In this exclusive interview, Graham Hatfull, PhD, the Eberly Family Professor of Biotechnology at the University of Pittsburgh and Howard Hughes Medical Institute (HHMI) Professor discusses phage biology and details of the case study published in Nature Medicine in May that used phages to successfully treat a Mycobacterium abscessus infection in a teenage girl.

GEN: How did you first get involved in the story that made headlines last month of the 15-year-old girl whose mycobacterium infection was halted through treatment with phages?

Graham Hatfull: For us, this particular endeavor started when we were contacted by a colleague in London. They had a couple of patients at the hospital [Great Ormond Street Hospital for Children (GOSH)] who had serious infections with Mycobacterium abscessus, and it is difficult to treat with antibiotics because they are highly resistant and do not respond to treatment. Both of the patients were of a similar profile. Both had cystic fibrosis (CF) and double lung transplants and a disseminated infection with this particular organism. But, they had different strains of that species.

The reason why James [Soothill] contacted us and sent us the strains is because I had met James at a phage therapy meeting in Tbilisi, Georgia, in 1997. James remembered that my lab works on phages that infect mycobacteria. So, they sent us the two bacterial strains with, I would say, the rather distant prospect of seeing if we had bacteriophages that might infect those strains. I was not thinking in any great specificity about what we would do if we found them, but we were certainly interested and willing to look.

GEN: How did you acquire your enormous collection of phages?

Hatfull: We help to run (together with the Howard Hughes Medical Institute) a large program called the SEA-PHAGES (Science Education Alliance-Phage Hunters Advancing Genomics and Evolutionary Science) program that is now in its 11th year. It is designed to provide authentic research experiences to first-year undergraduate college students. They isolate new phages from the environment and characterize them including sequencing and annotating the genomes. This is where the bulk of phage discovery is happening. This is not a small program. It is about 150 institutions, ranging from community colleges to R1 research institutions. This year, over 5,000 students are engaged in phage discovery and genomic characterization. All of the phages are sent to us and archived which is how we now have such a large archive of 15,000 phages. We have been doing phage genome sequencing for over 30 years, but we now do it at a large scale so that we can understand these questions of diversity and evolution.

Graham Hatfull, PhD

So, this program is incredibly important to the students who are really motivated by the idea that their phages may be used therapeutically. And, out of the three phages used in this study, one of them came from Providence College, which is a member of the SEA-PHAGES program. But, whether or not one or more of these phages are actually used to make people better, they are excited to be part of this large enterprise to understand these big questions of phage diversity. SEA-PHAGES has not only been a fun, and interesting and highly productive program to be a part of, but we also have detailed assessment that shows that students who participate are much more likely to continue in science education. Persistence is such a problem in STEM, generally and particularly with underrepresented minorities, this opens the door to broad inclusion of any students who want to do this, promotes persistence, and has the prospect of influencing the diversity of our scientific community.

GEN: Let us talk about how you found the three phages that were used in the patient.

Hatfull: Once we received the bacterial strains, we started testing for phages. The bottom line is that finding the phages for these strains is difficult. For one of the strains, we found a phage a couple of months later, but the patient had already died. These are very serious infections. She was quite poorly and, although we found a phage, we were too late to even consider being of any help. For the other strain, the patient hung in there long enough that we managed to find three phages that we then grouped into a cocktail. We prepared them, went through the extraordinary paperwork that is required to get regulatory approval, and then treated the patient intravenously about 6 months later. And, she has done very well.

We know a lot about phages. There are people that think that you should only use obligatory lytic phages therapeutically. But we do not have a lot of those. We have very few phages at all that infect these types of strains. And the ones that we do have are mostly temperate. So, the best that we can do is to engineer them to make them lytic and then use them. In order to help this patient, we needed to accept that it is ok to use engineered phages, derivatives of temperate phages, that we have made lytic, without any obvious deleterious outcomes.

GEN: How did you engineer the phages?

Hatfull: We used a recombineering system that we have developed, that allows us to remove genes that will convert the phages from being temperate into lytic—giving the phages the ability to kill efficiently. When we do the engineering, we can do it in such a way as to precisely delete those genes, without leaving any scars or additional material. And, because we have not added any sequences to them that are not normally there, they are technically not recombinant, so they are not genetically modified organisms (GMOs). Using phages that are GMOs is a whole difficult and complex registry landscape than using those that are technically engineered but not GMOs. We developed these tools a decade ago, but no one cared about them until we could put them to good use in a clinical application.

The background is that recombineering was developed for making E. coli mutants quite a long time ago and works on the basis of using phage-encoded recombinases to juice up the recombination system of the host. We tested and showed, as many other people did, that the E. coli systems do not work in Mycobacterium strains. We were able to find the genes encoding the enzymes for the phages of the mycobacteria to make them work in a homologous system. By using mycobacteriophage enzymes, we could get this to work in mycobacteria relatively easily and we described those systems for making mutants in multiple mycobacterial strains. Then we adapted that system to engineer the phages themselves. It is a trivial process, really. We make a DNA substrate which contains the mutation and then we combine that with phage genomic DNA and coelectroporate the two DNAs into the strain that has the recombineering functions. We used these juiced up recombineering systems to recover plaques and screen them with PCR to find those with the mutant allele. We have been able to make lots of different types of mutants in this way. For therapeutic purposes, we just deleted genes. But we can also make point mutations for our research.

GEN: How many genes do you have to delete to change the phage from lysogenic to lytic?

Hatfull: Basically, one. If we had our druthers, we would probably take out others. But the key gene is the phage repressor which temperate phages use to form lysogens. So, we just go in and remove the repressor gene, or a large part of the repressor gene.

The three phages in the cocktail had different things done to them. One of them, called Muddy, was used as is. It needed no changes to kill very efficiently. For another phage, called ZoeJ, we did what I just described. We engineered out the repressor gene to make it lytic and used that. For the third one, we made the lytic derivative by deleting part of the repressor gene, but then we also did forward genetics in order to isolate a mutant in which the host range has been changed. So, the phage learned to infect the particular strain we are trying to kill.

I would just add parenthetically that the fact that we had to do all of that is interesting to me. And, it illustrates something about the challenges that these strains pose to doing this. We have a collection of about 15,000 individual phage isolates here in Pittsburgh and about 10,000 of those infect Mycobacterium smegmatis. We have sequenced about 3,000 total phages. So, we have a lot of phages and a lot of sequenced genomes for which we understand their diversity. We relied on all of this information to try to find the phages that infect this one particular strain.

We found Muddy relatively quickly. But, finding the others was challenging and we had to go through these engineering tricks in order to get them. In fact, we have done the same kind of approach for other clinical isolates of Mycobacterium abscessus and it has been difficult to find any phages that work well on many of those other strains. So, we kind of got lucky with this one particular strain and we managed to bootstrap our way through to three phages. We have not been that successful with most of the other strains that we have looked at. It was a tour de force in some ways in order to be able to get to a point to be able to even contemplate doing the therapy.

GEN: How did you decide on the number of phages to use? Where does the number 3 come from?

Hatfull: I do not think that there is anything special about the number 3 per se, but here is the rationale. We worry substantially about the emergence of phage resistance and the eventual failure or relapse due to resistance. If you have one phage, that is likely to happen quite readily. If you have two, you will counter it more effectively, and with three, you should counter even yet more effectively. The idea is to battle resistance by using the phages combinatorially. In the Patterson case, they had used a couple of cocktails and saw resistance even when putting a relatively high pressure with several different phages. So, three seemed like a number which we felt we would have some confidence that we would not see resistance.

GEN: Do you lay in bed worrying that resistance might rear its ugly head one of these days?

Hatfull: That is just one of the things that I worry about a lot. We have not seen resistance yet in the isolates that have been recovered from the patient. They retain sensitivity to the three phages, which is a good sign. And, we have since been able to isolate a fourth phage. The hope and the plans are that we will add that to the cocktail in order to, basically, add more pressure onto the bugs. But that is not because we see resistance. It is because we would like to help the phages do whatever they are doing as best we can, and we know so little about what is happening in vivo.

In terms of worry about things more broadly, this is a case study that obviously has uncertain outcomes. We could make a case, after having thought about it very deeply, that the chances of doing harm were fairly minimal.

To me, there are three possible outcomes. First, that there is no effect and no benefit whatsoever. Second, you get a complete cure and the whole thing is gone never to be seen again. And the third is that you buy time. We can rule the first one out. There clearly has been a clinical change. And, although we cannot prove that the clinical change was from the phages, it is consistent with the idea that the phages provided some benefit. But this is a CF patient with new lungs on immunosuppressive drugs. The long-term prospects remain, and always did remain unclear. At minimum, we have bought time. And, to me, there is value in having done that.

GEN: How many cases are out there like this, where a bacterial infection has been successfully treated with phages?

Hatfull: That is a good question. Let me subdivide the answer. There is one particular set of cases, of which I am sure there are many, and have been done mostly in the former Soviet Union or more recently in Poland, which are mostly topical applications for burns and other kinds of infections. The second class is intravenous injections, which is a whole different ball of wax. In some ways, this was started with the Tom Patterson case that many are familiar with. Chip Schooley, who worked as the physician on that case, worked very closely with us on all of these therapeutically related issues of our case. Chip has been involved in more than half a dozen cases like this. So, I am sure that there are more than just these two. And, I would not be surprised if it is dozens at this point. But I do not have a hard and fast number on that.

GEN: This case study made a big splash when it was published. Why is that?

Hatfull: I think there are two reasons. First of all, because the infection was with a mycobacterium strain. There is a whole set of infections with nontuberculosis mycobacteria—the NTMs—that have become recognized as clinically important. Those types of infections themselves are often highly resistant to antibiotics so they make a good target for phage approaches. The other reason is that this type of approach makes you think a little bit more about tuberculosis itself— which kills 1.7 million people a year. TB is caused by Mycobacterium tuberculosis, obviously a different species to the one we treated here. But, nonetheless, it is the same genus so there are a lot of similarities in the type of disease and pathologies. TB is something of a controversial topic in the phage world (as if everything else was not) because a lot of folks question whether phages would be useful against TB because the bacteria invade macrophages in the body and also in the granulomatous structures and, therefore, it may be difficult to get phages to those areas. So, there have been a lot of questions as to whether TB would be a good target for phage therapy or not. The fact that our efforts seemed to be ‘‘successful’’ raises the question as to whether we need to think more—and more deeply—about that question. The idea of using phages together with antibiotics is to have them work more effectively and quicker, with less resistance to the antibiotics. All of this comes into play when you think about a case where there is successful use of phages to control a mycobacterial infection. The other element to this, and people can regard this with different degrees of importance, is the direct use of phages that we have genetically engineered. It is a first, for whatever that is worth. It establishes a position and helps us to think about how one might extend that moving forward.

GEN: How long has it been since you started administering phages?

Hatfull: We started administering the phages in the middle of June [2018] and the patient is actually still receiving the phages. The pathology, and [chuckling] I should add that I am very much not a clinician, but we could show from the PET scans that the hazardous infection in the liver was essentially undetectable after about 6 weeks of phage treatment. The other part of the pathology is the skin nodules that are common to these infections. They have mostly resolved, but they are slower to resolve. That process is not yet complete, but we hope that it will come to resolution at some point.

GEN: Do you think that her immunosuppressive state may have served in her favor?

Hatfull: Possibly. Again, we are speculating here because we really do not know. But, plausibly.

We have looked to see if we could detect a neutralizing activity of the serum and have failed to find that. With an immunocompetent patient, whether the phages would remain active over such a long period of time is a question without a clear answer. So, you could make an argument that the fact that she is immunocompromised means that the phages need to be given over a long period of time, but it also means that you can do it over a long period of time. It is a double edge sword. Either way, you would prefer for the patient to be immunocompetent.

GEN: Have you met the patient or her parents?

Hatfull: Yeah. I met both the patient and her mum. There are two people in the lab, Bekah Dedrick and Carlos Guerrero, who did a lot of the nuts and bolts work in finding the phages, getting the mutants, doing the engineering, etc. Bekah and Carlos both went over to London when the phage was administered and actually got to meet the patient and the parents. They helped with a lot of the initial microbiology that was done in London with the first samples that were collected. Then I went over at a later time and met the patient and her mum. It was pretty cool. I am a scientist and a professor and I am sure that I must seem like a nerdy, lab rat, scientist type as opposed to the clinicians that they spend so much time with. I am sure that they think that we are kind of oddballs. But we do not work with patients. We work with test tubes. This was such an unusual opportunity to see if we could help a patient. To have that kind of impact, even if it is just for one patient, or one patient at a time, is pretty amazing. It is really off the radar from what we normally do. There is a broader question here about what the future looks like for phage therapy. And, in my mind, it is interesting but unresolved. Because there are two types of approaches. You have the personalized approach, like this one and the Patterson case, where you can identify phages that will affect that particular strain and that particular patient. The other approach is to design a pharmaceutical that would treat all of one type of infection, for example, all Pseudomonas infections, that would go through a clinical trial. The individualized cases are expensive in time and effort. But there is a rationale for understanding how they ought to work in many cases. So, that makes sense to me. For the broader application to treat a disease, rather than a patient, it is still a very complicated question. There is so much strain variation, and they are differentially susceptible to the phages. And, finding phages that will infect the vast majority of the strains that are out there is a conundrum. For example, in the randomized controlled double-blind clinical trial known as the PhagoBurn study published recently in The Lancet Infectious Diseases. Even though they used a cocktail of 12 phages, there was a subset of strains that appeared to be not susceptible to infection by the phages. This disease level control approach remains a difficult and unresolved question as to how well, and in what circumstances, this is going to be effective.


This interview was edited for clarity and length. To read more about the case study, please see the article “Phage Therapy Win: Mycobacterium Infection Halted.”