Kinetics of the multitasking high-affinity Win binding site of WDR5 in restricted and unrestricted conditions

Recent advances in quantitative proteomics show that WD40 proteins play a pivotal role in numerous cellular networks. Yet, they have been fairly unexplored and their physical associations with other proteins are ambiguous. A quantitative understanding of these interactions has wide-ranging signi ﬁ cance. WD40 repeat protein 5 (WDR5) interacts with all members of human SET1/MLL methyltransferases, which regulate methylation of the histone 3 lysine 4 (H3K4). Here, using real-time binding measurements in a high-throughput setting, we identi ﬁ ed the kinetic ﬁ ngerprint of transient associations between WDR5 and 14-residue WDR5 interaction (Win) motif peptides of each SET1 protein (SET1 Win )

Several research groups have previously examined interactions of WDR5 protein with Win-motif SET1 (SET1 Win ) peptides using a variety of approaches, such as analytical ultracentrifugation (AUC) [22], isothermal titration calorimetry (ITC) [21,24,65], surface plasmon resonance (SPR) [23], and X-ray crystallography [21,23,24,65]. These explorations have confirmed the high affinity of the Win binding site for different Win motif interaction partners [24,65]. In this study, we determined the kinetic fingerprint and affinities of these interactions using high-throughput optical and fluorescence approaches, which included biolayer interferometry (BLI), SPR, and steady-state fluorescence polarization (state-state FP).
WDR5 is a 334-residue protein that has a seven-bladed, WD-40 repeat-based β propeller structure surrounding a central cavity ( Figure 1A) [29,[66][67][68]. Each blade contains four anti-parallel β strands. A segment of the central cavity serves as the high-affinity Win binding site for the SET1 Win peptides ( Figure 1B; Supplementary Figures S1, S2) [21,23,24,65]. Here, we performed a systematic kinetic analysis of the interactions of WDR5 with six 14-residue SET1 Win peptides, which include an evolutionarily conserved Arg residue at P 0 (Table 1). Moreover, this Arg residue has been shown to be critical to WDR5-SET1 Win interactions (Supplementary  Table S1; Supplementary Figure S3) [22,26,69,70]. In addition, SET1 Win sequences contain a highly conserved six-residue Win motif peptide, at positions P −3 through P 2 , along with eight residues on their flanking sides, at positions P −7 through P −4 and P 3 through P 6 . These flanking residues diverge among SET1 family members, accounting for differences in the binding affinity of WDR5-SET1 Win interactions [24,65]. The highly conserved six-residue Win motif peptide has to insert into the WDR5 binding cavity to facilitate these highly specific WDR5-SET1 Win interactions [24,65]. Therefore, the choice of a 14-residue length of SET1 Win was based on the requirement of the minimal six-residue Win motif sequence, along with four residues on each flanking side. This also facilitated the interpretation of our results in light of a prior crystallographic study [24], which was conducted with an identical SET1 Win length.
Here, we employed multiple techniques to probe the effect of the surface immobilization on the kinetic fingerprint of WDR5-SET1 Win interactions as well as to establish the efficacy of each technique for the measurement of the kinetic rate constants and binding affinities. We provide a critical analysis of each approach with respect to the kinetics of the Win binding site. BLI and SPR require surface immobilization, while steady-state FP does not. On the other hand, steady-state FP is free from this limitation, but is unable to provide real-time kinetic measurements. Furthermore, the use of multiple approaches allowed us to provide quantitative and qualitative validations of our conclusions and to obtain more generalizable outcomes that were not restrained by any one approach. Finally, we also show that distinctions in the kinetic rate constants of these interactions are correlated with unique sequences on the SET1 Win peptides' flanking sides.

Protein expression and purification
Human WDR5 (UniProtKB -P61964; WDR5_HUMAN) was expressed and purified, as follows. pET3aTr vectors containing the 6×His-TEV-WDR5 sequence were transformed into Rosetta™ 2 BL21(DE3)pLysS (Novagen, Cat #71403) competent E. coli cells. Rosetta™ 2 BL21(DE3)pLysS (Novagen, Cat #71403) competent E. coli cells were grown overnight on Luria-Bertani (LB) agar carbenecillin/chloramphenicol selection plates at 37°C. An amount of 50 ml LB broth starter cultures (one per 10 L bioreactor growth) containing 50 mg/ml each of carbenecillin and chloramphenicol were inoculated with 5 colonies and grown for 3-5 h until turbid at 37°C. An amount of 10 L Luria broth bioreactors (Eppendorf BioFlo, Enfield, CT) containing 50 mg/ml each of carbenecillin and chloramphenicol, in addition to Antifoam 204 (Sigma-Aldrich, St. Louis, MO), were then inoculated at 37°C and 800 rpm with the entire turbid starter culture. When the culture attained OD 600 = 0.8, agitation was decreased to 600 rpm and the temperature was rapidly dropped to 18°C. Induction of target proteins was initiated with 100 mM Isopropyl-β-D-1-thiogalactopyranoside (IPTG; Goldbio, St. Louis, MO) at 18°C . After 12-24 h post-induction at 18°C, cells were pelleted by centrifugation at 4465×g using a J6-MI centrifuge (Beckman Coulter, Brea, CA) for 30 min at 4°C. The pellet was resuspended in 50 mM Tris-HCl ( pH  7.4), 300 mM NaCl, 10 mM βME, and 20 mM imidazole. The resuspended cells were frozen on dry ice and stored at −80°C until purification. Cell pellets were lysed with a Qsonica Sonicator Q700 (FisherBrand, Pittsburg, PA) on ice. An amount of 5 L of thawed, resuspended pellets were resuspended in 160 ml of 50 mM Tris-HCl ( pH 7.4), 300 mM NaCl, 3 mM DTT, and 30 mM imidazole (lysis/Ni-NTA Buffer A/dialysis buffer) containing an additional 200 ml of phenylmethylsulfonyl fluoride (PMSF; Sigma-Aldrich) and two Pierce™ Protease Inhibitor Mini Tablets (EDTA-free; Thermo Fisher Scientific, Waltham, MA). The cell suspensions were stirred at 4°C for 15 min until homogeneous and then sonicated for 10 min at 90% power (2 s on, 4 s off ). Lysate was then centrifuged for 1 h at 4465×g and 4°C, and the supernatant was stored at 2-8°C for purification. A Ni-NTA purification process was followed. A Kontes 25 × 200 mm column with 30-40 ml Ni-NTA resin (Qiagen, Hilden, Germany) was equilibrated with at least 10 resin-bed volumes (RBVs) of 0.22 mm filtered de-ionized (DI) H 2 O to remove ethanol, and then at least 5 RBVs of Ni-NTA Buffer A (above). After equilibration, lysate was added to the column and washed with 5-10 RBVs of Ni-NTA Buffer A. Sample was eluted using at least 5 RBVs of Ni-NTA Buffer B (50 mM Tris-HCl ( pH 7.4), 300 mM NaCl, 3 mM DTT, and 500 mM imidazole). An amount of 5 ml fractions were collected, analyzed via SDS-PAGE, and pooled. Then, sample was dialyzed, as follows. WDR5-containing fractions were pooled into 12-14 kDa MWCO dialysis tubing (Repligen, Waltham, MA), then 2.5 mg of GST-6H-TEV protease ( per 5 L of culture; expressed and purified in-house) was added to cleave the His-tag, and the solution was dialyzed against 4 L of dialysis buffer (above) overnight for 12-18 h at 2-8°C. Dialyzed protein was analyzed via SDS-PAGE to ensure complete cleavage. A negative Ni-NTA was then conducted. The resulting cleaved WDR5 solution was passed through a Bio-Rad Ni-NTA IMAC cartridge (Bio-Rad, Hercules, CA) using a Bio-Rad NGC chromatography system (Bio-Rad) to remove the cleaved His-tag and GST-6H-TEV protease. The flowthrough was collected, analyzed via SDS-PAGE and UV-Vis, and concentrated for size-exclusion chromatography (SEC) purification. As a final polish, the WDR5 proteins were purified via SEC using a HiLoad® 26/600 Superdex® 200 pg column (GE Healthcare, Chicago, IL) into 20 mM Tris-HCl ( pH 7.5), 300 mM NaCl, 1 mM ZnCl 2 , 1 mM TCEP. Fractions of interest were analyzed via SDS-PAGE and UV-Vis, pooled, concentrated where necessary, aliquoted to 1 ml, flash-frozen in liquid nitrogen, and stored at −80°C.
Peptide synthesis, labeling, purification, and analysis for BLI measurements All peptides for biolayer interferometry measurements were synthesized by GenScript (Piscataway, NJ). The peptides were purified to ≥95% purity. Amino acid analysis, purity confirmation, and solubility testing were provided by GenScript. All peptides were biotinylated at the N terminus. Their C terminus was amidated.

Peptide synthesis, labeling, purification, and analysis for SPR and FP measurements
For Fmoc-SPPS, peptides were synthesized at theoretical 100 mmol scale using the standard double coupling workflow pre-programmed on the Biotage Syro I peptide synthesizer (Biotage, Charlotte, NC). An amount of 278 mg of Rink Amide AM Resin LL (100-200 mesh, 0.36 mmol/g functionalization; Novabiochem (EMD Millipore)) was loaded into each 10 ml reactor vial, corresponding to the 100 mmol synthesis scale. Each resin aliquot was swelled with DMF for 30 min, followed by an initial resin deprotection step using 40% piperidine in DMF. Each double coupling cycle was comprised of two independent, 45 min coupling sub-steps that differed in the choice of activation reagents: (1) DIC/Oxyma (1 : 2 ratio with respect to Fmoc-[AA]), followed by (2) HBTU/DIPEA (0.95 : 2 ratio with respect to Fmoc-[AA]). For each coupling sub-step, four molar equivalents of each respective 0.5 M Fmoc-[AA] stock was added to each vial. Each coupling cycle terminated with a double Fmoc deprotection step using 40% piperidine in DMF. Coupling reactions proceeded with interspersed vortexing of the vials and under an air atmosphere at ambient temperature. Following the conclusion of peptide synthesis, resins were washed with three successive aliquots of DCM, followed by 3 min under vacuum to dry.
Each peptidyl-resin aliquot was swelled with DMF for 30 min and drained. DMF (1200 ml) was added to each aliquot, followed by 300 ml of DIPEA in NMP (six molar equivalents, 600 mmol), then followed by 600 ml of Lissamine Rhodamine B sulfonyl chloride in DMF (three molar equivalents, 300 mmol). The mixture was shielded from light and allowed to react with intermittent vortexing overnight at ambient temperature. Following conjugation, the resin bed was drained, washed successively with DMF until no further change in the color of the flowthrough was observed (faint pink), washed successively with DCM, and then held under vacuum for 3 min to dry.
Cleavage cocktail containing 90% TFA, 5% TIS, 2.5% DODT, and 2.5% H 2 O was freshly prepared. An amount of 4 ml of cleavage cocktail was added to each peptidyl-resin aliquot, sealed, and placed on a rocking platform to react for 4 h at ambient temperature. After incubation, the contents of each reactor vial were plunged into separate 50 ml conical tubes. The resin was then treated with an additional 2 ml of cleavage cocktail and allowed to react on the rocking platform for 30 min at ambient temperature. Following the second incubation, the contents of each reactor vial were pooled into their respective 50 ml conical tubes and the resin containing reactor vials were discarded. Cleavage aliquots were triturated by the fast addition of ∼45 ml cold (−80°C) diethyl ether. Precipitate was compacted by centrifugation at 1000×g for 10 minutes at 0°C using an Allegra X-22R centrifuge (Beckman Coulter, Brea, CA). The supernatant was discarded, and the peptide pellets were washed with a second ∼30 ml aliquot of cold diethyl ether, centrifuged, and decanted as before. The resulting peptide pellets were flash frozen in liquid nitrogen and lyophilized using a FreeZone 2.5L lyophilizer (Labcono, Kansas City, MO) overnight to remove residual solvents. The crude lyophilized peptide was stored at −80°C until purification.
Crude peptide aliquots were purified using reversed-phase chromatography through two stages: (1) Flash chromatography using a Biotage Isolera One (Biotage AB, Uppsala, Sweden), (2) Semi-preparative HPLC using a Waters 2695 separations module equipped with a Waters 2996 photodiode array detector (PDA). (1) Biotage Isolera One purifications: Each crude peptide aliquot was solubilized in 2.5 ml total of either DMSO or DMF (DMF only if the peptide contained oxidizable Cys or Met residues) and loaded onto a Biotage Sfär C 18 Samplet for 25 g Column (Biotage, Uppsala, Sweden) with the aid of a vacuum. Each samplet was placed in a Biotage Sfär Bio C 18  , and 560 nm (rhodamine) using a Waters 2996 photodiode array detector (PDA). Sample purity was determined by baseline integration using Waters Empower 3 software (>90% purity in all cases). Rhodamine containing fractions were analyzed by MALDI-TOF mass spectroscopy for the identity and purity confirmatory tests. Target fractions were flash-frozen in liquid nitrogen and lyophilized using a FreeZone 2.5L lyophilizer (Labcono) for three days. The resulting purified, lyophilized peptides were reconstituted as concentrated stocks in ultrapure water, aliquoted, flash-frozen in liquid nitrogen, and stored at −80°C until use.
For MALDI-TOF mass spectroscopy of the purified peptides, 1 ml of peptide stock in H 2 O (1-100 mM) was mixed with 9 ml spotting matrix (10 mg/ml α-cyano-4-hydroxycinnamic acid (CHCA) in 50 : 50 ). An amount of 2 ml of each peptide spotting solution was spotted onto a Bruker MTP 384 Target Plate, which was calibrated using Anaspec Peptide Mass Standard Kit, and allowed to dry at room temperature for at least 30 min prior to analysis. Samples were analyzed on a Bruker Autoflex iii Mass Spectrometer as an average of 1200 shots using 35% laser power, an m/z range from 840-6000 Da with suppression <400 Da in linear mode, 3.6× detector gain, 2.00 sample rate, and medium gating strength.

Biolayer interferometry (BLI)
Octet RED384 (FortéBio, Fremont, CA) was used for the BLI studies [71][72][73]. Streptavidin (SA) sensors were presoaked in buffer for ∼30 min. The buffer solution contained 150 mM NaCl, 20 mM Tris-HCl, 1 mM TCEP, 1 mg/ml bovine serum albumin (BSA), pH 7.5. An amount of 5 nM tagged peptide was then loaded onto sensors for 15 min. Sensors were then dipped in buffer again for 5 min to wash off unbound peptides from the surface. A 3-fold serial dilution of WDR5 was conducted ranging from 0.1 mM to 9 mM for the association process and then placed into the buffer solution for the dissociation process. The association and dissociation processes were ∼200 and ∼600 s long, respectively. For all WDR5 concentrations, unloaded sensors were run concurrently as controls and were used to subtract the baseline and the drift in the sensorgrams to extract the binding curves. The BLI experiments were performed at 24°C. All reagents were prepared in the above-mentioned buffer and were loaded into 96-well flat bottom black plates for the sensorgram recordings. The binding curves were fitted using the Octet Data Analysis software (FortéBio). The curves of the association process, which were recorded for various analyte concentrations, [C], were fitted using the following equation [74]: Here, Y 0 and Y ∞ are the BLI response signals at time zero and infinity, respectively, of the association process. t denotes the cumulative time of the association reaction. k obs is the apparent first-order reaction rate constant of the association process. The curves of the dissociation process were fitted using the following equation: Here, Y 0 and Y ∞ are the BLI response signals at time zero and infinity, respectively, of the dissociation process. k off indicates the dissociation rate constant. Finally, the association rate constant, k on , was determined using the slope of the linear curve [75,76]: Global fitting, which was conducted using several analyte concentrations, provided the corresponding k on and k off values. The dissociation constant, K D , were indirectly determined using the k on and k off values. Three distinct BLI measurements were conducted for all inspected interactions.

Surface plasmon resonance (SPR)
All SPR experiments [77][78][79] were conducted on a Cytiva Biacore 8K (Cytiva Life Sciences, Marlborough, MA). All buffers and dilutions were freshly made in-house using ultrapure water obtained from an IQ 7000 Milli Win , and SETd1B Win , the affinity constants, K D , were calculated indirectly using K D = k off /k on . For MLL1 Win , a plot of relative response versus the MLL1 Win analyte concentration was constructed and data were fitted using a four-parameter logistic regression to obtain the K D . Therefore, an affinity analysis (relative response vs. concentration dose-response curve) was used to calculate the K D , in this instance due to kinetic rate constants falling outside of the instrument detection range. All interactions were independently determined in triplicate (e.g. separate ligand immobilizations). Experimental data and fits were plotted using GraphPad Prism 8 (GraphPad Software).

Steady-state fluorescence polarization (steady-state FP) measurements
All steady-state fluorescence polarization (FP) measurements were recorded using a SpectraMax i3x plate reader (Molecular Devices, San Jose, CA). [80,81] HPLC-purified peptides, which contained an N-terminal sulforhodamine B and a C-terminal amide, were reconstituted as concentrated stocks in ultrapure water and used in all subsequent experiments. All steady-state FP assays were conducted in a buffer containing 20 mM Tris-HCl ( pH 7.5), 150 mM NaCl, 1 mM TCEP, 0.005% Tween 20 and plated in black untreated 96-well polystyrene microplates (Corning Inc., Corning, NY). Steady-state FP assays were conducted in triplicate, 24-point serial dilution for each of the six SET1Win peptides against WDR5. An amount of 200 ml of WDR5 stock solutions, ranged from 21.6 to 131 mM, were added to each well on the first column of one of two adjacent black 96-well dilution plates, and 100 ml of the assay buffer was added to wells A2-H24 over the two plates. WDR5 variants were then diluted down the two plates by transferring 100 ml from each well to the next, for a total of 24, 2-fold dilutions which range from low mM to low pM (variant specific). Following dilutions, 100 ml of the appropriate 20 nM labeled SET1 Win peptide, which was dissolved in the assay buffer, was added to each well on each set of plates at a final concentration of 10 nM. The steady-state FP anisotropy was measured on the plates after a 1 h incubation at room temperature in the dark. The resulting dose-response data were averaged and fitted using a four-parameter logistic regression to obtain the binding affinity (K D ) for each interaction. Data were plotted and analyzed using GraphPad Prism 8 (GraphPad Software, San Diego, CA).

Molecular graphics
All cartoons showing molecular graphics were prepared using the PyMOL Molecular Graphics System (Version 2.4.0 Schrödinger, LLC).

Results and discussion
The kinetic fingerprint of WDR5 -SET1 Win interactions We first explored the association (k on ) and dissociation (k off ) rate constants of WDR5-SET1 Win interactions using BLI measurements [71,72]. This technology probes ligand-receptor interactions that lead to accumulations of bimolecular complexes at the surface of the BLI sensor. This process is facilitated by immobilizing one interacting partner (ligand) onto the surface of the BLI sensor and supplying the analyte partner (receptor) from solution. In this way, the association and dissociation phases of the ligand-receptor complex are optically measured in real time using alterations in the interference pattern between reflected light waves at the surface of the BLI sensor. The N-terminus of each SET1 Win was biotinylated and their C-terminus was amidated (Table 1). A nine-residue Gly/Ser-rich peptide spacer was inserted between the biotinylated site and the SET1 Win sequence. This 3 nm-long spacer ensures that there is a satisfactory distance between the BLI sensor and SET1 Win for WDR5 to interact without steric hinderance from the sensor surface. The association binding curves were acquired through a 3-fold serial dilution of WDR5 ( Figure 2). After the BLI response reached a saturation level, individual dissociation binding curves were recorded when sensors were placed in a WDR5-free buffer. The association and dissociation BLI phases underwent a time-dependent single-exponential increase and decrease, respectively. The binding curves were fitted using the Octet Data Analysis software (Materials and Methods; eqns. (1)-(3)). The equilibrium dissociation constants using BLI, K D-BLI , were indirectly determined using k on and k off .

Association rate constants
The BLI-determined k on values were in the order of 10 4 M −1 s −1 , clearly indicating slow association kinetics for a protein-peptide system ( Table 2). It is likely that this outcome resulted from two distinct physical restrictions: (i) tethering SET1 Win onto the surface of the BLI sensor, thus reducing its local mobility, and (ii) partitioning of SET1 Win into the WDR5 cavity. The latter physical process was illustrated in early crystallographic studies [21], which revealed that an MLL1 Win peptide penetrates into the WDR5 cavity to undergo a bimolecular association process. Here, we were unable to acquire an accurate mean value of k on for the WDR5-MLL1 Win interaction due to its relatively fast k off constant, whose binding time constant was shorter than the time-resolution limit of this approach (∼1 s). Furthermore, k on measured by BLI for various SET1 Win peptides were within the same order of magnitude, suggesting a similar insertion mechanism of the six-residue Win motif peptide into the Win binding site (Table 1; Supplementary Figures S1, S2) [24,65].

Dissociation rate constants
In contrast, k off values spanned between two and three orders of magnitude (Table 2), highlighting significant distinctions in the WDR5-SET1 Win interactions. This assumes that the k off of the WDR5-MLL1 interactions is faster than 1 s −1 , showing that MLL1 Win has the weakest interaction with WDR5 among all SET1 Win peptides. On the other hand, the strongest WDR5-SET1 Win interactions were monitored with MLL2 Win and MLL3 Win , which had average k off values of (7.7 ± 0.2) × 10 −3 s −1 and (5.3 ± 0.1) × 10 −3 s −1 , respectively. This finding provides unusually long binding times of ∼130 s and ∼190 s, respectively. It was previously demonstrated that WDR5-SET1 Win interactions have a common archetype: a highly conserved Arg residue at P 0 (Table 1) [22]. This Arg residue is the pivotal player of WDR5-SET1 Win interactions, contributing to most of the binding affinity through a complex network of contacts with neighboring residues (Supplementary Table S1; Supplementary Figure S3) [22,24]. To test this hypothesis, we conducted BLI measurements using a control MLL3 Win peptide, whose native form exhibited the longest binding time with WDR5. The key Arg residue at P 0 was replaced by an Ala residue, resulting in the R4710A MLL3 Win mutant. No interaction was detected with this R4710A MLL3 Win mutant (Supplementary Figure S4), confirming the critical role of Arg at P 0 for the strength of WDR5-SET1 Win interactions. In addition, this finding validates the efficacy of our BLI measurements for examining the kinetic landscape of the Win binding site of WDR5.
Does the tethering restriction of SET1 Win onto the surface of the BLI sensor impact the kinetic rate constants?

Association rate constants
We postulated above that the slow association kinetics of WDR5-SET1 Win interactions were caused at least in part by the immobilization of SET1 Win onto the surface of the BLI sensor. To assess this hypothesis, we next employed SPR [77][78][79] as an orthogonal, label-free approach for examining these interactions of SET1 Win peptides with WDR5. SPR is an optoelectronic technique that monitors accumulation of bimolecular ligand-receptor complexes onto the surface of the SPR sensor by changes in the refractive index. In this case, SET1 Win analyte is supplied by using a flow-driven fluidic device. WDR5 was immobilized onto the surface of the SPR chips as the 'receptor' (Materials and Methods). In this way, we recorded the association and dissociation phases in real time when SET1 Win was not immobilized onto a surface (Figure 3). Biacore TM Software was used to analyze and fit the SPR sensorgrams using a 1 : 1 binding interaction model to provide the k on and k off rate constants (Supplementary Figure S5). In accord with our expectation, k on values obtained by SPR were significantly increased by almost an order of magnitude to quantities greater than 10 5 M −1 s −1 (Table 3; Supplementary  Figures S6, S7A). Again, k on measured by SPR for various SET1 Win peptides were within the same order of magnitude, potentially indicating an identical insertion mechanism of the six-residue Win motif peptide into the Win binding site (Table 1; Supplementary Figures S1, S2) [24,65]. In addition, this insertion mechanism prevails no matter whether SET1 Win is in either physically restricted (e.g. BLI) or unrestricted (e.g. SPR) conditions.

Dissociation rate constants
Interestingly, with the exception of SETd1A, the k off values acquired by the SPR measurement of WDR5-SET1 Win interactions closely resembled those measured by BLI (Table 3; Supplementary Figure S7B). k on was not quantitatively determined. Although WDR5-MLL1 Win interactions were detectable using a BLI measurement (Figure 2), no accurate quantitative determination was made due to the limited time resolution of this approach. In this case, we assume that the k on was in the same order of magnitude with the k on of the other SET1 Win peptides (∼10 4 M −1 s −1 ); 2 k off was not quantitatively determined due to a fast dissociation rate constant. The upper-limit value for the detection of k off is 1 s −1 according to instrument specifications; 3 K D-BLI was not quantitatively determined due to the limited time resolution of the approach. In this case, K D-BLI determined by BLI for WDR5-MLL1 Win interactions should be greater than ∼10 5 nM. This value results from dividing the upper-limit value of detection of k off (ND**) by the value of the k on approximation (ND*);Numbers represent mean ± s.d. determined from three independent experiments.
Moreover, this finding provides indirect evidence that WDR5 did not undergo denaturation upon its immobilization onto the surface of the SPR sensor. Tethering SET1 Win onto the surface of the BLI sensor via its C-terminus was considered inconvenient, because prior crystallographic information indicated the interaction of the C-terminus residues (Table 1) with the WDR5 surface [24]. However, does the SET1 Win tethering onto the surface of the BLI sensor via its N-terminus impose an additional physical restriction on the binding mechanism? Since the k off measured by BLI and SPR are almost identical, we conclude that the SET1 Win tethering onto the surface of the BLI sensor via its N-terminus did not produce any additional restrain on the detachment mechanism of SET1 Win from the high-affinity Win binding site.

Comparisons of binding affinities of WDR5-SET1 Win interactions in restricted and unrestricted conditions
Next, we asked how immobilization-free (i.e. unrestricted) conditions influence the binding affinities. It is expected that the equilibrium dissociation constant slightly decreases if both WDR5 and SET1 Win move freely in solution. Hence, we determined equilibrium dissociation constant values using steady-state FP measurements, K D-FP (Materials and Methods) [80][81][82]. This fluorescence technique monitors changes in the polarity of the emitted light due to modifications in the rotational diffusion coefficient of a fluorescently labeled ligand. For example, the formation of the ligand-receptor complex is accompanied by a decline in the rotational mobility of the ligand, increasing the polarity of the emitted light. Here, Sulforhodamine B, an optically stable and bright fluorophore [83], was chemically attached to the N terminus of SET1 Win via a nine-residue Gly/Ser-rich flexible spacer, whereas their C terminus was amidated. WDR5 concentration-dependent, steady-state FP binding curves were collected using a plate reader in a high-throughput setting ( Figure 4). In this way, we determined the immobilization-free K D-FP (Supplementary Table S2). In accord with our anticipation, K D-FP values were slightly lower than those acquired by SPR, K D-SPR , meaning that WDR5-SET1 Win interactions appeared somewhat stronger when probed in unrestricted conditions. k on was not quantitatively determined. Although the WDR5-MLL1 Win interactions were detectable using an SPR measurement (Figure 3), no accurate quantitative determination was made due to the limited time resolution of the approach. In this case, we assume that the k on was in the same order of magnitude with the k on of other SET1 Win peptides (∼10 5 M −1 s −1 ); 2 k off was not quantitatively determined due to a fast dissociation rate constant. The upper-limit value for the detection of k off is 0.5 s −1 according to instrument specifications. The Biacore 8K+ cannot measure rate constants of dissociation, k off , faster than 0.5 s −1 ; 3 Here, K D-SPR was determined using a steady-state SPR measurement (Supplementary Figure S6); Values represent mean ± s.d. acquired from at least three independent experimental determinations (separate receptor immobilizations). For MLL2 Win , n = 4 independent experimental determinations were used. The N terminus of the SET1 Win peptides was tagged with sulforhodamine B, whereas the C terminus was amidated. The final concentration of labeled SET1 Win peptides in each well was 10 nM. Three independent experiments were conducted to obtain the WDR5 dose response.
The K D-FP values are also in agreement with results of prior ITC measurements (e.g. in immobilization-free conditions) at either lower [65] or higher salt concentrations [24]. It is worth mentioning that we preferred replacing Arg by Ser at the C terminus of MLL1 Win and MLL4 Win to avoid potential rebinding events of the native Arg at P 6 to the WDR5 cavity (Table 1). Regardless, we employed similar peptide sequences and identical buffer conditions to obtain meaningful comparisons among the three approaches. Then, we have discovered the following relationship for each WDR5-SET1 Win interaction: K D-BLI > K D-SPR > K D-FP ( Figure 5; Supplementary Table S2). In addition, numerical values of K D-FP replicated similar trends noted with different SET1 Win peptides using both BLI and SPR.

Tentative interpretations
The kinetics and affinity measurements of the WDR5-SET1 Win interactions were conducted with no significant variation among independently recorded BLI sensorgrams. The precision of this technique allowed us to quantitatively compare kinetics that were within the same order of magnitude. The SET1 Win motif is located ∼60 residues N-terminal to the SET domain, both of which are at the C-terminal end of the large SET1 proteins. These proteins range between ∼1700-5500 amino acids in length. The rotational and diffusional rates of SET1 proteins relative to WDR5 are slow enough that these proteins can be considered as stationary compared with WDR5. Therefore, a more physiologically-relevant approach that recapitulates the binding of the full-length SET1 is that in which the 14-residue SET1 Win peptide is attached onto a surface, whereas WDR5 is free in solution. In this way, the actual k on measured by BLI would be conceivably closer to that value in physiological conditions. In addition, it should be mentioned that SET1 Win is physically restrained at both ends when it is part of the SET1 subunit, suggesting an even lower k on value. However, SET1 Win may have stabilized conformations that promote binding within the context of the large SET1 subunit. In other words, the unrestricted SET1 Win may undergo significantly greater degrees of freedom to adopt numerous non-productive conformations. Furthermore, binding of SET1 proteins to WDR5 may be coupled with additional interaction pockets that can amplify the binding strength either by increasing k on or by decreasing k off or both. For example, prior sedimentation velocity experiments have shown that a 225-residue MLL1 Win -containing polypeptide exhibits a binding affinity of ∼120 nM to WDR5 [22].
k on values measured by SPR are yet at least an order of magnitude lower than those predicted for a proteinpeptide interaction system (10 7 -10 8 M −1 s −1 ) [84][85][86][87][88]. Here, WDR5-SET1 Win interactions require a precise insertion of the six-residue, Arg-containing SET1 Win peptide into the WDR5 cavity. Therefore, we interpret that k on values are limited by the entropic penalty determined by the SET1 Win partitioning into the Win binding site. The high-affinity Win binding site features a conical geometry with a maximum internal diameter of ∼1.5 nm, as measured from side chain to side chain. For example, MLL1 Win partitions ∼1.0 nm into the  (Table 2).
Win binding site (Figure 1) [24]. In support to this interpretation, previous single-molecule studies have shown that k on values of the interactions of peptides, up to 25 residues in length, with a narrow 2 nm-wide nanopore are in the order of 10 5 M −1 s −1 [89,90]. This is the same order of magnitude with k on values that we determined by SPR.
However, the k on is a composite parameter, which includes contributions of the diffusion-limited rate factor, k on0 , and electrostatic free energy of the WDR5-SET1 Win complex, ΔG* el . Zhou and coworkers have demonstrated that the rate constant of association of two proteins can be accurately computed using the following expression: k on = k on0 exp(−ΔG* el /k B T), where k B and T are Boltzmann's constant and the absolute temperature, respectively [91,92]. For this equation, they employed kinetic-rate theory of rigid-body docking and Poisson-Boltzmann formalism [93][94][95]. Using the same approach [93] and crystallographic information of the WDR5-SET1 Win complex [24], we found k on values of several SET1 Win peptides (Supplementary Table S3). For example, computed k on values for MLL2 Win and MLL3 Win were 2.9 × 10 5 M −1 s −1 and 1.2 × 10 5 M −1 s −1 , respectively. These values compare well with our corresponding experimental data determined by SPR, which were (3.7 ± 0.3) × 10 5 M −1 s −1 and (4.9 ± 0.4) × 10 5 M −1 s −1 , respectively. Yet, computed k on values for MLL1 Win and MLL4 Win were in the order of 10 4 M −1 s −1 . The TransComp server: Web Server for Predicting Protein Association Rate Constants [91,93] was not able to produce computed k on values for SETd1A Win and SETd1B Win , likely because these WDR5-SET1 Win complex formations do not undergo a single-step association mechanism. This outcome suggests that the WDR5-SET1 Win interactions of SETd1A Win and SETd1B Win exhibit some subtle structural distinctions with respect to MLL1 Win , MLL2 Win , MLL3 Win , and MLL4 Win , which is in accord with prior crystallographic data [24,65].

Concluding remarks, practical implications, and future prospects
In summary, we present a detailed kinetic fingerprint of the multitasking high-affinity Win binding site of WDR5, a protein with major regulatory implications in the methylation of H3K4 and in multiple physical associations with other proteins. This study reveals slow kinetics of association and dissociation of the SET1 Win peptides with WDR5. It is known that WDR5 bridges the interaction between the SET domain of large SET1 subunits and other WRAD 2 constituents. A long-binding time of WDR5-SET1 Win interaction, meaning slow dissociation kinetics of the high-affinity Win binding site, is a pivotal mechanism by which WDR5 assists the functional integrity of the multisubunit WRAD 2 complex. Furthermore, slow dissociation rates detected in this study point out a fundamental consideration for future therapeutics aimed at targeting interactions of the Win binding site with other proteins. The stability of these multisubunit complexes would decrease the opportunities for inhibitors to interfere in WDR5-SET1/MLL interactions. Consequently, this stresses the need for both fast association rates and slow dissociation rates when designing potential inhibitors to modulate WDR5 function. Newly designed small-molecule drugs would need to bind strongly to the WDR5 cavity and stave off other binding partners. Slow dissociation rate constants also reiterate the efficacy of SET1 win peptidomimetics as a fundamental platform for such drugs. Modifications of these sequences that enhance the association rates, while maintaining the disassociation rates, could be very effective at inhibiting Win binding site interactions. A good place to start would be the alteration of either the net charge or charge distribution of these peptides for amplifying the rate constants of association. Any approach employed in this work can be used in the highthroughput screening of libraries of small-molecule compounds against WDR5-SET1 Win interactions. However, WDR5 is a more relevant target for drug discovery when immobilized onto the SPR sensor, where the free inhibitor in solution would better mimic in vivo function. This experimental design enables determinations of kinetic rate constants, contrasting the steady-state FP measurements. Our results also indicate that WDR5 did not undergo denaturation upon its immobilization onto the SPR sensor. In the future, it would be desirable to extend these kinetic studies to full-length WRAD 2 subunits, because of the suitability of these high-throughput approaches for examining long-lived protein-protein interactions. For instance, it would be interesting to conduct SPR measurements, in which a full-length SET1 subunit is immobilized on the chip surface and use WDR5 as an analyte in solution.

Data Availability
All supporting data, materials, and sequence information are included within the main article and its Supplementary File.