Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

The polyketides FK506 and FK520 (a C-21 ethyl derivative of FK506) are potent immunosuppressants produced by Streptomyces. They act by binding first to the immunophilin FK506 binding protein 12 (FKPB12) through the "FKBP-binding domain" (Fig. 1). The binary complex then binds calcineurin through the "effector" domain and inhibits the phosphatase activity of calcineurin, and thus the activation of nuclear factor of activated T cells. Hence, FK506 and FK520 inhibit the activation of T cells and suppress the immune system.

View larger version (25K): [in this window] [in a new window]   Fig. 1.   Structure of the FKBP12-FK506-calcineurin ternary complex. The residues that have close contact with the methoxy groups at carbons 13 and 15 are illustrated. The main interactions are with Trp 352 of calcineurin (a bifurcated hydrogen bond, shown with a dotted line, 3.0 Å to the O at C-13 and 3.1 Å to the O at C-15 of FK506) and a possible hydrophobic contact between the 15-methoxy and Leu 343 of calcineurin (4.1 Å). Note also that the methoxy group at carbon 13 is close to a hydrogen bonding network comprising His 87 (FKBP12), a water molecule, and the carbonyl of the calcineurin backbone at Trp 352.

FK506 and FK520 also accelerate the rate of nerve regeneration (Gold, 2000; Hamilton and Thomas, 2000), being very potent in promoting neurite outgrowth in PC12 cells, SH-SY5Y cells, and primary neuronal cultures (Lyons et al., 1994; Gold et al., 1999). The neurotrophic property of these compounds has been established in a variety of animal models (for review, see Gold, 2000), including, sciatic nerve injury (Gold et al., 1994; Archibald et al., 1999; Doolabh and Mackinnon, 1999; Lee et al., 2000), spinal cord injury (Madsen et al., 1998; Wang and Gold, 1999), and toxic-chemical models of Parkinson's disease (Hamilton et al., 1997; Steiner et al., 1997b; Costantini et al., 1998; Emborg et al., 2001). Nevertheless, for treatment of neurodegenerative diseases, it would clearly be necessary to reduce or eliminate the immunosuppressive effects of these drugs. Nonimmunosuppressant neuroimmunophilin ligands have been shown to retain the potent neuroregenerative properties of FK506 itself (Gold et al., 1997; Steiner et al., 1997a).

One approach to obtaining nonimmunosuppressive neuroimmunophilin ligands is to design molecules that bind to FKBP12 but not to calcineurin. Although FKBP12 may not be the important neuroimmunophilin in nerve regeneration (Gold et al., 1999), where studied, binding to FKBP12 tracks neurite outgrowth (Hamilton and Thomas, 2000). Small molecule binders of FKBP12 have been discovered either by ab initio synthesis (Steiner et al., 1997b), or by chemical modification of the FK506 effector domain (Steiner et al., 1997a). The ab initio approach has provided many compounds with biological activity, but these entities must still surmount significant barriers to become therapeutic agents. Chemical modification exploits the well known and pharmacologically favorable properties of FK506, but the opportunities for performing specific chemical modifications are limited due to its complex structure. An alternative approach to modify the effector domain is to engineer the polyketide biosynthetic genes.

The three-dimensional structure of the FKBP12-FK506-calcineurin ternary complex reveals interactions between the 13- and 15-methoxy groups of FK506 and calcineurin (Fig. 1). First, the N1 nitrogen of the conserved Trp 352 of the calcineurin A-subunit forms a bifurcated H-bond with both the 13- and 15-methoxy groups of FK506. Mutation of this residue in yeast calcineurin prevents the FKBP12-FK506 complex from binding while retaining calcineurin phosphatase activity (Cardenas et al., 1995; Hemenway and Heitman, 1999). Second, the conserved Leu 343 of the A-subunit is in proximity (4.2 Å) to the methyl moiety of the 15-methoxy group of FK506. Finally, the 13-methoxy group is close to a polar H-bond network formed between His 87 of FKBP12, a water molecule, and the carbonyl of Trp 352 of the calcineurin backbone.

We surmised that modification of the 13- and 15-methoxy groups should disrupt calcineurin binding without affecting FKBP binding. These methoxy groups are not amenable to specific chemical modification, so the aim was to change them by manipulating the FK520 polyketide biosynthetic genes.

The FK520 polyketide synthase (PKS) from Streptomyces hygroscopicus contains 10 modules that catalyze sequential additions of two-carbon units into a polyketide chain. Subsequently, pipecolic acid is added, and the chain is cyclized to give FK520 (Hopwood, 1997; Wu et al., 2000). Each module has the expected complement of domains found in PKSs, including an acyltransferase that selects the particular acyl-CoA precursor for each round of chain extension. In the biosynthesis of FK506 and FK520, the methoxy groups at carbons 13 and 15 are introduced via unusual methoxymalonyl precursors by the acyltransferase domains at modules 7 and 8 of the PKS (Reeves et al., 2002). Thus, by exchanging the methoxymalonate-specific acyltransferase domains with heterologous acyltransferase domains that recognize malonate, methylmalonate, or ethylmalonate, the engineered PKS should direct synthesis of FK520 analogs with a hydrogen, methyl, or ethyl in place of the methoxy groups at positions 13 and 15.

We describe the manipulation of the FK520 gene cluster in S. hygroscopicus to produce analogs of FK520 altered at carbons 13 and 15 that retain high affinity for FKBP12 but bind poorly to calcineurin. Two nonimmunosuppressive FK520 analogs were tested for neurite outgrowth activity in vitro and one of these was further tested in vivo in the rat sciatic nerve crush model.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial Strains, Vectors, and Culture Conditions. The FK520-producing strain S. hygroscopicus (ATCC 14891) and its engineered clones were cultured as described previously (Wu et al., 2000). The phage vector KC515 and its cloning host S. lividans TK24 were obtained from the John Innes Centre (Norwich, UK) and cultured as described previously (Kieser et al., 2000).

Replacement of Acyltransferase Domains in FK520 PKS Cluster. The "acyltransferase-swap-cassettes" contained specified heterologous acyltransferase domains flanked by 1.5-kb fragments of DNA identical to the sequences on either side of the acyltransferase 7 domain of the FK520 gene cluster. DNA fragments were generated by the polymerase chain reaction (PCR), using Pfu DNA polymerase (Stratagene, La Jolla, CA) as recommended by the manufacturer, with 10% dimethyl sulfoxide in the reaction mixture. The temperature profile was as follows: one cycle at 96°C for 10 min, 30 cycles of PCR (denaturation for 1 min at 96°C, anneal for 1 min at 65°C, and extension for 2 min at 72°C), and one cycle at 72°C for 10 min. The oligonucleotides were as follows: sequence upstream of acyltransferase 7 (BglII-AvrII), forward 5'-GAAGATCTCACGATGCCGCGCTGCTCGAC-3', and reverse 5'-GCTCTAGACCTAGGGCGGTCGGTTTCGGGCCAG-3'; sequence downstream of acyltransferase 7 (XhoI-NsiI), forward 5'-GCTCTAGACTCGAGCCCACCTCCCGGGCCGATG-3', and reverse 5'-TGCATGCATGGCGTGTCGTCGGCTATGAGC-3'; rapamycin acyltransferase 3 (AvrII-XhoI), forward 5'-GGCCTAGGCGGGCGGGCGTGTCGTCCTTC-3', and reverse 5'-CCCTCGAGCCAGTACCGCTGGTGTTGGAA-3'; rapamycin acyltransferase 2 (AvrII-XhoI), forward 5'-ATCCTAGGCGGGCRGGYGTGTCGTCCTTCGG-3', and reverse 5'-ATCTCGAGCCAGTASCGCTGGTGYTGGAAGG-3'; and fkb (FK520) acyltransferase 4 (AvrII-XhoI), forward 5'-CGGAATTCCCTAGGCGCGCCGGTGTGTCCTCCTTC-3', and reverse 5'-CGGAATTCCTCGAGCCAGTAGCGCTCATGGTGGAAGG-3'. PCR products were cloned, verified by sequence analysis, and used for assembly of the acyltransferase-swap cassettes. The cassettes were inserted into KC515, a delivery vector based on the broad host range phage C31 (Kieser et al., 2000). Recombinant phage vectors were introduced into S. hygroscopicus as described for double crossing-over to generate the desired domain substitutions (Wu et al., 2000; Reeves et al., 2002).

Detection of FK520 Analogs. Strains were screened for FK520 analog production by online extraction LC/MS using a system comprised of a CTC PAL HTS Autosampler (Leap Technologies, Inc, Carrboro, NC) configured with a 10-port switching valve, a high-performance liquid chromatography (Beckman System Gold), and an API100 LC single quadrupole mass spectrometry-based detector (PE Sciex, Ontario, Canada) equipped with an atmospheric pressure chemical ionization source (source temperature, 350°C; ring voltage, 250 V; and orifice voltage, 20 V). Samples were prepared for analysis by adding an equal volume of methanol to whole broth followed by centrifugation to remove insolubles, and online extraction as described previously (Reeves et al., 2002). Analysis of compounds consisted of generating extracted ion chromatograms (XICs) of the ions expected for each analog and constructing chromatograms from the XICs (Table 1). Once the compounds were detected, cultures were scaled up to provide material for complete structure characterization by high-resolution mass spectrometry and NMR spectroscopy (J. R. Carney and G. Ashley, manuscript in preparation).

Purification of FK520 Analogs. Strains were grown in 20-liter cultures, as described previously (Regentin et al., 2002). Compounds were extracted from the fermentation broth by filtering the centrifuged culture broth through a column of XAD-16 absorbent. The absorbent was washed with water and then eluted with methanol. The methanolic eluate was concentrated to a volume of ca. 200 ml and then poured slowly into 1.5 liters of ether with stirring. The mixture was stored overnight at 4°C, filtered to remove the orange precipitate, and concentrated. The residue was dissolved in ether, dried over MgSO₄, filtered, and evaporated. The residue was chromatographed using an S100 chromatograph (ISCO, Lincoln, NE) with a 35-g column of silica gel. The elution program consisted of 80:20 hexanes/acetone for 10 min, a linear gradient to 70:30 hexanes/acetone over 5 min, and 70:30 hexanes/acetone for 30 min, collecting 13-ml fractions at a flow rate of 17 ml/min. Fractions were analyzed by thin layer chromatography (70:30 hexanes/acetone) using FK520 as a standard and staining with cerium molybdate. Fractions containing the product were pooled and evaporated. Final purification was by preparative high-performance liquid chromatography using an InertSil ODS-3 column (20 × 50 mm, 5 µm; MetaChem, Torrance, CA) at a flow rate of 8 ml/min and a linear gradient from 50:50 water/acetonitrile to 100% acetonitrile over 15 min, monitoring 290 nm. The product-containing fractions were pooled and evaporated to dryness. Compounds were characterized by ¹H and ¹³C NMR together with LC/MS.

FKBP, Calcineurin, and T-Cell Assays. FKBP binding assays were as described previously (Carreras et al., 2001). Calcineurin phosphatase inhibition was determined according to established procedures (Klee et al., 1983; Salowe and Hermes, 1998). A typical reaction mixture contained 40 mM Tris-HCl pH 7.5, 6 mM MgCl₂, 0.1 mM CaCl₂, 0.1% (w/v) bovine serum albumin, 15 nM calcineurin (Calbiochem, San Diego, CA), 30 nM calmodulin (Sigma-Aldrich, St. Louis, CA), 26 µM FK520 analog, and 0 to 25 µM FKBP12. After a 30-min preincubation, reactions were initiated with 1 µM of a ³²PO₄-phosphorylated peptide substrate of calcineurin phosphatase (synthetic RII phosphopeptide, ~5000 dpm/pmol) and initial rates of ³²PO₄ release determined (Kawai et al., 1993). Rate data were fit to an equation that corrects for depletion of free ligand via complex formation to obtain IC₅₀ values (Segel, 1975). Inhibition of T-cell response to concanavalin A was measured using peripheral blood mononuclear cells isolated from normal human donors according to standard procedures (Peterson et al., 1998). Drugs were tested for T-cell cytotoxicity using XTT as described previously (Scudiero et al., 1988).

SH-SY5Y Cell Cultures. The human neuroblastoma cell line SH-SY5Y was maintained and treated with analogs as described previously (Gold et al., 1999), with the following modifications. Cells were plated (15,000 cells/well) in duplicate wells and treated with 10 ng/ml nerve growth factor (NGF) to induce process outgrowth in the presence or absence of 1.0 nM FK520, or 0.1, 1.0, 3.0, or 10 nM of one of the FK520 analogs. Sixty to 180 cells (representing 15-20 fields/well) were randomly photographed (magnification, 45×) at 96 and 168 h, and neurite lengths were measured on photographic prints using an HI-Pad digitizing tablet (Houston Instrument, GTCO CalComp, Inc., Columbia, MD); only cells in the periphery were photographed because the cell density was too great in the center of the wells. Most cells showed one process, but in those rare instances where the process branched the length of the longest branch was measured. Only processes greater than twice the cell body diameter were included in the analysis. It should be noted that by 168 h, many cell processes contact other cells in the well, limiting their growth. Thus, the data reported at 168 h should be considered to represent minimal values (i.e., the differences between groups are likely to be greater than those reported herein). The entire experiment was repeated, and results reported herein are a composite of both. A third experiment showed similar results at 96 h, but was not included in the data analysis because the cell density was too great by 168 h to visualize individual processes.

Embryonic Hippocampal Cell Cultures. Neurons were obtained from rat pups on embryonic day 18.5 (Banker and Cowan, 1977) and treated as described previously (Gold et al., 1999). Cells were seeded onto coverslips (500 cells/coverslip) coated with poly(L-lysine) (Sigma-Aldrich). The coverslips (3/well) were inverted onto 24-well plates (Falcon Plastics, Oxnard, CA) that had been precoated with a monolayer of cortical astrocytes. Cells were treated with 10 nM NGF, or 1 nM FK520, or 1 nM FK520 analog. Eighty to 135 hippocampal neurons (identified by their characteristic polarity and dendrites) were randomly photographed (9-12 frames/coverslip at magnification, 45×) at 48 and 72 h. For each neuron, the axon (defined as the longest process) was measured on photographic prints using an HI-PAD digitizing tablet (Houston Instrument); only processes more than three times the cell body length were measured. The entire experiment was repeated, and results reported herein are a composite of both.

Nerve Crush Assay and Recovery. Nine 6-week-old (young adult) male Sprague-Dawley rats had their sciatic nerve crushed twice in one location at the level of the hip as described previously (Gold et al., 1997). Each set of three rats received daily subcutaneous administration of vehicle alone (30% DMSO), 1 mg/kg/day 13-Me-18-OH FK520, or 5 mg/kg/day 13-Me-18-OH FK520 for 18 days. Three additional age-matched (8-week-old) normal rats served as noninjured controls.

Statistical Analysis for Neurite Outgrowth and Nerve Regeneration. From the cell culture and axonal area data, cumulative histograms of the distribution were constructed, and mean values and standard errors were calculated. Histograms were compared using a Mann-Whitney U test ( = 0.05) (WINKS 4.62; TexaSoft, Cedar Hill, TX), which makes no assumptions about the shape of the distribution. Mean values were compared using a one-way analysis of variance followed by Newman-Keuls multiple comparison text for comparison of individual values (WINKS 4.62; TexaSoft). All values are mean ± S.E.M.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Genetically Engineered Analogs of FK520. We used genetic engineering of the FK520 PKS cluster in S. hygroscopicus (ATCC 14891) to effect changes at the 15 position of 13-H FK520 (Reeves et al., 2002). Three analogs were prepared by replacing the acyltransferase domain of module 7: 13,15-bis-desmethoxy FK520 (13-H-15-H FK520) using the malonyl-specific acyltransferase domain of module 2 from the rapamycin PKS cluster (Aparicio et al., 1996); 13-H-15-Me FK520 using the methylmalonyl-specific acyltransferase domain of module 3 from the rapamycin PKS cluster; and 13-H-15-Et FK520 using the ethylmalonyl-specific acyltransferase domain of module 4 from the FK520 cluster (Wu et al., 2000).

Immunosuppressive Properties of FK520 Analogs. The analogs were assayed for 1) binding to human FKBP12 and FKBP52, 2) inhibition of human calcineurin phosphatase by the FKBP12-ligand binary complex, 3) suppression of T-cell activation, and 4) cytotoxicity toward T cells (Table 2). 13-H FK520 retained FKBP12 and FKBP52 binding, calcineurin phosphatase inhibition, and suppression of T-cell activation essentially similar to FK520. In contrast, 13-Me FK520 retained FKBP12 and FKBP52 binding, but had about 20-fold diminished calcineurin phosphatase inhibition and greatly reduced suppression of T-cell activation. The 13-H-15-H/Me/Et FK520 analogs each retained FKBP12 and FKBP52 binding, but lost calcineurin phosphatase inhibition (IC₅₀ value of approximately 10 µM), and with the exception of the 15-Et analog, were nonimmunosuppressive by T-cell assays. The apparent IC₅₀ value of 630 nM for 13-H-15-Et FK520 in T-cell assays is likely due to a noncalcineurin-dependent suppressive effect (IC₅₀ value of 10 µM for calcineurin phosphatase inhibition), perhaps a manifestation of the observed toxicity of this analog (IC₅₀ value of 2500 nM). All other compounds were noncytotoxic up to the concentrations indicated (Table 2). These results demonstrate that replacement of the C-13 methoxy by methyl, or the C-15 methoxy by H, Me, or Et, does not greatly effect FKBP binding, but prevents calcineurin phosphatase inhibition and results in compounds that are nonsuppressive in T-cell activation assays. Likewise, 13-Me-18-OH FK520 retained FKBP binding, lost over 10-fold in calcineurin phosphatase inhibition and did not suppress T-cell activation.

13-Me FK520 Analogs Promote Neurite Outgrowth in Cell-Based Assays. 13-Me FK520 and 13-Me-18-OH FK520 were chosen for in vitro neurite outgrowth studies. Both analogs enhanced the NGF-dependent neurite outgrowth in cultures of the human neuroblastoma cell line SH-SY5Y at nanomolar concentrations (Table 3; Fig. 2). In a dose-response curve, 13-Me-18-OH FK520 showed a decrease in nerve outgrowth from SY5Y cells at 10 nM compared with 0.1 nM, reminiscent of the bell-shaped dose-response for neurite outgrowth observed for FK506 (Wang et al., 1997). 13-Me-18-OH FK520 (1 nM) also stimulated axon-like outgrowth in primary hippocampal cell cultures (average length, 569 ± 308 µm at 72 h; n = 135 cells measured; P < 0.05), an ~18% increase compared with control cells (485 ± 246 µm at 72 h; n = 88 cells measured; P < 0.05). For both SH-SY5Y cells and hippocampal neurons, the distributions for the whole population were significantly shifted to longer (more rapidly growing) processes (data not shown).

View larger version (75K): [in this window] [in a new window]   Fig. 2.   13-Me-18-OH FK520 stimulates neurite outgrowth in SH-SY5Y cells in a dose-dependent manner. Compounds were added to cells along with 10 ng/ml NGF and the mean neurite length was determined at 96 h. n is number of cells measured. , mean lengths that are different from the NGF treatment group at the 0.05 significance level; NT, no treatment.

Neuroregenerative Properties of 13-Me-18-OH FK520 in Vivo. 13-Me-18-OH FK520 was chosen for further study in the rat sciatic nerve crush model for peripheral nerve regeneration. The sciatic nerve was crushed at the hip to impair the ability of the animal to bear weight on the paw of the injured limb. At the higher dose of 5 mg/kg/day, the first signs of recovery (the ability to right the paw on the damaged limb) were seen at 8.2 days; at 1 mg/kg/day, at 11.6 days; and for the vehicle alone control group, at 14 days (Fig. 3). The time to a more complete recovery (the ability to walk on the hind feet and toes) was significantly shorter only at the higher dose of 5 mg/kg/day (12 versus 17 days for the vehicle alone control). At day 18, footprints were analyzed by determining the distance between the second and fourth digits; the 5-mg/kg/day treatment group showed a statistically significant improvement compared with the vehicle alone control group (8 ± 0.4 versus 6.8 ± 0.3 mm; P < 0.05). This compares favorably with the toe spread for normal uninjured animals (9-10 mm), indicating a near complete recovery.

View larger version (94K): [in this window] [in a new window]   Fig. 3.   13-Me-18-OH FK520 speeds functional recovery of rats after sciatic nerve crush. The time taken for the rats to right the paw (onset of recovery) was shorter for both 1- and 5-mg/kg/day treatment groups than for the vehicle alone control group. The time taken to walk on the toes and the front of the paw on the injured limb (full recovery) was shorter at the higher dose of 5 mg/kg/day than the vehicle control. Values are ± S.E.M. , mean times that are different from the vehicle alone treatment group at the 0.05 significance level.

View larger version (17K): [in this window] [in a new window]   Fig. 4.   13-Me-18-OH FK520 stimulates regeneration of larger axons. Cumulative histograms showing axonal areas in the soleus nerve of rats 18 days after sciatic nerve crush (a shift to the right indicates larger axons). a, axonal areas for all axons in the soleus nerve. Both the 1- and the 5-mg/kg/day treatment groups have larger axons than the vehicle-alone treated rats. b, axonal areas for the largest 30% of axons shown in a. Note that the greatest shift is present in the 5-mg/kg/day treatment group.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

We sought to discover nonimmunosuppressive neuroimmunophilin ligands by developing analogs of FK520 that bound to FKBP12 but could not inhibit calcineurin phosphatase. The crystal structure of the FKBP12-FK506-calcineurin complex shows that the oxygens of the methoxy groups at carbons 13 and 15 of FK506 are each within 3 Å of a conserved Trp 352 of the calcineurin A-subunit, and form a bifurcated hydrogen bond with the hydrogen of the indole nitrogen of the Trp residue. The methoxy groups do not interact with FKBP12 (Griffith et al., 1995; Hemenway and Heitman, 1999). By replacing the methoxymalonate-specific acyltransferase domains of modules 7 and 8 of the FK520 PKS with acyltransferase domains of differing specificity, we generated analogs of FK520 that possessed an H or Me in place of the 13-methoxy group; and a series of 13-H FK520 analogs with an H, Me, or Et in place of the 15-methoxy group. We also prepared 13-Me-18-OH FK520 because the 18-OH is known to disrupt binding of FK520 to calcineurin while retaining its ability to stimulate nerve regeneration in cell culture and animal models (Dumont et al., 1992; Steiner et al., 1997a).

As anticipated from the structure, all of the analogs bound to FKBP12 about as well as the parent FK520, although the Me and Et groups at position 15 caused some loss in binding relative to the OMe. The analogs also bound to FKBP52, a proposed molecular target for nerve regeneration (Gold et al., 1999), and their binding affinities to FKBP52 generally tracked that to FKBP12, albeit with markedly reduced affinity. The FKBP12-13-H FK520 binary complex inhibited the phosphatase activity of calcineurin phosphatase with a similar potency to the parental FKBP12-FK520 complex. Likewise, 13-H FK520 and FK520 showed equipotent suppression of T-cell activation. In contrast, FKBP12-13-Me FK520 lost about 20-fold potency in inhibiting calcineurin phosphatase, and 13-Me FK520 did not suppress T-cell activation at high concentrations. Retention of immunosuppressive properties by the 13-H analog was somewhat unexpected because we had predicted that loss of the H-bond of the 13-methoxy group to Trp 352 would result in loss of calcineurin phosphatase inhibition. That FKBP12-13-H FK520 binds to calcineurin argues that either the H-bond from the 13-methoxy group to Trp 352 of calcineurin is not important for binding, or that the hydrogen bond from the 15-methoxy is sufficient. The decrease in calcineurin binding by FKBP12-13-Me FK520 can be explained either by untoward effects of the hydrophobic methyl group at C-13 on the polar H-bond network connecting FKBP12 and calcineurin, or by loss of the H-bond (Fig. 1).

When the 15-methoxy group of 13-H FK520 was further modified to H, Me, or Et, there was a very large loss in the potency of the corresponding FKBP12-FK520 analog binary complex in calcineurin phosphatase inhibition (ca. 300-fold), correlating with losses in the ability of the analogs to suppress activation of T cells. The 15-Et substituent is isosteric with the methoxy group; it lacks the oxygen that H-bonds to Trp 352, but retains the methyl that can interact with the Leu 343 of calcineurin. Thus, one can speculate that the decrease of calcineurin phosphatase inhibition with the 15-substituted analogs is primarily due to the loss of the hydrogen bond from the Trp 352 of calcineurin. Based on the known effects of 18-OH FK520 as a neuroimmunophilin ligand, and the aforementioned properties of 13-Me FK520, we likewise prepared the 13-Me-18-OH FK520 analog. As expected, FKBP12 binding remained very tight, but there was a complete loss of calcineurin phosphatase inhibition and of T-cell suppression.

We chose 13-Me FK520 and 13-Me-18-OH FK520 to represent this class of compounds in studies on nerve regeneration because 13-Me FK520 was produced in higher titers by its corresponding engineered S. hygroscopicus strain than the other nonimmunosuppressive analogs; we anticipate that any of the other analogs reported herein would have served as well. The analogs were tested in the human neuroblastoma cell line SH-SY5Y and found to be active in stimulating neurite outgrowth at low nanomolar concentrations comparable with that of FK506 and FK520 (Gold et al., 1999). When 13-Me-18-OH FK520 was tested over a range of concentrations, a decrease in neurite outgrowth was observed at the highest concentration, similar to that reported for FK506 or for NGF (Chang et al., 1995; Gold et al., 1999). 13-Me-18-OH FK520 likewise caused a ~20% increase in nerve growth in primary rat hippocampal cell cultures. In the rat sciatic nerve crush model, 13-Me-18-OH FK520 caused a dose-dependent decrease in the time required for animals to regain the use of the affected limb, and a selective increase in axonal calibers and the numbers of myelinated axons. The increase in the extent of myelination indicates the presence of more mature nerve fibers in the distal nerve and suggests that the compound accelerates the rate of axonal regeneration, as found previously with other neuroimmunophilin ligands (Gold et al., 1995, 1997).

Genetic engineering has provided a tool to enable substitutions at various positions in the FK520 polyketide macrocycle that would have been impossible to access by chemical methods. A second advantage of the technology reported herein is that it allows for production of large quantities of complex molecules by fermentation for potential clinical application. These genetically engineered nonimmunosuppressive analogs are expected to retain many of the advantages of FK506, including bioavailability and blood-brain barrier penetration, and at the same time lose the toxic effects associated with calcineurin inhibition and immunosuppression (Peterson et al., 1998). The work reported herein sets the stage for a detailed analysis of these nonimmunosuppressive neuroimmunophilin ligands to determine their efficacy in treatment of various neurological disorders.