Targeting the checkpoint molecule CTLA-4 using antibodies has been relatively successful as a cancer immunotherapy, however, clinical benefits often come with severe and sometimes dose-limiting immune-related adverse events (irAEs). In a paper recently published in Cell Research, Zhang, Du, and Liu et al. investigated why different CTLA-4 targeting antibodies induce different degrees of both efficacy and irAEs, and developed a strategy that may improve both the safety and efficacy of anti-CTLA-4 antibodies.
While the exact mechanism of CTLA-4-targeted antibodies like ipilimumab and an IgG1 variant of tremelimumab (TremeIgG1) is still ambiguous, they likely work in part by blocking CTLA-4 from interacting with CD80/CD86, thereby allowing CD28 co-stimulation by these ligands in TCR-activated T cells. The same antibodies may also work in part by binding to CTLA-4 en masse on Tregs, flagging them for depletion via antibody-dependent cellular cytotoxicity (ADCC). While these particular antibodies can be effective in inducing immune responses to cancer, they are also prone to inducing serious irAEs. By contrast, some other anti-CTLA-4 antibodies, like HL12 and HL32 (previously identified by the authors), are less prone to inducing irAEs. Investigating this distinction, Zhang, Du, and Liu et al. noticed an interesting correlation: irAE-prone anti-CTLA-4 antibodies led to significant downregulation of CTLA-4 on target cells, while non-irAE-prone antibodies did not.
Further investigating the difference in CTLA-4 downregulation between different antibodies, the researchers found that all four anti-CTLA-4 antibodies localized with CTLA-4, leading to rapid internalization of the complex, but that only ipilimumab and TremeIgG1 subsequently drove CTLA-4 to the lysosomes. Based on prior evidence that pH drops significantly as endocytosed antibody-target complexes transition into lysosomes, the researchers suspected that pH sensitivity of the different antibodies might affect the fate of CTLA-4. They found that the capacity of ipilimumab and TremeIgG1 to bind to CTLA-4 was largely unaffected by pH within a range of 4-7. Meanwhile, HL12 and HL32 were highly sensitive to pH and progressively lost their binding capacity and dissociated from CTLA-4 once the pH dropped below 6.5 – the pH of early endosomes. Interestingly, the dissociation of HL12 and HL32 from CTLA-4 seemed to allow the CTLA-4 molecule to escape lysosomal degradation.
Even in the absence of ligand binding, CTLA-4 is known to be constitutively internalized and recycled through a mechanism involving interactions with the LRBA protein. Given that HL12 and HL32 induce internalization of CTLA-4, but don’t induce lysosomal degradation or reduction of CTLA-4 on the cell surface, Zhang, Du, and Liu et al. suspected that ipilimumab and TremeIgG1 interfered with CTLA-4 recycling, while HL12 and HL32 did not. To test this hypothesis, the researchers tested the anti-CTLA-4 antibodies in cells with a mutant CTLA-4 that was incapable of binding LRBA and being recycled. In these cells, all four antibodies induced comparable CTLA-4 downregulation. Furthermore, in cells with wild-type CTLA-4, complexes of bound ipilimumab/CTLA-4 did not associate with LRBA. These results suggest that LRBA is essential to CTLA-4 recycling, and that bound antibodies prevent CTLA-4 from binding to LRBA, thereby preventing CTLA-4 recycling and instead driving the complex towards lysosomal degradation.
To determine whether CTLA-4 recycling was related to the reduced incidents of irAEs observed with HL12 and HL32, the researchers used Primaquine (PQ) to block endocytic recycling. In vitro, PQ increased CTLA-4 downregulation in response to HL12, but not ipilimumab. In mice treated with anti-PD-1 (to sensitize them to irAEs) and either control antibody, TremeIgG1, or H12, the inclusion of PQ increased the toxicity of HL12, but did not further increase the toxicity of the already toxic TremeIgG1. This indicated that disrupted CTLA-4 recycling results in robust CTLA-4 downregulation and that CTLA-4 downregulation is related to the induction of irAEs.
In an effort to reduce the risk of irAEs in CTLA-4-targeted therapies, Zhang, Du, and Liu et al. tested whether they could make TremeIgG1 more sensitive to pH by replacing tyrosine with histidine in the complementary determining regions of the antibody. Using this strategy, they were able to generate variants of TremeIgG1 that were more sensitive to pH than HL12. The more pH-sensitive variants reduced the amount of antibody-associated CTLA-4 in cells, increased the amount of membrane-associated CTLA-4, and did not appear to drive CTLA-4 towards lysosomal degradation. This indicated that increased pH sensitivity allowed antibodies to dissociate from CTLA-4 after endocytosis, leaving room for CTLA-4 to interact with LRBA and be recycled to the cell surface rather than degraded in the lysosomes. In young, humanized (CTLA-4h/h) mice treated with anti-PD-1 (to sensitize them to irAEs) the pH-sensitive TremeIgG1 variants were less toxic than wild-type TremeIgG1, showing less evidence of severe irAEs and increasing the two-month survival rate from 10% to 86%. Although not demonstrated directly, the authors speculate that the reduced downregulation of CTLA4 in peripheral Tregs is responsible for the reduced irAEs with pH sensitive antibodies.
With the pH-sensitive variant TremeIgG1 antibodies showing a dramatically stronger safety profile than wild-type TremeIgG1, the researchers wondered whether the efficacy or bioavailability of the pH-sensitive antibodies was also altered. Tracking the antibodies within the cell, the researchers found that while pH-insensitive antibodies were degraded in lysosomes along with CTLA-4, pH-sensitive antibodies, such as the variant TremeIgG1, dissociated from CTLA-4 under the acidic conditions in the endosome and instead entered into recycling vesicles and re-accumulated in the supernatant. This led to an increased bioavailability of antibodies, which was associated with increased ADCC.
Both pH-sensitive and pH-insensitive CTLA-4-targeting antibodies are effective at treating small tumors. In the setting of large, established MC38 colon carcinoma tumors, however, pH-sensitive antibodies (HL12 and HL32) were faster and more effective than pH-insensitive ipilimumab at depleting Tregs, and showed more potent antitumor efficacy. Similar results were observed when pH-sensitive TremeIgG1 variants were compared against pH-insensitive wild-type TremeIgG1, with the pH-sensitive variants showing enhanced bioavailability and ADCC in vitro, and increased Treg depletion and enhanced antitumor efficacy, including a higher rate of complete tumor rejection, in vivo.
Overall, Zhang, Du, and Liu et al. show that using pH-sensitive antibodies to target CTLA-4 allows for recycling of both the antibody and the target molecule, thereby enhancing both the efficacy and safety of the treatment. Altering existing antibodies in a way that sensitizes them to changes in pH might serve as a simple and effective way to improve the clinical application of CTLA-4-targeted antibodies.
by Lauren Hitchings
This week, we asked our 3 questions to first co-author Yan Zhang and lead authors Pan Zheng and Yang Liu.
What prompted you to tackle this research question?
Immunotherapy-related adverse events (irAE) are becoming a bottleneck that prevents immunotherapy to reach its full potential to cure cancer. Among anti-CTLA-4 antibodies that cause tumor rejections, we have identified those that are toxic and those that are quite safe, but we were not sure what makes them different.
What was the most surprising finding of this study for you?
A eureka moment came to us when Yan followed the fate of anti-CTLA-4 antibodies inside the cell: the toxic antibodies moved quickly into big clusters and the antibodies that cause no toxicity also got inside the cells, but remained within much smaller vesicles within the cells! The differential trafficking of the antibodies inspired a series of experiments that clearly showed that the toxic antibodies drove CTLA-4 into lysosomal degradation, while the safe anti-CTLA-4 antibodies allowed recycling of CTLA-4 molecules. Further studies showed that it is the pH-dependent interaction between anti-CTLA-4 antibodies and their target that makes an antibody safe in humans. This conclusion was reached after extensive pharmacological and genetic studies in vivo and through extensive antibody studies in vitro.
What was the coolest thing you’ve learned (about) recently outside of the lab?
Science is cool