Bioactive recombinant human oncostatin M for NMR-based screening in drug discovery

Oncostatin M (OSM) is a pleiotropic, interleukin-6 family inflammatory cytokine that plays an important role in inflammatory diseases, including inflammatory bowel disease, rheumatoid arthritis, and cancer progression and metastasis. Recently, elevated OSM levels have been found in the serum of COVID-19 patients in intensive care units. Multiple anti-OSM therapeutics have been investigated, but to date no OSM small molecule inhibitors are clinically available. To pursue a high-throughput screening and structure-based drug discovery strategy to design a small molecule inhibitor of OSM, milligram quantities of highly pure, bioactive OSM are required. Here, we developed a reliable protocol to produce highly pure unlabeled and isotope enriched OSM from E. coli for biochemical and NMR studies. High yields (ca. 10 mg/L culture) were obtained in rich and minimal defined media cultures. Purified OSM was characterized by mass spectrometry and circular dichroism. The bioactivity was confirmed by induction of OSM/OSM receptor signaling through STAT3 phosphorylation in human breast cancer cells. Optimized buffer conditions yielded 1H, 15N HSQC NMR spectra with intense, well-dispersed peaks. Titration of 15N OSM with a small molecule inhibitor showed chemical shift perturbations for several key residues with a binding affinity of 12.2 ± 3.9 μM. These results demonstrate the value of bioactive recombinant human OSM for NMR-based small molecule screening.


Expression optimization of 6-His-OSM
Our initial attempts to express OSM in BL21(DE3) host were characterized by inconsistent expression despite normal cell growth before and after induction. Specifically, freshly transformed cells were growing well before and after induction with IPTG in LB or defined minimal growth media (M9) reaching high optical density, however, the expression of OSM was unpredictable with yield varying from zero to modest. Well known methods of induction regulation like change of temperature and concentration of IPTG (0.1mM, 0.5 mM, 0.7 mM and 1.0 mM at 18 °C) did not influence the inconsistent expression. Methods reducing basal expression of a toxic protein (pLysS host, 1-2% glucose (labeled as C6 in the figures) in growth medium) proved to be ineffective as well. Based on these observations we excluded OSM toxicity as the source of unpredictable expression and reasoned that misfolding of OSM was causing proteolytic degradation by host proteases. Such proteolytic digestion is especially commonly observed upon overexpression of heterogeneous proteins with multiple disulfide bonds 1 . The M9 minimal media has been suggested to reduce levels of E. coli endogenous proteases, 2 but alone did not improve the OSM expression in our case. However, we noticed that supplementing minimal growth media with high concentration of glucose (150 mM), guaranteed OSM expression in BL21 with moderate yield (Fig. S1A &  S1C). A possible explanation is that here glucose acted as a chemical chaperone stabilizing protein in a folded state. By analogy, osmolytes such as sorbitol and ethylene glycol have been previously utilized in the expression of recombinant proteins to increase an amount of soluble protein in E. coli 3 . As an alternative method we tested autoinduction which afforded OSM in the yield comparable to the expression in M9/glucose (Fig. S1B). Nevertheless, addition of glucose and autoinduction still did not appear as feasible strategies for isotopically labeled protein production. It is noted that while we routine use a low level of algal lysate in large scale growth to decrease the time to induction (Fig. S4), in the initial expression optimizations shown in Figs S1 & S2, algal lysate (Isogro) was not used.

Solubility of 6-His-OSM
To estimate the amount of soluble OSM expressed under different conditions, cell pellets were subjected to three freeze-thaw cycles after which a pellet and a supernatant were separated and analyzed by SDS-PAGE. The relative amounts of pellet and soluble fraction were calculated by densitometric analysis using ImageJ software. The ratio of OSM protein in the pellet:supernatant was estimated to be 2:1 in BL21 M9/C6 (Fig. S2A), and 1:1 in SHuffle LB, SHuffle LB/C6 and SHuffle M9 (Fig. S2B).

Addition of algal lysate decreases time to induction
Minimal defined medium (M9) supplemented with algal lysates at low levels (0.5 g/L) increases the growth rate of E. coli in minimal media (decreases the time to induction). This adds a nominal cost to the overall culture ($30 -$50/L).

Replicate expression and purification of MBP-OSM
We performed expression and purification of rhOSM from rich and minimal defined, isotope enriched ( 15 N, and 13 C , 15 N) media in triplicate. To balance time and cost, the replicates were performed in 100 mL cultures. The yields were lower (2-3 mg/L of culture) than what we routinely obtain from 1 L cultures (8.6 ± 1.9 mg/L of culture over 7 preps; both LB and 15 N ). The loss occurred at the amylose affinity column step, which is clear from the SDS-PAGE gels (Fig. S7). We suspect that this was due to a new batch of amylose resin and smaller bead volumes, which changed the dynamic binding capacity of the amylose resin, leading to breakthrough of the MBP-OSM. What is clear however, is that even when breaking in a new resin, there is low variation in protein yields between replicates.  Input material was loaded onto amylose resin and eluted using a linear gradient of 0 -10 mM maltose. Vertical lines indicate load (where input material was loaded onto the column), wash (where wash buffer was run over the column after input material was loaded), and gradient (where the elution gradient program was initiated). Purified rhOSM peaks elute between 80 and 120 minutes.

Protein long-term stability assay
To identify buffer conditions that would promote OSM stability and prevent autolysis, OSM samples (33 μM) were stored at room temperature in 50 mM sodium phosphate, 100 mM sodium chloride, pH 6.6 with different buffer additives and assessed for degradation by SDS-PAGE analysis at 1 and 2 weeks. Figure S15. Effect of additives on long-term rhOSM stability. SDS-PAGE of rhOSM stored for two weeks at room temperature in 50 mM NaPO4, 100 mM NaCl, pH 6.6 with different additives. Prominent degradation products are indicated at 17 and 7 kDa. Lane labels: MW -MW ladder; No add -no additive; Arg/Glu -50 mM L-arginine/50 mM L-glutamic acid; Arg -50 mM L-arginine; Pro -100 mM proline; His -15 mM L-histidine; Gln -100 mM L-glutamine; Imd -5 mM imidazole.

Chemistry
General Methods. All solvents and chemical reagents were obtained from commercial suppliers and used without further purification, unless otherwise stated. 4,5-dibromofuran-2-carbaldehyde (Compound 1), was purchased from Small Molecules, Inc. or was prepared according to the procedure of Chiarello et al. 5 and 3,4methylenedioxyphenyl boronic acid was obtained from Combi-Blocks. N,N-Dimethylformamide (DMF) was distilled from P2O5 and stored over activated 3 Å molecular sieves. All glassware used in reactions was ovendried (140°C) and cooled in a desiccator, fitted with septa and purged with anhydrous nitrogen immediately prior to use, unless otherwise indicated. All solvents and chemical reagents were transferred with a nitrogen-flushed syringe or cannula, unless otherwise indicated. 1 H and 13 C NMR spectra were obtained on either a Bruker Avance III Ultrashield Plus 600 MHz spectrometer or a Bruker Avance III Ultrashield 300 MHz spectrometer; chemical shifts of synthesized products were measured using residual solvent peaks as reference. High resolution mass spectrometry (HRMS) was conducted on a Bruker Daltonics maXis quadrupole time-of-flight spectrometer. IR data was obtained using a PerkinElmer FT-IR spectrometer with all samples analyzed using attenuated total reflection. Thin layer chromatography (TLC) was completed using 250 μm silica gel glass-backed plates with fluorescence indicator. Flash chromatography was completed using 230-400 mesh silica gel.

Fluorescence Quenching
Fluorescence quenching was used as a method to detect and quantify SMI binding to OSM. The fluorophore, W187, is located at the C-terminus of rhOSM1-187, is the only Trp in the human OSM gene, and is distal to the proposed SMI binding site. In the crystal structure, W187 is packed in the middle of the 4-helix bundle and is in proximity of Tyr10, Arg67, Phe70, Leu 71, Met 128, and Phe184 (Supplemental Fig. S16A). In the presence of increasing SMI, the fluorescence decreases, indicating quenching of W187 (Supplemental Fig.   S16B). Negative deviation from a linear Stern-Volmer plot of (! ! /!) obtained from OSM titrated with OSM-SMI-10B (Supplemental Fig. S16C) illustrates that the quenching of W191 is likely a result of indirect or allosteric effects with a fraction of the fluorophore accessible to quenching or only partial quenching. 4 We interpret this as slight allosteric structural changes due to binding of the SMI at the distal end of the 4-helix bundle (which is