Category: BLOG

Working with very large or multi-domain proteins can feel like wrestling a spaghetti monster – they’re floppy, prone to aggregation and cellular degradation and thus often low in yield.

A classic structural biology trick for taming such divas is “divide and conquer.” Rather than expressing a hundreds of kDa beast in one piece, you might break the gene into domains or functional modules and express those separately. Truncated constructs can ensure higher stability during expression which dramatically improves protein yield. On top of that, removing unstructured regions or internal flexibilities often gets you a more tractable, crystallizable protein.

And as a bonus, well-behaved protein fragments often crystallize when the full-length protein would not, yielding valuable structural insights.

Read on LinkedIn

The phenyl ring is electron rich with continuous density from the delocalized π-system. The side-on view with a clear planarity.

The piperidine ring is saturated, which is reflected by the electron density map. The chair conformation is clearly resolved in the crystal structure.

The imidazole ring has only a bump of its ring, but it is resolved at this resolution.

The solved structures are part of our internal MAGNET platform development activities. The molecules have been designed by our AI engine for fragment evolution and we solved them using an established SmartSoak system.

Read on LinkedIn

PROTACs have rapidly matured into a leading modality as proximity therapeutics, leveraging the ubiquitin–proteasome system to eliminate proteins by enforcing proximity between an E3 ligase and a protein of interest. Mechanistically, their activity is governed not only by binary affinities, but by the formation, stability, and productive geometry of the ternary complex.

In vitro, however, activity and tractability often hinge on a deceptively simple step – formation of a productive E3-PROTAC-target ternary complex. Even when binary binding is detectable, ternary assembly can be limited by cooperativity, kinetic barriers, conformational heterogeneity, and competition from non‑productive binary species. As a result, apparently “small” experimental choices (ratios, order of addition, incubation time, buffer composition, and purification strategy) can dominate whether you recover a discrete, well‑behaved complex.

For one of our incubation set-ups we found that assembling the reaction with target protein and PROTAC in defined molar excess (4x in the example shown) increased the population of the ternary species captured by SEC. Equally important was column selection: matching the SEC matrix to the hydrodynamic size of the complex improved resolution and enabled efficient removal of excess unbound protein. This substantially improved recovery and sample homogeneity.

Read on LinkedIn

We have highlighted one of our previous posts we optimized the yield in protein production by 330% compared to published literature: https://lnkd.in/dmrgTqpp

VCB serves as a critical component in PROTAC design by recruiting the target proteins via VHL. Formation of a ternary complex involving the target protein is required for targeted protein degradation (TPD). There is a need for novel chemical matter for VHL recruitment. Enhanced molecular characteristics of next-generation VHL binders could broaden the efficacy of VCB-mediated TPD.

After crystallizing the protein, a SmartSoak system was set up. We observed that SmartSoak improves the diffraction of the crystal by 0.5 Å. We SmartSoak’ed an in-house fragment library and identified novel binders. Four binders at EloC, one binder at the EloC-EloB (cryptic pocket) and three binders at VHL (cryptic and orthosteric).

Interestingly, two fragments bind at the cryptic pocket of VHL, inducing a more pronounced opening of the pocket compared to the published structures. The two fragments are well-positioned for linking.

Read on LinkedIn

…over the years, we have handled protein crystals in many different morphologies and behaviour.

Working successfully across such a diversity takes broad expertise, automation and reliable technology.

By successfully applying SmartSoak® technology in high-throughput co-structure determination workflows, we generate the structural data that helps medicinal chemistry and structure-based drug design move faster.

Our experience spans across:
– diverse target classes: E3 ligases, transcription factors, GTPases, helicases, phosphatases, and kinases,
– different binding mechanisms: non-covalent and covalent,
– different type of ligands: small molecules, macrocycles, molecular glues, and PROTACs.

Crystal diversity is part of the job – turning it into actionable structural insight is where expertise matters.

Read on LinkedIn

Even with the right construct in the right host, induction conditions can be the deciding factor between soluble expression and insoluble product. A common pitfall is inducing with a high IPTG concentration and incubating at 37 °C, which can drive expression faster than the cell can fold the protein properly, often increasing aggregation.

In practice, “less is more” is frequently true. Inducing at lower temperature (e.g., 18–20 °C overnight) slows translation and can improve folding, boosting soluble yield. IPTG concentration is another key variable: in many cases, lower levels (e.g., ~0.05–0.2 mM rather than 1 mM) are sufficient to initiate expression while reducing cellular stress. The induction point matters as well – inducing at mid-log phase (OD₆₀₀ ~0.6) versus later can impact both yield and protein quality.

A strong alternative is auto-induction media, which gradually induce expression as cultures grow to higher density, which is useful for obtaining good yields with minimal intervention and for screening conditions. Finally, don’t overlook aeration and shaking speed: oxygen transfer (flask-to-volume ratio, baffling, rpm) can significantly influence bacterial growth and expression performance.

Tuning these parameters takes some iteration, but the payoff can be substantial: higher soluble yields and better-behaved protein.

Read on LinkedIn

Covalent campaigns don’t usually stall because the compounds are not active. They stall because the co-structure never arrives or the resolution is not sufficient to resolve the binding mode.

In our example, protein crystals of a human reductase are formed under acidic conditions, the  cysteine is protonated and not reactive. Changing the environment of crystals towards higher pH for soaking dissolves crystals in conventional set ups.

In this case, SmartSoak was applied and a high-throughput soaking system was established at a pH which facilitates cysteine reactivity. The ligands appeared in the active site indicating a novel mode of action at 1.3 Å resolution.

Our patented SmartSoak technology delivers co-structures reliably and with short turnover cycles. It has been successfully applied for covalent and non-covalent campaigns, for fragments, leads, drug candidates, peptides and macrocycles.

Read on LinkedIn

Got a crystal hit? 💎
Now the real work begins – optimizing that condition to grow larger, well-diffracting crystals. Start by varying one factor at a time around the hit: perform grid screens  of precipitant, salt, buffer and protein concentrations as well as pH in a finer range around the initial hit point.

For example, if a crystal appeared in 20% PEG 3350, 0.1 M HEPES pH 7.5, try 15–25% PEG3350 and pH 7.0–8.0 in small increments. Small tweaks can have big effects – even 0.1 pH units or a few percent precipitant can turn crystals on or off. Also consider additive screens (small molecules or ions that might stabilize the crystal lattice) and try different drop ratios or seeding (if crystals are small, rare or have non-optimal morphology). Document every tweak – optimization is an iterative, data-driven process. If you have different morphologies, then always remember that you can’t judge a book by it’s cover. Sometimes the ugliest crystals might exhibit the best diffraction. Therefore it is important to not just optimize “by eye” but also check diffraction quality during the optimization.

Read on LinkedIn

Even a perfectly designed construct can underperform if the protein has “hidden” features that complicate expression. It’s wise to scout your protein sequence for any post-translational modification (PTM) sites or targeting signals before you start the experiment.

Glycosylation sites, for example, won’t be processed in E. coli, which could lead to misfolding or formation of inclusion bodies for some eukaryotic proteins. Signal peptides or transit peptides might misdirect a protein in an unexpected way if you’re using a heterologous system.

One dramatic example is the presence of a GPI-anchor signal: a small motif (often a hydrophobic stretch at the extreme C-terminus, with a preceding glycine as a linkage site) that, in mammalian cells, will covalently attach your protein to the cell membrane. If you’re expressing such a protein intending for it to be secreted, a GPI anchor will thwart you by tethering it to the cell surface.

The good news is these issues can often be fixed by slight modifications: remove or mutate a signal sequence or anchor site, or choose a host that can handle the PTM. Modern bioinformatics tools (UniProt databases, signal peptide predictors, etc.) are your allies – they can flag N-glycosylation motifs, phosphorylation sites, protease cleavage sites, and more. Being forewarned allows you to engineer around the problem (for instance, mutating certain residues to prevent a modification) before you waste weeks on a non-expressing construct.

Take-Home Points
1. Analyze the sequence upfront: Identify motifs like glycosylation sites, GPI-anchor signals, transmembrane helices, or low-complexity regions that might affect expression or solubility.
2. Plan construct design accordingly: If a problematic motif is non-essential, consider removing or altering it.
3. Use the right host: When PTMs are desired, use a system that provides them (yeast, insect, or mammalian cells); when they’re undesired, ensure your strategy mitigates their impact.

Read on LinkedIn

It is usually assumed that buffer components are inert – just there to keep pH stable. Sometimes the buffer is not just buffering. Here’s an example of a buffer bound right in the protein’s active site. It showed up clear as day in the electron density map.

This finding underscores that buffers aren’t always passive bystanders. Any buffer component could essentially act like an weak or competitive inhibitor – occupying the protein functional hotspot where ligand would bind. It’s a reminder that buffer chemicals can directly interact with proteins.

Takeaway: Don’t always assume your buffer is inert. Nothing in our experiments should be taken for granted as “just background.” Buffer components can influence protein structure and function – potentially affecting their behavior in assays or how and how strong a ligand binds.

Read on LinkedIn