Polymyxin B enhances ISS-mediated immune responses across ...
Published on: Mar 4, 2016
Transcripts - Polymyxin B enhances ISS-mediated immune responses across ...
Cellular Immunology 229 (2004) 93–105
Polymyxin B enhances ISS-mediated immune responses
across multiple species
Jason D. Marshall,¤ Debbie Higgins, Christi Abbate, Priscilla Yee, Glen Teshima, Gary Ott,
Tracy dela Cruz, David Passmore, Karen L. Fearon, Stephen Tuck, and Gary Van Nest
Dynavax Technologies Corporation, Berkeley, CA, USA
Received 15 December 2003; accepted 22 April 2004
Available online 24 June 2004
The immunostimulatory eVects of bacterial DNA on mammalian cells have been localized to unmethylated CpG motifs, and syn-
thetic CpG-containing oligodeoxynucleotides that mimic these eVects are known as immunostimulatory sequences (ISS). We have
found that the polycationic antibiotic, polymyxin B (PMXB), associates with ISS and serum albumin in vitro and forms microparti-
cles that greatly increase the activity of ISS on plasmacytoid dendritic cells (PDCs). SpeciWcally, ISS/PMXB greatly enhanced IFN-
production from PDCs and other activities downstream of IFN- , including IFN- secretion, NK lytic activity, and the expression of
genes dependent upon IFN- /IFN- . This ampliWcation was speciWc for the IFN- pathway since other ISS activities, including B
cell proliferation, B cell IL-6 secretion, and PDC maturation, were not aVected by PMXB. Both the polycationic peptide and lipo-
philic fatty acid side chain domains of PMXB, as well as the presence of a third party stabilizing agent such as albumin or Tween 85,
were required for particle formation and enhanced ISS activity. The ISS-enhancing activity of PMXB was observed across multiple
species (human, primate, and mouse) and in vivo (primate, mouse). These data illustrate the usefulness of formulating ISS with a cat-
ionic lipopeptide such as PMXB, which focuses and greatly ampliWes the ISS-induced pathway of IFN- -mediated responses.
2004 Elsevier Inc. All rights reserved.
Keywords: CpG DNA; Plasmacytoid dendritic cells; Adjuvants
1. Introduction 2,4-diaminobutyric acids (DABs), which theoretically
contribute a net charge of +5 to each molecule (Fig. 1).
The Wrst members of the family of antibiotics known The cyclization of the peptide is conducted through the
as polymyxins were isolated from strains of Bacillus poly- -amino and carboxyl groups of one DAB residue, which
myxa and Bacillus colistinus. These approximately also acts as the attachment point via the -amino group
1.2 kDa molecules share a structural consistency deWned for a tripeptide (in PMXB) that contains two more
by a decapeptide attached through an amide bond to a DABs. A fatty acyl derivative is attached to the -amino
fatty acid residue . The peptide portion is remarkably terminus of the tripeptide and can be either 6-methyloc-
conserved between members and consists of a heptapep- tanoic acid (PMXB1) or 6-methylheptanoic acid
tide ring with a further 1–3 residues branching oV line- (PMXB2). Other molecules with a similar basic structure
arly. In polymyxin B (PMXB)1 and colistin, the peptide as PMXB have been isolated from Bacillus species but
portion of the molecule is rich with positively charged have not been clinically developed; these include circulin,
Corresponding author. Present address: Dynavax Technologies Corporation, Berkeley, CA, USA. Fax: 1-510-848-5694.
E-mail address: firstname.lastname@example.org (J.D. Marshall).
Abbreviations used: BSA, bovine serum albumin; CMS, colistin methanesulfonate; cPLGA, cationic poly(D,L-lactide-co-glycolide); FAM, Xuo-
rescein amidite; HBsAg, hepatitis B surface antigen; HSA, human serum albumin; ISS, immunostimulatory sequence; MSA, mouse serum albumin;
ODN, oligodeoxynucleotide; PDC, plasmacytoid dendritic cell; PMXB, polymyxin B; PMXB-9, polymyxin-B nonapeptide; PO, phosphodiester; PS,
phosphorothioate; SN, supernatant; TLR, Toll-like receptor.
0008-8749/$ - see front matter 2004 Elsevier Inc. All rights reserved.
94 J.D. Marshall et al. / Cellular Immunology 229 (2004) 93–105
receptor family . TLRs have evolved to recognize
repeating, invariant microbial motifs such as lipopep-
tides, peptidoglycans, Xagellin, and double-stranded
RNA. Known human TLR9-expressing cells include
memory B cells and plasmacytoid dendritic cells (PDCs).
Activation of TLR9 on PDCs promotes maturation and
the secretion of cytokines such as IFN- , TNF- , IL-6,
and IL-8, while simultaneously sending anti-apoptotic
signals [5,6]. B cells respond to ISS by upregulating the
expression of several activation markers, including CD80,
CD86, and CD40, proliferating, and secreting IL-6 and
TNF- [7–9]. The increased production of IFN- by
PDCs can also lead to downstream eVects such as the
activation of monocytes and the induction of lytic activ-
ity and IFN- production from NK cells [10,11]. ISS thus
activates the innate immune system to respond with anti-
microbial defense measures while also contributing to
adaptive immunity by priming PDCs to promote Th1
development and cytotoxic T cells.
ISS are subdivided into three basic classes termed A,
B, and C [12–14]. CpG-A are characterized by a central
CpG-containing palindromic phosphodiester (PO) core
Xanked by phosphorothioate (PS) poly-guanosine
sequences that enable these ISS to aggregate in large clus-
Fig. 1. Chemical structure of polymyxin B. PMXB is composed of a
decapeptide, containing a heptapeptide ring linked to a linear tripep-
ters, presumably through a combination of duplex for-
tide which is attached to the fatty acid, 6-methyloctanoic acid, through mation and guanosine-tetrad stacking . This
an amide bond. 2,4-Diaminobutyric acid (DAB) residues contribute a aggregation is critical for CpG-A activity and promotes
net charge of +5 to each molecule. highly elevated IFN- secretion from PDCs and
increased NK cell activity. However, this ISS class exhib-
octapeptin, brevistin, cerexin, polypeptin, and stendomy- its poor activity at promoting PDC maturation or acti-
cin . vating puriWed B cells. On the other hand, the CpG-B
Polymyxins demonstrate antibiotic activity against a class does not form high-order structures and is generally
diverse array of organisms including bacteria, yeasts, and all PS with one or more CpG motifs and without strict
protozoa [1,2]. PMXB is especially active on gram-nega- Xanking sequence requirements. CpG-B exhibit activities
tive bacteria and exerts toxicity through bacterial mem- that are largely mutually exclusive with CpG-A; they
brane permeabilization and the rapid eZux into the potently activate B cells and promote PDC maturation
extracellular space of small charged or polar cytoplasmic but are poor at eliciting IFN- production from PDCs
molecules such as nucleosides, bases, and ions . Studies (reviewed in ). The CpG-C class are also PS ODNs
with PMXB derivatives have shown that both the cyclic which require the following sequence characteristics: at
peptide and acyl moiety are required for optimal mem- least one TCG motif at or close to the 5 end and a palin-
brane disruption. This disruption appears to be caused by dromic sequence that is optimally at least 12 bases long
interaction between PMXB and bacterial phospholipids and contains at least two and preferably more CpG dinu-
and lipopolysaccharide in the outer membrane . The cleotides that can be separated by 0–3 bases. The func-
polycationic peptide and fatty acid tail portions of tional capacity of CpG-C covers all known ISS
PMXB confer upon the molecule the amphipathic prop- responses, including those of CpG-A and CpG-B [14,16].
erty of solubilizing in water or lipid membranes. This, combined with their lack of structural complexity,
Although both domains of PMXB are required for anti- makes CpG-C ISS leading candidates for many clinical
bacterial activity, the peptide portion alone can bind bac- applications.
terial endotoxin , an activity routinely utilized as a Our laboratory has investigated various mechanisms
means for preventing LPS stimulation in mammalian cell for enhancing the eVects of CpG-B and CpG-C ISS and,
cultures . in particular, the induction of IFN- from PDCs. We
Much attention has been focused on the potential of previously determined that ODNs become tightly
unmethylated CpG motif-containing immunostimulatory adsorbed to the surface of cationic poly(D,L-lactide-co-
sequences (ISS) to enhance mammalian immunity. ISS glycolide) (cPLGA) microparticles through ionic
activates several immune functions through recognition interactions . Furthermore, cPLGA greatly enhanced
by the ISS-receptor TLR9, a member of the Toll-like the uptake of ODNs into PDCs and B cells and also
J.D. Marshall et al. / Cellular Immunology 229 (2004) 93–105 95
ampliWed the various activities that ISS has been shown generated with 1018, 1040, or C274 were sized by laser
to exert in vitro, particularly IFN- / induction. We light scattering (Malvern Instruments, Worcestershire,
sought to demonstrate whether another polycationic UK) and proportions of ODN and PMXB free and
molecule, PMXB, might also similarly associate with ISS bound were measured by HPLC (gel permeabilization
ODNs and whether this association would amplify ISS chromatography for ODN and reverse-phase chroma-
activity. We found that PMXB does indeed complex with tography for PMXB; Fig. 2). In the case of these pre-
ISS ODNs under deWned conditions to form microparti- made formulations, the volume of particles required to
cles and that this conWguration leads to greatly ampliWed get Wnal concentrations of 20 g/ml ISS and 100 g/ml
IFN- production by PDCs as well as immune events PMXB was added to the culture. To obtain an ISS/
downstream of IFN- , including increased IFN- secre- PMXB preparation depleted of uncomplexed PMXB,
tion, cytokine/chemokine gene expression, and NK lytic ISS/PMXB was dialyzed for 18 h in a Slide-A-Lyzer cas-
activity. Additionally, we show that ISS/PMXB enhances sette (Pierce) in buVer (0.4% Tween 85, 0.4% oleic acid in
the immune activity of various species and demonstrate 100 mM sodium phosphate, and 150 mM sodium chlo-
its eVectiveness in vivo in baboons and mice under cir- ride, pH 7.5). A 71% reduction in free PMXB was
cumstances where other ISS-formulations such as ISS/ obtained as assayed by reversed-phase HPLC. ODN and
cPLGA are poorly active. cPLGA (Boehringer Ingelheim Chemicals) were also pre-
mixed at Wnal concentrations of 20 and 100 g/ml, respec-
tively, for 15 min at r.t. before they were added to the cul-
2. Methods ture, as described previously .
2.1. Oligodeoxynucleotides 2.3. PBMC preparation
PS ODNs were synthesized as described . All Peripheral blood was collected from healthy volun-
ODNs had 05 endotoxin units/mg of ODN, determined teers by venipuncture using heparinized syringes. BuVy
by Limulus amebocyte lysate assay (Bio-Whittaker). coats were obtained from the American Red Cross.
Sequences: 1018 (CpG-B): 5 -TGACTGTGAACGT PBMCs were isolated by centrifugation through a Ficoll
TCGAGATGA; C274 (CpG-C): 5 -TCGTCGAACGT (Pharmacia) density gradient and cultured in RPMI 1640
TCGAGATGAT; 1040 (CpG-B control): 5 -TGACTG (Bio-Whittaker) supplemented with 10% heat-inactivated
TGAACCTTAGAGATGA; C661 (CpG-C control): 5 - human AB serum (Gemini) plus 50 U/ml penicillin, 50 g/
T GCTTGCAAGCTTGCAAGCA; C264 (CpG-A) 5 -G ml streptomycin, 300 g/ml glutamine, 1 mM sodium
GtgcatcgatgcagGGGGG. Uppercase denotes PS, lower- pyruvate (Bio-Whittaker), 50 mM Hepes, and 1£ non-
case denotes PO, and CpG motifs are underlined.
2.2. ISS/PMXB formulations and other reagents
PMXB (Sigma) was administered in a concentration
range of 1–500 g/ml. Optimal ISS enhancement was
observed at 100 g/ml PMXB and all data reported below
were generated using that dose. Colistin, colistin
methanesulfonate (CMS), polymyxin B nonapeptide
(PMXB-9) (all from Sigma), and rENP (recombinant
Endotoxin Neutralizing Protein; Cat. No. 910140, Sei-
kagaku America) were also all used at 100 g/ml. Gener-
ally, PMXB and ODNs were administered to cell cultures
separately in small volumes and therefore the PMXB
(100 g/ml) complexed with ODN (20 g/ml) within the
cell culture itself. In other cases, ISS, PMXB, and a third
stabilizing component [human serum albumin (HSA),
bovine serum albumin (BSA), mouse serum albumin Fig. 2. Physicochemical characterization of ISS/PMXB. Preparations
(MSA), or keyhole limpet hemocyanin (KLH)] were pre- of ISS/PMXB were generated by addition of 2 mg/ml ODN (1018,
mixed at a w/w ratio of 1:5:20 (ISS:PMXB:stabilizer; all 1040, and C274), 5 mg/ml PMXB, 4 mg/ml Tween 85, and 4 mg/ml
stabilizers from Sigma). Additionally, preparations of oleic acid. ISS/PMXB preps were sized by laser light scattering and
data reported as volume average distribution of diameters ( m). The
ISS/PMXB were generated using the non-ionic detergent
percentages of ODN and PMXB determined to be complexed or
Tween 85 (Sigma) as a stabilizer, in which 2 mg/ml ISS, uncomplexed were calculated after dissolution of pellets/supernatants
5 mg/ml PMXB, 4 mg/ml Tween 85, and 4 mg/ml oleic from ISS/PMXB formulations centrifuged for 5 min at 12000g in alka-
acid (Sigma) were pre-mixed. Preparations of ISS/PMXB line:isopropanol:glycerol.
96 J.D. Marshall et al. / Cellular Immunology 229 (2004) 93–105
essential amino acids (Bio-Whittaker). For cytokine . Primer sequences for ubiquitin, MIG, MCP-2, IP-10,
secretion, PBMCs were cultured at 0.5 £ 106/well (2 £ 106/ IFN- , and IFN- have been reported . Threshold
ml) in 96-well Xat-bottomed plates in duplicate with ISS cycle (CT) values for each gene were normalized to ubiq-
ODNs at a concentration range of 0.16–20 g/ml for 24 h, uitin. The negative control for each experiment, stimula-
determined by previous studies to be the optimal time- tion with medium alone, was assigned a value of 1 and all
point for ISS-induced cytokine secretion. Cell-free SNs data expressed as fold induction over the negative con-
were harvested and cytokine content was assayed by trol.
ELISA. Culture conditions (including cell concentration,
ODN concentration, and period of culture) to induce 2.8. ISS uptake assay
optimal activity in each ISS functional assay were pre-
determined. ISS uptake into PDCs was measured as described else-
where . In brief, FAM-labeled 1018 was incubated
2.4. Detection of cytokines by ELISA with monocyte-depleted PBMCs (monocytes were
removed because they act as a non-speciWc sink for
Cell-free medium was harvested and assayed for cyto- ODNs) for 2–4 h, then excess FAM-1018 was removed by
kine content via commercial kits. IFN- and IL-6 were washing and cells were stained with BCDA-4-PE, then
assayed via CytoSet antibody pairs (BioSource). Limits analyzed by FACScan.
of maximal/minimal detection were 4000/2 pg/ml for both
assays. IFN- was assayed via ELISA kit (PBL Biomedi- 2.9. In vivo hepatitis B surface antigen immunization and in
cal Laboratories) and limit of maximal/minimal detection vitro activity assays for mouse and baboons
was 13160/102 pg/ml. All kits were used according to
manufacturers' instructions. Mouse in vivo studies were performed at Northview
Biosciences (Hercules, CA) using 8-week-old female
2.5. Statistical analysis BALB/c mice (Charles River). Ten mice/group were
injected twice intramuscularly in the quadriceps at 2-
Statistical signiWcance was calculated using the Mann– week intervals with HBsAg alone (1 g), HBsAg + 1018
Whitney test with unpaired and non-parametric values (5 g), HBsAg + 1018/cPLGA (5 g/250 g DOTAP/
(GraphPad InStat). Symbols representing signiWcance PLGA), or HBsAg + 1018/PMXB or 1040/PMXB (5 g/
are: ***p 0 0.001; **p 0 0.01; *p 0 0.05; and ns, p 1 0.05. 25 g) stabilized with Tween 85. Blood was collected
retro-orbitally 2 weeks after each injection and serum
2.6. Cell puriWcation and ISS functional assays was prepared for antibody determination. Samples were
analyzed for anti-HBsAg IgG1 and IgG2a content by
For PDC enrichment, PBMCs were depleted of CD3-, ELISA. Earlier studies on serum from naïve mice veriWed
CD19-, and CD11b-expressing cells by MACS labeling that their anti-HBsAg levels are undetectable (0120).
and two passages over an LS column (Miltenyi Biotec). Plates were coated with 1 g/ml HBsAg, and goat anti-
The resultant populations were 1–2% BCDA-2+ CD123+, mouse IgG1 or IgG2a biotin-conjugated antibody
indicating that the PDC population had been enriched (Southern Biotechnology) was used for detection. The
approximately 5- to 10-fold. B cells were puriWed from titer is deWned as the reciprocal of the serum dilution that
PBMCs through CD19-positive selection by MACS and gave an ELISA absorbance of 0.5 OD using four-param-
used in proliferation assays as described . NK lytic eter analysis (Softmax Pro97, Molecular Devices).
activity was assayed through lysis of K562 target cells. In Splenocytes from immunized mice were harvested at
brief, PBMCs were stimulated with 10 g/ml week 6 and used for cytokine secretion and CTL assays.
ISS § 100 g/ml PMXB for 48 h then were co-cultured To measure CTL activity, splenocytes were stimulated
with 51Cr-loaded K562 tumor target cells at a range of with 1 g/ml HBsAg MHC class I H-2d-restricted peptide
eVector:target ratios for 4 h. 51Cr release was measured by (IPQSLDSWWTSL synthesized by Research Genetics)
a TopCount NXT scintillation counter (Packard), and % for 6 days, then co-cultured with 51Cr-labeled, peptide-
speciWc lysis was calculated as [(experimental loaded P815 target cells at a range of eVector:target ratios
release ¡ spontaneous release) /(maximum for 4 h. 51Cr release was measured by scintillation
release ¡ spontaneous release) £ 100]. All assays were counter. To characterize cytokine production, additional
previously optimized for ISS dose. splenocytes were stimulated with HBsAg (5 g/ml) for 4
days, then supernatants were harvested and analyzed for
2.7. Gene expression assay and analysis IFN- and IL-5 content by ELISA. Baboon studies were
performed at Southwest Foundation for Biomedical
Human PBMCs were cultured with 5 g/ml Research (San Antonio, TX). Five animals per group
ISS § 100 g/ml PMXB at 2 £ 106/ml for 4 h. Total RNA (male and female, 10–18 kg/animal) were immunized i.m.
was extracted and RT-PCR was performed as described with HBsAg (20 g), HBsAg + 1018 (1 mg), or
J.D. Marshall et al. / Cellular Immunology 229 (2004) 93–105 97
HBsAg + 1018/PMXB (0.86 mg/3 mg) in volumes of 1 ml. PBMCs, dependent upon the class of ISS . CpG-B
Peripheral blood serum was collected at day 28 and (e.g., 1018) induce low levels of IFN- and IFN- while
analyzed for anti-HBsAg conformational epitope via CpG-C (e.g., C274) induce slightly higher levels of IFN-
AUSAB EIA (Abbott Laboratories), according to manu- and very high levels of IFN- . In the presence of 100 g/
facturer's protocols. ml PMXB, however, induction of IFN- by both classes
of ISS is potently enhanced approximately 50- to 100-
fold (Fig. 3A). IFN- induction by 1018 is also markedly
3. Results increased by PMXB at higher doses. The activity of the
strong IFN- -inducer C274 was modestly increased
3.1. PMXB enhances ISS activity on PBMCs (Fig. 3B). These eVects are CpG-speciWc as little or no
activity was observed from cells stimulated with the con-
Previous studies have indicated that ISS ODNs can trol ODNs 1040/PMXB or C661/PMXB. CpG-A is an
induce variable amounts of IFN- and IFN- from ISS class that also provides an extremely potent stimula-
Fig. 3. PMXB greatly enhances IFN- /IFN- -inductive potential of ISS in CpG-speciWc manner. PBMCs were stimulated with 20, 4, 0.8, and 0.16 g/
ml ODN (open circles) or ODNs + 100 g/ml PMXB (closed circles) for 24 h. The ODNs chosen are representatives of the CpG-B class (1018), CpG-
C class (C274), and CpG-A class (C264). Also shown are the negative controls for class B (1040) and class C (C661). Supernatants were assayed for
(A) IFN- and (B) IFN- levels via ELISA. Data are expressed as means of four donors § SEM and are representative of three experiments. Statisti-
cal signiWcance was computed vs ODN alone.
98 J.D. Marshall et al. / Cellular Immunology 229 (2004) 93–105
tion for IFN- production . PMXB was observed to determine whether the presence of PMXB enhances ISS
greatly enhance CpG-A-induced IFN- but not IFN- . activity diVerently according to cell type. PBMCs
Since the IFN- -inducing capacity of CpG-A is depen- enriched for PDCs by lineage depletion were stimulated
dent upon its super-aggregated structure , it is possi- with ISS § PMXB for 24 h then analyzed for cytokine
ble that PMXB cannot associate in the same way with secretion. As expected, Fig. 4A shows that stimulation
this class of ISS as it does with monomeric CpG-B/C. with either CpG-B 1018 or CpG-C C274 alone resulted in
To determine whether the ISS-enhancing ability of 4- or 8-fold higher levels of IFN- secretion, respectively,
PMXB was dependent upon its LPS-inhibiting function, compared to non-depleted PBMCs (as presented in
we analyzed activity of ISS in the presence of PMXB or Fig. 3B). This IFN- production was conWrmed to be
recombinant endotoxin neutralizing protein (rENP), a PDC-speciWc by intracellular co-staining for IFN- and
protein originally isolated from Limulus that neutralizes BCDA-2 and by speciWc PDC-depletion, which abro-
endotoxin from gram-negative bacteria. Although rENP gated IFN- secretion (data not shown). The PDC-spe-
neutralized the LPS-mediated induction of IL-12 simi- ciWc IFN- was elevated to even higher levels in the
larly to PMXB, it had no ability to alter ISS activity (data presence of PMXB, especially in the case of the 1018.
not shown), suggesting that PMXB is not enhancing the However, the presence of PMXB had no eVect on the
eVects of ISS through the inactivation of endotoxin. ability of ISS to enhance DC maturation by upregulation
of CD86 (Fig. 4B) and CD80 (data not shown) on PDCs.
3.2. PMXB/ISS induces a similar gene pattern to ISS Therefore, it appears that the enhancing eVect of PMXB
is targeted speciWcally towards PDC secretion of IFN-
ISS stimulation induces a well-deWned and speciWc without altering other eVects of ISS on the PDC. NK
pattern of gene activation in human PBMCs , includ- cells are indirectly activated by ISS through signals trans-
ing cytokines, chemokines, and antiviral factors. We mitted from the PDC. The exact nature of this signal has
determined whether that pattern was altered during not been yet identiWed, although IFN- appears to play a
PMXB enhancement by using TaqMan RT-PCR tech- major role . We sought to determine whether the lytic
niques to measure the mRNA expression of a panel of activity of NK cells was also enhanced in the presence of
genes in PBMCs. As reported in Table 1, PMXB clearly PMXB. Fig. 4C indicates that NK cells exhibit increased
boosted the expression of various genes in ISS-speciWc lytic activity vs K562 target cells in the presence of 1018
fashion including IFN- and IFN- , and chemokines and that this activity is further ampliWed by PMXB.
such as interferon-inducible protein-10 (IP-10 or When highly puriWed B cells were subjected to stimu-
CXCL10), monokine induced by -interferon (MIG or lation with ISS/PMXB, no improvement in B cell func-
CXCL9), and monocyte chemoattractant protein-2 tions compared to ISS alone was observed. Neither ISS-
(MCP-2 or CCL8). Other genes not induced by ISS (e.g., enhanced proliferation (Fig. 4D) nor the induction of IL-
MCP-1, MIP-3 , and IL-1 ) were also not enhanced in 6 secretion (data not shown) from B cells were found to
the presence of PMXB (data not shown). Therefore, be increased by the presence of PMXB; in fact, the pres-
PMXB increases the ISS-transduced signal for gene acti- ence of PMXB led to marked decreases in B cell prolifer-
vation quantitatively but does not alter it qualitatively. ation (constitutive and ISS-speciWc) but not IL-6
production. Reduced B cell proliferation appeared to be
3.3. PMXB enhances ISS activity in PDCs, not B cells due to prolonged exposure of the B cells to non-com-
plexed PMXB, since an ISS/PMXB preparation from
The known ISS-responsive cells in human PBMCs are which unbound PMXB had been removed via dialysis
PDCs and B cells. We examined the eVect of ISS/PMXB showed no reduction in B cell proliferation (Fig. 4E).
formulations on enriched or puriWed cell populations to These studies suggest that ISS stimulation activates at
PMXB enhances the induction of gene expression by ISS
Stimulus IFN- IFN- IP-10 MIG MCP-2
Mean SEM Mean SEM Mean SEM Mean SEM Mean SEM
1018 66.6 52.0 1.5 0.2 3.3 2.5 2.0 1.6 1.2 1.0
1018/PMXB 1484 962 46.3 24.5 14.7 7.2 17.0 12.2 4.7 2.4
C274 7271 3870 7.6 3.2 61.4 31.0 19.0 16.2 13.0 8.2
C274/PMXB 19,705 8159 196.0 73.0 110.6 42.9 73.8 48.4 73.9 17.4
C661 1.9 0.8 1.1 0.2 0.8 0.2 0.5 0.1 0.1 0.0
C661/PMXB 2.0 0.9 9.3 1.8 5.4 2.4 3.1 1.4 0.6 0.4
PBMCs were stimulated with 5 g/ml 1018, C274, or C661 § PMXB for 4 h and then RNA was extracted and analyzed via TaqMan RT-PCR.
Gene expression was normalized to ubiquitin expression. Data are presented as the mean of fold induction over medium control (given the value of
1.0) with SEM for four donors, and are representative of three separate experiments.
J.D. Marshall et al. / Cellular Immunology 229 (2004) 93–105 99
Fig. 4. PMXB enhances the ISS-mediated induction of IFN- from PDC-enriched cells and activation of lytic activity from NK cells. (A) PBMCs
were enriched for PDCs through lineage depletion and cultured with 20 g/ml ISS § PMXB for 24 h, then supernatants were assayed for IFN- via
ELISA. Data are expressed as mean of eight donors + SEM. (B) MACS-puriWed PDCs were stimulated with 5 g/ml ISS § PMXB for 18 h, then ana-
lyzed for CD86 expression by FACScan. The plot with Wlled-in area represents unstimulated PDCs. (C) PBMCs from two donors were cultured with
10 g/ml 1018 § 100 g/ml PMXB for 48 h. These cells were then harvested and co-cultured with 51Cr-loaded K562 target cells at 50:1–6:1 E:T ratios
for 4 h. Released 51Cr was then measured in supernatants via a scintillation counter; data are reported as (experimental value ¡ spontaneous release)/
total release. This is one representative experiment of three. (D–E) MACS-puriWed B cells were cultured with 5 g/ml ODNs § PMXB for 96 h. Pro-
liferation was assessed by [3H]thymidine incorporation. Proliferation data are expressed as means of four donors + SEM.
least two separate signaling pathways: one that results in required to confer activity upon the ISS + PMXB formu-
elevated production of type I interferon by the PDC and, lation. Indeed, when ISS, PMXB, and HSA were com-
in turn, heightened downstream IFN-dependent events bined at a 1:5:20 ratio in PBS, the result was a stable
such as NK activation and chemokine expression, which solution of suspended particles (size range 100 nm–10 m,
can be ampliWed by PMXB; and another that results in see Fig. 2) that exhibited ISS-enhancing activity on
many other known ISS functions including dendritic cell PBMCs. On the other hand, withholding the addition of
maturation and B cell activation, which is unaltered by HSA resulted in immediate precipitation of the
PMXB. ISS + PMXB, which became highly adsorptive to poly-
propylene and exhibited poor ISS-enhancing activity. To
3.4. ISS/PMXB form particles that depend on the presence conWrm the requirement of albumin for ISS-enhancing
of a stabilizing agent activity, ISS + PMXB preparations were generated using
albumins of other species, BSA and MSA. An
We initially observed that ISS formulated with PMXB ISS + PMXB prep was also generated with a non-albu-
did not exhibit enhanced activity on cells cultured in min protein, KLH, which has an isoelectric point of 5.3,
serum-free media and instead immediately formed an similar to that of HSA (5.8). ISS + PMXB formulations
aggregated precipitate, which was not observed in cul- that utilized HSA, BSA, MSA, or KLH demonstrated
tures containing serum. Therefore, we hypothesized that both the capacity to form a stable suspension of particles
a third component, found to be serum albumin, was and comparable ISS-enhancing activity (data not shown).
100 J.D. Marshall et al. / Cellular Immunology 229 (2004) 93–105
We suspected that these proteins played the role of stabi- intact. Conversely, the PMXB-9 molecule lacks the
lizing agent in the ISS + PMXB formulations and could hydrophobic tail but retains the positively charged cyclic
be substituted with another immunologically inert com- peptide domain. Therefore, we used these two derivatives
ponent. Tween 85 is a detergent that minimizes protein as a means of analyzing the importance of each domain
aggregation and is commonly used as an excipient for for the observed enhancement of ISS activity. Both CMS
pharmaceuticals. We combined ISS, PMXB, and Tween (which did not form particles with ISS) and PMXB-9
85/oleic acid and observed that, like HSA, the presence of (which did) exhibited substantially reduced ISS-enhanc-
Tween 85 allowed for the formation of particles of a sim- ing activity in comparison to PMXB or colistin (Fig. 5).
ilar size range. In comparison to ISS/PMXB/HSA preps, These data indicated that both the polycationic and lipo-
ISS/PMXB/Tween 85 preps displayed a similar ability to philic domains of the polymyxin molecule are required
enhance IFN- /IFN- production from human PBMCs for optimal ISS-enhancing activity.
(data not shown). We further examined whether activity
could be localized to small or large particles through sep- 3.6. PMXB does not enhance uptake of ISS into PDCs
aration of ISS/PMXB/HSA preparations by either centri-
fugation at 2000 rpm for 5 min or by passage through a To investigate whether the enhancement of ISS-
2.7 m Wlter. Both techniques result in a rough division induced IFN- secretion by PMXB was attributable to
between 01 m particles and 11 m particles. We found an enhancement in ISS uptake by PDCs, we examined
that both populations exerted comparable IFN- /IFN- - uptake of Xuoresceinated-ISS into PDCs in the presence
inducing activity on human PBMCs (data not shown), of cPLGA or PMXB (Fig. 6). As has been previously
indicating that ISS-enhancing properties are not
restricted to one size of particle.
3.5. Both polycationic and hydrophobic domains of PMXB
are required for optimal activity
Members of the polymyxin family (including PMXB
and colistin) have two major domains: (1) a cyclic peptide
that contains Wve positively charged amino acid residues
and (2) an N-terminal lipophilic fatty acyl derivative. In
order to determine whether one or both domains were
required in order to enhance ISS activity, we examined
the ISS-enhancing potential of two derivatives of the Fig. 6. PMXB does not enhance uptake of ISS into PDCs. Monocyte-
polymyxin family, colistin methanesulfonate (CMS) and depleted PBMCs were cultured with 2 g/ml FAM-conjugated
polymyxin B nonapeptide (PMXB-9). CMS is produced 1018 § 100 g/ml PMXB or § 100 g/ml cPLGA for 4 h, then washed
by a sulfomethylation reaction in which the primary thoroughly to remove surface-bound 1018. Cells were stained with
BCDA-2-PE then analyzed via FACScan. Dead cells were gated out
amine groups of the 2,4-diaminobutyric acids are reacted
via FSC vs SSC and double-positive cells (FL1-intracellular 1018,
with formaldehyde and then sodium bisulWte . This FL2-BCDA-2) acquired. This histogram is gated on the BCDA-2+
results in a molecule in which the cationic amino acids population. Data shown are from one representative donor of four.
are now negatively charged but the hydrophobic tail is Lines for 1018 and 1018/PMXB are virtually superimposed.
Fig. 5. Both domains of the PMXB molecule are required for optimal ISS-enhancing activity. PBMCs were stimulated for 24 h with 20 g/ml 1018 or
C661 and 100 g/ml PMXB, colistin, colistin methanesulfonate (CMS), or polymyxin B nonapeptide (PMXB-9). Supernatants were assayed for IFN-
/IFN- levels via ELISA. Data are expressed as means of six donors + SEM. Numbers are for statistical relevance: IFN- : 1 vs 2: *; 1 vs 3: **; 2 vs
5: *; 3 vs 6: **; 1 vs 4: ns; 1 vs 5: ns. IFN- : 1 vs 2: ***; 1 vs 3: *; 2 vs 5: ***; 3 vs 6: *; 1 vs 4: ns; 1 vs 5: ns.
J.D. Marshall et al. / Cellular Immunology 229 (2004) 93–105 101
shown , cPLGA greatly enhanced ISS uptake into in vivo models of immunogenicity with hepatitis B sur-
PDCs and this mechanism may account, in part, for its face antigen (HBsAg). Mice were immunized on days 0
ability to enhance ISS activity by simply allowing more and 14 with HBsAg, HBsAg + 1018, or HBsAg + a prep-
ISS molecules to enter the PDC and contact intracellu- aration of 1018 or 1040 complexed with PMXB and
larly expressed TLR9. In contrast, the presence of PMXB Tween 85/oleate. HBsAg alone induced modest titers of
did not enhance ISS uptake into PDCs and may have HBsAg-speciWc IgG1 levels 2 weeks post-second immu-
slightly inhibited it. This suggests that PMXB may be nization, while the presence of 1018 elevated IgG1 and
enhancing IFN- levels through a mechanism other than especially IgG2a titers (Fig. 7B). Mice immunized in the
increasing the total amount of ISS entering the PDC. Not presence of 1018/PMXB, however, demonstrated a
surprisingly, B cells also demonstrated no enhanced much greater ampliWcation of IgG2a levels than were
uptake of ISS in the presence of PMXB (data not shown), observed with 1018 alone, and this was even visible at 2
similar to PDCs. weeks post-Wrst immunization (Fig. 7A), thus verifying
that PMXB can greatly enhance and accelerate 1018
3.7. PMXB boosts ISS-induced hepatitis B surface activity in vivo as well as in vitro. Although 1040/
antigen-speciWc IgG in vivo PMXB control immunization led to an increase in IgG1
levels (presumably due to CpG-unrelated eVects on B
To conWrm these in vitro Wndings, PMXB was exam- cells as also observed in Fig. 4D), it led to no increase in
ined for ISS-enhancing activity in murine and baboon IgG2a levels, an indication that the IgG2a eVect was
Fig. 7. PMXB enhances the ISS eVect on hepatitis B surface antigen-speciWc antibody production in immunized mice and baboons. (A–B) BALB/c
mice were immunized in four groups of 10 animals each with HBsAg § 1018, 1040/PMXB, or 1018/PMXB. Immunizations (i.m.) took place on days
0 and 14. Blood was collected for serum analysis of HBsAg-speciWc Ab titers at days 14 (A) and 28 (B). Naïve animals exhibited undetectable levels
(0120) of anti-HBsAg Ab. Data are reported as means of 10 animals + SEM. Statistical signiWcance was computed vs HBsAg immunization. (C)
Baboons were immunized with HBsAg § 1018 § PMXB and blood serum samples were analyzed for anti-HBsAg 4 weeks later. Data are reported as
means of 5–10 animals. Numbers indicate animals sero-positive for anti-HBsAg (110 mIU/ml).
102 J.D. Marshall et al. / Cellular Immunology 229 (2004) 93–105
both CpG-speciWc and PMXB-dependent. Baboons 3.8. ISS/PMXB exhibits diVerential eVects on human and
immunized with HBsAg + 1018/PMXB also showed mouse immune activities in comparison to ISS/cPLGA
enhanced serum titers of anti-HBsAg IgG in compari-
son with HBsAg + 1018 injection (Fig. 7C), further vali- We have previously described the elevating eVects of
dating the in vivo properties of ISS/PMXB as a vaccine ISS associated with cPLGA on IFN- induction from
adjuvant. human PBMCs . To compare the similar ISS-enhanc-
Fig. 8. ISS/PMXB induces immune activity equivalent to or greater than ISS/cPLGA formulations. (A) Human PBMCs were stimulated for 24 h
with 20 g/ml ODNs § 100 g/ml PMXB or cPLGA. Supernatants were assayed for IFN- /IFN- levels via ELISA. Data are expressed as means of
8–12 donors + SEM. Statistical signiWcance for 1018 was computed vs medium and for all other conditions vs 1018. (B) Human B cells were cultured
with ODNs § PMXB or cPLGA for 96 h. The ISS/PMXB prep used in this assay had been previously dialyzed to remove free PMXB molecules. Pro-
liferation was assessed by [3H]thymidine incorporation. Proliferation data are reported as percentage of 1018-induced proliferation and are expressed
as means of four donors. Statistical signiWcance was computed vs 1018 for all conditions. (C) BALB/c mice were immunized in four groups of 10 ani-
mals each with HBsAg § 1018, 1018/cPLGA, or 1018/PMXB. Immunizations (i.m.) took place on days 0 and 14. Blood was collected for serum anal-
ysis of HBsAg-speciWc Ab titers at day 14. Data are reported as means of 10 animals + SEM. Statistical signiWcance was computed vs HBsAg
immunization. (D) Splenocytes were collected from these mice on day 42 and stimulated with HBsAg for 4 days after which IFN- /IL-5 secretion
was measured by ELISA. Statistical signiWcance was computed vs HBsAg immunization. (E) Day 42 splenocytes were also stimulated with HBsAg
peptide for 6 days, then co-cultured with 51Cr-P815 target cells for 4 h to measure cytolytic activity. Data are reported as mean of Wve mice.
J.D. Marshall et al. / Cellular Immunology 229 (2004) 93–105 103
ing properties of ISS/PMXB, we utilized the two formu- response to ISS was also not directly aVected by PMXB,
lations in a variety of in vitro and in vivo models of ISS although B cell responses might be aVected indirectly in
activity. Although formulation with either PMXB or vivo through PDC-derived signals. These results suggest
cPLGA greatly boosted the IFN- /IFN- -inducing prop- that ISS/PMXB targets the PDC-derived IFN- arm of
erties of 1018 upon human PBMCs (Fig. 8A), the formu- the immune response for augmentation. The mechanism
lations diverged in regards to B cell activation. While by which PMXB achieves these eVects does not appear to
1018/cPLGA enhanced human B cell proliferation over be due to enhanced uptake of ISS ODNs by the PDC.
1018 stimulation alone, 1018/PMXB does not signiW- Studies were performed to elucidate the structure of
cantly alter this activity (Fig. 8B). However, 1018/PMXB associated ISS/PMXB. When ISS and PMXB are added
proved much more eVective at increasing HBsAg-speciWc together in saline, precipitation immediately occurs,
IgG2a titers in HBsAg-immunized mice than did 1018/ yielding a highly aggregated and insoluble substance that
cPLGA and furthermore skewed the IgG1:IgG2a ratio lacks ISS activity when applied to cultured cells (data not
towards a much more pronounced IgG2a-dominant Th1- shown). Since this precipitation is not observed when ISS
like proWle (Fig. 8C), a result that was continued follow- and PMXB are added directly to cultured cells, we sur-
ing the second immunization as well (data not shown). mised that the presence of serum in the culture may be
This Th1 skew was also reXected in cytokine production required for allowing the appropriate conWguration to
by splenocytes from HBsAg-immunized mice (Fig. 8D). occur and express activity. Indeed, when HSA is present
Administration of 1018/PMXB maintained the IFN- - during the mixing of ISS and PMXB in saline (1:5:20
dominant proWle established by 1018 alone, while cells molar ratio ISS:PMXB:HSA), we observed the forma-
from mice treated with 1018/cPLGA reverted back to tion of discrete particles conforming to a bimodal size
higher levels of IL-5 and lower levels of IFN- . Neither distribution and ranging in size from 100 nm to 10 m
was 1018/cPLGA administration able to support spleno- (Fig. 2). These particles also showed enhanced ISS activ-
cyte CTL activity from HBsAg-immunized mice, unlike ity for IFN- / production when added to PBMCs, simi-
1018/PMXB, which permitted the strong CTL activity larly to adding ISS and PMXB separately to the culture.
induced by 1018 (Fig. 8E). Therefore, ISS/PMXB is PS ODNs are known to bind serum albumins with micro-
clearly superior to ISS/cPLGA in the mouse in vivo molar Kd , and thus HSA may assist the formation of
model since it maintains or boosts the eVects of ISS, while ionic bonds between negatively charged ISS and posi-
such activity is generally abrogated by association with tively charged PMXB. In addition, HSA may act as a sta-
cPLGA. We did not observe enhancement of IFN- bilizing agent that counteracts the tendency for large ISS/
secretion or CTL activity by PMXB in comparison to PMXB aggregates to form. The role of the stabilizing
1018 alone in this system (Figs. 8D and E), and this may agent may be played by several molecules, including
indicate that PMXB is less able to boost T cell-mediated, other albumins, KLH, and even the surfactant Tween 85,
adaptive immunity compared to the potent enhancement an excipient commonly used in many pharmaceutical for-
of innate immune responses reported in Figs. 3–6 and mulations to minimize protein–protein interactions.
Table 1. The chemical properties of PMXB that are required
for optimal enhancing ability are its polycationic domain
and its lipophilic fatty acid side chain. The ISS ODNs
4. Discussion used herein have a net negative charge of 19–21 and
should tightly associate with +5 positively charged
This study demonstrates that enhanced ISS potency PMXB molecules that are Wve times in molar excess. This
can be achieved by formulating ISS ODNs with PMXB tight association leads to the formation of microparticles.
and a stabilizing agent. ISS from the CpG-B and CpG-C Since the hydrophobic tail of PMXB is known to per-
classes exhibited the most dramatic PMXB-mediated meabilize bacterial cell membranes , it is possible that
increase in IFN- / activity, while less enhancement was the ISS/PMXB complex enters the PDC through this
observed with CpG-A, perhaps due to the diYculty in means. On the other hand, PMXB-9, which lacks a lipo-
formulating particles with a class of ISS known to be philic sidechain, is also known to have some cell mem-
already highly aggregated . We observed that those brane-permeabilizing activity , although it is unclear
immune responses thought to be downstream of and whether it could accomplish this if its positively charged
dependent on IFN- production were the same ones also region were bound (and neutralized) by ISS. Since
enhanced by PMXB. These included enhancement of enhanced uptake of ISS does not occur when PMXB-for-
IFN- -inducible chemokine expression and NK cell acti- mulated (Fig. 6), the partial activity enhancement
vation. Interestingly, we found that other measures of observed with PMXB-9 and the fact that PMXB-9 forms
PDC activation that are not dependent on IFN- , i.e., particles with ISS (data not shown) point to clustered ISS
increased expression of the activation markers CD80/ polyvalent presentation as the mechanism for elevated
CD86, were not enhanced by PMXB. Furthermore, these IFN- induction, rather than increased entry of ISS into
studies indicated that the B cell arm of the immune the cell interior through membrane permeabilization.
104 J.D. Marshall et al. / Cellular Immunology 229 (2004) 93–105
CMS is gradually hydrolyzed and converted to colistin in TNF- /IL-6 production, and B cell proliferation [14,17].
an aqueous solution of human plasma over several days This divergence of elevated IFN- induction from other
 and this may explain its partial enhancement of ISS ISS functions implies separate pathways initiated by the
activity in our system (Fig. 5). ISS receptor and that perhaps ISS presented in arrays or
Therefore, we propose that PMXB and related mole- aggregate form crosslinks TLR9 and ampliWes only the
cules complex with ISS via interaction between the poly- pathway that leads to high IFN- transcription.
cationic domain of PMXB and the negatively charged PMXB, colistin, and related molecules have been
DNA oligonucleotides. Tight association between multi- administered as antibiotics to humans for over 40 years,
ple molecules of both components occurs in such a way especially to cystic Wbrosis patients . We have shown
that large aggregate microparticles are formed with that in addition to its bactericidal properties, PMXB also
super-aggregation and subsequent precipitation coun- demonstrates a unique ability to enhance the activity of
tered by the presence of stabilizers such as HSA or Tween ISS ODNs. This enhancement is speciWcally directed to
85. Such microparticles enter the interior of the PDC, IFN- production from PDCs without aVecting other
possibly by permeabilization of the cell membrane using ISS functions, such as B cell activation. Although the
remaining free polycationic charge or the lipophilic fatty activity of ISS/PMXB formulations on human cells in
acid side chain. Alternatively, we cannot rule out that the vitro is well duplicated by the previously described ISS/
complex may be endocytosed by an unidentiWed receptor cPLGA microparticles , ISS/PMXB complexes are
on the PDC cell surface. Once inside, the ISS/PMXB plainly much more eVective at carrying out ISS activities
complex presents multiple ISS motifs in multimeric fash- in vivo in mice (Fig. 8), while ISS/cPLGA particles are
ion, a conWguration observed to consistently induce very ineVective, perhaps because they are sequestered at the
elevated levels of IFN- expression from PDCs. site of injection. Although the A-class of ISS is also
Although ISS complexed to PMXB is also able to enter B known for greatly amplifying PDC-derived IFN- , CpG-
cells, we see no increase in ISS-induced activity, indicat- A are not recognized by B cells, while ISS/PMXB can
ing that B cell-speciWc ISS pathways are not inXuenced by directly induce both PDC and B cell activity. Further-
multimeric presentation of ISS motifs. more, CpG-A activity is dependent upon the formation
We conducted studies to clarify the mechanism by of heterogeneous populations of aggregates comprised of
which PMXB increases the potency of ISS. PMXB does a variable number of CpG-A molecules , thereby
not appear to be enhancing the eVect of ISS through making GMP-level characterization diYcult. On the
increased ODN uptake into the PDC population; other hand, ISS/PMXB microparticles can be easily sepa-
instead, the signal for PDC-internalized ISS after 4 h rated by size into much more homogeneous populations.
incubation in the presence of PMXB appears to be equiv- Renal and neurological toxicities have been reported
alent to or even somewhat less than that in the absence of stemming from use of polymyxins in humans [2,22]; how-
PMXB. Nor does PMXB appear to be shuttling ISS ever, the dose required for antibacterial eYcacy is 2.5–
ODNs to diVerential compartments within the PDC (J. 5.0 mg/kg/day for several consecutive days . We esti-
Marshall, unpublished observations). This suggests an mate a delivery of PMXB in ISS formulations of approx-
alternative possibility to explain PMXB's mechanism of imately 0.2 mg/kg, 1–2 times per week (or less often for
action: that the conWguration in which ISS is presented vaccines), which should signiWcantly decrease its toxicity
may be crucial for the enhanced eVect. Although the proWle. Furthermore, we have proved that ISS/PMXB
other ISS-enhancer we studied, cPLGA, does greatly formulations are fully functional when stabilized with
enhance ISS uptake into PDCs , the increased IFN- Tween 85, an approved pharmaceutical excipient. The
levels induced by this formulation may not depend on IFN- -dominant proWle of activity induced by ISS/
that process. cPLGA also presents numerous ISS ODNs PMXB formulations might be particularly useful for
in a multivalent display on the surface of the nanoparti- treating disorders such as asthma or B cell lymphomas, in
cle, which may share conWgurational similarities with which the Th1-inducing and antitumor eVects of IFN-
ISS/PMXB/HSA particles, and possibly it is this aggre- are desirable without the concurrent elevated polyclonal
gated form that is critical to achieve enhanced IFN- activation of disease-mediating B cells. In addition, our
production in the case of both types of formulation. Fur- preliminary studies with in vivo HBsAg immunization in
ther evidence supporting this hypothesis is provided by mice and baboons also demonstrate the potent activity of
observations concerning the activity of CpG-A  and ISS/PMXB as a vaccine adjuvant.
short heptameric ISS linked to Ficoll . Both of these
formulations also present a large number of ISS motifs in
multimeric form, and both also induce high levels of Acknowledgments
IFN- from PDCs in a CpG-speciWc manner. Another
commonality to these various forms of clustered ISS is We are grateful to F. Barrat, E. Hessel, and H. Kanz-
that they are not nearly as enhancing, if at all, for other ler for helpful discussion; to Cindy Fressola and Sherry
CpG-dependent functions such as PDC maturation, Kelly (Advanced Bioscience Resources) for phlebotomy
J.D. Marshall et al. / Cellular Immunology 229 (2004) 93–105 105
services; to Holli Aaron (CRL Molecular Imaging Cen-  D. Verthelyi, K.J. Ishii, M. Gursel, F. Takeshita, D.M. Klinman,
ter, University of California at Berkeley) for Xuorescent Human peripheral blood cells diVerentially recognize and respond
to two distinct CpG motifs, J. Immunol. 166 (2001) 2372–2377.
microscopy services; and to N. Khounchanh for techni-  A. Krug, S. Rothenfusser, V. Hornung, B. Jahrsdorfer, S. Black-
cal assistance. well, Z.K. Ballas, S. Endres, A.M. Krieg, G. Hartmann, IdentiWca-
tion of CpG oligonucleotide sequences with high induction of
IFN-alpha/beta in plasmacytoid dendritic cells, Eur. J. Immunol.
31 (2001) 2154–2163.
References  J.D. Marshall, K. Fearon, C. Abbate, S. Subramanian, P. Yee, J.
Gregorio, R.L. CoVman, G. Van Nest, IdentiWcation of a novel CpG
 D.R. Storm, K.S. Rosenthal, P.E. Swanson, Polymyxin and related DNA class and motif that optimally stimulate B cell and plasmacy-
peptide antibiotics, Annu. Rev. Biochem. 46 (1977) 723–763. toid dendritic cell functions, J. Leukoc. Biol. 73 (2003) 781–792.
 M.E. Evans, D.J. Feola, R.P. Rapp, Polymyxin B sulfate and colis-  A.M. Krieg, CpG motifs in bacterial DNA and their immune
tin: old antibiotics for emerging multiresistant gram-negative bac- eVects, Annu. Rev. Immunol. 20 (2002) 709–760.
teria, Ann. Pharmacother. 33 (1999) 960–967.  G. Hartmann, J. Battiany, H. Poeck, M. Wagner, M. Kerkmann,
 M. Vaara, Agents that increase the permeability of the outer mem- N. Lubenow, S. Rothenfusser, S. Endres, Rational design of new
brane, Microbiol. Rev. 56 (1992) 395–411. CpG oligonucleotides that combine B cell activation with high
 H. Hemmi, O. Takeuchi, T. Kawai, T. Kaisho, S. Sato, H. Sanjo, IFN-alpha induction in plasmacytoid dendritic cells, Eur. J.
M. Matsumoto, K. Hoshino, H. Wagner, K. Takeda, S. Akira, A Immunol. 33 (2003) 1633–1641.
Toll-like receptor recognizes bacterial DNA, Nature 408 (2000)  K. Fearon, J.D. Marshall, C. Abbate, S. Subramanian, P. Yee, J.
740–745. Gregorio, G. Teshima, G. Ott, S. Tuck, G. Van Nest, R.L. CoV-
 M. Bauer, V. Redecke, J.W. Ellwart, B. Scherer, J.P. Kremer, H. man, A minimal human immunostimulatory CpG motif that
Wagner, G.B. Lipford, Bacterial CpG-DNA triggers activation potently induces IFN-g and IFN-a production, Eur. J. Immunol.
and maturation of human CD11c(¡), CD123(+) dendritic cells, J. 33 (2003) 2114–2122.
Immunol. 166 (2001) 5000–5007.  J. Li, R.W. Milne, R.L. Nation, J.D. Turnidge, K. Coulthard,
 M. Kerkmann, S. Rothenfusser, V. Hornung, A. Towarowski, M. Stability of colistin and colistin methanesulfonate in aqueous
Wagner, A. Sarris, T. Giese, S. Endres, G. Hartmann, Activation media and plasma as determined by high-performance liquid
with CpG-A and CpG-B oligonucleotides reveals two distinct reg- chromatography, Antimicrob. Agents Chemother. 47 (2003) 1364–
ulatory pathways of type I IFN synthesis in human plasmacytoid 1370.
dendritic cells, J. Immunol. 170 (2003) 4465–4474.  R.S. Geary, T.A. Watanabe, L. Truong, S. Freier, E.A. Lesnik,
 H. Liang, Y. Nishioka, C.F. Reich, D.S. Pisetsky, P.E. Lipsky, N.B. SiouW, H. Sasmor, M. Manoharan, A.A. Levin, Pharmaco-
Activation of human B cells by phosphorothioate oligodeoxynu- kinetic properties of 2 -O-(2-methoxyethyl)-modiWed oligonu-
cleotides, J. Clin. Invest. 98 (1996) 1119–1129. cleotide analogs in rats, J. Pharmacol. Exp. Ther. 296 (2001)
 M. Bauer, K. Heeg, H. Wagner, G.B. Lipford, DNA activates 890–897.
human immune cells through a CpG sequence-dependent manner,  J.D. Marshall, E.M. Hessel, J. Gregorio, C. Abbate, P. Yee, M.
Immunology 97 (1999) 699–705. Chu, G. Van Nest, R.L. CoVman, K.L. Fearon, Novel chimeric
 G. Hartmann, A.M. Krieg, Mechanism and function of a newly immunomodulatory compounds containing short CpG oligode-
identiWed CpG DNA motif in human primary B cells, J. Immunol. oxyribonucleotides have diVerential activities in human cells,
164 (2000) 944–953. Nucleic Acids Res. 31 (2003) 5122–5133.
 M. Gursel, D. Verthelyi, D.M. Klinman, CpG oligodeoxynucleo-  T. Jensen, S.S. Pedersen, S. Garne, C. Heilmann, N. Hoiby, C.
tides induce human monocytes to mature into functional dendritic Koch, Colistin inhalation therapy in cystic Wbrosis patients with
cells, Eur. J. Immunol. 32 (2002) 2617–2622. chronic Pseudomonas aeruginosa lung infection, J. Antimicrob.
 S. Yamamoto, T. Yamamoto, S. Iho, T. Tokunaga, Activation of Chemother. 19 (1987) 831–838.
NK cell (human and mouse) by immunostimulatory DNA  J.A. Bosso, C.A. Liptak, D.K. Seilheimer, G.M. Harrison, Toxicity
sequence, Springer Semin. Immunopathol. 22 (2000) 35–43. of colistin in cystic Wbrosis patients, DICP 25 (1991) 1168–1170.