The Specific FKBP38 Inhibitor N-(N′,N′-Dimethylcarboxamidomethyl)cycloheximide Has Potent Neuroprotective and Neurotrophic Properties in Brain Ischemia*♦

FK506 and FK506-derived inhibitors of the FK506-binding protein (FKBP)-type peptidylprolyl cis/trans-isomerases (PPIase) display potent neuroprotective and neuroregenerative properties in various neurodegeneration models, showing the importance of neuroimmunophilins as targets for the treatment of acute and chronic neurodegenerative diseases. However, the PPIase activity targeted by active site-directed ligands remainsed unknown so far. Here we show that neurotrophic FKBP ligands, such as GPI1046 and N-[methyl(ethoxycarbonyl)]cycloheximide, inhibit the calmodulin/Ca2+ (CaM/Ca2+)-regulated FKBP38 with up to 80-fold higher affinity than FKBP12. In contrast, the non-neurotrophic rapamycin inhibits FKBP38·CaM/Ca2+ 500-fold less affine than other neuroimmunophillins. In the context of the high expression of FKBP38 in neuroblastoma cells, these data suggest that FKBP38·CaM/Ca2+ inhibition can mediate neurotrophic properties of FKBP ligands. The FKBP38-specific cycloheximide derivative, N-(N′,N′-dimethylcarboxamidomethyl)cycloheximide (DM-CHX) was synthesized and used in a rat model of transient focal cerebral ischemia. Accordingly, DM-CHX caused neuronal protection as well as neural stem cell proliferation and neuronal differentiation at a dosage of 27.2 μg/kg. These effects were still dominant, if DM-CHX was applied 2-6 h post-insult. In parallel, sustained motor behavior deficits of diseased animals were improved by drug administration, revealing a potential therapeutic relevance. Thus, our results demonstrate that FKBP38 inhibition by DM-CHX regulates neuronal cell death and proliferation, providing a promising strategy for the treatment of acute and/or chronic neurodegenerative diseases.

inhibit the PPIase activity of members of the FK506-binding protein (FKBP) family and therefore account for numerous physiological effects. Recently, we have shown that FKBP38 is special among the PPIases, because it displays enzymatic activity exclusively in complex with calmodulin/Ca 2ϩ (CaM/Ca 2ϩ ), thus defining the protein as a CaM/Ca 2ϩ -dependent PPIase (3). Only the FKBP38⅐CaM/Ca 2ϩ has affinity for FK506.
In general, the interpretation of effects caused by FK506 in cells is difficult, because FK506 inhibits not only the enzymatic activity of FKBPs, but also the protein phosphatase activity of calcineurin (CaN, PP2B). CaN inhibition is mediated by complex formation with FK506⅐FKBP complexes and is thought to be the initial process leading to immunosuppression (20 -22). CaN inhibition by immunophilin-immunosuppressant complexes is used to prevent allograft rejection in transplantation medicine, to treat autoimmune diseases and to circumvent graft-versus-host diseases. Additionally, inhibition of the protein phosphatase was the proposed basis of FK506-mediated neuroprotection, because the FKBP ligand rapamycin, which has no effects on CaN activity, did not exhibit neuroprotective properties (12,17,23).
In contrast, monofunctional inhibitors of FKBPs, such as GPI1046, GPI1048, GPI1485 (Guilford Pharmaceuticals and Amgen), and V10,367 (Vertex Pharmaceuticals) have been developed, that have no influence on CaN activity, while neuroprotective and neuroregenerative effects of FK506 remain conserved. In the central nervous system, GPI1046 promotes protection and sprouting of serotonin-containing nerve fibers in the somatosensory cortex following parachloramphetamine treatment, induces regenerative sprouting from spared nigrostriatal dopaminergic neurons following MPTP toxicity in mice or 6-hydroxydopamine toxicity in rats, and alleviates the rotational abnormality in 6-hydroxydopamine-treated rats (8,19). Other monofunctional FK506 derivatives increase branching from developing * 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. ࡗ This article was selected as a Paper of the Week. 1 To whom correspondence should be addressed. Tel. dopamine neurons in culture, enhance neurite outgrowth of fetal dopamine transplants, increase nerve regeneration, and accelerate functional recovery following peripheral nerve injury (24,25). Although the dramatic neuroprotective and neurotrophic effects of FKBP inhibitors imply an involvement of FKBP activity in neuronal cell death, the nature of the FKBP representing the primary target of these drugs remained enigmatic. Until now, 16 FKBPs have been identified in the human genome, two of which, FKBP12 and FKBP52, have been proposed to mediate neurotrophic actions of neuroimmunophilin ligands (12,26). However, there is accumulating evidence that FKBP12 inhibition does not affect neuronal protection and regeneration. Accordingly, it was shown that FK506 retained its neurite outgrowthpromoting properties in hippocampal cultures from FKBP12 knock-out mice (26).
In this report, using (i) comparative PPIase inhibition studies, (ii) expression analysis of FKBPs in neuroblastoma cells, and (iii) application of a novel, highly specific FKBP38 inhibitor in the rat focal cerebral ischemia model, we show that FKBP38⅐CaM/Ca 2ϩ inhibition constitutes the molecular basis of FKBP ligand-mediated neuroprotective and neurotrophic function.

Cloning, Expression, and Purification of FKBPs
To obtain recombinant FKBP13 the nucleotide sequence encoding the amino acids 27-142 (lacking endoplasmic reticulum transport and retention signals) was amplified by PCR with the following primers: 5Ј-AATTTCATGAAAAGGAAGATGCAGATCGGGGTC-3Ј and 5Ј-GCTAAAGCTTACAGCTCAGTTCGTCGCTCTATT-3Ј. The PCR product was subcloned into a pSTBlue-1 vector (Novagen), digested with BspHI/HindIII (New England Biolabs), ligated with a pET28a vector, and transformed into BL21(DE3) Rosetta cells (Novagen). To express FKBP38 in Escherichia coli, the corresponding sequence of the amino acids 1-336 (lacking the membrane anchor) was amplified by PCR using the following primers: 5Ј-AGTAAGTCATGAGACAAC-CCCCGGCGG-3Ј and 5Ј-ACGTAAGCTTAAAACAGCCA CTTC-CATGG-3Ј. The PCR product was subcloned into a pSTBlue-1 vector, digested with BspHI/HindIII, and cloned into a pET28a vector and transformed into BL21(DE3) Rosetta cells. The resulting constructs were verified by restriction analysis and DNA sequencing. The expression clones of human FKBP51 and FKBP52 were kindly provided by J. Buchner (Technical University of Munich). Recombinant human FKBP12 and FKBP12.6 were produced using the plasmid pQE60 (Qiagen) in E. coli strain K12 M15/pREP4, as described previously (27).
Protein expression of the FKBPs was induced by addition of isopropyl ␤-D-thiogalactopyranoside to a final concentration of 1 mM and incubation for 4 h. Subsequently, cells were harvested by centrifugation at 4°C for 15 min at 5000 ϫ g. The bacterial pellet from a 6 liters of culture was resuspended in 200 ml of lysis buffer (10 mM Hepes, pH 7.5, 150 mM NaCl) and French pressed. Next the supernatant was centrifuged at 4°C for 45 min at 35,000 ϫ g. Supernatants containing FKBP51 or FKBP52, respectively, were applied to a nickel-nitrilotriacetic acid column (His Trap HP, 1 ml; Amersham Biosciences) equilibrated with 20 mM Tris/ HCl, pH 8.0, using the N-terminal His 6 -fusion of the two proteins. The separation was performed according to the manufacturer's instructions. Fractions were analyzed by 12.5% (w/v) SDS-PAGE and staining with Coomassie blue. FKBP-containing fractions were dialyzed against 10 mM Hepes buffer, pH 7.8, 1.5 mM MgCl 2 , 150 mM KCl and loaded on a HiLoad 16/60 Superdex 200 pg (Amersham Biosciences) according to the manufacturer's instructions. The protein fractions were analyzed by 12.5% (w/v) SDS-PAGE and staining with Coomassie Blue.
Cell lysates containing FKBP12, FKBP12.6, or FKBP13, respectively, were applied to a Fractogel EMD DEAE-650 (Merck) column and a Reactive Blue 2-Cl 6B (Merck) column equilibrated with 10 mM Tris buffer, pH 8.0. Protein was eluted from the Reactive Blue 2-Cl 6B column by 1 M NaCl. Fractions were analyzed by 15% (w/v) SDS-PAGE and staining with Coomassie Blue. FKBP-containing fractions were dialyzed against 10 mM Hepes buffer, pH 7.5, and applied to a Fractogel EMD SO 3 Ϫ -650 column. Protein was eluted by 1 M NaCl and analyzed by 15% (w/v) SDS-PAGE and staining with Coomassie Blue.
To purify FKBP38, cell lysate was dialyzed in 10 mM Hepes, pH 7.5, 2 mM CaCl 2 and applied to a CaM-Sepharose (Amersham Biosciences) column. Protein was eluted by 5 mM EGTA and analyzed by 12.5% (w/v) SDS-PAGE and staining with Coomassie Blue. FKBP38-containing fractions were dialyzed against 10 mM Hepes buffer, pH 7.8, 1.5 mM MgCl 2 , 150 mM KCl and loaded on a HiLoad 16/60 Superdex, 200 pg (Amersham Biosciences), according to the manufacturer's instructions. The protein fractions were analyzed by 12.5% (w/v) SDS-PAGE and staining with Coomassie Blue.
The purified FKBPs were subsequently analyzed by matrix-assisted laser desorption ionization time-of-flight mass spectrometry and N-terminal protein sequencing, confirming the identity of the proteins.
Peptide substrates used were obtained from Bachem (Heidelberg, Germany). FK506 was purchased from Calbiochem.

Measurements of PPIase Activity
PPIase activity was measured using protease-coupled assays, as described previously (28). FKBP38 1-336 ⅐CaM/Ca 2ϩ PPIase activity was assayed in a reaction mixture containing 1 M FKBP38 1-336 , 5 M recombinant human CaM, and 2 mM CaCl 2 . Inhibition constants for the PPIase activity of the hFKBP38 1-336 ⅐CaM/Ca 2ϩ complex by low molecular weight inhibitors were determined by a competition assay using recombinant FKBP12 (3). PPIase activity of 12 nM FKBP12 was inhibited either by 20 nM FK506, 23 nM rapamycin, 1140 nM GPI1046, or 2400 nM N-[methyl(ethoxycarbonyl)]cycloheximide and subsequently recovered by addition of FKBP38 1-336 ⅐CaM/Ca 2ϩ . Due to the competition of both FKBPs for inhibitor binding, inhibition constants were determined using the Dynafit software (29).
The entire amount of the crude product N-[methyl-(tert-butyloxycarbonyl)]cycloheximide was dissolved in 5 ml of a 2.2 M solution of the zinc chloride diethyl ether complex in methylene chloride (Fluka) and stirred for 2 h at room temperature. After evaporation of the solvent and dissolving of the resulting residue in 5 ml of tetrahydrofuran, the product was precipitated by addition of diethyl ether. The crude product was further purified by preparative HPLC to yield 479 mg of compound 2 (79% relating to the applied amount of cycloheximide). 100 mg (295 mol) N-(carboxymethyl)cycloheximide, 306 mg (588 mol) of PyBOP (520 g/mol) and 72 mg (882 mol) of dimethylamine hydrochloride were dissolved in 20 ml of methylene chloride and cooled to 0°C. After dropwise addition of 130 l N-morpholine in 5 ml of methylene chloride, the resulting reaction mixture was stirred for 1 h at 0°C and subsequently for 12 h at room temperature. Following evaporation of the solvent, the crude product was purified by preparative HPLC to yield 67 mg of N-(NЈ,NЈ-dimethylcarboxamidomethyl)cycloheximide (62%).
The identity of the synthesized compounds was verified by ESI-MS and NMR.
NMR experiments were performed in CdCl 3 at 25°C using a DRX 500 spectrometer (Bruker, Rheinstetten, Germany). 1 H NMR data of were acquired with a 5-mm inverse triple-resonance probe with XYZgradient capability at 500.13 MHz resonance frequency; 13 C NMR data were obtained with a 5-mm broadband probe at 125.76 MHz resonance frequency. Standard one-dimensional and two-dimensional pulse sequences were employed. All spectra were processed with the XWIN-NMR 3.5 software (Bruker) and referenced to chloroform. 1

Quantitative PCR
mRNA from SH-SY5Y, PC3, and human Jurkat cells was prepared using the RNeasy mini kit (Qiagen) according to the manufacturer's protocol. Ribonuclease inhibitor (Calbiochem) was added to the mRNA preparations. The prepared mRNA was reverse-transcribed into cDNA using oligo(dT) primers (Novagen) and the Omniscript RT kit (Qiagen).

Protein Synthesis Assay
The eukaryotic translation assay was performed for 60 min at 30°C using the FlexiRabbit reticulocyte lysate system (Promega) according to the manufacturer's instructions. To prevent degradation of sample luciferase RNA, 40 units of RNasin ribonuclease inhibitor was added to the reaction. The concentration of synthesized firefly luciferase was determined by luminescence measurement in a scintillation counter (Wallac1450, PerkinElmer Life Sciences) at 25°C.

Investigation of Drug Effects on Focal Cerebral Ischemia
Animals-Studies were performed on male Sprague-Dawley rats (250 -280 g) obtained from Harlan Winkelmann (Borchen, Germany). The animals were maintained under constant environmental conditions with ambient temperature of 21 Ϯ 2°C and relative humidity of 40%. They were housed with a 12-h light-dark cycle and given food and water ad libitum.
Induction of Focal Cerebral Ischemia-The procedure for the induction of focal cerebral ischemia by occlusion of the middle cerebral artery via intracerebral microinjection of endothelin 1 was modified from that published previously by Sharkey and Butcher (30). Anesthesia was induced with halothane in a mixture of nitrous oxide/oxygen (70:30) and maintained with 2-3% halothane during the following procedures: rats were placed in a Kopf stereotaxic frame and further anesthetized via nose cone. For the induction of focal cerebral ischemia, a burr hole was drilled (1 mm diameter) into the skull (coordinates: anterior 0.90 mm from bregma, lateral 5.2 mm to satura sagittalis) and a 29-gauge cannula was lowered 7.5 mm below the dura according to the rat brain atlas of Paxinos and Watson (31). To induce occlusion of the middle cerebral artery, rats received an injection of 60 pmol of endothelin 1 (ED-1, Sigma) in 3 l of 0.1 M phosphate-buffered saline over a time period of 5 min. After a further 5 min, the cannula was slowly withdrawn. For the intracerebroventricular application of DM-CHX, a second burr hole was drilled into the skull (coordinates: posterior 0.80 mm from bregma, lateral 1.5 mm to satura sagittalis) and a 29-gauge cannula was lowered 4.5 mm below the dura. Intracerebroventricular applications of DM-CHX were performed either during (i) ischemia, 6 and 24 h after reperfusion; (ii) 2, 6, and 24 h; or (iii) 6 and 24 h after reperfusion. Rats were maintained at 37-38°C throughout the operation procedure. For controls, following a middle incision in sham operated animals, a burr hole was drilled into the skull, but no endothelin 1 was injected.
Determination of Infarct Volume-After a survival time of 7 days after eMCAO, animals were anesthetized by an intraperitonial injection of pentobarbital and perfusion-fixed transcardially with saline followed by 4% paraformaldehyde in 0.1 M phosphate buffer. Brains were then removed carefully, post-fixed in the same fixative for 2 h, and placed in a rodent brain matrix (rat, Activational Systems Inc., Scientific Instrumentation). 1-mm coronal brain slices were cut with a razor blade at 14 predetermined anterior-posterior levels. After cryoprotection in 30% sucrose, slices were rapidly frozen in isopentane and stored at Ϫ80°C. Four to five cryostat sections (30 m) from each brain slice were cut in acryo-microtome and stained with toluidine blue.
The extent of cortical and striatal damage following ischemic injury was documented with microphotographic images from the Nissl stained slices showing the anterior-posterior level according to the brain atlas of Paxinos and Watson (31). The extent of the infarct area at each level was calculated by integrating the area of damage at each stereotactic level and the distances between the various levels. Using a light microscope (Nikon, Eclipse TE 3000) equipped with a 4ϫ objective, the image analysis was performed with Lucia software, Version 4.2.1. The data were statistically analyzed by non-paired Student's t tests. Data are represented as mean Ϯ S.E. Statistical significance was accepted at the level of 0.001 or 0.01 of probability. MAY 26, 2006 • VOLUME 281 • NUMBER 21

JOURNAL OF BIOLOGICAL CHEMISTRY 14963
Quantification of Cell Proliferation and Neuronal Differentiation-For labeling mitotic cells, the thymidine analogue 5-bromo-2Ј-deoxyuridine (BrdUrd, Sigma) was used. Immediately after eMCAO surgery, animals received an intraperitoneal BrdUrd injection of 50 mg per kg body weight. This treatment was repeated during the following 5 days. Preparation of cryosections was performed as described above 7 days after eMCAO. For BrdUrd and NeuN labeling, sections were washed three times with 0.1 M phosphate-buffered saline (PBS) and incubated in 2 N HCL for 60 min at 37°C. Unspecific binding was blocked by 1.5% goat normal serum (Alexis, Gruenberg, Germany). Slices were then incubated with rat anti-BrdUrd (1:50 Ladder Rung Walking Test-To assess loss and recovery of function after focal cerebral ischemia, a ladder rung walking test, according to Metz et al. (32), was performed. A first series of training and test sessions were performed 2 days and 1 day before eMCAO surgery, respectively. Another series of training and test sessions was applied 6 and 7 days after eMCAO. In the training sessions, all animals crossed the ladder five times using a regular rung pattern of 2-cm distance from rung to rung. The total ladder length was 1 m. The following day, the same animals were tested three times on an irregular rung pattern where the distance between rungs varied between 1 and 4 cm. The rung pattern was changed between each test run to avoid memory and learning effects. Each session was monitored. Errors of hind and fore limb placements were counted and scored according to the following scheme (which includes score and type of misplacement): 0, correct placement; 1, limb position on same rung was corrected; 2, limb withdrawn from the rung and replaced to the same or adjacent rung; 3, fall after limb slipped off the rung; 4, fall after limb missed the rung. Final results were calculated as sum of error scores divided through the total number of single steps per animal and run. Five to eight animals per treatment group were used.
Data Analysis-Data were statistically analyzed by a non-paired Student's t test. Data are represented as mean Ϯ S.E. ***, p Ͻ 0.001 or **, p Ͻ 0.01 (see Fig. 4). (ethoxycarbonyl)]cycloheximide (ME-CHX), and the non-neurotrophic rapamycin, we measured the concentration dependence of residual PPIase activity of FKBP12, FKBP12.6, FKBP13, FKBP25, FKBP38⅐CaM/Ca 2ϩ , FKBP51, and FKBP52 using a standard oligopeptide-based PPIase assay and a PPIase competition assay (2,3,33). Inhibition constants were calculated from the plot of residual PPIase activity against inhibitor concentration on the basis of a 1:1 stoichiometry. With the exception of FKBP38⅐CaM/Ca 2ϩ , the standard oligopeptide-based PPIase assay was utilized. The PPIase competition assay is based on the recovery of FKBP12 activity in the standard PPIase assay in the presence of FKBP38⅐CaM/Ca 2ϩ using succinyl-ALPF-pnitroanilide as a substrate (3). Briefly, FKBP12 activity is inhibited by the low molecular weight compound. Activity is regained by the addition of different concentrations of FKBP38⅐CaM/Ca 2ϩ , because its active site competes with FKBP12 for the ligand (Fig. 1, A and B). To compare results from both assays, we determined inhibition constants of FKBP12 and FKBP13 with both the standard and the competition assay. The inhibition constants varied by less than 4% in both assays.

FKBP38⅐CaM/Ca 2ϩ Is Inhibited Potently by GPI1046 and N-[Methyl(ethoxycarbonyl)]cycloheximide-With the FKBP inhibitors GPI1046, N-[methyl
Interestingly, the CaM/Ca 2ϩ -activated FKBP38 exhibits high affinity to the neuroprotective FKBP ligands GPI1046 and ME-CHX with K i values in the low nanomolar concentration range, whereas the nonneurotrophic rapamycin is bound with at least 10-fold lower affinity (Table 1). In contrast, other FKBPs, such as the constitutively active FKBP12, are usually inhibited by rapamycin with an inhibition constant of about 1 nM. GPI1046 and ME-CHX bind to constitutively active FKBPs with at least 30-fold lower affinity. Therefore, FKBP38 is unique among the FKBPs by displaying low affinity to rapamycin, while neuroprotective FKBP ligands show preferential binding to FKBP38⅐CaM/ Ca 2ϩ , indicating a role of FKBP38 in neuroprotection, sprouting of nerve fibers, and functional nerve recovery.
FKBP38 Is Highly Expressed in Neuroblastoma Cells-To link (i) known neurotrophic properties of FKBP inhibitors, (ii) the cellular abundance of FKBPs, and (iii) the respective inhibition constants of FKBP inhibitors, we examined the most relevant FKBPs at both the mRNA and protein levels in three different cell lines: neuroblastoma, prostate carcinoma (PC3), and human Jurkat cells. Real-time PCR was performed using total cellular mRNA preparations from these cells. In neuroblastoma cells, we identified large amounts of FKBP38 mRNA and high FKBP38 protein concentrations ( Fig. 2A, B). The protein level was less pronounced in Jurkat cells. FKBP12, FKBP13, and FKBP52 were abundant in SH-SY5Y cells as well, but their mRNA levels and protein concentrations were lower than those of FKBP38. Furthermore no other FKBP showed an expression pattern comparable with that of FKBP38 with pronounced expression in neuroblastoma cells. For instance, the mRNA of FKBP12 and FKBP51 were found at higher levels in Jurkat cells compared with neuroblastoma cells. This fact was confirmed for FKBP12 at the protein level. The levels of FKBP13 and FKBP25 mRNA indicate that the two proteins are the most abundant of the tested FKBPs in PC3 cells, whereas FKBP38 mRNA levels are higher in neuroblastoma and Jurkat cells.
These data suggest that FKBP38 is specifically expressed at high concentration in neuroblastoma cells, whereas other immunophilins present in neuronal cells, such as FKBP12 and FKBP51, can be found at higher concentrations in other cell lines. In context of the inhibition constants of rapamycin-, GPI1046-, and ME-CHX-mediated inhibition for the different FKBPs, the FKBP expression pattern determined in three different cell lines implies that GPI1046 and ME-CHX target the FKBP38⅐CaM/Ca 2ϩ complex in neuroblastoma cells, whereas rapamycin likely forms complexes with FKBP12, FKBP13, and FKBP52. Therefore specific inhibitors of FKBP38⅐CaM/Ca 2ϩ may have potential to serve as neurotrophic drugs.
N-(NЈ,NЈ-Dimethylcarboxamidomethyl)cycloheximide Targets FKBP38⅐CaM/Ca 2ϩ in Vivo-Although ME-CHX was best in discriminating between the most abundant FKBPs and FKBP38⅐CaM/Ca 2ϩ among the inhibitors listed in Table 1, the low metabolic stability of this compound makes it less suitable for physiological experiments. ME-CHX could not be detected in the lysate of SH-SY5Y cells after 30-min incubation, because the ethyl ester group is particularly prone to esterase-catalyzed decomposition (Fig. 3A). We found that the -C(ϭO)OCH 2 CH 3 to -C(ϭO)N(CH 3 ) 2 substitution resulting in DM-CHX (Fig. 3B) imparted resistance toward metabolic conversion in a cell lysate. DM-CHX was stable in SH-SY5Y lysate on a 24 h time scale (Fig.  3A). Cellular uptake of DM-CHX in SH-SY5Y cells was validated by measuring an intracellular concentration of 1 M after 2-h incubation in the medium containing 40 M DM-CHX. The K i value for DM-CHXmediated inhibition of FKBP38⅐CaM/Ca 2ϩ revealed 6-fold lower affin-ity of the novel CHX derivative compared with ME-CHX (Table 1). However, the substitution improved the inhibitory specificity of DM-CHX in comparison to ME-CHX and other FKBP ligands, showing discrimination between FKBP38⅐CaM/Ca 2ϩ and constitutively active FKBPs ranging from 3 ϫ 10 3 (FKBP52) to 1.5 ϫ 10 2 (FKBP25).
Given the well established role of biotinylated compounds as affinity labels, the synthesis of N-{methylcarboxamido-[N,NЈ-␣,-bis(ethylene)decaethyleneglycol]--(NЈ-biotinamidyl)}cycloheximide (Biot-CHX) was performed. We found that Biot-CHX inhibits FKBP38⅐CaM/ Ca 2ϩ with a K i value of 726 Ϯ 34 nM. The affinity of the compound for FKBP38⅐CaM/Ca 2ϩ was not influenced when Biot-CHX was bound to streptavidin. To verify whether CHX derivatives preferentially bind to   MAY 26, 2006 • VOLUME 281 • NUMBER 21 endogenous FKBP38 in neuroblastoma cells, Biot-CHX bound to a streptavidin-Sepharose affinity matrix was incubated with SH-SY5Y cell lysate (Fig. 3C). Unmodified streptavidin-Sepharose beads served as control (Fig. 3C, lanes 2 and 3). FKBP38 alone, and not FKBP12, interacted with the CHX affinity matrix, suggesting that soluble CHX derivatives may preferentially capture the FKBP38⅐CaM/Ca 2ϩ complex in vivo. While general inhibition of the eukaryotic protein synthesis has been well established for the parent compound CHX (34), a similar effect of DM-CHX has not yet been tested. Synthesis of firefly luciferase in rabbit reticulocyte lysate in the presence of various CHX derivatives has been used to evaluate their effect on the eukaryotic translation machinery (Fig. 3D). As expected, CHX is highly inhibitory in this assay. DM-CHX has only a moderate influence on protein synthesis, exhibiting an IC 50 value Ͼ Ͼ750 M. Based on these results the influence of DM-CHX on mammalian protein synthesis is negligible at the concentrations required for inhibition of endogenous FKBP38. Furthermore, DM-CHX is not only cell-permeable but also chemically stable in cell lysates. Thus DM-CHX is a highly discriminating active site-directed inhibitor of the FKBP38⅐CaM/Ca 2ϩ complex that does not influence protein translation at therapeutic concentrations.

DM-CHX Induces Neuroprotection and Cell Proliferation in a Model of Transient Cerebral
Ischemia-To address the question whether well known neuroprotective and neurotrophic properties of monofunctional, nonimmunosuppressive FKBP inhibitors are mediated by specific inhibition of the FKBP38⅐CaM/Ca 2ϩ complex, we studied the effects of DM-CHX in a rat brain model involving endothelin-induced focal cerebral ischemia (eMCAO). In this model, the intracerebral injection of the vasoconstrictor endothelin-1 close to the middle cerebral artery causes a transient vessel occlusion and thus a focal brain ischemia lasting for 1-2 h. This approach reflects the effects of ischemic attack in many stroke patients. Here, DM-CHX was administered at different points during and/or after eMCAO.
In the first series of experiments, DM-CHX was applied in parallel to the endothelin injection, as well as 6 and 24 h post-insult. Control animals were exposed to eMCAO but received PBS without the drug. In general, histological analysis was performed after a survival time of 7 days after eMCAO. Under these conditions, DM-CHX reduced the eMCAO-induced infarct volume by 43.8% compared with controls (Fig. 4A).
To assess whether the neuroprotective effect of DM-CHX remains effective with increasing time intervals between eMCAO and initial drug application, the delay between both events was increased in a stepwise manner. Therefore, further animal groups received their first DM-CHX treatment 2 h or 6 h after eMCAO (Fig. 4A). A lower but still significant infarct volume reduction of 34.2% was found when the DM-CHX was administered 2, 6, and 24 h after eMCAO. In contrast, we were not able to detect a significant infarct volume reduction if DM-CHX was administered only 6 and 24 h after eMCAO (Fig. 4A). These results reveal a potential therapeutic window of DM-CHX between 2 and 6 h. The extent of neuroprotection caused by the DM-CHX was comparable with that observed with the selective N-methyl-D-aspartate antagonist MK801, a well known neuroprotective compound lacking, however, a significant therapeutic time window (data not shown).
Monofunctional FKBP inhibitors were shown to promote functional innervation and induce regeneration of lesioned sciatic nerve axons and myelin levels (8,35,36). Therefore, we analyzed the neural stem cell proliferation and differentiation 7 days after eMCAO (Fig. 4, B and C). BrdUrd and NeuN were used to label proliferating and neuronal cells, respectively. We found that the number of proliferating (and therefore BrdUrd-positive) cells, as well as the number of newly differentiated neurons, were significantly increased by DM-CHX in the ipsilateral brain hemispheres (Fig. 4C). The most pronounced stimulation of cell proliferation and neuronal differentiation was observed with DM-CHX applications during eMCAO, 6 and 24 h after eMCAO. However, this effect was still significant if the time window between eMCAO and first drug injection was extended up to 6 h.
To verify whether FKBP38⅐CaM/Ca 2ϩ inhibition can preserve neuronal function after focal cerebral ischemia, animals exposed to eMCAO and DM-CHX treatment were analyzed in a ladder rung walking test. In this test, animals have to pass a horizontal ladder of 1 m in length containing rungs with variable gaps. Using this setup, skilled walking performance, forelimb and hindlimb placing, stepping, and interlimb coordination were scored before and 7 days after eMCAO (37). DM-CHX improved the walking performance of injured animals significantly. The necessary time window between eMCAO and onset of DM-CHX administration was between 2 and 6 h (Fig. 4D), indicating functional relevance of FKBP38⅐CaM/Ca 2ϩ inhibition in neuroprotection post-insult.

DISCUSSION
Our results demonstrate that inhibition of Ca 2ϩ /CaM-dependent folding helper enzyme FKBP38 by the novel inhibitor DM-CHX dramatically decreased neuronal damage in a rat eMCAO model. In addition, FKBP38⅐CaM/Ca 2ϩ inhibition has marked neurotrophic effects on neural stem cell proliferation and differentiation in rat brain. The specificity of targeting of the CaM/Ca 2ϩ -dependent PPIase FKBP38 in the presence of other FKBPs by DM-CHX and other CHX derivatives was inferred from binding experiments using Biot-CHX.
Among the human FKBPs, FKBP38 deserves particular attention, representing the first example of a PPIase that is not constitutively active, but requires activation by a second messenger (3). Currently, FKBP38 is the only human FKBP that has been found to localize to the membranes of endoplasmic reticulum and mitochondria, where it interacts with the apoptosis regulator Bcl-2. This interaction depends on the enzymatic activity of FKBP38 and affects the anti-apoptotic function of Bcl-2 in neuronal apoptosis (3). Active site-directed inhibitors of FKBP38 interfere with the interaction of FKBP38⅐CaM/Ca 2ϩ and Bcl-2 and thus with the pro-apoptotic activity of FKBP38 in neuronal cells. Furthermore, FKBP38 is highly expressed in neuroblastoma cells, whereas other cell lines display significantly lower FKBP38 mRNA and protein levels.
An activity-based competition method was adopted that allowed the measurement of K i values for specific FKBP38 inhibitors on a subset of human FKBPs. In this assay, a high k cat /K m value of the respective enzyme/substrate combination is not required for the determination of tight binding constants. Moreover, the assay renders possible a comparative analysis of K i values for a broad range of FKBPs under similar  MAY 26, 2006 • VOLUME 281 • NUMBER 21 experimental conditions. In contrast, other assay conditions have resulted in widely scattered inhibition data, as reflected by a 2700-fold difference in IC 50 values reported for the FKBP12/GPI1046 interaction (38,39). Using our method, we could identify high affinity FKBP38⅐CaM/Ca 2ϩ inhibition by the neuroprotective compounds GPI1046 and ME-CHX. However, both FKBP ligands inhibit constitutively active FKBPs with at least 7-fold higher inhibition constants. Additionally, ME-CHX showed significant discrimination among the different human FKBPs, binding to FKBP38⅐CaM/Ca 2ϩ with up to 120-fold higher affinity than to other FKBPs. This discrimination implies that intracellular ME-CHX will efficiently bind to the FKBP38⅐CaM/Ca 2ϩ complex even in the presence of high concentrations of other FKBPs. The discriminating power of ME-CHX, and in part of GPI1046 as well, distinguishes both neuroprotective FKBP ligands clearly from FK506, which inhibits human FKBPs with comparable inhibition constants (5, 40 -43).

FKBP38 Inhibitor Mediates Neuroprotection
Rapamycin, in contrast, exhibits low affinity for the FKBP38⅐CaM/ Ca 2ϩ complex, even though the low molecular weight compound shows high affinity for constitutively active FKBPs, revealing inhibition constants below 1 nM. Hence, rapamycin will not target FKBP38 in cells expressing considerable concentrations of constitutively active FKBPs, as exemplified by SH-SY5Y cells. Consequently, neurotrophic properties should not be attributed to rapamycin in the stroke model. Interestingly, studies examining the reported neuroprotective properties of rapamycin in a rat MCAO stroke model induced by either endothelin-1 or microvascular clip failed to detect significant effects of rapamycin administration on infarct volume even at concentrations of 2 mg per kg body weight (12,44). Furthermore, in gerbils, rapamycin had no effect on the reduction of lesions induced by global ischemia. These results point strongly to a potential of FKBP38 inhibition by low molecular weight compounds such as ME-CHX, in protection of neuronal tissue against stroke. Importantly, the neuroprotective and neuroregenerative effects of FK506, and its open-chain derivatives in various models of neuronal disorders (5, 8, 19), could therefore also be interpreted in terms of FKBP38 inactivation. A, effect of DM-CHX on infarct volume 7 days after eMCAO. DM-CHX was applied (icv) either 1) in parallel to eMCAO as well as 6 and 24 h after eMCAO/reperfusion; 2) 2, 6, and 24 h after eMCAO; or 3) only 6 and 24 h after eMCAO. Cortical infarct volumes of the ipsilateral brain hemisphere were measured 7 days after eMCAO. B, influence of DM-CHX on neurogenesis 7 days after eMCAO. Proliferating cells and neurons were stained with BrdUrd (green fluorescence) and NeuN (red fluorescence, right panel only), respectively. Thus, newly differentiated neurons were double-labeled with BrdUrd and NeuN (yellow spots, right panel). Animals were exposed to eMCAO without DM-CHX (upper panel) or with DM-CHX treatment during eMCAO, 6 and 24 h after reperfusion (lower panel). For each image, a 20ϫ objective was used. C, time window of DM-CHX effect on cell proliferation and differentiation. Proliferating, BrdUrd-positive cells (a columns) and BrdUrd/NeuN-double stained neurons (b columns) were counted in the ipsilateral brain hemisphere 7 days after eMCAO. As in A, DM-CHX was administered 1) in parallel to eMCAO as well as 6 and 24 h after eMCAO; 2) 2, 6, and 24 h after eMCAO; or 3) only 6 and 24 h after eMCAO. D, motor behavior deficits of DM-CHX-treated and untreated animals in a ladder rung walking test. Bars represent walking errors per step of treated and untreated animal groups. The schedule of DM-CHX administration was the same as in A and C. Data were statistically analyzed by a non-paired Student's t test and are presented as mean Ϯ S.E. ***, p Ͻ 0.001 or **, p Ͻ 0.01.
Notably, cycloheximide (CHX), the parent compound of ME-CHX, was reported to have significant neurotrophic effects at concentrations that lowered the protein synthesis rate by only 20 to 40% (45). For instance, CHX prevents apoptosis of sympathetic neurons in culture, inhibits trophic factor deprivation-induced PC12 cell death (46 -48), and protects cultured cortical neurons against oxidative-induced death and adult cortical neurons against ischemic injury in vivo (49 -52). Additionally, CHX prevents excitotoxicity-induced death of retinal ganglion cells, glutamate toxicity in PC12 cells, amyloid ␤-peptide toxicity, and 3-nitropropionic acid toxicity in cortical neurons (53)(54)(55)(56). The observed CHX-mediated neuroprotection is not likely to be caused by the influence of CHX on protein expression. Interestingly, the neurotrophic properties of CHX seem to alter a signaling pathway that involves Bcl-2 activity (45). In our inhibition assay, CHX proved to be an exceptionally strong inhibitor of FKBP38⅐CaM/Ca 2ϩ with a K i value of 2.7 nM (data not shown), whereas FKBP12 is inhibited with a K i value of 3.4 M, therefore indicating exclusive binding to FKBP38⅐CaM/Ca 2ϩ in neuronal cells.
However, CHX cannot be used to prevent neurodegeneration, because the compound effectively inhibits the eukaryotic protein translation machinery (57,58). The CHX derivative ME-CHX, even though lacking the drawback of protein translation inhibition, suffers from low intracellular stability. In contrast, the recently developed DM-CHX is suitable for FKBP38 inhibition in cells, as it exhibits (i) increased stability in cell lysates and (ii) exceptionally low potential to inhibit protein translation among tested CHX derivatives. Although the affinity of DM-CHX toward the FKBP38⅐CaM/Ca 2ϩ complex is 6-fold lower compared with the affinity of the parent compound, the discrimination potential of DM-CHX is the highest among known FKBP ligands. For instance, the inhibition constant for the interaction of DM-CHX and FKBP38⅐CaM/ Ca 2ϩ is more than 3000-fold lower than the constant of the DM-CHX⅐FKBP52 complex. Based on the prevention of FK506-mediated neurotrophic action by monoclonal antibodies against FKBP52 (26), the involvement of FKBP52 in FKBP ligand-mediated neurotrophic effects was suggested. However, a comparison of the inhibition constants of DM-CHX interaction with both neuroimmunophilins excludes the possibility of FKBP52-mediated DM-CHX effects in the presence of FKBP38 and strongly suggests a FKBP38-specific targeting by DM-CHX in cells.
In this respect the striking effects of DM-CHX application on eMCAO-induced neuronal cell death is remarkable. Prevention of neuronal cell death, resulting from a combination of reduced infarct volume and increased neuronal proliferation by DM-CHX-mediated FKBP38 inhibition, might represent an ideal therapeutic strategy. The therapeutic potential of FKBP38 inhibitors for acute neurodegenerative diseases or injuries is further stressed by (i) a considerable time window of 2-6 h between eMCAO and initial drug application and (ii) the correlation between molecular and cellular events such as neuroprotection and proliferation and a functional outcome parameter such as motor behavior deficits. We believe that the improved motor behavior results primarily from the DM-CHX-induced infarct volume reduction because FKBP38 inhibition by either low molecular weight inhibitors or siRNA also leads to a dramatic reduction of neuroblastoma cell death after apoptosis induction via different pathways (3). Further studies are required to investigate in detail both short and long term effects as well as the consequences of FKBP38 inhibition on cell proliferation and differentiation. Nevertheless, our results demonstrate that FKBP38 inhibition by DM-CHX has significant therapeutic potential for the treatment of ischemia-induced neurodegeneration.