Use of pyrrolobenzodiazepines and related covalent-binding DNA-interactive molecules as ADC payloads: Is mechanism related to systemic toxicity?

Antibody-drug conjugates (ADCs) consist of monoclonal antibodies (mAbs) or antibody fragments conjugated to biologically active molecules (usually highly cytotoxic small molecules) through chemical linkers. Although no ADCs containing covalent-binding DNA-interactive payloads have yet been approved (although two containing the DNA-cleaving payload calicheamicin have), of those in clinical trials systemic toxicities are beginning to emerge. This article discusses the observed toxicities in relation to the structures and mechanisms of action of payload type.

The use of molecular dynamics simulations to evaluate the DNA sequence-selectivity of G-A cross-linking PBD-duocarmycin dimers.

The pyrrolobenzodiazepine (PBD) and duocarmycin families are DNA-interactive agents that covalently bond to guanine (G) and adenine (A) bases, respectively, and that have been joined together to create synthetic dimers capable of cross-linking G-G, A-A, and G-A bases. Three G-A alkylating dimers have been reported in publications to date, with defined DNA-binding sites proposed for two of them. In this study we have used molecular dynamics simulations to elucidate preferred DNA-binding sites for the three published molecular types. For the PBD-CPI dimer UTA-6026 (1), our simulations correctly predicted its favoured binding site (i.e., 5'-C(G)AATTA-3') as identified by DNA cleavage studies. However, for the PBD-CI molecule ('Compound 11', 3), we were unable to reconcile the results of our simulations with the reported preferred cross-linking sequence (5'-ATTTTCC(G)-3'). We found that the molecule is too short to span the five base pairs between the A and G bases as claimed, but should target instead a sequence such as 5'-ATTTC(G)-3' with two less base pairs between the reacting G and A residues. Our simulation results for this hybrid dimer are also in accord with the very low interstrand cross-linking and in vitro cytotoxicity activities reported for it. Although a preferred cross-linking sequence was not reported for the third hybrid dimer ('27eS', 2), our simulations predict that it should span two base pairs between covalently reacting G and A bases (e.g., 5'-GTAT(A)-3').

From Anthramycin to Pyrrolobenzodiazepine (PBD)-Containing Antibody-Drug Conjugates (ADCs).

The pyrrolo[2,1-c][1,4]benzodiazepines (PBDs) are a family of sequence-selective DNA minor-groove binding agents that form a covalent aminal bond between their C11-position and the C2-NH2 groups of guanine bases. The first example of a PBD monomer, the natural product anthramycin, was discovered in the 1960s, and the best known PBD dimer, SJG-136 (also known as SG2000, NSC 694501 or BN2629), was synthesized in the 1990s and has recently completed Phase II clinical trials in patients with leukaemia and ovarian cancer. More recently, PBD dimer analogues are being attached to tumor-targeting antibodies to create antibody-drug conjugates (ADCs), a number of which are now in clinical trials, with many others in pre-clinical development. This Review maps the development from anthramycin to the first PBD dimers, and then to PBD-containing ADCs, and explores both structure-activity relationships (SARs) and the biology of PBDs, and the strategies for their use as payloads for ADCs.

Effect of hairpin loop structure on reactivity, sequence preference and adduct orientation of a DNA-interactive pyrrolo[2,1-c][1,4]benzodiazepine (PBD) antitumour agent.

The pyrrolobenzodiazepines (PBDs) are a family of covalent-binding DNA-interactive minor-groove binding agents with a thermodynamic preference for binding to 5'-Pu-G-Pu-3' sequences (Pu = Purine) but a kinetic preference for 5'-Py-G-Py-3' (Py = Pyrimidine). Using HPLC/MS methodology and a range of designed hairpin-forming oligonucleotides, the kinetics of reaction of a C8-bis-pyrrole pyrrolobenzodiazepine (PBD) conjugate (GWL-78, 2) with sixteen isomeric oligonucleotides has been evaluated, each containing a single PBD binding site in one of two locations. The PBD-binding base-pair triplets were designed to include every possible combination of A and T bases adjacent to the covalently-reacting guanine, with the set of hairpins consisting of isomeric pairs containing the same sequence in the hairpin stem but with either hexaethylene glycol (HEG) or TTT loops. The PBD 2 reacted most rapidly with TGT and TGA sequences, with the possibility that adducts might form in both the 3'- and 5'-directions with some sequences according to modelling studies. A faster reaction rate was observed for all hairpins containing the HEG loop except one (Seq 10) when the PBD binding triplets were located either near the loop or adjacent to the 5'-end. Modelling studies have suggested that this difference in reactivity could be due to the structural flexibility of the HEG loop allowing both A-ring-3' and A-ring-5' adducts to form, while a TTT loop should favour only A-ring-5' adducts due to steric considerations. These findings contrast with the results reported by Nguyen and Wilson for the interaction of non-covalent DNA-binding molecules with DNA hairpins, where the loop structure was found to have little effect on interaction in the main stem of the hairpin.

Crispene E, a cis-clerodane diterpene inhibits STAT3 dimerization in breast cancer cells.

Crispene E, a new clerodane-type diterpene, inhibited STAT3 dimerization in a cell-free fluorescent polarisation assay and was found to have significant toxicity against STAT3-dependent MDA-MB 231 breast cancer cell line and selectively inhibited the expression of STAT3 and STAT3 target genes cyclin D1, Fascin and bcl-2. Molecular docking studies suggest the molecule inhibits STAT3 by interacting with its SH2 domain. The compound has been isolated from Tinospora crispa and characterized using standard spectroscopic techniques.

Investigation of the protein alkylation sites of the STAT3:STAT3 inhibitor Stattic by mass spectrometry.

STAT3 (Signal Transducer and Activator of Transcription factor 3) is constitutively active in a wide range of human tumours. Stattic is one of the first non-peptidic small molecules reported to inhibit formation of the STAT3:STAT3 protein dimer complex. A mass spectrometry method has been developed to investigate the binding of Stattic to the un-phosphorylated STAT3ßtc (U-STAT3) protein. Alkylation of four cysteine residues has been observed with possible reaction at a fifth which could account for the mechanism of action.

Characterization of a novel lytic protein encoded by the Bacillus cereus E33L gene ampD as a Bacillus anthracis antimicrobial protein.

Lytic proteins encoded by bacterial genomes have been implicated in cell wall biosynthesis and recycling. The Bacillus cereus E33L ampD gene encodes a putative N-acetylmuramoyl-l-alanine amidase. This gene, expressed in vitro, produced a very stable, highly active lytic protein. Very low concentrations rapidly and efficiently lyse vegetative Bacillus anthracis cells.

DNA sequence-selective G-A cross-linking ADC payloads for use in solid tumour therapies.

Antibody-Drug Conjugates (ADCs) are growing in importance for the treatment of both solid and haematological malignancies. There is a demand for new payloads with novel mechanisms of action that may offer enhanced therapeutic efficacy, especially in patients who develop resistance. We report here a class of Cyclopropabenzindole-Pyridinobenzodiazepine (CBI-PDD) DNA cross-linking payloads that simultaneously alkylate guanine (G) and adenine (A) bases in the DNA minor groove with a defined sequence selectivity. The lead payload, FGX8-46 (6), produces sequence-selective G-A cross-links and affords cytotoxicity in the low picomolar region across a panel of 11 human tumour cell lines. When conjugated to the antibody cetuximab at an average Drug-Antibody Ratio (DAR) of 2, an ADC is produced with significant antitumour activity at 1 mg/kg in a target-relevant human tumour xenograft mouse model with an unexpectedly high tolerability (i.e., no weight loss observed at doses as high as 45 mg/kg i.v., single dose).

Single nucleotide polymorphisms for assessing genetic diversity in castor bean (Ricinus communis).

BACKGROUND: Castor bean (Ricinus communis) is an agricultural crop and garden ornamental that is widely cultivated and has been introduced worldwide. Understanding population structure and the distribution of castor bean cultivars has been challenging because of limited genetic variability. We analyzed the population genetics of R. communis in a worldwide collection of plants from germplasm and from naturalized populations in Florida, U.S. To assess genetic diversity we conducted survey sequencing of the genomes of seven diverse cultivars and compared the data to a reference genome assembly of a widespread cultivar (Hale). We determined the population genetic structure of 676 samples using single nucleotide polymorphisms (SNPs) at 48 loci. RESULTS: Bayesian clustering indicated five main groups worldwide and a repeated pattern of mixed genotypes in most countries. High levels of population differentiation occurred between most populations but this structure was not geographically based. Most molecular variance occurred within populations (74%) followed by 22% among populations, and 4% among continents. Samples from naturalized populations in Florida indicated significant population structuring consistent with local demes. There was significant population differentiation for 56 of 78 comparisons in Florida (pairwise population phiPT values, p < 0.01). CONCLUSION: Low levels of genetic diversity and mixing of genotypes have led to minimal geographic structuring of castor bean populations worldwide. Relatively few lineages occur and these are widely distributed. Our approach of determining population genetic structure using SNPs from genome-wide comparisons constitutes a framework for high-throughput analyses of genetic diversity in plants, particularly in species with limited genetic diversity.

UPLC-based assay to assess the hydrophobicity of Antibody-Drug Conjugate (ADC) payloads.

Antibody-Drug Conjugates (ADCs) consist of antibodies attached to cytotoxic small molecules or biological agents (i.e., payloads) through chemical linkers which may be cleavable or non-cleavable. The development of new ADCs is challenging, particularly the process of attaching the linker-payload construct to the antibody (i.e., the conjugation process). One of the major problems associated with conjugation is high hydrophobicity of the payload which can lead to low yields of the ADC through aggregation and/or lower than desired Drug-Antibody Ratios (DARs). We report here a UPLC-based assay that can be used to study the physicochemical properties of ADC payloads at an early stage of development, and to provide information on whether the hydrophilic-hydrophobic balance is suitable for conjugation or further physicochemical optimization is required. The assay is relatively simple to establish and should be of use to those working in the ADC area.

A Novel Antibody-Drug Conjugate (ADC) Delivering a DNA Mono-Alkylating Payload to Chondroitin Sulfate Proteoglycan (CSPG4)-Expressing Melanoma.

Despite emerging targeted and immunotherapy treatments, no monoclonal antibodies or antibody-drug conjugates (ADCs) directly targeting tumor cells are currently approved for melanoma therapy. The tumor-associated antigen chondroitin sulphate proteoglycan 4 (CSPG4), a neural crest glycoprotein over-expressed on 70% of melanomas, contributes to proliferative signaling pathways, but despite highly tumor-selective expression it has not yet been targeted using ADCs. We developed a novel ADC comprising an anti-CSPG4 antibody linked to a DNA minor groove-binding agent belonging to the novel pyrridinobenzodiazepine (PDD) class. Unlike conventional DNA-interactive pyrrolobenzodiazepine (PBD) dimer payloads that cross-link DNA, PDD-based payloads are mono-alkylating agents but have similar efficacy and substantially enhanced tolerability profiles compared to PBD-based cross-linkers. We investigated the anti-tumor activity and safety of the anti-CSPG4-(PDD) ADC in vitro and in human melanoma xenografts. Anti-CSPG4-(PDD) inhibited CSPG4-expressing melanoma cell growth and colony formation and triggered apoptosis in vitro at low nanomolar to picomolar concentrations without off-target Fab-mediated or Fc-mediated toxicity. Anti-CSPG4-(PDD) restricted xenograft growth in vivo at 2 mg/kg doses. One 5 mg/kg injection triggered tumor regression in the absence of overt toxic effects or of acquired residual tumor cell resistance. This anti-CSPG4-(PDD) can deliver a highly cytotoxic DNA mono-alkylating payload to CSPG4-expressing tumors at doses tolerated in vivo.

Single nucleotide polymorphism typing of Bacillus anthracis from Sverdlovsk tissue.

A small number of conserved canonical single nucleotide polymorphisms (canSNP) that define major phylogenetic branches for Bacillus anthracis were used to place a Sverdlovsk patient's B. anthracis genotype into 1 of 12 subgroups. Reconstruction of the pagA gene also showed a unique SNP that defines a new lineage for B. anthracis.

Differentiation of Clostridium botulinum serotype A strains by multiple-locus variable-number tandem-repeat analysis.

Ten variable-number tandem-repeat (VNTR) regions identified within the complete genomic sequence of Clostridium botulinum strain ATCC 3502 were used to characterize 59 C. botulinum strains of the botulism neurotoxin A1 (BoNT/A1) to BoNT/A4 (BoNT/A1-A4) subtypes to determine their ability to discriminate among the serotype A strains. Two strains representing each of the C. botulinum serotypes B to G, including five bivalent strains, and two strains of the closely related species Clostridium sporogenes were also tested. Amplified fragment length polymorphism analyses revealed the genetic diversity among the serotypes and the high degree of similarity among many of the BoNT/A1 strains. The 10 VNTR markers amplified fragments within all of the serotype A strains but were less successful with strains of other serotypes. The composite multiple-locus VNTR analysis of the 59 BoNT/A1-A4 strains and 3 bivalent B strains identified 38 different genotypes. Thirty genotypes were identified among the 53 BoNT/A1 and BoNT/A1(B) strains, demonstrating discrimination below the subtype level. Contaminating DNA within crude toxin preparations of three BoNT/A subtypes (BoNT/A1 to BoNT/A3) also supported amplification of all of the VNTR regions. These markers provide clinical and forensics laboratories with a rapid, highly discriminatory tool to distinguish among C. botulinum BoNT/A1 strains for investigations of botulism outbreaks.

Evaluating Burkholderia pseudomallei Bip proteins as vaccines and Bip antibodies as detection agents.

Burkholderia pseudomallei is a biothreat agent and an important natural pathogen, causing melioidosis in humans and animals. A type III secretion system (TTSS-3) has been shown to be critical for virulence. Because TTSS components from other pathogens have been used successfully as diagnostic agents and as experimental vaccines, it was investigated whether this was the case for BipB, BipC and BipD, components of B. pseudomallei's TTSS-3. The sequences of BipB, BipC and BipD were found to be highly conserved among B. pseudomallei and B. mallei isolates. A collection of monoclonal antibodies (mAbs) specific for each Bip protein was obtained. Most recognized both native and denatured Bip protein. Burkholderia pseudomallei or B. mallei did not express detectable BipB or BipD under the growth conditions used. However, anti-BipD mAbs did recognize the TTSS needle structures of a Shigella strain engineered to express BipD. The authors did not find that BipB, BipC or BipD are protective antigens because vaccination of mice with any single protein did not result in protection against experimental melioidosis. Enzyme-linked immunosorbent assay (ELISA) studies showed that human melioidosis patients had antibodies to BipB and BipD. However, these ELISAs had low diagnostic accuracy in endemic regions, possibly due to previous patient exposure to B. pseudomallei.

Fatal pneumonia among metalworkers due to inhalation exposure to Bacillus cereus Containing Bacillus anthracis toxin genes.

Bacillus cereus pneumonia is unusual in nonimmunocompromised hosts. We describe fatal cases in 2 metalworkers and the associated investigation. Anthrax toxin genes were identified in B. cereus isolates from both patients using polymerase chain reaction. Finding anthrax toxin genes in non-Bacillus anthracis isolates has, to our knowledge, only been reported once previously.

Sequence-selective binding of C8-conjugated pyrrolobenzodiazepines (PBDs) to DNA.

DNA footprinting and melting experiments have been used to examine the sequence-specific binding of C8-conjugates of pyrrolobenzodiazepines (PBDs) and benzofused rings including benzothiophene and benzofuran, which are attached using pyrrole- or imidazole-containing linkers. The conjugates modulate the covalent attachment points of the PBDs, so that they bind best to guanines flanked by A/T-rich sequences on either the 5'- or 3'-side. The linker affects the binding, and pyrrole produces larger changes than imidazole. Melting studies with 14-mer oligonucleotide duplexes confirm covalent attachment of the conjugates, which show a different selectivity to anthramycin and reveal that more than one ligand molecule can bind to each duplex.

Covalent Bonding of Pyrrolobenzodiazepines (PBDs) to Terminal Guanine Residues within Duplex and Hairpin DNA Fragments.

Pyrrolobenzodiazepines (PBDs) are covalent-binding DNA-interactive agents with growing importance as payloads in Antibody Drug Conjugates (ADCs). Until now, PBDs were thought to covalently bond to C2-NH2 groups of guanines in the DNA-minor groove across a three-base-pair recognition sequence. Using HPLC/MS methodology with designed hairpin and duplex oligonucleotides, we have now demonstrated that the PBD Dimer SJG-136 and the C8-conjugated PBD Monomer GWL-78 can covalently bond to a terminal guanine of DNA, with the PBD skeleton spanning only two base pairs. Control experiments with the non-C8-conjugated anthramycin along with molecular dynamics simulations suggest that the C8-substituent of a PBD Monomer, or one-half of a PBD Dimer, may provide stability for the adduct. This observation highlights the importance of PBD C8-substituents, and also suggests that PBDs may bind to terminal guanines within stretches of DNA in cells, thus representing a potentially novel mechanism of action at the end of DNA strand breaks.

Identification of Genome-Wide Mutations in Ciprofloxacin-Resistant F. tularensis LVS Using Whole Genome Tiling Arrays and Next Generation Sequencing.

Francisella tularensis is classified as a Class A bioterrorism agent by the U.S. government due to its high virulence and the ease with which it can be spread as an aerosol. It is a facultative intracellular pathogen and the causative agent of tularemia. Ciprofloxacin (Cipro) is a broad spectrum antibiotic effective against Gram-positive and Gram-negative bacteria. Increased Cipro resistance in pathogenic microbes is of serious concern when considering options for medical treatment of bacterial infections. Identification of genes and loci that are associated with Ciprofloxacin resistance will help advance the understanding of resistance mechanisms and may, in the future, provide better treatment options for patients. It may also provide information for development of assays that can rapidly identify Cipro-resistant isolates of this pathogen. In this study, we selected a large number of F. tularensis live vaccine strain (LVS) isolates that survived in progressively higher Ciprofloxacin concentrations, screened the isolates using a whole genome F. tularensis LVS tiling microarray and Illumina sequencing, and identified both known and novel mutations associated with resistance. Genes containing mutations encode DNA gyrase subunit A, a hypothetical protein, an asparagine synthase, a sugar transamine/perosamine synthetase and others. Structural modeling performed on these proteins provides insights into the potential function of these proteins and how they might contribute to Cipro resistance mechanisms.

Computational studies support the role of the C7-sibirosamine sugar of the pyrrolobenzodiazepine (PBD) sibiromycin in transcription factor inhibition.

The pyrrolo[2,1-c][1,4]benzodiazepines (PBDs) are a group of sequence-selective, DNA minor-groove binding agents that covalently attach to guanine residues. Originally derived from Streptomyces species, a number of naturally occurring PBD monomers exist with varying A-Ring and C2-substituents. One such agent, sibiromycin, is unusual in having a glycosyl residue (sibirosamine) at its A-Ring C7-position. It is the most cytotoxic member of the naturally occurring PBD family and has the highest DNA-binding affinity. Recently, the analogue 9-deoxysibiromyin was produced biosynthetically by Yonemoto and co-workers.1 Differing only in the loss of the A-Ring C9-hydroxyl group, it was reported to have a significantly higher DNA-binding affinity than sibiromycin based on DNA thermal denaturation studies, although these data have since been retracted.2 As deletion of the C9-OH moiety, which points toward the DNA minor groove floor, might intuitively be expected to reduce DNA-binding affinity through the loss of hydrogen bonding, we carried out molecular dynamics simulations on the interaction of both molecules with DNA over a 10 ns time-course in explicit solvent. Our results suggest that the two molecules may differ in their sequence-selectivity and that 9-deoxysibiromycin should have a lower binding affinity for certain sequences of DNA compared to sibiromycin. Our molecular dynamics results indicate that the C7-sibirosamine sugar does not form hydrogen bonding interactions with groups in the DNA minor-groove wall as previously reported, but instead points orthogonally out from the minor groove where it may inhibit the approach of DNA control proteins such as transcription factors. This was confirmed through a docking study involving sibiromycin and the GAL4 transcription factor, and these results could explain the significantly enhanced cytotoxicity of sibiromycin compared to other PBD family members without bulky C7-substituents.

Pyrrolobenzodiazepines (PBDs) do not bind to DNA G-quadruplexes.

The pyrrolo[2,1-c][1,4] benzodiazepines (PBDs) are a family of sequence-selective, minor-groove binding DNA-interactive agents that covalently attach to guanine residues. A recent publication in this journal (Raju et al, PloS One, 2012, 7, 4, e35920) reported that two PBD molecules were observed to bind with high affinity to the telomeric quadruplex of Tetrahymena glaucoma based on Electrospray Ionisation Mass Spectrometry (ESI-MS), Circular Dichroism, UV-Visible and Fluorescence spectroscopy data. This was a surprising result given the close 3-dimensional shape match between the structure of all PBD molecules and the minor groove of duplex DNA, and the completely different 3-dimensional structure of quadruplex DNA. Therefore, we evaluated the interaction of eight PBD molecules of diverse structure with a range of parallel, antiparallel and mixed DNA quadruplexes using DNA Thermal Denaturation, Circular Dichroism and Molecular Dynamics Simulations. Those PBD molecules without large C8-substitutents had an insignificant affinity for the eight quadruplex types, although those with large π-system-containing C8-substituents (as with the compounds evaluated by Raju and co-workers) were found to interact to some extent. Our molecular dynamics simulations support the likelihood that molecules of this type, including those examined by Raju and co-workers, interact with quadruplex DNA through their C8-substituents rather than the PBD moiety itself. It is important for the literature to be clear on this matter, as the mechanism of action of these agents will be under close scrutiny in the near future due to the growing number of PBD-based agents entering the clinic as both single-agents and as components of antibody-drug conjugates (ADCs).