Modulation of Lymphocyte Potassium Channel K_V1.3 by Membrane-Penetrating, Joint-Targeting Immunomodulatory Plant Defensin

Abstract

We describe a cysteine-rich, membrane-penetrating, joint-targeting, and remarkably stable peptide, EgK5, that modulates voltage-gated K_V1.3 potassium channels in T lymphocytes by a distinctive mechanism. EgK5 enters plasma membranes and binds to K_V1.3, causing current run-down by a phosphatidylinositol 4,5-bisphosphate-dependent mechanism. EgK5 exhibits selectivity for K_V1.3 over other channels, receptors, transporters, and enzymes. EgK5 suppresses antigen-triggered proliferation of effector memory T cells, a subset enriched among pathogenic autoreactive T cells in autoimmune disease. PET-CT imaging with ¹⁸F-labeled EgK5 shows accumulation of the peptide in large and small joints of rodents. In keeping with its arthrotropism, EgK5 treats disease in a rat model of rheumatoid arthritis. It was also effective in treating disease in a rat model of atopic dermatitis. No signs of toxicity are observed at 10–100 times the in vivo dose. EgK5 shows promise for clinical development as a therapeutic for autoimmune diseases.

Introduction

Ion channels underlie electrical activity of excitable cells in the nervous system, heart, and muscle. Their role in nonexcitable cells was less appreciated until patch-clamp analysis made it possible to record electrical signals from single cells.¹ Ion channels entered the immunological realm in 1984 when human T lymphocytes, quintessential nonelectrically excitable cells, were found to express voltage-dependent potassium (K_V) currents that were required for T cell proliferation.²⁻⁵ The KCNA3 gene encoding the K_V1.3 channel was identified in 1990.^6,7 The functional channel in lymphocytes is a tetramer of K_V1.3 subunits complexed to a tetramer of K_Vβ2 accessory subunits.^8,9 K_V1.3 is part of a signalosome in the plasma membrane that clusters at the immunological synapse and serves to couple external stimuli with intracellular signaling cascades that are essential for cellular homeostasis and activation in T cells.¹⁰⁻¹⁶ The channel performs similar functions in B lymphocytes.^17,18 K_V1.3 is up-regulated in activated effector memory T cells (T_EM cells), a subset enriched among pathogenic autoreactive T cells in many autoimmune diseases.¹⁹⁻²¹ K_V1.3 inhibitors suppress proliferation, cytokine secretion, and in vivo migration of T_EM cells without compromising protective immune responses provided by other T cell subsets, and they treat disease in animal models of autoimmune disorders.²⁰⁻³⁰

Cysteine-rich peptides from scorpions, sea anemones, and parasitic worms block K_V1.3 by binding to a shallow basin at the outer entrance to the K_V1.3 pore and quell the flow of K⁺ ions.³¹⁻³⁶ Dalazatide (a.k.a. ShK-186), a derivative of the sea anemone peptide ShK,²⁰ is the only K_V1.3-selective inhibitor to advance to human clinical trials (ClinicalTrials.gov identifiers NCT02435342, NCT02446340). It proved safe in phase I trials and ameliorated symptoms in patients with plaque psoriasis.²⁰ These studies demonstrate the therapeutic potential of K_V1.3 inhibitors for autoimmune diseases.

Plant defensins are cysteine-rich peptides with many functions including antimicrobial, antifungal, and anti-insect activity.³⁷⁻⁴⁶ They also modulate sodium, calcium, and potassium channels, which could possibly contribute to their defensive functions.⁴⁷⁻⁴⁹ Here, we describe EgK5, a designed analogue of plant defensins from grapevine and oil palm, that enters plasma membranes, binds to K_V1.3, and suppresses K_V1.3 currents. EgK5 inhibits T_EM cell proliferation, accumulates in small and large joints, and treats rheumatoid arthritis in a rat model without signs of toxicity.

Results

Structure Determination of VvK1 and EgK1

Using tBLASTN searches we identified plant defensins from grapevine (Vitis vinifera, VvK1) and oil palm (Elaeis guineensis, EgK1) with sequence similarity to scorpion toxin inhibitors of K_V1.3 (Figure 1a). We synthesized these peptides (Supporting Information, Figure S1). VvK1 generated two peaks on reverse phase HPLC, which were further characterized by NMR (see below), while EgK1 gave a single peak (Figure S1). We determined the structures of these peptides by X-ray crystallography. VvK1 (peak 1) and EgK1 crystallized as dimers; VvK1 (peak 2) crystals were not of diffraction quality. We determined VvK1’s (peak 1) structure at 1.3 Å resolution by molecular replacement using plant defensin MtDef4 (PDB 2LR3), and the structure of EgK1 (2.1 Å) was solved by molecular replacement using the structure of VvK1 (Figure 1b; Table S1). The structures of the two defensins are remarkably similar to each other (Figure 1b) and to the K_V1.3-blocking scorpion toxin kaliotoxin (KTx, PDB 2UVS)³² (Figure 1c). Although the defensins crystallized as dimers, one-dimensional (1D) ¹H nuclear magnetic resonance (NMR) studies suggested that in solution at pH 2–10 they were monomers. Good chemical shift dispersion was observed in the 1D ¹H NMR spectra of VvK1 (peaks 1 and 2) and EgK1, indicating that these peptides adopted stable, folded structures in solution. Single sets of resonances in the ¹H NMR spectra indicated the presence of a single conformation (Figures S2–S5). The conformations of VvK1 peaks 1 and 2 were stable at pH 5–10, but unfolded partially at pH < 5 (Figures S2, S4, S5). Slight peak broadening was observed in the spectrum of VvK1 peak 1 compared to VvK1 peak 2 (Figure S2). In addition, the chemical shifts of a few resonances in the VvK1 peak 1 spectrum were quite distinct from those in the VvK1 peak 2 spectrum (Figures S4, S5); such differences between two peptides with the same sequences and masses may have arisen from racemization of the C-terminal cysteine residue during peptide synthesis.⁵⁰ Hydrodynamic radii calculated from translational diffusion measurements made by NMR (Table S2)⁵¹ confirmed that the defensins are monomeric in solution at pH 5 and 20 °C, in contrast to their dimeric structures in crystals.

Figure 1.

Design of EgK5, an analogue of plant defensins from grapevine and oil palm. (a) Alignment of K_V1.3-blocking scorpion toxins (KTx and MgTx) and plant defensins (VvK1, EgK1, AtPDF2.3). Identical residues in plant defensins are highlighted in gray. The critical pore-blocking lysine in scorpion toxins (KTx-K27 and MgTx-K28) is highlighted in blue. Plant defensins contain K33 in the corresponding position. The [K/R]G[F/L]RRR motif is shown. EgK6 was designed to make EgK5 more like K_V1.3-blocking scorpion toxins. The GKCKGLRRRC motif of EgK5 was replaced with GKCMNGKC (resembling K_V1.3 pore binding motif of scorpion toxins), the length was shortened to 33 amino acids, and the Cys3–Cys47 disulfide bridge was replaced with a Cys18–Cys33 bridge to mimic HsTx1/α-KTX-6.3. (b) Structure of VvK1 (blue, peak 1) and EgK1 (red). VvK1 and EgK1 showed buried surface areas of 990 and 1030 Ų, respectively. Both peptide dimers were categorized as assemblies that fall into the gray area of complexation criteria, which may or may not be stable in solution. (c) Overlay of VvK1 (blue), EgK1 (red), and KTx (green). K27 of KTx and K33 of VvK1 and EgK1 are shown as sticks. RMSDs of 1.14 Å (VvK1-KTx) and 1.08 Å (EgK1-KTx). (d) The orientations of side chains and interactions of contact pairs that are formed during MD simulations of VvK1 and EgK1. The average contact distances are listed in Table S3. EgK5 contains five substitutions (Q6A, K11E, A17D, R35K, and F37L) designed to make the peptide more resistant to protease digestion. (e) Stability of EgK1 (▲) EgK5 (△), VvK1 (peak 2, ■), VvK1 (peak 1, □), VvK1_10–44 (○) and ShK (●) tested against trypsin, chymotrypsin, and pancreatin in simulated intestinal fluid. Reactions were conducted in triplicate (n = 3). EgK1 and VvK1 (both peaks) were more resistant to trypsin, chymotrypsin, and pancreatin than the sea anemone peptide, ShK. Both peaks of VvK1 were more stable than EgK1, and VvK1 (peak 2) was more stable than VvK1 (peak 1). Truncated VvK1 (VvK1_10–44) was drastically less protease resistant, emphasizing the importance of the Cys3–Cys47 bridge between the N- and C-termini. No peptide was affected by pepsin in simulated gastric fluid.

Design of EgK5 and Assessment of Its Stability

We performed molecular dynamics simulations to understand the stabilizing forces in VvK1 and EgK1. To make these defensins more like the sequence-related K_V1.3-blocking scorpion toxins α-KTx-3.1 and α-KTx-2.2 (Figure 1a), we deleted the Cys3–Cys47 disulfide bond, which is present in the defensins but not the scorpion toxins. Removal of the Cys3–Cys47 disulfide bridge caused truncated VvK1_10–44 to be less stable than full-length VvK1 (RMSDs with respect to the crystal structure are 2.1 and 1.4 Å, respectively), indicating that minimization of the structure might not be successful. Besides the four disulfide bonds, intrapeptide interactions contribute to stability (Figure 1d, Table S3, Figure S6).

On the basis of structural and MD comparisons of EgK1 and VvK1, we designed EgK5 to enhance stability. EgK5 contains six substitutions: Q6A, K11E, R16G, A17D, R35K, and F37L (Figure 1a). We made the K11E and A17D substitutions to establish the E11-R40 and D17-C34 interactions of VvK1. The Q6A change was made to eliminate R40 side chain-switching from E11 to Q6. The R16G, R35K, and F37L substitutions were performed to make EgK1 more like VvK1. MD simulations of EgK5 showed the predicted new ionic bonds E11–R40 and D17–C34 and preservation of existing ionic bonds in EgK1 (R1–C47, E4–R35, and K9–E27) (Table S3). Synthetic EgK5 gave one peak on reverse phase HPLC (Figure S1). EgK5 was resistant to trypsin, chymotrypsin, and pancreatin in simulated intestinal fluid (Figure 1e). EgK5 was also stable in rat and human plasma (Table S4), and stable at extreme pH for several days (Figures S7a,b).

EgK5, VvK1, and EgK1 Do Not Block K_V1.3 Despite Their Structural Similarity to K_V1.3-Blocking Scorpion Toxins

Despite their sequence similarity to K_V1.3-blocking scorpion toxins (Figure 1a), EgK5, VvK1, and EgK1 applied on the extracellular side in patch-clamp experiments of <1 h duration did not block K_V1.3 (Figure 2a). In contrast, K_V1.3 inhibitors margatoxin³² (MgTx, α-KTx-2.2, 1 nM), Psora-4⁵² (1 μM), and PAP-1²⁵ (1 μM) blocked the K_V1.3 currents completely (Figure 2a, Figure S8a). In these patch-clamp experiments, EgK5 did not alter the voltage-dependence of activation, deactivation, or inactivation (Figure S8b–d) demonstrating that it is not a gating modifier. Further, MgTx blocked K_V1.3 with the same concentration–response curve in the absence or presence of an excess of EgK5 (Figure 2b), indicating that EgK5 does not bind to the channel’s external vestibule.³² AtPDF2.3, a recombinant plant defensin from Arabidopsis thaliana with sequence similarity to EgK5 (Figure 1a), was also reported to not block K_V1.3 expressed in Xenopus oocytes, although it blocked K_V1.2 and K_V1.6 expressed in Xenopus oocytes in a manner similar to block by scorpion toxins.⁴⁹ EgK5 (10 μM) had no effect on K_V1.2 and other potassium channels expressed in mammalian cells (Figure S8e). These results indicate that EgK5, EgK1, and VvK1 do not block K_V1.3, unlike sequence- and structurally related scorpion toxins.

Figure 2.

EgK5 binds to K_V1.3 and enters cell membranes. (a) EgK5, EgK1, and VvK1 do not block K_V1.3 when applied from the extracellular side (<1 h). In contrast, MgTx (1 nM) and Psora-4 (1 μM) when applied from the extracellular side (<1 h) block K_V1.3 currents. (b) In a competition experiment, no change was seen in the dose–response curve of MgTx block of K_V1.3 in the absence or presence of an excess of EgK5. (c) SDS-PAGE analysis of purified C-terminally His₁₀-tagged human K_V1.3. (d) MALDI-TOF/TOF mass spectra of EgK5 (top and middle panel; 5.2 kDa) or EgK6 (bottom panel; 3.5 kDa) from a cobalt resin affinity purification in the absence (top panel) or presence (middle and bottom panels) of human K_V1.3 (in detergent and phospholipids). (e) Biolayer interferometry data showing the association and dissociation between His₆-tagged K_V1.3 and EgK5 at the concentrations indicated. The dashed line denotes the time at which the sensors were transferred to the control buffer. (f) Surface rendering of the electrostatic potential of EgK5, highlighting Arg38, Arg39, and Arg40 in the KGLRRR motif. (g) (Left) L929 fibroblasts stained with TAMRA-EgK5 (red), actin (green), and Hoechst (blue). Orthogonal XZ and YZ projections are shown with the merged image. Z-stack images were acquired at 1 μm intervals (Supplementary video 1). Scale bar = 5 μm. (Right top) Intensity plot through a representative cell showing presumptive plasma membrane and intracellular region, actin (green) and TAMRA-EgK5 (red). The histogram plots the distribution of EgK5 intensities between presumptive membrane and intracellular regions (n = 6). (Right bottom) Average membrane intensity of TAMRA-EgK5 at 1, 3, and 24 h (n > 80 per time point). (h) Trypan blue and 7AAD dye exclusion on mouse K_V1.3-expressing L929 fibroblasts and activated human T cells treated with various concentrations of EgK5 for 24 and 72 h.

EgK5 Binds to Purified K_V1.3

We tested if EgK5 could bind to purified K_V1.3 (Figure 2c and Figure S8f–h) using a mass spectrometry method used previously for scorpion toxin binding to the KcsA potassium channel.⁵³ Purified human K_V1.3 containing a C-terminal His₁₀-tag in detergent and phospholipids was incubated with EgK5 and then bound to HisPur cobalt resin. After the flow-through was collected, the beads were washed to remove unbound EgK5, followed by elution of K_V1.3 with imidazole. All fractions were analyzed by MALDI-TOF mass spectrometry focusing on the low mass range appropriate for EgK5 (∼5.2 kDa). In the K_V1.3-binding experiment, EgK5 was seen in the K_V1.3-containing eluted fractions and not in the flow-through and wash fractions (Figure 2d). In the control experiment where K_V1.3 was omitted, EgK5 was detected in the flow-through and wash fractions, and only to a small extent in the first eluted fraction, indicating that EgK5 did not bind nonspecifically to the cobalt beads (Figure 2d). EgK6 (∼3.5 kDa), a truncated and modified analogue of EgK5 (Figure 1a), did not bind K_V1.3 (Figure 2d) or block K_V1.3 (Figure S8m). As an orthogonal test, we used biolayer interferometry (BLI) to measure EgK5 binding to K_V1.3. His₆-tagged K_V1.3 in detergent was anchored to the BLI sensor tip.⁵⁴ The addition of EgK5 to K_V1.3 increased the optical interference signal in a concentration dependent manner (Figure 2e). Association and dissociation curves were fitted with a single exponential and the K_D was determined to be 0.51 μM (Figure 2e). These results confirm that EgK5 binds with submicromolar affinity to K_V1.3.

EgK5 Enters Plasma Membranes and Does Not Cause Cell Cytotoxicity

Arginine-rich peptides and some cysteine-rich peptides enter plasma membranes and some enter the cytoplasm.⁵⁵⁻⁶¹ EgK5 contains 10 positively charged residues (Figure 1a) and a positively charged electrostatic potential with a KGLRRR basic residue-patch (Figure 2f) resembling the RGFRRR motif that allows plant defensins to enter membranes.⁵⁵ To test this idea, we attached the fluorophore TAMRA (tetramethylrhodamine) via a stable amide bond to the N-terminus of EgK5 (TAMRA-EgK5) and tested its ability to enter plasma membranes and the cytoplasm.⁵⁶ K_V1.3-expressing L929 cells were incubated with TAMRA-EgK5 for 1, 3, and 24 h, washed extensively to remove any free TAMRA-EgK5, stained for actin to delineate the cellular outline and with Hoechst dye to delineate the nucleus, and visualized by confocal microscopy. We used Zeiss Zen 2 imaging software to measure pixel intensities along axes drawn through each cell to assess presumptive membrane and intracellular expression. The pixel intensity profile of a representative cell is shown in Figure 2g. TAMRA-EgK5 is primarily seen in the membrane overlapping actin staining, and a smaller fraction is seen intracellularly (Figure 2g, Supplementary video 1). The average intensity measured for multiple cells shows TAMRA-EgK5 mainly in the plasma membrane where it accumulates over time (Figure 2g). Similar results were seen with human T cells (Figure S8i). These results suggest that TAMRA-EgK5 enters and remains in the plasma membranes, and a minor fraction enters the cytoplasm. Although plant defensins are reported to interact with membrane phospholipids and form carpet-structures resulting in cell lysis,⁴⁰⁻⁴⁶ EgK5 did not alter the viability of K_V1.3-expressing L929 fibroblasts and T lymphocytes (Figure 2h; Figure S8j), and did not cause toxicity in vivo (Tables S5–S7).

EgK5 Induces K_V1.3 Current Run-down by a PIP₂-Dependent Mechanism

Since EgK5 binds to purified K_V1.3 (Figure 2d,e) and enters plasma membranes over 1–24 h (Figure 2g), we next tested whether prolonged exposure to EgK5 would result in K_V1.3 channel-modulation. L929 fibroblasts stably expressing cloned mouse K_V1.3 channels were exposed to vehicle (control) or EgK5 (10 μM) for 24 h, and K_V1.3 currents were then measured during 20 depolarizing pulses. Figure 3a shows that EgK5-treated cells exhibit significant run-down of K_V1.3 current but there is little change in current amplitude in control cells. When K_V1.3 current amplitude at each pulse is normalized to the amplitude at the first pulse, EgK5-treated cells show significant run-down of K_V1.3 compared to control cells (Figure 3b). L929 cells incubated without or with EgK5 (10 μM) for 24 h showed similar average membrane capacitance (without EgK5 17 ± 5 pF; with EgK5 18 ± 6 pF). Average current amplitude at pulses 18–20 normalized for membrane capacitance (pA/pF) showed that EgK5 at 1 and 10 μM reduced K_V1.3 current density compared to control cells, this reduction being evident as early as 3 h (Figure 3c,d). The viability of cells used in all these patch-clamp experiments was greater than 99% (Figure 2h) indicating that the significant cell-to-cell variation in EgK5-induced K_V1.3 current run-down (Figure 3c,d, Figure S8k,l) is not due to reduction in cell viability. Similar results were obtained with human T cells (Figure S8k). In cells exposed to EgK5 for longer periods (48 h, 72 h), lower concentrations (0.01 μM, 0.1 μM) were effective in modulating both rodent and human K_V1.3 (Figure S8l) and these cells exhibited >99% viability (Figure 2h). The increased sensitivity of mouse K_V1.3 to 0.1 μM EgK5 compared to human K_V1.3 may be due to sequence differences in these channels, or to differences in the cells in which they were expressed (adherent L929 cells for mouse K_V1.3 versus activated T cells for human K_V1.3 where it is present as a complex with K_Vβ2).

Figure 3.

EgK5 modulates K_V1.3 currents in a PIP₂-dependent mechanism. (a) K_V1.3 currents in cells treated with vehicle (control) or EgK5 (10 μM) for 24 h. Current amplitude at pulses 1, 3, 10, and 20 are shown. (b) K_V1.3 currentamplitude measured during 20 depolarizing pulses in L929-fibroblasts treated with vehicle (control, blue) or EgK5 (10 μM, red) for 24 h. The current amplitude at each pulse is normalized to the amplitude at the first pulse. n = 69 cells. (c) EgK5 (1 and 10 μM) treatment for 24 h reduces K_V1.3 current densities (pA/pF). pA was the average K_V1.3 current amplitude at pulses 18–20. (d) EgK5 (10 μM) reduces K_V1.3 current-density after 3 and 6 h treatment. (e,f) Cells were treated with vehicle (control, n = 22) or EgK5 (10 μM, n = 26) for 24 h and then patch-clamped. (e) K_V1.3 currents during 20 depolarizing pulses. Amplitude at each pulse is normalized to the first pulse. Cells were patch-clamped with standard internal solution (purple) or internal solution containing PIP₂ liposomes (red). (f) Current amplitudes during pulses 18–20 normalized to the first pulse from cells shown in part e. (g) Cells were patch-clamped with standard internal solution (black) or internal solution containing EgK5 (10 μM, red), n = 15. (h) Confocal microscopy revealed equivalent cell surface K_V1.3 staining following exposure to EgK5 (10 μM) or control medium for 24 h. Scale bar = 5 μm. (i) EgK5 (10 μM, 24 h) did not affect K_V1.5, K_V11.1, K_Ca3.1, and K_Ca1.1 currents. (a–h) L929-fibroblasts stably expressing cloned mouse K_V1.3 were used. (a–i) Cells with capacitances of 10–30 pF (L929-K_V1.3, CHO-K_V11.1, and HEK293-K_Ca3.1) or 3–12 pF (MEL-K_V1.5) were included for data analysis. (j) EgK5 (1–100 nM) suppressed antigen-triggered ³H-thymidine incorporation, a measure of proliferation, by ovalbumin-specific rat T_EM cells.⁶⁶ Data are representative of at least two (e–i) or three (a–d,j) independent experiments. All bar graphs depict mean ± SEM. Statistical analysis for (c,d) 1-way ANOVA, (h,i), Student t test, and (j) 1-way ANOVA with Tukey posthoc test. *p < 0.05, **p < 0.01, ***p < 0.001.

PIP₂ (phosphatidylinositol 4,5-bisphosphate), a low-abundance phospholipid of the inner leaflet of plasma membranes, is a cofactor of many ion channels and is required for channel function.⁶²⁻⁶⁵ PIP₂ depletion results in current run-down, and currents can be restored by application of PIP₂ liposomes to the intracellular surface.⁶²⁻⁶⁵ In K_V channels, PIP₂ interacts with the linker connecting the voltage-sensor to the pore domain, and PIP₂ depletion causes current run-down by increasing voltage-sensitivity and decreasing current amplitude.⁶³ To test if PIP₂ depletion is involved, cells were treated with EgK5 (10 μM) or vehicle (control) for 24 h, and K_V1.3 currents were measured over 20 depolarizing pulses with either standard internal solution or PIP₂ liposome-containing internal solution. Current amplitudes were normalized to the amplitude at the first pulse (Figure 3e). To quantify run-down, we compared average normalized current amplitude at pulses 18–20 (Figure 3f). In control cells, there was no K_V1.3 run-down in the presence or absence of PIP₂ (Figure 3e,f). In EgK5-treated cells, K_V1.3 currents ran down in cells patched with standard internal solution, and PIP₂ reversed this K_V1.3 current-decay (Figure 3e,f). These results suggest that PIP₂ depletion underlies EgK5-induced K_V1.3 run-down. Inclusion of EgK5 in the internal solution and recording of K_V1.3 currents for 20 pulses following break-in to the whole-cell configuration did not result in K_V1.3 current run-down (Figure 3g). Longer experiments were not possible because membranes become leaky and recording becomes difficult 30–60 min following break-in. To assess if EgK5’s effect on K_V1.3 was due to channel internalization, we stained cells with a K_V1.3-specific antibody (extracellular epitope) without permeabilization following prolonged exposure to EgK5 or vehicle control and measured surface K_V1.3 expression. There was no difference between the two conditions indicating that EgK5 does not cause channel-internalization (Figure 3h). In summary, EgK5, a cysteine-rich and remarkably stable-analogue of plant defensins from grapevine and oil palm, enters plasma membranes of mammalian cells, binds to K_V1.3, depletes PIP₂, and in this way reduces the pool of K_V1.3 channels capable of gating, thereby causing K_V1.3 run-down. Future studies will determine the precise location of EgK5 in the plasma membrane, and define its site of interaction on K_V1.3 and the biophysical mechanism underlying its modulation of the channel.

EgK5 Exhibits Selectivity for K_V1.3 over Other Molecular Targets

We tested the specificity of EgK5 by electrophysiology and binding assays. To assess specificity, K_V1.5 and K_V11.1 (hERG) were used because they are critical for cardiac safety and because K_V11.1 is modulated by PIP₂,⁶²⁻⁶⁵ K_Ca3.1 because it, together with K_V1.3, is essential for T cell homeostasis, and K_Ca1.1 because it is a target for modulating synovial fibroblasts and therefore germane to the studies on the rat model of collagen-induced arthritis described below. Prolonged exposure to EgK5 (10 μM) did not affect any of these channels (Figure 3i). Further studies are required to determine if EgK5 causes current run-down of other ion channels known to be modulated by PIP₂.⁶²⁻⁶⁵ Binding studies were performed with EgK5 (1 μM) on a panel of 88 additional molecular targets to further assess EgK5’s specificity. These included channels, transporters, receptors, and kinases; many assays were done over long incubation times and some on native tissues (Table S8). Eighty-five targets were unaffected (Table S8). EgK5 at 1 μM inhibited binding of [¹²⁵I]omega-conotoxin GVIA to the rat Ca_V2.2 calcium channel by 78%, and the binding of [¹²⁵I]NDP-α-MSH to melanocortin receptors mouse MC1R and human MC4R by 63 and 67%, respectively, but at a 10-fold lower concentration (0.1 μM) inhibition of radioligand binding to all three targets was <10% and IC₅₀ values were determined to be >0.1 μM (Table S8). Ca_V2.2 calcium channels and MC4R are expressed in the nervous system, while MC1R is present in skin and hair. Although EgK5 did not exhibit in vivo toxicity (Tables S5–S7), further studies are required to exclude potential off-target effects of EgK5 on Ca_V2.2, MC1R, and MC4R.

EgK5 Suppresses the Proliferation of T_EM Cells

K_V1.3 channels play a critical role in T_EM cells, a subset implicated in many autoimmune diseases.^10,19−30 We tested whether EgK5 could suppress the proliferation of an ovalbumin-specific T_EM cell line that induces ovalbumin-induced dermatitis in rats.^27,66 EgK5 reduced K_V1.3 current-density (pA/pF) in human and rodent T cells (Figure S8k,l) and suppressed antigen-driven [³H]-thymidine incorporation of the ovalbumin-specific T_EM cell line at 0.1 μM (Figure 3 j).

Pharmacokinetic and Biodistribution Analysis Shows that EgK5 Targets Joints

We studied the pharmacokinetic and biodistribution profiles of EgK5. Following intravenous injection, EgK5 gave an apparent in vivo half-life of 1.3 h, and after subcutaneous injection of 1 mg/kg, EgK5 concentrations decayed over ∼8 h (Figure 4a; Tables S9,S10). For biodistribution studies, we labeled an EgK5 analogue containing a C-terminal azido-phenylalanine with ¹⁸F ([¹⁸F]EgK5-10-FADIBO). The labeled peptide was injected subcutaneously or intravenously into healthy rats and mice, and PET-CT imaging was performed. Rapid kidney uptake and appearance of the labeled peptide in the bladder was observed in both rats (Figures 4b,c) and mice (Figure S9a,b) suggesting urine as a route of excretion. However, pharmacokinetic studies suggested that only about 1% of EgK5 was detected in the urine; the reason for this difference is unclear. HsTX1[R14A] and dalazatide, cationic peptide inhibitors of K_V1.3, also accumulate in the kidney and bladder, and are believed to be excreted in the urine.^67,69 K_V1.3 expression in the kidney⁷⁰ may also contribute to the accumulation of K_V1.3-modulating peptides in the kidney. There was a smaller but significant hepato-biliary clearance and clearance via the intestines (Figure 4c, Figure S9b). Selective uptake in the bone marrow was observed but uptake in bone was low, indicating little defluorination in vivo (Figure 4d, Figure S9c). Retention in brain, muscle, blood, spleen, thymus, and lymph nodes was low. Interestingly, EgK5 showed significant accumulation in large (shoulder, elbow, knee, ankle, wrist) and small (fingers, toes) joints in both rats and mice (Figures 4b,d, and Figures S9a,c). EgK5 shares sequence-similarity with recently described joint-targeting cysteine-dense peptides.⁷¹ Structural analysis of these cysteine-dense peptides by crystallography revealed that the distribution of positive charge and their cysteine-stabilized tertiary folds underlie their arthrotropism.⁷¹Figure 4e shows the protein sequence alignment of EgK5, VvK1, and EgK1 with two arthrotropic peptides (CDP-11, CDP-09R) and two peptides (CDP-29, CDP-71R) that do not accumulate in joints. EgK5, EgK1, CDP-11, and CDP-09R contain a net positive charge greater than six, VvK1 has a net positive charge of 5, while CDP-29 and CDP-71R have a net positive charge of 1 and −1, respectively (Figure 4e). A comparison of the electrostatic potential shows that EgK5, EgK1, and CDP-09R contain large positively charged patches with only small negatively charged patches, while CDP-29 and VvK1 have larger negative patches (Figure 4f). These data suggest that EgK5’s arthrotropic property may be due to its positively charged electrostatic potential and disulfide-bridge-stabilized tertiary structure, although more work will be required to determine exactly why EgK5, CDP-11, and CDP-09R accumulate in joints.

Figure 4.

Stability, pharmacokinetic, and biodistribution profiles of EgK5. (a) Pharmacokinetic profile of EgK5 after intravenous and subcutaneous injection. Data shown as mean ± SD, n = 3 for each profile. (b) Representative coronal image displaying relative distribution of [¹⁸F]EgK5-10-FADIBO in rats summed from 90 to 120 min postinjection. (c,d) Excretion profile and biodistribution of [¹⁸F]EgK5-10-FADIBO in rats measured by longitudinal PET imaging (data shown as %ID/g ± SD, n = 4). (e) Alignment of protein sequences of EgK5 and joint-targeting cysteine-dense peptides (CDP-11, CDP-09R) and nonarthrotropic cysteine-dense peptides (CDP29, CDP-71R). Positively charged (blue), negatively charged (red), and cysteine (yellow) residues are shown. (f) Electrostatic potential of EgK5, VvK1, EgK1, CDP-09R [PDB ID: 6AY7] and CDP-29 [PDB ID: 6ATY]).

Toxicological Analysis Shows that EgK5 is Safe in Rodents

We assessed EgK5’s toxicological profile in rats administered subcutaneous injections of EgK5 at 0.1 or 1 mg/kg for 2 weeks. EgK5 was well tolerated with no evidence of mortality, adverse clinical signs, or changes in mean body weights, mean feed consumption, hematology, coagulation, clinical chemistry, and urinalysis (Tables S5–S7). There were no gross changes in any organ examined, and no histopathology findings in the EgK5-treated group.

EgK5 Treats Collagen-Induced Arthritis in Rats, a Model of Rheumatoid Arthritis

Given the joint accumulation of EgK5, we evaluated the peptide in rat collagen-induced arthritis, a T_EM cell-mediated model of rheumatoid arthritis.³⁰ Lewis rats immunized with collagen and incomplete Freund’s adjuvant developed arthritis within a week after receiving a booster of collagen and adjuvant emulsion. Treatment was started at disease-onset, after the first signs of arthritis (usually a swollen ankle joint = score of 2). At this time, rats were placed in vehicle or treatment groups. Rats were injected subcutaneously every other day with EgK5 (0.1 mg/kg) or vehicle. Vehicle-treated rats developed clinical scores of approximately 30, indicative of moderate inflammatory arthritis (Figure 5a). Rats treated with EgK5 exhibited significantly reduced clinical scores and bone erosions on X-ray compared to vehicle-treated rats (Figures 5a,b). Histological examination showed substantial inflammatory infiltrates of joints, cartilage damage, and bone destruction in vehicle-treated rats, which was reduced with EgK5 treatment (Figure 5c,d). VvK1, the defensin from grapevine, also ameliorated collagen-induced arthritis in rats, but not as effectively as EgK5 (Figure S10), possibly because its electrostatic potential (Figure 4f) makes it is less arthrotropic than EgK5. These results suggest that EgK5, due to its accumulation in joints and modulation of T_EM-mediated immune responses, may be a potential treatment for rheumatoid arthritis. Further, EgK5 shows no evidence of toxicity in vivo at 10 times the effective therapeutic dose.

Figure 5.

EgK5 treats collagen-induced arthritis in rats. (a) EgK5 (0.1 mg/kg) treatment started at disease onset and administered on alternate days by subcutaneous injection) reduced clinical scores of arthritis measured with a standard scale in which a score of 2 was given for each mildly red or swollen ankle or wrist, a score of 5 for each severely swollen ankle or wrist, a score of 1 for each red or swollen toe or knuckle joint, and the maximum possible score per paw was 15. (b) X-rays revealed that bone erosions and joint damage (indicated by yellow arrows) was decreased following EgK5 treatment. (c) Hematoxylin/eosin stained synovial hind paw joints of healthy rats, or rats with arthritis treated with vehicle or EgK5. The yellow arrowheads show inflammatory cells in vehicle-treated rats with arthritis that are significantly reduced in EgK5 treated rats. (d) Safranin-O/fast green staining of cartilage. In healthy joints the cartilage is stained red. Yellow arrowheads identify areas of cartilage that are destroyed in vehicle-treated rats with arthritis. EgK5 reduced joint damage as evidenced by the presence of intact cartilage. Scale bars = 100 μm. Statistics method: two-way ANOVA, ***p < 0.001.

EgK5 Treats Ovalbumin-Induced Dermatitis in Rats, a Model of Atopic Dermatitis

K_V1.3 inhibitors have been reported to be effective in rat ovalbumin-induced dermatitis,^22,27,72 a T_EM cell-mediated delayed type hypersensitivity model of atopic dermatitis.⁷³ We tested EgK5 in this model by administering EgK5 (0.01 and 0.1 mg/kg) or vehicle by subcutaneous injection at the time of ovalbumin challenge. Vehicle-treated rats developed severe inflammation as shown by almost a doubling in ear thickness. Ear thickness was reduced in EgK5-treated rats at both doses, indicative of decreased inflammation (Figure 6a). Hematoxylin and eosin staining showed no immune infiltrates in the ears of healthy animals (Figure 6b). Ears from vehicle-treated rats contained significant immune infiltrates, which were reduced with EgK5 treatment (Figure 6b).

Figure 6.

EgK5 treats ovalbumin-induced dermatitis. (a) Clinical scores. There was no difference between males (n = 6) and females (n = 5); we therefore pooled them. Statistics method: two-way ANOVA **p < 0.01. (b) Hematoxylin and eosin staining of ear sections. Representative sections are shown for each group of rats in addition to healthy control animals. Yellow arrows indicate edematous areas containing immune infiltrates. Scale bar = 200 μm. EgK5 (0.01 and 0.1 mg/kg) reduced clinical scores, immune infiltration, and edema in the ear.

Discussion

K_V1.3 channels in lymphocytes are widely regarded as therapeutic targets for autoimmune diseases.^10,29,35 Here, we describe a peptide modulator of K_V1.3 with a novel mechanism of action. We determined the structures of defensins from grapevine (VvK1) and oil palm (EgK1) that share sequence and structural similarity to K_V1.3-blocking scorpion toxins. Guided by molecular modeling and molecular dynamics simulation, we designed EgK5, a remarkably stable analogue of VvK1 and EgK1. EgK5 is resistant to intestinal proteases, stable over a wide pH range, and stable in both human and rodent plasma. EgK5 binds to purified K_V1.3 with submicromolar affinity, but it does not block K_V1.3 when applied acutely from the extracellular side, and competition experiments with MgTx (α-KTx-2.2) indicate that EgK5 does not bind in the channel’s external vestibule where many animal toxins bind.³¹⁻³⁵ EgK5 also does not block the channel when applied from the intracellular side, suggesting that the binding site is not internal. Using a TAMRA-tagged EgK5 analogue, we showed that EgK5 enters and accumulates in the plasma membrane. Prolonged exposure to EgK5 (3–24 h) results in concentration-dependent and specific suppression of K_V1.3 currents, which is not due to cytotoxicity or internalization of K_V1.3. This suppression is due to EgK5-induced K_V1.3 current run-down, which is prevented by the application of PIP₂ liposomes to the internal solution in patch-clamp experiments. These results suggest that EgK5 enters membranes, binds to K_V1.3 and decreases the number of ion-conducting K_V1.3 channels in the cell membrane by depleting PIP₂. By modulating K_V1.3, EgK5 inhibited antigen-triggered proliferation of T_EM cells, which are implicated in autoimmune diseases.

EgK5 was effective in treating disease in collagen-induced arthritis, a model for rheumatoid arthritis. EgK5 administered alone on alternate days reduced the arthritis score almost to baseline, and decreased inflammatory infiltration in joints, cartilage damage, and bone destruction. Dalazatide, a K_V1.3 inhibitor with picomolar affinity, was less effective in this model.³⁰ Dalazatide reduced the arthritis score by about 50% when administered alone, and only when combined with the K_Ca1.1 channel blocker iberiotoxin, which targets synovial fibroblasts, was the score reduced close to baseline.³⁰ The improved efficacy of EgK5 is possibly due to its accumulation in small and large joints, a property not exhibited by dalazatide.⁶⁹ Other cysteine-rich peptides have recently been shown to target joints by binding to the proteoglycan matrix of cartilage, their positively charged electrostatic potential and cysteine-stabilized tertiary structure being essential for arthrotropism.⁷¹ These peptides when conjugated to disease-modifying antirheumatic drugs (DMARDs) were effective in delivering the drugs to joints.⁷¹ EgK5 shares sequence similarity and a positively charged electrostatic potential like these joint-targeting cysteine-rich peptides. EgK5’s novel mechanism of K_V1.3 modulation, immunomodulatory activity, enhanced stability, and propensity to target joints makes it an attractive candidate to be both a therapeutic and a carrier to deliver DMARDs to joints for the treatment and pain relief of arthritic diseases.

EgK5 might have use in the treatment of other autoimmune diseases. The prevalence of T_EM cells in human atopic dermatitis lesions⁷³ and EgK5’s efficacy in a rat model of this disease suggest that EgK5 might be useful for this indication. Another potential clinical indication is lupus. In patients with active lupus nephritis, urinary T cells show the T_EM cell phenotype⁷⁴ and K_V1.3 is expressed in these cells with expression being higher in active lupus nephritis compared to inactive lupus or healthy controls.⁷⁵ Further, EgK5 accumulates in the kidney, the target organ. EgK5’s remarkable stability suggests it would be suitable for incorporation into slow-release depot formulations. While we have not determined immunogenicity of EgK5, other cysteine-rich peptides elicit weak immune responses in vivo possibly because their disulfide-constrained scaffolds hinder processing and antigen presentation by antigen-presenting cells.^20,35,69

Materials and Methods

Synthesis and Folding of Peptides

VvK1, EgK1, EgK5, EgK6, ShK, truncated VvK1, and EgK6 were produced synthetically with Fmoc solid-phase peptide chemistry using Fmoc-Cys(Trt)-Chem-Mtrix resin. These peptides were synthesized from the C-terminal amino acid in a linear format to the final N-terminal amino acid in a vectoral fashion. Following purification by preparative RP-HPLC, the linear peptides were oxidatively folded (Figure S1), and fractions deemed >95% pure by analytical HPLC and with the correct mass spectrum were pooled and lyophilized. EgK5 was conjugated to 5-TAMRA (tetramethy-rhodamine) via a AEEA (aminoethyloxyethyloxy acetyl) linker placed at the N-terminus of EgK5. Following oxidative folding, the cyclized product was purified by preparative RP-HPLC, and fractions with a purity of >90% were pooled, lyophilized, and characterized by analytical HPLC and ESI-MS (theory, 5808.5; experimental, 5806.8). VvK1, EgK1, EgK5, ShK, and truncated VvK1 were prepared by Peptides International (Kentucky, USA), and EgK6 and EgK5-TAMRA were prepared by AmbioPharm Inc. (South Carolina, USA). Details are provided in the Supporting Information.

Crystallization and Structure Determination of VvK1 and EgK1

VvK1 and EgK1 (33 and 40 mg/mL, respectively, in ultrapure water) were screened with commercial crystallization reagent kits (Hampton Research, Aliso Viejo, USA) and nanoliter-scale sitting-drop vapor diffusion using a Mosquito liquid handler robot (TTP Labtech, Melbourne, UK). VvK1 crystals were obtained at 20 °C by sitting drop vapor diffusion. Based on Rfree values during initial refinement, space group P43 was chosen over enantiomorphic P41. Space group P43 were grown in 15% PEG 4000, 0.2 M MgCl₂, 0.1 M Tris-HCl pH 8.5. Crystallographic data were collected at the MX1 beamline of Australian Synchrotron, integrated with XDS,⁷⁶ and diffraction intensities scaled with AIMLESS.⁷⁷ Tomato defensin TPP3 (PDB ID: 4UJ0)⁷⁸ was used as a molecular replacement model to solve VvK1’s structure with MOLREP.⁷⁹ Refinement was carried out using REFMAC5⁸⁰ and PHENIX⁸¹ and interspersed with manual model rebuilding sessions using Coot.⁸² The quality of the structure was analyzed with Molprobity.⁸³ EgK1 crystals were obtained by the hanging drop method in 0.2 M sodium tartrate dibasic dehydrate, 20% w/v PEG 3350. Prior to data collection, crystals were soaked for a few seconds in a cryoprotecting solution containing 20% (v/v) glycerol before being mounted and cooled to 100 K in liquid nitrogen. Crystallographic data for EgK1 were collected at the Proxima 2A beamline of the SOLEIL synchrotron (Saint Aubin, France). Integration, scaling, and merging of the intensities were carried out using programs MOSFLM and SCALA from the CCP4 suite.⁸⁴ Data were processed using XDS⁷⁶ in the triclinic P1 space group, and the structure was solved by molecular replacement using Phaser MR (CCP4 suite) with VvK1 as the search probe, and refined using BUSTER TNT (GlobalPhasing, Inc.). Figures were prepared using PyMOL.⁸⁵ Statistics for data collection and model refinement are presented in Table S1. The refined coordinates were deposited in the PDB under accession codes 7C2P (EgK1) and 7C31 (VvK1).

NMR Spectroscopy

Peptide stability was assessed by 1D ¹H NMR. To assess self-association of peptides in solution, translational diffusion coefficients were measured using a longitudinal eddy-current delay simulated echo sequence with presaturation for water suppression⁸⁶ and dioxane as a reference compound.⁵¹ Hydrodynamic radii and translational diffusion coefficients were calculated as described.⁸⁷ Further details are provided in the Supporting Information.

Molecular Dynamics Simulations

MD simulations of VvK1, truncated VvK1(10–44), EgK1, and EgK5 were performed for 100 ns in a water box for conformational sampling using NAMD (nanoscale molecular dynamics) software (http://www.ks.uiuc.edu/Research/namd/). At least three layers of water were included around peptides, and the system was ionized with KCl to 0.15 M and neutralized with additional Cl^– ions. The system was equilibrated with restrained peptide atoms at 300 K to obtain the correct water density with 1 atm pressure coupling. Then side chain and backbone atoms were slowly relaxed simultaneously from k = 5 kcal/mol/ Ų to 0, simulating the system for 200 ps at each step.⁸⁸ Final MD simulations were performed for 60 to 80 ns to sample different conformations of the peptides. Electrostatic potential was calculated with the Coulombic Surface Coloring function in UCSF Chimera using PDB files for VvK1, EgK1, CDP-09R, and CDP-29. For EgK5, the APBS program generated the electrostatic potential map and the pqr file was used to visualize and render the image in UCSF Chimera.

Assessment of Peptide Stability

EgK1, EgK5, VvK1, (peaks 1 and 2), truncated VvK1_10–44, and ShK peptides (25 μM) were incubated in the presence of trypsin or α-chymotrypsin (200 nM in 50 mM Tris, 100 mM NaCl, pH 7.4), pepsin in simulated gastric fluid (SGF, 200 nM in 35 mM NaCl, 225 mM HCl, pH 2.0), and pancreatin in simulated intestinal fluid (SIF, 2 mg/mL in 30 mM KH₂PO₄ and 100 mM KCl). Reactions with trypsin and chymotrypsin were quenched with 5% trifluoroacetic acid (1:5 ratio) while reactions with pepsin and pancreatin were quenched with 200 mM glycine/NaOH pH 11 (1:3 ratio). Samples were taken following 0, 0.5, 1, 2, 4, 6, 8, and 16 h of incubation at 37 °C. Peptide digestions were quantified by LC–MS (Shimadzu, Kyoto, Japan). Globular peptides are known to be stable to pepsin at low pH, hence contryphan-Vc1 was used as a control substrate for pepsin.⁸⁹ Reactions were conducted in triplicate (n = 3). Peptide stability was also determined in rat and human plasma as described in the Supporting Information.

Purification of Human K_V1.3 and Binding Studies with EgK5

Human K_V1.3 with a C-terminal His₁₀-tag was purified from Tetrahymena thermophila membranes by nickel-affinity chromatography and dialyzed against 20 mM HEPES pH 7.4, 150 mM KCl, 0.2 mM TCEP, 0.01% dodecylmaltoside (DDM), 0.0025% phosphatidyl choline (PC), and 0.2 μg/mL Psora-4.⁹⁰ The protein was analyzed for purity and structural integrity using size-exclusion chromatography (Superdex 200 Increase 5/150 GL) followed by running peak fractions on SDS-PAGE. Purified K_V1.3 in detergent (0.025% DDM) was incubated for 24 h with EgK5 or EgK6 (2:1 molar ratio = EgK5 (or EgK6):K_V1.3 tetramer) in 20 mM HEPES pH 7.4 buffer, containing 150 mM KCl, 0.025% DDM, and phospholipids (0.00036% PIP₂; 0.0025% PC/POPE [98:2]; PC = phosphatidyl choline, PIP₂ = phosphatidylinositol 4,5-bisphosphate, POPE = 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphoethanolamine) and then bound to HisPur cobalt resin (Thermo Scientific) for 90 min. After the flow-through was collected, the beads were washed three times with the above buffer to remove unbound peptide, followed by two elutions of K_V1.3 with 1 M imidazole. A control experiment where K_V1.3 was omitted was performed similarly. All fractions were analyzed by ABI 4800 Proteomics Analyzer MALDI TOF/TOF mass spectrometer (Applied Biosystems) using MALDI matrix α-cyano-4-hydroxycinnamic acid. For BLI, human K_V1.3 protein with an N-terminal His₆-tag was expressed in insect Sf9 cells. Whole cell extracts were obtained by resuspending cells in buffer (20 mM HEPES pH 7.4, 150 mM KCl) containing 1% DDM for 1 h at 4 °C followed by centrifugation. The binding of EgK5 to the His₆-K_V1.3 channel protein was measured and analyzed on an Octet Red96e instrument (FortéBio) at 25 °C. The buffer used in the BLI binding assay was 20 mM HEPES pH 7.4, 150 M KCl, 0.025% DDM, 0.0025% PC/POPE (98:2), and 0.1% BSA. Buffer-equilibrated anti-Penta-HIS (HIS1K) biosensors were loaded with the cell extract. A duplicate set of sensors without protein loading was used as a background binding control. The assay was performed with a working volume of 200 μL in black 96-well plates. Binding was performed using different concentrations of EgK5 (5, 2.5, 1.25, 0.625, and 0.312 μM). The data were fitted by the Octet Red96e software and analyzed by Microsoft Excel.

Electrophysiology Studies

The effects of EgK5 on K_V1.3 and other channels were evaluated by patch-clamp using a QPatch HTX automated electrophysiology platform (Sophion, Denmark) and/or by manual patch-clamp using protocols shown in Table S11. Peptides were prepared as 1–10 mM stock solutions in P6N buffer (10 mM sodium phosphate, 0.8% w/v, NaCl, and 0.05% v/v Tween 20, pH 6), diluted with 0.1% (w/v) bovine serum albumin (BSA) in external or internal buffer. For <1 h exposure experiments, peptides (EgK5, VvK1, EgK1, and MgTx) were added to patched cells in the whole-cell conformation after the second saline wash when currents were stable. Controls (MgTx, Psora-4, and PAP-1) were applied after the last dose of peptides. For 1–24 h exposure experiments, cells were seeded into T75 flasks for 72 h, then spent media was removed, and cells were incubated at 37 °C with serum-free media containing varying concentrations of EgK5 diluted in P6N buffer for 30 min before being replaced with complete media containing the specific concentrations of EgK5 for the indicated time points. Control cells were treated in a similar protocol with media containing P6N buffer at the same final concentration. For 72 h exposure experiments, cells were seeded, treated with media containing EgK5 or P6N buffer the next day, cultured for 48 h, and replenished with freshly prepared media containing EgK5 or P6N for another 24 h. Studies were also performed on human peripheral blood T lymphocytes isolated from buffy coats obtained from Health Sciences Authority, Singapore, using Lymphoprep (Axis Shield) density gradient centrifugation. Experiments were approved by the Nanyang Technological University Institutional Review Board (IRB-2015-07-039). Human T lymphocytes from healthy donors were activated for 48 h with anti-CD3/anti-CD28 antibodies (Thermo Fisher Scientific, USA) and IL-2 (Peprotech, USA), then treated with EgK5 in P6N buffer for 24 or 72 h and patch-clamped. Control cells were incubated in P6N buffer at the same final concentration.

Selectivity Analysis by Enzyme and Radio-Ligand Binding Assays

EgK5 was tested on additional molecular targets in enzyme, radioligand-binding, and patch-clamp assays, as described in Table S8.

Confocal Microscopy and Image Analysis

L929 fibroblasts expressing mouse K_V1.3 were treated with medium (control) or 10 μM EgK5 for 24 h, and fixed with 4% (v/v) paraformaldehyde at room temperature for 15 min. Cells were blocked with 5% (w/v) BSA and stained with guinea pig anti-K_V1.3 (extracellular) antibody (Alomone #AGP-005 at 1:1000 dilution) without permeabilization with Triton-X100. Mean fluorescence intensity analysis of K_V1.3 extracellular staining was carried out using Imaris software 9.5 (Andor-Bitplane) using the surface module. To test if EgK5 could cross the cell membrane, L929 fibroblasts expressing mouse K_V1.3 were treated with medium (control) or 10 μM TAMRA-EgK5 for 1, 3, and 24 h, washed 5 times with PBS, and fixed with 4% (v/v) paraformaldehyde at room temperature for 15 min. Cells were stained with Alexa Fluor 488 Phalloidin (ThermoFisher Scientific) for actin and with Hoechst (Sigma). Confocal microscopy was performed using the ZEISS LSM 800 Airyscan microscope with a Plan-Apochromat 40x/1.3 oil DIC (UV) VIS-IR M27 objective lens (Carl Zeiss). All imaging parameters were the same for all image acquisitions. Airyscan was used for T cell image acquisition (Figure S8i). Figures and intensity over distance plots were prepared with ZEN lite 2.1 (Carl Zeiss). Intensities were normalized for each cell against background. The presumptive membrane area was defined as the peak region of actin signal. The intracellular area was defined as the region between the two peak regions of actin signal. TAMRA-EgK5 accumulation within the membrane region was quantified using Imaris software 9.5. The cell membrane area was defined using the surface module based on actin staining, and then the mean intensity of TAMRA-EgK5 was measured within the defined area.

Assessment of Cell Viability

The viability of cells treated with EgK5 (0.1, 1, 10 μM) or control medium for 1–3 days was determined using the CellTier 96 Aqueous One Solution Cell Proliferation Assay kit (Promega) according to the manufacturer’s instruction. At the end of each assay, MTS reagent (3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium) was added to cells for 2 h at 37 °C before absorbance readings were recorded at 490 nm using a microplate reader (Tecan). Each experiment was performed in triplicate. For trypan blue dye exclusion, cells were mixed with trypan blue at room temperature for 2–4 min and examined visually immediately to quantify cells with clear (viable) or blue (nonviable) cytoplasm. Activated human T cells were treated with EgK5 or control medium for 1 and 3 days, stained with 7AAD (Biolegend) at room temperature for 10 min, and analyzed by a BD LSR Fortessa X-20 flow cytometer and Flowjo software (OR, USA).

T_EM Cell Proliferation Assay

Lewis rat ovalbumin-specific CD4⁺ T_EM cells transduced with GFP⁶⁶ (gift from Dr. Alexander Flügel, University Medical Center Göttingen, Germany) were stimulated with ovalbumin, and [³H]-thymidine incorporation was measured. T_EM cells (5 × 10⁴) were placed in each well of a 96-well plate and incubated with EgK5 for 45 min. They were stimulated for 72 h with 10 μg/mL ovalbumin in the presence of 2 × 10⁶ irradiated thymocytes as antigen-presenting cells. [³H] thymidine was added to the cells for the last 16–18 h of stimulation. DNA was harvested on glass filters and the incorporated [³H] thymidine was quantified by a β-scintillation counter (Beckman Coulter).

Pharmacokinetics and Biodistribution

The pharmacokinetics of EgK5 was assessed in Sprague–Dawley rats, and biodistribution was assessed in both Sprague–Dawley rats and Balb/c mice as described in the Supporting Information.

Collagen-Induced Arthritis

The therapeutic efficacy of EgK5 was determined in collagen-induced arthritis in male and female Lewis rats (8–11 weeks old).³⁰ Assays were approved by the institutional animal use and care committee at Baylor College of Medicine (approval AN-5341). EgK5 was dissolved in P6N buffer and administered by subcutaneous injection once every 2 days. We used 2-way ANOVA for statistical analysis. Male and female Lewis rats (8–11 weeks old) were purchased from Envigo (Indianapolis, IN) and housed in a facility approved by the Association for Assessment and Accreditation of Laboratory Animal Care (AAALAC) with water and food ad libitum. Rats were immunized with 200 μL of 2 mg/mL porcine collagen II (Chondrex) in a 1:1 emulsion with incomplete Freund’s adjuvant and received a booster injection of 100 μL the same emulsion 7 days later. Rats were monitored daily for clinical signs of arthritis. Arthritis severity was scored using a standard scale in which a score of 2 is given for each mildly red or swollen ankle or wrist, a score of 5 for each severely swollen ankle or wrist, and a score of 1 for each red or swollen toe or knuckle joint. The maximum score per paw is 15. Rats were randomly placed in the vehicle-treated or EgK5-treated groups at time of disease onset, which was defined as the occurrence of at least one red or swollen joint. Rats received vehicle (500 μL/rat) or EgK5 (0.1 mg/kg in vehicle; 500 μLrat) by subcutaneous injection in the scruff of the neck starting from the onset of clinical signs and continued every other day until the end of the trial. At the end of the trial, the rats were humanely euthanized for collection of their hind limbs. Paws were X-rayed using an In-Vivo Xtreme Imaging System (Bruker BioSpin, Billerica, MA, USA) to assess the presence of abnormal bone structures and erosions. Paws were then fixed in 10% buffered formalin, decalcified, sectioned, and stained with either hematoxylin and eosin to detect inflammatory infiltrates or with Safranin O/Fast Green to detect cartilage. Images of synovial joints were taken with an Olympus BX41 microscope equipped with an Olympus Q Color 5 camera at 10× magnification (Olympus, Center Valley, PA, USA).

Ovalbumin-Induced Dermatitis

The therapeutic efficacy of EgK5 was determined in ovalbumin-induced dermatitis, a delayed-type hypersensitivity model of atopic dermatitis,⁷² in 18 male and 15 female Lewis rats (8–11 weeks old). The assays were approved by the institutional animal use and care committee at Baylor College of Medicine (approval AN-4950). EgK5 was dissolved in P6N buffer and administered by subcutaneous injection at the time of ovalbumin dermal-challenge. We used 2-way ANOVA for statistical analysis. Lewis rats were divided into three groups, with each group consisting of 6 males and 5 females. Since there is a large difference in weight between the two sexes, rats were distributed in the three groups to match average weights. They were immunized with 200 μL of a 1:1 emulsion of 2 mg/mL ovalbumin (Sigma) and complete Freund’s adjuvant (Difco/Becton Dickinson, Franklin Lakes, NJ). Seven days later, 20 μg of ovalbumin in saline was subcutaneously injected into the pinna of one ear and saline was administered in the other ear under isoflurane anesthesia. They were immediately treated with vehicle or EgK5 (10 or 100 μg/kg) subcutaneously in the scruff of the neck. Ear thickness was recorded 24 h later by an investigator blinded to the treatment groups using a spring-loaded micrometer (Mitutoyo, Melville, NY). The difference in thickness between the two ears for each rat was used as a measure of inflammation. Ears were collected after measurements, fixed in 10% buffered formalin, embedded in paraffin, sectioned, and stained with hematoxylin and eosin to detect inflammatory infiltrates.

Toxicity Analysis

The systemic/local toxicity profile of EgK5 was assessed in Sprague–Dawley rats in a non-GLP study conducted in accordance with the Principles of Good Laboratory Practices at Syngene Limited (Bengaluru, India), a GLP-compliant facility following Standard Operating Procedures. A dose volume of 5 mL/kg body weight was used across the dose groups. The in-life phase observations/parameters evaluated included mortality check, clinical signs observation, detailed clinical examination, body weights, and feed consumption. After completion of the treatment period, on day 15, blood and urine samples were collected from all animals for clinical pathology evaluation (hematology and coagulation, clinical chemistry, or urinalysis). Subsequently, all animals were euthanized in a stratified manner by exsanguination under deep isoflurane anesthesia. All animals were subjected to gross examination, and organ weights were determined. All the study plan specified tissues were collected for histopathological evaluation.

Supporting Information Available

Supporting Information includes Supporting Information and Methods, . (Table S8, and electrophysiological protocols Table S11). The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsptsci.0c00035.

• Methods including synthesis and folding of peptides, NMR spectroscopy, assessment of peptide stability, analysis of EgK5 in plasma and urine samples, pharmacokinetic studies in Sprague–Dawley rats, radiochemical synthesis of [¹⁸F]EgK5-10-FADIBO, imaging of radiolabeled [¹⁸F]EgK5-10-FADIBO, and biodistribution in rodents, selectivity screen protocols; Tables S1–S11 and Figures S1–S10 (PDF)

• Z-stack images of L929 fibroblasts stained with TAMRA-EgK5 (red), actin (green), and Hoechst (blue) acquired at 1 μm intervals. Scale bar = 5 μm (AVI)

Author Contributions

K.G.C. identified VvK1 and EgK1 sequences and conceived and managed the project. M.W.P. and V.D. synthesized VvK1, EgK1, EgK5, EgK6, truncated variants, and TAMRA-EgK5. S.C.C., C.M.W., D.L., E.S., and J.L. determined the structures of VvK1 and EgK1 by crystallography. R.S.N., B.K., and R.A.V.M. performed ¹H 1D and 2D-NMR experiments on the defensins and determined their stability to gastric and pancreatic proteases and over a range of pH. S.K., D.P., and S.Y. performed the molecular dynamics simulation studies and contributed to the design of EgK5. K.G.C., S.K., R.S.N., and M.W.P. designed EgK5. O.S.T., N.X.R., Z.Z., H.N., H.W., C.B., and K.G.C. designed and performed the patch-clamp experiments. S.B., N.J.H., J.B., and P.C. purified K_V1.3 and performed the EgK5 and EgK6 binding assays. R.P. and S.A.C. developed a pharmacokinetic assay for EgK5, determined its pharmacokinetic properties in rats, and determined its stability in rat and human plasma. E.G.R., J.L.G., P.W.T., P.S., and B.R. radio-labeled the peptide and performed the biodistribution studies. M.R.T. and C.B. assessed EgK5′s effect on T_EM cell-proliferation and evaluated EgK5 and VvK1 in the rat models of rheumatoid arthritis and atopic dermatitis. K.G.C., O.S.T., and S.B. wrote the paper and prepared figures, with contributions from all other authors.

The authors declare no competing financial interest.