Lipstick on the Pig

The rise of antibiotic resistance regularly prompts calls for another look at phage therapy, a phenomenon that dates back at least to the 1960s. Although resistance rates have skyrocketed in the intervening decades, phage therapy remains decidedly earthbound.

Bacteriophage are viruses that infect and (often) kill bacteria. Phage therapy was intensively pursued in the 1920s. Despite tremendous advances in preventing infectious diseases with vaccines and better sanitation, there was still little that doctors could do to treat active infections, and infectious diseases remained far and away the leading killers in all countries. Anything that attacked bacteria and left human cells alone, as phage most assuredly do, could be a world-changing medical advance. 

Doctors jumped at the chance. More than 2000 scientific papers describing phage and phage therapy were published by 1930. It is likely that no other medical technology was as intensively studied in the first decades of the 20th century. Phage therapy was given every possible opportunity to succeed.

Instead it failed. The usual story is that phage therapy was eclipsed by penicillin. But its demise came more than a decade before penicillin became available. By 1934, a comprehensive review of the state of phage therapy concluded that it was a failure, and that it was time to move on.

Some of the sources of failure were really limitations of the state of biotechnology at the time, such as contamination with endotoxin. These failures could be readily remedied with modern best practices. But other limitations are intrinsic to phage and bacterial and human biology: limited host range, poor biodistribution, immunogenicity, rapid evolution of resistance in bacterial targets. Remedies and workarounds can be found for any one or two of these limitations. Fixing them all is implausible. That doesn’t mean that phage therapy can never work, but it does mean that its applications are likely to be limited.

But there’s always a chance for a game-changing breakthrough, right? How about incorporating the newest, coolest, hottest technology into phage? Maybe that will do the trick and push phage therapy through the valley of death and get it into the clinic.

Tim Lu and colleagues at MIT have attempted to do just this. They incorporated CRISPR-Cas technology into phage, turning them into nanorobots that precisely target those bacteria which harbor genes for virulence factors and antibiotic resistance. The bad guys get wiped out or get converted to good guys. The good bugs, including members of the same species that are not resistant, are spared. This is precision medicine brought to infectious disease therapy. Rather than decimate our gut microbiota with antibiotics, good and bad bugs alike, only the bugs with malign intent are marked for destruction. It is the Minority Report made real in the micro world

Lu is a leading exponent of synthetic biology, and the level of skill in his lab is matched by few and exceeded by none. But arming phage with CRISPR does exactly nothing to address the limitations of phage therapy.

CRISPR programs phage to destroy particular bacterial genes. If the genes are on the bacterial chromosome, the bacteria are killed. If the genes are on extrachromosomal DNA (as so many resistance genes are), the genes are eliminated but the bacteria are spared. This makes phage killing of bad bugs exquisitely specific

But specificity has never been a barrier to adoption of phage therapy. In fact, phage tend to be too specific. In order to target all the strains within a species, it is usually necessary to concoct a “cocktail” of half a dozen or so different phage variants. The thing you have to remember is that the level of genetic diversity in a bacterial species far outstrips the diversity in a plant or animal species. All humans – as diverse as we imagine ourselves to be – are 99.5% identical. All E. coli, by contrast, share only a core set of genes that comprises maybe 40% of their genome. Two strains of E. coli, in other words, can be about as genetically related to each other as a dog is to a dogwood tree.

Phage that target only bacterial strains that are virulent or that are antibiotic resistant have been around for years. Bacteriophage that are highly specific to E. coli O157:H7, the strain that causes deadly hemolytic-urinary syndrome, were described in 1999. Although these phage and others have found commercial use in food decontamination, none have been tested in a clinical trial for therapeutic efficacy.

Other researchers have reported the identification of phage that exclusively target methicillin-resistant Staph aureus (MRSA), which kills some 19,000 Americans each year, while sparing susceptible strains, as well as all other bacteria. So far as I know, no one has even attempted to commercialize this technology, let alone move it into the clinic.

The exquisite specificity of the CRISPR phage technology is a liability, not an asset. CRISPR targets very specific gene sequences. A change of one or two nucleotides in a gene renders it invisible to CRISPR. This means that (1) bacteria can evolve resistance to CRISPR phage in short order and (2) pre-existing variants will escape destruction.

Let’s take the second scenario first. Much of the concern about multi-resistant “superbugs” centers on Gram-negative bacteria (like E. coli) which harbor metallo-ß-lactamase genes that break down antibiotics. Many hundreds of different variants of these genes are known to science, and it is likely that thousands more exist. The notion that all, or even most of these variants can be targeted on the basis of their DNA sequences is a fantasy

Lu and colleagues are not fools, and they did consider the possibility that bacteria could spontaneously acquire mutations that render them resistant to attack by CRISPR. But they didn’t consider this possibility seriously enough. They confirmed that these sorts of mutations are very rare indeed, occurring at a frequency of less than 1 in 100,000. But the thing about bacteria is that there are lots of them, and the Law of Large Numbers applies: anything that is not impossible is inevitable. Antibiotic resistance mutations typically occur in bacterial populations at a frequency of 1 in 100 million to 1 in 10 billion. Maybe resistance would not evolve during the course of a single treatment with CRISPR phage, but I doubt it. Bacteria and phage have been battling it out for billions of years, and neither one has gotten, or will ever get, the upper hand. It is likely enough that bacteria will use their own CRISPR systems – which provide immunity to bacteriophage infection – to defeat attacks by CRISPR phage. I’ll bet that an enterprising high school student could demonstrate this for a science fair project.

Using CRISPR to create phage that specifically attack resistance and virulence genes in bacterial populations is clever. But the thing about developing therapeutics is that cleverness counts for very little. Drug development is hard, boring, repetitious work. It’s all about solving dozens or hundreds of problems that pop up at every step of the way that cause a drug candidate to lose efficacy or cause side effects. The problems and limitations of phage therapy have been known for decades. As a therapeutic platform, it has a distinctly porcine cast. Slapping a little CRISPR on its lips does nothing to change this.