A diversity of assembly mechanisms of a generic amyloid fold

Protein misfolding and amyloid assembly have long been recognized as being responsible for many devastating human diseases. Recent findings indicate that amyloid assemblies may facilitate crucial biological processes from bacteria to mammals. This review focuses on the mechanistic understanding of amyloid formation, including the transformation of initially innocuous proteins into oligomers and fibrils. The result is a competing folding and assembly energy landscape, which contains a number of routes by which the polypeptide chain can convert its primary sequence into functional structures, dysfunctional assemblies, or epigenetic entities that provide both threats and opportunities in the evolution of life.

Identification of the protein receptor binding site of botulinum neurotoxins B and G proves the double-receptor concept

Botulinum neurotoxins (BoNTs) cause muscle paralysis by selectively cleaving core components of the vesicular fusion machinery within motoneurons. Complex gangliosides initially bind into a pocket that is conserved among the seven BoNTs and tetanus neurotoxin. Productive neurotoxin uptake also requires protein receptors. The interaction site of the protein receptor within the neurotoxin is currently unknown. We report the identification and characterization of the protein receptor binding site of BoNT/B and BoNT/G. Their protein receptors, synaptotagmins I and II, bind to a pocket at the tip of their H(CC) (C-terminal domain of the C-terminal fragment of the heavy chain) that corresponds to the unique second carbohydrate binding site of tetanus neurotoxin, the sialic acid binding site. Substitution of amino acids in this region impaired binding to synaptotagmins and drastically decreased toxicity at mouse phrenic nerve preparations; CD-spectroscopic analyses evidenced that the secondary structure of the mutated neurotoxins was unaltered. Deactivation of the synaptotagmin binding site by single mutations led to virtually inactive BoNT/B and BoNT/G when assayed at phrenic nerve preparations of complex-ganglioside-deficient mice. Analogously, a BoNT B mutant with deactivated ganglioside and synaptotagmin binding sites lacked appreciable activity at wild-type mouse phrenic nerve preparations. Thus, these data exclude relevant contributions of any cell surface molecule other than one ganglioside and one protein receptor to the entry process of BoNTs, which substantiates the double-receptor concept. The molecular characterization of the synaptotagmin binding site provides the basis for designing a novel class of potent binding inhibitors.

Conformational conversion during amyloid formation at atomic resolution

Numerous studies of amyloid assembly have indicated that partially folded protein species are responsible for initiating aggregation. Despite their importance, the structural and dynamic features of amyloidogenic intermediates and the molecular details of how they cause aggregation remain elusive. Here, we use ΔN6, a truncation variant of the naturally amyloidogenic protein β₂-microglobulin (β₂m), to determine the solution structure of a nonnative amyloidogenic intermediate at high resolution. The structure of ΔN6 reveals a major repacking of the hydrophobic core to accommodate the nonnative peptidyl-prolyl trans-isomer at Pro32. These structural changes, together with a concomitant pH-dependent enhancement in backbone dynamics on a microsecond-millisecond timescale, give rise to a rare conformer with increased amyloidogenic potential. We further reveal that catalytic amounts of ΔN6 are competent to convert nonamyloidogenic human wild-type β₂m (Hβ₂m) into a rare amyloidogenic conformation and provide structural evidence for the mechanism by which this conformational conversion occurs.

Understanding the complex mechanisms of β2-microglobulin amyloid assembly

Several protein misfolding diseases are associated with the conversion of native proteins into ordered protein aggregates known as amyloid. Studies of amyloid assemblies have indicated that non-native proteins are responsible for initiating aggregation in vitro and in vivo. Despite the importance of these species for understanding amyloid disease, the structural and dynamic features of amyloidogenic intermediates and the molecular details of how they aggregate remain elusive. This review focuses on recent advances in developing a molecular description of the folding and aggregation mechanisms of the human amyloidogenic protein β(2)-microglobulin under physiologically relevant conditions. In particular, the structural and dynamic properties of the non-native folding intermediate I(T) and its role in the initiation of fibrillation and the development of dialysis-related amyloidosis are discussed.

The biological activity of botulinum neurotoxin type C is dependent upon novel types of ganglioside binding sites

The seven botulinum neurotoxins (BoNT) cause muscle paralysis by selectively cleaving core components of the vesicular fusion machinery. Their extraordinary activity primarily relies on highly specific entry into neurons. Data on BoNT/A, B, E, F and G suggest that entry follows a dual receptor interaction with complex gangliosides via an established ganglioside binding region and a synaptic vesicle protein. Here, we report high resolution crystal structures of the BoNT/C cell binding fragment alone and in complex with sialic acid. The WY-motif characteristic of the established ganglioside binding region was located on an exposed loop. Sialic acid was co-ordinated at a novel position neighbouring the binding pocket for synaptotagmin in BoNT/B and G and the sialic acid binding site in BoNT/D and TeNT respectively. Employing synaptosomes and immobilized gangliosides binding studies with BoNT/C mutants showed that the ganglioside binding WY-loop, the newly identified sialic acid-co-ordinating pocket and the area corresponding to the established ganglioside binding region of other BoNTs are involved in ganglioside interaction. Phrenic nerve hemidiaphragm activity tests employing ganglioside deficient mice furthermore evidenced that the biological activity of BoNT/C depends on ganglioside interaction with at least two binding sites. These data suggest a unique cell binding and entry mechanism for BoNT/C among clostridial neurotoxins.

Broad substrate specificity of the amide synthase in S. hygroscopicus--new 20-membered macrolactones derived from geldanamycin

The amide synthase of the geldanamycin producer, Streptomyces hygroscopicus, shows a broader chemoselectivity than the corresponding amide synthase present in Actinosynnema pretiosum, the producer of the highly cytotoxic ansamycin antibiotics, the ansamitocins. This was demonstrated when blocked mutants of both strains incapable of biosynthesizing 3-amino-5-hydroxybenzoic acid (AHBA), the polyketide synthase starter unit of both natural products, were supplemented with 3-amino-5-hydroxymethylbenzoic acid instead. Unlike the ansamitocin producer A. pretiosum, S. hygroscopicus processed this modified starter unit not only to the expected 19-membered macrolactams but also to ring enlarged 20-membered macrolactones. The former mutaproducts revealed the sequence of transformations catalyzed by the post-PKS tailoring enzymes in geldanamycin biosynthesis. The unprecedented formation of the macrolactones together with molecular modeling studies shed light on the mode of action of the amide synthase responsible for macrocyclization. Obviously, the 3-hydroxymethyl substituent shows similar reactivity and accessibility toward C-1 of the seco-acid as the arylamino group, while phenolic hydroxyl groups lack this propensity to act as nucleophiles in the macrocyclization. The promiscuity of the amide synthase of S. hygroscopicus was further demonstrated by successful feeding of four other m-hydroxymethylbenzoic acids, leading to formation of the expected 20-membered macrocycles. Good to moderate antiproliferative activities were encountered for three of the five new geldanamycin derivatives, which matched well with a competition assay for Hsp90α.

Spectral phase control of interfering chirped pulses for high-energy narrowband terahertz generation

Highly-efficient optical generation of narrowband terahertz radiation enables unexplored technologies and sciences from compact electron acceleration to charge manipulation in solids. State-of-the-art conversion efficiencies are currently achieved using difference-frequency generation driven by temporal beating of chirped pulses but remain, however, far lower than desired or predicted. Here we show that high-order spectral phase fundamentally limits the efficiency of narrowband difference-frequency generation using chirped-pulse beating and resolve this limitation by introducing a novel technique based on tuning the relative spectral phase of the pulses. For optical terahertz generation, we demonstrate a 13-fold enhancement in conversion efficiency for 1%-bandwidth, 0.361 THz pulses, yielding a record energy of 0.6 mJ and exceeding previous optically-generated energies by over an order of magnitude. Our results prove the feasibility of millijoule-scale applications like terahertz-based electron accelerators and light sources and solve the long-standing problem of temporal irregularities in the pulse trains generated by interfering chirped pulses.

Optical generation of terahertz radiation is needed for many applications, but gaining high efficiency is still a challenge. The authors report a method to overcome dispersion effects in interfering chirp pulses used for THz pulse production by tuning their relative spectral phase, enabling 0.6 mJ of THz energy output.

Kinetic characterization of the disulfide bond-forming enzyme DsbB

DsbB is an integral membrane protein responsible for the de novo synthesis of disulfide bonds in Escherichia coli and many other prokaryotes. In the process of transferring electrons from DsbA to a tightly bound ubiquinone cofactor, DsbB undergoes an unusual spectral transition at approximately 510 nm. We have utilized this spectral transition to study the kinetic cycle of DsbB in detail using stopped flow methods. We show that upon mixing of Dsb-B(ox) and DsbA(red), there is a rapid increase in absorbance at 510 nm (giving rise to a purple solution), followed by two slower decay phases. The rate of the initial phase is highly dependent upon DsbA concentration (k(1) approximately 5 x 10(5) M(-1) s(-1)), suggesting this phase reflects the rate of DsbA binding. The rates of the subsequent decay phases are independent of DsbA concentration (k(2) approximately 2 s(-1); k(3) approximately 0.3 s(-1)), indicative of intramolecular reaction steps. Absorbance measurements at 275 nm suggest that k(2) and k(3) are associated with steps of quinone reduction. The rate of DsbA oxidation was found to be the same as the rate of quinone reduction, suggestive of a highly concerted reaction. The concerted nature of the reaction may explain why previous efforts to dissect the reaction mechanism of DsbB by examining individual pairs of cysteines yielded seemingly paradoxical results. Order of mixing experiments showed that the quinone must be pre-bound to DsbB to observe the purple intermediate as well as for efficient quinone reduction. These results are consistent with a kinetic model for DsbB action in which DsbA binding is followed by a rapid disulfide exchange event. This is followed by quinone reduction, which is rate-limiting in the overall reaction cycle.

Identification and Characterization of Botulinum Neurotoxin A Substrate Binding Pockets and Their Re-Engineering for Human SNAP-23

Botulinum neurotoxins (BoNTs) are highly potent bacterial proteins that block neurotransmitter release at the neuromuscular junction by cleaving SNAREs (soluble N-ethyl maleimide sensitive factor attachment protein receptors). However, their serotype A (BoNT/A) that cleaves SNAP-25 (synaptosomal-associated protein of 25 kDa) has also been an established pharmaceutical for treatment of medical conditions that rely on hyperactivity of cholinergic nerve terminals for 25 years. The expansion of its use to a variety of further medical conditions associated with hypersecretion components is prevented partly because the involved SNARE isoforms are not cleaved. Therefore, we examined by mutational analyses the reason for the resistance of human SNAP-23, an isoform of SNAP-25. We show that replacement of 10 SNAP-23 residues with their SNAP-25 counterparts effects SNAP-25-like cleavability. Conversely, transfer of each of the replaced SNAP-23 residues to SNAP-25 drastically decreased the cleavability of SNAP-25. By means of the existing SNAP-25-toxin co-crystal structure, molecular dynamics simulations, and corroborative mutagenesis studies, the appropriate binding pockets for these residues in BoNT/A were characterized. Systematic mutagenesis of two major BoNT/A binding pockets was conducted in order to adapt these pockets to corresponding amino acids of human SNAP-23. Human SNAP-23 cleaving mutants were isolated using a newly established yeast-based screening system. This method may be useful for engineering novel BoNT/A pharmaceuticals for the treatment of diseases that rely on SNAP-23-mediated hypersecretion.

Botulinum neurotoxin G binds synaptotagmin-II in a mode similar to that of serotype B: tyrosine 1186 and lysine 1191 cause its lower affinity

Botulinum neurotoxins (BoNTs) block neurotransmitter release by proteolyzing SNARE proteins in peripheral nerve terminals. Entry into neurons occurs subsequent to interaction with gangliosides and a synaptic vesicle protein. Isoforms I and II of synaptotagmin were shown to act as protein receptors for two of the seven BoNT serotypes, BoNT/B and BoNT/G, and for mosaic-type BoNT/DC. BoNT/B and BoNT/G exhibit a homologous binding site for synaptotagmin whose interacting part adopts helical structure upon binding to BoNT/B. Whereas the BoNT/B-synaptotagmin-II interaction has been elucidated in molecular detail, corresponding information about BoNT/G is lacking. Here we systematically mutated the synaptotagmin binding site in BoNT/G and performed a comparative binding analysis with mutants of the cell binding subunit of BoNT/B. The results suggest that synaptotagmin takes the same overall orientation in BoNT/B and BoNT/G governed by the strictly conserved central parts of the toxins' binding site. The surrounding nonconserved areas differently contribute to receptor binding. Reciprocal mutations Y1186W and L1191Y increased the level of binding of BoNT/G approximately to the level of BoNT/B affinity, suggesting a similar synaptotagmin-bound state. The effects of the mutations were confirmed by studying the activity of correspondingly mutated full-length BoNTs. On the basis of these data, molecular modeling experiments were employed to reveal an atomistic model of BoNT/G-synaptotagmin recognition. These data suggest a reduced length and/or a bend in the C-terminal part of the synaptotagmin helix that forms upon contact with BoNT/G as compared with BoNT/B and are in agreement with the data of the mutational analyses.

Description of spatio-temporal couplings from heat-induced compressor grating deformation

High average power high-intensity laser systems can suffer from a heat-induced deformation of the final compressor gratings, which introduces wavefront aberrations and spatio-temporal couplings to the pulse. Here, we use a simple numerical description, that was first introduced by Li et al. (Appl. Phys. Express, 10, 102702, 2017 and Optics Express, 26, 8453, 2018), to calculate the resulting degradation of the peak intensity and the 3-dimensional deformation of the laser pulse as a function of average power, and verify the results using experimental data. For a typical 100 TW-class laser we find that non-negligible pulse distortions can occur at an average power as low as 2.7 Watts. An open source implementation of our numerical description is available for researchers to estimate the effects of spatio-temporal couplings for their specific laser configuration.

Optimal Beam Loading in a Laser-Plasma Accelerator

Applications of laser-plasma accelerators demand low energy spread beams and high-efficiency operation. Achieving both requires flattening the accelerating fields by controlled beam loading of the plasma wave. Here, we optimize the generation of an electron bunch via localized ionization injection, such that the combination of injected current profile and averaged acceleration dynamics results in optimal beam loading conditions. This enables the reproducible production of 1.2% rms energy spread bunches with 282 MeV and 44 pC at an estimated energy-transfer efficiency of ∼19%. We correlate shot-to-shot variations to reveal the phase space dynamics and train a neural network that predicts the beam quality as a function of the drive laser.

Spatio-spectral couplings in saturated collinear OPCPA

Ultrafast laser pulses featuring both high spatio-temporal beam quality and excellent energy stability are crucial for many applications. Here, we present a seed laser with high beam quality and energy stability, based on a collinear optical parametric chirped pulse amplification (OPCPA) stage, delivering 46 µJ pulses with a 25 fs Fourier limit at 1 kHz repetition rate. While saturation of the OPCPA stage is necessary for achieving the highest possible energy stability, it also leads to a degradation of the beam quality. Using simulations, we show that spectrally dependent, rotationally symmetric aberrations dominate the collinear OPCPA in saturation. We experimentally characterize these aberrations and then remove distinct spatial frequencies to greatly improve the spectral homogeneity of the beam quality, while keeping an excellent energy stability of 0.2 % rms measured over 70 hours.

Abstract 3268: Costimulatory T-cell engagement by PRS-342, a GPC3/4-1BB bispecific molecule, leads to activation of T-cells and tumor growth inhibition in a HCC humanized mouse model

Background: Increasing evidence shows that 4-1BB is a key costimulatory immunoreceptor and a highly-promising therapeutic target in cancer. Current antibody-based approaches showed immune cell activation not only in tumor tissues, but also in the periphery, which is associated with dose-limiting on-target toxicity. To overcome this limitation, we generated PRS-342, a GPC3/4-1BB bispecific molecule based on the Anticalin technology. This molecule is designed to promote 4-1BB clustering by bridging 4-1BB-positive T cells with GPC3-expressing tumor cells. GPC3 is an oncofetal protein with high tumor selectivity and high expression in not only hepatocellular carcinomas, but also in a variety of other tumors with high medical need.

Methods: Anticalin therapeutics are 18 kD proteins derived from human lipocalins. We utilized phage display to generate an Anticalin protein binding to 4-1BB with high affinity and specificity. The PRS-342 bispecific construct was generated by genetic fusion of a 4-1BB-specific Anticalin protein to a humanized high affinity GPC3-targeting monoclonal antibody with an engineered IgG4 backbone.

Results: We show that the bispecific molecule PRS-342 retains its ability to bind both targets (4-1BB and GPC3) with similar affinity to the parental building blocks and is capable of binding both targets simultaneously. Biophysical characterization of PRS-342 reflect its good drug-like properties. Using in vitro assays based on mixed culture of human pan-T cells and GPC3-expressing tumor cell lines, PRS-342 induced T-cell costimulation leading to increase production of IL-2 with EC50 in the sub-nanomolar range. In contrast, monospecific binding to GPG3 or 4-1BB by benchmarks or single building blocks was not able to activate T cells in this assay. These data demonstrate the ability of PRS-342 to bind both targets simultaneously, which is necessary for clustering of 4-1BB. PRS-342 was also evaluated for activity in a HepG2 mouse xenograft engrafted with human PBMCs with results supporting its differentiated MoA compared to relevant benchmark controls.

Conclusion: PRS-342 was designed to elicit 4-1BB costimulatory effects in a tumor-localized manner. Here we report potent T-cell activation that is strictly dependent on the presence of GPC3-positive tumor cells. Collectively our in vitro and in vivo data support the continued development of PRS-342.

Citation Format: Birgit Bossenmaier, Corinna Schlosser, Rachida Siham Bel Aiba, Christian Barthels, Benjamin Weiche, Timo Eichner, Michelle Yegres, Shane A. Olwill. Costimulatory T-cell engagement by PRS-342, a GPC3/4-1BB bispecific molecule, leads to activation of T-cells and tumor growth inhibition in a HCC humanized mouse model [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 3268.

Wavefront degradation of a 200 TW laser from heat-induced deformation of in-vacuum compressor gratings

High-repetition-rate high-power laser systems induce a high average power heat deposition into the gold-coated diffraction gratings. To study the effects of the thermal expansion of in-vacuum Pyrex gratings on the laser properties, we scan the pulse energy and repetition rate of a 200 TW laser system while monitoring the laser wavefront. Through the measured changes in laser divergence and focusability, we define an average power limit below which the in-vacuum compressor can be used with no degradation of the laser focus quality.

Frequency-doubled Q-switched 4 × 4 multicore fiber laser system

Frequency doubling of a Q-switched Yb-doped rod-type 4 × 4 multicore fiber (MCF) laser system is reported. A second harmonic generation (SHG) efficiency of up to 52% was achieved with type I non-critically phase-matched lithium triborate (LBO), with a total SHG pulse energy of up to 17 mJ obtained at 1 kHz repetition rate. The dense parallel arrangement of amplifying cores into a shared pump cladding enables a significant increase in the energy capacity of active fibers. The frequency-doubled MCF architecture is compatible with high-repetition-rate and high-average-power operation and may provide an efficient alternative to bulk solid-state systems as pump sources for high-energy titanium-doped sapphire lasers.

Active energy compression of a laser-plasma electron beam

Radio-frequency (RF) accelerators providing high-quality relativistic electron beams are an important resource enabling many areas of science, as well as industrial and medical applications. Two decades ago, laser-plasma accelerators1 that support orders of magnitude higher electric fields than those provided by modern RF cavities produced quasi-monoenergetic electron beams for the first time2-4. Since then, high-brightness electron beams at gigaelectronvolt (GeV) beam energy and competitive beam properties have been demonstrated from only centimetre-long plasmas5-9, a substantial advantage over the hundreds of metres required by RF-cavity-based accelerators. However, despite the considerable progress, the comparably large energy spread and the fluctuation (jitter) in beam energy still effectively prevent laser-plasma accelerators from driving real-world applications. Here we report the generation of a laser-plasma electron beam using active energy compression, resulting in a performance so far only associated with modern RF-based accelerators. Using a magnetic chicane, the electron bunch is first stretched longitudinally to imprint an energy correlation, which is then removed with an active RF cavity. The resulting energy spread and energy jitter are reduced by more than an order of magnitude to below the permille level, meeting the acceptance criteria of a modern synchrotron, thereby opening the path to a compact storage ring injector and other applications.

Evolutionary optimization and long-term stabilization of a white-light seeded two-stage OPCPA seed laser

Ultrafast laser systems, such as optical parametric chirped pulse amplifiers (OPCPA), are complex tools. Optimizing laser performance for a given application is often plagued by intricate couplings between different output parameters, making simultaneous control of multiple pulse properties difficult. Here, we experimentally demonstrate an autonomous tuning procedure of a white-light seeded two-stage OPCPA using an evolutionary strategy to reliably reach an optimized working point. We use the data collected during the tuning procedure to calibrate a performance model of the laser system, which we then apply to stabilize the intricately coupled laser output energy and spectrum simultaneously. Our approach ensures reliable day-to-day operation at optimized working points without manual tuning. We demonstrate shot-to-shot energy stability of <0.18 % rms, in combination with <25 pm rms wavelength stability and <0.2 % rms bandwidth stability during multi-day operation.

Out-of-plane multilayer-dielectric-grating compressor for ultrafast Ti:sapphire pulses

Extreme heat loads on optics, in particular the final pulse compression gratings, are a major hurdle to overcome in the ongoing push towards high average power (kW) and high repetition rate (kHz) operation of terawatt-class Ti:sapphire lasers. Multilayer dielectric (MLD) diffraction gratings have been suggested as a potential alternative to traditionally gold-coated compressor gratings, which are plagued by high energy absorption in the top gold layer. However, to support the required bandwidth (and ultimately the desired pulse duration) with MLD gratings, the gratings have to be operated in an out-of-plane geometry near the Littrow angle. Here, we report on the design of an MLD-based out-of-plane test compressor and a matching custom stretcher. We present a full characterization of the MLD compressor, focusing on its spectral transmission and the significance of laser pulse polarization in the out-of-plane geometry. To demonstrate compression of 40 μJ pulses centered at 800 nm wavelength to 26 fs pulse duration, we use the compressor with an MLD and gold grating configuration, and fully characterize the compressed pulses. Extrapolating our results indicates that MLD-grating-based out-of-plane compressors can support near-transform-limited pulses with sub-30 fs duration and good quality, demonstrating the viability of this concept for kW-level ultrafast Ti:sapphire laser systems.

Reducing wavelength jitter in white-light seeded femtosecond optical parametric chirped-pulse amplifiers

White light generation (WLG) in bulk material can be used as a versatile broadband seed source for optical parametric chirped pulse amplification (OPCPA) stages. In this case, it is beneficial to optimize the performance of the WLG seeder in combination with the subsequent OPCPA stage. Here, we characterize how small variations in the drive pulse energy affect the white light seeder performance, in particular the wavelength stability of the amplified OPCPA spectrum. To isolate the influence of drive pulse energy on the OPCPA central wavelength, we introduce a phase-sensitive amplitude measurement and carefully optimize the drive pulse focus position (at the bulk) to minimize the sensitivity to the jitter of the drive pulse energy. With additional active stabilization of the delay between pump and seed pulses in the OPCPA, we achieve a central wavelength jitter of 2 × 10-4. With this performance, our source is an ideal front-end for applications in laser-plasma acceleration or free-electron laser seeding.

Holistic in silico developability assessment of novel classes of small proteins using publicly available sequence-based predictors

The development of novel therapeutic proteins is a lengthy and costly process, with an average attrition rate of 91% (Thomas et al. Clinical Development Success Rates and Contributing Factors 2011-2020, 2021). To increase the probability of success and ensure robust drug supply beyond approval, it is essential to assess the developability profile of new potential drug candidates as early and broadly as possible in development (Jain et al. MAbs, 2023. https://doi.org/10.1016/j.copbio.2011.06.002 ). Predicting these properties in silico is expected to be the next leap in innovation as it would enable significantly reduced development timelines combined with broader screens at lower costs. However, developing predictive algorithms typically requires substantial datasets generated under very defined conditions, a limiting factor especially for new classes of therapeutic proteins that hold immense clinical promise. Here we describe a strategy for assessing the developability of a novel class of small therapeutic Anticalin® proteins using machine learning in conjunction with a knowledge-driven approach. The knowledge-driven approach considers developability attributes such as aggregation propensity, charge variants, immunogenicity, specificity, thermal stability, hydrophobicity, and potential post-translational modifications, to calculate a holistic developability score. Based on sequence-derived descriptors as input parameters we established novel statistical models designed to predict the developability scores for Anticalin proteins. The best models yielded low root mean square errors across the entire dataset and were further validated by removing input data from individual screening campaigns and predicting developability scores for those drug candidates. The adoption of the described workflow will enable significantly streamlined preclinical development of Anticalin drug candidates and could potentially be applied to other therapeutic protein scaffolds.