by Dr. Shaohua Zhao
DR. ZHAO: Good afternoon. I know this is the last talk for today. I will keep it short and speak as fast as I can.
I will talk about the molecular characterization of --- isolates. I think previously our team and Dr. Paula Cray has talked about the --- isolates on the human side, as well as the animal side. There are many aspects --- as the interest area.
We are focusing on the three areas. On molecular subtyping, detection and identified of antimicrobial resistance genes and study the mechanisms of antibiotic resistance and resistance gene transfer.
I would like to give you even more about the molecular subtyping work at CVM, but before I do that I will just briefly talk about the history --- identification of antimicrobial resistance and mechanisms of resistance.
We have used PCR and followed it by DNA sequence analysis to detect and identify the antimicrobial resistant gene. We are particularly interested in the --- integron --- resistant mechanism. Integron is a mobile --- plays a significant role to spread the antimicrobial resistant gene among the gram lactam bacteria.
So we use PCR to detect the integron, then we use a DNA sequence analysis to identify what kind of specific gene is in the gene cassette. So we have identified many resistance genes that are associated with resistance to beta-lactam, aminoglycosides, trimethoprin, chloramphenicol, and sulphonamides.
We also identified the many beta lactam resistance in genes including the members of the blaCMY gene. As the previous talker has mentioned, the blaCMY gene is so important to public interest because of the cause --- third generation --- drug of choice to treat --- Paula talked about a microarray.
Right now at CVM we are studying the development of microarray to detect the antimicrobial resistance gene. Microarray has a greater capacity. They can simultaneously detect 100s of 1000s of genes in a single DNA chip. We have selected more than 100 resistance genes and virulence genes and selected 75 --- on each gene put on the array.
So, once this project that --- increase all capacity to detect the antimicrobial resistance gene.
And the studies of mechanisms of antibiotic resistance. For certain resistance profile we like to find out what is the mechanism of that. We are particularly interested to identify the resistance. You know, we are they located at. Is it a plasmid or chromosome? If it is a plasmid, can it transfer to a new different strain or species by conjugation or transformation?
So we have identified many plasmid-borne integron medicated resistance genes transferred by conjugation and transformation. We also clone and express the blaCMY gene into the new host to see if they are resistant to capacity. We find out that the blaCMY gene can resist to a 1st, 2nd and 3rd generation of cephalosporins.
Not just one particular antibiotics. Actually, there is a whole bunch of antibiotics that they can cause resistance.
We are also interested in -- you know, for fluoroquinolone resistance mechanism. As you know, fluoroquinolone resistance is caused by the --- gene, the --- mutation. So we are the PCR and follow the DNA sequence analysis to identify where the mutation occur. If it was a single mutation or a double mutation or the factors there for fluoroquinolone.
Dr. White and McDonough has a collaboration with the University of --- by gene knock-out mutagenesis. They have identified several efflux pumps, such as acrA, acrB and tolC. If highly expressing those efflux pumps, they can cause resistance to tetracycline and chloramphenicol.
Also, expressing high levels of those efflux pumps will --- mutation that can cause highly resistance to ciprofloxacin -- dramatically increased.
Now I would like to talk a little bit about molecular subtyping. The goal of molecular subtyping is to determine genetic relationships among Salmonella and Campylobacter obtained from NARMS isolates, identify emerging multi drug resistance serotypes or strain that is associated with human diseases and share the DNA fingerprinting data with other public health agencies through PulseNet.
The technique we have been using at CVM include Pulsed Field Gel Electrophoresis, multi local sequence, plasmid profile and repetitive PCR and ribotyping. We compare those methods in terms of the discriminatory power, reproducibility, cost and PFGE comes out as still the best method.
These methods is standardized by PulseNet. As previously talked this morning about the best methods, I would to just briefly talk about PulseNet. PulseNet is the National Molecular Subtyping Network for Foodborne Disease Surveillance. It is a collaborative between CDC, the Public Health Laboratory and USDA and FDA.
The goal of this program is to reduce the burden of foodborne illness by assisting investigations and improving outbreak detection through the rapid linking of cases by DNA fingerprinting patterns comparison and leading to faster intervention and establishment of control measure.
Since PulseNet was established, according to statistic models, the size of the outbreak has significantly reduced. Two thirds of cases could have been prevented because of the PulseNet, because when you find out the source you can recall the product or the meat from the market.
Right now PulseNet in the USA have over 17 laboratories participating. It includes all the 50 state public health laboratories. Some of the bigger states, like California or Texas, they have county or the city laboratories participating. We have two USDA and eight FDA laboratories, including six --- and CVM.
This is a database from CVM. We have over 4,000 database entries. Salmonella is the biggest database. We have over 2,700 data entry. Out of that about 1016 isolates are from NARMS isolates. That includes 1999, and in 2000 Paula sent us about 600 --- for us to type.
And then we have Campylobacter, 1,3000 isolates and 761 is from NARMS retail meat. All of this data was submitted to the PulseNet. We have E. coli and Vibrio. Those are mostly from internal responsibility activity.
Now, I would like to share with you some PFGE profiles for multi drug --- Typhimurium. This is actually the PFGE DNA fingerprint pattern. This is a denogram generated by software --- based on the DNA fingerprinting pattern.
If there is a streak line here, that means they have a distinguishing pattern. That is 100 percent pattern similarity. Here is an antimicrobial resistance profile. Each black square shows the resistance to particular antimicrobials.
This is the CVM number, from different state and the source and the isolation date. We do not have -- put the FDA’s and CVM’S PFGE pattern because we compare the database with PulseNet. So this pattern number is from CDC.
This comparison here is not to say this particular -- this isolate is associated with outbreak. It is just to say in particular that this pattern is not associated with human disease because there is no epidemiology information to link it together. We just want to say particular --- cause human disease.
Okay. Look at this. This is a chicken isolate, isolates from New York in 2003. We also have two human isolates from New York that show identical pattern. Again, because of state labs -- some state labs do not provide the antimicrobial data here. So we just leave it empty, but that does not mean --- particular sensitive strain.
There is another cluster here. This is another identical pattern from the ground turkey. We also have human isolates. This cluster here is a typical DT104 pattern here because they are resistant to ampicillin, chloramphenicol, cipromycin --- and tetracycline. We have this isolates isolated from chicken breast, pork, chicken breast again and pork. They all have human isolates and show identical patterns here.
Okay. Some random Newport -- you know, everybody has some mentioned it in the previous talks. It is emerging in multi drug resistance, the serotype. In our retail meat isolates 42 percent of them are resistant to nine or more antimicrobial. So it is quite a resistance compared to Typhimurium.
Again, we have pork isolates, ground beef isolates, ground turkey isolates. In each of these patterns we have human isolates which match this pattern.
This is a PFGE profile of Campylobacter jejuni from Iowa. I think Terry has talked about the Iowa Retail Meat Study. So we did it by --- because what happened with Campylobacter is sometimes it was -- when inside the discriminatory power is limited. So we used the second --- to confirm the commonality. So you can say we have --- there is chicken isolates, we have human isolates that show the same pattern and also turkey isolates and chicken isolates shows the PFGE pattern. All it is from the Human Iowa Retail Meat.
So, in summary of our study, molecular studies on mechanisms of resistance and resistance transfer have provided valuable information for better understanding the emerging and dissemination of antimicrobial resistance in bacterial pathogens. Our data shows that certain multi drug resistance to Salmonella and Campylobacter were frequently present in the retail foods and also recovered from human patients.
For the success of the NARMS program to collaborate on molecular characterization I think there are three important elements. First of all, isolates and data sharing is so critical to determine the emergence of antimicrobial resistance of foodborne pathogens. We really hope eventually we all do the PFGE and we somehow link this data together.
Like Tom this morning mentioned, hopefully the three sites and the data can be put on the front page to say, okay, this is the emerging serotype. What is the commonality on the three. Hopefully, we can some day be able to do that comparison.
Of course, we have to use standard methods to make it comparable our subtyping a DNA fingerprinting pattern.
Lastly is linking the molecular subtyping to other epidemiological data to provide valuable information on origin and transmission of multi drug resistant pathogens. Thank you.
DR. ALTEKRUSE: My understanding is that all of the isolates of the three arms are going into the PulseNet now? Is that correct?
DR. ZHAO: No. According to this morning, Tom or Tim said that --- isolates or maybe it is already in the PulseNet, but that is not identified because the number was not matching. Is that correct?
DR. CHILLER: Yes. Basically, from -- traditionally, NARMS has not requested surveillance isolates to be pulsed, because remember we are getting one out of every 20 isolates. The majority of the isolates that are in our public health lab are sporadic. So maybe 20 percent outbreak, eight percent sporadic.
We are getting majority sporadic cases and those come into us and no one has ever asked state labs to pulse those specific isolates before.
As Tim mentioned, what probably has been happening is a lot of times states pulse their isolates anyway. I mean, Minnesota pulses everything. So the data is actually in there, but there have been discrepancies of the I.D. number that was sent NARMS versus what sent to PulseNet.
We are sort of moving forward on two fronts. The first front is we have sort of a national campaign to standardize I.D. numbers from states. Any isolate that comes to CDC for any reason, if we could use one I.D. number, whether it is for susceptibility testing, pulsed field, MLST, whatever, if we could use one I.D. number, then we could link to all that data.
DR. ZHAO: But CVM isolates --- every single Salmonella and the Campylobacter --- PulseNet.
DR. ALTEKRUSE: And then my question -- the other part of it is the FSIS slaughter HACCP isolates. Those are going into VetNet. Correct? Which is a component of PulseNet. So, is it now possible to do a query?
DR. FEDORKA-CRAY: Not yet. But there will be when we get our database.
DR. ALTEKRUSE: Things are -- it started?
DR. FEDORKA-CRAY: Yes. We don’t have a database up yet. I mean, we have a hard time even querying our own stuff right now in that database. When it is up, then we will be able to use -- there will be cross talks between the systems.
DR. ALTEKRUSE: I just want to agree with something you said, and that is that the real value of NARMS will increase tremendously when that data sharing of molecular information is possible.
DR. ZHAO: Absolutely.
DR. FEDORKA-CRAY: Right. We discussed with Swami -- we have been discussing this for three years now. The idea to keep it separately was based mostly on a personnel and a funding issue. Swami doesn’t have the personnel, nor does he have any monies to deal with animal isolates, per se.
So this was a way to get it up and going and have somebody else worry about a database and keeping it going, and they are going back and redoing parts of their database now and trying to change these numbers over and renaming some old patterns because they were done in a different way, where our database gets to start clean.
Again, it is like you want to improve the wheel instead of reinvent the wheel, and that is where the VetNet system comes in.
DR. ZHAO: You know, Paula, my understanding is that eventually your VetNet will connect with the PulseNet. Right?
DR. FEDORKA-CRAY: I’m sorry? What?
DR. ZHAO: Will connect with PulseNet.
DR. FEDORKA-CRAY: Yes.
DR. ZHAO: And we have access to compare -- to access your database in other words?
DR. FEDORKA-CRAY: Sure.
DR. ZHAO: So we can --- right now we can just easily compare our data to PulseNet, but the --- we can compare all data to your data and you can compare it,
DR. FEDORKA-CRAY: Yes. That is the idea. We would all be able to look at each other’s databases. Right now we don’t have access to CDC’s database. At some point in time we will have access to their database to be able to query theirs, just like a PulseNet site. They will be able to query ours. Hopefully we will be able to query yours, you will be able to query ours and PulseNet’s too. So that would be the idea. An integrated system.
DR. ZHAO: I think that is very important. When you have the capability to compare the three arms, you can make sense what is commonality, what is human, what is animal and what is retail food and that would make it meaningful.
DR. McEWEN: (Microphone not turned on.) I wonder how useful it is to know that. My understanding is that PulseNet is good for helping you identify --- local or geographically disbursed. The further you get from Atlanta it seems people are less enamored with PFGE and ask questions about how discriminatory it is I guess or how peaceful --- genetics.
I am wondering how well all of the efforts going into --- we have talked a bit about looking at some similarities between animals and people and that sort of thing, but I wondered if ---
DR. BARRETT: I think the critical issue in understanding how this will be helpful is to think in terms of clustering instead of matching. When we are talking about an outbreak, you know, is this strain from this person the same as this one, we want them to match. But I don’t think we are going to see a lot of that in animal and meat and human isolates.
But I think we will see clustering where you will see very clearly -- the Kentucky is a good example. The chicken Salmonella Kentucky isolates cluster separately from the human isolates. I would say that is a pretty strong indication that those human isolates are probably not coming from chickens.
I think we will be able to make those sorts of observations by seeing these cluster more tightly; this group clusters more tightly than that group. You know, they are most closely related. There is probably a link here.
As far as matching, there will be some, like you have seen, but I don’t think that is the strength of it. I don’t think that is what we want to look at.
DR. McEWEN: (Microphone not turned on.) -- how you define the clusters? I don’t know --- distribution and --
DR. BARRETT: Yes. I would say that is true. I think that it is sort of an early area. That is what we will do. You know, at two percent, boy, I don’t see anything here. They are just randomly scattered. Let’s go to five percent. What do we see?
To some extent the only way you can do that is to have the information already that you would like to have to validate it, and I am not sure, with NARMS, where that will come from. I think in other avenues of PulseNet we are able to do that and perhaps we can transfer that to NARMS.
DR. ZHAO: I think a lot of good example is Salmonella Newport. In most of the multi drug resistance to Newport, compared with the CDC database, most of them were cattle. But if you look at the susceptible strain for the human, --- isolates from the turkey. You know what I’m saying here?
So you can say it is a multi drug resistance from animal through the food and to the human. You know, comes from which animal? The PFGE will provide the information of this linkage together.
DR. McEWEN: (Microphone not turned on.) So, you think it may be helpful in --- of exposure to different species ---
DR. ZHAO: Yes. Not for all, but for certain serotypes. Not for all, but for certain serotypes. Also, I think that the PFGE --- has a very important --- to say certain pattern only from certain store or from certain brand of the type of meat. So maybe if you put those information together maybe you can provide us to say that this MDR is due to the retail store cross contamination or from slaughter or originally from animal.
So, if you put the PFGE with the --- pattern, plus this other information, maybe we can find out where the MDR come from there.
DR. KOTARSKI: I wondered. When you find matches between humans and a certain retail store or if it is an animal species or whatever, how many matches become compelling evidence that, for example, all of the Newport was in humans that was susceptible is coming actually from turkey?
How many matches do you see to come forward with that hypothesis or come forward even with a conclusion?
DR. ZHAO: Well, we just see a phenomena --- how many of them are --- turkey isolates are most susceptible compared with the cattle isolates. When you compare the -- look at the one particular pattern and compare with the --- database, then you have a match. --- pattern.
But this has to be maybe to compile more data --- ground beef we will --- phenomenon.
DR. KOTARSKI: So, it allows hypothesis generation conclusions?
DR. ZHAO: Yeah.
DR. WHITE: And I think too, Sue, there is no magic number. There is not like five or three. I mean, the more data we generate, the larger our associations become or not become too.
DR. ALTEKRUSE: What becomes compelling? Is it finding it in food from a patient? It seems like you are saying the denograms are exploratory, but it is not about causation. It is not even really associations. It is just looking at a pattern.
DR. ZHAO: Yeah. For this comparison, not necessarily this particular isolate are associated with --- epidemiology of information is much more critical to say this outbreak is caused by a particular strain. You have to have where the human --- they have a time, a place or have that information linked together to make that --- linkage or calculation.
DR. WHITE: I think the best case scenario would be if we could be real time with pulsed field and from the FoodNet sites, active disease surveillance, human disease and the retail from those same sites, and we could do pulsed field immediately and submit those patterns. Then we would have a much shorter time frame for comparison.
Right now we are submitting patterns a year to two years later. I mean, you can’t do any type of analysis, except it looks like it came from this.
DR. ALTEKRUSE: So, when will you all have that in place?
DR. WHITE: Remember, we have a budget of so much money.
DR. FEDORKA-CRAY: We are giving FSIS our in two weeks.
DR. ZHAO: --- for our --- and we are not participating --- because we have a web --- communication every day. So if we have a match mostly we feel guilty --- they are asking for last month’s --- pattern. We have the same pattern, but it is last year.
DR. WHITE: I think the last thing, as Tim mentioned as well, the more data we generate, the more we start to see maybe particular associations with particular patterns with particular foods or particular animals over time. So, the more we generate that and see these differences -- say like between like the Kentuckys with chicken breast.
They are very different from the Kentuckys you see from ground beef or swine, and you start to see these outbreaks of human disease, we will be able to say, well, we can steer the public health officials in that direction possibly. We are not going to come out and say it is chicken.
DR. ZHAO: Yeah. That is a really good example. We have a high frequency isolate from the animal arm and the retail meats, but they have very low incidence on the human side. But if you look at the PFGE profile, they are very different.
DR. ALTEKRUSE: Shaohua, with the case of Campylobacter, for example, it seems like there are so many different laboratories that have different preferences for how they subtype their Campylobacter. How do you arrive at a preferred method?
DR. ZHAO: Well, CDC’s --- already. But our experience with Campylobacter, the second --- is very --- because the discriminatory pattern for Campy. One enzyme is not good enough. So we find out that you use two --- for Campy.
DR. ALTEKRUSE: But, for example, in Europe I believe they use amplified polymorphism.
DR. ZHAO: Yeah. You know, but those methods I think -- first of all, it is not standardized and it --- problems. So I think the PulseNet --- so we can make the data comparison, otherwise you compare apples to oranges ---
DR. ALTEKRUSE: But even in the United States, for example, there are some people who have been very strong advocates of FLA sequencing. I don’t know. That is my question. How do you arrive at a pathogen specific method?
DR. ZHAO: Well, CDC -- you know, PulseNet has a --- R&D group where they --- methodology and look at more distinct power methods.
For example, even PFGE’s golden standard for --- is not powerful enough so that --- additional methods are --- a certain serotype or other methods it is better than PFGE, and therefore, it can be --- they are also looking for other methods.
But for the time being, I think the PFGE is the gold standard, you know, --- for Campy.
DR. ALTEKRUSE: Thank you.
DR. FEDORKA-CRAY: I think that is one of the things that I wanted to point out. If you talk to Swami, he sits there and doubts that PFGE will be used by PulseNet in five years, but there will be an evolution of methods and there will be other methods.
Tim is working on other methods. Our labs are working on other methods and there will be others. I think microarray technology might be something that comes on or comes up faster, and there will be more standardized methods over time.
So, even Swami doesn’t think that PFGE will stay as the gold standard over time and that there will be other things developed, other techniques refined and made more standard.
DR. ZHAO: Yeah. This is a concern because the PulseNet --- deal with over --- state labs --- in terms of the costs, that is another important concern besides the methods --- for the time being --- it is one of the best methods I have seen.
DR. BARRETT: Over the years I have been accused many times of being PFGE-centric. I really don’t like it, PFGE. The bottom line is that it has worked out the best for what it has been used for, and it continues to.
DR. ZHAO: Yeah. I just went to the Canada --- Study. Actually, Bob --- talk about the molecular subtyping and compare --- PFGE is their number one. You know, there is no question about it. You know, --- and the other countries and whatever, and everybody comes to the conclusion, for the time being, that PFGE is their number one.
DR. YOUNGMAN: Okay. Thank you very much, Shaohua.
I guess this concludes things for today. I am sure some of you will want to talk afterward, but I am aware of the time and feel that we need to close for today.
We would like to reconvene tomorrow morning at 8:30, and we will start by talking about a recap and then we will also be talking about international issues related to NARMS.
Thank you very much to all of you for coming today, and we will see you tomorrow morning at 8:30.
(Whereupon, at 4:54 p.m., the hearing was recessed, to reconvene Friday, June 24, 2005, at 8:30 a.m.)