Authors: Jennifer D. Stone*, David H. Aggen, Hans Schreiber, and David M. Kranz
Using a patients immune system to target cancer and maintain long-term surveillance is an attractive goal. Although T cells can respond potently to tumors, immune tolerance mechanisms often result in deletion or inactivation of those T cells expressing specific T cell receptors (TCRs) against effective target epitopes. To overcome this problem, two approaches using antibodies against cancer-associated epitopes have been designed to re-direct T cell activity. In one approach, the binding domain (scFv) of an anti-cancer antibody is linked to intracellular signaling domains, yielding a chimeric antigen receptor (CAR) that is expressed on ex vivo-activated T cells for adoptive therapy. In the second approach, the anti-cancer scFv is fused to another scFv that binds the TCR/CD3 complex, yielding an injectable product called a bispecific T cell engager (BiTE). Here, we directly compared the sensitivity of each strategy directed against a unique tumor epitope on the transmembrane glycoprotein OTS8. While both approaches showed efficient anti-tumor activities, CAR-targeted T cells were consistently more sensitive than BiTE-targeted T cells to low numbers of antigens per target cell. Using a panel of targets with varying levels of OTS8, we have constructed a sensitivity scale, providing a guide for applying these approaches under various conditions.
I'd like to thank the organizers again.
This has been a very unique and pleasurable meeting.
I am Jennifer Stone, as was said,
and I'd like to talk about a couple
cancer immunotherapeutic strategies today
and how they compare to one another.
Specifically, chimeric antigen receptors
and bispecific T cell engagers.
So we can look at cancer as sort of a failure
of the adaptive immune response
to surveil properly for mutated cells.
So in that sense, the immune system malfunction
can lead to cancer.
We know that in the adaptive immune system,
T cells can frequently respond potently to tumors,
but similarly frequently they don't.
And that's because the normal T cell receptor
peptide MHC interaction is susceptible
to a lot of different tolerance mechanisms.
There's central tolerance,
because a lot of these tumors epitopes
are very similar to self.
There's also immunosuppression effects in a tumor
and possibly dysfunctional antigen processing
and presentation,
causing a lot of these MHC peptide complexes
to be downregulated or not present the proper peptides
to allow T cells to recognize them.
However, it would be very desirable
to have these T cells actually respond.
So how can we retarget T cells
to effectively fight tumors?
And I'm gonna talk about two strategies today.
The first is a chimeric antigen receptor,
which consists of a targeting moiety or binding moiety,
which is fused to an intracellular signaling domain
all in one chain.
And this can recognize a tumor associated antigen or TAA,
which could be any protein on the surface of a tumor cell,
including a peptide MHC complex,
as you might have heard earlier from Dave Agin.
The other strategy I'd like to talk about
is called the bispecific T cell engager or a BITE.
And this is a soluble bivalent T cell engager
that binds to an invariant portion
of the T cell receptor complex
and crosslinks it to a tumor associated antigen.
So I'll go a little more in depth into these two strategies.
The first strategy, CARs or chimeric antigen receptors,
as I said, consists of a binding moiety,
which is frequently a single chain FV
or the binding portion of an antibody
and is fused to these intracellular signaling domains,
which can consist of part of the T cell receptor complex
and lots of co-stimulatory signals.
And the nice thing about this strategy
is we have evidence that it works.
In recent clinical trial success,
Carl June and his colleagues were able to treat
and cure a small number of refractory human leukemias
with long-lived anti-CD19 CAR transduced cells.
And CD19 is a differentiation antigen present on B cells,
both cancerous and regular healthy ones.
The second strategy I'd like to talk about is the BITE.
And this is just a little scheme of the BITE.
This would be a single chain FV
against a tumor associated antigen,
which is covalently linked to a single chain FV
against the T cell receptor complex.
And the nice thing about this is it also works.
There was recent clinical success
for a BITE called Blinatumumab.
And this was marketed by Micromet,
which was recently bought by Amgen.
And in a very small trial as well,
100% objective responses were seen
at 1,000-fold lower mass doses than rituximab,
which is sort of the gold standard antibody treatment
for B cell leukemias and lymphomas.
And 16 of 21 of their patients became totally negative
for minimal residual disease.
And that's using this as a single agent.
So I'm going to be comparing the sensitivity
of these two strategies against a single tumor epitope.
And I'll talk a little bit about this epitope
and where it comes from.
So what we know is tumors present immunogenic antigens.
And there have been lots of conserved
tumor-associated antigens identified
in a variety of different categories.
I'll just point out CD19 and CD20
are differentiation antigens that are present on B cells
for B cell leukemias and lymphomas.
Overexpressed antigens here are antigens
that are normally expressed in healthy tissues
but are dramatically upregulated in some tumors.
And specifically HER2, I'd like to mention.
But if we could get our ideal tumor epitope,
what it would be is an epitope that's commonly expressed
on a lot of different cancers
and yet is exclusively found on tumors
and not anywhere in healthy tissue,
embryonic, cancer testes, anywhere else.
And if we had such an epitope,
we would want to use a very sensitive specific strategy
to redirect T cell responses to it.
We have a candidate epitope for this ideal scenario.
237 is an antibody that we're using
as our targeting molecule.
It recognizes a glycopeptide,
which is part of a larger transmembrane glycoprotein.
And this is just the crystal structure
of the single-chain FV binding to the peptide and the sugar.
And the affinity for this glycopeptide is 140 nanomolar.
What we're recognizing is actually
an unusual glycosylation, a TN-like antigen,
in the context of this OTS8 protein.
And OTS8 is a protein that's upregulated
on many tumor types.
And it's thought to correlate with invasiveness.
So this aberrant glycosylation results
from a mutant cosmic chaperone.
So it's gotta have both the OTS8 protein
and the cosmic chaperone mutation
in order for this epitope to be revealed.
This epitope was first found
in this murine fibrosarcoma, AG104A.
And if wild-type cosmic chaperone
is transduced into that cell line,
the resulting cell line, A cosmic, is 237 negative.
So you can see, nice bright staining here
and basically nothing here.
So what we were able to benefit from
is a panel of cells with different levels of 237 epitope.
So you can see here I've got six cell lines
characterized for their 237 epitope staining.
AG104A is the cell line I already mentioned
with the cosmic rescue mutant.
Neuro2A is a cosmic mutant cell line
that does not express OTS8.
But if you transduce OTS8 in,
it becomes very positive for this epitope.
And Jurkat is a human T cell transformed line
that is also cosmic mutated.
And when we transduce OTS8 in,
we get a higher level of 237 staining.
If you look at this plot here
with a titration of 237 staining,
you can see that most of these OTS8 cosmic mutant
cell lines titrate out at about the same level.
But Jurkat, this line here, which is mutated for cosmic
and does not express OTS8,
seems to have a cross-reactive epitope
of a different affinity.
This is another important consideration
that you may have cross-reactive epitopes in the body.
So we were able to quantitate these levels
by comparison to calibrated beads.
And we've got tens of thousands per cell
for Neuro2A and AG104A,
just under 10,000 for this Jurkat with OTS8.
And this cross-reactive epitope
quantitates out to just under 1,000.
So now I'll go through the two strategies
and how we constructed them.
First, we have the chimeric antigen receptor
against the 237 epitope.
And this is actually constructed by Dave Agin,
who spoke earlier today.
And we've got the 237 single-chain FV
linked to intercellular signaling domains.
And this is cloned into a retroviral vector
and transduced into purified ex vivo primary cells.
So the cells are conditioned and transduced in vitro
and could then be returned into a patient.
And we get very high transduction efficiencies
in both CD4 and CD8 cells.
And by testing, we were able to show
that specific responses were seen to 237 positive cells,
but not to 237 negative cells.
So for comparison, we also constructed
the bispecific T cell engager to this very same epitope.
And here we have 2C11, which is an anti-CD3,
so anti-T cell receptor complex antibody.
The single-chain FV of that
linked to the single-chain FV of 237.
And this protein was purified from mammalian cells,
and we get nice pure protein and high yield.
If we test just the anti-CD3 portion by coating on a plate,
we see that the anti-CD3 single-chain FV
stimulates very similarly to the full antibody.
And if we then put it in with T cells
with an irrelevant T cell receptor,
we can take those T cells
and make them lyse 237 positive cells specifically.
So this strategy also seems to be working.
Finally, I'd like to compare directly side-by-side
these two strategies.
So here I have a slide of cytokine release
induced by 237 targeting strategies.
And here we're looking at just AG104A and ACOSMIC,
so the wild type and the cosmic rescue mutant.
On the left, I have T cells targeted by 237 CAR,
and on the right, 237 BITE.
And then for each, we have CD8 and CD4 tested.
So what you can see is we have specific responses
all the way across this panel,
but we have decreasing sensitivity
as we go from CAR to BITE and from CD8 to CD4.
If we expand this to the full panel of cells,
we see the same trend across the rows,
but we also see a trend of decreasing sensitivity
as we go from high antigen levels per cell
to lower antigen levels per cell.
However, if we look at a different response,
and instead of cytokine release, we look at target lysis,
I'd like to point out this Jurkat cell line,
which had the secondary cross-reactive epitope.
And now, even though this is OTS8 negative,
we get very efficient lysis
by especially 237 CAR redirected T cells,
and even a little bit by BITE T cells.
So this epitope does cause responses
from both targeting strategies.
So given all of this data,
we were able to construct a sensitivity scale
for targeting this epitope via CAR versus BITE.
And this is just a summary of what we've seen before.
Cytokine release is less sensitive than target lysis,
CD8 is more sensitive than CD4,
and CAR is more sensitive than BITE.
And each bar is stretching from high antigen levels
down to the minimum antigen level per cell
we expect to see the response.
If we look at other tumor-associated epitopes
on this same sensitivity scale,
we can see CD19, which is the epitope
that is being targeted by both the clinical CAR
and BITE strategies I mentioned earlier.
And you can see it would be very efficient for target lysis,
but you might not get quite the cytokine release
from CD4 or CD8 BITEs that you would from the CARs.
I'd also like to mention HER2.
HER2 is an upregulation antigen for breast cancer.
And here we have the levels on normal breast,
and you can get up to a 40-fold increase of level
on about 15 to 30% of tumors.
This normal level is already within the sensitivity range
of both CAR and BITE.
So it's very important to know this sort of level
before you approach a strategy like this.
And in a very unfortunate incident,
HER2-directed CAR T-cells were inserted into a patient,
resulting in very rapid lethal effects
due to expression of HER2 on the lung.
So in conclusion, we were able to have T-cell targeting
to the 237 epitope, which is specific via CAR or BITE.
Stimulation of transduced T-cells with 237 CAR
is more sensitive than stimulation mediated by the BITE,
and that was across the board.
And then CD4 T-cells were active
in the same assays as CD8 T-cells,
although they were less sensitive also across the board.
And the choice of CAR or BITE will have to depend
on the antigen density,
specificity of the epitope expression,
and of course other factors such as ease of delivery
and manufacture, et cetera.
I'd like to acknowledge the members of the Kranz Lab,
especially Dave Agin, who constructed the 237 CAR.
Our collaborator, Andrea Schoedinger,
who was the characterizer of the 237 epitope
and made several of these 237 target cell lines.
Our ongoing collaboration with the Schreiber Lab
at the University of Chicago,
helping us move in vivo
with some of these targeting strategies,
and our funding from the Cancer Research Institute
and the NIH.
Thank you very much, and I'll take any questions.
I have a quick question.
Okay.
I don't really see the difference
between CD4 and CD8 population
and how they're the less sensitive
to the CD4 population.
Do you know what mechanism causes that?
We're not really sure why they're less sensitive
in this case, since these are not
co-receptor dependent reactions.
But we did see this in all of the cases that we tested,
that both target lysis and cytokine secretion
were less potent from the CD4 T-cells.
And that could just be based on
the way that they function normally.
All right.
I think let's thank our speaker once more,
and then we'll bring up the next student.
Thank you.