The pharmacokinetic consequences of human IgG1 expression in human FcRn Tg32 mice

Greg Christianson - The Jackson Laboratory

Talk Title: The pharmacokinetic consequences of human IgG1 expression in human FcRn Tg32 mice

Transcript:

Hello everyone, one of the purposes of today's talk is to introduce you to a new model that we've made and published last fall. It was in the same issue of MABS as Delphine's publication. And it's where we have produced a new model of tg32 that expresses a chimeric igg one antibody. And let me give you some background. You saw this in Delphine's presentation, the FcRn recycling pathway, and just to add more to that while in the recycling, the antibodies and albumin return to circulation via the atypical side of the cell the same pathway is used for transcytosis of IgG and albumin, where both of those molecules then gain access to surrounding tissues. And so FcRn facilitates bio distribution, and here's a replot of some Gayda from AB dish.

And it's looking at the Kd binding of the four isotypes of human IgG by SPR. And here is human FcRn and the KD values for the four isotypes and the pattern that results. And here's non-human primate where there's a higher affinity for human FcRn and a similar pattern. And they then looked at the wild type mouse FcRn the pattern is different and the affinity is even higher and here's some data from Avery. And in this study they had a panel of antibodies and they focused in on the non-human primate, the clearance levels of the set of antibodies.

And you can see that they rated them from lowest to highest clearence left to, right. And then when they compared that to a subset of those antibodies in humans, the clearence was similar and in wild type mice, the pattern was fairly erratic. So that model wild type mice are not a great model for predicting human antibody PK. So in Derry Roopenian's lab, we generated two transgenics the Tg 32, which in the mice, they lack mouse, FcRn and the transgene is driven by an indogenous human FcRn promoter. So the expression is tissue specific. And as a result, it provides the longest half-life of our models. And again, that tissue specificity makes it often desirable for modeling PK therapeutic antibodies, the other model Tg276, again, it lacks mouse FcRn expression, and that transgene is driven under the promoter of CAG.

And so the expression is ubiquitous, and we made that model wondering what we might be able to get in terms of results, because we wondered if ubiquitous expression might make it be an excellent model for human FcRn And the result was, this has lower, the TG276 has lower FcRn function. But what has turned out is that we've determined that it can play a role though in screening antibodies, that it's able to discriminate subtle PK differences, especially among hopefully matched fc variants and I'll show some data. So it can be best suited for selection of lead candidates from a large pool here's data from Avery where Tg 32 clearence is shown to outperform the wild type mouse. And it's compared in this publication versus the non-human primate. And you can see that both the hemizygous and homozygous do very well at correlating with the non-human primates and in Heraeus publication. It's half-life that is looked at and being compared to human in this case, a panel of antibodies, and it's 276, the Gg 276 is being evaluated. And it, in this set of antibodies does equally well to the non-human primate for matching or correlating with human half-life.

We wanted to do our own smaller study. We looked at both homozygous and heterozygous function for both Tg32 and Tg276. And we analyze three different therapeutic antibiodies belatacept was the first one. And you can see that the homozygous protects belatacept better than the hemizygous in both the TG 32 and TG276, and with no FcRn function the half-life is the lowest. We also looked at epi linear map. And again, you see the same sort of pattern where the homozygous is functions and protects better than the hemizygous in both trans genes and the lack of FcRn yields very low Half-Life.

And the last antibody that we looked at was pembrolizumab, and it yielded this nice dataset with very good or long half-life from the TG 32 homozygous and TG 32 hemizygous, and Tg276, homozygous are similar in the middle And the lowest protection is from the Tg276 hemizygous. So when we line up this dataset from lowest to highest in terms of half-life in the gray circle data, that's the human half-life data. That's at the top, you get this linear function, and you can see that the blue circles from the TG 32 homozygous are linear and look very similar in terms of their slope. And the data then goes down from there. But in general, we see from our Tg32 and 276, we do see an increase from left to right matching the human data while the wild type mouse does not have that pattern Ipilimumab is higher than pembrolizumab. So if we look at the correlation then of each of these in the upper left is the TG32 home, and it has excellent correlation with human half-life. And the values that are the correlations are very good, but decrease as you go down the panel of models that TG 32 hemizygous and the Tg276 homozygous and heterozygous, but when you compare them to the wild type data on the right, you can see they all perform much better.

And so I wanted to show you some data. This is demonstrating the utility of the TG276, although it doesn't protect as well as the Tg 32 just as I remembering some of Delphine's data where she had a wild type and maybe an LS mutant that were not different as if it had maximized the protection. So they didn't show any differences between those two antibodies, the mutant and the wild type. That's where the Tg276 can come into play in this version. We had variant one and variant two and they're closed and open circles. And the homozygous is on the left hemizygous is on the right. And the blue is TG 32 and the green is Tg276. And let's see if I can get Nope, I can't do it.

I'm just going to get my cursor going, but I'll describe it. You can see the blue plots at the top on the left in the homozygous TG 32, the variants don't really discriminate. Well, they've both maxed out that in vivo system and the TG 32 homozygous, and the same can be said for the TG 32 hemizygous, on the right, the blue symbols. But if you then go to the left and the homozygous Tg276, which is the green symbols, you can see that the variants do separate and that's because there isn't as efficient protection. So then if there's subtle differences, they will show up in this last maximize protection that the Tg276 has. And you see the same on the right and the hemizygous where there's a discrimination between variant one and two, which is the green symbols. So that's the advantage of using the Tg276 model if you have closely matched variance.

And now another model we have is the albem knockout. And the purpose of this model is that mouse albumin binds human FcRn that much greater affinity than human albem. And you can see this on the left panel, the open triangle you have TG 32 protecting human albumin and it's intermediate and it's protection. And that's because mouse albumin is competing very avidly. And yet in the closed squares which are at the top of the left plot, that's the TD 32 where the it's a mouse albumen knockout with no mouse, abdomen competing. You get very good protection and have more data here.

We did some studies and our service put human albumin into both the TG 32. And that's both knockout for albumin or not. And albumin in our hands had a half-life approximately 14 days while the TG 32 with mouse albumen being expressed has a half-life of about four days. And no FcRn protection yields about a one day half-life, we also looked at Tanzeum. And Tanzeum yielded a half-life of about nine days in the TG, 32 of your min knockout, while you have mouse albumin competing, it drives the elimination of human albumin. And the half-life is about one and a half days, and Tanzeum has only a 0.6 day half-life in when, FcRn is not functioning.

And there was a lot of discussion about ADA. And so there are models available for both TJ 32, and we are bringing a TG 276 immunodeficient model out. But here we can see that when Enbrel was put into these mice, they yielded a very similar data. And we've seen this over and over again that ADA is eliminated because the, either the skid mutation or the rag1 knockout won't allow antibodies to be produced. And so no ADA can occur. And yet the PK results are unaffected. And this was also published by Myzithras she used Humira or one of their own antibodies on the right and Humira on the left. And you can see that just as Delphine's data, you'll have dropout of the concentration that you can see in this case around day four or five, and on the right upper panel, you can see that monoclonal antibody X, you can see that there is dropout of the antibody concentrations of plasma concentrations between day four and 10 or so. And so in these situations, she had TG 32, as well as black skid. It TG32 skid as well as black skid. And you can see that complete datasets were produced both in the left and the right, demonstrating the utility of these models. And we also have the Tg32 albumin knockout skit because occasionally albium and conjugates, for example, can be immunogenic. And you can see here that both the TG 32 albumin knockout with, and without the skid mutation, yield comparable data for half-life.

So I'm excited to introduce you to our new model, our latest model. And this is the Tg 32 that expresses a Kymera, an IgG one antibody. And so Y is expression of indogenous human IgG import And looking at this goes back to the beginning of the concept of FcRn Bramble published back in 1964, he had observed other people's data that both in mouse and rabbit that as the level of IgG went up in the animals that the half-life went down and he plotted this out and he predicted that there was a receptor involved in the recycling of IgG and that receptor was saturate and that it would account for the data that hehad observed. And a few years later, 1970 Wildman published this same sort of effect in human data. He had human patients who had myelomas, so they had different IgG isotypes that were at very high concentrations in circulation, as well as patients who were hypo gamma, globulin, anemic. So they had very low IgG levels and saw this same pattern here. And so it demonstrated, again that there's a saturable receptor that's involved in the recycling, and it also demonstrates that how much antibody is there in circulation will impact how long the individual IgG or albumin entities are able to persist in circulation.

And so we did this study back in early two thousands, where we put different variants FC variants of trastuzumab into TG 32 mice. And we dose them at different levels of, of these antibodies either with a mouse or the human and I plotted this data out. And you can see that the I 253, a, the red plot, that's a mutant that is not well-protected or does not bind to give an FcRn. And you can see, as the mice were dosed at higher and higher levels of this, it did not impact the half-life of human IgG. One that's being plotted here. And yet wild type that was dosed in, or the affinity improved variance of the blue and green on the left plot, you can see that the higher the dose, the shorter, the half-life of the human IgG, one tracer molecule, and yet on the right panel, you can see that mouse IgG one, which does not have an affinity and is not protected by human FcRn its levels are not impacted by the amounts of these different antibodies being dosed into the mice. So it is FcRn specific.

And so for our to make a mouse, a TG 32 mouse that expresses a Comeric human IgG, one molecule, the human IgG one locus was transferred into the TG 32 mouse. And the result was an antibody schematic here where the variable region was mouse. And the rest of the antibody is human. What we did was we immunized the mice with DMP KLH and Franz agiment and naive TG 32 HFC mice had 600 micrograms of Chimeric IgG one in circulation, while the immunized mice had 4.2 milligrams per mil of the Chimeric antibody. And yet the mouse IgG one was lacking in either the naive or immunized TG 32 HFC mice, demonstrating that the mouse Ighg-1 gene locus had been replaced, and the mice that were immunized produced antibodies that work specific to DNP.

And here we look at all the isotypes of mouse IgG, excuse me. Yes, mouse IgG. We have IgM on the left two, a two B and IGA on the right. And you can see that for the TG32 HFC naive, or TG 32 naive, or the FcRn knockout, all three, there was no significant difference in the levels for each of these isotypes. And only when the Tg32 HFC was immunized were all of these isotypes elevated. So here we show the impact of the Ig competition, IgG competition on trastuzumab PK. And we, the mice that had been immunized as well as naive TJ 32 and naive TG, 32 HFC. And we as a positive control, we pre-loaded TG 32 mice with polyclonal human IgG antibody, and you can see the plasma levels on the left, the percent remaining and the half lives on the right of trastuzumab with no competition. The half-life was around nine days. And with naive levels in the TG32 HFC mouse, the levels of trastuzumab or the half-life of trastuzumab was reduced somewhat and more greatly reduced when immunized and similar to the half-life was reduced similar to the TG 32 mice that were preloaded with IgG.

And we also put a human, another human IgG 1 antibody into these mice. It it's a specificity is licisime. So it was easily tracked within the mice, along with the indogenous chimeric antibody or the mice that were preloaded with polyclonal human IgG. And you see a similar pattern where TG 32 naive mice looking at the half-life there, the orange bar, and they yeilded at about 11 day 11 and a half day half-life. And that was reduced with both the naive levels of Crimea KGG one, and the higher levels induced by immunization in the TG, 32 HFC mice, and pre-loading polyclonal antibodies, polyclonal human IgG lowered the half-life even more so in summary, the chimeric IgG1 mice, and excuse me, chimeric IgG1 in these mice has replaced the mouse IgG one the chimeric IgG1 one can be induced to respond to immunization.

And the other mouse IgG isotypes are all at normal levels. So there's no trans gene of fact expressing their expression, trastuzuman and HuLys 11 half-life demonstrate naive levels of chimeric IgG, one compete for FcRn protection and the elevated Comeric IgG levels, augment FcRn competition, and the TG 32 HFC advantage it more closely models, humans due to competition because in the TG 32 model, they are a new incompetent. So they have mouse, IgG and circulation, but again, mouse IgG has no affinity for giving an FcRn. So there's no level of competition that exists in humans. So future studies will determine if the TG 32 HFC yields more accurate PK than the TG 32, or the Tg276. And this model can also be used to generate therapeutic monoclonal antibodies, should someone want to investigate this possibility. So that ends my talk and I'm open for questions.