But a compelling hypothesis still needs proof. Did a polymer really form, and how did it grow? These questions are hard to answer. The product won’t dissolve, and conventional mass spectrometry (MS) saw nothing above 210 Da.

A new Journal of the American Chemical Society study from Zhejiang University and Jianghan University (China) reports this first microdroplet-driven polymerization. Among the many analytical techniques applied in the study, only the Waters Xevo CDMS provided direct molecular-weight evidence, showing the molecular-weight distribution shift from approximately 100 kDa to 300 kDa within 30 minutes.

Why does this matter

For a new polymerization pathway, molecular weight is the ultimate proof point. It confirms that a polymer has formed and provides a window into how the reaction unfolds. Tracking molecular-weight growth helps researchers understand the underlying mechanism and evaluate whether the process could eventually be optimized and scaled.

The analytical challenge

The product to be analyzed was an amorphous, highly cross-linked powder, nearly insoluble in eight common solvents, which complicates the solution-based methods typically used to size polymers.

Spectroscopy filled in the chemistry. FTIR, Raman, XPS, and solid-state NMR confirmed the C=C, C–S, S–S, and C=S bonds, and pyrolysis GC-MS identified the fragments released on heating. But these methods described the building blocks, not the size of the assembled polymer.

Conventional MS hit a wall too. ESI-MS of the reaction filtrate found small sulfur ions but no polymer, and online microdroplet MS saw species of no more than four carbons, all below 210 Da. As mass and heterogeneity rise, conventional MS loses the charge-state resolution that accurate mass depends on.

What CDMS delivered

CDMS measures the mass-to-charge ratio and the charge of each ion individually, so it can size large, heterogeneous species without resolved charge states. 

Applied to samples taken at 10, 20, and 30 minutes, it enabled the team to:

• Confirm polymer formation – detecting macromolecular populations in the tens to hundreds of kilodaltons, roughly a thousandfold larger than anything conventional MS could see in the same reaction.
• Track growth over time – showing the overall distribution shift from about 100 kDa at 10 minutes to about 300 kDa at 30 minutes, with the high-mass population increasing markedly as the reaction proceeded.
• Resolve coexisting populations – distinguishing species near 40 kDa and 105 kDa at 10 minutes, and a separate higher-mass, higher-charge population emerging by 30 minutes.

Sample preparation was minimal. The team filtered the reaction suspension through a 0.45-µm membrane, loaded it into a 2-µm glass capillary, and analyzed it directly, with no dissolution, derivatization, or upstream separation.

Building the full picture

No single method could tell this story alone. Each answered a different question, and CDMS answered the one in between: how big the polymer gets and how it grows.

When a sample is insoluble, cross-linked, or polydisperse, CDMS adds the molecular-weight evidence that completes the picture, wherever the question is how big, and how heterogeneous, including for these applications:

• Polymers
• Supramolecular assemblies   
• Viral particles 
• Gene therapy vectors  
• Protein complexes

Read the full study, “Spontaneous Room-Temperature Polymerization of Carbon Disulfide in Water Microdroplets”, and see the CDMS data in the Supporting Information, Figures S14–S17.

In a recent webinar, nitrosamines expert Jason Brown of Brown Pharma Consulting joined Dr. Marian Twohig, Consulting Scientist at Waters to explore this changing landscape. From emerging nitrosamine considerations and the intersection with E&L to practical LC-MS/MS strategies for achieving demanding limits for small dialkyl nitrosamines, they examined where the industry has been, the challenges laboratories face today, and what may come next.

This blog continues the conversation by answering some of the questions that attendees asked during the webinar.

Q: Can you provide some more details around the use of a delay column and its effect on retention time, pressure, and compatibility with an analytical column?

A: The retention properties of the delay column should be equal to or greater to those of the analytical column to ensure sufficient delay of the interfering component, thus preventing co‑elution with the analyte. The overall retention time difference observed after the delay column is added to the flow path will be proportional to the volume of the packed bed column, the volume of the connecting tubing, and the flowrate of the pump.¹ The observed pressure difference will also depend on the added components. In the experimental details described in the webinar, the delay column had a particle size of 5 µm, with dimensions of 2.1 x 50 mm and an experimental flowrate of 300 µL/min, bringing the method pressure to less than 6000 psi. The added pressure was well within the maximum pressure of the analytical UHPLC pump, which is 15,000 psi.

Q: For extraction of nitrosamines from IV bags, how do you decide the volume of sample to be used for extraction?

A: In this iteration of the work, we analyzed the IV bag solutions in their intended for-use conditions, enabling the detection of low ng/mL detection of the small molecule nitrosamines.² If the application volumes of the solutions contained in the IV bags are high, a concentration step might be necessary to achieve the sensitivity required to meet regulatory limits. In this case, SPE or liquid-liquid extraction could be used to concentrate the target nitrosamines, thus ensuring the required sensitivity level is achieved by the detector used for the analysis.

The small molecule nitrosamines vary in their polarity. Therefore, the SPE chemistry used should be suitable to retain and provide good recovery for all the nitrosamines in the testing method.

Q: Do you have suggestions to improve the retentivity of polar nitrosamines in the analytical method?

A: For robust method development of polar compounds using reversed-phase liquid chromatography (RPLC), the target analyte should exhibit a minimum retention factor (k′) of 1.0 to ensure adequate separation from the column void volume. Ideally, a k′ value of 2.0 or greater should be achieved to provide improved retention and method robustness.

The retention factor, also referred to as the capacity factor (k′), is a measure of the extent to which an analyte is retained on the chromatographic column. It reflects the amount of time the analyte spends interacting with the stationary phase relative to the mobile phase. Higher k′ values indicate stronger interactions with the stationary phase, resulting in increased retention. As k′ describes the elution position of an analyte relative to the void volume, it is a useful parameter for assessing the likelihood of interference from matrix components or other compounds that elute near the void and may impact analyte detection.

Several approaches can be used to increase analyte retention. Selection of a stationary phase with enhanced polar retention characteristics is often beneficial. Examples include columns specifically designed for polar analytes, such as Waters XSelect HSS T3, Atlantis Premier BEH C₁₈ AX, and CORTECS T3 Columns. These phases can provide improved retention of polar compounds, including under highly aqueous mobile-phase conditions.

Method development should begin by reducing the proportion of organic solvent in the mobile phase. Where column chemistry permits, operation under 100% aqueous conditions can further promote retention of polar analytes. Adequate column re-equilibration is also critical, with a minimum of 10 column volumes at the initial mobile-phase conditions recommended between injections.

Achieving sufficient retention improves method robustness by ensuring the analyte is well separated from the void volume, thereby reducing the potential impact of early-eluting interferences on analyte sensitivity, accuracy, and reproducibility. If acceptable retention cannot be obtained for the target nitrosamines using RPLC, alternative retention mechanisms should be considered. For example, hydrophilic interaction liquid chromatography (HILIC) may provide improved retention and selectivity for highly polar compounds.

Q: How did you arrive at the acceptable ion ratios to be used for the final method? Does the software help with monitoring these ratios in routine analysis?

A: waters_connect for Quantitation Software can be configured to measure and monitor MRM ion ratios. The expected MRM ion ratios can be specified in the processing method or, alternatively, reference standards can be designated in the sample list and used to automatically calculate the expected ion ratios. Rule sets are configured in the software and used to flag when result parameters, including MRM ion ratios, are outside the specified acceptable criteria for a component. In the rule set, the tolerances applied to the measured versus the expected ion ratio are specified. If a sample produces peaks outside the specified tolerances, it will be flagged (on a component-by-component basis) as having an exception. The Ion Ratios Check column can be seen in the data summary table along with the pass/fail status applied to the detected components.

Q: Have you had any discussions with regulators around the use of delay columns?

A: We have not discussed the use of a delay column with the regulatory agencies. The use of a delay column was included in the method development strategy to enable the removal of an interfering peak eluting at the same retention time and contributing a signal from every MRM transition of NDBA, which would have caused ambiguity in the detection of NDBA in the final assay.

Q: Is the agency accepting the justifications for higher specification limits for NDBA in high-volume parentals based on toxicological evaluation and lifetime exposure strategy?

A: At the time of the recording, no. However, a recent FDA guidance changed that stance. So, Less-Than-Lifetime limit adjustments are now accepted. This must be done on a product-by-product basis and based on treatment duration. To be clear, this guidance applies to all marketed drugs, not just parenterals.³

Q: If my lab already does nitrosamine analysis, can we perform E&L analysis, too? What products and processes are really in scope?

A: The specific skills needed to perform non-targeted analysis differ greatly from those needed in targeted nitrosamine analysis. Additionally, while there is some equipment overlap, it’s rarely complete. This even extends to MS needs, where full spectral capabilities are a necessity. The efforts to make a typical MS lab into an E&L-specific lab are difficult to overcome without a large budget and substantial investment into the analysts.

Scope is a great question. The new draft Container Closure System Guidance from the FDA expanded this, recently.⁴ So, all drug substances and drug products are in scope. Additionally, all manufacturing process steps for large molecule and biologics are in scope. Essentially, that only leaves out excipient raw materials and small molecule production. I would look and hope for that to change, as it relies on some assumptions I would not agree with, but I would not expect to see that change soon. Keep in mind that even items “in scope” must go through risk identification and risk assessment steps. Both steps remove components from the workflow, leading to laboratory analysis.

Q: What tests should/can/must we request from our packaging material suppliers for nitrosamine risk assessments for tablets, syrups, suppositories, and injectables?

A: Right now, the risk of leachable nitrosamines from packaging is believed to be entirely related to nitrocellulose use in labels directly affixed to the primary packaging and, again, is believed to be contained to blisters and infusion bags. Some have suggested including questions regarding potential nitrosamines in technical requests to packaging and labelling suppliers. As part of a risk-based approach, this is easily justified.

Q: What are the risks associated with nitrosamines formation inside the body? Is there any guidance on this?

A: This is an interesting and complex topic. Certainly, there is a real, understood possibility of nitrosamine formation in the GI tract. There have been a few studies that have actively looked into this. To my knowledge, there have not been any exhaustive studies using modern LC-MS technologies. While the GI tract certainly has all the factors necessary for nitrosation: temperature, pH, and nitrite quantities, and there are also primary amines, which will react with the nitrite preferentially. I led a proof-of-concept study a few years ago that showed large-scale nitrosamine formation in multiple biorelevant dissolution media. However, it is impossible to fully replicate the in vivo complexities. It should also be noted that many vulnerable amines are present in many common foods.

To the point, much more is needed to fully understand patient risk. Contributions from gastric formation, including food, is relevant.

Q: Recently published papers have shown the impact of NOx on nitrosamine formation at small, analytical-scale samples. Does this impact the interpretation of data for large-scale bulk material?

A: Rok Gharek and team authored a couple of great papers on this subject, and I had the opportunity to discuss this directly with him and one of his coauthors.⁵ Part of the study explored how NOx saturated various levels of a tube filled with an API or excipient to show that it was not just surface exposure. Their conclusion was that while NOx can penetrate deep into various levels of a bulk container, that the available NOx was consumed by the outermost layers of the bulk.

Honestly, a good resampling of the bulk material, using a sample thief to take from the core, is the only way forward. Some work should be considered in the lab to determine which materials are even at risk, as the papers referenced showed that some materials are at much higher risk than others.

Watch the full webinar to get more details about the changing analytical landscape for nitrosamines testing.

Special thank you to Jason Brown of Brown Pharma Consulting for his presentation in the webinar and for his assistance answering these questions.

References

1. The Use of a Delay Column to Mitigate Persistent Interference Peaks in the Trace Level Quantitation of NDBA in Normal Saline Intravenous Infusion Bags Using UHPLC-MS/MS
2. Identification and Quantitation of N-Nitrosamines in Normal Saline Intravenous Infusion bags Using UHPLC-MS/MS
3. Application of Less-Than-Lifetime Adjustment to Nitrosamine Impurities
4. Container Closure Systems for Human Drugs and Biological Products
5. Grahek, R., et al. (2026). “Excipient nitrite levels are not static material attributes: Nitrite generation from environmental NOx and implications for drug product manufacturing.” Journal of Pharmaceutical Sciences 115(11).

Views expressed here are of the primary and contributing authors alone, and do not represent the views of Waters Corporation. Please perform your own due diligence while making any decisions based on regulatory guidelines.

The pace of artificial intelligence (AI) advancement has shifted what can be achieved, accelerating the ability to move past ideas to present reality. We are now in an era where technology, when executed well and in line with industry standards, can move an organization beyond simply generating quality data. The advanced automation achieved with AI creates a new level of consistency and trust in both data and decisions. The goals of speeding up medication delivery, flagging potential issues earlier, and advancing risk management and patient safety are no longer a future concept.

In our daily lives, we use AI for speed and convenience. But in regulated labs, the currency is trust. When deployed properly, AI can help eliminate human variability, deliver more consistent results, and build greater confidence in both the data and the decisions that follow. But proper deployment is more than just introducing new technology, it means ensuring AI is validated, operates consistently, meets industry and regulatory requirements, and produces results that can be relied upon. The future belongs to life science organizations that can successfully combine these benefits, improving efficiency without compromising trust.

Read our recent blog post to get a closer look at expectations for using AI in regulated labs.

The discussion is no longer hypothetical

Regulators have clearly outlined their expectations for the use of AI and machine learning (ML):

• Every action must operate within a framework of scientific rigor, compliance, and quality
• Context of use must be clear to ensure it is operated as intended
• AI must be explainable by the scientists who use it
• Every result must be traceable
• Every decision must be defensible by quality leaders

Waters has taken the leap to end the experimentation phase of AI/ML in a regulated laboratory.

For decades, Empower Chromatography Data System (CDS) has been the trusted backbone of labs where compliance and data integrity are non-negotiable. The success of an analytical lab has been defined by its ability to generate quality data–collecting, managing, protecting, and proving its integrity. Leveraged to standardize and optimize chromatographic workflows, Empower CDS has helped life sciences companies realize a myriad of efficiency gains over 30-plus years. Our Waters Informatics team has focused on where AI/ML could bring the next benefit to labs. The answer is data review and chromatographic peak integration.

AI is already demonstrating its potential to improve consistency and reduce human variability. A 2020 study published in Nature³ found that an AI system outperformed radiologists in breast cancer prediction for mammograms, reducing both false positives and false negatives.

In regulated labs, the same principle applied, AI must not only identify potential issues but do so reliably and within the required standards. Our first step was in 2025 with the introduction of AI/ML Anomaly Detection in Empower Data Viewer on the Waters Cloud Platform. By using machine learning to compare target results across chromatography data and expedite surfacing deviations for review, we are taking an important step towards bringing AI/ML directly into the analytical workflow. The capability is available in Empower Data Viewer today, with validated, compliant deployments to follow.

Our next step with Empower Software Solutions takes AI/ML a step further, providing a mechanism to consistently integrate chromatograms based on your definition of accurate data with machine learning. For a pilot study of 400 chromatograms, manual interventions dropped from 90 interventions to four after AI-powered peak integration was implemented.⁴ AI-powered peak integration increases the speed, consistency, and scalability of data processing, reducing the manual effort and variability that slow analytical labs without eroding productivity or confidence in results.

It is built for high-throughput, regulated labs and complements established Empower Software workflows. It will be available for early access in December 2026 while Waters works toward validated, compliant deployment.

From potential to reality and beyond

Waters is at the forefront of the shift toward cloud-enabled, AI/ML-driven laboratory informatics and is delivering on the promise of AI/ML with integrated, scalable capabilities inside established lab workflows, so you can realize the benefits of intelligent automation today, with Waters advancing these capabilities toward a validated, compliant deployment.

Empower Anomaly Detection and ML Peak Integration represent our ongoing commitment to deliver future-ready software solutions that adapt to changing regulations, innovations, and markets to improve outcomes for labs, particularly those in regulated environments. These innovative capabilities enable your organization to move from experimenting with AI/ML to using it for practical applications that improve the speed, trust, and scalability of data review and processing.

The benefits go beyond just access to new technology—it is unlocking the ability to perform data processing in new ways that were previously impractical or not possible.

Waters intends to continue expanding AI and machine learning functionality across its software platform. Waters goal is to make the compliant data foundation the trusted starting point as laboratories progress toward agent-driven and increasingly autonomous operations.

Learn more about how Empower Software Solutions deliver the compliant-ready results that your lab needs.

References

1. Dataiku, Global AI Confessions Report: CEO Edition 2026
2. Define Ventures, Inside the C-Suite Report, 2025
3. https://www.nature.com/articles/s41586-019-1799-6
4. Waters internal pilot study with a global pharmaceutical partner, 2025. Data on file, Waters Corporation

The biological challenge

Tuberculosis (TB) remains one of the world’s deadliest infectious diseases, claiming more than 1.2 million lives in 2024 alone. As drug-resistant strains continue to emerge, there is an urgent need for new therapeutic targets. One promising target is the bacterial proteasome activator Bpa, a protein found in Mycobacterium tuberculosis with no known human counterpart.

However, understanding how Bpa recognizes and delivers substrates for degradation has proven challenging because it exists in multiple, closely related oligomeric assemblies that are difficult to distinguish using conventional analytical methods.

Why the complete picture remained elusive

Multiple studies using X-ray crystallography, cryo-EM, SEC-MALS, and conventional native mass spectrometry described Bpa as forming a dodecameric ring. However, these approaches average together structurally similar species, making it difficult to distinguish closely related assemblies that differ only subtly in mass and size.

Because a potential 11-subunit ring differs from the dodecamer by less than 10% in mass, alternative oligomeric states could remain hidden in plain sight, obscuring a complete understanding of Bpa structure and function.

The CDMS finding that changed the story

CDMS revealed that apo Bpa exists as a mixture of previously unrecognized 11-subunit and 12-subunit rings and established a 12:2 binding stoichiometry between Bpa and HspR. It further showed that substrate binding favors the fully assembled dodecamer, providing critical evidence that guided subsequent cryo-EM analysis.

These findings were made possible because CDMS directly measures both mass and charge for individual ions, allowing closely related assemblies to be separated even when they appear as a single population in conventional mass spectrometry.

cryo-EM turned the CDMS discovery into structural insight

cryo-EM transformed the CDMS discovery into structural insight by revealing how the native substrate HspR engages Bpa. Guided by CDMS data, researchers resolved both previously hidden 11-mer and established 12-mer Bpa assemblies and obtained a 3.9-Å structure of the Bpa:HspR complex.

These structures showed that HspR preferentially associates with the fully assembled dodecamer and localized the region responsible for substrate recognition. Importantly, the undecameric assembly was not recovered in the initial cryo-EM analysis.

It was the unexpected population identified by CDMS that prompted researchers to refine their classification and reconstruction strategy, enabling a more complete view of the system’s structural landscape.

The key takeaway

Hidden structural populations can hold the key to biological function. This study showed how CDMS revealed what conventional methods missed and how cryo-EM translated that discovery into structural understanding, providing new insight into a validated tuberculosis drug target.

Read the full article, “Resolving Hidden Stoichiometries in Bacterial Proteasome Activator (Bpa)-Substrate Complexes by Cryo-EM and Charge Detection Mass Spectrometry,” to see how you can use CDMS and cryo-EM to advance your protein analysis.

A recent Analytical Chemistry article by scientists from Pfizer and Megadalton Solutions demonstrates how charge detection mass spectrometry (CDMS) helped to close this gap. The researchers directly measured intact mRNA constructs up to 4522 nucleotides, while simultaneously assessing sequence integrity, nucleotide modifications, poly(A) tail length, and the impact of lipid nanoparticle (LNP) formulation.

Why does this matter

As mRNA therapeutics expand beyond vaccines into gene editing, oncology, and other advanced modalities, developers need analytical tools that provide a more complete picture of the molecule itself.

This study shows that CDMS can bridge a long-standing gap by offering direct measurements of intact mRNA while simultaneously providing information on several critical quality attributes (CQAs). Rather than replacing established methods, CDMS complements them by providing something they cannot: a direct view of the whole therapeutic molecule.

The analytical challenge

Standard workflows digest mRNA into small fragments before MS analysis—useful for confirming sequence composition, but blind to whether the full-length transcript survived intact. That blind spot grows with size: many therapeutic mRNAs exceed 4000 nucleotides, and their highly charged phosphate backbones attract counterions that broaden signals and collapse the charge-state resolution that conventional MS depends on.

Native MS has been pushed to mRNAs around 800 nt, and single-ion Orbitrap methods to roughly 2100 nt, but overlapping charge states and adduct formation still obscure the low-abundance truncated species that quality assessment depends on finding.

What did CDMS find

CDMS overcomes these limitations by measuring both mass and charge for individual ions, eliminating the need for conventional charge-state resolution.  With this approach applied to four constructs, eGFP, luciferase, β-galactosidase, and Cas9 transcripts, the researchers showed that CDMS delivered:

• Confirm intact mRNA length and integrity, measuring constructs from 997 to 4522 nucleotides with approximately 0.5% uncertainty, while measured mass tracked nucleotide count with an R² of 0.99998.
• Verify chemical modifications, accurately detecting the mass shifts associated with methoxyuridine incorporation. For luciferase mRNA, an 8.3 kDa shift was observed, closely matching the 8.2 kDa theoretical value.
• Differentiate poly(A) tail lengths, resolving hemagglutinin mRNA constructs containing tails from 0 to 74 nucleotides with an uncertainty of approximately ±8 nucleotides, supported by orthogonal CE and LC-MS measurements.
• Assess formulation integrity, showing that mRNA extracted from LNPs was statistically indistinguishable from the original material, indicating that encapsulation did not measurably alter the mRNA cargo

A practical insight

The study also identified conditions that improved performance. Ammonium-based buffers consistently produced more accurate mass measurements than water alone, while higher-charge mRNA populations delivered the most reliable and highest-resolution results. These insights provide useful guidance for future nucleic acid analyses.

The key takeaway

A single CDMS workflow can confirm intact mRNA length, assess sequence integrity, verify nucleotide modifications, differentiate poly(A) tail lengths, and evaluate the impact of LNP formulation, all through direct measurement of the intact molecule.  

Read the full study, “Advancing Charge Detection Mass Spectrometry for Characterizing Large, Intact mRNA Constructs”, to discover how, with mRNA therapeutics continuing to grow in complexity, CDMS is emerging as a powerful addition to your analytical toolbox.

Bringing AI closer to the lab

Lab software supports highly specialized work, but routine tasks can still involve moving through multiple applications, menus, and setup steps. Creating acquisition methods, building sample lists, assigning instruments, submitting work, and retrieving results all require time and attention. For experienced users, that can mean repetitive administration, whereas for newer analysts it can create a steep learning curve and greater dependence on expert colleagues.

Waters AI Link introduces a more intuitive way to engage with these capabilities. Instead of requiring users to navigate every step manually, it makes supported waters_connect Software functions available through natural language. A scientist or analyst can describe what they want to accomplish, while the underlying tools translate that intent into defined actions within the platform.

What is Waters AI Link?

Waters AI Link is a Model Context Protocol (MCP) server that converts waters_connect Software  capabilities into governed AI tools. MCP is an open standard that enables AI assistants to connect with software capabilities in a consistent way. For Waters customers, this creates the foundation for conversational interaction with laboratory workflows without replacing the scientific applications, validated assets, or domain expertise on which laboratories rely.

Waters AI Link currently brings together capabilities supporting quantitative acquisition and processing workflows. These tools enable activities such as creating and editing sample lists, working with acquisition methods, assigning instruments, accessing analytical results, and generating reports. As coverage expands, the goal is a connected, end-to-end experience from method development through acquisition, processing, and result retrieval.

Designed to make everyday work simpler

The value of conversational AI is most tangible when it removes friction from real tasks. With Waters AI Link, a request that once necessitated a series of manual setup steps can now be expressed in a single conversation. Teams can more readily reuse existing methods and sample structures, helping them build on their validated work rather than starting again. Results and supporting information can also be presented more directly, reducing the time spent moving between screens to find the right data.

• For analysts: A simpler route through routine workflows.
• For scientists and method developers: Less time on repetitive setup and more time for scientific judgement.
• For laboratory managers: More consistent execution and reduced reliance on a small number of experienced users.

The result is not AI for its own sake, but a practical way to help laboratories work more efficiently and make established capabilities easier to access.

Governed AI for a scientific environment

In laboratory informatics, ease of use cannot come at the expense of control. Waters AI Link uses a governed approach:

• It does not give an AI model unrestricted access to the waters_connect Software.
• Each tool is explicitly defined and constrained.
• Every action remains subject to the platform’s authentication, permissions, and access controls.
• Users can only perform actions through AI that they are authorized to perform in waters_connect Software.

This governed approach distinguishes Waters AI Link from an uncontrolled agent. Natural language becomes an additional interface to trusted platform capabilities—not a shortcut around product ownership, validation, or laboratory policy. The domain applications remain in the authoritative environment for deep scientific work, while Waters AI Link provides a more accessible conversational entrance.

An open foundation that grows with the laboratory

By building on the open MCP standard, Waters is creating an extensible foundation rather than a point solution tied to a single AI client. MCP-compatible assistants can access published tools without a separate custom integration for every use case. A plugin model will allow additional waters_connect Software capabilities to be made available through the same governed mechanism, extending the potential reach of conversational AI across laboratory workflows over time.

Waters brings together analytical instruments, scientific data, and a connected informatics platform. Waters AI Link provides an additional layer that allows AI to interact with those capabilities in a controlled and useful way. By combining natural-language access, open architecture, and platform-level governance, Waters is helping lead the introduction of AI into lab software on terms that reflect the needs of modern science.

Waters AI Link is an important step towards the AI-enabled lab: one where routine complexity is reduced, trusted scientific assets are easier to reuse, and people can focus more of their time on insight and decision-making. As its tool library and workflow coverage continue to grow, the same foundation can support an expanding range of laboratory experiences, making AI a practical, governed part of everyday work.

Explore how waters_connect Software advances your science through connected and accessible data.

Introduction

Grain, feed ingredients, and finished feed are routinely tested for mycotoxins to support quality, compliance, and risk management decisions. Recent peer-reviewed research suggests that the question may no longer be whether mycotoxins are present, but whether testing programs capture the full range of contaminants present in a sample.

Many monitoring programs have historically focused on individual contaminants, such as aflatoxin, deoxynivalenol (DON), fumonisins, or zearalenone. While these toxins remain important, a growing body of literature indicates that frequently multiple mycotoxins occur together. Testing programs only report the mycotoxins measured, which may not provide the full picture. In other words, a result can be accurate and still be incomplete.

Across both grain and feed systems, multiple mycotoxins are increasingly understood to be the norm rather than the exception.

This has practical implications across the supply chain, where materials are evaluated at multiple points as they move from grain to finished feed. Decisions at intake, in the laboratory, and in formulation are based on the data available at that moment. When that data is limited in scope, the picture guiding those decisions may not always be complete.

Co-occurrence begins in grain

Large-scale survey data indicate that co-occurrence is common at the earliest stages of the supply chain. In a seven-year study of corn grain and corn silage in the United States, Weaver et al. (2021) reported that 90.2% of corn grain samples and 96.8% of corn silage samples contained two or more mycotoxins, with an average of 4.8 and 5.2 mycotoxins per sample, respectively.

Regional data reinforces these findings. Fusilier et al. (2022), analyzing maize grain harvested in Michigan, reported that every sample contained multiple mycotoxins, with at least four detected in one year and six in another, with several toxins frequently occurring together rather than in isolation.

These results indicate that co-occurrence is not an isolated condition. It is a consistent feature of grain entering feed and food supply chains. Identifying a single mycotoxin therefore does not necessarily characterize the contaminant profile of the sample.

Co-occurrence extends beyond harvest

Co-occurrence does not end at grain intake. It continues through feed ingredients and finished feed, where contamination may reflect a combination of field conditions, storage environments, and ingredient blending.

Pokoo-Aikins et al. (2026) evaluated poultry feed and feed ingredients across more than 570 samples and reported fumonisins in 86% of samples, DON in 45%, zearalenone in 36%, and aflatoxin in 10%. All corn and finished feed samples contained at least one mycotoxin, and more than half contained multiple contaminants.

Importantly, the study found that co-contamination patterns were consistent in both corn and finished feed. This indicates that mycotoxin risk is not confined to a single stage of the supply chain but persists throughout it.

When incomplete screening becomes visible

The limitations of single-analyte testing are not always apparent at the point of initial screening. In many cases, they become visible when materials are evaluated again in a different context.

A shipment may meet criteria for one mycotoxin while other contaminants remain unexamined, or an ingredient may be accepted for use based on partial information. In some cases, additional testing is triggered later or materials already in process are reconsidered as new information becomes available.

In practice, this often means that a result initially considered acceptable may need to be revisited when additional information becomes available.

In these situations, testing has not failed. It has provided a view aligned to the scope of the assay performed. For operations responsible for grain quality, formulation, and production, that distinction becomes operationally significant.

Operational impact and economic context

At the operational level, the effects of mycotoxin contamination are not abstract. They can appear as discounts, rejected loads, additional testing requirements, and adjustments to ingredient use or production plans.

Recent studies continue to demonstrate how frequently these decisions occur under real market conditions. For example, Postma et al. (2023) reported that only about 41% of harvested maize fell below the 20 μg/kg threshold, while 38% fell between 20 and 300 μg/kg—a range that may trigger discounting or rejection at the grain elevator.

The implication is straightforward. Incomplete visibility does not eliminate risk. It can shift when and where that risk becomes visible, often at the point where a decision must be confirmed, reconsidered, or delayed.

Expanding insight within operational constraints

The prevalence of co-occurrence is well established. The challenge for many organizations is how to respond without introducing additional operational burden.

Expanding monitoring scope can require additional extractions, increased testing time, and greater laboratory workload. For facilities managing high sample volumes, those constraints are significant.

As a result, the priority is not simply broader testing. It is broader understanding that remains practical within existing workflows. Approaches that allow multiple mycotoxins to be evaluated from a single extraction can expand visibility without proportionally increasing complexity.

A practical path forward

The question facing feed and grain operations is no longer whether multiple mycotoxins are present. The question is whether current testing approaches provide a complete enough picture to support confident decisions.

A result may be accurate for the analyte tested, but still incomplete in the context of co-occurrence. Bridging that gap requires approaches that align analytical capability with operational reality.

Expanding mycotoxin monitoring with Waters VICAM Solutions

In a supply chain where multiple mycotoxins are common, the most important risk may still be the one that hasn’t been measured. The evidence is clear: multiple mycotoxins are frequently present within the same sample. The remaining challenge is not awareness, but how to expand visibility without adding time, complexity, or additional testing steps.

Waters VICAM Myco 5-in-1 PLUS workflow, used with the Vertu Touch platform, enables the analysis of six mycotoxins from a single extraction. By consolidating multiple assays into one workflow, food safety and quality teams can reduce total testing time while maintaining precise, quantitative results.

Results are delivered in less than 10 minutes per assay, excluding sample preparation, allowing decisions to be made within the same workflow rather than across multiple testing steps.

The ability to evaluate multiple mycotoxins in a single method reflects a shift toward testing strategies that better align with the reality of co-occurrence.

Take the next step

Expanding the scope of mycotoxin testing does not require expanding complexity.

With Waters VICAM, one extraction enables a more complete mycotoxin profile without adding additional steps, time, or complexity.

• One extraction, six mycotoxins

• Reduced testing time by eliminating separate assays

• Quantitative results delivered rapidly within a single workflow

Multi-mycotoxin monitoring enables a more complete view of contamination risk while preserving the speed and efficiency required for everyday operations.

Visit VICAM.com or connect with a VICAM representative to learn more about Myco 5-in-1 PLUS.

Request a quote

References

1. Weaver, A.C., Weaver, D.M., Adams, N., & Yiannikouris, A. (2021). Co-Occurrence of 35 Mycotoxins: A Seven-Year Survey of Corn Grain and Corn Silage in the United States. Toxins, 13(8), 516.

2. Fusilier, K., Chilvers, M.I., Limay-Rios, V., & Singh, M.P. (2022). Mycotoxin Co-Occurrence in Michigan Harvested Maize Grain. Toxins, 14(7), 431.

3. Pokoo-Aikins, A., Mitchell, T.R., Read, Q.D., Shanmugasundaram, R., & Glenn, A.E. (2026). Mycotoxin Occurrence in Poultry Feed and Feed Ingredients. Journal of Applied Poultry Research.

4. Postma, K., et al. (2023). Economic benefit of co-regulation to manage aflatoxin risk in maize. Heliyon.

PROTACs combine two binding elements with a linker to selectively target proteins for degradation. Their larger size, flexibility, and multiple functional groups can create new analytical considerations compared to more conventional small molecules.

When developing methods, the question isn’t whether liquid chromatography still works. It does. The question you should ask is how to approach method development when the molecule itself is more complex.

Complexity changes the method development challenge

A PROTAC can behave differently from a traditional small molecule in ways that are not always predictable from molecular weight alone. Retention, selectivity, peak shape, recovery, and reproducibility become even more critical to monitor.

This is particularly relevant as the pharmaceutical industry explores new therapeutic modalities and is working to establish reliable analytical workflows for development, stability testing, and quality control.

For analytical scientists, that can mean moving beyond incremental changes to a single chromatographic condition.

A systematic approach can help identify what works rather than relying on what you expect to work.

A case study in systematic method development

Our recently published application note on PROTAC ARV-825 provides an example with the method developed in just over two days.

ARV-825 is a heterobifunctional molecule containing pomalidomide and birabresib connected by a polyethylene glycol linker. To evaluate its stability, the compound was subjected to forced degradation conditions, generating additional components that needed to be separated and characterized.

Rather than optimizing one chromatographic condition through trial and error, the study applied a systematic approach.

The process began by evaluating retention at different pH conditions. Low-pH conditions provided better retention of lower-level degradants, while the primary ARV-825 peak was relatively unaffected. This information dictated the mobile-phase conditions for the next stage of screening, low-pH mobile phase additives.

The next step was to deliberately introduce chromatographic selectivity by screening multiple stationary phases and mobile-phase combinations.

This distinction is important. Changing the gradient changes how compounds move through the column. Changing the stationary phase can change how they interact with the column in the first place.

For complex molecules and unknown degradation products, that difference can be critical.

Look beyond the chromatogram

The ARV-825 work also highlights why LC-MS can be valuable during method development.

When degradation products are not available as individual standards, UV detection alone may not provide enough information to confidently track what is being separated. In the study, mass detection was used to help identify and track components by their mass, providing additional information during the separation process.

The study also evaluated against standard stainless-steel hardware using the same stationary phase. Differences in retention and selectivity were observed for some degradants, demonstrating that interactions with metal surfaces can become another variable in the chromatographic equation. The MaxPeak Premier Columns also demonstrated improved peak-area reproducibility in the study.

For complex molecules, controlling variables early in method development can help reduce the risk of having to revisit the method later.

A more deliberate path to robust methods

The ARV-825 example demonstrates a broader principle for analytical scientists working with emerging drug modalities.

When the molecule is different, the method-development strategy may need to be different too.

A systematic workflow can provide a logical path from retention and pH screening to stationary-phase selectivity to optimization and detection, while creating a more traceable basis for the final method.

In the ARV-825 study, the combination of systematic screening and MaxPeak Premier Columns produced baseline resolution of the detected degradants and the main ARV-825 peak, with the method developed in just over two days.

As drug discovery continues to expand into more complex molecular architectures, analytical workflows will need to evolve alongside them. The goal is not to make chromatography more complicated. It is to make method development more deliberate, reproducible, and efficient.

For scientists characterizing PROTACs and other emerging small-molecule modalities, the right combination of systematic screening, chromatographic selectivity, surface control, and complementary detection can help turn molecular complexity into a manageable analytical challenge.

Ultimately, better analytical methods help provide the reliable information needed to advance new therapies and, in turn, accelerate the benefits of pioneering science to improve human health.

Learn more about how MaxPeak Premier Solutions can enhance your PROTACs analysis.

Why does this matter

CDMS is uniquely suited for analyzing very large and heterogeneous biomolecules because it measures both the mass-to-charge ratio (m/z) and the charge (z) of individual ions. Until now, however, high-performance ELIT-CDMS systems have largely been limited to specialized academic laboratories. This new benchtop platform makes high-resolution CDMS more accessible by combining advanced performance with a compact, user-friendly design.

The challenge with large and heterogeneous biomolecules

Traditional native MS works well for many proteins, but it becomes increasingly difficult to analyze very large and heterogeneous particles.

As molecular size increases, charge-state distributions overlap, making accurate mass determination challenging. This challenge is especially important for AAV gene therapy vectors, virus-like particles (VLPs), lipid nanoparticles (LNPs), vaccines, heavily glycosylated proteins, and large protein assemblies.

Alternative techniques, such as analytical ultracentrifugation (AUC), size-exclusion chromatography with multi-angle light scattering (SEC-MALS), and mass photometry, can provide useful information, but often lack the resolution, throughput, or particle-level detail needed for these complex samples.

How CDMS makes a difference

Unlike conventional mass spectrometry, CDMS directly measures individual ions and determines both their charge and m/z value. This allows the mass of each particle to be calculated directly, overcoming the limitations caused by overlapping charge states.

The new ELIT-CDMS platform offers: Ion optics optimized to cover a broad m/z range (up to m/z 250,000), detection of ions carrying up to ~3000 charges, mass measurements extending into the hundreds of megadaltons, real-time data processing and visualization, and high-resolution measurements enabled by charge quantization.

Improved charge precision delivers higher resolution

A key innovation highlighted in the study is the instrument’s ability to achieve highly precise charge measurements. By improving charge precision, the system can accurately assign individual charge states, significantly increasing mass resolution. The researchers demonstrated mass resolutions of up to approximately 160, representing the highest mass resolution reported for a commercially available CDMS platform.

Strong performance for AAV characterization

A major focus of the study was AAV analysis, a critical application in gene therapy development. The platform successfully distinguished empty, full, and overfilled AAV capsids, quantified empty-to-full ratios, resolved partially filled particle populations, analyzed multiple AAV serotypes, and delivered highly reproducible results across repeated measurements

The instrument also showed strong quantitative performance and demonstrated approximately 20-fold higher sensitivity when operating in Target Enhancement Mode (TEM), making it suitable for lower-concentration samples.

Beyond gene therapy: Vaccines and VLPs

The authors also demonstrated the platform’s capabilities using chikungunya and dengue virus-like particles. CDMS revealed intact particles, degradation products, and aggregates across a wide mass range. For chikungunya VLPs, particles with masses up to approximately 350 MDa were observed.

For dengue VLPs, researchers used controlled heating experiments to investigate particle dissociation and stability, gaining structural insights that would be difficult to obtain using conventional approaches.

Looking ahead

As advanced therapeutics continue to grow in size and complexity, analytical technologies must evolve as well. This study demonstrates how commercial benchtop ELIT-CDMS can provide:

• Direct measurement of heterogeneous megadalton assemblies
• High-confidence AAV characterization
• Structural insights for VLPs and vaccines
• Broad mass range coverage
• Ease of use for both industry and academic laboratories

The introduction of a commercial benchtop CDMS platform represents an important milestone for the field. By making high-resolution analysis of very large biomolecules more accessible, CDMS is poised to become an increasingly valuable tool for gene therapy, vaccine development, nanoparticle characterization, and other emerging biopharmaceutical applications.

Read the full paper, “A Charge Detection Mass Spectrometer for the Analysis for Megadalton-sized Molecules”, and see how a new benchtop ELIT-CDMS platform is helping researchers overcome the challenges of characterizing the world’s largest and most heterogeneous biomolecular assemblies.

A recent bioRxiv paper by Dr. Charlotte Uetrecht and colleagues at the Centre for Structural Systems Biology (CSSB) in Hamburg, Germany, in collaboration with Waters scientists, demonstrates how charge detection mass spectrometry (CDMS) is pivotal for uncovering a non-classical capsid assembly invisible to any other method.

In the study, CDMS was used to analyze two human norovirus strains, with cryo-EM and bottom-up proteomics providing orthogonal validation. 

The core problem: Heterogeneity defeats conventional MS

Norovirus VLPs can form T=1, T=3, T=4, and other geometries. Combined with N-terminal processing of the VP1 capsid protein, this creates overlapping mass distributions that conventional native MS cannot resolve—coexisting particle populations, including non-icosahedral species, are entirely obscured.

How CDMS solves the problem

CDMS determined the mass and charge of individual ions simultaneously, by passing the need for charge state resolution.  The authors compared results obtained from the electrostatic linear ion trip (ELIT)-based Xevo CDMS and the Orbitrap-based Direct Mass Technology (DMT) and reported that:

• Both platforms identified the same major assemblies in GI.1 Norwalk and GII.17 Kawasaki VLPs
• Waters Xevo CDMS produced masses closer to theoretical values, attributed to its dual m/z and charge calibration that accounts for adduct mass
• DMT combined with UniDec deconvolution (UCD) achieved higher apparent mass resolution, at the cost of greater workflow complexity and risk of deconvolution artifacts

CDMS and cryo-EM jointly reveal a Novel GII.17 capsid

Combining CDMS and cryo-EM enabled discovery and structural validation of a novel viral assembly state that would have been difficult to identify with either technique alone.

• CDMS detected three distinct capsid populations in GII.17 Kawasaki: classical T=3 (~10.65 MDa), T=4 (~14.25 MDa), and an unexpected intermediate species (~12.58 MDa).
• Mass analysis estimated ~212 VP1 subunits for the intermediate particle, suggesting a non-icosahedral assembly inconsistent with traditional Caspar-Klug symmetry.
• cryo-EM confirmed the structural identity of this species as a prolate (oval-shaped) capsid and yielded the first images of GII.17 Kawasaki T=3, T=4, and prolate assemblies.

Instrument comparison guides future strategy

Stepwise inlet heating in the Xevo CDMS reduced measured mass toward theoretical values—a practical tool for separating true particle mass from adduct contributions.

Key takeaways:

• Xevo CDMS offers a streamlined workflow with live data visualization and mass accuracy within ±1%
• DMT plus UCD deconvolution reduces peak width by ~50%, but requires careful application to avoid artifacts
• Analyte-matched calibration standards are critical—both platforms showed mass deviations linked to calibrant size mismatch

Read the full paper, “Applying distinct CDMS strategies to observe non-classical virus capsid assembly”, to see how two CDMS platforms together reveal the full complexity of norovirus VLP populations.