Proteomics as a complementary approach to measure norovirus infection in clinical samples

Abstract

Norovirus (NoV) is a leading cause of global acute gastroenteritis, particularly in young children, with no current licensed vaccine. Epidemiological studies have shown that asymptomatic cases are common, and infected patients may test positive for prolonged periods; however, the impact of these phenomena on transmission and public health measures remains unclear. A major limiting factor is our ability to measure infection, which is constrained to real-time reverse transcription polymerase chain reaction or antibody-based assays, both of which are susceptible to loss of detection by rapid NoV evolution. This review highlights the potential for proteomics to overcome current technical limitations and advance basic science discovery and clinical research. Importantly, proteomics-based protein detection can span NoV, host, and microbiome proteins and could help identify host or microbiome factors that correlate with disease outcome. Further developing proteomics tools to complement existing diagnostic technologies will improve our ability to assess NoV pathogenesis and transmission, as well as therapeutic efficacy.

1. Introduction

Norovirus (NoV) is a leading cause of non-bacterial acute gastroenteritis (AGE) around the globe(Ahmed et al., 2013, 2014; Cannon et al., 2021). In particular, NoV poses a heavy health burden on young children under the age of five years, ranking second only to rotavirus as cause of AGE-related death worldwide and first in countries with high coverage of rotavirus vaccination (Black et al., 2024; Challappa et al., 2016; MARSH et al., 2018; Rouhani et al., 2016; Santos et al., 2017; Yori et al., 2009; Zeng et al., 2010).

The detection and study of NoV infections is particularly active in Peru. A birth cohort study in Lima, Peru found that 80 % of children had experienced at least one NoV infection by year one, and more than 70 % had experienced an AGE event by year two(Saito et al., 2014). In a recent study conducted in 3 hospitals in Lima, Peru, preliminary findings indicate that around 40 % of children under the age of five years visiting an emergency room (ER) because of an AGE episode were positive for NoV(Ingunza et al., 2023). The health burden of NoV may be further underrepresented, as one study found that ~17 % of AGE cases in Lima, Peru which were initially thought to be pathogen-negative turned out to be NoV positive (Rivera et al., 2010), indicating that the AGE burden of NoV may need to be reassessed(Lanata et al., 2025).

In addition to low- and middle-income countries (LMICs), NoV poses a large health and economic burden on high income countries(Lopman et al., 2015). In the US alone, NoV disease accounts for ~$2 billion in economic losses and healthcare expenses each year(Lopman et al., 2015). NoV outbreaks are a regular occurrence in the US, and, at the time of writing, there is a larger than average number of outbreaks in the 2024–2025 season reported in the US Centers for Disease Control and Prevention (CDC) NoroSTAT database (www.cdc.gov/norovirus/php/reporting/norostat-data.html), highlighting the continued impact of NoV on global health.

After an acute NoV illness, patients, especially the immunocompromised(Henke-Gendo et al., 2009), continue to test positive for NoV by RT-PCR for prolonged periods, suggesting extended virus shedding. This phenomenon has been observed in both experimental human infection studies(Atmar et al., 2008; Kirby et al., 2014) and in longitudinal observational studies(Henke-Gendo et al., 2009; Rouhani et al., 2016). The clinical significance of this prolonged detection remains to be determined, particularly whether this molecular detection equates to transmissible and infectious viruses. The development of complementary tools to interrogate this phenomenon is a driving motivation for this review.

A non-enveloped RNA virus, NoV exhibits widespread genetic diversity, which impedes formation of naturally acquired immunity and complicates treatment strategies(Ford-Siltz et al., 2021). There are at least 10 NoV genogroups, of which GI and GII are commonly associated with disease in humans(Winder et al., 2022). Furthermore, these genogroups encompass a number of distinct NoV genotypes(Cannon et al., 2017). NoV infection treatment is largely limited to supportive care, as there is currently no licensed vaccine or anti-viral therapeutic, though several vaccine candidates are in development and in clinical trials (Kawagishi et al., 2023; Vesikari et al., 2022). Given that NoV prevalence is globally widespread, an effective NoV vaccine would bring substantial gains in public health and represents a high-value target for development(Armah et al., 2023; Mirelman et al., 2015).

The current gold standard for detecting NoV in patient samples is real time reverse transcription polymerase chain reaction (RT-PCR), which uses DNA primers to amplify and detect specific sequences from complementary DNA that has been reverse-transcribed from viral RNAs(Liu and Moore, 2020). Like most RNA viruses, NoV evolves by acquiring point mutations in its genome due to its error-prone RNA-dependent RNA polymerase, as well as through recombination(Graaf et al., 2016; Green, 2016). NoV antigenicity consistently evolves to evade the human immune response and persist in the human population. These changes in the viral genome can reduce the complementarity of the primers and probes used in RT-PCR detection of NoV, potentially decreasing the detection rate of currently circulating strains and limiting the utility of these method(Vinjé, 2015).

RT-PCR cannot distinguish between free-floating viral RNA versus RNA from an infectious virion(Wang et al., 2021). As a result, patients may test positive for extended periods, even after symptoms have resolved or in the absence of symptoms. The relevance of these molecular detections remains unclear. Currently, cell culture models for growing NoV are not widely available, so it is difficult to determine whether NoV RNA detected through RT-PCR identifies an infectious virus or just viral RNA remnants. A similar phenomenon has been observed in other viral diseases, including SARS-CoV-2 infections during the COVID-19 pandemic(Bruce et al., 2022; Cevik et al., 2021).

An important issue in NoV epidemiology is the differentiation between infectious and non-infectious NoV among RT-PCR-positive patients. Once NoV diagnoses are made, it is unclear whether and for how long patients are shedding infectious virus in stool or shedding lingering viral RNA or potentially non-viable virions. These distinctions are highly relevant for understanding viral transmission and accurately estimating the duration of contagiousness. Determining the typical length of infectious NoV shedding may directly influence public health approaches to contain NoV outbreaks around the globe. Development of tools to measure this distinction will also be crucial in evaluating the effectiveness of vaccines and therapeutics to curtail the spread of NoV.

Based on work from various RNA virus fields, it is evident that viral RNA detection alone does not provide enough information to determine infectivity. A major obstacle to testing the correlation of NoV viral RNA detection with infectious NoV is the lack of a widely accessible virus culture model. Current culture models use specialized human intestinal enteroid systems(Estes et al., 2019), which have allowed progress in the understanding of NoV infectivity. Nevertheless, those methods may be slow, technically challenging, and genotype-dependent(Ettayebi et al., 2021; Jones et al., 2015; Winder et al., 2022). In the meantime, it is necessary to assess alternative methods that can complement RNA-based diagnostics.

1.1. Measuring NoV proteins for detection

1.1.1. Norovirus proteins

Viral proteins, expressed during the replication cycle of the virus, offer complementary targets for detecting NoV infection. NoVs encode two structural capsid proteins (VP1, VP2) and six nonstructural proteins (NS1/2, NS3-7) (Fig. 1). VP1 and VP2 form the capsid protecting the viral RNA genome and are the main immunogenic targets in the anti-NoV antibody response (reviewed in(Ford-Siltz et al., 2021)). The nonstructural proteins replicate the viral RNA genome inside infected host cells and modify the host environment to enhance infection (Campillay-Véliz et al., 2020). NS proteins are typically intracellular and expressed only during active viral replication, though a recent study found that murine NoV NS1 is secreted in the gastrointestinal environment of infected mice(Lee et al., 2019).

Fig. 1.

Human norovirus (NoV) virion and genome composition.

A. NoV virions contain an outer protein shell composed of viral protein 1 (VP1) multimers, interacting with VP2 proteins on the interior of the virion. Inside the virion, the single-stranded, positive-sense viral RNA (+ssRNA) is bound to the VPg protein at the 5′ end.

B. During infection, several open reading frames (ORFs) are expressed. ORF1 encodes for nonstructural proteins 1–7 (NS1-7), which carry out various important functions for viral replication within the host cell, though a cleavage product of NS1/2 is secreted(Lee et al., 2019). ORF2 and ORF3 encode for VP1 and VP2 proteins respectively. NTPase: nucleotide-triphosphatase, VPg: viral protein genome-linked, RdRP: RNA-dependent RNA polymerase.

Measuring viral proteins as correlates of infection has proven to be effective in other viral diseases, including COVID-19, where virus culture models have been more tractable and allow for direct testing of infectious virus in patient samples. Several groups have found that specific protein antigen-based testing was a more effective correlate of infection than RT-PCR(Pekosz et al., 2021; Pickering et al., 2021). That previous work supports the validity of using NoV proteins as markers for infectious virions in the absence of direct titering assays.

1.1.2. Immuno-based detection

Detection of NoV protein antigens has commonly been performed by antibody-based immunoassays, such as rapid antigen tests and enzyme-linked immunosorbent assays (ELISAs), which capture a target antigen with antibodies anchored to a solid surface(Kirby et al., 2010; Vinjé, 2015). These methods are limited by the number and quality of antibodies to NoV genotypes pre-selected for testing, biasing testing based on reagent availability and genotype predictions(Ambert-Balay and Pothier, 2013). This method is also potentially susceptible to antigenic changes in NoV VP1/VP2 caused by evolutionary pressure from the human immune system(Vinjé, 2015). These changes can disrupt the antigen-antibody interaction necessary for ELISAs and antigen tests and potentially affect detection sensitivity, requiring lengthy generation and validation of new antibodies. Furthermore, these diagnostics focus on just one viral protein antigen, while all other proteins present in a sample, ranging from viral nonstructural to host to microbiome proteins, are ignored and narrow the scope of information available from any one sample. These limitations highlight the need for unbiased, non-antibody-based methods which are less susceptible to viral protein evolution and can measure the full breadth of proteins present in samples.

1.1.3. Proteomics-based methods for norovirus detection

Mass spectrometry (MS) proteomics is a powerful, comprehensive tool to interrogate the protein composition of biological samples. While it has yet to become a common diagnostic method in clinical settings, we believe it offers several potential advantages for measuring viral infection in clinical samples. MS proteomics does not rely on antigen-antibody interactions for protein measurement and is not affected by viral antigen changes like immune-based detection methods. Evolutionary changes in viral protein amino acid sequences can be accounted for in post-sample analysis by updating the sequence databases used to identify proteins in a sample. This eliminates the need to generate new primers and probes for each new genotype prior to testing like in RT-PCR methods.

Several recent studies from other viral fields offer precedents for applying MS proteomics to clinical research. For instance, a multiplexed targeted assay for flavivirus diagnostics was recently reported, which capitalizes on the amino acid differences in the secreted flavivirus protein NS1 to differentiate between multiple flaviviruses(Wee et al., 2019). This method could be advantageous over serology-based methods, which can be confounded by cross-reactive antibodies. The COVID-19 pandemic further accelerated the development of proteomics-based clinical diagnostic methods. Using proteomics approaches, SARS-CoV-2 proteins have been detected from a variety of patient samples, including nasopharyngeal swabs(Cardozo et al., 2020), bronchoalveolar lavage(Zecha et al., 2020), and gargle solution(Ihling et al., 2020). For a comprehensive review on SARS-CoV-2 mass spectrometry diagnostics see Grossegesse et al.(Grossegesse et al., 2020).

While still a burgeoning field, the application of MS proteomics to measure NoV infection is underway. The first MS characterization of NoV compared two methods, matrix-assisted laser desorption ionization and nanospray MS, for capsid detection(Colquhoun et al., 2006). The authors benchmarked their targeted assay using virus-like particle (VLP) standards, detecting 16 target peptides of the VP1 capsid protein. To extend their method to complex stool samples, the authors applied a virus purification protocol to NoV-negative stool samples and then spiked known amounts of NoV VLPs into extracts. They were able to detect viral peptides at as low as 250 fmol of capsid protein, validating the sensitivity of MS to detect low abundance viral proteins in human stool and opening the door to detect NoV proteins in NoV-positive patient samples.

A key requirement to identify NoV in human samples is the ability to detect and distinguish between various viral strains and genotypes. A recent report established the feasibility of an MS-based method to distinguish between three separate strains of the GII.4 genotype(Chu et al., 2021). VLPs encoding the GII.4 2006b, 2006a, or 2009a strain VP1 proteins were recombinantly produced and analyzed by liquid chromatography-mass spectrometry (LC/MS). All three VLP types were detected with up to 98 % sequence coverage of VP1, and 10 post-translational phosphorylation modification sites were identified, supporting the applicability of MS to distinguish between NoV genotypes and make de novo observations on viral protein characteristics.

Most recently, a targeted parallel reaction monitoring (PRM) assay was applied to differentiate between six NoV genotype VP1 proteins recombinantly expressed(Kimura et al., 2022). An inclusion list of twelve unique target peptides was created to identify viral proteins extracted from NoV-positive human stool samples from four patients experiencing AGE. Heavy isotype-labeled peptide standards were spiked into digested protein extracts and used for absolute quantification of target peptides in subsequent PRM analysis. The genotype of patients (GII.2, GII.3, and GII.4) was correctly detected and matched as compared to RT-PCR detection. Furthermore, host and microbiome proteins were also detected in non-targeted MS analysis of the same samples, potentially lending useful information to other protein correlates of AGE disease. While this study only analyzed a small number of clinical samples, this work provides proof-of-principle for using MS methods to systematically analyze a larger group of patient samples.

Given the often-low abundance of viral proteins in clinical samples (Grossegesse et al., 2020), particularly complex stool samples, antigen sensitivity is a crucial component of detection. Traditional antigen tests can underperform on sensitivity with clinical samples. A recent study of 358 clinical samples found that a commercial routine diagnostic antigen test achieved only 27.5 % versus the manufacturer reported 92 % sensitivity(Thangjui et al., 2020).

In contrast, the high sensitivity of a mass spectrometer lends itself to detect and quantify viral proteins. Colquhoun et al. were able to detect viral capsid peptides at as low as 250 fmol using instrumentation available in the early 2000s(Colquhoun et al., 2006). Since then, MS instrumentation has advanced significantly(Yates, 2019), enabling more sensitive detection of smaller amounts of viral protein. Chu et al. were able to measure as low as 3.6 fmol of VP1 protein from purified virus-like particles while still retaining over 94 % sequence coverage (Chu et al., 2021). This high sequence coverage in such low protein amounts could enable phylogenetic analysis. Most recently, Kimura et al. measured as low as 5.99 fmol VP1/μg protein from clinical patient stool samples that were PCR positive (Ct value = 18)(Kimura et al., 2022). One can estimate the lower limit of detection of proteomic analysis based on the number of VP1 molecules detected in this study. Kimura et al. injected 20 μg of a 545 μg protein sample extracted from 100 to 200 μL (100–200 mg) of feces for MS analysis and were able to measure 5.99 fmol VP1/μg protein. Given that one NoV virion is composed of 90 VP1 dimers, 5.99 fmol VP1/μg protein corresponds to 3.6 x 10⁹ VP1 molecules/μg protein (5.99 x 10⁻¹⁵ mol VP1/μg protein x 6.022 x 10²³ mol⁻¹). This equates to 2 x 10⁷ virions/μg protein (3.6 x 10⁹ VP1 molecules/μg protein ÷ 90 VP1 dimers ÷ 2) as the lower limit of detection. Higher protein input amounts and improvements in stool extraction and concentration could further increase the sensitivity, which will be necessary to enable the analysis of samples with low copy numbers. Work with other viruses, such as flaviviruses (Wee et al., 2019) and coronaviruses (Grossegesse et al., 2020), has shown that accurate and meaningful measurements can be made with small sample volumes relevant to clinical samples. Head-to-head comparisons of sensitivity and specificity between RNA-based and MS proteomics detection methods for NoV are still lacking, and these studies should be performed to benchmark the performance of proteomics to existing methods.

1.2. Future perspectives on proteomics for NoV studies

1.2.1. Potential breadth of microbe detection using quantitative proteomics

Moving forward, quantitative proteomics is poised as a powerful tool for making new discoveries with a systems-level perspective on clinical samples. Improvements in instrumentation are enabling both wider breadth of microbe detection and higher sensitivity to detect low-abundance proteins. Applications of proteomics to systems biology studies of infection will likely parallel the resurgence in metagenomics and Next Generation Sequencing (NGS) for studying microbes, including pathogens, in clinical samples. Advances in sequencing have enabled important research on the evolution of circulating NoV strains(Brown et al., 2016; Tohma et al., 2021).

Likewise, proteomics can be used to detect a variety of viruses, bacteria, and other microbes, including pathogens, within a single sample. This breadth of coverage will be essential in disentangling the dynamics of co-infections, which can occur with NoV infection, either with different genotypes of NoV or entirely separate pathogens. Unlike common clinical PCR detection methods, such as Taqman array cards, which require predefined primer sets, proteomics can survey the entire pathogen proteome within one MS run, and the data can be continually re-mined to quantify various pathogens of interest. Quantitative proteomics could be particularly useful in characterizing the influence of the microbiome on NoV dynamics, as the large and complex swath of microbes affects the patient’s gastrointestinal environment. A recent analysis of a NoV human challenge study found that there are significant differences in gut microbiome composition between asymptomatic and symptomatic participants infected with NoV(Patin et al., 2020). With recent advances in metaproteomics (the application of proteomics to measure the microbiome)(Long et al., 2020; Stamboulian et al., 2022), applying proteomics to better understand the mechanism of interplay between the gut microbiome and NoV is becoming feasible.

Furthermore, for a given viral pathogen, a variety of protein targets can be measured, enabling better insight into replication dynamics. Beyond structural proteins, NS proteins offer alternative targets for viral replication, given that these proteins are only produced once host cells are infected(Campillay-Véliz et al., 2020). Of particular interest for NoV is NS1, which was recently reported to be secreted during murine NoV infection(Lee et al., 2019). Lee et al. found that secretion of NS1 plays an important role in suppressing the host interferon-gamma response and helps mediate pathogenesis during infection. Furthermore, vaccinating mice with NS1 was protective against subsequent murine NoV challenge. While NoV NS1 secretion and function in human infection remains to be shown, Lee et al. did find that recombinant human norovirus (G1.1) NS1 is secreted in a human cell culture model (HEK293T cells) when expressed recombinantly using a FLAG-NS1/2 expression system. Thus, NS1 may be an alternative viral protein to track in human stool samples as a potential indicator of active infection. It remains to be tested whether NS1 or other NS proteins can be detected in human stool samples, given their likely lower abundance. Additionally, the kinetics of NS proteins in excreted samples and their stability will be important factors to determine. Quantitative assessments of NS protein levels and kinetics in excreted stool samples using proteomics could provide valuable information on NoV infection progression.

1.2.2. Identification of host protein markers

Beyond pathogen-specific proteins, proteomics can be used to measure host proteins present in clinical stool samples. Detection of proteins that correlate with AGE severity can identify potential biomarkers, which may help predict clinical outcomes and better our understanding of disease progression. Non-targeted proteomics workflows can be used to characterize thousands of host proteins present in samples in an unbiased manner, enabling discovery of interesting proteins without previous knowledge of a role in AGE disease. Previous work has assessed the dynamics of micronutrients and inflammatory response biomarkers in serum of adult NoV challenge participants(Williams et al., 2019). That study measured a select few predetermined inflammation protein markers and micronutrients by sandwich ELISA and HPLC. More recent work has applied multiplexed MS assays to quantify targeted blood and stool protein markers in the context of inflammatory bowel disease (Vidova et al., 2021). Expanding on the principle of this study with proteomics analyses of NoV stool samples will allow for a more detailed profiling of changes in host factors in response to NoV infection.

1.2.3. Measuring NoV components in patient samples to understand infectivity

As stated previously, NoV infections may be asymptomatic, and prolonged RNA positivity for NoV has been detected in convalescent patients even after symptoms have resolved. It is unclear whether these patients are still shedding viable virus, or if this observation is either a technical artifact or a result of lingering viral RNA in stool samples. Characterization of the viral proteome within these samples may complement RT-PCR positivity data to elucidate whether patients are shedding components of infectious NoV (Fig. 2). To test this approach, it would be necessary to examine sequential samples from AGE patients with RT-PCR positivity to measure viral and host proteins in stool samples throughout, and ideally beyond, their clinical recovery. This sampling approach will allow for comparison of the correlation of viral and host factors with viral RNA, as well as overall disease severity. Furthermore, longitudinal studies may offer insight into the kinetics of viral RNA and protein abundance for the duration of infection, both in the presence and absence of clinical symptoms.

Fig. 2.

Measurable metrics from NoV patients to potentially determine infectivity.

Stool samples from NoV-infected patients contain a varied mixture of infectious NoV virions, viral RNA, viral proteins, host proteins, and microbiota proteins. In the absence of widely accessible NoV titer assays, measuring host, viral, and bacterial proteins with proteomics by mass spectrometry can potentially provide an abundance of information on the dynamics of NoV infectivity in combination with traditional measurements of viral RNA (by RT-PCR) and symptoms. Future work is needed to measure these metrics in a clinical cohort and interrogate the relationship and potential correlations between protein, RNA, and symptoms in order to assess the implications for transmission and disease outcomes.

A logical extension of this detection framework would be to understand how therapeutic or prophylactic interventions, such as NoV vaccines, affect the progression of disease. While a primary objective of vaccines is to reduce the incidence and severity of disease, some vaccines may also mediate indirect effects by reducing transmissibility of NoV by infected individuals. An important, complementary objective would then be to measure whether vaccination alters the intensity or length of viral shedding. In the absence of widely utilized NoV diagnostics, the detection of viral RNA and protein abundance in patient samples can be used in clinical trials to assess whether vaccination can reduce infectious virus shedding. Such information could also help clinicians decide how to implement non-therapeutic interventions to curb NoV spread even when vaccines are being used.

1.2.4. Challenges and future perspectives

The primary obstacles to widespread use of MS to diagnose clinical samples are accessibility and scalability. MS proteomics requires specialized training and expensive equipment to operate. In addition, the speed of high-throughput ELISAs currently outpaces MS sample preparation and analysis. Nevertheless, recent technological developments can help close some of this gap. Automation of MS sample preparation significantly improves the scalability of high-throughput sample preparation and reduces technical variation introduced by manual sample preparation(King et al., 2022). Advances in label-free and data-independent acquisition methods for MS analysis can also reduce the associated costs and improve the quantity of information gleaned from each sample(Meyer and Schilling, 2017; Verma et al., 2021). Additionally, alternative technologies for single-molecule proteomics, such as protein nanopore sequencing(Martin-Baniandres et al., 2023), single-molecule Edman fluorosequencing, and solid-state biosensors(Alfaro et al., 2021), are being explored, though these are still in development.

However, the high costs of MS and automation instruments make it unlikely that clinical settings, particularly those in low- and middle-income countries (LMICs), will be able to have and operate their own MS instrument in the near future. Rather, this presents an opportunity for collaboration between basic science laboratories within LMICs or laboratories abroad to engage in discovery-oriented research. Clinical trials and prospective cohort studies produce an abundant bank of human samples that can be mined for basic science discoveries on the nature of infectious disease in addition to the primary clinical objectives of those studies. This approach can multiply the utility of each sample collected and elevate the significance of proteomics research. The ultimate goal of these partnerships should be to find new trends of pathogenesis or biological targets that can be translated back to the clinic. This new knowledge may inform the development of alternative diagnostics, as in the case of additional point-of-care tests, or for the evaluation of treatment effectiveness, such as in vaccine clinical trials.

In conclusion, MS-based proteomics is a powerful, comprehensive tool to explore the dynamics of NoV infection and could open the door to discovering new patterns and targets of pathogenesis that can aid in the evaluation of novel therapeutics.

Data availability

No data was used for the research described in the article.