylation site of Ubc9 which is phosphorylated by CDK1/cyclin B, a solution-based in vitro phosphorylation of Ubc9 was initially prepared, the end-reaction of Ubc9 phosphorylation was firstly reduced with 10 mM DTT at room temperature for 60 min. Alkylation reaction was then carried out by adding 55 mM iodoacetanide at room temperature for 20 min. Reduced and alkylated Ubc9 was subsequently digested with trypsin 37uC overnight. Resulting tryptic Ubc9 phopshopeptides were subjected to phosphopeptide enrichment procedures supplied by Clontech. Enriched Ubc9 phosphopeptides were acidified TFA for further ZipTips TM desalting procedure (Millipore).

LC-ESI-MS/MS Analyses
Peptide samples were reconstituted in buffer A (0.1% FA in H2O) and analyzed on Waters Synapt G2 HDMS (Waters, Milford, MA). Samples were injected onto a 2 cm6180 mm capillary trap column and separated on a 75 mm65 cm nanoACQUITY 1.7 mm BEH C18 column using a nanoAC-QUITY Ultra-Performance LC system (Waters, Milford, MA). The column was maintained at 35uC and bound peptides were eluted with a gradient of 5-40% buffer B (buffer A, 0.1% FA in H2O; buffer B, 0.1% FA in ACN) for 30 min. The MS was operated in ESI positive V mode with a resolving power of 10000 and calibrated with a synthetic human [Glu1]-Fibrinopeptide B solution (0.5 pmol/mL, Sigma-Aldrich) delivered through the NanoLockSpray source. Data acquisition was performed using data directed analysis (DDA) method. The DDA method included one full MS scan (m/z 350-1600, 1 s/ scan) and three MS/MS scans (m/z 100-2000, 1 s/scan) performed sequentially on the three most intense ions present in the full scan mass spectrum.  Figure S3 Non-phosphorylated Ubc9 as a negative control analyzed by CID tandem mass spectra. Precursor ion m/z 508.92 (charge state +3) representing the peptide MLFKDDYPSSPPK. 0b0 and 0y0 ions series represent fragment ions containing the N-and C-termini of the peptide, relatively. (TIF)

Supporting Information
Text S1 Supporting methods. (DOC)