Design

We first sought to construct a plasmid that is compatible with the lactobacillus system, settling on the following vector comprised of a mechanism that is similar to what our lactobacillus species, Lactobacillus plantarum, processes: a two-component regulatory system homologous to the SppK/SppR system. Our goal was to replace the NucA site (constructed for the study Genome-wide analysis of signal peptide functionality in Lactobacillus plantarum WCFS1 by Mathiesen G, Sveen A, Brurberg MB, Fredriksen L, Axelsson L, Eijsink VG) with our IL-10 mutation variations (original, mutation 1, and mutation 2 IL-10 sequences), created through ThermoMPNN and other modeling softwares that can be found here.

Figure 1. A simple design of our plasmid sourced from Addgene, detailing the locations of the sppK and sppR system (Left). The origin of replication of the same plasmid is labeled along with restriction enzymes (Right). Source: pLp_3050sNuc was a gift from Vincent Eijsink & Geir Mathiesen (Addgene plasmid # 122030 ; http://n2t.net/addgene:122030 ; RRID:Addgene_122030)

Further plasmid details are provided that can be found in the link here. Aspects to note include that it (1) is a spp-based expression vector, (2) has a selectable marker of erythromycin, and (3) has a sppA promoter.

Final sequences were submitted into ApE as a visualization of the expected plasmid produced. We created a total of three constructs that include components (aside from the vector backbone) that include: a 3050 signal peptide designed specifically for gene expression in Lactobacillus plantarum, the IL-10 DNA sequence, a V5 epitope, and an 8x His tag.

Figure 2. Visualization original IL-10 plasmid’s DNA sequence on ApE (a plasmid editor)

Figure 3. Visualization mutation 1 IL-10 plasmid’s DNA sequence on ApE

Figure 4. Visualization mutation 2 IL-10 plasmid’s DNA sequence on ApE

Previous Next

Test

After assembling our plasmids, we attempted to transform them into Lactobacillus plantarum to evaluate whether the constructs could be taken up and maintained by the host. However, the transformation did not yield any viable colonies on selective media, indicating that either the transformation was unsuccessful or the plasmids were not maintained in the cells.

Figure 5. Transformation plates showing unsuccessful growth of Lactobacillus plantarum in erythromycin antibiotic. Multiple selective plates were prepared, each divided into three regions corresponding to the original IL-10, M1-IL-10, and M2-IL-10 constructs. After transformation, no bacterial colonies were observed in any of the regions, indicating that the plasmids were not successfully taken up or maintained by the host.

To investigate the issue, we extracted DNA from the attempted Lactobacillus transformants and analyzed it by agarose gel electrophoresis. The gel showed no visibly apparent bands beyond debris corresponding to the expected plasmid size, suggesting that the plasmids had not been successfully introduced into or propagated within the Lactobacillus cells.

Figure 6. Gel electrophoresis image of DNA extracted from attempted Lactobacillus plantarum transformants. The DNA ladder is visible on the right, but no distinct bands corresponding to the expected plasmid size were observed in the sample lanes. Some debris and background signal were captured on the gel, indicating that the transformation did not yield detectable plasmid DNA.

This outcome highlighted a key limitation in our first cycle: while the plasmid assembly itself appeared correct, the delivery of the construct into Lactobacillus required further optimization, such as refining the transformation protocol or reassessing plasmid compatibility with the host.

Build

To assemble our Lactobacillus-compatible plasmids, we first grew up the pLp_3050sNuc backbone to obtain sufficient plasmid for downstream assembly. The vector was then linearized by restriction enzyme digestion at the NucA site, which we targeted for replacement with our IL-10 constructs.

Rather than PCR-amplifying the inserts, we used pre-ordered synthetic DNA fragments for the three IL-10 variants (original, M1, and M2), each designed to include the 3050 signal peptide for Lactobacillus expression, along with a V5 epitope and an 8×His tag. These synthetic inserts were joined with the digested backbone using enzyme digest and Gibson Assembly, chosen for its efficiency in creating seamless junctions between overlapping sequences.

Learn

From this first cycle, we learned that our lack of observable growth and the absence of plasmid bands on the gel left open two possibilities: either the plasmid design or assembly was not fully successful, or the transformation into Lactobacillus plantarum was ineffective. Without clear evidence of plasmid recovery, we could not pinpoint which factor was primarily responsible for the failure.

Given time and budgetary constraints, we determined that additional rounds of troubleshooting in Lactobacillus would delay our progress toward demonstrating proof-of-concept. As a result, we strategically pivoted to using the more tractable E. coli system for Cycle 2, which offered well-established protocols, lower costs, and faster turnaround for testing expression and protein stability. This decision allowed us to focus our limited resources on obtaining interpretable experimental data while keeping longer-term Lactobacillus optimization in view for future work.

Design

Learning from our first iteration, our second cycle consisted of constructs specific to E. coli, which we were more confident in working with. We utilized a pre-existing expression vector, pET-21(+), as the backbone, specifically for its resistance to antibiotic ampicillin. We have attached an image of the vector below:

Figure 7. The pET-21(+) vector as detailed by Twist Biosciences. Components of the plasmid include a lac operator, 6x His tag, T7 promoter, ampicillin resistance gene, and an origin of replication. Source: Twist Biosciences, Expression Vector Catalog

Notes about the plasmid: T7 RNA polymerase driven transcription vector for expression in E. coli. The vector, which lacks the ribosome binding site and ATG start codon, is designed for protein expression from translation signals carried by the cloned DNA. Certain features we chose it for include a lac repressor/lac operator, and can be induced by adding lactose or IPTG.

As one might note, one of the key things the plasmid lacks is a ribosome binding site and ATG start codon. Therefore, aside from the components of our plasmid we incorporated in the first cycle (3050 signal peptide, the IL-10 DNA sequence, a V5 epitope, and an 8x His tag), we also added a start codon and ribosomal binding site.

Below is also a sample construct of our designed gene insert. The SD sequence is our ribosomal binding site, the ATG (start codon) included at the start of the signal peptide and IL10 component, and the V5 epitope and 8x His tag at the end. We should also note that the 8x His tag here also corresponds with a 6x His tag that could be found in the vector, but were advised to simply increase the elution concentration in our washing procedures for a cleaner yield.

Figure 8. Visualization of our construct (starting from ribosomal binding sites to 8x His tag) on ApE; Excludes the backbone.

Build

To test our plasmid, we used a transformation protocol to introduce it into two strains of E. coli, DH5-α and BL21(DE3). Transforming it into DH5-α allowed us to propagate and maintain the plasmid, ensuring stable storage and the creation of a long-term glycerol stock. In contrast, we chose BL21(DE3) as our expression host because it carries a chromosomal copy of the T7 RNA polymerase gene under control of the lacUV5 promoter. This means that when induced with IPTG or lactose, BL21(DE3) cells produce T7 RNA polymerase, which specifically recognizes the T7 promoter in our pET-21(+) plasmid and drives robust transcription of our IL-10 construct. This setup enabled us to pair the cloning efficiency of DH5-α with the high-level protein expression capacity of BL21(DE3), providing an effective workflow for plasmid maintenance and subsequent expression testing.

We used an ampicillin antibiotic to test for bacteria that took up the correct plasmid, verified in the images below.

Figure 9. Pictures depict pET M2 IL-10 plasmid in DH5-α (top left), pET M1 IL-10 plasmid in DH5-α (top center), pET original IL-10 plasmid in DH5-α (top right), pET M2 IL-10 plasmid in BL21(DE3) (bottom left), pET original IL10 plasmid in BL21(DE3) (bottom center), and pET M1 IL-10 plasmid in BL21(DE3) (bottom right).

We next induced expression in verified BL21(DE3) colonies by adding IPTG, preparing the cultures for downstream protein analysis. To do this, individual colonies were first picked and grown in LB medium supplemented with ampicillin until mid-log phase, at which point IPTG was introduced to initiate expression of our IL-10 construct. These induced cultures were subsequently harvested for assays to confirm and evaluate protein production.

Test

To assess whether our IL-10 constructs were expressed in E. coli BL21(DE3), we analyzed both total cell lysates and culture supernatants from induced and non-induced cultures using SDS-PAGE followed by anti-His Western blotting. The Coomassie-stained SDS-PAGE gel (Figure 8) shows the overall protein profiles for the samples. Lanes loaded with induced lysates (lanes 3, 5, 7) and supernatants (lanes 9, 11, 13) did not display an obvious new dominant band compared to their non-induced counterparts, suggesting that recombinant IL-10 was not highly abundant in the crude lysate or secreted fraction at detectable levels by Coomassie staining alone.

Figure 10. SDS-PAGE analysis of IL-10 expression in BL21(DE3). Coomassie-stained 4% stacking / 15% resolving SDS-PAGE gel comparing total lysates (lanes 2–7) and culture supernatants (lanes 8–13) from BL21(DE3) expressing wild-type or mutant IL-10, with or without IPTG induction. Lanes 14–15 contain BSA standards. No prominent new band corresponding to IL-10 (~18 kDa) is apparent in induced samples relative to controls.

The Western blot probed with an anti-His antibody (Figure 9), however, revealed a weak but specific signal at approximately the expected size of IL-10 (~18 kDa) in the induced lysate lanes compared to non-induced controls, consistent with low-level expression of the His-tagged IL-10 variants. The absence or faintness of signal in the supernatant lanes suggests that most of the recombinant protein remained intracellular or was produced at levels below the blot’s detection limit in the culture medium. The strong positive bands observed in lanes 14–15 (purified mCherry-His controls) validated the antibody and detection system. These results together indicate that the constructs were expressed but at relatively low levels under the tested conditions, motivating further optimization of induction parameters and potentially the inclusion of secretion-enhancing elements.

Figure 11. Western blot detection of His-tagged IL-10 in BL21(DE3). Blot probed with mouse anti-His (1:5000) and HRP-conjugated goat anti-mouse secondary antibody. Weak but distinct bands near ~18 kDa were detected in induced lysate lanes, while supernatant lanes showed minimal signal. Purified mCherry-His in lanes 14–15 served as a positive control.

Learn

From this cycle, we learned that while our constructs were successfully introduced and expressed in BL21(DE3), the detected protein levels were relatively low and largely confined to the intracellular fraction. The weak Western blot signal and absence of strong bands in the Coomassie gel suggested that the constructs were functional but sub-optimally expressed under the tested induction conditions.

Although these results highlighted the potential benefit of reconstructing or refining the plasmids (e.g., adding stronger expression or secretion elements), we did not have sufficient time during this cycle to redesign and rebuild new constructs. We kept this consideration in mind as we moved forward. Instead, we prioritized growing an additional batch of induced BL21(DE3) cultures to harvest sufficient protein for downstream testing, focusing on evaluating protein stability using ELISA assays.

Design

In our third cycle, we shifted our focus from plasmid construction to engineering more stable forms of IL-10 to incorporate into our plasmid (so in a sense it is in lieu with our cycle 2). Our goal was to design IL-10 variants that could retain their structure and function under harsh temperature conditions.

We began by selecting a crystal structure of IL-10 from the Protein Data Bank (PDB ID: 1ILK) as our reference model. This 3-D structural template allowed us to visualize the native fold and the interface of the two subunits that make up the active dimer.

Figure 12. Crystal structure of human IL-10 homodimer (PDB ID: 1ILK). The two identical subunits of IL-10 are shown in green and orange ribbon representation. Each monomer contributes to the characteristic intertwined four-helix bundle, forming the biologically active homodimer. This structure, determined by X-ray crystallography at high resolution, served as the starting template for identifying and modeling thermostabilizing point mutations in our design cycle. Source: RCSB Protein Data Bank, https://www.rcsb.org/structure/1ILK

To search for stabilizing mutations, we used ThermoMPNN, a machine-learning tool that predicts how point mutations influence a protein’s thermal stability (reported as ΔΔG, where negative values indicate improved stability). We uploaded the IL-10 structure to a Google Colab implementation of ThermoMPNN, toggled off the “include cysteines” setting to avoid disulfide-bond misfolding, and tested the multichain-interface option to ensure we preserved dimer integrity. Out of nearly 7,000 predicted single-amino-acid changes, about 1,000 were predicted to increase stability, becoming the focus of our analysis.

Figure 13. ThermoMPNN stability heatmap for IL-10 (PDB ID: 1ILK). Predicted ΔΔG values (kcal/mol) for all single-amino-acid substitutions. Blue indicates stabilizing mutations, while red indicates destabilizing mutations. Clusters of stabilizing sites were notably observed around residues 32-38 and 133-139, guiding candidate selection for thermostable IL-10 variants.

We next used PyMOL to screen the top ~400 mutations, removing those that caused steric clashes or disrupted key hydrogen bonds, salt bridges, or dimer geometry. PyMOL also allowed us to confirm that favorable interactions (such as hydrophobic packing and receptor-binding contacts) were preserved. Because IL-10’s activity depends on binding its receptors (IL-10RA and IL-10RB), we used AlphaFold3 to ensure our candidate mutations did not affect receptor engagement. The modeled complexes showed interface scores (ipTM) similar to wild type, indicating that binding affinity was maintained.

We then combined several top mutations into two final designs and evaluated them with the ESM protein language model, which predicted slightly lower folding free energies (ΔG ≈ 4.26 vs. 4.36 kcal/mol for wild type), suggesting improved thermostability.

Overall, by integrating PyMOL, AlphaFold3, and ESM, we developed two IL-10 variants predicted to resist unfolding at higher temperatures while retaining receptor function. These designs guided the constructs tested in later wet-lab cycles. More about protein design can be found in our Dry Lab subcategory.

Build

For the third cycle, we focused on producing the newly designed IL-10 variants in the same bacterial expression system used in the previous cycle. We transformed the plasmids carrying the two selected mutant IL-10 sequences into E. coli BL21(DE3) cells, the standard strain for T7 promoter–driven expression. Successfully transformed colonies were grown in LB medium supplemented with ampicillin to select for plasmid maintenance.

To induce protein production, cultures were grown to mid-log phase and then treated with IPTG, which activates T7 RNA polymerase in BL21(DE3) and drives transcription of the IL-10 constructs. After induction, cells were harvested for downstream analysis and planned stability testing of the mutant proteins. This approach mirrored the procedure established in Cycle 2.

Test

To assess the yield and apparent stability of our expressed IL-10 variants, we collected both the cell lysate and the culture supernatant separately from induced BL21(DE3) cultures expressing original IL-10, M1-IL-10, and M2-IL-10. The cells were first disrupted using a continuous-flow cell disruptor to efficiently release soluble intracellular protein, and the lysate was clarified by centrifugation to remove debris.

Figure 14. Collected supernatant fractions from induced BL21(DE3) cultures expressing IL-10 variants. Shown are the clarified supernatant fractions from cultures expressing original IL-10 (O-IL-10), M1-IL-10, and M2-IL-10, kept on ice prior to purification. The corresponding cell lysate fractions (not pictured here) were collected in similar 50 mL tubes and resuspended in PBS for parallel purification and downstream ELISA analysis.

All three constructs supernatant and lysate were then subjected to nickel-affinity purification using the NEB Ni-NTA bead protocol. The His-tagged IL-10 proteins bound to the resin, were washed to remove non-specific contaminants, and eluted with an imidazole-containing buffer optimized for recovery. Due to our plasmid construct containing greater amounts of His tags, we eluted with a solution containing greater imidazole than the protocol recommended. To examine thermal stability, the purified eluates were split into two aliquots and incubated at 37 °C and 55 °C for 10 minutes before proceeding to quantification.

Following this treatment, all samples were analyzed using a commercial human IL-10 ELISA kit to quantify protein recovery and to evaluate the relative thermal stability of the wild-type and mutant IL-10 constructs. This approach provided a quantitative comparison across both expression localization (lysate vs. supernatant) and construct performance under mild heat stress. By comparing ELISA signals across these conditions, we were able to begin assessing whether our engineered mutations improved IL-10 recovery and stability.

Figure 15. IL-10 ELISA plate showing color development prior to (left) and after addition (right) of stop solution. The standard curve wells are the furthest column on the left, showing decreasing concentrations of IL-10. The remaining wells contain the lysate and supernatant samples from original IL-10, M1-IL-10, and M2-IL-10 after purification and thermal treatment for two batches in triplicates. The darker the color, the more present the protein.

Figure 16. Raw absorbance data from IL-10 ELISA. Image of the Gen5 plate-reader output showing the optical density (OD_{450}) values for the IL-10 ELISA plate. These raw OD_{450} values were used to generate the standard curve and calculate relative IL-10 concentrations for downstream analysis.

Learn

This cycle was successful, as we were able to express, purify, and detect our IL-10 variants using ELISA after nickel-affinity purification. Importantly, the workflow (beginning with separate collection of lysate and supernatant, followed by thermal treatment and quantitative ELISA) validated our ability to not only recover the protein but also assess its stability under different conditions. The quantitative outcomes from the ELISA are presented in the results tab, providing key insights into the relative performance of the wild-type and mutant IL-10 constructs.