Cloning of La BP

To produce La BP for the Curli-BP system, we followed a streamlined cloning workflow, which was subsequently applied to all other REE:

PCR Amplification

The La BP coding sequence and the pET28a backbone were amplified separately using primers designed for Gibson assembly.

Figure 2. Agarose gel showing amplified backbone (~5.3 kb) and La BP (8DQ2 ; ~750 bp). 5kb plus DNA ladder (100–5000 bp range) was used as a size reference.

Gibson assembly and transformation

PCR products were assembled via Gibson assembly and transformed into E. coli DH5α. Correct assembly of La BP constructs was confirmed by sequencing of selected DH5α colonies.

Figure 3. Sequencing verification of La-BP in DH5α. Sequencing of the La binding protein construct confirmed that the full-length gene, including the DogCatcher fusion and backbone elements, was correctly assembled in DH5α. No errors or unexpected mutations were observed, ensuring that the construct is suitable for transfer to the protein production strain and subsequent expression.

Verified plasmids were then transferred into E. coli BL21(DE3) for protein expression. Glycerol stocks of the working strain were prepared to enable subsequent functional testing.

This workflow ensured reproducible assembly of La BP and provided a robust starting point for downstream protein expression and testing in the Curli-BP system.

For the first pilot experiment, we prepared both Curli fibers and binding proteins using E. coli BL21. At that time, we only had Gd BP lysate available, so we tested the system with Gd BP and metal. In all subsequent experiments, however, the system was tested with the Lanthanum-binding protein and La metal. The components were mixed in PBS and incubated for three hours with shaking, a duration optimized for DogTag–DogCatcher interactions to covalently attach BPs to CsgA subunits (Zakeri et al., 2012). Once assembled, the system and appropriate controls were exposed to different Gd concentrations and incubated for 30 minutes at room temperature. After incubation, the samples were pelleted and the supernatant collected for the first measurement, allowing determination of the remaining REE in solution and thus the amount captured. To quantify the remaining metal, we used the Arsenazo III assay, where the dye forms colored complexes with REE under acidic conditions that can be measured spectrophotometrically. The detailed protocol and results are described in the Results section.

Figure 4. Gadolinium remaining in the supernatant after exposure to and capture by the Curli-BP system at varying Gd³⁺ concentrations (5–50 µM). Measurement of Gd³⁺ remaining in the supernatant following incubation with the Curli-BP system in this pilot experiment. The figure includes control samples with Curli + BP without metal and metal-only solutions, highlighting systematic background absorbance observed in the no-metal control. Only one replicate per sample was included in this pilot experiment.

In this pilot experiment, we observed an unexpected issue in the control condition (Curli + BP without metal), which exhibited an apparent absorbance corresponding to ~40 µM of metal, as shown in Figure 4. Furthermore, the Curli + BP samples incubated with 50 µM metal showed absorbance values closer to ~100 µM, whereas the expected outcome would have been a decrease relative to the control. By contrast, the metal-only controls approximately matched the expected concentrations, indicating that the issue did not arise from the metal dilutions. Overall, these data show that Curli + BP samples systematically displayed elevated absorbance. After accounting for the baseline absorbance of the no-metal control, the estimated capture values were ~45 µM for the 50 µM condition and ~10 µM for the 5 µM condition, essentially reflecting the background signal from the control. This strong bias prevented us from reliably interpreting the results of this experiment.

We identified a source of false positives, most likely due to residual unbound binding proteins or other components present in the cell lysate that interfered with the Arsenazo assay. To address this, additional washing steps after the 3 h Curli-BP incubation and before exposure to REE should be introduced to remove this bias. This finding emphasized the need for systematic step-by-step optimization of each protocol stage in order to minimize artifacts and establish reliable conditions for REE capture and release measurements.

The following sections summarize all the iterative cycles undertaken to optimize our protocol, improving system efficiency while enhancing workload management, time efficiency, sustainability, and the precision of our results.

To perform this test, curli fibers and binding proteins were combined in experimental and negative control tubes for 3 h, before direct exposure to the metal. All samples were prepared in 1X PBS, which was maintained as the buffer throughout the experiment. Negative controls contained only curli fibers and binding proteins, while positive controls contained only La³⁺, serving as a reference for the arsenazo assay.

After the initial incubation, samples were washed three times. La³⁺ was then added to reach the desired final concentrations, including to positive controls, which were topped up with PBS to achieve the correct volume and concentration. All tubes were incubated for an additional 30 minutes to allow the binding proteins to capture free La³⁺ ions.

Finally, tubes were centrifuged, and 470 μL of supernatant was collected from each for absorbance measurement using the arsenazo assay.

During the first measurement of this experiment, the positive control triplicates (REE only) at 125 µM showed very low to undetectable absorbance at OD₆₅₀ (0–0.002), far below the expected value of ~0.700. The same pattern was observed across all other conditions, indicating that a step in the protocol compromised the results, rendering the arsenazo assay unable to detect the metal.

To test our hypothesis, we prepared triplicates of tubes containing only La³⁺ dissolved in PBS and incubated them for different time periods (0 min, 30 min, and 1 hour). As a control, we prepared the same set of tubes but diluted the La³⁺ in water instead of PBS. To specifically test for a centrifugation bias, we also made a second triplicate of the 30-minute PBS samples and subjected these to centrifugation before measuring absorbance.

Figure 5. Absorbance of La³⁺ (125 µM, measured with Arsenazo III) "in PBS", "in water" after incubation and centrifugation. Absorbance at OD₆₅₀ was measured after 0, 30, and 60 minutes of incubation. An additional condition was included at 30 minutes with centrifugation prior to measurement. Only one replicate per sample was included in this experiment.

Figure 5 revealed that in PBS, the 30-minute samples subjected to centrifugation showed almost no absorbance, whereas the non-centrifuged 30-minute samples displayed high absorbance. This effect was not observed in the water controls. Furthermore, PBS samples showed a marked decrease in absorbance over time, while water samples exhibited only a slight decrease.

We compared Tris, Tris-HCl, TBS, HEPES, saline water, and MilliQ water (control) using the same setup as PBS versus water, with 50 µM La³⁺.

Figure 6. Absorbance of 50 µM La³⁺ solutions in various buffers after incubation and centrifugation. Absorbance at OD₆₅₀ was measured at 0 and 30 minutes of incubation, with an additional 30-minute condition subjected to centrifugation prior to measurement. Tested buffers included Tris, Tris-HCl, TBS, HEPES, saline water, and MilliQ water, allowing comparison of buffer effects on absorbance stability and consistency. Only one replicate per sample was included in this experiment.

HEPES clearly stood out: it gave consistent absorbance values across time points and remained stable after centrifugation, making it the best choice for the interaction and capture steps. Tris was stable but yielded lower absorbance.

As HEPES appeared to be a promising candidate, we tested using it throughout the entire workflow (lysis, DogTag-DogCatcher interaction, washing) to determine whether it could maintain pellet stability and reliable measurements, which would also simplify the protocol by requiring only a single buffer.

Figure 7. Lanthanum capture and recovery by the Curli-BP system at varying Curli:BP ratios (250 µM La³⁺) using HEPES buffer throughout the experiment. Panel A: Lanthanum remaining in the supernatant after exposure to and capture by the Curli-BP system. Conditions included Curli alone with and without REE, and Curli-BP at ratios of 1:100, 1:250, and 1:500. The figure shows the La³⁺ concentration in the supernatant following incubation with the Curli-BP system under different Curli:BP ratios. Panel B: Lanthanum recovered into the supernatant from the pelleted Curli-BP system after ion release using low-pH buffer (pH 2.4). Four conditions are shown: Curli alone without REE, and Curli-BP at ratios of 1:100, 1:250, and 1:500. Low-pH HEPES and low-pH MilliQ water were compared, showing no difference in release efficiency. Error bars represent the mean ± standard deviation of three technical replicates, except for the Curli-alone condition, which was tested once.

In Figure 7 panel A, when HEPES buffer was used from the start instead of PBS, the pellets became fragile and captured measurements were unreliable, with higher BP amounts occasionally resulting in lower apparent binding. However, in panel B, the release assay displayed a consistent trend, with increased BP proportion corresponding to greater metal capture, although overall capture was lower than in the standard workflow. For the release step, low-pH HEPES (pH 2.4) and low-pH MilliQ water (pH 2.4) were compared, showing no difference in release efficiency.

Figure 8. Lanthanum remaining in the supernatant after exposure and capture by the Curli-BP system (50 µM La³⁺) at varying Curli:BP ratios (1:25 –1:1000). La³⁺ remaining in the supernatant after 30 minutes of incubation with the system. Conditions include REE only, Curli alone with and without REE, and Curli + BP 1:25-1:1000 with and without REE, illustrating the fraction of metal not captured by the system. Error bars represent the mean ± standard deviation of three technical replicates, except for the Curli-alone condition, which was tested in duplicates.

The results show that nearly all La³⁺ was removed from the supernatant after exposure to the Curli-BP system, indicating effective capture of the metal ions. Curli fibers alone also bound some La³⁺, but only a small fraction compared to the complete system.

Figure 9. Lanthanum recovered into the supernatant from Curli-BP system after ion release at varying ratios (200 µM La³⁺, Curli:BP ratios 1:10-1:1000). Measurement of La³⁺ released into the supernatant after exposure of pelleted Curli-BP system to a low-pH buffer (pH 2.4). The figure includes multiple Curli:BP ratios (1:100, 1:500, 1:1000), illustrating the experimental setup used to assess recovery of bound metal ions at higher initial REE concentrations. Error bars represent the mean ± standard deviation of three technical replicates.

These results demonstrate that increasing BP proportions enhances metal ion capture at higher La³⁺ concentrations, confirming that binding is primarily driven by the BP rather than the curli fibers. This distinction is important, as our focus is on the binding capacity of the BP. Between ratios of 1:500 and 1:1000, the amount of REE released from the system reaches a plateau, suggesting the system may be approaching saturation.