Human BPI: One Protein's Journey from
Laboratory into Clinical Trials
Challenges and frustrations attend efforts to bring
this and other proteins with promising activity into clinical use as
Protein pharmaceuticals, including interferons, growth
hormones, and erythropoetins, typically show high activity and
specificity at relatively low concentrations. Such properties help make
them invaluable drugs within their respective therapeutic categories.
Developing protein pharmaceuticals, however, can be an unpredictable
process, one that is further complicated because of the need to identify
highly specific conditions for maintaining product integrity and
functionality. Diversity of protein structure and dramatic differences
in physical characteristics make it difficult to describe general rules
for stabilizing protein functions; therefore, each protein needs to be
evaluated on a case-by-case basis. Although laborious and expensive,
this process remains an essential step when converting basic biological
research findings into therapeutic products.
Occasionally, the need for a new drug coincides with
research that identifies protein candidates suited to that therapeutic
purpose, overcoming the barriers to clinical development. For instance,
the emergence of antibiotic-resistant pathogens increases the worldwide
clinical demand for novel antimicrobials and other anti-infective
agents. Meanwhile, over several decades, researchers studying
polymorphonuclear leukocytes (PMN), the primary phagocytic white blood
cells responsible for the innate immune response, have discovered a
multitude of potent antimicrobial agents, many of which are peptides or
Among these antimicrobial proteins, the
bactericidal/permeability-increasing protein (BPI) is one of the most
promising, and researchers studying it report significant progress in
bringing it along the pathway from lab to clinic. However, despite its
potent anti-infective activity and other advantages, this journey has
proved far from straightforward.
Discovery and Early Characterization of BPI
In 1978, researchers at New York University in New York,
N.Y., isolated a protein from the primary granules of human PMNs that
increases the permeability of the outer membrane of gram-negative
bacteria (GNB) and thus can effectively kill many species within this
class. Peter Elsbach and Jerry Weiss, who applied the name BPI, showed
that its potent activity against GNB is related to its high-affinity
binding for lipopolysaccharide (LPS, also known as endotoxin), the major
component of the outer membrane of such bacteria. As part of the innate
immune response, PMNs phagocytose bacteria and release the contents of
their granules, including BPI, into the phagocytic compartment. BPI
immediately arrests bacterial growth, increases the permeability of the
outer and inner bacterial membrane, and eventually kills the bacterial
cells. The fate of bacteria treated with purified BPI closely resembles
what is seen within a PMN.
Exactly how BPI kills bacteria is not understood.
However, its initial interaction with GNB resembles that of a number of
cationic antimicrobial peptides, including polymyxin B (Fig. 1). When
the highly cationic (pI > 9.5) BPI protein binds to negatively
charged LPS molecules, divalent cations that normally bridge the
phosphorylated head groups of LPS are released, disrupting its organized
packing within the outer leaflet of the bacterial membrane.
Subsequently, the outer membrane changes in several additional ways. For
instance, it becomes permeable to hydrophobic antibiotics such as
actinomycin D. This is believed to happen when phospholipids in the
inner leaflet of the outer membrane flip into the outer leaflet,
creating small areas of phospholipid bilayers that actinomycin D is
known to easily cross. Moreover, free Ca2+ activates
endogenous phospholipases, which can degrade newly accessible areas of
phospholipid bilayers, contributing to a time-dependent deterioration of
the outer membrane. Initially, the effects of BPI can be reversed by
adding high concentrations of Mg2+, but eventually the
changes in cellular metabolism and the inner membrane become
irreversible, killing the bacterium.
Several Reasons Why BPI Is a Promising Antimicrobial
BPI catches the attention of people searching for new
antimicrobial agents for several reasons. First, it is highly potent and
specific against GNB, a class that includes many important human
pathogens. On a molar basis, BPI is at least 10 times more effective
than any other known mammalian antimicrobial protein or peptide. In
addition, unlike some highly effective peptide cytotoxins, BPI is
nontoxic to eukaryotic cells, even at high concentrations. Another
important feature of BPI is that it maintains activity under
physiological conditions and in the complex milieu of body fluids.
Although many mammalian peptides and proteins are active as
antimicrobial agents in vitro, few retain their activity in vivo.
Moreover, BPI not only is a potent bactericidal agent,
but it can also neutralize the biological activity of LPS, according to
Elsbach, Weiss, and others. LPS, on or derived from GNB outer membranes,
is widely recognized for its toxicity to humans and other higher
organisms. When introduced into the bloodstream, as might happen during
severe infections or following trauma, LPS can trigger a powerful
inflammatory response, leading to a potentially fatal condition known as
septic shock. Once an infection proceeds to the stage of septic shock,
very little can be done for patients in a clinical setting since the
underlying problem is no longer the bacteria, but rather an
out-of-control host immune response.
While many proteins bind with high affinity to LPS,
including several monoclonal antibodies that were tested clinically,
none effectively reduces the inflammatory effects of LPS in vivo. BPI
can effectively neutralize the toxic effects of LPS in vitro, in whole
blood ex vivo, and in animals, most likely mediated by its ability to
block the interactions of LPS with inflammatory signaling systems. Its
combined bactericidal and LPS neutralization activities, coupled with
the growing need for new anti-infective compounds, convinced scientists
and managers at XOMA Ltd., a biopharmaceutical company in
Berkeley, Calif., to bring BPI into clinical trials.
Facing the Challenge of Producing Stable, Highly
Before BPI could enter clinical testing, the developers
faced issues such as the need to develop an efficient, high-yield scheme
for purifying the protein while retaining its biological functions.
After overcoming a number of obstacles, researchers at XOMA eventually
developed an effective expression system and an efficient purification
BPI's potent bactericidal activity interferes with
expression in a number of convenient bacterial systems. Fortunately,
large amounts of active protein can be prepared using mammalian cell
cultures, such as transfected CHO-K1 cells, which secrete BPI into the
growth media. XOMA researchers discovered that adding S-Sepharose beads
to the growth media would capture this secreted protein, a method that
increases its recovery following fermentation.
The large size and complex structure of BPI also pose
challenges. BPI contains 456 amino acids, its molecular weight is
greater than 50 kDa, and its crystal structure reveals a complex
three-dimensional fold (Fig. 2). Elsbach and Weiss observed that, after
being stored several months at 4șC, samples of BPI can generate two
fragments of nearly equal size, corresponding to its N- and C-terminal
domains. Furthermore, BPI's N-terminal portion of about 200 amino acids
contains its bactericidal, LPS-binding, and LPS-neutralizing activitiesthus
greatly reducing the necessary size of genetic constructs for producing
the functional portion of this protein. These observations led to
creation of rBPI23, a recombinant 23-kDa fragment derived from the
N-terminal region of human BPI.
After successfully producing and purifying this version
of BPI, XOMA researchers faced regulatory challenges specifying
standards for product stability and homogeneity. They approached these
tasks in two ways: changing the protein itself and changing the
protein's environment. A highly purified protein often may display
considerable heterogeneity because of glycosylation, oxidation, or other
posttranslational chemical changes. However, although rBPI23 lacks
glycosylation sites, it does contain three cysteine residues that can
form disulfide bonds. During prolonged storage, rBPI23 forms covalent
dimers when neighboring protein molecules become linked via an
intermolecular disulfide bond.
Site-directed mutagenesis evaluating the roles played by
the three cysteine residues in rBPI23 demonstrated that two of these
residues are disulfide bonded and essential for maintaining its
biological functions. However, when the third cysteine is mutated,
rBPI23 no longer can form dimers, but still retains its biological
activity. Thus, XOMA chose a further modified version, designated
rBPI21, for clinical studies. This 21-kDa N-terminal fragment (residues
1-193), in which the cysteine at position 132 was replaced with alanine,
is biologically active but does not form dimers. Follow-up biophysical
studies identified a suitable buffer for stabilizing rBPI21, consisting
of poloxamer (Pluronic-F68) and polysorbate (Tween-80) in
citrate-buffered saline (pH 5.0). This mixture is used for shipping and
storing rBPI21 at 4șC, under which conditions it is stable for more
than two years.
Evaluating BPI in Animal Studies, Clinical Trials
Understanding and manipulating the pharmacodynamic
properties of BPI (or other therapeutic proteins) present considerable
challenges. To treat life-threatening GNB infections, BPI is
administered intravenously. However, it is rapidly cleared from the
bloodstream of mammals, primarily by the liver. Even so, single
injections, repeated injections, or sustained infusions of rBPI21 can
protect animals from infections, even when they are injected with live
bacteria. Some studies indicate that rBPI21's rapid clearance in human
patients may be due to interactions with heparin, which is a negatively
charged molecule, similar in some ways to LPS. If this is the case, it
seems unlikely that rBPI21 can be modified in a straightforward way to
reduce its clearance rate without affecting function and clinical
efficacy. Fortunately, however, rBPI21 acts in synergy with several
antibiotics, suggesting that it might retain activity even at reduced
concentrations in the bloodstream of patients.
During the 1990s, clinical researchers conducted Phase I
and II trials of rBPI21, under the trade name NEUPREX|Pr, confirming its
safety and also determining that it does not provoke an immune response
in humans, presumably because it is a natural component of the host
defense system. Following a promising Phase I trial in 1997, the U.S.
Food and Drug Administration granted orphan drug status to rBPI21 for
treating meningococcal sepsis. This deadly systemic infection is caused
by Neisseria meningitidis and often causes outbreaks among
children and young adults.
Meningococcemia is characterized by a massive release of
LPS into the bloodstream of infected patients, resulting in a high rate
of morbidity and mortality. Although relatively rare, the lack of other
effective therapies and the fact that most meningococcemia patients are
otherwise healthy made it an attractive system for testing BPI's
combined bactericidal and LPS-neutralization activities. With FDA
permission, clinical testing of rBPI21 for meningococcemia proceeded
directly to a Phase III trial.
The results of the Phase III clinical trial for
meningococcemia were reported in The Lancet (356:961,
2000). This clinical trial, which was headed by Michael Levin of the
Imperial College School of Medicine in London, England, and Brett Giroir
of Children's Medical Center in Dallas, Tex., and sponsored by XOMA,
enrolled 393 pediatric patients, making it the largest double-blind
study ever conducted for this disease. Despite evidence of clinical
benefit, BPI did not significantly reduce deaths among the treated
children compared to the control group. In this study, the overall
mortality rate was substantially lower than expected. Because of the
rapid progression of the disease, many patients died before reaching the
medical centers where the trials were being conducted.
Meanwhile, clinical researchers continue to investigate
BPI. In early 2000, Baxter Healthcare Corp., Deerfield, Ill., acquired
worldwide rights to rBPI21 for treating meningococcemia, similar
infections, and other conditions for which its profile might be suited.
In July 2001, Baxter initiated a Phase II clinical trial of rBPI21 for
Crohn's disease. This autoimmune condition is characterized by
inflammation of the bowel, and appears to be complicated by episodes of
LPS toxicity from intestinal bacteria. Crohn's disease is chronic, so
this clinical trial will not likely be complicated by a need for rapid
responses as occurred during the meningococcemia trial.
BPI's difficult journey from the lab and into clinical
trials illustrates some important issues in modern biopharmaceutical
drug development. Efforts to harness human proteins as biopharmaceutical
agents entail considerable energy and ingenuity to produce homogeneous,
functional, and stable products that meet regulatory standards. Other
challenges, including methods for delivering such proteins into patients
and dealing with their sometimes rapid clearance from the bloodstream,
only add to the complexities of these investigational efforts.
Additionally, the experience with BPI in clinical trials underscores
that trial design and clinical endpoints can be as important as having a
highly active and carefully formulated protein that works safely and
While FDA officials have not approved BPI for clinical
use, parts of its evaluative journey undoubtedly are being repeated by
other human-derived proteins that are considered promising
pharmaceuticals candidates. In the aggregate, these studies may
eventually lead to more general methods for expediting the
laboratory-to-clinic journeys of proteins that may well be useful for
treating diseases in humans.
Bauer, R. J., N. Wedel, N. Havrilla, M.
White, A. Cohen, and S. F. Carroll. 1999.
Pharmacokinetics of a recombinant modified amino terminal fragment of
bactericidal/permeability-increasing protein (rBPI21) in healthy
volunteers. J. Clin. Pharmacol. 39:1169-1176.
Beamer, L. J., S. F. Carroll, and D.
Eisenberg. 1997. Crystal structure of human
BPI and two bound phospholipids at 2.4 Angstrom resolution. Science 276:1861-1864.
Elsbach, P., and J. Weiss. 1995.
Prospects for use of recombinant BPI in the treatment of gram-negative
bacterial infections. Infect. Agents Dis. 4:102-109.
Elsbach, P., and J. Weiss.
1998. Role of the bactericidal/permeability-increasing protein in host
defense. Curr. Opin. Immunol. 10:45-49.
Hancock, R. E. W., and D. S. Chapple.
1999. Peptide antibiotics. Antimicrob. Agents Chemother. 43:1317-1323.
Levin, M., P. A. Quint, B. Goldstein, P.
Barton, J. S. Bradley, S. D. Shemie, et al. 2000.
Recombinant bactericidal/permeability-increasing protein (rBPI21) as
adjunctive treatment for children with severe meningococcal sepsis: a
randomized trial. Lancet 356:961-967.
Wang, W. 1999.
Instability, stabilization, and formulation of liquid protein
pharmaceuticals. Int. J. Pharm. 185:129-188.