Targeted inhibition of the Shroom3–Rho kinase protein–protein interaction circumvents Nogo66 to promote axon outgrowth

Recombinant proteins

GST-SD2, a GST fusion to the mouse Shroom3 SD2 domain (aa 1,563–1,986) expressed in
pGST-SD2, and HisSUMO-R2C1, a His-Sumo epitope fusion to mouse ROCK R2C1 (aa 698–957)
were produced in Escherichia coli. Briefly, E. coli were lysed by sonication in PBS+ buffer (GST purification: PBS with 0.1 mM phenylmethylsulfonyl
fluoride, 14 µg/mL aprotinin, 0.1% ?-mercaptoethanol, 1 µM leupeptin, 1 µM pepstatin)
(His purification: PBS with 0.1 mM phenylmethylsulfonyl fluoride, 14 µg/mL aprotinin,
0.1% ?-mercaptoethanol, 1 µM leupeptin, 1 µM pepstatin, 25 mM imidazole). Triton X-100
was added to the lysate at 1% of the final volume. Lysates were incubated with prewashed
glutathione agarose or HisPur Ni-NTA resin (Thermo Scientific) for 1 h at 25°C. Purified
protein was eluted with GST elution buffer (50 mM Tris buffer with 100 mM reduced
glutathione, pH 8) or His elution buffer (PBS+ with 250 mM Imidazole). HisSUMO-R2C1
was dialyzed overnight at 4°C in PBS and stored in 25% glycerol at ?20°C. GST-SD2
was dialyzed for 3 h at 4°C in PBS with three buffer changes. The GST epitope tag
was removed using His-TEV (S219V)-Arg Protease overnight at a concentration of 1 µg
TEV protease per 100 µg of GST-SD2. TEV protease and free GST was removed from purified
SD2 by incubation overnight at 4°C with prewashed glutathione agarose and HisPur Ni-NTA
resin. SD2 was stored at ?20°C in 25% glycerol.

HisSUMO-R2C1 was biotinylated (NHS-PEO₄-Biotinylation Kit, Pierce). Briefly, biotinylation reactions were carried out at
a 20:1 molar ratio of NHS-PEO₄-biotin to HisSUMO-R2C1 in PBS (pH 7.4). HisSUMO-R2C1 was labeled for 2 h at 4°C.
After the incubation, the unreacted NHS-PEO₄-biotin was removed with buffer exchange in PBS (pH 7.4) using Zeba Desalt Spin Columns
(2 mL, MWCO = 7,000 Da) (Pierce). The average extent of labeling for HisSUMO-R2C1
was estimated to be four biotin molecules per 1 mol of protein using the HABA assay,
a measurement of the extent of biotinylation, as per the manufacturer’s protocol.
Biotin-R2C1 was stored at ?20°C in 25% glycerol.

For the pull-down assay in Figure 4c, wild type or mutant GST-SD2 (5 ?g), bound to glutathione agarose resin, was incubated
overnight with His-Sumo-R2-C1 (12.5 ?g). Pull down samples were washed 3× in Triton
IP buffer and 1× in PBS prior to SDS-PAGE. Proteins were detected by Coomassie G-250
(Gel Code Blue, Pierce). 1 ?g of SD2 or R2-C1 was included as a gel loading (input)
control.

Affinity determination and competition assays

Apparent binding affinity (K_d) was determined by immobilizing 0.5 µg of SD2 diluted in 75 µL PBS for 16 h at 4°C
on 96-well Immulon 2B high binding plates (Thermo Scientific). Plates were blocked
for 1 h at 25°C in SuperBlock T20 (TBS) Blocking buffer (Thermo Scientific). Concentrations
of Biotin-R2C1 diluted in TBS-1 (20 mM Tris HCl, 150 mM KCl, 0.5% Triton X-100, pH
7.9) with 0.5% bovine serum albumin (BSA) were added from 0 to 1,778 nM for a total
of 11 concentration points for 1 h at 25°C. Unbound protein was removed with four
washes in TBS-2 (20 mM Tris HCl, 300 mM KCl, 0.5% Triton X-100, pH 7.9). High Sensitive
NeutrAvidin-HRP was added at a dilution of 1:40,000 in TBS-3 (25 mM Tris HCl, 8.25 mM
Tris Base, 154 mM NaCl, 2% BSA, 0.05% Tween-20) for 1 h at 25°C. Excess SUMO antibody
was removed with 4 washes in TBS-T (25 mM Tris HCl, 137 mM NaCl, 2.7 mM KCl, 0.1%
Tween-20). TMB substrate (Pierce) was added for 15 min and quenched with 0.18 M H₂SO₄. Absorbance was measured at 450 nm using a SpectraMax M5 microplate reader. The K_d was calculated using GraphPad Prism 4.0 using a hyperbolic fit with a non-zero intercept
(?A = ?A_max*[R2C1]/(K_d + [R2C1]). ?A = absorbance change; ?A_max = maximum absorbance change; [R2C1] = R2C1 concentration. For competition ELISAs,
100 ng of Biotin-R2C1 was incubated for 1 h at 25°C with unlabeled R2C1 (1–10 µg),
and processed as described above.

High-throughput screening

Primary screen

Initial assay development was performed in 96-well Immulon 2B high-binding plates
(Thermo Scientific). The assay was then optimized for high-throughput screening in
384-well plates. 20,000 compounds (ChemDiv) were screened in the Center for Chemical
Genomics at the University of Michigan. All reagent additions were performed using
Thermo Labsystems Multidrop, and plate washes were performed using Bio Tek EL406 washer/aspirator.
150 ng of SD2 diluted in PBS (pH, 7.4) was immobilized for 16 h at 4°C on 384-well
high-binding plates (Perkin Elmer). Unbound protein was removed with two washes of
PBS (pH 7.4). Plates were blocked for 1 h at 25°C in SuperBlock T20 (TBS) Blocking
buffer (Thermo Scientific), followed by two washes with PBS (pH 7.4). 20 µL of buffer
A (20 mM Tris HCl, 150 mM KCl, 0.05% Triton X-100, 0.5% BSA, pH 7.9) was added to
all wells. 200 nL of compounds were pin-tooled (one per well) into columns 3–22 resulting
in a final concentration of 10 µM using the Biomex FX (Beckman). 200 nL of DMSO was
added to control columns 1–2 (negative control) and 23–24 (positive control). 5 µg
of unlabeled R2C1 in 20 µL of buffer A was added to columns 23–24 as a positive control
for inhibition. After 30 min at 25°C, 10 ng of Biotin-R2C1 in 10 µL of buffer A was
added to all wells and incubated for 1 h at 25°C. Unbound protein was removed with
three washes (buffer A supplemented with 300 mM KCl and 0.5% Triton X-100). 40 µL
of 1:40,000 NeutrAvidin-HRP diluted in TBS-3 was added to all wells and incubated
for 45 min at 25°C. Plates were washed three times in TBS-T, followed by the addition
of 20 µL of TMB substrate for 5 min. The TMB reaction was quenched with 20 µL 0.18 M
H₂SO₄. Absorbance was measured at 450 nm using an automated PHERAstar plate reader (BMG
Labs). Hits were defined as percent inhibition greater than 3 standard deviations
(3 SD) from the mean of the negative control for inhibition (159 actives). Additional
active samples based on percent effect values and standard deviation values were also
analyzed (6 and 15 samples, respectively).

Dose response and hit selection criteria

Dose–response confirmation (180 compounds) was performed following the ELISA screening
platform. Compound dilutions of 100, 59.8, 35.9, 21.5, 12.9, 7.69, 4.61, and 2.70 µM
were delivered using the Mosquito X1 (TTP Labtech) in duplicate. Compounds with at
least 30% inhibition and a pIC50 of 3.5 were considered active (74 compounds). Compounds
with greater than 22% promiscuity and less than 60% efficacy were removed. The application
of these selection criteria resulted in 36 compounds. IC₅₀ values were calculated with GraphPad Prism 4.0 using nonlinear regression and the
log (inhibitor) vs. response, variable slope equation. Clustering was performed using
DataMiner by the Tripos algorithm OptiSim under the criteria of 65% or greater similarity.

Dose response with fresh powder

32 of the 36 hits from the primary screen were available for purchase from ChemDiv
and were diluted in DMSO and stored at ?20°C. 150 ng of SD2 diluted in 25 ?L PBS was
immobilized for 16 h at 4°C on 384-well high-binding plates (Perkin Elmer). Following
blocking, serial compound dilutions of 1,000, 300, 100, 30, 10, 3, 1, and 0.3 ?M were
delivered in triplicate and incubated for 30 min at 25°C. 10 ng of biotin-R2C1 was
added (1.5 ng in Figure 4) and incubated for 1 h at 25°C then the ELISA was completed as described in primary
screen. 27 of the 32 compounds were confirmed as active hits. Compounds with an IC₅₀ of less than 30 ?M were prioritized (nine compounds). The IC50 values were calculated
using GraphPad Prism with nonlinear regression and the log (inhibitor) vs. response,
variable slope equation.

DTT and reversibility experiments

After plating 150 ng SD2, 1 mM DTT was added and was present throughout the assay.
CCG-17444 was added at a concentration of 75 ?M prior to addition of 1.5 ng biotin-R2-C1.
To test reversibility, wells were washed five times for 2 min each with buffer A after
a 30-min incubation of 75 ?M CCG-17444 with SD2, prior to addition of biotin R2-C1.
Biotinylated R2-C1 was detected with neutravidin HRP as described above. Absorbance
was measured at 450 nm using a Tecan safire² microplate reader.

Neurite outgrowth experiments

P19-derived neurons

P19 neurons were generated by transfection of pluripotent P19 embryonal carcinoma
cells with the neural basic helix-loop-helix transcription factor Ngn2. P-19 derived
neurons express neuronal markers, adopt a neuronal morphology, and are electrically
excitable 21]. A GFP expression vector is included in the transfection and readily enables the
identification of the transfected cells. GFP is present both in the cell body and
in neurites. For neurite length measurements, the length of the longest neurite per
cell (ca. 3× cell body or 50 ?m) was measured; GFP positive neurites are positive
for the axon marker Tau 16] and the neuron specific tubulin TuJ1 21], 22]. P19 cells, grown in minimal essential medium-? supplemented with 7.5% calf serum,
2.5% fetal bovine serum, and penicillin–streptomycin were plated the day before transfection
to a density of 2 × 10⁵ cells/well of a 12-well dish and transfected the next day with 2.5 µg of total DNA
(0.75 µg of Ngn2, 0.75 µg of GFP, and 1 µg of pUI4/UI5 RNAi expression vectors). In
Figure 5, 1.25 µg of control or an SD2 expression vector (SD2 amino acids 1,768–1,986) was
included with Ngn and GFP. RNAi vectors targeting POSH and Shroom3, as well as control
luciferase RNAi vectors, have been described previously (CS2 Vector Resource, RRID:nif-0000-3027)
14], 16], 43], 44]. 4–6 h after transfection, cells were re-split 1:5 or 1:6 to laminin-coated dishes
(2 ?g/ml, Invitrogen). 18–20 h later, the cells were transferred into Opti-mem supplemented
with 1% FBS and penicillin–streptomycin. CCG-17444 was added 48 h after transfection
at final concentration of 25 ?M (Figures 3, 5). 72 h after transfection, cells were fixed in 3.7% formaldehyde in PBS and stained
for GFP to identify transfected cells 14], 16], 21]. Similar results are obtained when neurite length is determined by staining for GFP
or neuron specific markers.

Cerebellar granule neurons

CGNs were isolated and plated in the presence or absence of Nogo66 and neurite length
determined 30 h post-plating, as described 9], 14], 45], 46]. All care and procedures for mice were performed in accordance with the guidelines
and approval of the University of Michigan University Committee on Use and Care of
Animals. 12-well dishes (Corning) were coated with 10 ?g/ml poly-L-lysine for 4 h then overnight with 2 ?g/ml laminin at room temperature, or laminin
plus bacterially expressed His-SUMO Nogo66 (0.75 ?g/cm²) at 4°C. Postnatal d8 cerebellar granule neurons (CGNs) were nucleofected as described
previously with a total of 6 ?g DNA 14]. 6 h post nucleofection, Y-27632 (10 ?M, Calbiochem, San Diego, CA, USA), a ROCK
1/2 inhibitor, or CCG-17444 (25 ?M of 3347-0032, ChemDiv, San Diego, CA, USA), were
added to CGNs in culture media [DMEM (Invitrogen) supplemented with 2% B27 (Invitrogen),
and 1% penicillin/streptomycin (Invitrogen)]. Cells were fixed in 3.7% formaldehyde
in PBS 24 h post inhibitor addition. Cells were stained with anti-GFP primary antibody
[Molecular Probes (Invitrogen) Cat# A11122 RRID:AB_221569] followed by detection with
Alexa Fluor 488 goat anti-rabbit (Life Technologies Cat# A11034 RRID:AB_10562715).

Measurement of neurite length

Photographs of GFP-positive neurons were imaged with a digital camera; the length
of the longest neurite per cell (50 ?m or greater, ~3× the cell body) was measured
using the polyline function in MicroSuite Special Edition imaging software. For assessment
of inhibition by Nogo66, neurites less than 50 ?m were also measured. At least 100
neurons per condition per experiment were quantitated for neurite length in three
independent experiments.