Inhibition of axotomy-induced neuronal apoptosis by extracellular delivery of a Bcl-XL fusion protein.

Bcl-2 and Bcl-XL prevent neuronal apoptosis during development, neurodegenerative disease, and trauma. To test a new anti-apoptosis strategy for neuroprotection, we engineered nontoxic components of anthrax toxin into a Bcl-XL delivery system. Delivery of Bcl-XL by this system prevented apoptosis of cultured rat cerebellar granule cells and macrophages, and the prevention depended on both the Bcl-XL and the anthrax toxin receptor binding/translocation moieties. Furthermore, neuronal death in vivo in a retinal ganglion cell model of axotomy-induced apoptosis was inhibited by administration of this fusion protein. Thus, Bcl-XL protein can be delivered into cells from the medium or interstitial space, offering a new way to block apoptosis upstream of many caspases and the mitochondria dysfunction phase of apoptosis.

Bcl-2 and Bcl-XL prevent neuronal apoptosis during development, neurodegenerative disease, and trauma. To test a new anti-apoptosis strategy for neuroprotection, we engineered nontoxic components of anthrax toxin into a Bcl-XL delivery system. Delivery of Bcl-XL by this system prevented apoptosis of cultured rat cerebellar granule cells and macrophages, and the prevention depended on both the Bcl-XL and the anthrax toxin receptor binding/translocation moieties. Furthermore, neuronal death in vivo in a retinal ganglion cell model of axotomy-induced apoptosis was inhibited by administration of this fusion protein. Thus, Bcl-XL protein can be delivered into cells from the medium or interstitial space, offering a new way to block apoptosis upstream of many caspases and the mitochondria dysfunction phase of apoptosis.
Bcl-XL and Bcl-2 inhibit apoptosis of neurons and many other types of cells (18). Bcl-XL, an anti-apoptotic member of the Bcl-2 family (19), is globally expressed in the developing and mature mouse brain at levels higher than those of Bcl-2 (20). Bcl-XL knockout mice die in utero around embryonic day 13 and exhibit extensive apoptotic death of neurons in the brain, spinal cord, and dorsal root ganglia (21). In contrast, Bcl-2 is widely expressed in the developing nervous system, although, in the adult, expression persists at high levels only in the peripheral nervous system (22). Whereas the brains from embryonic homozygous Bcl-2 knockout mice appear to be grossly normal (23), the facial motoneurons and spinal cord sensory and sympathetic neurons exhibit degeneration postnatally (24). Thus, Bcl-2 appears to be necessary for postnatal peripheral neuronal survival, whereas Bcl-XL plays a more important role in preventing death in the central nervous system.
Although Bcl-2 and Bcl-XL have potential to inhibit neurodegeneration, it is not now clinically practical to deliver genes to the brain. However, the transient delivery of anti-apoptotic proteins into neurons could help prevent neuronal death associated with traumatic injury and neurological diseases. Previously, Bcl-XL fused to the diphtheria toxin receptor binding domain, Bcl-XL-DTR, was found to prevent apoptosis in vitro induced by staurosporine, ␥-irradiation, and poliovirus (34). However, as diphtheria toxin receptors are present only on human and primate cells and the Bcl-XL-DTR protein does not inhibit apoptosis in rat and mouse cells (34), in vivo utility of the Bcl-XL fusion protein in rodent models could not be evaluated.
A nontoxic derivative of anthrax toxin has been shown to be a useful system to deliver peptides into the cytosolic compartment of mammalian cells (35). Anthrax toxin comprises three components: lethal factor (LF), 1 edema factor (EF), and protective antigen (PA) (35). PA binds to an unidentified cell surface receptor existing on the cells of many species including rodents (35,36) and, after being proteolytically activated, binds and transports LF or EF into cells (35). PA alone is not toxic (35). The first 254 residues of LF (LFn), which constitute the PAbinding domain, can mediate the delivery of heterologous peptides into the cytosol (35,(37)(38)(39)(40)(41)(42)(43). Initially, two splicing isoforms were discovered for human Bcl-X: Bcl-XL and Bcl-XS (19). Shortly thereafter, an alternatively spliced form of Bcl-XL lacking the C-terminal membrane anchor, Bcl-X⌬TM, was identified in mouse cells (44). Bcl-XL and Bcl-X⌬TM both inhibit apoptosis, whereas Bcl-XS promotes apoptosis. We fused human Bcl-XL and a truncated Bcl-XL lacking C-terminal membrane anchor (⌬Bcl-XL, similar to mouse Bcl-X⌬TM) to LFn and explored their effects on apoptosis in vitro and in vivo.

EXPERIMENTAL PROCEDURES
Construction of Prokaryotic Expression Plasmids-The gene for LF from codons 34 to 288 (LFn) (45) was amplified by PCR from the template of pET15b/LFn (38). The genes of full-length human Bcl-XL and C-terminal truncated human Bcl-XL from codons 1 to 209 (⌬Bcl-XL) (19) were amplified by PCR, and the LFn gene was fused to the 5Ј end of the Bcl-XL gene or the ⌬Bcl-XL gene by a second round of PCR. A stop codon was introduced immediately after the last codon of fulllength Bcl-XL or codon 209 of Bcl-XL. The fused DNA fragments, LFn-Bcl-XL and LFn-⌬Bcl-XL, were cut with NdeI and XhoI and separately inserted into the prokaryotic expression vector pET15b cut with * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
NdeI and XhoI (Fig. 1A). A histidine tag and a thrombin cleavage site were linked to the N terminus of LFn-Bcl-XL or LFn-⌬Bcl-XL. Similarly, the ⌬Bcl-XL gene alone was also cloned into pET15b at the NdeI and XhoI sites. All the constructs were verified by DNA sequencing.
Construction of Eukaryotic Expression Plasmids, Transfection, Western Blotting, and Biologic Activity Assay-The genes of LFn-Bcl-XL, LFn-⌬Bcl-XL (Fig. 1A), full-length Bcl-XL, and ⌬Bcl-XL were separately cloned into the eukaryotic expression vector pcDNA3 and verified by DNA sequencing. COS-7 cells were transfected separately with the above four plasmids, and 12 h after transfection the cell lysates were loaded on SDS-PAGE, visualized by immunoblotting with anti-Bcl-XL monoclonal antibodies (2H12, Trevegen, Gaithersburg, MD), and developed by using enhanced chemiluminescence (Amersham Pharmacia Biotech) (Fig. 1B). As reported, COS-7 cells were co-transfected with one of the four above plasmids and the reporter plasmid, pEGFP-3C, which contains the green fluorescence protein (GFP) gene (46). The cells were treated with 0.8 M staurosporine (STS) 12 h later. The dead and living cells expressing GFP were then counted at different times after STS treatment as reported previously (34,46).
Protein Expression, Purification, SDS-PAGE, and Western Blotting-The proteins LFn, LFn-Bcl-XL, LFn-⌬Bcl-XL, and ⌬Bcl-XL were individually expressed in Escherichia coli BL21(DE3) (Novagen, Inc.) and purified with the His⅐Tag binding purification kit (Novagen, Inc.). The transformed BL21(DE3) were cultured at 37°C in LB medium until the A 600 reached 0.5-0.8, then treated with 1 mM isopropyl-1-thio-␤-Dgalactopyranoside, and cultured for another 3 h. The cells expressing LFn-Bcl-XL were pelleted and lysed with a French press. The inclusion bodies were collected and dissolved in 6 M guanidine⅐HCl. His⅐Tag binding resin (Novagen) was used to purify LFn-Bcl-XL. LFn-Bcl-XL was refolded by dialysis against, or dilution into, 100 mM Tris⅐acetate (pH 8.0) plus 0.5 M arginine, concentrated with polyethylene glycol 15,000 -20,000, and dialyzed against PBS. The cells expressing LFn, LFn-⌬Bcl-XL, and ⌬Bcl-XL were pelleted and disrupted with a French press. The cytosol was loaded on the His⅐Tag binding column. The eluted proteins were dialyzed against PBS. PA was purified as reported (47). The proteins were run on SDS-PAGE gels and stained with Coomassie Blue or visualized by immunoblotting with anti-Bcl-XL monoclonal antibodies (2H12) and developed by using enhanced chemiluminescence (Fig. 2).
J744 Macrophage-like Cell Culture, Treatment, and Apoptosis Assay-J744 macrophage-like cells at 10 5 /ml were plated in 96-well plates in 100 l of RPMI 1640 with 10% fetal calf serum/well and cultured overnight. The cells were treated with 0.1 M STS along with different combinations of proteins or with PA or with PBS. The apoptotic and living cells were counted with Hoechst 33342 as reported elsewhere (34).
Cerebellar Granule Cell Culture, Treatment, and Viability Detection-Cerebellar granule cells were prepared from 8-day-old Sprague-Dawley rat pups (15-19 g, Taconic Farms, Germantown, NY) as described by Levi et al. (48). The cells at 4 ϫ 10 5 /ml were plated in 96-well plates in 100 l of basal Eagle's medium with 10% fetal calf serum, 2 mM glutamine, and 25 mM KCl per well. The cells were treated with 2.5 g/ml Ara-C 24 h later to eliminate non-neuronal cells and cultured for 6 more days. The cerebellar granule cells were treated with PBS, 0.1 M STS alone, 0.1 M STS along with LFn-⌬Bcl-XL (47 g/ml) plus PA (39 g/ml), or with 0.1 M STS along with different protein combinations. The apoptotic and living cells were counted with Hoechst 33342.
Western Blotting of Transfected Bad in J774 Cells Treated with the Fusion Protein-J774 cells were transfected or co-transfected with the gene for GFP-bad, the genes for GFP-bad and Bcl-XL, or the genes for GFP-bad and ⌬Bcl-XL. The GFP-bad gene was in pEGFP-3C, and Bcl-XL or ⌬Bcl-XL was in pcDNA3. 12 h later, the cells transfected with GFP-bad were treated with proteins LFn-⌬Bcl-XL (60 g/ml) plus PA (55 g/ml). Five hours later, cell lysates were made and loaded onto SDS-PAGE, immunoblotted with antibody against phospho-Bad (Ser-136) 2 or antibody against Bad (Cell Signaling Technology, Beverly, MA) and developed with enhanced chemiluminescence.
Optic Nerve Section and Intraocular Protein Injection-The P0 pups of Fisher 344 rats were used for the present study. P0 is defined as the day of birth. The intracranial lesion of unilateral optic nerve was performed as reported (3). Briefly, a P0 pup was anesthetized by hypothermia. Under a dissecting microscope, an incision over the right eye was cut and a piece of bone flipped up. The right optic nerve was sectioned after suctioning the overlying tissue. The section site of the optic nerve is about 3 mm away from the eyeball. A piece of gelfoam was put in the hole, the bone was placed back, and the incision repaired with superglue.
Immediately after the operation, LFn-⌬Bcl-XL (0.65 g) along with PA (0.35 g) or PA alone (0.35 g) or LFn-⌬Bcl-XL (0.65 g) alone or PBS in a volume of 350 nl was injected through ora serrata into the posterior chamber of the eye by using a microinjector with a pulled micropipette. The pup was warmed up with a light lamp until the recovery, and then returned to the mother.
Histology-24 h after sectioning of the optic nerve, the right eyes were taken out under deep anesthesia with sodium pentobarbital, fixed in 4% paraformaldehyde for ϳ30 h, embedded in paraffin, and cut at 6 m. The eyes taken from the normal pups in the same litters were processed in the same way and taken as controls. The sections were rehydrated, stained with 0.2% cresyl violet, dehydrated, and mounted with DPX mountant.
The pyknotic and living cells of the entire retinal ganglion cell layer of three sections per retina were counted by the use of a 40ϫ objective. The pyknotic cells were identified as reported (3). The values were presented as the percentage of pyknotic cells versus total cells (see Fig. 6B). hydrophobic tail (LFn-⌬Bcl-XL), was made to examine whether this tail, which targets Bcl-XL to mitochondria, would help or hinder cell entry via the anthrax toxin (Fig. 1A). To compare the potential bioactivity of these two fusion proteins, the genes were transfected into mammalian cells prior to apoptosis induction by STS, a potent protein kinase inhibitor. LFn-⌬Bcl-XL inhibited apoptosis induced by staurosporine similarly to fulllength LFn-Bcl-XL whereas native Bcl-XL or ⌬Bcl-XL transfected into cells was slightly more active (Fig. 1C). The expression of the four transfected proteins was confirmed by Western blotting (Fig. 1B).

Characterization of Proteins Purified from E. coli by SDS-PAGE and Western
Blotting-To prepare these Bcl-XL fusion proteins for extracellular delivery, the proteins were expressed in E. coli and purified by affinity chromatography to near homogeneity, as shown by SDS-PAGE analysis (Fig. 2A). The control proteins LFn and ⌬Bcl-XL were also expressed in E. coli and purified. The composition of LFn-Bcl-XL and LFn-⌬Bcl-XL was confirmed by Western blotting with anti-Bcl-XL antibody (Fig. 2B).
Biological Activity of Purified LFn-Bcl-XL and LFn-⌬Bcl-XL in Vitro-The biological activity of the LFn-Bcl-XL and LFn-⌬Bcl-XL proteins was initially evaluated in tissue culture. LFn-Bcl-XL or LFn-⌬Bcl-XL and the anthrax toxin receptor binding and entry domain (PA) were added to the medium of J774 cells immediately after apoptosis was induced by STS. Cells treated with STS died by apoptosis over the following 36 h as shown in Fig. 3A. However, when the cells were treated with LFn-Bcl-XL plus PA or LFn-⌬Bcl-XL plus PA, most of the cell death was inhibited. Controls were performed to evaluate the requirements for apoptosis inhibition. Fig. 3B shows that J774 cells treated with LFn alone, ⌬Bcl-XL alone, LFn-⌬Bcl-XL without PA, and PA without LFn-⌬Bcl-XL were not protected from apoptosis induced by STS, whereas LFn-⌬Bcl-XL with PA prevented more than half of the cell death (p Ͻ 0.001). This indicates that ⌬Bcl-XL uses the anthrax toxin entry pathway to access the cell cytosol. Interestingly, LFn protein fused to Bcl-XL without the C-terminal hydrophobic tail appeared to prevent apoptosis to an extent similar to LFn protein fused to full-length Bcl-XL protein, indicating that the C terminus of Bcl-XL is not essential for the anti-apoptosis activity. This is consistent with certain previous studies in which C terminustruncated Bcl-XL retained apoptosis inhibition activity (44,46,49) and with the results shown in Fig. 1C where transfection with ⌬Bcl-XL prevents apoptosis. Similarly, Bcl-2 lacking the C-terminal hydrophobic tail retains anti-apoptosis activity (50 -52), although truncating the C-terminal tail can impair the potency of apoptosis inhibition (52,53).
We determined the stability of LFn-⌬Bcl-XL apoptosis inhibition by adding it to the cell media at various times prior to STS treatment. The pretreatment of cells with LFn-⌬Bcl-XL up to 10 h before STS treatment is as effective as the treatment at the same time in preventing cells from apoptosis (Fig. 3C). Thus, the protein efficacy remains stable for at least 10 h. The protein was also partially effective at preventing cell death when added up to 1 h after apoptosis induction (Fig. 3C). The potency of LFn-⌬Bcl-XL was measured by examining its dose response in apoptosis protection. When less potent doses of STS are applied to cells that initiate only 15-20% apoptosis after 30 h, LFn-⌬Bcl-XL can block over 80% of the cell death (Fig. 3D). When the dose of PA is kept constant at 33 g/ml, the dose of LFn-⌬Bcl-XL that is half-maximal at apoptosis prevention is less than 2 g/ml or less than 40 nM (Fig. 3D). Thus, LFn-⌬Bcl-XL offers an extremely potent mechanism to prevent cell death.
As the C-terminal truncated LFn-⌬Bcl-XL is more soluble and easier to purify, we explored the potential of this protein in models of neuronal death in vitro and in vivo. We examined the effect of LFn-⌬Bcl-XL on apoptosis of primary rat cerebellar granule cells. The LFn-⌬Bcl-XL protein combined with PA inhibited STS-induced apoptosis of these neurons by about 30% after 45 h (p ϭ 0.0017) (Fig. 4). However, ⌬Bcl-XL alone, LFn-⌬Bcl-XL alone, PA alone, and PA plus LFn lacked significant bioactivity. Thus, in primary neuron cultures, as in the cultured macrophage cell line, ⌬Bcl-XL protein can be delivered via the anthrax toxin entry pathway to prevent apoptosis.
The Blockage of Bad Phosphorylation by LFn-⌬Bcl-XL-We have recently found that overexpression of Bcl-XL or ⌬Bcl-XL inhibits Bad phosphorylation at serine 136. 2 We determined one cellular activity of LFn-⌬Bcl-XL by examining the effect of LFn-⌬Bcl-XL on this phosphorylation of Bad. As shown in Fig.  5, LFn-⌬Bcl-XL protein, like transfection with Bcl-XL and ⌬Bcl-XL, effectively blocks Bad phosphorylation. This is consistent with a cytosolic delivery of Bcl-XL by LFn and demonstrates one similar cytosolic bioactivity between LFn-⌬Bcl-XL protein and endogenous overexpression of Bcl-XL or Bcl-XL lacking the C terminus. Biological Activity of Purified LFn-⌬Bcl-XL in Vivo-This new strategy to block cell death was explored in an animal model of neuronal apoptosis. Neonatal rat retinal ganglion cells (RGCs) die by apoptosis within 24 h after optic nerve section (3). Retinal ganglion cells were axotomized, and within 5 min a protein mixture containing 0.35 g of PA and 0.65 g of LFn-⌬Bcl-XL was injected into the ipsilateral eye. Control mice were not axotomized, axotomized and injected with PBS, or axotomized and injected with LFn-⌬Bcl-XL alone or PA alone. Mice were sacrificed 24 h later, and the eyes were examined histologically. As seen in Fig. 6A, a great number of pyknotic cells, i.e. apoptotic cells (3), were found in the retinal ganglion cell layer 24 h after axotomy. However, when eyes were injected with LFn-⌬Bcl-XL and PA, much of the cell death was inhibited (Fig. 6A). LFn-⌬Bcl-XL alone or PA alone caused no obvious prevention of cell death. To quantitate the extent of cell death, the living and pyknotic cells in the entire retinal ganglion cell layers of three sections from one eye in each of 4 -10 mice/group were counted. The results are shown in Fig. 6B. LFn-⌬Bcl-XL plus PA inhibited more than half of the cell death caused by neuronal axotomy in vivo (p Ͻ 0.001), whereas PA alone or LFn-⌬Bcl-XL alone had no significant effect. DISCUSSION Bcl-XL engineered to enter cells from the extracellular milieu inhibits neuron death in vitro and in vivo. Because neuron apoptosis occurs during many neurodegenerative diseases such as Alzheimer's disease (11)(12)(13), amyotrophic lateral sclerosis (17,54), and Huntington's disease (15,16), and exacerbates neuron loss from stroke (9,10,55) and traumatic injury to the optic nerve (3,7,8), spinal cord (4), and brain (5,6), the Bcl-XL fusion protein has potential to promote the recovery of injured neurons and delay the progression of these diseases.
Although delivery of the gene for Bcl-2 or Bcl-XL is not now clinically practical, methods of delivering protein to the central nervous system via direct infusion into the brain, cerebrospinal fluid, spinal cord, or peripheral nerves have been established in animals (63,64) and man (65,66). Thus, delivery of proteins that prevent apoptosis may inhibit certain neurodegenerative conditions. Here, we demonstrate that a single intravitreal administration of both LFn-⌬Bcl-XL and PA prevents 60% of the apoptosis induced by optic nerve section in the RGC layer of neonatal rats. Because only one concentration and single dose injection of both LFn-⌬Bcl-XL and PA have been tested in the present study, it is possible that a higher apoptosis-preventing effect could be achieved using increased protein dosage or increasing injection frequency.
It has been demonstrated that some neurotrophic factors promote the survival of neurons in vivo (67), but the clinical utility of neurotrophic factors thus far has not been dramatic. Nerve growth factor was confirmed to have therapeutic effects only on some sensory neuropathies (68,69). Brain-derived neurotrophic factor in ALS patients had no effect on survival but exhibited a statistically significant benefit for those ALS patients with early respiratory impairment and altered bowel function (70). Ciliary neurotrophic factor exhibited mostly negative effects in ALS patients (71).
Direct delivery of LFn-⌬Bcl-XL may have an advantage over neurotrophic factors to increase neuron survival. In the nervous system, various neurotrophic factors protect different subsets of neurons only during certain developmental periods. For example, nerve growth factor maintains the survival of sympathetic neurons only during neonatal maturation. Because Bcl-XL is widely expressed in the central nervous system during development (20) and its overexpression protects against a broad range of neurotoxic insults (32,33,56), LFn-⌬Bcl-XL may have broad potential to prevent apoptosis of different types of neurons resulting from different insults at different stages of development. Here we show LFn-⌬Bcl-XL protects cerebellar granule cells, a macrophage-related cell line, and retinal ganglion cells from apoptosis. Previously, we found Bcl-XL-DTR prevents apoptosis caused by poliovirus, radiation, and STS (34).
The finding that the combination of LFn-⌬Bcl-XL and PA dramatically inhibits apoptosis in the J774 macrophage cell line also indicates that the fusion protein can be successfully delivered to non-neuronal cell lines to block cell death. Thus, a large number of uses outside the nervous system may also be considered as potential applications of Bcl-XL protein delivery.