1 mg of FLT3L-WT protein was oxidatively stressed by incubating with 1 mM AAPH (2,2′-Azobis (2-amidinopropane) dihydrochloride) (Cayman Company, Cat# 0554100-6) in low-ionic strength histidine-acetate buffer, pH 5.5, for 16 h at 40°C. After 16 h incubation, the reaction was quenched with excess methionine (Met) at a ratio 20:1 (Met to AAPH). Control samples were prepared by the addition of water in place of AAPH. Control and stress samples were buffer-exchanged prior to analysis by LC-MS/MS peptide mapping.

A 250 µg sample of control and stressed protein was reduced with 20 mM DTT in 6M guanidine hydrochloride, 360 mM Tris, and 2 mM EDTA at pH 7.0 for 1 h. After the reduction, samples were cooled to room temperature and iodoacetic acid was used for alkylation with a final concentration of 50 mM for 15 min in darkness. Samples were then buffer exchanged into the digestion buffer (composed of 50 mM Tris, 2 mM CaCl₂, pH 7.5) and subjected to trypsin digestion for 1 h at 37°C (where the enzyme to substrate ratio was 1:20 by weight). The reaction was terminated by adding a final concentration of 2.2 mM L-Methionine along with 3% formic acid. Samples were then analyzed using a Vanquish UHPLC (Thermo Fisher) paired with an Exploris240 (Thermo Fisher) mass spectrophotometer. Separation was performed on an Acquity UPLC peptide CSH C18 column consisting of 1.7 µm, 130 Å particles (Waters), at a flow rate of 0.2 mL/min at 77°C. The mobile phase A was 0.1% formic acid in water and mobile phase B was 0.1% formic acid in acetonitrile. The separation gradient was set as follows: 1% mobile phase B for 2 minutes, 1-13% mobile phase B over 5 minutes, 13-35% mobile phase B over 35 minutes, 35-95% mobile phase B over 2 minutes, and then 95% mobile phase B for 2 minutes. Mass spectrometry data was collected in positive ion mode using a Top 8 data dependent scan with a resolution of 60000 for MS scans and 30000 for MS2 scans. An external calibration of the instrument was performed prior to sample analysis. Mass spectrometry data was processed using PMI-Byos software (Protein Metrics) using the PTM workflow. Relative quantitation was calculated using the top two charge states for both the native tryptic peptide and the modified counterpart. The percent of tryptophan oxidation following oxidative stress were then calculated.

White 96 well plates (Corning) were coated overnight at room temperature (RT) with 75 μL poly-L-ornithine solution (Millipore), washed 3 times with PBS and dried. OCI-AML5 cells were seeded at 3000 cells/well in 75 μL assay media (RPMI 1640, 5% heat inactivated FBS, 1 × Glutamax). FLT3L-Fc was serially diluted 1:10 with 10 nM starting concentration for 9 points. 75 μL of FLT3L-Fc diluted solutions were added to assay wells in triplicate. Maximum and neutral control wells included 10 nM FLT3L-Fc-WT or assay media alone. Assay plates were incubated at 37°C, 5% CO₂ for 7 days. 50 μL of CellTiter-Glo Luminescent cell viability reagent (Promega) was added. Plates were incubated for 10 min and luminescence was measured with a Tecan Infinite M1000 Microplate Reader. Raw values were normalized to the maximum and neutral control wells. Dose response curves were plotted, and 50% effective concentration (EC50) values were determined with GraphPad Prism using a four-parameter logistic fit.

Potential immunogenic T cell epitopes were computationally predicted using NetMHCIIpan-3.2 (34) and NetMHCIIpan-4.0 (35) for the nine common HLA-DR alleles: HLA-DRB1*01:01, HLA-DRB1*03:01, HLA-DRB1*04:01, HLA-DRB1*07:01, HLA-DRB1*08:01, HLA-DRB1*09:01, HLA-DRB1*11:01, HLA-DRB1*13:01, and HLA-DRB1*15:01. For each FLT3L-Fc variant, the FLT3L region was computationally digested into all possible overlapping 12-20-mer peptides. These peptides were considered to be strong binders or presented on HLA-DR if the percentile rank score for the predicted binding core (a 9-mer peptide) was within the top 10% (43, 44). Among these binding cores, likely tolerized epitopes were excluded if their sequences were identically aligned to 9-mer peptides from the human reference proteome using PEPMatch (45). Among the remaining binding cores, those that were predicted to bind to >1 HLA-DR molecules were considered potentially immunogenic CD4 T cell epitopes (43, 44). For each epitope, global population coverage was calculated using allele frequencies from the Immune Epitope-Database (IEDB) population coverage tool (46).

DC generation and loading was conducted following a previously published method (47). In brief, monocytes were isolated from human blood from anonymous healthy volunteers participating in the Genentech blood donor program, after receiving a written, informed consent from the Western Institutional Review Board. Monocytes were differentiated into immature DCs using interleukin-4 (IL-4) at 50 ng/ml and granulocyte-macrophage colony-stimulating factor (GM-CSF) at 3 ng/ml for five days. After incubation with 100 µg/mL of the test antibody and 1 µg/mL of lipopolysaccharide (Sigma) for 24 h, the DCs were harvested and washed with fluorescence-activated cell sorter (FACS) buffer (phosphate-buffered saline containing 2% fetal bovine serum and 0.02% sodium azide). For external staining, cells were stained with goat F(ab’)₂ anti-human IgG (H+L) conjugated to AlexaFluor-647 (Jackson ImmunoResearch Laboratories) for 30 min on ice, washed three times with FACS buffer and then fixed with 4% paraformaldehyde for 10 min. The cells were washed and resuspended in the FACS buffer before analysis. For total staining, cells were permeabilized and fixed with Fixation and Permeabilization Solution (BD Biosciences), washed three times with Permeabilization Solution prior to staining with the same goat F(ab’)₂ anti-human IgG (H+L) conjugated to AlexaFluor-647 for 30 min on ice. Cells were washed with permeabilization wash buffer and resuspended in FACS buffer before analysis. The cells were analyzed with an Attune NxT flow cytometer (Invitrogen). The data was analyzed using FlowJo (BD Biosciences, version 10.7.2) and plotted in Prism (v9.5.1). The internalization index was calculated by subtracting the fluorescence signal of the median fluorescence intensity (MFI) of externally stained cells from the MFI of the total stained cells.

The mature monocyte derived DC (MoDC) was obtained by incubating immature MoDC with 1 µg/mL of LPS overnight. To stain the surface FLT3 receptors on both immature and mature DCs, 20 μL of PE conjugated anti-human CD135 antibody (BD PharmingenTM) or isotype control (BD PharmingenTM) were added to 200 μL of cells (10⁶ cells/mL) respectively and incubated on ice for 30 minutes. The cells were washed and resuspended in the FACS buffer (phosphate-buffered saline containing 2% fetal bovine serum and 0.02% sodium azide) before analysis. The cells were analyzed with an Attune NxT flow cytometer (Invitrogen). The data was analyzed using FlowJo (BD Biosciences, version 10.7.2) and plotted in Prism (v9.5.1).

Cryopreserved peripheral blood mononuclear cells (PBMCs) from 48 healthy donors were thawed at 37°C water bath using Aim-V media (Gibco, Cat# 12055091) supplemented with 3% human serum (Sigma-Aldrich, Cat# H3667). Cells were then seeded in a 24-well plate with Lonza X-vivo15 media (Lonza, Cat# 04-418Q) at a concentration of 4X10⁶ cells/mL, with each well containing 1 mL medium. Following seeding, testing agents were then added to the designated wells at a final concentration of 50 ug/mL for each group. All cells were cultured in a 37°C incubator with 5% CO₂ for five days. On day six, BrdU reagent (BD Bioscience, Cat# 557891) was added to the cells, and they were further cultured for an additional 24 hours. The following day, cells were harvested and then stained with Aqua Live/Dead (Thermo Fisher, Cat# L34957), anti-BV421-CD3 (BD Bioscience, Cat# 562426), anti-PE-Cy-7-CD4 (BD Bioscience, Cat# 557852), anti-APC-CD14 (BD Bioscience, Cat# 555399), anti-APC-CD19 (BD Bioscience, Cat# 555415) antibodies. After surface staining, cells were washed with 0.5% BSA-PBS and then fixed and permeabilized by Cytofix/Cytoperm buffer (BD Bioscience, Cat# 554714) according to the manufacturing protocol. Cells were then stained with anti-BrdU (BD Bioscience, Cat# 557891) antibody followed by flow cytometry analysis on Attune (Invitrogen™). FACS files were processed using FlowJo software (Tree Star, Inc.; Ashland, OR) and downstream data analysis was performed by R studio.

To determine whether any of the in silico predicted epitopes resembled microbial sequences, we used NCBI protein BLAST (BLASTP) to search the non-redundant protein sequences (nr) database (51). We analyzed predicted epitopes from FLT3L-Fc-WT and FLT3L-Fc-W144D (e.g. LVALKPWIT and LVALKPDIT, respectively) using BLASTP with the following the short input sequence settings: word size = 2, expect value = 200000, hitlist size = 100, gap open costs = 9, gap extension cost = 1, matrix = PAM30, window Size = 40 and threshold = 11. The search results were filtered to identify 8- and 9-mer sequences that were identical to bacterial protein sequences.

Given the limitations associated with the rapid clearance of FLT3 ligand in clinical settings, we engineered a FLT3L-Fc fusion protein to extend its half-life and enhance its pharmacokinetic profile (40). During molecule assessment of FLT3L-Fc, we identified a tryptophan residue (W144) within the FLT3L portion of the protein ( Figure 2A ) that exhibited high susceptibility to oxidation when stressed with AAPH (2,2′-Azobis (2-amidinopropane) dihydrochloride) ( Figure 2B ). To address this oxidation liability, which may impact the biological activities of the molecule, we computationally mutated W144 to other residues except Methionine and Cysteine without oxidation risk. To reduce the risk of generating new T cell epitopes, we aimed to identify the muteins with the lowest immunogenicity risk. Using NetMHCIIpan-3.2 (34), the FLT3L-Fc variant with the lowest predicted immunogenicity risk had W144 replaced by aspartic acid (W144D mutein, Supplementary Table S1 ). Thus, the W144D mutein was selected as the lead candidate, as it not only mitigates oxidation liability but also presents the lowest predicted immunogenicity risk among the tested variants.

Based on the in silico prediction results, we generated the W144D mutein to experimentally quantify its biological activity and immunogenicity risk. To assess whether the single point mutation W144D impacted the biological potency of the parent FLT3L-Fc protein, we used the human OCI-AML5 cell line, which has been reported to proliferate in response to human FLT3L stimulation (52, 53). Notably, both FLT3L-Fc-WT and W144D mutein induced similar dose-dependent proliferation of OCI-AML5 cells, with EC₅₀ potency values of 6.0 pM and 6.3 pM, respectively ( Figure 2C ). This indicated that the W144D mutation did not significantly impact the molecule’s biological potency, thereby emerging as a promising candidate. To experimentally validate the low immunogenicity risk predicted for W144D, we used a suite of in vitro assays, including DC internalization, T cell proliferation, MAPPs and peptide binding assays.

First, we assessed immunogenicity risk using a DC internalization assay. DC-based internalization assay has demonstrated a correlation between the magnitude of internalization and the immunogenicity risk of biotherapeutics (54). The characterization of internalization of FLT3L-Fc variants in monocyte derived DCs (MoDCs) was based on a previously reported assay (47), with the modification of using anti-human IgG (H+L) as the detection reagent. Two FLT3L-Fc variants were evaluated together with two control molecules: bococizumab (high internalization benchmark, clinical ADA 48%) and bevacizumab (low internalization benchmark, clinical ADA 0.6%) (55). As shown in Figure 3A , surprisingly, both FLT3L-Fc-WT and W144D showed very high levels of internalization, comparable to bococizumab. There was no statistical difference between WT and W144D mutein. Given that DCs can capture antigens through receptor-mediated endocytosis and/or non-specific constitutive macropinocytosis and considering that the FLT3 receptor is known to undergo rapid internalization upon dimerization by FLT3 ligand (56), it is important to examine the FLT3 receptor expression level on MoDCs to investigate the internalization mechanism. Interestingly, surface staining of both immature and mature MoDCs using flow cytometry did not reveal any detectable FLT3 receptor ( Figure 3B ), suggesting the high internalization of FLT3L-Fc variants in MoDCs is unlikely to be mediated by receptor-mediated endocytosis. Instead, the uptake may be facilitated by macropinocytosis, which plays a crucial role in DC mediated antigen presentation to CD4⁺ T cells. Nonetheless, the precise mechanism remains to be further elucidated.

Activation of T cells is a valuable marker for immunogenic responses, so we performed a reported PBMC-based CD4⁺ T cell proliferation assay (28) to further evaluate the immunogenicity risk for the two FLT3L-Fc variants. PBMCs isolated from 48 healthy donors, representing a diverse set of MHCII alleles ( Supplementary Table S2 ) and exhibiting a broad genotype distribution of DRB1 alleles ( Supplementary Figure 1 ), were stimulated with human natural FLT3 ligand, FLT3L-Fc-WT, and the W144D mutein. Additionally, bococizumab and bevacizumab served as positive and negative benchmarks, respectively. T cell proliferation was subsequently assessed by flow cytometry ( Figure 4A ). The stimulation index (SI) was calculated as the ratio of Brdu⁺CD4⁺ cells percent in the testing group over Brdu⁺CD4⁺ cells percent in the untreated group to define the positive response of the test molecules for each donor. As shown in Figures 4B, C , the positive control bococizumab induced a very high T cell response, with approximately 66% of donors testing positive, while the negative control bevacizumab exhibited a low stimulation index with around 5% positive donors. As expected, the natural ligand did not activate CD4⁺ T cells with only 8.3% of donors showing positive response. Interestingly, the results revealed a stark contrast between the two FLT3L-Fc variants. FLT3L-Fc-WT induced very low T cell response, comparable to bevacizumab and natural ligand, suggesting low immunogenic potential. Conversely, the W144D mutein elicited robust T cell responses, with around 48% of donors testing positive, indicating a high risk of immunogenicity. These findings suggest that while the W144D mutation does not impact the protein’s functional activity, it may introduce new epitopes capable of activating T cells, thereby increasing the risk of an immunogenic response. Furthermore, we analyzed the specific alleles from donors who exhibited positive responses to the W144D mutein. Interestingly, T cell activation induced by the W144D mutein was not limited to T cells associated with specific MHCII alleles. Instead, it occurred across T cells from donors with a diverse range of MHCII alleles as shown in Supplementary Table S2 . These findings further underscore the increased risk of immunogenicity associated with the W144D mutein, but not the WT variant.

When the experimental assays were being performed, an updated pMHCII presentation prediction model was released, NetMHCIIpan-4.0. This model was shown to significantly outperform the initial model we used, NetMHCIIpan-3.2 (19, 35). Using NetMHCIIpan-4.0, both FLT3L-Fc-WT and W144D mutein were subsequently re-assessed for immunogenicity risk by searching the FLT3L region for potential immunogenic linear CD4⁺ T cell epitopes. The FLT3L sequence was computationally digested into all possible overlapping 12-20-mer peptides. A pMHCII pair was considered to be strongly presented if the elution percentile rank score was less than 10%. From these pMHCII pairs, the predicted binding cores were removed from further consideration if their sequences were identically aligned to 9-mer peptides from the human reference proteome (45). Among the remaining binding cores, those predicted to bind to more than one MHCII molecule were considered potentially immunogenic T cell epitopes. For FLT3L-Fc-WT, the FLT3L region did not have any potential immunogenic T cell epitopes predicted. In contrast, the same region in the W144D mutein was predicted to have two potential T cell epitopes, both of which overlapped with the mutated residue ( Table 1 ), indicating a high immunogenicity risk for W144D. Compared to the previous prediction from NetMHCIIpan-3.2, the predictions from the newer pMHCII presentation model, NetMHCIIpan-4.0, were more consistent with the results from the T cell proliferation assay.

To further evaluate the immunogenic potential of the FLT3L-Fc variants and specifically identify peptide regions from these variants bind to MHCII, we conducted a MAPPs assay. CD14⁺ monocytes were isolated from PBMCs of healthy human donors. Fifty percent of the donors ( Supplementary Table S3 ) were selected based on sharing at least one HLA-DRB1 allele in common with donors that scored a positive T cell response in the T cell proliferation assay ( Figure 4 ) and the other 50% of donors were selected randomly. The monocytes were subsequently differentiated into DCs. DCs were treated with FLT3L-Fc variants, matured, and lysed. HLA-DR molecules were immunoprecipitated from the lysates and peptides bound to HLA-DR proteins were acid eluted from the pMHCII complex and detected by mass spectrometry. MAPPs enabled the identification and relative quantification of HLA-DR presented peptides corresponding to the FLT3L-Fc variants, thus gaining a deeper understanding of their immunogenic potential. Similar peptide clusters were detected for both FLT3L-Fc-WT, natural ligand, and W144D mutein ( Supplementary Figure 2A ). Peptide clusters containing the W144 residue in both the WT and W144D mutein were detected by MAPPs ( Figure 5A ). Interestingly, two peptide cores (LVALKPDIT and LKPDITRQN) identified by in silico prediction ( Table 1 ) overlapped with the peptides detected by MAPPs in the W144D mutein. Comparisons between WT and mutein proteins showed broadly overlapping peptide clusters in the mutation region, but with more peptide spectrum matches (PSM) in the FLT3L-Fc-W144D variant ( Figure 5B ). Furthermore, presentation of peptide clusters containing W144D was only observed in donors 1, 6, 7 and 10. Donors 6, 7, and 10 share the HLA-DRB1*01:01 allele ( Supplementary Figure 2B ), which was also found in some donors who had positive T cell responses in the T cell proliferation assay. Although we observed a 3-5x greater mass spectrometry signal for the peptide ETSEQLVALKPDIRT (W144D) compared to ETSEQLVALKPWIRT (WT) in donors 6 and 10, as determined by area-under-the-curve (AUC) of the corresponding peak ( Supplementary Figure 2C ), it was difficult to establish a statistically significant correlation between MHCII alleles with positive T cell responses and MAPPs from our correlation analysis.

Next, we performed two-dimensional liquid chromatography-mass spectrometry (2D-LCMS) to further evaluate the efficiency of in vitro peptide exchange with MHCII proteins and peptides of interest. To evaluate peptide binding across multiple common MHCII alleles, we used recombinant MHCII proteins containing an exchangeable CLIP peptide variant. These proteins were exposed to low pH conditions to enable the exchange of the CLIP peptide with the epitope peptides. Peptide exchange was monitored using 2D-LCMS, and the propensity of each peptide to bind the MHCII allele of interest was determined by comparison of the mass spec peak intensities for the exchanged peptides. In the first dimension, pMHCII complexes were separated on an analytical size exclusion column from excess peptides in the reaction mixture and then collected in a sampling loop. In the second dimension, the purified pMHCII collected in the sampling loop was injected onto a reversed-phase column to separate bound peptides from the MHCII complex. These steps enabled us to evaluate the binding efficiency between the MHCII proteins and the peptides of interest ( Figure 5C ). The top pair of peptides, ETSEQLVALKPWITR and ETSEQLVALKPDITR, identified from the MAPPs assay for the WT and W144D mutein, respectively, were synthesized as representatives and examined using 2D-LCMS analysis. The results demonstrated that both the WT peptide and the peptide containing the W144D mutation promiscuously bind to MHCII alleles ( Figure 5D ; Supplementary Figure 3 ).

Of the eight MHCII alleles tested, the highest recovery of complex and mass spec intensity for W144D was observed in DRB1*01:01 and DRB1*04:01; for which only DRB1*01:01 was the same donor MHCII allele that W144D T cell epitopes were detected by MAPPs. Taken together, these findings suggest stronger peptide binding is not always directly proportional to greater presentation for all epitopes. Additionally, the peptide binding data and T cell assay results indicate that T cells specific to the WT sequence are likely tolerized or clonally deleted during development. In contrast, the T cell response to the W144D mutein suggests that the D mutation is recognized as a foreign antigen by cognate TCRs, thereby inducing a strong T cell response and enhancing immunogenicity risk.

In addition to the increased immunogenicity risk associated with a higher number of T cell epitopes, some T cell epitopes can carry additional risk due to their similarity to epitopes present on microbial agents (57). To understand the potential mechanism by which a single point mutation could introduce new T cell epitopes that significantly enhance the immunogenicity response, we conducted a sequence similarity alignment against bacteria proteins using NCBI protein BLAST (Basic Local Alignment Search Tool). One of the in silico predicted epitopes, LVALKPDIT, was identical to sequences from 2 bacterial strains. In addition, this epitope included 8-mer subsequences (VALKPDIT and LVALKPDI) that were identical to sequences from 70 different bacterial strains ( Supplementary Figure 4 ; Supplementary Table S4 ). Among these, approximately 41% of the strains belonged to the Pseudomonadota family, a major phylum of Gram-negative bacteria found in human gut microbiota (58). In contrast, no bacterial sequences were identical to the corresponding sequence (LVALKPWIT) in the WT molecule. This finding may explain the high immunogenic response induced by the W144D mutein.