Correcting specificity in an antibody therapeutic: Off-target binding detected and rectified to counter toxicity risk for clinical stage antibody


Monoclonal antibodies feature prominently in the recent wave of immunotherapy development due to their potential for high potency, long half-life and highly specific modulation of a single drug target. However, binding to non-target receptors could drive unexpected and potentially severe toxicities in newly-developed immunotherapies. If human off-target reactivity is not identified by in vitro screening or through classical pre-clinical toxicology, the issue may only be uncovered after lengthy and costly clinical development.

The anti-PD1 antibody SHR-1210 (also known as camrelizumab) has had the expected biological activity in early clinical studies for solid tumours; however, its toxicity profile is different to other anti-PD1 immunotherapies. The majority of patients receiving camrelizumab have developed capillary hemangioma, which is a benign tumour formed by a collection of excess blood vessels. This observation led the team at UltraHuman Eight Ltd to hypothesize that the target-binding domains of SHR-1210-IgG1 might interact with previously unidentified and unpredictable receptors that are associated with vascular development or tissue differentiation.

Materials and Methods

The full Retrogenix screening technology was used to investigate the binding profile of SHR-1210-IgG1 (an effector function null human IgG1 carrying the v-domains of SHR-1210, provided by UltraHuman Eight Ltd, Edinburgh, UK).

“Primary Screen”: The SHR-1210-IgG1 antibody was screened at a concentration of 5ug/ml against approximately 5000 human plasma membrane proteins that were over-expressed in HEK293 cells on Retrogenix’s cell microarrays. Antibody binding was detected using an AF647 fluorescent secondary anti-human IgG1 antibody.

“Confirmation Screen”: Vectors encoding all hits from the primary screen were sequenced to confirm their correct identities. These hits, along with appropriate controls, were then re-expressed on custom slides and treated with either: the test antibody, control antibodies or a negative control treatment of secondary antibody only (no test molecule). A further confirmation screen was performed using live cell transfectants with a flow cytometry readout.

These confirmation screens were later repeated using the novel, lead IgGs that were generated through the re-engineering of the SHR-1210-IgG1 antibody by the UltraHuman Eight team.


The results for SR-1210-IgG1 confirmed consistent binding to the primary target, PD1, along with three additional, unexpected hits (see figure 1). These were:

  • VEGFR2 (KDR)
  • FZD5 – frizzled class receptor 5
  • ULBP2Figure_1_Retrogenix_case_study

Figure 1. Off-target binding analysis – Retrogenix confirmation screen. Plasmids encoding PD1 and the putative SHR-1210-IgG1 off-target binding proteins identified in the primary screen were arrayed and used to transfect HEK-293 cells. Effective transfection of all plasmids was confirmed by screening for the co-encoded marker ZS green (a). Separate slides were then probed in duplicate using SHR-1210-IgG1 (b), Isotype IgG1 (c), Rituximab (d), and no primary antibody (e). These analyses confirmed that only SHR-1210-IgG1 exhibited binding to PD1, but also exhibited unexpected off-target binding to VEGFR2 (KDR), FZD5 and ULBP2 proteins.

Following modification of the antibody by UltraHuman Eight, all of the novel IgG leads that were selected for rescreening showed specific binding to PD1 only and no binding to the additional receptors; VEGFR2, FZD5 or ULBP2. This was shown both by cell microarray (data not shown) and by flow cytometry, (data for one lead IgG versus SHR-1210-IgG1 are shown in figure 2).



Figure 2. Specificity analyses by flow cytometry using transiently-transfected HEK-293 cells. Analyses of binding specificity were performed on HEK-293 cells transiently transfected with plasmids encoding either human PD1, human VEGFR2 (KDR), human FZD5 or human ULBP2. All plots show the target of interest transfected cells (grey line) versus ZS green marker-only transfected cells (black line). These analyses confirmed that SHR-1210-IgG1 exhibited binding to PD1, VEGFR2, FZD5 and ULBP2 but not ZS-green. The designer lead IgGs all showed binding to PD1 but exhibited no measurable signal on ZS-green, VEGFR2, FZD5 or ULBP2 transfected cells (data for only one lead shown (IgG1-06D02) – further data can be found in full paper).


The initial off-target screening work uncovered highly selective, low intensity binding to three human receptors (in addition to PD1) that may explain the molecular mechanisms behind the clinical toxicity that has been observed with SHR-1210-IgG1. These interactions were validated by orthogonal binding assays, and a functional test of SHR-1210-IgG1 against VEGFR showed agonistic activity. Notably, off-target binding to VEGFR2 and possibly frizzled class receptor 5 (FZD5) may stimulate vascular neogenesis, leading to development of hemangioma. Importantly, the off-target binding of VEGFR2 actually led to potent activation of that receptor, which is known to be a key signalling event that stimulates capillary blood vessel proliferation. A third off-target, ULBP2, is an MHC class 1-related protein which is a known ligand for natural killer cell activating receptors; however the consequence of modulating ULBP2 during cancer therapy with this anti-PD1 antibody is currently unknown.

Having proven that SHR-1210-IgG1 target binding domains exhibit relevant off-target reactivity, the UltraHuman Eight group investigated further and identified that the antibody’s polyspecificity existed prior to the humanisation process and was exacerbated during humanisation. Subsequent engineering of the antibody through mutagenesis and reselection generated novel antibodies that targeted PD1 and showed improved pharmacological properties, but which did not bind to the three off-targets originally identified.


The potential impact of off-target issues in antibody development is increasingly being recognised. The authors of this study have demonstrated that clinically-relevant polyspecificity is a phenomenon that can be identified and can be rapidly engineered out of a final therapeutic protein, potentially leading to improved potency and reduced toxicity.

The full paper, published in the journal ‘mAbs’, is entitled: “Anti-PD1 ‘SHR-1210ʹ aberrantly targets proangiogenic receptors and this polyspecificity can be ablated by paratope refinement.” For a link to the full text please click here.

To arrange a presentation or a meeting to discuss our off-target screening capabilities then please get in touch.

Contact us about a project