Key Points

• NRPS Library Characterization: Characterising all 150+ parts of our Part collection via high-throughput LC-MS.

• Integrated Chemical & Functional Analysis: Combined LC-MS, HPLC, NMR, and bioactivity assays for thorough validation.

• Structural Confirmation: Purified peptides analyzed by HPLC and 1D/2D NMR to confirm identity and purity.

• Bioactivity Screening: Evaluated antimicrobial potential of all library peptides, identifying new-to-nature hits.

• Module Insights: MS² analysis revealed unexpected amino-acid incorporation, informing module functionality.

• Accessible Workflow: Developed a handbook adapting the screening workflow for labs with standard equipment.

Graphical Abstract

Created in BioRender. Solakoudi, E. (2025) https://BioRender.com/z53ogfu

Precision Measurement for NRPS Biosynthesis: An Integrated Analytical Framework

Accurate measurement is fundamental to synthetic biology. It drives the Design–Build–Test–Learn (DBTL) cycle and makes data reliable and comparable across different labs. This precision is vital in natural product discovery, where tiny structural changes can completely change a molecule's biological function.

Older screening methods (like bioassays, TLC, HPLC-UV)^[1][2] lack the necessary sensitivity and specificity to analyze the low-yield, complex mixtures characteristic of microbial production. To address this, the field has increasingly adopted Mass Spectrometry (MS)-based techniques^[3].

Natural products produced by Non-ribosomal Peptide Synthetases (NRPS) are inherently challenging to measure because engineering them often reduces yields while the host continues producing wild-type compounds alongside truncated or side products. Combined with their broad molecular diversity across multiple compound classes, these mixed and low-abundance products are difficult to detect and distinguish in complex cellular extracts.^[4].

Our Robust Measurement Strategy

For a robust and precise assessment of NRPS system performance, we adopted a framework that leverages three complementary, sequential techniques.

By combining this chemical framework with standardized bioactivity assays, we have created a complete and highly reliable strategy for assessing our engineered biosynthetic pathway.

Uncovering the Library’s Production Profile

Our project focuses on a modular NRPS library designed to produce a wide range of novel peptides. Each NRPS unit is responsible for selecting and adding a specific amino acid to the growing chain, determining the sequence and properties of the final product. Measuring these output peptides was our main goal, as their presence directly reflects the proper function of the engineered NRPS constructs.

Confirming peptide production was essential not only to validate the functionality of each hybrid NRPS but also to guide downstream steps such as purification, structural analysis, and bioactivity testing. Given the diversity of variants in the library, we required a high throughput analytical approach to efficiently detect and verify the predicted products. This strategy provided a clear picture of library performance and helped identify promising candidates for further study.

LC-MS/MS Baseline Validation

Before selecting samples for detailed downstream analysis, we first validated the LC-MS screening workflow using multiple controls: blanks (MeOH:ACN), extraction blanks (solvent processed through the full extraction workflow), negative controls (NC) (media without cells), empty-vector controls (both uninduced and induced), and uninduced cultures of the NRPS constructs. A positive control (PC), using a known peptide-producing NRPS variant, confirmed that the LC-MS system and extraction process were functioning as expected (Fig. 1).

Chromatograms from blanks, extraction blanks, and media-only controls showed no detectable peptide peaks within the expected retention-time window, while the positive control exhibited clear signals with the predicted masses 644.30 and 721.33. Uninduced cultures displayed only low-level background peaks unrelated to the target peptides. Based on these results, we concluded that uninduced cultures serve as a reference to identify peptide-specific signals enabling confident high-throughput LC-MS screening across the entire NRPS library.

🔍

To exemplify our technical validation, we selected a native NRPS library variant (Chaiyaphumine), predicted to produce the peptide PAA-T-f-a-P-W (Fig. 2), representing the typical behavior of our library and used here to illustrate our standardized measurement workflow.

🔍

Standardized Protocol

This standardized workflow, applied across the entire NRPS library, was followed here for the selected native variant (Chaiyaphumine) to ensure reproducible sample preparation and reliable LC-MS analysis (general production parameters are detailed in 'Characterization our Parts').

1. Precultures: Three plasmids containing the starter, elongation, and termination units of the NRPS native variant were triple-transformed and grown overnight in 5 mL LB medium.

2. Cultivation: The culture was inoculated in 10 mL XPP3 production medium, supplemented with all necessary precursors, cofactors and XAD 16N beads for capturing and concentrating the peptides.

3. Beads Harvest: After a 72-hour production phase at 25 °C and 200 rpm, the production beads were harvested.

4. Extraction and Clarification: Peptides were isolated using a high-efficiency solvent extraction with 10ml MeOH:ACN (1:1), followed by centrifugation at 4000rpm to remove cell debris.

5. Final Extract Preparation: 100ul from the resulting purified extract was subsequently transferred to LC-MS vials and prepared for LC-MS analysis.

This standardized preparation minimized experimental variability and ensured that each extract was in an optimal state for high-fidelity LC-MS detection and accurate mass analysis.

For screening the entire NRPS library, we developed a miniaturized 4 mL high-throughput version of this expression protocol, enabling parallel production in deep-well plates (see High-Throughput Expression Protocol).

🔍

LC-MS/MS Validation

To verify the functional output of our NRPS library, extracts were analyzed using Liquid Chromatography-Tandem Mass Spectrometry (LC-MS/MS), combining high-resolution separation (LC) with sensitive mass detection (MS/MS). This approach enabled rapid screening of library variants, detection of minor analogs, and reliable peptide confirmation. For most variants, MS¹ data was sufficient to confirm production. However, one candidate displayed an unexpected signal, which we subsequently examined using MS² to unambiguously identify the incorporated amino acids.

Native Cluster (Chaiyaphumine) LC-MS/MS Analysis

In the base peak chromatogram (BPC) for the expression of Chaiyaphumine, the two most prominent peaks at the retention times of 7.3 min and 8.9 min (labelled 1 and 2 in Fig. 4a) immediately stands out. Inspecting the mass spectra taken from the fractions at these retention times reveals the prominent masses 645.22 and 721.26, corresponding to the [M+1] peaks of expected products Chaiyaphumine D and Chaiyaphumine A (Fig. 4b). To streamline this analysis process and to make sure that we do not miss low-yield products, we used expected ion chromatograms (EICs) instead of manually analyzing the prominent peaks. EICs only show a peak when a specific mass is detected. For this experiment, the EIC for the mass 645.3 ± 0.3 only shows the peak at 7.3 min and the EIC for the mass 721.3 ± 0.3 only shows the peak at 8.9 min (Fig. 4c).

🔍

To enable high-throughput analysis, we developed a custom script that automatically predicts product structures and calculates their expected masses for all peptides synthesized in an experiment. We also used the batch processing function of the LC-MS analysis software to automate the creation of EICs for these expected masses. This significantly accelerated peak identification and made large-scale screening of the NRPS variants more efficient and consistent.

Interpretation

LC-MS analysis confirmed that the predicted Chaiyaphumine peptide was successfully produced by the NRPS module combination, demonstrating both the functionality and reliability of the NRPS module and thereby the hybrid NRPS. The LC-MS workflow proved to be highly specific and sensitive, enabling clear distinction between closely related Chaiyaphumine variants and providing sufficient resolution to screen all library members. Notably, one variant exhibited an unexpected mass signal, prompting further MS² analysis for unambiguous structural confirmation.

MS² Analysis: Resolving a Special Case

While MS¹ screening efficiently confirmed peptide production for most library variants, one candidate displayed an unexpected mass signal that raised the possibility of altered amino acid incorporation. Inspection of the MS² spectrum of this candidate allowed us to interpret the unexpected signal.

Sample and Preparation

As a part of our library screening, we investigated an NRPS unit predicted to incorporate glutamine (BBa-25JQ5Y0E). Insertion of this unit into the XUT4 position of the Chaiyaphumine synthetase should yield the peptides C2-T-f-a-Q-W and PAA-T-f-a-Q-W (Fig. 5).

🔍

LC, MS¹ and MS² Analysis

Analysis of the chromatogram as described above showed flat EICs for both predicted product masses, indicating that neither of the expected products was formed (Fig. 6). However, the BPC still showed a very prominent peak at a retention time of 7.1 min. Clearly, the NRPS produced a different peptide than expected.

🔍

We analyzed both the MS¹ and MS² spectra of this peak (Fig. 7). This showed us that the [M+1] peak of the unknown compound was 780.35. The fragment peaks were consistent with subsequent losses of tryptophane, arginine and alanine fragments which all occur in the expected peptide besides arginine. This immediately suggested that (BBa_25JQ5Y0E) incorporates arginine instead of glutamine and we quickly found that the peptide PAA-T-f-a-R-W is consistent with the expected mass.

🔍

We created the EICs with the expected [M+1] peaks of C2-T-f-a-R-W and PAA-T-f-a-R-W and found that the C2 product was also formed (Fig. 8).

🔍

Similarly, we repeated the analysis of all other experiments where we made use of this unit. XUT^I2 and XUT^I3 insertions of this unit into the Chaiyaphumine synthetase also produced peptides whose masses were consistent with arginine incorporation, further cementing our hypothesis. Based on these MS² results, we updated the NRPS cluster annotation to reflect the actual amino acid incorporated, and the corresponding peptide structure is shown in Fig. 9.

This example nicely shows that our analysis does more than just provide a binary output of whether an engineered NRPS works - it can guide us toward a better understanding of our parts.

🔍

Definitive Structural Measurement and Purity Assessment

While LC-MS screening provided high-throughput verification across the NRPS library, final structural confirmation and quantitative purity assessment required a more robust approach. Achieving NMR-level confidence necessitated sufficient material, prompting a shift in the measurement strategy.

An NRPS variant with reliably high production was strategically selected for upscaling, enabling the isolation of enough peptide for High-Performance Liquid Chromatography (HPLC) purification and subsequent Nuclear Magnetic Resonance (NMR) spectroscopy analysis. This workflow allowed unambiguous verification of peptide identity, and precise assessment of compound purity.

Scaled-Up Sample Preparation for High-Resolution Analysis

🔍

To obtain sufficient material for HPLC purification and NMR characterization, we strategically selected two NRPS variants:

1. The native chaiyaphumine cluster (Fig. 10(a)) – previously analyzed as our LC-MS example was chosen because, as the natural biosynthetic cluster, it was expected to yield the peptide efficiently.

2. A high-producing module-exchanged variant (Chaiyaleucine) (Fig. 10(b))– identified during our LC-MS screening was included to demonstrate that our workflow could be extended to non-native constructs that showed promising production levels.

For both variants, cultivation was scaled up to 1 L under the same production conditions as described in our Standardized Protocol to ensure reproducibility. Induction timing, media composition, and harvesting were identical to the small-scale workflow. (NRPS figures and peptides)

The scaled-up cultures were processed using the same solvent-based extraction protocol (upscale expression protocol), enabling direct comparability with the LC-MS screening data. This approach yielded sufficient material for HPLC fractionation and NMR-level structural verification without introducing additional process variability.

LC-MS Confirmation Prior to Purification

After extraction, the samples were analyzed by LC-MS to verify the presence and mass accuracy of the target peptides. The Fig. 11 shows the Extracted Ion Chromatograms (EICs), with the peaks corresponding to the two variants, Chaiyaphumine D and A (m/z = 645.22 and 721.26, respectively) and Chaiyaleucine D and A (m/z = 661.33 and 737.36, respectively) highlighted in different colors. Both variants exhibited clear, well-resolved peaks with retention times in the 7–9 minute range, confirming successful detection and precise separation of the expected peptide species.

This quick quality control step ensured that only productive upscaled cultures proceeded to HPLC purification, saving time and resources. Verified samples were then purified by HPLC and prepared for NMR to achieve definitive structural confirmation.

🔍

Sample Preparation for HPLC

After LC-MS confirmation of the upscaled 1 L cultures, the peptide extracts were concentrated using a rotary evaporator at 40°C, 175 mbar, rotation 120 rpm until the volume was reduced to approx. 400 mL. The concentrated extracts were then flash freezed and lyophilized to obtain a dry peptide powder, which was weighed to quantify the yield of the unpurified extract.

The dry peptide powder was re-dissolved in 2ml DMSO. The individual optimized gradient was predicted from the LC-MS data to shorten the runtime and improve the separation and ensure optimal peak resolution.

HPLC Purification

High-performance liquid chromatography (HPLC) was employed to purify the target peptide from the crude extract, effectively separating it from minor analogs and other contaminants to enable accurate downstream NMR analysis. Preparative HPLC was carried out on an Agilent 1290 Infinity II Autoscale Preparative LC System equipped with a ZORBAX Eclipse XDB-C18, PrepHT column (AG-977250-102, Agilent) using a binary solvent system of acetonitrile (ACN) and water containing 0.1 % formic acid. A linear gradient was applied, increasing ACN at 0.5 % per minute over 30 minutes (final gradient 40–60 % ACN, flow rate 20 mL/min), with the initial ACN concentration optimized individually for each peptide. For each run, 600 µL of the crude extract was injected for purification.

The elutes from the HPLC were pooled together in a 250ml round bottom flask, were flash frozen and lyophilized. The lyophilized products (Fig. 12) were later weighed to find the yield. The weight of the dried peptide is given in Table 2.

🔍

Following HPLC purification, LC-MS analysis (Fig. 13) of the purified fractions confirmed that the target peptide was the predominant species, validating the success and efficiency of the purification process.

🔍

Sample Preparation for NMR

The dried peptide powders were dissolved in 550 µL of deuterated DMSO and briefly sonicated to ensure complete dissolution. The resulting solutions were then transferred into NMR tubes for analysis.

NMR Analysis

To verify the structure of our peptides, we performed a nuclear magnetic resonance (NMR) analysis. In this section, we will explain how we used different NMR experiments to verify the complete structure (except stereochemistry) of a Chaiyaleucine A. We performed the same set of measurements on Chaiyaphumine A, but since an NMR analysis of Chaiyaphumine has already been published^[5], we only had to compare the spectra to verify that we obtained the correct product.

NMR measures the energy transitions of nuclear spins in a strong magnetic field. Not all atomic nuclei have spins – most notably the most abundant carbon isotope ¹²C, which can thus not be observed by NMR. Most NMR techniques measure ¹H, the most common hydrogen isotope, or ¹³C, a stable carbon isotope with a natural abundance of around 1.1 %.

We started our structure deduction with an inspection of the ¹H spectrum of our compound (Fig. 14). In this spectrum, each peak corresponds to a chemically distinct hydrogen atom in the sample. The two most prominent peaks belong to water and our solvent. We used deuterated DMSO as solvent, which means that the ¹H isotopes have been replaced by ²H (deuterium), but it is normal to see a peak of residue ¹H atoms that remained in the solvent molecules.

🔍

The smaller peaks that are annotated in the spectrum could all be assigned to hydrogen atoms of Chaiyaleucine (a complete overview of all assigned peaks is given at the end of this section). Each peak has multiple features. Its position on the x axis is called the chemical shift δ and is given in ppm. This is a representation of the environment of the hydrogen atom – hydrogens bound to heteroatoms or to aromatic rings will normally have a higher chemical shift than those bound to aliphatic carbon atoms. One peak of the spectrum can directly be assigned only based on the chemical shift: The hydrogen atom in the N-H group of the indole in tryptophan (which is bound to a heteroatom and an aromatic ring) is well known to have a significantly higher chemical shift than all other peaks that normally occur in peptides – we could immediately see that this must be the peak at 10.81 ppm.

The integral is the area under a peak, and it is a representation of the number of chemically equivalent hydrogen atoms. We calibrated them by setting the integral of the already assigned indole-NH peak to 1.00. The four methyl groups in Chaiyaleucine (one in alanine, one in threonine, two in leucine) are directly obvious because they have integrals around 3. The integrals around 2 visible in the spectrum are overlaps of two peaks. The hydrogens of CH₂ groups will appear as two different peaks (each with an integral of 1) because the molecule is chiral, making the two hydrogen atoms diastereotopic. Finally, ten aromatic protons of the two phenyl groups, one aromatic proton from the indole and one peptide bond NH proton overlap in the region from 7.16 to 7.35 ppm. Since we could not resolve these peaks, we integrated the entire region.

Finally, the peaks have a substructure called J-coupling depending on the number and type of hydrogen atoms that are at most three bonds away. The information about the type and size of these couplings is also given in the table at the end of this section. To give an example of how this can be useful, the peaks corresponding to CH₂ groups will have unusually high coupling constants because the two chemically distinct hydrogen atoms of the CH₂ group couple with each other.

Using correlation spectroscopy (COSY), it can be determined more exactly which hydrogen atoms couple with each other (Fig. 15). COSY is a so-called 2D NMR method, which means that both the x-axis and the y-axis contain a 1D spectrum – in the case of COSY, the same ¹H-NMR spectrum. Then, there is only a visible peak in the spectrum if the hydrogen atom corresponding to the chemical shift on the x-axis and the hydrogen atom corresponding to the chemical shift on the y-axis couple.

This allowed us to assign all resolved ¹H-NMR peaks to the hydrogen atoms in Chaiyaleucine. To give an example, we noticed in the COSY spectrum that two peaks belonging to methyl groups coupled with the same peak at 1.63 ppm. The only place where two methyl groups are bound to the same atom is in leucine, so we knew that these three peaks belonged to the leucine isopropyl group. The peak at 1.63 ppm also coupled with the two peaks of the CH₂ group, those coupled with each other and with the α-hydrogen and that coupled with the NH-proton. Through this logic, we could assign all peaks belonging to leucine and we proceeded similarly with the other amino acids.

🔍

Furthermore, we used two 2D-NMR methods that correlate a ¹H-spectrum with a ¹³C spectrum. First, Heteronuclear Single Quantum Coherence NMR-spectroscopy (HSQC) shows a peak whenever the hydrogen atom corresponding to the chemical shift on the x-axis is directly bound to the ¹³C atom corresponding to the chemical shift on the y-axis (Fig. 16). This let us assign all ¹³C-NMR peaks except for the aromatic carbon atoms since the aromatic region in the ¹H-NMR was not resolved and the carbon atoms that had no hydrogen substituents, most notably those from the six carbonyl groups involved in amide and ester bonds.

🔍

Following these analyses, we were sure that our product contained all expected amino acids, but crucially, the order of the amino acids was not clear. We used Heteronuclear Multiple Bond Correlation NMR-spectroscopy (HMBC) to verify that our product was not a constitutional isomer (Fig. 17). HMBC is similar to HSQC, but it shows the coupling of hydrogen nuclei with ¹³C nuclei across two or sometimes three bonds. This meant that we could observe couplings with the carbon atoms of the carbonyl groups. Crucially, each of the ¹³C peaks of carbonyl groups coupled with the N-H hydrogen of one amino acid that participated in the peptide bond and the α-hydrogen of the other! Since we had already assigned all these ¹H peaks to the amino acids via COSY, this gave us the full amino acid sequence of Chaiyaleucine and showed that our assumed structure was indeed correct.

🔍

Production Titer Determination

To generate a standard for calibration, we initially attempted solid-phase peptide synthesis (SPPS) of the target compound. However, the cyclization step was unsuccessful, and the synthetic route could not yield the complete peptide. This experience also underscored how challenging SPPS can be for such cyclic peptides, highlighting the superior capability of NRPS systems in complex peptide synthesis.

Building on this insight, and after confirming the identity and purity of our NRPS-produced peptides through HPLC and NMR, we used these purified compounds to construct calibration curves for accurate quantification of production titers. This approach was applied only to the variants for which sufficient purified material was available. While LC-MS screening verified peptide production across the entire library, absolute titer determination was feasible only for these selected purified peptides.

To prepare the calibration standards, the purified peptide was dissolved in DMSO to a final concentration of 4 mg/mL to create a concentrated stock solution. From this stock, a serial dilution series was prepared, starting at 20 mg/L (1:200 dilution), followed by five consecutive 1:1 dilutions to yield 10 mg/L, 5 mg/L, 2.5 mg/L, 1.25 mg/L, and 0.625 mg/L standards. Each dilution was analyzed by LC-MS, and the resulting peak areas were plotted against their known concentrations to generate a robust calibration curve (Calibration Curve Protocol) (Fig. 18).

🔍

This calibration curve enabled absolute quantification of peptide titers in both small- and large-scale production cultures. As shown in Fig. 19, yields were determined for the Chaiyaphumine and Chaiyaleucine derivatives over a three-day production period. Extracts harvested at 24 h, 48 h, and 72 h were analyzed by LC-MS, and their peak areas were mapped onto the calibration curve to calculate peptide yields at each time point. Based on these measurements, the three-day production duration aligns well with observed peptide accumulation, confirming the practicality of our production workflow.

🔍

This approach provides quantitative insight into NRPS productivity, supporting optimization of culture conditions and guiding selection of high-yielding variants for downstream structural and bioactivity analyses.

Bioactivity Assay: Screening for Antimicrobial Potential

A central aim of our project was not only to produce a diverse set of nonribosomal peptides but also to evaluate their potential as novel antimicrobial candidates. While LC-MS, HPLC, and NMR were used to confirm structural identity and production levels only for selected representative variants, all produced peptides in the library were subjected to bioactivity assays. This approach allowed us to focus resource-intensive structural analyses on a few key examples while still functionally assessing the full diversity of our NRPS-derived peptides.

The peptide extracts used for bioactivity testing were prepared from 50 mL production cultures grown without antibiotics (see Library Screening Protocol for details). Extraction was performed with 5 mL of MeOH:ACN (1:1), resulting in an initial 10-fold concentration of the culture supernatant. After extraction, the clarified organic phase was transferred to 24-well deep-well plates together with 100 µL of DMSO added to each well. The plates were placed in a speed-vac (50 mbar, 25 °C) and dried overnight. The next day, the dried residues were re-dissolved in 100 µL of DMSO, yielding a final preparation that was approximately 500-fold concentrated relative to the original culture volume.

These concentrated DMSO extracts were transferred to sterile vials for testing. Bioactivity was assessed against bacterial strains using drop-spot and filter-disk assays: for drop-spot assays, 2 µL of extract was pipetted directly onto marked spots on inoculated agar plates and allowed to dry; for disk assays, 10 µL of extract was applied to sterile filter disks placed on the agar surface. Relative bioactivity was evaluated by observing inhibition zones to identify peptides exhibiting measurable antimicrobial effects.

Representative bioactivity assay results from the Hit to Lead optimization experiment are shown in Fig. 20, illustrating visible inhibition zones produced by selected peptide extracts against MRSA and E. faecalis. Clear growth suppression around the filter disks demonstrates that few of our NRPS-derived new to nature peptides exhibit measurable antimicrobial effects. This visual evidence highlights the functional relevance of the screening workflow and enabled us to rapidly pinpoint promising candidates for potential follow-up studies.

🔍

This integrated screening workflow extended our chemical analyses and provided a robust framework for high-throughput peptide characterization. LC-MS confirmed the presence of predicted peptides across the NRPS library, validating the functionality of our part collection and supporting the discovery of new-to-nature compounds. HPLC and NMR further verified structural identity and purity for selected variants, while bioactivity assays assessed the functional relevance of all produced peptides, including ~210 library members.

One variant displayed an unexpected mass profile, revealing a discrepancy in amino-acid incorporation and highlighting the importance of careful measurement for understanding module behavior. These measurements were crucial throughout the project, guiding decisions on production feasibility, duration, and selection of variants for detailed HPLC, NMR, and bioactivity testing. By providing accurate, quantitative insights, they reinforced the design–build–test–learn cycle, ensuring that planning, interpretation, and follow-up experiments were reliably grounded in data.

Outlook and Accessibility

The integrated measurement workflow established here provides a strong foundation for future exploration of NRPS libraries. With rapid LC-MS verification, structural validation via HPLC and NMR, and functional screening through bioactivity assays, this approach can be scaled to larger or more diverse libraries, enabling systematic discovery of novel peptides. Bioactivity assays could also be expanded to include a broader range of microbial strains or alternative target organisms, increasing the likelihood of identifying new compounds with promising functional activity. Building on our automated EIC extraction script, future adaptations could expand to full LC-MS/MS pipelines for faster screening and integrated quantification, explore additional microbial hosts and culture conditions, and integrate advanced analytical modalities such as high-resolution ion mobility spectrometry to resolve complex peptide variants and minor analogs. Together, these directions offer a roadmap for both expanding the discovery potential of NRPS-derived peptides and refining the measurement workflow for broader applicability.

To make peptide discovery approachable for teams without access to advanced instrumentation, we developed the Accessible Workflow for Peptide Discovery Handbook, which outlines a simplified version of our screening workflow. This companion guide allows exploration of NRPS libraries and functional bioactivity testing using common laboratory equipment and simpler tools.