Efficacy of alcohol‐based hand sanitizers against human norovirus using RNase‐RT‐qPCR with validation by human intestinal enteroid replication

In vitro suspension assay followed by RNase/RT‐qPCR to test hand sanitizer efficacy

The commercial products tested by suspension assay followed by RNase‐RT‐qPCR were more efficacious against HuNoV (sample NV14‐017) compared to the benchmark of 60% ethanol (Table 1; P < 0·05). In addition, when there was no additional soil added to the suspension (~2·5% faecally derived residual soil only), Product A outperformed Product B (P < 0·05). These results are consistent with those of another study (Park et al. 2010), which found that exposure to 70% ethanol produced variable reduction in viral RNA, ranging from 0·5 to 3·0 log₁₀ GEC reduction for GII.4 HuNoV strains. A prior study characterizing the efficacy of various concentrations of ethanol and an alcohol‐based hand sanitizer on HuNoV in suspension and on human fingerpads respectively also using the RNase‐RT‐qPCR method, showed limited product efficacy (Liu et al. 2009). In that study, <0·5 log₁₀ reduction in HuNoV GEC was shown for ethanol suspensions, and a <0·4 log₁₀ reduction in HuNoV GEC was shown for fingerpad studies with the commercial hand sanitizer tested. The efficacy of alcohol‐based hand sanitizers against HuNoV has been debated in the literature, and in the absence of an in vitro assay to determine infectivity, researchers have had to rely on RNase‐RT‐qPCR or cultivable surrogates to estimate product efficacy (Macinga et al. 2008; Liu et al. 2009; Park et al. 2010; Shimizu‐Onda et al. 2013; Cromeans et al. 2014; Tuladhar et al. 2015).

In vitro suspension assay followed by HIE challenge to test hand sanitizer efficacy

Optimization experiments were necessary before it was possible to use the HIE model to test sanitizer efficacy. In a head‐to‐head comparison, commercial media out‐performed in‐house media, providing higher cell density (80–100 enteroids/field vs 20–30 enteroids/field respectively) and almost twofold faster cell growth (7 days vs 10–12 days). Although more expensive than the reagents produced in‐house, the commercial media had a streamlining advantage since there was not need to produce our own growth factors, resulting in significant savings in both time and labour costs.

Of the two GII.4 Sydney faecal suspensions used in this study, only the NV14‐117 inoculum produced efficient and consistent replication (2·8 ± 0·5 log₁₀ GEC increase at 72 hpi; no treatment control–Fig. 1) in the HIE, and only if first purified by serial filtration. The lack of replication of the NV14‐017 inoculum used in the RNase‐RT‐qPCR studies above was not due to differences in the viral RNA polymerase sequence (data not shown). The variable replication efficiency of various HuNoV strains to replicate in HIE has been reported by others (Ettayebi et al. 2016; Costantini et al. 2018) and appears to be the function of many different factors which range from differences in genotype to faecal sample to faecal sample variation.

Consistently, we found that it was necessary to use some means by which to remove residual disinfectant, surfactants, and/or neutralization buffer prior to challenging the HIE model. Otherwise, extensive cytotoxicity was observed. Treatment with spin columns, as per ASTM E1482‐12, produced a small (0·3 ± 0·15 log₁₀ GEC) but insignificant loss in virus titre (P > 0·05) compared to the positive control. Even after the use of spin columns for detergent removal, we could not overcome the cytotoxic effect of Product B in the HIE model, so further confirmation of the efficacy of this product by infectivity assay was not possible. This highlights the sensitivity of the HIE model to residual cytotoxic substances and the need to investigate alternative ways to more effectively remove myriad residual formulation components to facilitate more universal use of the HIE model in future disinfection studies.

In the HIE model, neutralization controls produced a 1·97 ± 0·27 log₁₀ HuNoV GEC increase (Fig. 1), suggesting the assay provided about 2 log₁₀ replication efficiency, which is similar to that reported in the literature (Ettayebi et al. 2016; Costantini et al. 2018). No virus replication in HIE was observed after treating GII.4 Sydney NV14‐117 with Product A as evidenced in the absence of a RT‐qPCR signal in the 72‐h post infection (hpi) assays (Fig. 1). At 1 hpi, 1·72 ± 0·50 log₁₀ HuNoV GEC was detected from samples treated with Product A, in comparison to 3·60 ± 0·60 and 3·41 ± 0·21 log₁₀ HuNoV GEC RT‐qPCR signal from the positive control samples and 60% ethanol samples respectively (data not shown), suggesting that treatment with Product A affected capsid binding efficiency. The samples treated with 60% ethanol produced an RT‐qPCR signal consistent with a 0·86 ± 0·16 log₁₀ GEC increase at 72 hpi, showing that ethanol produced some inactivation of HuNoV but did not completely eliminate its infectivity (all six of six replicates showing viral replication). This is consistent with another study which showed slightly reduced HuNoV inactivation (up to 0·7 log₁₀ GEC lower (72 hpi) than the neutralization control) after treatment with ethanol and no inactivation of HuNoV after treatment by isopropanol using the HIE model (Costantini et al. 2018).

It should be noted that the virus inactivation data collected using RNase‐RT‐qPCR and the HIE model are not directly comparable, and the use of the HIE model in this study was to validate the RNase‐RT‐qPCR results. While the HuNoV GII.4 Sydney genotype was used in both studies, the inocula consisted of two unique faecal suspensions obtained from two different infected individuals. Preliminary screening of several GII.4 Sydney‐positive stool samples revealed that sample ID NV14‐117 showed superior HIE replication efficiency, but for unknown reasons. Others have reported similar sample‐to‐sample variation in performance on the HIE model, even when the HuNoV strains were the same genotype (Ettayebi et al. 2016; Costantini et al. 2018). For the HIE model, it was also necessary to purify NV14‐117 by serial filtration prior to exposure to the sanitizers, consistent with other studies (Ettayebi et al. 2016). This step likely resulted in removal of some residual organic matter associated with the resuspended stool, as well as disaggregation of viral particles (Escudero‐Abarca et al. 2014). This could have provided conditions more favourable to virus inactivation compared to the unfiltered 20% faecal suspension used in the RNase‐RT‐qPCR experiments. Unfortunately, these sample manipulations were necessary in order to effectively use the HIE model for this application.

One of the major limitations of the current HIE model for HuNoV is its lack of quantifiability. Previous studies investigating the efficacy of alcohol‐based hand sanitizers against HuNoV surrogates report results as log₁₀ reductions thanks to the availability of quantifiable infectivity assays, such as plaque and TCID₅₀ assays (Macinga et al. 2008; Park et al. 2010; Shimizu‐Onda et al. 2013). In order to develop a quantifiable infectivity assay for HuNoV disinfection studies, a greater understanding of the mechanism of infection of HuNoV in the intestinal epithelium is required. The cellular factors that regulate HuNoV binding and uptake into intestinal cells for successful infection are still unknown (Haga et al. 2020). Ongoing research in this field is needed to develop a quantifiable infectivity assay for HuNoV.

HuNoV strains

GII.4 Sydney strains obtained as deidentified stool specimens from outbreaks were suspended 20% in PBS with clarification by centrifugation (3100 g for 5 min at 4°C). Residual soil (organic) load in the working suspensions was approximately 2·5% as previously reported (Moorman et al. 2017). Samples obtained from two different naturally infected individuals were used in this study, both with initial titres of 6–7 log₁₀ GEC per ml. Sample designation NV14‐017 was used without further preparation for in vitro suspension assays evaluated by RNase‐RT‐qPCR. Sample designation NV14‐117 was used in suspension assays tested using the HIE model. To facilitate infection and reduce cytotoxicity in the HIE studies, it was necessary to further purify this stock by serial filtration through 0·8, 0·45 and 0·22 µm filters (Millipore Sigma, Burlington, MA) prior to treatment with disinfectants, consistent with previous studies (Ettayebi et al. 2016).

Products screened

Two commercial alcohol‐based hand sanitizers (GOJO Industries, Inc., Akron, OH), PURELL^® VF PLUS™ (Product A) and PURELL VF 481™ (Product B) and 60% ethanol (v/v) were screened in parallel in this study (Table 2) by in vitro suspension assay (ASTM E1052‐11) followed by evaluation of virus inactivation (i) directly by RNase‐RT‐qPCR; and (ii) by HIE culture with pre‐ and post‐infection RT‐qPCR.

In vitro suspension assay followed by RNase‐RT‐qPCR for evaluation of product efficacy

Virucidal suspension assays of GII.4 Sydney sample NV14‐017 for the in vitro study were done in accordance with ASTM standard E1052‐11, with minor modifications. Briefly, a 25 µl volume of the 20% virus faecal suspension, and the same volume of the faecal suspension supplemented with an additional 2·5% soil (prepared according to ASTM standard E1053‐11, for an approximate total soil load of 5%) was mixed with 225 µl of the products, 60% ethanol, or PBS. After 30 and 60 s contact times, 20 µl of this solution was added to 180 µl of 10% D/E neutralization broth (Sigma‐Aldrich, St Louis, MO). Suspensions were held frozen at −80°C until assayed. Prior to RNA extraction, samples were treated with RNase to eliminate free RNA. For RNase pre‐treatment, 2 μl RNase One (Promega, Madison, WI) and 22 μl reaction buffer were added to 200 μl of the post‐neutralization sample eluate and incubated at 37°C for 15 min. Samples were placed on ice for 5 min to abolish RNase enzyme activity prior to RNA extraction, which was performed using the automated NucliSENS^® easyMag^® system (bioMerieux, Durham, NC) as per the manufacturer’s instructions. RNA was eluted in a 25 μl volume of proprietary NucliSENS elution buffer. Viral RNA was quantified by RT‐qPCR targeting the conserved OFR1‐ORF2 junction of GII HuNoV (Jothikumar et al. 2005). The resulting C_T values were extrapolated to log₁₀ genome equivalent copies (GEC) by comparison to a standard curve produced by RT‐qPCR amplification of HuNoV GII.4 Sydney RNA obtained from the initial inoculum. Results were presented as log₁₀ reduction in GEC of HuNoV.

Human intestinal enteroid monolayer production

Human intestinal enteroids (HIE; jejunal) were kindly provided by Dr Mary Estes, Baylor College of Medicine (BCM), Houston, TX. They were initially grown as multilobular 3D cultures in matrigel using in‐house media and were passaged after 7 days. In‐house media required maintenance of three separate cell lines to produce key growth factors: (i) Wnt3A cells (ATCC, CRL‐2647); (ii) 293T‐HA Rpo‐FC cells to produce R‐Spondin (Trevigen, Gaithersburg, MD); and (iii) 293 Noggin cells (provided by BCM colleagues). Complete media with growth factors (CMGF+) was required to grow undifferentiated enteroids, and CMGF+ with omission/reduction of key growth factors (CMGF−) was required to obtain differentiated enteroids. This system (Ettayebi et al. 2016) was simplified with the recent availability of commercial media (Intesticult, STEMCELL, Technologies, Kent, WA) used as per manufacturer recommendations.

Three‐dimensional cultures were dissociated into single cell suspensions in CMGF+ supplemented with 10 µmol l⁻¹ Y‐27632 (Sigma Aldrich, St Louis, MO) and plated into 96‐well plates coated with collagen IV (Corning Life Sciences, Tewksbury, MA) at a seeding density of 1–2 × 10⁵ cells/well as undifferentiated monolayers (Costantini et al. 2018). After 48 h the media was replaced with differentiation medium and after 4 days of incubation, a confluent monolayer of differentiated enteroids was available for infection studies.

In vitro suspension assay followed by HIE culture for evaluation of product efficacy

The virucidal suspension assays performed in this phase of the study were similar to the ASTM E1052‐11 described above except for: (i) the use of highly purified HuNoV GII.4 Sydney sample NV14‐117, which showed superior replication efficiency in the HIE model (data not shown); (ii) a smaller inoculum volume (10 µl) was used, with associated volume reduction for sanitizer and neutralization; (iii) CMGF‐ supplemented with 10% faetal bovine serum was used as the neutralizer; (iv) only Product A and 60% ethanol at the 60 s contact time without soil was tested; and (v) after neutralization, the samples were purified using detergent removal spin columns (Thermofisher, Rockford, IL) to reduce residual cytotoxicity, as per ASTM E1482‐12. Product B produced cytotoxic effects in the model despite the use of the detergent removal spin columns, so further studies with this product in the HIE model were not possible.

HIE infection

The positive control consisted of purified HuNoV GII.4 Sydney, strain NV14‐117, having a titre of 6–7 log₁₀ GEC per ml. Monolayer cultures were infected with 100 µl of diluted (i) untreated virus stock (no treatment control); (ii) virus stock exposed to neutralized product (neutralization control); and (iii) virus stock exposed to product followed by neutralization (treatment). These were then incubated for 1 h to facilitate virus binding, followed by washing with CMGF‐ and overlay with 100 µl of differentiation media containing 500 µmol l⁻¹ sodium glycochenodeoxycholate (Sigma Aldrich). For each set of infections, duplicate plates were prepared; one plate was removed and stored at −80°C immediately post infection (corresponding to 1 h post‐infection (hpi)) and another was incubated for 72 h prior to freezing (constituting 72 hpi). Frozen plates were subjected to RNA extraction using the Direct‐zol RNA kit (Zymo Research, Irvine, CA) according to the manufacturer’s instructions, with RNA elution in a 25 µl volume of DNase/RNase free water. Viral RNA was quantified by RT‐qPCR targeting the conserved ORF1‐ORF2 junction of GII HuNoV (Jothikumar et al. 2005). The resulting C_T values were extrapolated to log₁₀ GEC by comparison to a standard curve produced by RT‐qPCR amplification of HuNoV GII.4 Sydney RNA obtained from the initial inoculum. Results were presented as log₁₀ increase in GEC of HuNoV.

Statistical analysis

Suspension assay experiments with RNase‐RT‐qPCR enumeration were done in independent triplicates. Suspension assays followed by HIE infectivity assays were completed in two wells for two to six independent runs, yielding 4–12 replicates. When RNase‐RT‐qPCR was used as the quantification method, results were expressed as mean ± standard deviation of log₁₀ reduction in HuNoV GEC. When the HIE model was used, log₁₀ HuNoV GEC increase after 72 hpi was used to assess whether HuNoV infectivity was abolished by sanitizer treatment. Results were compared statistically using anova and the Tukey‐Kramer test (Sigma Plot; Systat Software Inc., San Jose, CA). Statistical significance was established at a level of P < 0·05.