Inhibition of Endotoxin Response by Synthetic TLR4 ... - Ingenta Connect

Current Topics in Medicinal Chemistry 2004, 4, 1147-1171

1147

Inhibition of Endotoxin Response by Synthetic TLR4 Antagonists Lynn D. Hawkinsa, William J. Christb and Daniel P. Rossignolc,* a

Eisai Research Institute, 4 Corporate Drive, Andover, MA, USA, bcurrent address: Ancora Pharmaceuticals, 700 Huron Avenue, Suite 18K, Cambridge MA, USA, cEisai Medical Research Inc., 55 Challenger Road, Ridgefield Park, NJ 07660, USA Abstract: Endotoxin, from the outer membrane of Gram-negative bacteria, has been implicated as the etiological agent of a variety of pathologies ranging from relatively mild (fever) to lethal (septic shock, organ failure, and death). While endotoxin (also known as lipopolysaccharide or LPS) is a complex heterogeneous molecule, the toxic portion of LPS (the lipid A portion) is relatively similar across a wide variety of pathogenic strains of bacteria, making this molecule an attractive target for the development of an LPS antagonist. Research over the past fifteen years focused on the design of various lipid A analogs including monosaccharide, acyclic and disaccharide compounds has lead to the development of E5564, an advanced, unique and highly potent LPS antagonist. E5564 is a stable, pure LPS antagonist that is selective against endotoxin-mediated activation of immune cells in vitro and in animal models. In Phase I clinical trials, we have developed an ex vivo endotoxin antagonism assay that has provided results on pharmacodynamic activity of E5564 in addition to the more typical safety and pharmacokinetic evaluations. Results from these assays have been reinforced by analysis of in vivo antagonistic activity using a human endotoxemia model. Results from all of these studies indicate that E5564 is an effective in vivo antagonist of endotoxin, and may prove to be of benefit in a variety of endotoxin-mediated diseases. This review discusses the evolution of synthetic LPS antagonists with emphasis on the SAR and development of E5564 and its precursors.

Key Words: Endotoxin, lipopolysaccharide, antagonist, sepsis, septic shock, Gram negative, E5564. I. INTRODUCTION Treating Sepsis Bacterial infection can be life threatening, requiring potential host organisms to protect themselves from infection by diligently maintaining an antimicrobial “state of awareness”. This awareness does not arise from a “learned” response such as through the development of antibodies but from an innate ability of our immune system to detect the presence of molecules specific to invading microorganisms. These bacteria-specific molecules include peptidoglycan and lipoteichoic acid from Gram-positive bacteria and lipopolysaccharide (LPS) from Gram-negative bacteria. LPS, also known as endotoxin, is an integral component of the outer membrane of Gram-negative bacteria. Insofar as LPS is unique to bacteria, common to Gram-negative pathogenic bacteria, and is essential for bacterial growth, it is not surprising that LPS is a “sentinel” molecule used by the host to detect infection by Gram-negative bacteria. Upon detection of endotoxin, the host responds with a rapid and robust antibacterial challenge that includes generation of a complex cascade of cellular mediators such as cytokines and chemokines among others [Figure (1)]. Secretory products released by monocytes, macrophages, neutrophils and other immune cells as part of this response can be directly bactericidal, can recruit other antimicrobial cells to assist in eradicating the infecting organism, or protect the host from *Address correspondence to this author at Eisai Medical Research Inc., 55 Challenger Road, Ridgefield Park, NJ 07660, USA; Tel. / Fax: 201 2872240 / 2340; E-mail [email protected] 1568-0266/04 $45.00+.00

further damage due to infection. Unfortunately, the continued presence or high levels of endotoxin in the blood (whether released from dead or dying bacteria or from the large amount associated with the normal flora within the lumen of the intestines) can result in a pathophysiological over-reaction resulting in release of toxic quantities of cellular mediators that can permanently harm the host. This reaction can result in the systemic inflammatory response syndrome (SIRS; ref[1]) and be followed by life-threatening responses such as vascular leak (leakage of fluid from blood vessels into surrounding tissues), tissue damage, hypotension, shock (commonly known as septic shock), organ dysfunction and failure, and death. A variety of approaches have been taken to alleviate the morbidity and mortality of patients due to severe sepsis and septic shock. These approaches aimed at interrupting events leading to a severe disease state are shown in Figure (2). Interventions that focus on different points in the cascade of events that lead to severe sepsis include: A) neutralizing LPS or blocking initial LPS-signaling events by preventing the generation of cell-surface signals; B) blocking the intracellular signals induced by endotoxin or the synthesis of cytokines and other cellular mediators; C) inhibiting the release of these cytokines and cellular mediators; D) blocking the receptors to the cellular mediators on their responsive target cells. E) inhibiting further “downstream” pathophysiological events such as acute respiratory distress or aberrant blood clotting. It is encouraging that after many years and many failed clinical trials, one treatment aimed at reducing pathological blood clotting has been found to significantly reduce mortality of the very sickest of septic © 2004 Bentham Science Publishers Ltd.

1148

Current Topics in Medicinal Chemistry, 2004, Vol. 4, No. 11

Hawkins et al.

Fig. (1). Sequence of events leading Endotoxin-induced sepsis and septic shock. Endotoxin released by bacteria is bound by lipopolysaccharide binding protein (LBP) and soluble CD14, and either transferred to serum lipoproteins (i.e. HDL) or to cell surface CD14. In turn, this complex stimulates TLR4/MD-2 to signal a proinflammatory response. This results in release of cytokines such as TNF-α, and interleukins such as IL-1β, IL-6, and IL-8 as well as other cellular mediators (NO and HMG-1). If this response goes out of control, toxic levels of these cellular mediators can trigger an inflammatory response that can induce endothelial cell activation, systemic hypotension, organ dysfunction and organ failure and death.

patients [2]. Discussion of each of the approaches is beyond the scope of this chapter, but detailed reviews can be found elsewhere that should provide an appropriate start to the comprehensive literature [3-6]. This chapter focuses on the strategy of blocking LPS at its cell-surface receptor by synthetic antagonists. II. BLOCKING ACTIVATION OF CELLS BY LPS AND LIPID A Elegant studies have provided a model to the sequence of events performed by the host to detect endotoxin with exquisite sensitivity and to “neutralize” the ability of LPS to trigger a response Figure (1). According to the current model [7-9], LPS initially binds to an LPS binding protein (LBP) in serum, which catalyzes the transfer of monomerized LPS from aggregate structures and in some cases from intact Gram-negative bacteria to membrane-bound CD14 (mCD14) on monocytes or monocytically-derived cells. mCD14 is not in integral membrane protein, but rather is bound to the membrane via a glycerolphosphatidylinositol (GPI) moiety, rendering CD14 unable to transduce signals across the membrane [10,11]. Final transmembrane signaling takes place when mCD14 next transfers or presents LPS to the transmembrane toll-like receptor-4, or TLR4 protein that works in conjunction with an obligate accessory protein MD2, forming a TLR4/MD-2 complex that initiates intracellular signaling triggering generation and release of a wide spectrum of cytokines and cellular mediators [12-18]. LPS may also be transferred by LBP to a soluble form of CD14 (sCD14) that can transfer endotoxin to mCD14 or TLR4 directly. A separate and opposing pathway that neutralizes LPS involves the transfer of LPS to lipoproteins Figure (1). Binding of LPS to HDL and other lipoproteins has been shown to dramatically reduce the ability of LPS to trigger

inflammatory response [19,20]. However, lipoproteins have been reported to be dramatically reduced in septic patients [21]. It is unknown if low cholesterol is a cause or effect of the underlying pathological septic response. Bacterial LPS is a large and complex molecule consisting of a polysaccharide (O-antigen) region, a core region that includes 3-deoxy-D-manno ketodeoxyoctulosonic acid (KDO) and a “lipid A” region (Figure (3). Lipid A from E. coli Figure (4) was first isolated by Westphal and Luderitz [22] and found to have toxicity, pyrogenicity and the ability to induce TNF equal to that of the parent LPS molecule [23]. Thus lipid A is considered the “toxicophore” of LPS. Wouldn’t antibodies to LPS reduce its toxicity? Bacterial “serotypes” have long-been described as strains of bacteria that induce a well-defined polyclonal antibody response. Many of these highly specific antibodies are directed against the entire LPS molecule (in general the O-antigen region), and can neutralize LPS from that serotype. Unfortunately, the narrow specificity of these antibodies for individual serotypes limits their usefulness. Other approaches to block the interaction of LPS with cell surface receptors used specific antibodies directed against LPS or parts of the LPS molecule with the hope that this interaction would enhance LPS clearance or neutralize the ability of LPS to activate cells [24,25]. More recent studies indicate that surgical patients having higher natural titres of antibodies to the core region of LPS appear to recover from surgery more readily and more completely [26,27]. However, the use of autologous antiendotoxin antibody or development of an immunization strategy has not been successfully explored. Antibodies directed against the more conserved lipid A region have proven in vitro to bind to a broad range of LPSs.

Inhibition of Endotoxin Response by Synthetic TLR4 Antagonists


1149

Fig. (2). Interventions that in the cascade of events that can lead to death due to sepsis. As described in the text, interventions to prevent death due to sepsis include: A) neutralizing LPS ; anti-LPS antibodies, bacterial increasing protein (BPI), adsorbing LPS to either soluble (conjugated) polymyxin B or polymyxin B bound to a column and blood treated “extracorporally”, increasing levels of plasma HDL. B) Blocking the intracellular signals induced by endotoxin or the synthesis of cytokines and other cellular mediators has been tested with kinase inhibitors such as inhibitors of P38 kinase, or processing inhibitors such as thalidomide and others. C) Inhibiting the release of cytokines and cellular mediators; as inhibitors of protein maturation (cell-surface protease inhibitors such as interleukin-1 converting enzyme). D) Blocking the receptors to the cellular mediators on their responsive target cells has been attempted with anti-TNF-α antibodies, soluble TNF-α receptors, and IL-1 receptor blockers (IL-1 receptor antagonist). Inhibiting further “downstream” pathophysiological events such as acute respiratory distress or aberrant blood clotting. To date, it is only here (reduction of pathological blood clotting) where an effective intervention has finally been found [97].

Unfortunately, all of these antibodies have shown varying degrees of activity both in vitro and in vivo, and have been ineffective in the clinic [28]. In addition it has been shown that antigenicity of chemically derived (acid-hydrolyzed) lipid A may be focused on C-6’ hydroxyl group which is newly generated when lipid A is formed by acid hydrolysis of LPS, and would be unavailable in vivo or in natural LPS. This would make antibodies to lipid A reactive only to hydrolyzed lipid A, leaving natural lipid A that is bound to the remaining polysaccharide region of LPS non-reactive [29]. Preliminary results from an active immunization strategy using a keyhole limpet hemocyanin (KLH) conjugated to synthetic monosaccharide analogs of Lipid X

show promise in mice [30]. Further investigation using more advanced models is warranted prior to determination of the potential of this vaccination strategy. The central role that lipid A plays in activating cells makes it clear that blocking the binding of lipid A at one or more of its binding proteins will block all downstream events leading to systemic inflammatory response and septic shock. Pharmacological receptor antagonists have been successfully shown to be derived from modification of a parent agonistic molecule [31]. For this reason, a logical first approach to developing an antagonist would be through modification of the core structure of natural lipid As derived

1150


Hawkins et al. OH O P OH OH O O O

NH O

OH

O OH

O

O O OH HO HO

OH O

HO HO

OH O

O

OH

O *

HO

O

OH HO

OH

HO O

O

O O

OH

HO HO OH OH O

O

O

n OH HO HO

O

O

O

OH

NHAc

NHAc

HO O O

H2 N

O

OH HO O

O HO HO O O P O P O O OH OH

HO HO O

O

O

O

O O

O

O PO OH OH CO2H

O

HO HO

O OH OH OH

O

O HO

HN

CO2H

NH2 O O HO P O HO O OH OH O HO O OH

OH

HO OH

O

O

O

OH

O

OH

O-Antigen

Outer Core

Inner Core

Lipid A Region

serotype specific repeating subunit

maybe absent in some species

highly charged functionalities

Toxicophore

Fig. (3). Representative structure of E. coli lipoplysaccharide. LPS is an integral component of the outer membrane of Gram-negative

from a variety of bacterial species. Features likely to be key to receptor recognition would be expected to be common to isolated toxic lipid As, non-toxic lipid As, and possibly their biosynthetic precursors. Studies toward elucidating the structure-activity relationships of substructures of E. coli lipid A have provided valuable insight into the significance of several portions of the molecule [32-36]. The reported antagonistic activities of the lipid A biosynthetic precursor, Lipid X, and the non-toxic, naturallyderived lipid As, Rhodobacter capsulatus and Rhodobacter (formerly Rhodopseudomonas) sphaeroides, both of which potently block the agonistic activity of E. coli lipid A [37,38] Figure (4) provided significant confirmatory support to this initial medicinal chemistry strategy. This chapter describes our efforts to develop pharmaceutically acceptable antagonists that would interact with the putative LPS receptor on the cell and block the action of LPS without demonstrating any “LPS-like” cellular activation on its own. III. SYNTHETIC LPS ANTAGONISTS A. General Strategy Prior to the initiation of a program to develop synthetic endotoxin receptor antagonists for the treatment of Gramnegative induced sepsis and related diseases, the following issues anticipated to be possible barriers to development were considered. What is the Most Appropriate Means of Administration? The majority of patients in severe sepsis and septic shock are treated in a hospital intensive care unit (ICU) and are so debilitated as to require intravenous (IV) administration of drugs and nutrition. Thus the best route of administration for an antagonist would be via IV. This route of administration also circumvents bioavailability issues usually encountered for drugs administered orally or via other routes.

What Assays Should be Used for Development? In vitro immune response to endotoxin is readily performed using purified monocytes/macrophages from peripheral human blood or blood from animal model systems. However, because therapeutic compounds are to be directly infused into blood, it must also be ascertained that soluble or cellular components of blood do not adversely affect activity. This requires use of an in vitro or ex-vivo bioassay that uses human whole blood as a tool to monitor the activities of the potential antagonists. Having a human “whole” blood assay to test antagonist activity allows us to measure the “preference” of antagonists for interaction with the targeted receptor instead of other consequential serum proteins, such as LBP or cholesterol sequestering proteins, that as described above, can sequester and “de-activate” LPS-like molecules. An in vivo model for testing sepsis therapies is also necessary. Unfortunately because all available sepsis models have one or more artificial components (e.g. rapid acute onset), no animal model for sepsis is considered ideal. The desired “test” animal for screening potential antagonists in vivo would be a rodent (mouse or rat). However, understanding the similarities and differences between the mouse infection model and the disease state in humans is important. Indeed this issue was deemed crucial to the development of the drug since, compared to humans, mice and rats are relatively insensitive to LPS and require substantially higher doses to induce the sepsis syndrome. Because the effective dose of antagonist may be dependent on the amount of LPS used as a “challenge”, efforts were first directed at reducing the amount of LPS needed to induce a septic state. This was done by activating or “priming” animals to enhance their sensitivity to lower doses of LPS. In addition, studies have demonstrated, that several natural and synthetic antagonists in human systems have been found to be agonists in the mouse systems [39]. Clearly, species-specific agonistic activity would confuse



1151

O HO HO

OH

P

O

O O

O

O

HO

O

O

O O

NH

O

O

NH

P

O

O O

OH

OH P OH

OH

O

OH

O

O O

NH

O

O

O

NH

O

P OH

O OH

O

OH

O

O OH

OH O

O

O O

OH O

O

O

E. coli Lipid A

R. capsulatus Lipid A RcLA

O HO HO

OH

P

O

O

O

O HO

O O

NH

O

O O

NH

O O

O OH

P

OH

OH

OH O

O O

R. s phaeroides Lipid A RsLA

Fig. (4). Comparison of Structures of Natural Agonistic and Antagonistic Lipid As from E. coli , R. sphaeroides proposed structure, and R. capsulatus proposed structure.

development of animal models to study antagonism, however, it is less obvious that agonistic activity (even if apparent at only very high doses) would make it difficult or impossible to test toxicity in a rodent model. Thus having a developmental candidate that is a pure antagonist in human as well as in the appropriate animal model systems would be required.

high cost of synthesizing and purifying such a complex molecule. Thus nanomolar or preferably sub-nanomolar antagonists were targeted. Naturally other pharmacological and toxicological issues that are encountered during drug development would also require attention.

What Milestones Should be Met Before Proceeding?

As discussed above, the initial strategy to produce an endotoxin antagonist was to use a known natural product antagonist or analog of an agonist as a lead structure. Prior to initiation of this strategy, confirmation of the structure and

High potency of our proposed target antagonist was deemed crucial to solve a number of issues such as likely

B. Medicinal Chemistry Strategy

1152


Hawkins et al.

biological activity of the natural product(s) was deemed essential prior to initiation of a medicinal chemistry program. The natural product(s) would be either: 1) acquired followed by a confirmation of the chemical structure and biologically activities, or 2) the proposed structure would be synthesized and followed by biological confirmation. Once confirmation was obtained, a synthetic and medicinal chemistry strategy was devised to address not only the biological issues described above but also the synthetic and medicinal chemistry issues required to convert the natural product into a drug. These additional issues included:

compounds with low antagonistic potency. In our hands, many monosaccharide antagonists such as I demonstrated good activity in vitro in assays utilizing cultured cells or primary cultures of human monocytes containing low concentrations of plasma. However, when tested in vivo in the mouse endotoxin challenge model or in vitro in the presence of higher concentrations of serum or plasma (such as in whole blood), the activity of the monosaccharides was attenuated. Furthermore, as described by Stuetz and coworkers at Sandoz [47,48], difficulties in purifying the final monosaccharides from undesirable, agonistic impurities made this approach extremely difficult.

• A robust synthetic route allowing one to obtain a variety of analogs based on the lead structure.

Workers at Sankyo reported results [49,50] on two similar tri-acyl monosaccharides containing an anomeric carboxylic acid and C-4 phosphate [II and III] that were shown to be potent inhibitors of LPS-induced TNF-α production in monocytes with IC50s of 0.005 and 0.017 µM, respectively. The use of an anomeric carboxyl group prevented the formation of undesirable, agonistic disaccharides as described by Auschauer et al.[47]. To date, the in vivo activities of these antagonistic monosaccharides have not been reported.

• Analogs that demonstrate chemical and physical stability during preparation and in vitro and in vivo biological evaluation. • Synthetic feasibility for large-scale synthesis (multikilograms). 1. Monosaccharide Antagonists Lipid X, Figure (5) a monosaccharide biosynthetic precursor of lipid A, was first isolated from a variant of E. coli and identified through the pioneering work of Raetz and colleagues [40,41]. Lipid X antagonizes LPS activity in vitro [42] and was reported to protect mice and sheep from LPSinduced lethality [43,44]. However, it was ineffective in vivo in a canine sepsis model [45]. Due its more simplistic structure, efforts to develop a monosaccharide-based endotoxin antagonist were deemed most appropriate. Unfortunately, these efforts met with limited success [46], yielding OH

O

OH O

O

HO

HO

O

O O

Surprisingly, our in-house assessment of monosaccharide activities in vitro using a simple competitive binding assay for radiolabelled LPS [51,52] indicated widely disparate structure activity relationships for inhibition of binding and LPS-mediated cellular activation. The implications of these findings are discussed in the Biology section below, but indicate that the use of a binding assay to screen LPS antagonists was not likely to be an effective strategy.

NH

O

P

O OH

HO HO

O

O O

NH

O O

OH O

OH

Lipid X

OH

OH O

P

P

OH

OH

O

O O

NH

I

OH

OH O

O X O

O O

X O

II (X = F) III (X = H)

Fig. (5). Structures of monosaccharide antagonists. Lipid X is a monosaccharide precursor of lipid A, I is the synthetic tri-substituted “righthalf” of lipid A, and II and III are previously described [49,50].


2. Disaccharide Antagonists The disappointing results obtained by us from our synthetic monosaccharide antagonists directed efforts toward the investigation of the non-toxic disaccharide lipid A natural products. The potential utility of disaccharide endotoxin antagonists was verified by reports describing the activity of lipid As from R. capsulatus, [37] and R. sphaeroides, RsLA [50,53,54]. Both lipid As possessed potent in vitro and in vivo anti-endotoxin activities that are desirable for further drug development. Preparation and structural verification of their proposed lipid A structures [Figure (4)], followed by biological confirmation of their antagonistic activities were undertaken. Significant effort was required for the preparation of the proposed lipid A disaccharides to confirm their chemical structures and biological activities. In addition it was most beneficial to establish a general synthetic method that allowed us to produce the targeted natural products and additional analogs to accomplish our ultimate goal. The elegant first synthesis of E. coli lipid A by Shiba’s group [55,56] laid the foundation of subsequent preparations of this and other lipid A-type disaccharides. Shiba’s strategy for the synthesis of E. coli lipid A was based on a convergent approach where appropriately functionalized and protected monosaccharides were condensed using classical KoenigsKnorr methodology to form the β-disaccharide at a late stage in the synthesis. This disaccharide intermediate was then further elaborated and finally deprotected to provide the desired natural product. The key to this strategy was the use of a suitable protection-deprotection paradigm to facilitate the incorporation of a variety of functionalities about both the glycosyl acceptor and donor at the appropriate time, to provide the desired β-disaccharide in high yield upon condensation of these monosaccharides, and to allow the desired functionalities to remain intact after final deprotection. The final deprotection of the fully functionalized disaccharide was executed via hydrogenolysis using palladium and platinum catalysts to provide synthetic E coli lipid A. The reported structures of the non-toxic lipid As from R. capsulatus (RcLA) and R. sphaeroides (RsLA) differ from toxic E. coli lipid A by the number of acyl chains on the glycosyl acceptor, the difference in length of the individual acyl chains, the presence of unsaturation on one of the acyloxy chains on RcLA and RsLA, and the presence of two β-oxo-acylamino chains on the C-2 and C-2’ positions on RcLA and one β-oxo-acylamino chain on the C-2 position on RsLA.. Although the overall structures of the three lipid As may look similar, the structural differences required a major modification to the protection/deprotection paradigm established in Shiba’s synthetic route. This is due to the incompatibility of the olefinic and β-oxo-functionalities to hydrogenolytic protecting group removal processes. Ultimately, we successfully established a convergent, efficient, and general route that permitted us to synthesize and confirm the structure proposed for the R. capsulatus lipid A. The key to this strategy was the development of an alternative protection/deprotection strategy by incorporating allyl- and allyloxycarbonyl-protecting groups as opposed to


1153

the phenoxy- and benzyl-protecting groups used in Shiba’s synthetic process. Synthetic RcLA (Table 1) appears to retain the antagonistic activities ascribed to the bacterially derived molecule. Although the natural product can be prepared on a laboratory scale, our experience demonstrated that the synthetic RcLA could not be readily synthesized and purified to pharmaceutical standards at the scale necessary for development (unpublished observations). In addition, the final product did not possess a suitable shelf life. Upon standing for extended periods or after treatment with acid or base, side products were generated that are weak agonists in murine systems. For example, deacylation at the C-3 or C-3’ position of the disaccharide generated weak agonists (IV and V). Additional side products included the elimination of the β-acyloxy chain on the C-3’ position (VI) along with the C-3 deacylation of the eliminated by-product (VII). To overcome the decomposition issues, a more robust disaccharide was envisioned. Most importantly it was determined that stabilizing RcLA-like molecules by conversion of C-3 and C-3' acyloxy functionalities to alkyloxy functionalities would be most appropriate. Shiozaki, et al. [57], showed that stabilization of E. coli lipid A with the generation of C-3 and C-3’-alkyloxy chains produced an analog which lost all endotoxin-like activity. Although Shiozaki’s results argued that conversion to ether linkages might reduce activity, ether conversion appeared necessary for the stabilization of a potential disaccharide antagonist. To our pleasant surprise the newly converted molecule, VIII, was found to be a potent LPS antagonist both in vitro and in vivo. During the preparation of synthetic RcLA and VIII, small amounts of inactive and agonistic impurities arose that appeared to be generated by the intramolecular nucleophilic attack of the free C-6’ hydroxy group toward the adjacent C4’ phosphate providing either a hydrolyzed or “dephospho” product, or a 4’,6’-di-O-cyclic phosphate product. Also, an intermolecular attack of the C-6’ hydroxy group on the C-1 position of a second disaccharide was observed that was similar to that observed from the impurity profile of Lipid X and its analogs. Obviously, these impurities complicated the final purification of the desired disaccharide. Since the natural lipid As are linked at the C-6’ position to a KDO moiety as part of the core region of the LPS molecule, it seemed logical that substitution on this position of the lipid A analog would be possible. Indeed, replacement of the C-6' hydroxyl group with a C-6’ methoxy group eliminated the formation of all of the possible impurities mentioned above. This final molecule, E5531, is a fully stabilized endotoxin antagonist with no detectable agonistic activities [58-60]. Lack of toxicity and mutagenicity, as well as availability of stable formulation established E5531 as a candidate for clinical development. In the clinic, E5531 showed promise as a candidate for the treatment of sepsis in being the first compound to completely block signs and symptoms of endotoxemia in normal volunteers. As little as 250 µg of E5531 completely inhibited human responses to 4 ng/kg LPS [61]. Unfortunately, further development of E5531 was hindered by difficulties in large-scale synthesis and

1154


Hawkins et al.

Table 1. OR 6 X2O R 5O

O

O NHR4

O

R 3O R2O

R1HN No.

sRcLA

X1

X2

O

O

P

OH

OH

IV

OH

O

Mouse Activityb

H

1-2