|Home | About | Journals | Submit | Contact Us | Français|
A new target strategy in the development of bacterial vaccines, the induction of antibodies to microbial outer membrane ferrisiderophore complexes, is explored. A vibriobactin (VIB) analogue, with a thiol tether, 1-(2,3-dihydroxybenzoyl)-5,9-bis[[(4S,5R)-2-(2,3-dihydroxyphenyl)-4,5-dihydro-5-methyl-4-oxazolyl]carbonyl]-14-(3-mercaptopropanoyl)-1,5,9,14-tetraazatetradecane, was synthesized and linked to ovalbumin (OVA) and bovine serum albumin (BSA). The antigenicity of the VIB microbial iron chelator conjugates and their iron complexes was evaluated. When mice were immunized with the resulting OVA-VIB conjugate, a selective and unequivocal antigenic response to the VIB hapten was observed; IgG monoclonal antibodies specific to the vibriobactin fragment of the BSA and OVA conjugates were isolated. The results are consistent with the idea that the isolated adducts of siderophores covalently linked to their bacterial outer membrane receptors represent a credible target for vaccine development.
Iron occurs in oxidation states from −2 to +6 depending on both pH and the nature of the ligating groups surrounding the metal.1 It is these dependencies that nature has exploited so effectively in enlisting the metal as a central component in a myriad of redox processes.2 In fact, life without iron is virtually nonexistent.3 However, while the metal composes some 5% of the earth’s crust, it is nevertheless difficult for living systems to access. In the biosphere, iron exists largely as Fe(III), in a variety of water insoluble forms, at pH 7. The concentration of free Fe(III) under these conditions is ≈1.4 × 10−9 M,4 somewhat lower than that required to support most life forms. In the presence of phosphate ions in culture media and potential animal hosts, the free Fe(III) concentration in solution drops even further, by a factor of 10.
Both prokaryotes and eukaryotes have overcome the problem of iron accessibility by developing iron-binding ligands and associated transport systems.5–12 Prokaryotes produce a group of iron chelators, siderophores, (generally low-molecular weight, iron-specific ligands) that they secrete into the environment.10 These ligands often present very large formation constants (e.g., 1048 M−1)13 and can effectively remove the metal from other donor arrays. The resulting metal complex, a ferrisiderophore, is then taken up by microorganisms,14–16 most often beginning with binding to an outer membrane receptor.17 This is followed by shuttling the iron complex through the periplasm and finally to the cytoplasm, where the iron is freed up. These ferrisiderophore transporters are energy-dependent, often exploiting the tonB system.18–20 In most instances, the desferrisiderophore is released to further gather iron.
There have now been over 500 different siderophores identified.21–25 While there are certainly exceptions, the two main classes of natural product iron chelators are hydroxamates,26–31 such as desferrioxamine (1) and catecholamides,13,32–36 including vulnibactin (2) and vibriobactin (3) (Figure 1). Some microorganisms can, in fact, utilize more than one type and/or class of siderophores.37
Iron acquisition becomes somewhat more problematic for microorganisms in an in vivo situation (e.g., in humans). Pathogens have additional iron acquisition hurdles to overcome beyond low metal solubility. Animals, for example, have an iron-withholding system: proteinaceous iron chelators that make iron acquisition difficult for microorganisms. There is little of the free metal available in animals. It is generally bound to heme10 (iron-containing enzymes)10 by transferrin12 (an iron shuttle protein) or stored in ferritin.11 In each instance, iron is not easily accessible to microorganisms.
The opportunistic microorganism Vibrio vulnificans nicely illustrates how pathogens can overcome host iron-withholding.20,38 The siderophore produced by Vibrio vulnificus,21 vulnibactin (2) (Figure 1), cannot remove iron from transferrin, the ever-present iron shuttle protein in plasma, in spite of the fact that 2 binds iron more tightly than transferrin. The chelator cannot access transferrin iron, as it is bound within the protein. To solve this problem, the microorganism secretes a protease, which cleaves transferrin, thus releasing iron. The metal is then sequestered by 2, and the ferrisiderophore is taken up via an intermembrane receptor, viuA.39–41
In fact, Vibrio vulnificus mutants without the vulnibactin transporter have reduced pathogenicity in mice.42 This uptake apparatus has been shown to have significant homology with the Vibrio cholerae receptor.20,38,39 However, while it seems clear from studies with genetically altered microorganisms that shutting down the siderophore iron-uptake system can slow growth and reduce pathogenicity, microorganisms can still access iron via other mechanisms.43–45 For example, Vibrio cholerae can utilize transferrin and heme as iron sources. The issue then becomes how useful a target the siderophore transport apparatus is in antimicrobial design strategies.
Miller has, in a series of classic studies, employed siderophores and the corresponding transporters as vectors for the delivery of antibiotics.46 Alternatively, Esteve-Gassent was able to demonstrate that a vaccine developed to treat eels infected with Vibrio vulnificus serovar E. contained antigens to the putative receptor for vulnibactin. Esteve-Gassent point out that the antibody could be blocking siderophore uptake, could trigger classical complement activation, or “mark bacteria for opsonophagocytosis.”47
There is now significant literature that supports the idea that many microorganisms present with outer membrane receptors for the binding and internalization of their ferrisiderophore complexes. It is not unreasonable to assume that on binding to the microbial receptors, the iron siderophore complex is at least initially exposed. If sufficiently antigenic, this “ferrisiderophore face” could represent a significant target in vaccine development. The question then becomes what should the expectations be regarding the antigenicity of a ferrisiderophore fixed to a large carrier molecule? If indeed it were very antigenic, this would merit the assembly of ferrisiderophores with functionality that allow for covalent linkage to the transporter and isolation of the adduct as a potential vaccine.
The antigenicity of a ferrisiderophore bound to a large carrier molecule is the focus of this manuscript. The specific questions addressed here are the following: Is it possible (1) to assemble a carrier siderophore conjugate, i.e., a protein carrier conjugate, 2) to raise antibodies to the conjugate in mice, and 3) to assess the antigenicity of the protein siderophore and its iron complex?
The current study focuses on the generation of antibodies against vibriobactin (3, VIB), the hexacoordinate iron chelator, a siderophore, responsible for iron utilization in Vibrio cholerae.48,49 This was attractive for two reasons: Vibrio cholerae represents an important pathological target,50–52 and we had established critical information about vibriobactin chemistry in earlier studies.32–36 Accordingly, we elected to investigate an ovalbumin (OVA)-vibriobactin protein conjugate (4, OVA-VIB) as an antigen.
The fundamental issue would be appending a tether to vibriobactin (3) (Figure 1), which would allow for fixing the ligand to a carrier protein, in this case, both OVA and bovine serum albumin (BSA). This demanded a synthetic approach very different from the assembly of vibriobactin itself.35 The OVA-VIB conjugate (4) would be used as an antigen to raise antibodies in mice, and the BSA-VIB conjugate (5) (Figure 2) would be utilized in an enzyme-linked immunosorbent assay (ELISA), first for the detection of serum polyclonal antibodies and, finally, vibriobactin-specific IgG monoclonal antibodies. Thus, choosing the appropriate activated tether for the vibriobactin protein conjugate was the first hurdle. While a number of different tethers were considered (e.g., acyl, halo, thiol), previous experience with hypusine antibody generation53 encouraged pursuit of a thiol-containing tether. The final ligand would be 1-(2,3-dihydroxybenzoyl)-5,9-bis[[(4S,5R)-2-(2,3-dihydroxyphenyl)-4,5-dihydro-5-methyl-4-oxazolyl]carbonyl]-14-(3-mercaptopropanoyl)-1,5,9,14-tetraazatetradecane (6), or vibriobactin thiol (Figure 2).
The first approach to vibriobactin thiol (6) began with the conversion of norspermidine (7) to polyamine reagent 12, that is, thermospermine54 protected at the aminobutyl terminus (Scheme 1). Specifically, norspermidine (7) was heated at reflux with 4-chloro-1-butanol in 1-butanol in the presence of K2CO3 and KI, producing linear triamino alcohol 8. The amine groups of 8 were masked as tert-butyl carbamates using di-tert-butyl dicarbonate in THF to generate alcohol 9 in 42% yield for two steps. In spite of the moderate yield, the conversion of 7 to 9 in Scheme 1 is considerably shorter than our previous route to N1-(4-hydroxybutyl)-N1,N4,N7-tris(tert-butoxycarbonyl)norspermidine (9).55 The hydroxyl of 9 was activated as its tosylate 10 in 92% yield by treatment with TsCl in CH2Cl2 and NEt3.55 Heating 10 with potassium phthalimide in DMF at 85 °C gave the fully protected tetraamine 11 in 71% yield. The carbamates of 11 were cleaved quantitatively with trifluoroacetic acid (TFA), affording reagent 12. Activation of 2,3-dimethoxybenzoic acid with 1,1′-carbonyldiimidazole (CDI) and acylation of the primary amine of 12 in CH2Cl2 and NEt3 gave masked catecholamide 13 in 70% yield. Next, the N-hydroxysuccinimide ester of N-tert-butoxycarbonyl-L-threonine56 (3 equivalents) was coupled to the internal nitrogens of 13 in DMF to produce triamide 14 in 55% yield. The phthalimide protecting group of 14 was removed in 65% yield with hydrazine hydrate in EtOH, generating primary amine 15, two equivalents of which were connected using 3,3′-dithiodipropionic acid (CDI in CH2Cl2), furnishing disulfide 16 in 40% yield. Threonylamine and catecholamide deprotection and also disulfide cleavage occurred in the presence of BBr3 in CH2Cl2, giving dihydrobromide salt 17 in 73% yield. Unfortunately, any attempt at cyclization of the threonyl fragments of 17 with ethyl 2,3-dihydroxybenzimidate34 resulted in complex mixtures, including thioesters, that were virtually impossible to separate, thus dooming the route of Scheme 1 to failure in the last step.
A catechol protecting group other than methyl, that is, one that could be removed concurrently with the BOC functionality while leaving the disulfide intact, was required. Thus, 2,3-dihydroxybenzoic acid (18) was converted to its trianion with NaH in DMF and treated with excess 4-methoxybenzyl bromide to make ester 19 in 62% yield. Hydrolysis of 19 with NaOH (aqueous) in dioxane produced 2,3-bis(4-methoxybenzyloxy)benzoic acid (20) in 90% recrystallized yield (Scheme 2). Activation of an equivalent of carboxylic acid 20 with CDI and stirring with N12-(phthaloyl)thermospermine (12) in CH2Cl2 and NEt3 afforded diamine 21 in 57% yield (Scheme 3).
The secondary amines of 21 were acylated with the active ester of N-(BOC)-L-threonine as before to generate tetraamine derivative 22 in 60% yield. The phthalimide functionality of 22 was cleaved in 90% yield with hydrazine hydrate in EtOH at room temperature, yielding primary amine 23, two equivalents of which were joined utilizing 3,3′-dithiodipropionic acid (CDI in CH2Cl2), resulting in disulfide 24 in 60% yield. The threonyl carbamates and the 4-methoxybenzyl ethers57 of 24 were simultaneously cleaved, using TFA in anisole and CH2Cl2; Sephadex LH-20 purification resulted in a 60% yield of disulfide 25, a tetrakis(TFA) salt. The threonyl moieties of 25 were next condensed with excess ethyl 2,3-dihydroxybenzimidate34 in refluxing EtOH to produce tetrakis(oxazoline) disulfide 26 in 20% yield. Finally, the disulfide bond of iron chelator 26 was reduced to the free thiol 6 in 60% yield, utilizing H2 (3 atm) over Pd black in CH3OH under iron-free conditions (Scheme 3).
Vibriobactin thiol (6), freshly generated from disulfide 26 (Scheme 3), was incubated with a maleimide-activated OVA (27) or BSA (28) protein carrier for 8 h. Unreacted maleimide was capped by conjugating it further with cysteine for 8 h, resulting in Michael adduct OVA-VIB (4) or BSA-VIB (5), respectively (Scheme 4). Both 4 and 5 were purified on a dextran desalting column. Positive fractions, as determined by the optical density (OD) at 280 nm, were analyzed for protein concentration using a Coomassie assay.58 Functionalities 29 and 30 were also generated (Scheme 4); 30 was used as a negative control in evaluating the capacity of 6 as an antigenic determinant when bound to BSA.
Protein desferrivibriobactin-OVA complex 4 was converted to the corresponding ferric siderophore complex 31 by mixing the conjugate with excess ferric nitrilotriacetate in phosphate buffer for 2 h. At this point, desferrioxamine (1) (Figure 1) was added to complex excess iron. The mixture was then purified on a G-25 Sepharose column. The same ferration procedure, including purification, was carried out on the maleimide-activated OVA (27), producing ferric protein complex 32. The iron content of the purified ferrivibriobactin protein adduct (31), the iron-treated maleimide protein (32), and the elution buffer were determined by inductively coupled plasma mass spectroscopy (ICP-MS). The background iron, elution buffer iron, and maleimide protein iron were subtracted from the iron content associated with the ferrivibriobactin OVA complex (31). From this measurement, and assuming that Fe(III) and vibriobactin form a 1:1 complex, the coupling efficiency of OVA maleimide (27) to 6 was 30%.
Vibriobactin (3) and vibriobactin disulfide (26) iron(III) complexes, 33 and 34 respectively, which were to be evaluated as potential antigens, were prepared as previously described.59 The resulting suspensions were separated on a small C-18 column, eluting with EtOH (aqueous). Colored fractions were pooled and lyophilized.
The procedures were similar to those of Kao and Klein60 and Simrell, et al.61 Briefly, the OVA-VIB conjugate (4) was used as an antigen to raise antibodies in mice. Antigenic response was determined via an ELISA. Once an adequate IgG response was observed, an immunized mouse was given a final, prefusion booster of the antigen without adjuvant. The mouse was euthanized four days later and antibody-forming cells from the animal’s spleen were fused to tumor cells grown in culture. The resulting hybridomas were screened for antibody production; antibody-producing hybridomas were cloned. Monoclonal antibodies were then produced and purified.
Two mice (M1 and M2) were immunized with 4. The immune response (serum titer) of these animals and non-immunized mice (normal mouse sera, NMS) was determined via ELISA for polyclonal antibodies against the following potential antigens: a) BSA-VIB conjugate (5), b) BSA-cysteine conjugate (30), c) vibriobactin thiol (6), d) vibriobactin disulfide (26), e) vibriobactin disulfide-iron complex (34), f) vibriobactin (3), and g) vibriobactin-iron complex (33). Twenty-three days after the second immunization, the serum from the mouse immunized with a higher dose of 4 (M2) seemed more active against antigen 5 than M1, with a higher p-nitrophenol OD at 405 nm at all dilutions (Table 1). However, by day 56, there was little difference in the reactivity of serum from M1 vs. M2. At this time, even after a 25600-fold dilution, the serum titers of the immunized mice against 5 were over 9 times greater than that of the NMS (Table 1).
The mouse sera also contained polyclonal antibodies against 30 that were nearly as active as antibodies against 5 (Table 1). Since cysteine was used to cap unreacted maleimide sites in the synthesis of 4, this was expected. As will be discussed below, this was not an issue for the purified monoclonal antibodies; we were able to select for antibodies specific against 4. The mouse sera did not react against any of the other antigens, c–g, (data not shown). This could have been attributed to a simple lack of activity in the case of antigens c–g or the nature of the ELISA itself. Antigens c–g are relatively low molecular weight, moderately water-soluble ligands. These compounds may not have adhered to the ELISA wells, or they may have been removed during the washing steps. In order to settle this issue, a series of competitive binding ELISAs were performed.
In the competitive binding ELISA, sera from immunized mice or non-immunized mice were first incubated with potential antigens f and g, or with 4, at antigen concentrations ranging from 0–250 μg/mL. If an antigen is an effective competitor, it will bind to the antibody during this initial incubation, leaving less antibody available to bind to a second antigen coated on the ELISA plate. This “competition” will be reflected in lower p-nitrophenol optical density values.
Antigen 4 was found to be an effective competitor at all concentrations tested (Figure 3, Table 2). The smaller antigens f and g were not effective competitors (data not shown), verifying the necessity for a large carrier molecule in order for the antibody to recognize vibriobactin (3). Unconjugated OVA (27) was not an effective competitor. The p-nitrophenol optical density of serum incubated with 27 against antigen 5 was 92% greater than serum first incubated with 4 at the same concentration (Figure 3). It is clear that the antibody “recognizes” the siderophore on the protein carrier.
Four months after the second immunization, mouse M2 was given a prefusion booster of 4. Fusion was done following the method of Simrell, et al.,61 except that the myeloma cell line used was Sp2/0. In short, spleen cells were fused with Sp2/0 cells such that the ratio of spleen cells to Sp2/0 cells was 7:1. After incubating for 11 days, the supernatants from the resulting hybridomas were tested via ELISA against antigen 5. The hybridomas that gave the strongest ELISA signal were further cultured for 7 days. Their supernatants were tested again via ELISA against 5, BSA-VIB-iron (35), and 30. The class of antibody (IgG or IgM) was also determined. The twelve most active hybridomas that were 5 and 35 positive and 30 negative are shown in Table 3. One hybridoma, 2D6, was IgM-positive; the remaining eleven hybridomas were IgG-positive (Table 3). Two of the most promising IgG-positive hybridomas, 5A6 and 2F10, were cloned.
Cells from the 5A6 and 2F10 hybridomas were diluted to a concentration of one or two cells per well. After incubating for four days, the plates were scanned microscopically and were scored for single colony and multiple colony wells. Ten days after seeding, supernatant from the wells that contained cells underwent screening via ELISA against 5. Supernatants from the single colony wells of 5A6 were not very active against 5 (data not shown). Because of this, the four most 5 positive multiple colony wells of 5A6 were pooled, diluted, and replated as single cells. After ten days, supernatants from the resulting clones were tested by an ELISA against 5. The ten single colony wells with the strongest ELISA positives (primary) against 5 are shown in Table 4. After incubation for an additional 5 days, the supernatants were tested again (secondary) via ELISA against 5, 35 and 30. All ten clones were highly active against 5 and 35 and showed little activity towards 30 (Table 4). A positive 5 and 35 response and a negligible 30 response were a clear indication that the antigenic determinants of the antibody were associated with the vibriobactin (3) segment of 5, and not with BSA (28) itself.
Recloning of the 2F10 hybridoma was unnecessary: 2F10–1A9 and 2F10–2A3 were highly active against 5 and 35 and poorly responsive to 30 (Table 5). These two clones, as well as two clones from 5A6 (5A6–2D5 and 5A6–1G8), were selected for further evaluation. The clone supernatants were assayed for the class of antibody, IgG or IgM, and were shown to be IgG-positive (Table 5). Multiple stocks from each cell line were frozen. 5A6–1G8 and 2F10–2A3 were tested for mycoplasma contamination and were found to be negative. Additional monoclonal antibodies (mAb) derived from 5A6–2D5 and 2F10–1A9 were produced and purified.
Competitive binding ELISA studies using the purified mAb were conducted. The mAb were first incubated with varying concentrations of antigen 4, from 0–250 μg/mL, prior to being transferred to an ELISA plate that had been coated with antigen 5. Antigen 4 was found to be an effective competitor at all concentrations tested (Figure 4, Table 6); 27 was not an effective competitor. The p-nitrophenol optical density of the mAb initially incubated with 27 against antigen 5 was approximately 90% greater than those of the mAb first incubated with 4 at the same concentration (Table 6). It is clear that the antibody “recognizes” the siderophore on the protein carrier.
One of the observations that stands out with the data from the antibody-containing hybridoma supernatants is the lack of difference in reactivity between 5 and its iron complex (35) (Table 3, Table 5). There are profound differences in structure between vibriobactin (3) and its 1:1 iron complex (33). Complex 33 would have all of the donor groups, (e.g., aromatic hydroxyls and oxazoline nitrogen) folded into the metal.48 Catecholamides such as 3 bind iron very tightly, with formation constants of nearly 1048 M−1.13 This means that because of the ubiquitous nature of iron, the iron complexes 31 and 35 were probably formed in vitro. In fact, it is likely that antibodies in animals are being formed against the OVA-VIB iron complex (31).
To support this idea, bile duct-cannulated rats were given 3 subcutaneously (s.c.) at a dose of 75 μmol/kg. The rodents’ bile and urine were collected for 48 h to determine if 3 sequestered and promoted the excretion of iron. The iron content of the bile and urine was determined using atomic absorption spectroscopy. We have used this model for many years to evaluate the efficiency with which iron chelators promote the excretion of iron from animals.62,63 Vibriobactin (3) was indeed found to sequester iron in vivo. Recall that 3 forms a tight 1:1 iron complex with Fe(III). Rats given 75 μmol/kg of 3 would be expected to clear 75 μg-atoms/kg of iron if the binding and clearance were 100% efficient. However, the iron clearing efficiency of 3 in the rats, i.e., the actual amount of iron cleared by the ligand vs. the theoretical iron clearance, was only 4.3 ± 1.1%. This implied that up to 95% of the free ligand remains, or is cleared uncomplexed. In the bile duct-cannulated rats this, of course, all unfolds in a matter of hours. However, in the mouse immunization experiments, 4 had weeks to become saturated with iron; the most important issue is the iron to ligand ratio. Assuming that 3.23 μg-atoms of iron/kg is available for chelation, in a 25 g mouse 1.45 × 10−9 moles of iron is available. The mice were effectively given 0.33 × 10−9 (M1) or 0.66 × 10−9 (M2) moles of 4. In view of the protracted exposure of the ligand to iron, it is difficult to imagine that all of the 3 bound to OVA would not also be iron-bound.
The current study focused on methodologies for assembling antigens that would allow for the assessment of the antigenic properties of vibriobactin fixed to large carrier molecules, e.g., OVA (27) and BSA (28). Choosing the appropriate activated tether for the vibriobactin protein conjugate was the first hurdle. While a number of different tethers were considered, a thiol analogue was chosen, 1-(2,3-dihydroxybenzoyl)-5,9-bis[[(4S,5R)-2-(2,3-dihydroxyphenyl)-4,5-dihydro-5-methyl-4-oxazolyl]carbonyl]-14-(3-mercaptopropanoyl)-1,5,9,14-tetraazatetradecane (6), or vibriobactin thiol (Figure 2).
The first attempt at synthesizing 6 (Scheme 1) unfortunately failed in the last step. Cyclocondensation of ethyl 2,3-dihydroxybenzimidate34 with the threonyl units of thiol 17 led to intractable mixtures, including thioesters. It became clear that the free thiol could not be released prior to this cyclocondensation. In a second approach (Scheme 2), a different catechol protecting group was employed in 2,3-bis(4-methoxybenzyloxy)-benzoic acid (20). This could be removed concurrently with the BOC functionality of the key intermediate 24 to produce 25, while leaving the requisite disulfide intact (Scheme 3). The threonyl moieties of 25 were next condensed with excess ethyl 2,3-dihydroxybenzimidate.34 Finally, disulfide iron chelator 26 was cleaved to the vibriobactin thiol (6), utilizing H2 (3 atm) over Pd black in CH3OH under iron-free conditions (Scheme 3). The thiol (6) was then incubated with a maleimide-activated OVA (27) or BSA (28) protein carrier, resulting in Michael adduct OVA-VIB (4) or BSA-VIB (5), respectively (Scheme 4). Conjugate 4 was mixed with an adjuvant and was successfully used as an antigen to raise antibodies in mice, and conjugate 5 was used in an ELISA, first for the detection of serum polyclonal antibodies, and ultimately vibriobactin-specific IgG monoclonal antibodies.
It is clear that further characterization of the antibody-antigen binding is required, e.g., stoichiometries, formation constants, etc. The results to date are consistent with the idea that vibriobactin (3) presents strong antigenic determinants when fixed to a large carrier molecule. The fact that 4 given to rodents produced an antibody so active against 5 begs the question regarding the antibodies’ potential in controlling the course of a vibrio infection in an animal model. We believe the data are further in keeping with the idea that covalently linking siderophore analogues to siderophore bacterial outer membrane receptors38 or to defined siderophore outer membrane receptor c-terminal peptide fragements represents a credible target for vaccine development. Although there are many potential pitfalls, nevertheless, because of the above data, we feel compelled to look at this approach. This will be the subject of a future manuscript.
The animal-related protocols were approved by the University of Florida Institutional Animal Care and Use Committee. Three female Balb/c ByJ mice (6–7 weeks old) were purchased from The Jackson Laboratory (Bar Harbor, ME). TiterMax Adjuvant was obtained from Sigma (St. Louis, MO). ELISA plates (MaxiSorp) were purchased from Nalge Nunc International Co. (Naperville, IL). An automatic microplate washer (Model EL404, Bio-Tek Instruments, Inc., Winooski, VT) and microplate reader (Spectromax Plus 384, Molecular Devices, Union City, CA) were utilized. The rabbit anti-mouse IgG (whole molecule), goat anti-mouse IgG (γ-chain specific) and goat anti-mouse IgM (μ-chain specific) antibodies were purchased from Sigma (St. Louis, MO). Dulbecco’s modified Eagle’s medium (HyClone) was obtained from Theromo Fisher Scientific (Waltham, MA). Hybridoma plates were incubated in a Forma Scientific Incubator, Model 3154 (Marietta, GA). A MycoAlert Kit (Lonza, Allendale, NJ) was used to assess potential mycoplasma contamination. Monoclonal antibodies were produced using hybridoma production media, BD Cell Mab Medium, Quantum Yield, (BD Biosciences, San Jose, CA) supplemented with 10% low IgG fetal bovine serum (HyClone, Theromo Fisher Scientific, Waltham, MA) in CELLine CL 350 flasks (Sartorius Stedim, New York, NY). Amicon Ultra-15 centrifugal filters with a 30 kDa cutoff were obtained from Millipore (Billerica, MA). Male Sprague-Dawley rats (400–450 g) were procured from Harlan Sprague-Dawley (Indianapolis, IN). Cremophor RH-40 was provided by BASF (Parsippany, NJ). An atomic absorption spectrometer, Perkin-Elmer model 5100 PC (Norwalk, CT), was used to determine the iron content of the rat bile and urine samples.
To produce antibodies to small molecules in animals, conjugation to larger carrier proteins (e.g., 27 or 28) is generally required. Compound 6 was conjugated with 27 using a linker to prepare 4, which was mixed with an adjuvant and used as an antigen to immunize two mice. One mouse (M1) received 50 μg of 4 per s.c. injection and the other mouse (M2) received 100 μg of 4 per s.c. injection. The mice were given a second immunization of 4 at the same doses four weeks later.
The method followed the approach of Kao and Klein.60 Briefly, the assay involved coating a potential antigen (50 μL/well) on the ELISA plates, utilizing solutions of antigens ranging in concentration from 1–40 μg/mL. The antigens: a) BSA-VIB conjugate (5), b) BSA-cysteine conjugate (30), c) vibriobactin thiol (6), d) vibriobactin disulfide (26), e) vibriobactin disulfide iron complex (34), f) vibriobactin (3), and g) vibriobactin iron complex (33) were diluted in blocking buffer (1% BSA in PBS with 0.02% azide). Normal mouse sera (NMS) or medium served as negative controls. Polyclonal serum from an immunized mouse (M2) in a 1:1000 dilution or hybridoma supernatant (Tables 4 and and5)5) were used as positive controls.
The plates were allowed to incubate overnight at 4 °C. An ELISA wash buffer (EWB) containing PBS with 0.02% azide and 0.5% Tween-20 was used to wash the plates. The plates were washed 4 times (300 μL/wash) using an automatic microplate washer. The wells were then blocked with 1% BSA in PBS with 0.2% azide for 1 h at room temperature and washed again. A 50 μL aliquot of diluted polyclonal mouse serum (1:200–1:25600), undiluted hybridoma supernatant, or purified mAb (0.11 μg protein/mL) was added to each well. The plates for this and each subsequent step of the ELISA were incubated with gentle agitation for 1 h at room temperature. The plates were then washed 4 times as above and rabbit anti-mouse IgG (whole molecule), conjugated to alkaline phosphatase, was added (50 μL/well); the IgG antibody was diluted (1:1000) in BSA-blocking buffer. The plates were washed 4 times as above and p-nitrophenyl phosphate at a concentration of 1.0 mg/mL, 100 μL/well was added. The plates were read using an ELISA plate reader at 405 nm, tracking the absorbance (OD) of the yellow water-soluble product, p-nitrophenol. A positive response was considered a test well with a p-nitrophenol OD value three times greater than that of the negative control.
The class of antibody (IgG or IgM) of the hybridoma supernatant (Table 3) or the clone supernatant (Table 5) was determined by an ELISA, replacing the rabbit anti-mouse IgG (whole molecule) with goat anti-mouse IgG (γ-chain specific) or goat anti-mouse IgM (μ-chain specific) at a dilution of 1:4000 (50 μL/well).
Competitive binding ELISAs were performed on 1) serum from immunized mice that contained polyclonal antibodies, and 2) purified mAb derived from the cloning of 5A6–2D5 and 2F10–1A9. Medium was used as a negative control, while 27 (10 μg/mL, 50 μL/well) served as a positive control for the competitive binding ELISA of the polyclonal serum (Table 2) and purified mAb (Table 6). Two types of plates were utilized: a conical bottom polypropylene plate was used for the incubation of the antibody-antigen mixture, whereas a 96-well MaxiSorp plate was used for the ELISA.
In brief, the polypropylene incubation plate was blocked with 300 μL of blocking buffer (1% BSA in PBS with 0.02% azide) and was allowed to incubate overnight at 4 °C. The following day, the blocking buffer was removed (flicked) from the plate, and the plate was blotted on a paper towel. A 96-well ELISA plate was coated with antigen 5 (10 μg/mL, 50 μL/well) that had been diluted in PBS with 0.02% azide. After incubating overnight at 4 °C, the plate was washed 4 times (300 μL/wash) as described above, blocked with 300 μL of blocking buffer for 1 h, and washed again.
Antigens d-g were diluted in blocking buffer in microcentrifuge tubes at antigen concentrations ranging from 0–250 μg/mL. The diluted antigens (50 μL/well) were transferred from the microcentrifuge tubes to the polypropylene incubation plate that had been blocked and washed. A 50 μL aliquot of the diluted polyclonal mouse serum (1:10000 dilution) or mAb (0.11 μg protein/mL) was added to the incubation plate wells containing the diluted antigens. The polypropylene plate was then incubated on a rocking platform for 2 h at room temperature, allowing time for the antigen-antibody complexes to form.
A portion of the antigen/antibody mixture (50 μL) was transferred from the polypropylene incubation plate to the 5-coated ELISA plate that had been blocked and washed. The ELISA plate was incubated with gentle agitation for 1 h at room temperature and was washed 4 times with the EWB. Rabbit anti-mouse IgG antibodies conjugated to alkaline phosphatase were added (50 μL/well); the IgG antibody was diluted (1:1000) in 1% BSA-blocking buffer. After 1 h, the ELISA plate was washed four times with the EWB to remove any unreacted IgG antibodies. p-Nitrophenyl phosphate substrate (1.0 mg/mL, 100 μL/well) was added. The ELISA plate was allowed to incubate with gentle agitation for 1 h at room temperature and was read at 405 nm, tracking the OD of the yellow water-soluble product, p-nitrophenol. Reduction in the OD value of a test well compared to the positive control reflects the effectiveness of the competitor binding to the antibody.
Four months after the second immunization, four days before fusion, a mouse (M2) was given a prefusion booster of 100 μg of 4 without adjuvant. This final immunization was given intraperitoneally (i.p.). On the day of fusion, the mouse was anesthetized, exsanguinated and euthanized. The spleen was removed and washed with Dulbecco’s modified Eagle’s medium to remove the antibody-forming cells. The medium was supplemented with 10% equine serum and 1× antibiotic-antimycotic (per mL: 100 I.U. penicillin, 0.10 mg streptomycin, 0.25 μg amphotericin B and 50 μg gentamycin). The fusion was performed by the procedures described in Simrell, et al.,61 except that the myeloma cell line used was Sp2/0 (a murine myeloma aminopterin-resistant cell line with a defect in purine metabolism). Spleen cells were mixed with myeloma cells at a 7:1 ratio. The fusion was performed with 50% polyethylene glycol 1500 followed by a controlled dilution with media. A centrifugation step (1500–1800 rpm for 8 minutes) was performed to pellet the fused cells. The pellet was resuspended in Dulbecco’s modified Eagle’s medium–high glucose, supplemented with 20% equine serum, 25% Sp2/0 myeloma conditioned medium and 1× hypoxanthine, aminopterin, thymidine (HAT) and was seeded in five 96-well plates at 2.8 × 105 cells per well. The plates were incubated at 37 °C, with a CO2 concentration of 7%.
Eleven days post-fusion, supernatants from the resulting HAT-resistant hybridomas were subjected to a primary ELISA screening to detect if the cells were secreting antibodies against 5. Cells from fifty-one wells that were positive in the primary screening against 5 were transferred to three 24-well plates and further incubated. The supernatant was assessed again one week later in a secondary ELISA screening against 5, 35, and 30 (Table 3). The hybridomas that were 5 and 35 positive and 30 negative were further tested as described above to determine the class of antibody, IgG or IgM (Table 3).
Two of the most promising hybridomas, 5A6 and 2F10, were cloned. Cells from the hybridomas were diluted to a concentration of one or two cells per well and were seeded in eight 96-well plates (four plates each for 5A6 and 2F10) over a feeder layer of irradiated 3T3 mouse fibroblasts. After incubating for four days, the plates were scanned microscopically and were scored for single colony and multiple colony wells. For the 5A6 cell line, 96 single colony wells and 60 multiple colony wells were found, whereas the 2F10 cell line presented 120 single colony wells and 47 multiple colony wells. Ten days after seeding, the supernatants from all 323 wells that contained cells underwent screening via ELISA against 5.
Supernatants from the single colony wells of 5A6 were not very active against 5 (data not shown). Because of this, the four most 5 positive multiple colony wells of 5A6 were pooled, diluted, and replated as single cells in four 96-well plates. Ten days after this recloning, the ten single colony wells with the strongest ELISA positives (primary) against 5 were chosen (Table 4) and cultured in a 24-well plate. After incubating for an additional 5 days, the supernatant was tested again (secondary) via ELISA against 5, 35 and 30. All ten clones were highly active against 5 and 35 and showed little activity toward 30 (Table 4).
Recloning of the 2F10 hybridoma was unnecessary. The ten single colony wells with the strongest ELISA response against 5 from the four 96-well plates were transferred to a 24-well plate. After incubating for an additional 5 days, the supernatants were tested against 5, 35 and 30. Two clones from 2F10 (2F10–1A9 and 2F10–2A3), as well as two clones from 5A6 (5A6–2D5 and 5A6–1G8), were chosen for further evaluation. Cell lines 5A6-1G8, 5A6-2D5, 2F10-2A3, and 2F10-1A9 were tested for mycoplasma contamination and were found negative (Table 5). Multiple stocks from each of the four cell lines were frozen.
5A6–2D5 or 2F10–1A9 cloned hybridoma cells were grown in hybridoma production media supplemented with 10% low IgG fetal bovine serum in CELLine CL 350 flasks. The flasks were incubated at 37 °C, with a CO2 concentration of 7%. One harvest per week was taken for three weeks. The cells were centrifuged at 2000 rpm for 15 min and the resulting supernatant was collected and purified by circulating through a 5 mL Protein G Sepharose 4B column. The supernatant recirculated through the column for 90 min (~23 passes). The column was then rinsed with PBS (70–150 mL). The antibodies were eluted using 0.1 M glycine, pH 2.8. Ten fractions of 3 mL each were collected. The eluted fractions were neutralized with 2.0 M Tris, pH 9.0. Absorbance was measured at 280 nm. Fractions with the highest absorbance readings were pooled, desalted, and concentrated using Amicon Ultra-15 centrifugal filters with a 30 kDa cut-off. The concentrated, purified monoclonal antibodies were recovered in a final volume of 0.6 mL in PBS. The protein concentration of the purified mAb were measured spectrophotometrically at 280 nm. Most mammalian antibodies (i.e., immunoglobulins) have protein extinction coefficients (εpercent) in the range of 12 to 15.64 The final protein concentration of the purified IgG antibody was estimated assuming a protein extinction coefficient of 14. For an IgG antibody with a molecular weight of approximately 150000, this corresponds to a molar extinction coefficient (ε) of 210000 M−1 cm−1.
Four rats were subjected to bile duct-cannulation as previously described.62,63 The rats were given 3 s.c. at a dose of 75 μmol/kg. The drug was solubilized in 40% Cremophor RH-40/water. Bile samples were collected from the rats at 3 h intervals for 48 h. The urine samples were taken at 24 h intervals. Sample collection and handling are as previously described.62,63 The iron content of the bile and urine were assessed by atomic absorption spectroscopy. Iron clearing efficiency was calculated as set forth elsewhere.65 The theoretical iron output of the chelator was generated on the basis of a 1:1 vibriobactin:iron complex.48
Reagents were purchased from Aldrich Chemical Co. (Milwaukee, WI), and Fisher Optima-grade solvents were routinely used and DMF was distilled. Organic extracts were dried with sodium sulfate and then filtered. Distilled solvents and glassware that had been presoaked in 3 N HCl for 15 min were employed in reactions involving chelators. Silica gel 70–230 from Fisher Scientific was utilized for column chromatography, and silica gel 40–63 from SiliCycle, Inc. (Quebec City, Quebec, Canada) was used for flash column chromatography. Sephadex LH-20 was obtained from Amersham Biosciences (Piscataway, NJ). An ICP-MS X 0675, manufactured by Thermo Electron Corporation (England), was used for the determination of the iron content of selected solutions/compounds. Derivatized OVA or BSA protein containing maleimide moieties 27 and 28, respectively, were obtained from Pierce (Rockford, IL).66 NMR spectra were obtained at 400 MHz (1H) or 100 MHz (13C) on a Varian Mercury-400BB. Chemical shifts (δ) for 1H spectra are given in parts per million downfield from tetramethylsilane for organic solvents (CDCl3 not indicated) or sodium 3-(trimethylsilyl)propionate-2,2,3,3-d4 for D2O. Chemical shifts (δ) for 13C spectra are given in parts per million referenced to 1,4-dioxane (δ 67.19) in D2O or to the residual solvent resonance in CDCl3 (δ 77.16). Coupling constants (J) are in hertz. The base peaks are reported for the high resolution mass spectra. Elemental analyses were performed by Atlantic Microlabs (Norcross, GA) and were within ±0.4% of the calculated values.
Derivatized proteins 2766 or 2866 (2 mg each) were dissolved in the conjugation buffer (100 μL) and incubated with a freshly prepared solution of 6 (2 mg in 200 μL of 50% aqueous DMSO) for 8 h. Unreacted maleimide was capped by conjugating it further with cysteine (2 mg in 50 μL degassed water) for 8 h. After incubation, 4 and 5 respectively were purified on a dextran desalting column (5 mL), eluting with 30% DMSO/purification buffer. Fractions (0.5 mL) were collected and the OD was measured at 280 nm. Positive fractions of each conjugate were pooled, and the protein concentration of each conjugate was estimated by a Coomassie assay.58
Palladium black (0.127 g) was added to a solution of 26 (0.231 g, 0.133 mmol) in anhydrous degassed EtOH (5 mL), and the mixture was stirred under H2 at 45 psi for 24 h. After filtration through Celite and washing the solids with degassed EtOH (2 × 5 mL), the filtrate was concentrated in vacuo to afford 0.137 g (60%) of 6 as a brown solid: 1H NMR δ 1.35–1.75 (m, 10 H), 1.8–2.2 (m, 4 H), 2.47–2.59 (m, 2 H), 2.82–2.91 (m, 2 H), 3.02–3.91 (m, 12 H), 4.72–5.0 (m, 2 H), 5.21–5.36 (m, 2 H), 6.58–6.77 (m, 3 H), 6.84–7.2 (m, 3 H), 7.24–7.8 (m, 3 H); HRMS m/z calcd for C42H53N6O12S, 865.3358 (M + H); found, 865.3466.
4-Chloro-1-butanol (16.5 g, 0.150 mol) was introduced to a mixture of 7 (39.4 g, 0.300 mol), KI (2.475 g, 15.0 mmol), and K2CO3 (10.5 g, 75.0 mmol) in anhydrous 1-butanol (375 mL). The mixture was stirred at 125 °C for 24 h, slowly cooled to room temperature, filtered and concentrated under vacuum to afford 8 as a viscous oil, which was dissolved in 50% aqueous THF (500 mL). Di-tert-butyl dicarbonate (130.8 g, 0.600 mol) in THF (100 mL) was added with continuous stirring. After 16 h, volatiles were removed, and the residue was dissolved in H2O (300 mL) and extracted with EtOAc (200 mL, 3 × 100 mL). The combined organic layers were washed with 0.5 M citric acid (100 mL), H2O (100 mL) and saturated NaCl (100 mL) and were concentrated under reduced pressure. Column chromatography using 49:49:2 hexanes/EtOAc/CH3OH provided 31.73 g (42%) of 9 as a viscous colorless oil. 1H NMR δ 1.44, 1.45, and 1.46 (3 s, 27 H), 1.5–1.8 (m, 9 H), 3.02–3.34 (m, 10 H), 3.67 (t, 2 H, J = 5.9), 4.78 and 5.27 (2 br s, 1 H); 13C NMR δ 25.20, 27.84, 28.57, 28.60, 28.61, 29.82, 37.72, 44.10, 45.01, 47.01, 62.57, 79.57, 155.75, 156.16, 174.84; HRMS m/z calcd for C25H50N3O7, 504.3648 (M + H); found, 504.3640. Anal. (C25H49N3O7) C, H, N.
10 was prepared from 9 by our previous reaction conditions.55 HRMS m/z calcd for C32H56N3O9S, 657.3659 (M + H), found, 657.3672. Anal. (C32H55N3O9S) C, H, N.
Potassium phthalimide (1.138 g, 6.15 mmol) was added to 10 (2.70 g, 4.10 mmol) in DMF (50 mL), and the reaction mixture was heated at 90 °C for 48 h. Solvent was removed in vacuo, and the residue was treated with H2O (50 mL) and extracted with CHCl3 (3 × 50 mL). The combined organic phase was washed with saturated NaCl and concentrated under reduced pressure. Column chromatography with 20% EtOAc/CHCl3 generated 1.84 g (71%) of 11. 1H NMR δ 1.43–1.45 (3 s, 27 H), 1.52–1.76 (m, 8 H), 3.06–3.29 (m, 10 H), 3.71 (t, 2 H, J = 6.8), 7.70–7.73 (m, 2 H), 7.83–7.87 (m, 2 H); 13C NMR δ 25.98, 27.46, 28.48, 29.49, 37.40, 37.60, 43.71, 44.87, 46.52, 78.88, 79.43, 79.65, 123.25, 123.41, 132.10, 133.99, 134.11, 155.45, 156.03, 168.41. HRMS m/z calcd for C33H52N4O8, 633.3858 (M + H); found, 633.3920. Anal. (C33H52N4O8) C, H, N.
TFA (25 mL) was added to 11 (3.46 g, 5.47 mmol) in CH2Cl2 (25 mL) with ice bath cooling, and the solution was stirred for 1 h at 0 °C and 1 h at room temperature. After removal of volatiles in vacuo, the residue was treated with toluene and dried by high vacuum to give 3.69 g (quantitative) of 12 as a white solid: 1H NMR (D2O) d 1.72–1.75 (m, 4 H), 2.05–2.15 (m, 4 H), 3.08–3.19 (m, 10 H), 3.71 (t, 2 H, J = 6.4), 7.81–7.87 (m, 4 H); 13C NMR δ (D2O) 23.24, 23.56, 24.37, 25.47, 37.10, 37.53, 44.90, 45.20, 45.29, 47.79, 123.91, 131.71, 135.37, 171.17; HRMS m/z calcd for C18H29N4O2 333.2285, (M + H, free amine); found, 333.2324. Anal. (C24H31F9N4O8 •1.5 H2O) C, H, N.
CDI (0.563 g, 3.48 mmol) was added to a solution of 2,3-dimethoxybenzoic acid (0.633 g, 3.48 mmol) in CH2Cl2 (5 mL). After being stirred for 1 h, the solution was cooled to 0 °C and was added to a suspension of 12 (2.82 g, 4.18 mmol) and NEt3 (2.46 g, 24.4 mmol) in CH2Cl2 (30 mL). The reaction mixture was stirred for 15 h at room temperature, diluted with CH2Cl2 (100 mL), and washed with 8% NaHCO3 (50 mL). The organic phase was concentrated in vacuo, and the residue was subjected to flash chromatography eluting with 5% concentrated NH4OH/MeOH to afford 1.20 g (70%) of 13 as a colorless oil: 1H NMR δ 1.48–1.57 (m, 2 H), 1.62–1.85 (m, 6 H), 2.61–2.73 (m, 8 H), 3.53 (q, 2 H, J = 6.0), 3.70 (t, 2 H, J = 7.6), 3.88 (s, 3 H), 3.89 (s, 3 H), 7.03 (dd, 1 H, J = 8.0, 1.6), 7.14 (t, 1 H, J = 8.0), 7.64 (dd, 1 H, J = 8.0, 1.6), 7.69–7.71 (m, 2 H), 7.82–7.84 (m, 2 H), 8.15 (br, 1 H, J = 7.2); HRMS m/z calcd for C27H37N4O5, 497.2765 (M + H); found, 497.2671. Anal. (C27H36N4O5) C, H, N.
A solution of freshly prepared N-tert-butoxycarbonyl-L-threonine N-hydroxysuccinimide ester56 (4.17 g, 13.2 mmol) in DMF (20 mL) was added to a solution of 13 (2.2 g, 4.4 mmol) in DMF (20 mL). After the mixture was stirred for 72 h, the solvent was removed under vacuum and the residue was taken up in CHCl3 (50 mL). The organic layer was washed with 5% NaHCO3 (3 × 50 mL), H2O (50 mL), and saturated NaCl (50 mL) and was concentrated in vacuo. Flash chromatography, eluting with 10% EtOH/EtOAc, furnished 2.17 g (55%) of 14 as a white foam: 1H NMR δ 1.14–1.22 (m, 6 H), 1.38–1.46 (m, 18 H), 1.54–2.02 (m, 8 H), 2.90–3.61 (m, 10 H), 3.68–3.76 (m, 2 H), 3.89 (s, 3 H), 3.91 (s, 3 H), 3.98–4.14 (m, 2 H), 5.44–5.68 (m, 2 H), 7.04 (d, 1 H, J = 8.4), 7.14 (d, 1 H, J = 8.0), 7.62–7.67 (m, 1 H), 7.70–7.72 (m, 2 H), 7.84–7.86 (m, 2 H); HRMS m/z calcd for C45H67N6O13, 899.4767 (M + H); found, 899.4783. Anal. (C45H66N6O13) C, H, N.
Hydrazine hydrate (5 mL) was added to a solution of 14 (0.350 g, 0.389 mmol) in EtOH (10 mL), and the reaction mixture was stirred at room temperature for 16 h. Solid was filtered, and the filtrate was concentrated in vacuo. The residue was taken up in 1 N NaOH (10 mL) and extracted with CHCl3 (3 × 10 mL), and organic extracts were concentrated. Flash chromatography eluting with 1% concentrated NH4OH/MeOH gave 195 mg (65%) of 15 as a viscous solid: 1H NMR δ 1.14–1.22 (m, 6 H), 1.38–1.46 (m, 18 H), 1.54–2.02 (m, 8 H), 2.68–2.76 (m, 2 H), 2.90–3.61 (m, 10 H), 3.89 (s, 3 H), 3.91–3.94 (m, 3 H), 3.98–4.14 (m, 2 H), 4.44–4.62 (m, 2 H), 5.46–5.64 (m, 2 H), 7.04 (d, 1 H, J = 8.4), 7.14 (t, 1 H, J = 8.0), 7.62–7.67 (m, 1 H); HRMS m/z calcd for C37H65N6O11, 769.4713 (M + H); found, 769.4709. Anal (C37H64N6O11) C, H, N.
CDI (0.10 g, 0.65 mmol) was added to a solution of 3,3′-dithiopropionic acid (0.06 g, 0.065 mmol) in CH2Cl2 (2 mL). After stirring for 2 h, a solution of 15 (0.60 g, 0.78 mmol) in CH2Cl2 (5 mL) was added to the reaction mixture. The solution was stirred for 15 h at room temperature and was diluted with CH2Cl2 (25 mL). The organic layer was washed with 1 N NaOH (15 mL) and saturated NaCl (15 mL) and was concentrated in vacuo. Flash chromatography using 15% EtOH/CHCl3 generated 0.218 g (40%) of 16 as a pale-brown solid: 1H NMR δ 1.10–1.25 (m, 12 H), 1.29–1.50 (m, 36 H), 1.52–2.05 (m, 16 H), 2.50–2.65 (m, 4 H), 2.90–3.08 (m, 4 H), 3.09 -3.60 (m, 24 H), 3.85–3.98 (m, 12 H), 3.99–4.60 (m, 8 H), 5.40–5.65 (m, 4 H), 6.98–7.15 (m, 2 H), 7.18–7.20 (m, 2 H), 7.60–7.69 (m, 2 H); HRMS m/z calcd for C80H135N12O24S2, 1711.9155 (M + H); found, 1711.9170.
Boron tribromide in CH2Cl2 (1 M, 7.56 mL, 7.56 mmol) was added to a mixture of 16 (340 mg, 0.199 mmol) in CH2Cl2 (20 mL) at −78 °C, and after being stirred for 1 h, the reaction mixture was warmed to room temperature and was stirred for 15 h. The reaction mixture was quenched cautiously at 0 °C with H2O (10 mL) and was stirred for 2 h. The aqueous layer was extracted with CH2Cl2 (20 mL) and was concentrated in vacuo, and the residue was subjected to a Sephadex LH-20 column eluting with 40% EtOH/toluene to give 115 mg (73%) of 17 as a white solid: 1H NMR δ 1.15–1.25 (m, 6 H), 1.46–2.01 (m, 8 H), 2.56–2.63 (m, 2 H), 2.94–3.00 (m, 2 H), 3.03–3.68 (m, 12 H), 4.04–4.20 (m, 2 H), 4.24–4.39 (m, 2 H), 6.83 (t, 1 H, J = 8.4), 7.15 (d, 1 H, J = 8.0), 7.30 (d, 1 H, J = 7.6); HRMS m/z calcd for C28H49N6O8S, 629.3333 (M + H, free amine); found, 629.3329. Anal. (C28H50Br2N6O8S •H2O) C, H, N.
Sodium hydride (60%, 4.69 g, 0.117 mol) was added in portions to 18 (5.47 g, 35.5 mmol) in DMF (150 mL) with ice bath cooling, and the mixture was stirred at 0 °C for 45 min and at room temperature for 1 h. A solution of 4-methoxybenzyl bromide (25.0 g, 0.124 mol) in DMF (50 mL) was added to the reaction mixture over 30 min. After the mixture was stirred for 20 h, quenching with H2O (30 mL) at 0 °C was performed, and solvents were removed under high vacuum. The concentrate was dissolved in EtOAc (200 mL), which was washed with H2O (100 mL) and saturated NaCl (100 mL); solvent was removed in vacuo. Flash chromatography, eluting with 5:1 hexanes/EtOAc, gave 11.33 g (62%) of 19 as a white solid, mp 108 °C: 1H NMR δ 3.77 (s, 3 H), 3.79 (s, 3 H), 3.81 (s, 3 H), 4.95 (s, 2 H), 5.03 (s, 2 H), 5.25 (s, 2 H), 6.76 (d, 2 H, J = 8.8), 6.86 (d, 2 H, J = 8.8), 6.89 (d, 2 H, J = 8.4), 7.04 (t, 1 H, J = 7.8), 7.11 (dd, 1 H, J = 8.0, 2.0), 7.18 (d, 2 H, J = 8.8), 7.33–7.36 (m, 5 H); 13C NMR δ 55.32, 55.37, 55.40, 66.81, 71.20, 75.31, 112.97, 113.61, 114.02, 118.16, 123.93, 127.08, 128.19, 128.75, 129.51, 129.73, 130.31, 130.41, 148.43, 152.95, 159.43, 159.62, 159.70, 166.44; HRMS m/z calcd for C31H30NaO7, 537.1889 (M + Na); found, 537.1884.
A solution of 19 (8.60 g, 16.71 mmol) in dioxane (84 mL) and 2 N NaOH (42 mL) was stirred for 24 h at room temperature. The reaction mixture was concentrated in vacuo. The residue was stirred with H2O (100 mL) and then acidified to pH 2 with 1 N HCl. The white solid was filtered, washed with hexane, and was recrystallised from EtOAc/hexanes to generate 5.93 g (90%) of 20 as a white crystalline solid, mp 129 °C: 1H NMR δ 3.8 (s, 3 H), 3.85 (s, 3 H), 5.12 (s, 2 H), 5.20 (s, 2 H), 6.83 (d, 2 H, J = 8.8), 6.96 (d, 2 H, J = 9.2), 7.18 (t, 2 H, J = 8.0), 7.22–7.27 (m, 2 H), 7.41 (d, 2 H, J = 8.4), 7.73 (dd, 1 H, J = 1.2, 8.0); 13C NMR δ 55.36, 55.37, 55.44, 55.46, 71.42, 114.26, 119.12, 123.04, 124.37, 125.00, 126.87, 128.02, 129.73, 131.21, 147.17, 151.45, 159.94, 160.44, 165.45; HRMS m/z calcd for C23H22NaO6, 417.1332 (M + Na); found, 417.1332.
CDI (0.239 g, 1.48 mmol) was added to a solution of 20 (0.583 g, 1.48 mmol) in CH2Cl2 (2 mL). After being stirred for 1 h, the solution was cooled to 0 °C and added to a suspension of 12 (1.0 g, 1.48 mmol) and NEt3 (1.05 g, 10.4 mmol) in CH2Cl2 (2 mL) at 0 °C. The solution was stirred for 15 h at room temperature and was diluted with CH2Cl2 (50 mL). The reaction mixture was washed with 8% NaHCO3 (25 mL) and was concentrated. Flash chromatography eluting with 5% concentrated NH4OH/MeOH afforded 0.597 g (57%) of 21 as a colorless oil: 1H NMR δ 1.48–1.72 (m, 8 H), 2.52–2.63 (m, 8 H), 3.36 (q, 2 H, J = 6.0), 3.69 (t, 2 H, J = 7.2), 3.80 (s, 3 H), 3.84 (s, 3 H), 4.98 (s, 2 H), 5.07 (s, 2 H), 6.83 (d, 2 H, J = 8.8), 6.93 (d, 2 H, J = 8.4), 7.13 (d, 2 H, J = 4.8), 7.23 (m, 3 H), 7.4 (d, 2 H, J = 8.4), 7.68–7.7 (m, 2 H), 7.81–7.83 (m, 2 H), 8.12 (t, 1 H, J = 7.2); HRMS m/z calcd for C41H49N4O7, 709.3596 (M + H); found, 709.3611. Anal. (C41H48N4O7) C, H, N.
A solution of freshly prepared N-tert-butoxycarbonyl-L-threonine N-hydroxysuccinimide ester56 (1.12 g, 3.54 mmol) in DMF (10 mL) was added to a solution of 21 (1.0 g, 1.41 mmol) in DMF (10 mL). After the mixture was stirred for 72 h at 40 °C, the solvent was removed under vacuum, and the residue was taken up in CHCl3 (50 mL). The organic layer was washed with aqueous 5% NaHCO3 (3 × 50 mL), H2O (50 mL), saturated NaCl (50 mL) and was concentrated in vacuo. Flash chromatography, eluting with 10% EtOH/EtOAc afforded 0.945 g (60%) of 22 as a white foam: 1H NMR δ 1.14–1.21 (m, 6 H), 1.37–1.49 (m, 18 H), 1.52–1.8 (m, 8 H), 3.10–3.60 (m, 10 H), 3.68–3.74 (m, 2 H), 3.79–3.81 (m, 3 H), 3.84 (s, 3H), 3.97–4.11 (m, 2 H), 4.32–4.60 (m, 2 H), 5.02 (s, 2 H), 5.07 (s, 2 H), 5.44–5.62 (m, 2 H) 6.82–6.86 (m, 2 H), 6.93 (d, 2 H, J = 8.4), 7.11–7.13 (m, 2 H), 7.22–7.26 (m, 3 H), 7.39 (d, 2 H, J = 8.4), 7.69–7.72 (m, 2 H), 7.81–7.85 (m, 2 H), 8.02 (br s, 1 H); HRMS m/z calcd for C59H79N6O15, 1111.5598 (M + H); found, 1111.5650. Anal. (C59H78N6O15) C, H, N.
Hydrazine hydrate (5 mL) was added to a solution of 22 (250 mg, 0.225 mmol) in EtOH (10 mL), and reaction mixture was stirred at room temperature for 16 h. Solid was filtered, and the filtrate was concentrated under high vacuum. The residue was taken up in 1 N NaOH (10 mL) and extracted with CHCl3 (3 × 10 mL), and organic extracts were concentrated. Flash chromatography, eluting with 1% concentrated NH4OH/MeOH afforded 198 mg (90%) of 23 as a white solid: 1H NMR δ 1.11–1.22 (m, 6 H), 1.37–1.43 (m, 18 H), 1.58–2.02 (m, 8 H), 2.71 (q, 2 H, J = 6.0), 2.90–3.60 (m, 10 H), 3.80–8.81 (m, 3 H), 3.84 (s, 3 H), 3.98–4.14 (m, 2 H), 4.30–4.60 (m, 2 H), 5.00–5.08 (m, 4 H), 5.42–5.65 (m, 2 H), 6.82–6.86 (m, 2 H), 6.93 (d, 2 H, J = 8.4), 7.13–7.19 (m, 2 H), 7.22–7.26 (m, 3 H), 7.4 (d, 2 H, J = 8.4); HRMS m/z calcd for C51H77N6O13, 981.5543 (M + H); found, 981.5589. Anal. (C51H76N6O13) C, H, N.
CDI (0.063 g, 0.372 mmol) was added to a solution of 3,3′-dithiopropionic acid (0.039 g, 0.186 mmol) in CH2Cl2 (2 mL). After the mixture was stirred for 2 h, a solution of 23 (0.550 g, 0.56 mmol) in CH2Cl2 (5 mL) was added followed by NEt3 (0.025 g, 0.25 mmol). The solution was stirred for 15 h at room temperature and was diluted with CH2Cl2 (25 mL). The organic layer was washed with 8% NaHCO3 (20 mL) and saturated NaCl (20 mL) and concentrated in vacuo. Flash chromatography, using 8% CH3OH/CHCl3, generated 0.238 g (60%) of 24 as a pale-brown solid: 1H NMR δ 1.10–1.23 (m, 12 H), 1.25–1.47 (m, 36 H), 1.48–2.00 (m, 16 H), 2.52–2.70 (m, 4 H), 2.80–3.01 (m, 4 H), 3.07–3.55 (m, 24 H), 3.80–3.81(m, 6 H), 3.83 (s, 6 H), 3.95–4.17 (m, 4 H), 4.27–4.62 (m, 4 H), 5.01–5.04 (m, 4 H), 5.07 (2 s, 4 H), 5.42–5.65 (m, 4 H), 6.82–6.86 (m, 4 H), 6.93 (2 d, 4 H, J = 8.4), 7.12–7.18 (m, 4 H), 7.22–7.27 (m, 6 H), 7.4 (d, 4 H, J = 8.4); HRMS m/z calcd for C108H159N12O28S2 2137.0859 (M + H), found 2137.0867. Anal. (C108H158N12O28S2) C, H, N.
TFA (25 mL) was added to a mixture of 24 (1.16 g, 0.547 mmol) and anisole (5 mL) in CH2Cl2 (25 mL) with ice bath cooling, and the solution was stirred for 1 h at 0 °C and 1 h at room temperature. After removal of volatiles in vacuo, the concentrate was subjected to a Sephadex LH-20 column, eluting with 40% EtOH/toluene to give 0.54 g of 25 (60%) as a white solid: 1H NMR (D2O) δ 1.24–1.30 (m, 12 H), 1.34–1.60 (m, 8 H), 1.74–2.21 (m, 8 H), 2.57–2.66 (m, 4 H), 2.84–2.93 (m, 4 H), 3.0–3.64 (m, 24 H), 4.04–4.20 (m, 4 H), 4.24–4.38 (m, 4 H), 6.80–6.86 (m, 2 H), 7.03–7.07 (m, 2 H), 7.18–7.21 (m, 2 H); HRMS m/z calcd for C56H95N12O16S2, 1255.6425 (M + H, free amine); found, 1255.6417. Anal. (C64H98F12N12O24S2 •2.5 H2O) C, H, N.
Ethyl 2,3-dihydroxybenzimidate34 (0.231 g, 1.27 mmol) was added to a solution of 25 (0.35 g, 0.21 mmol) in anhydrous EtOH (15 mL). The mixture was heated at reflux under N2 for 36 h and was concentrated in vacuo. Column chromatography on Sephadex LH-20, eluting with 15% EtOH/toluene afforded 0.072 g (20%) of 26 as a gray solid: 1H NMR δ 1.35–1.75 (m, 20 H), 1.8–2.2 (m, 8 H), 2.47–2.59 (m, 4 H), 2.82–2.91 (m, 4 H), 3.02–3.91 (m, 24 H), 4.72–5.0 (m, 4 H), 5.21–5.36 (m, 4 H), 6.58–6.77 (m, 6 H), 6.84–7.02 (m, 6 H), 7.08–7.24 (m, 6 H); HRMS m/z calcd for C84H103N12O24S2, 1728.6669 (M + H); found, 1728.7344. Anal. (C84H102N12O24S2) C, H, N.
Maleimide-activated protein 2866 (2 mg in 100 μL buffer) and cysteine (2 mg in 200 μL of degassed water) were incubated for 8 h at room temperature (Scheme 4). After incubation, 30 was purified on a dextran desalting column, eluting with a purification buffer. Fractions (0.5 mL) were collected and the OD was measured at 280 nm. Positive fractions were pooled and protein concentration was estimated by Coomassie assay.58
A 1-mL solution of 4 (1.685 mg/mL) was incubated at pH 7.4 with 1 mM FeNTA (500 μL) at room temperature. After the mixture was incubated for 2 h, excess 1 was added and the mixture again incubated for 2 h. The mixture was subjected to a G-25 Sepharose column, using 30% DMSO/phosphate buffer as an eluting solvent. Fractions (0.5 mL) were collected and the first colored band (fractions 3–6) was eluted; the OD was checked at 280 nm. Protein-containing fractions were pooled (4 mL) and subjected to ICP-MS for estimation of iron content.
The iron complexes of compounds 3 and 26 were prepared by the previously reported method.59
This paper is dedicated to Dr. William R. Weimar in recognition of his many contributions to our research program over the years. The project described was supported by Grant Numbers R37DK049108 (National Institute of Diabetes and Digestive and Kidney Diseases) and R01AI074068 (National Institute of Allergy and Infectious Diseases). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institute of Diabetes and Digestive and Kidney Diseases, The National Institute of Allergy and Infectious Diseases, or the National Institutes of Health. We thank Elizabeth M. Nelson for her technical assistance and Miranda E. Coger for her editorial and organizational support. We acknowledge the spectroscopy services in the Chemistry Department, University of Florida, for the mass spectrometry analyses and Diane G. Duke of the Hybridoma Core Laboratory, Interdisciplinary Center for Biotechnology Research, University of Florida, for the ELISA assessments of antibodies against the OVA-VIB conjugate.