Characterization of New Cell Permeable C3-Like Proteins That Inactivate Rho and Stimulate Neurite Outgrowth on Inhibitory Substrates

five new C3-like chimeric designed cross These the addition of short transport peptides to the carboxyl terminal of C3 and tested a measuring neurite outgrowth of PC-12 cells plated on inhibitory All five dose-response using C3-05, shift assays showed C3-05 ability to ADP-ribosylate Western blots and immunocytochemistry used to the of C3 C3-05 was also effective at neurite outgrowth in well the disassembly and in These new active C3 into different Tat (C3-02, C3-03); Antennapedia homeodomain (C3- 04); a proline-rich fusogenic peptide sequence (C3-05); and a highly basic, arginine-rich sequence corresponding to the reverse Tat sequence (C3-06). We provide evidence that these five new C3-like chimeric proteins all cross the plasma membrane, inactivate Rho, and stimulate neurite outgrowth on inhibitory substrates.


INTRODUCTION
Rho GTPase regulates the actin cytoskeleton and cell motility in response to extracellular signals. Initial studies using Swiss 3T3 fibroblasts demonstrate the ability of Rho to regulate the formation of actin stress fibers and focal adhesion complexes in non-neuronal cells (1). In neurons, Rho plays a key role in determining the response of axons to growth inhibitory proteins. GTPases have two conformations: a GDP-bound inactive state and a GTP-bound active state (2). The activation of Rho in neurons causes growth cone collapse, neurite retraction and cell body rounding (3)(4)(5). Treatment with C3-ADP-ribosyltranferase (C3) 1 , a specific inhibitor of Rho, stimulates axon growth and regeneration (6,7). To be effective, this 24 kDa protein must cross the plasma membrane and interact with intracellular Rho, however, C3 does not easily enter cells. To date, various methods have been used to help facilitate the entry of C3 into cells. In experiments using fibroblasts, C3 is microinjected into individual cells (1), whereas in studies using neuronal cells, triturating (8), or scrape loading techniques (7) are used to aid cellular entry. The need for such disruptive methods to inactive Rho by C3, and the inability to treat all cell types with equivalent techniques has limited the use of C3 as a tool for biochemical studies on Rho signalling. One solution has been to create a fusion protein that increases the efficiency of C3 delivery across the membrane. One such protein, a fusion between C3 and the B subunit of diphtheria toxin (DT), binds to cell surface DT receptors and is internalized by an endocytosis-mediated mechanism (9). This fusion protein is only effective in cells that contain DT receptors, therefore excluding most rodent cells (9). A fusion protein between the bacterial toxin C2 and C3transferase, C2IN-C3, also increases the ability of C3 to cross the cell membrane by receptor-mediated endocytosis (10). However, C2IN-C3 cannot independently cross the cell membrane because it requires the presence of the C2 toxin-binding component, C2II, to enter cells (10,11). Both of these fusion proteins enter cells by receptor-mediated endocytotic pathways, and therefore, may be trapped within vesicles, which may lessen efficient interaction with Rho.
Small peptides can act as carriers by transporting large protein cargo across cell membranes. Such peptides are part of larger proteins that are able to cross biological membranes. A series of different classes of transport peptides exist: 1) the human immunodeficiency virus transcription activator (Tat) contains a region spanning amino acids 37 to 72, which translocates its cargo to both the cytosol and nucleus (12,13). A shorter Tat sequence, spanning amino acids 48 to 60, is also effective (14). 2) The third helix of the Antennapedia homeodomain (Antp), a Drosophila homeoprotein, possesses the ability to cross biological membranes. Experiments using biotinylated forms of this 16 amino acid peptide have confirmed its ability to penetrate cells and locate in both the cytosol and nucleus (15)(16)(17). 3) Prolines are functional participants in some transport peptides (18). Proline residues act as helix breakers and form turn structures within peptides. Peptides rich in proline can form conformations that help in membrane translocation (19). 4) Peptides of 7 amino acids in length or longer that contain basic, arginine-rich sequences can act as effective transport peptides (20). 5) The hydrophobic regions of several membrane transport sequences (MTS) can translocate across the cellular membrane and accumulate in the nucleus (21). 6) Recently, a short amphipathic peptide carrier, Pep-1, was also shown to be able to translocate across cellular membranes (22). We have made and tested four different classes of transport peptides: PCR, a 60 amino acid region encoding the full-length homeodomain, using primers: 5'GGAATCCCGCAAACGCGCAAGGCAG 3' and 5'TCAGTTCTCCTTCTT CCACTTCATGCG 3'. The PCR product was cloned into pSTBlue-1 and subcloned into pGEX-4T/C3 using EcoRI, creating C3-05. Sequencing of this construct revealed a deletion mutation that altered the primary amino acid sequence, giving a proline-rich sequence resembling fusogenic peptides (23), and thus it was kept and tested. C3-06 was constructed by oligonucleotide sequences that coded for a highly basic and arginine-rich peptide corresponding to the reverse Tat sequence, 5'AATTCAGAAGGAAAC AAAGAAGAAAAGAAGACTGCAGGC 3'and 5' GGCCGCCTGCAGTCT TCTTTTTCTTCTTTGTTTCCTTCTG 3'. These oligonucleotide sequences were annealed and ligated into pGEX-4T/C3, at EcoRI and NotI restriction sites. Plasmids were transformed into XL-1 blue competent cells except C3-06, in which DH5α competent cells were used. Plasmids were sequenced through the fusion region to the end of the peptide.  Recombinant C3 and C3-like chimeric proteins were purified by affinity chromatography. Bacteria were grown in L-broth (10 g/L Bacto-Tryptone, 5 g/L yeast extract, 10 g/L NaCl) with 50 µg/ml ampicillin (Roche, Québec, Canada), in a shaking incubator for 1.5 hrs at 37°C and 300 rpm. Isopropyl βdithiogalactopyranoside (IPTG), (Gibco, Burlington, Ontario, Canada) was added to a final concentration of 0.5 mM, to induce the production of recombinant protein, and then the cultures were incubated for another 6 hours at 37°C and 250 rpm. The bacteria were pelleted by centrifugation in a GSA rotor (Sorval, Superspeed Centrifuge) at 7000 rpm for 6 minutes at 4°C. Each pellet was re-suspended in 10 ml of buffer A (50 mM Tris, by guest on March 24, 2020 http://www.jbc.org/ pH 7.5, 50 mM NaCl, 5 mM MgCl 2 , 1 mM DTT plus PMSF (1 mM). All re-suspended pellets were pooled and transferred to a 100 ml plastic beaker on ice. The bacterial suspension was sonicated for 6 x 20 seconds, on ice, using a Branson Sonifier 450 probe sonicator (VWR, Québec, Canada). The lysate was centrifuged twice in a Sorvall SS-34 rotor at 16,000 rpm for 12 minutes at 4°C to clarify the supernatant. Glutathione-agrose beads (Sigma, Oakville, Ontario, Canada) were added to the cleared lysate and the preparation was placed on a rotator for 2-3 hours. The beads were washed 4 times with Buffer B, (Buffer A + 100 mM NaCl) and 2 times with Buffer C (Buffer B + 2.5mM

Preparation of Recombinant Proteins
CaCl 2 ). The final wash was removed until a thick slurry was created, and 20 U of thrombin (Calbiochem, San Diego, California) was added and the beads were shaken overnight at 4°C. The beads were loaded into an empty 20 ml column and 1 ml aliquots were collected after elution with PBS. The fractions containing the protein peak were pooled. To remove the thrombin form the protein sample, 100 µl of p-aminobenzamidine agarose beads (Sigma, Oakville, Ontario, Canada) were added and left mixing for 45 minutes at 4°C. The protein was centrifuged to remove the beads and then concentrated using a centriprep-10 concentrator (Millipore, Ontario, Canada). The concentrated protein was desalted by PD-10 column containing Sephadex G-25M (Amersham Pharmacia, Québec, Canada) and 10, 0.5 ml aliquots were collected. The appropriate aliquots were pooled, sterilized by filtration, and stored at -80°C. Concentration of The specificity of the antibody was tested by western blot (Fig. 1). PC-12 cells were grown and plated on myelin-coated slides as described above. The cells were treated with 10 µg/ml of C3-05, or C3 added directly to the media and incubated at 37ºC for 24 hours. The cells were fixed with 4% PFA and probed with an anti-C3 antibody followed by FITC staining.
Lysates were clarified by centrifugation and 10 µg of protein was separated on 12% acrylamide gels. After transfer to nitrocellulose, the membranes were either blocked with TBS containing 0.1% Tween 20 (TBS-T) and 3% BSA and incubated in blocking buffer with an anti-RhoA antibody (1:1000) (Santa Cruz, Santa Cruz, California), or blocked with 5% powered milk and incubated in blocking buffer with an anti-C3 antibody (1:4000). The signals were revealed by an HRP-based chemiluminescent reaction (Pierce, Rockford, IL). Membranes probed with anti-RhoA antibody were stripped and re-probed with an anti-Cdc42 antibody (1:1000) (Santa Cruz, Santa Cruz, California).
Pull down assays to detect Rho-GTP  The activity assays were preformed as previously described (25,26). PC-12 cells were grown on poly-l-lysine, or myelin coated 6 well culture dishes. After the cells settled (3-6 hours at 37°C), the media was aspirated and fresh media containing the test C3-like chimeric proteins was added to the cultures. NIH 3T3 cells were plated in 6 well culture dishes and incubated at 37ºC for 3-6 hours. After the cells settled, the media was aspirated and fresh media containing the test C3-like chimeric proteins was added to the cultures. At indicated times (Fig. 8B), the cells were washed with ice cold Tris buffered saline (TBS) and lysed in RIPA buffer (50 mM Tris pH 7.2, 1% Triton X-100, 0.5% sodium deoxycholate, 0.1% SDS, 500 mM NaCl, 10 mM Densitometry values for untreated cells plated on myelin were normalized to correspond to100% Rho activation. C3-05 treated cells were calculated as the percent inactivation compared to the normalized values.

RESULTS
We chose to test a number of different strategies to design C3-like chimeric proteins that could cross the plasma membrane by receptor independent mechanisms.
Five C3-like chimeric proteins were constructed by adding DNA sequences encoding known membrane translocating peptides derived from Tat (C3-02, C3-03), Antennapedia (C3-04), a proline-rich fusogenic peptide (C3-05) and a basic, arginine-rich peptide (C3-06) to the 3' end of the C3 cDNA. All five cDNAs encoding the C3-like chimeric proteins were expressed as GST fusion proteins in E.coli, purified, and their molecular weights verified by SDS-PAGE gel (eg. Fig. 1A). To confirm the presence of C3 in all five constructs, western blots using a polyclonal antibody raised against C3 were completed (eg. Fig.1B).
PC-12 cells typically extend neurites in response to NGF, but when plated on inhibitory substrates, this outgrowth is inhibited and the cells remain round (7). The ability to inactivate Rho and promote neurite outgrowth on inhibitory substrates was used as a bioassay to test the effectiveness of the new C3-like chimeric proteins. First, we examined the dose-response profile of unmodified C3. In previous experiments to inactivate Rho, we determined that scrape-loading was necessary to treat PC-12 cells with C3 (7). We found that even with the scrape-loading technique, high concentrations by guest on March 24, 2020 http://www.jbc.org/ of C3 were required to stimulate neurite outgrowth on myelin (Fig. 2). When C3 was added to the culture medium of pre-plated cells, C3 had no significant effect (Fig. 2).
To test the ability of the new C3-like chimeric proteins to promote neurite outgrowth, we performed dose response experiments with PC-12 cells plated on myelin substrates (Fig. 3). In these experiments, the C3-like chimeric proteins were added directly to the culture medium after pre-plating cells. This procedure was carefully preformed to avoid any mechanical disruption of the cells. To establish the effective concentration ranges, preliminary experiments included a test concentration of 0.00025 µg/ml were completed, but none of the C3-like chimeric proteins were effective at this dose. Surprisingly, concentrations of 0.0025 µg/ml of C3-03, C3-05, and C3-06 lead to significant increases in both the number of cells extending neurites and the length of neurites compared to cells plated on myelin without treatment (Fig. 3A-B). This effective dose is 10,000-fold lower than that required with unmodified C3 using scrapeloading techniques (Fig. 2). At the highest concentrations tested, C3-03 was not effective, perhaps because of toxicity associated with the Tat sequence (14). C3-02 and C3-04 promoted significant neurite outgrowth at concentrations of 0.25 µg/ml and 2.5 µg/ml respectively and an increase in neurite length was observed at concentrations between the range of 0.025 µg/ml and 2.5 µg/ml (Fig.3A-B). At all concentrations, C3-05 and C3-06 stimulated neurite outgrowth on myelin substrates, with C3-05 giving the best results.
These results indicated that all of the new C3-like proteins had some capacity to penetrate cells, inactivate Rho, and promote neurite outgrowth on inhibitory substrates. As C3-05 gave the best results, it was used for further testing.
ADP-ribosylation of Rho causes it to migrate with a larger apparent molecular weight on SDS gels (7,27). To study the ability of C3-05 to ADP-ribosylate Rho, we examined the electrophoretic mobility of RhoA by Western blot of cell lysates treated for 24 hours with 25 µg/ml of scrape-loaded C3, or 10 µg/ml of C3-05 added directly to the media (Fig. 4, top). Both scrape loaded C3 and C3-05 caused a similar molecular weight shift, confirming the ability of C3-05 to ADP-ribosylate Rho (Fig. 4, top). As a control for the specificity of this molecular weight shift effect, we stripped and re-probed the same blots for another member of the Rho GTPase family, Cdc42 (Fig. 4, bottom). Cdc42 did not show any change in mobility after treatment with C3-05, demonstrating that C3-05 maintains the same ADP-ribosylation specificity as unmodified C3.
We compared by immunocytochemistry the ability of C3 and C3-05 to enter cells and C3-05 was clearly detected in the cell lysates of both fibroblast cell lines tested (Fig. 5B).
By immunocytochemistry of PC-12 cells plated on myelin, we observed no intracellular staining with unmodified C3, but staining was visible when cells were treated with C3-05 ( Fig. 5B-C). These results further confirm the permeability of C3-05. Rho in fibroblasts, we treated CHO cells and NIH 3T3 cells with varying concentrations of C3-05, and stained the cells with phalloidin to visualize actin stress fibers (Fig. 6).
When CHO and NIH 3T3 cells were plated in serum containing medium in the absence of C3-05, well-formed actin stress fibers were present (Fig. 6). In both cell lines the addition of C3-05 at concentrations as low as 0.0025 µg/ml dramatically reduced actin stress fiber formation. At concentrations of 0.025 µg/ml, or 0.25 µg/ml disassembly of actin stress fibers were almost complete. At 25 µg/ml, the highest concentration tested, CHO cells treated with C3-05 showed an altered morphology (Fig. 6). In NIH 3T3 cells Rho activation is correlated with the formation of actin stress fibers (Fig.4). Pull down assays of homogenates prepared from NIH 3T3 cells grown in the presence of serum showed high Rho activation levels, in agreement with the spread morphology of these cells (Fig 8C). Treatment with varying concentrations of C3-05 at 0.0025 µg/ml, 0.025 µg/ml and 0.25 µg/ml for 24 hours decreased Rho activation ( Fig.   8C-D). These results confirm that C3-05 inactivated Rho in fibroblasts.
To further support the ability of C3-like chimeric proteins to promote neurite outgrowth on inhibitory substrates, we examined the response of primary cultures plated on inhibitory substrates to C3-05 treatment. Purified retinal ganglion cells (RGCs) were plated on myelin, or CSPG substrates and treated with varying concentrations of C3-05 for 24 hours. During the RGC dissection great care was taken in order to try to limit the amount of mechanical manipulation of the cells, however, the isolation protocol requires that some triturating take place in order to dissociate and separate the cells. When RGCs are plated on inhibitory substrates, they maintained a similar round appearance to PC-12 cells plated on myelin. Treatment of RGCs with C3-05 promoted neurite outgrowth and increased neurite length on both myelin and CSPG substrates (Fig. 9A-F). In contrast to the wide range of concentrations shown to be effective in experiments with PC-12 cells, a narrower range of C3-05, 0.025 µg/ml to 50 µg/ml, promoted neurite outgrowth and increased neurite length on myelin (Fig. 9A-D). In the case of RGCs plated on CSPG effective concentration ranges of 0.0025 µg/ml to 50 µg/ml were observed (Fig. 9A,B, E,F).

DISCUSSION
Here we report the construction of five new C3-like chimeric proteins, all of which posses the ability to translocate across the plasma membrane to ADP-ribosylate and inactivate Rho. By a bioassay in which PC-12 were cells plated on growth inhibitory myelin substrates, we have shown that, to varying extents, all five C3-like chimeric proteins promoted neurite outgrowth. Based on our experiments with unmodified C3, which must be scrape loaded, we suggest that the ability of these new C3-like proteins to promote neurite outgrowth at such low concentrations is due to their increased cellular permeability. The differences observed in promoting neurite outgrowth between the five C3-like chimeric proteins might result from the different methods used by the various transmembrane carrier peptides to enter cells.
We tested two different transport sequences derived from the Tat protein of the human immunodeficiency virus (HIV). This protein has been reported to enter cells, carry protein cargo into cells, and even cross blood brain barrier (13). The mechanism whereby Tat transports cargo across the plasma membrane is still not completely understood. Tat internalization is not decreased at 4 °C, or in the presence of endocytosis inhibitors (14). There is uncertainty, however, whether its uptake is receptor mediated because Tat binds to specific cell membrane proteins (28). We found that C3-03, the longer Tat peptide sequence, was more efficient at promoting neurite outgrowth then C3-02, the shorter Tat peptide. However, the longer sequence may have some cellular toxicity (14), a finding consistent with the decreased ability to promote neurite growth at high concentrations (Fig. 3).
The third helix of the Antennapedia homeodomain can cross cell membranes by both energy and receptor independent mechanisms (16). Antp is a basic peptide that interacts with the charged phospholipids on the outer side of the cell membrane, causing destabilization of the lipid bilayer and the formation of inverted micelles. The formation of this hydrophobic structure allows the Antp and protein cargo to travel freely across the membrane, releasing the transported protein inside the cell once the hydrophobic pocket opens (16,17,29). One drawback to this family of transport peptides is that they lose their translocating ability when they bind to double stranded DNA (30). We found that C3-04 containing the Antp sequence was an effective carrier, but only within a narrow concentration range (Fig. 3).
Proline-rich peptides can also act as receptor independent delivery peptides.
Fusogenic peptides contain both hydrophobic and hydrophilic amino acids, which form amphiphilic α-helical structures. A critical component of these proteins are proline residues (19,23). Studies where site-directed mutation changing single proline residues of the PH-30α fusogenic protein, active in sperm-egg fusion, shows that prolines are critical for the fusogenic activity (23,31). The membrane translocating sequence (MTS) of Kaposi fibroblast growth factor, a known transport peptide, contains 3 proline residues spaced 5 to 7 amino acids apart (21). The spacing of prolines in this MTS peptide are similar to that in C3-05, which also possesses 3 prolines spaced 6 to 8 amino acids apart.
Furthermore, these proline residues may explain why C3-05 was the most effective C3like chimeric protein tested. When proline residues were added to Antp translocating sequences, Antp and its cargo were only present in the cytoplasm and not in the nucleus (32). The possibility that proline residues may restrict the membrane translocating peptides and their cargo to the cytoplasm would increase the ability of C3-05 to inactive Rho, a cytoplasmic protein.
Highly basic, arginine-rich peptides are another class of transport peptides. A simple string of seven, or more arginines covalently linked to a fluorescein moiety at the N-terminal was able to cross the cell membrane when analyzed by flow cytometry (21).
These transport peptides were more effective than both Tat and Antennapedia when compared directly (20). C3-06 contains a basic transport sequence, which corresponding to the reverse Tat sequence. As previously reported (20), we also found it to be a more effective transport sequence than either Tat, or Antennapedia. C3-06 contains three arginine residues at the amino end of the peptide end, compared to the Tat translocation peptide used in C3-02, which contains one arginine, and two lysine residues at the amino terminal. The increased ability of C3-06 to penetrate PC-12 cells and promote neurite outgrowth suggests that N-terminal residues are important for cellular uptake.
Furthermore, arginine residues are more effective then lysine residues in inducing cellular uptake (20).
Neurite outgrowth and neurite length profiles of PC-12 cells treated with C3-02 and to a lesser degree C3-03 and C3-04 had a normal distribution, (Fig.3) showing increased neurite outgrowth when cells were treated with low to moderate concentrations, but not with high concentrations. High concentrations of C3-02, C3-03 and C3-04 may have toxic effects on PC-12 cells. Previously, two other permeable C3 fusion proteins have been produced, one with Diphtheria-toxin B subunit, called DC3B (9) and another that is a C2 toxin-C3 fusion protein, called C2IN-C3 (10). Our three most effective C3like chimeric proteins, C3-03, C3-05, and C3-06 all worked at a much lower dose than DC3B (0.6 µg/ml) (9) and C2IN-C2 (0.2-0.3 µg/ml) (10), being effective at 0.0025 µg/ml. The lower effective dose of our C3-like proteins might be because they enter cells by receptor-independent mechanisms, and therefore should not be trapped within endocytotic compartments. In addition, C2IN-C3 is not independently cell permeable because the C2II, the binding component of C2 toxin, must be present to induce the uptake of the C2IN-C3 fusion protein by endocytosis. Furthermore, C2IN is a relatively large peptide consisting of 225 amino acids, which when attached to C3-transferase nearly doubles its molecular weight. All of the transport sequences used in this paper are under 50 amino acids, which may enhance uptake, as the total size of the protein is not dramatically increased.
When neuronal cells are plated on myelin they become round and do not grow neurites (7). Previously, we have suggested that myelin-derived growth inhibitory proteins directly activate Rho (33). Here we demonstrate the first evidence that inhibitory substrates activate Rho. GTP-bound Rho assays showed that myelin alone activated Rho when compared to cells plated on poly-l-lysine substrates (Fig. 7). Cells plated on myelin showed a 4 to 5-fold increase in cellular active Rho compared to cells plated on poly-llysine (Fig. 7). Treatment with C3-like chimeric proteins not only reversed the myelin induced Rho activation, but sustained this decrease for 36 to 48 hours. For all 3 concentrations tested, peak Rho inactivation appeared 24 hours after treatment, and then began to decrease. Possibly, the decrease occurred because the C3-like chimeric proteins were no longer active, or had all been taken up. We did not test whether the addition of more C3-05 to the culture media could sustain Rho inactivation for longer periods.
In previous in vivo experiments using unmodified C3, a robust period of regeneration of retinal ganglion cells in the optic nerve was observed after treatment (7).
Future studies will address the ability of these new C3-like proteins to help axon regeneration and repair after CNS injury. These C3-like chimeric proteins, therefore, may improve the extent of regeneration in the central nervous system following spinal cord injury.
The pathological progression of cancer involves abnormal cell growth, resulting in the formation of tumors, and increased cell motility, causing invasive properties and metastasis. Recent studies provide evidence that Rho A, B and C, all substrates for C3, play a role in both tumor development and metastatic progression by regulating the growth and motility of cells (34)(35)(36)(37)(38)(39)(40). For example, over expression of Rho is implicated in tumor formation of neck squamous cell cancer, aggressive ductal adenocarcinoma of the pancreas and inflammatory breast cancer (34,37,38). In culture, fibroblasts transfected with active Rho develop alterations in morphology and grow at higher densities than untransfected cells (35,39). The regulation of cell proliferation by active Rho in fibroblasts is inhibited by C3 treatment. This inhibition of cell proliferation is evident one day after treatment with C3, correlating with the ADP-ribosylation of Rho (41). Other studies suggest Rho plays a role in tumor metastasis. In contrast to neuronal cells, where Rho activation inhibits cell motility (Fig. 3C)