Bacterial infections are a big problem for us. Although we have lots of antibiotics, bacteria develop resistance at an alarming rate – which makes the discovery of new antibiotics, or new strategies to use with pre-existing antibiotics, very urgent. My current research project has given me the opportunity to find new ways to kill these bad bugs by building a “Trojan Horse.”
When we take a closer look at these “bugs,” bacteria can be divided into two major groups based on their Gram staining result: Gram-positive and Gram-negative. The difference upon staining is due to their different cell wall structures. Gramnegative bacteria have two layers of cell membrane while Gram-positives have only one layer (Figure i). Because the additional membrane serves as a permeability barrier, Gramnegative bacteria are generally difficult to treat. Small nutrients like sugars and amino acids enter the Gram-negative cells through channels called porins that span the outer membrane. These porins only allow passage of hydrophilic molecules that are relatively small in size (up to 500 Dalton). Although some antibiotics can enter Gram-negatives through porins, antibiotics that are larger than this cut-off or that exhibit hydrophobic properties cannot enter and thus do not work well against Gram-negative bacteria. Moreover, antibiotic resistance has emerged via mutations that either decrease the channel size of porins or decrease the number of porins present on the outer membrane, both of which lower the successful passage of antibiotic drugs. In the long run, if we can overcome this barrier and better deliver antibiotics into Gram-negative bacteria, we will achieve better treatment.
Figure i. Cell wall structures of Gram-positive and Gram-negative bacteria. The Gram-positive cell wall is composed of a single layer membrane and thick peptidoglycan. The Gram-negative cell wall has two membranes. Porin and other protein channels are shown in the figure.
To address this problem, we have to take another look at the bacteria. To support their growth, bacteria need to transport nutrients, including some large hydrophobic molecules like vitamin B12, sugars, and amino acids. How do those molecules pass the permeability barrier? It turns out that bacteria have dedicated transport machineries specifically for these nutrients. Of particular interest is the transport system responsible for iron uptake. Iron is an essential nutrient for almost all living organisms; however, the “free” iron concentration in the environment is very low due to its low solubility and toxicity. In order to obtain enough iron, bacteria make a class of small molecules, siderophores, which bind to iron with high affinities. Bacteria secrete siderophores to the environment to catch iron; once “caught,” the bacteria utilize sophisticated machinery involving multiple proteins to actively transport the iron-bond siderophores into the cell. Afterwards, iron is released from the siderophores, with the help of other proteins, for cellular use. For most bacterial pathogens, iron acquisition is essential for replication in its host, and siderophore transport machinery is considered to be important for virulence. It is a promising target to hijack.
Here comes our “Trojan Horse.” If we can disguise the antibiotic (or other cargo) with a siderophore molecule, the machinery may be fooled into transporting the drug through the outer membrane permeability barrier – and into the bacteria – allowing the antibiotic to do its job. Moreover, this siderophoremediated iron transport mechanism is not present in humans, so this strategy has high delivery specificity and potentially low toxicity. The delivery specificity can be further developed to select specific pathogenic bacterial strains and leave the commensals intact, based on the fact that different bacteria utilize different siderophores.
Figure ii. a) Chemical structure of enterobactin (Ent) and a peptide-Ent conjugate Microcin E492m. b) Ent transportation machinery from E. coli. The outer membrane receptor FepA recognizes Ent-Fe; with the help of ExbD/ExbB/TonB system, FepA deliver Ent-Fe to the periplasm, where FepB shuttles it to the inner membrane; FepD/FepG/FepC transport Ent-Fe to the cytosol using the energy from ATP hydrolysis; Fes disassemble Ent and Fe is released for cellular usage. We aim to find out if, when a cargo is attached to Ent, the resulting conjugate will still be transported by this machinery.
For my project, I started with a particular siderophore called enterobactin, or Ent, which is among the strongest iron chelators created by nature (Figure iia). Its uptake machinery is well understood (Figure iib), and it is used by many pathogenic bacteria. In fact, some bacterial species (e.g., Klebsiella) have applied this “Trojan Horse” approach by attaching antimicrobial peptides to Ent (Microcin E492m, Figure iia) to kill othercompetitors; this example indicates that proper modification on Ent can afford transport via the same machinery. When I started my project, it was unknown whether other types of cargos, like antibiotics, could be delivered into the cell in this way and whether there was a size limitation that the transport machinery could tolerate. In order to investigate these questions, the first challenge was to figure out how to modify Ent using chemical synthesis.
First, I designed and executed a synthetic route to modify Ent and attach cargos to it. I started with very simple building blocks and aimed to develop a synthetic route that is modular. By doing so, I obtained a key intermediate with a versatile chemical handle, Bn6Ent-R, which can be used to attach various cargo molecules through different types of reactions (Figure iii). As a preliminary study, I chose the very robust acid-amine coupling reaction to attach a series of cargos with various sizes and shapes to probe the tolerance of the Ent uptake machinery. Another chemical handle I used was azide, which is needed for a highly efficient and selective cycloaddition reaction (or “Click” reaction) to link complicated cargos (e.g., vancomycin) that are not compatible with the acid-amine coupling reactions (Figure iii, Ent-cargo).
Figure iii. Syntheses and structures of Ent-cargo conjugates.
I prepared 11 Ent-cargo conjugates with different-sized cargos, and demonstrated that they still maintain iron-binding ability. When testing on bacteria, I found that the conjugates are transported by the Ent-uptake machinery with different efficiencies. The delivery seems to be size-dependent, which is not a surprise, because the Ent transporter on the outer membrane (FepA, Figure iib) has a channel-like structure with defined size. This discovery is important for us for designing the real “Trojan Horse,” because it tells us the promiscuity of the transport machinery.
Next, we selected the first weapon to load: b-lactams. This class of antibiotics is well studied, has the proper size, and is usually used for Gram-positive bacteria, but is less efficient against Gram-negative bacteria due to the permeability issue described earlier. I attached two of them (ampicillin and amoxicillin) to Ent using the “Click” reaction. Gratifyingly, when testing them against several Gram-negative bacterial species, I observed a 10-1000 fold activity increase from the conjugates compared to the parent antibiotics. Besides, these conjugates are able to kill bacteria much faster (Figure iv), most likely due to the active transport mechanism. Several control experiments were also performed and confirmed that the activity increase is indeed due to Ent mediated delivery.
Figure iv. a) Growth of a pathogenic E. coli strain in the presence of Ent-Ampicillin or Ampicillin (Ent-Amp or Amp) at different concentrations. OD600 values represent the bacterial cell density in the growth culture. Ent-Amp inhibits the bacterial growth at ~10-8 M concentration; Amp requires a 1000-fold higher concentration (10-5 M) to exhibit the same effect. b) The growth of the same strain monitored over time in the presence of Ent-Amp (5 μM) or Amp (50 μM) or no compound (control). After approximately 1 hour, Ent-Amp caused the bacterial population to drop significantly; Amp needs longer than 3 hours to cause the same effect even at a 10-fold higher concentration.
With these exciting results, we are now working on expanding the strategy to other types of siderophores and other bacteria. We also plan to load other cargos, including toxins, fluorophores, and biotin for different types of applications. I really enjoyed the journey from designing and building new molecules to finding their activity in living organisms. And I feel it is very promising that said “Trojan Horse” approach will have a clinical impact and help people in the future.