Salivary flow and turbidity development inconsistently associated with lower taste intensity of vegetables and juices

Abstract

The same phytochemicals that stimulate aversive sensations are often also responsible for purported health benefits in fruits and vegetables. Prior work indicates that some salivary proteins may reduce aversiveness of phytochemicals. In rodents, the salivary binding proteins have been shown to reduce bitter taste of polyphenols and alkaloids, but work in humans has focused primarily on polyphenol astringency (dry, rough, or puckery sensations). In this study, we tested if tastes of vegetable products might correlate to either salivary flow rate or the polyphenol binding capability of saliva, as measured by turbidity development when saliva is mixed with tannic acid. Participants (N=26) provided chewing-stimulated saliva samples and rated five juices and two chopped vegetables for bitterness, sourness, and sweetness intensity. Saliva was mixed with tannic acid and light absorbance was measured for quantification of haze development. Greater absorbance corresponded to less bitterness for one green vegetable juice blend, less sweetness for two green vegetable juices and chopped kale, and less sourness from cranberry juice. Greater salivary flow corresponded to less bitterness from chopped brussels sprouts, and less sweetness from one green vegetable juice blend and chopped kale. These findings indicate that greater salivary flow rate and presence of certain salivary binding proteins is not universally associated with lower aversive tastes from phytochemical-containing foods. Whether associations between these salivary properties are ingredient specific or simply not robustly related to taste in commercial products should be further investigated.

1. Introduction

Despite widespread recommendations and emphasis on the importance of consuming vegetables as part of a healthy dietary pattern, vegetable consumption in the United States is still low. Almost 90 percent of the U.S. population consumes less than the Dietary Guidelines recommended intake (U.S. Department of Agriculture and U.S. Department of Health and Human Services, 2020). While there are many possible barriers to vegetable consumption, two of the most commonly cited barriers include time (related to buying and preparing vegetables) and preference (related to poor taste of vegetables, or competing tastier foods) (De Leon et al., 2020).

One way to reduce the time burden of vegetable consumption could include the use of vegetable juices, though this must be balanced against the sugar content that is commonly high in these drinks. One randomized controlled trial successfully implemented the addition of daily vegetable juice consumption to help participants reach recommended intakes (Shenoy et al., 2010). Furthermore, some consumers already purchase 100% vegetable or mixed fruit and vegetable juices. In one report, among consumers who purchased juice or juice drinks over the previous three months, 34% and 38% purchased 100% vegetable juice or fruit and vegetable juice blends, respectively (Formanski, 2021). Extra vitamin consumption was the most cited reason for buying more juice/juice drinks, and reducing sugar intake was the most cited reason for buying less juice/juice drinks (Formanski, 2021), indicating that lower sugar vegetable juices may have consumer appeal. However, the barrier of poor taste for vegetables remains.

Vegetables are recommended as a part of a healthy diet largely because of substantial epidemiological evidence associating vegetable consumption with decreased risk of many cancers, cardiovascular disease, and stroke, among others (Van duyn & Pivonka, 2000). Plant-based bioactive compounds called phytochemicals are responsible for many of these protective roles (Blekkenhorst et al., 2018). Some of the prominent phytochemicals of interest for human health include polyphenols and isothiocyanates found in Brassica vegetables such as broccoli, Brussels sprouts, cabbage, and kale (Raiola et al., 2018). Unfortunately, these compounds and their derivatives are the main contributors of bitter, astringent, stinky, or otherwise generally aversive sensory features of these foods (Drewnowski & Gomez-Carneros, 2000). For example, bitter and/or stinky isothiocyanates are created in Brassica vegetables upon tissue disruption and activation of an enzyme called myrosinase (Wieczorek et al., 2018). However, vegetables are not universally disliked (Cox et al., 2012), so exploring potential influencing characteristics for their perception is warranted.

Saliva is one factor that has documented influence on taste perception. Saliva is an “unavoidable ingredient” (Mosca & Chen, 2017) during food and beverage intake that aids in processing and solubilization, among other duties (Humphrey & Williamson, 2001; Running, 2018). Many salivary characteristics have been studied related to taste perception, including effects of flow rate, buffer capacity, protein composition, and enzyme activity. Salivary protein interactions with polyphenols or alkaloids have been demonstrated to influence associated taste and astringency sensations, and the concentrations of these proteins may be associated with dietary patterns (Crawford & Running, 2020; Davis & Running, 2021; Dinnella et al., 2009, 2010; Fleming et al., 2016; Horne et al., 2002; Martin et al., 2018; Martin, Nikonova, et al., 2019; Martin et al., 2020; Torregrossa et al., 2014). Horne et al. demonstrated that development of haze in mixtures of saliva and tannic acid (a polyphenol) was inversely related to astringency ratings of tannic acid solutions (Horne et al., 2002). This turbidity development, an indication of the complexation of salivary proteins with polyphenols, corresponded to a protective effect of salivary proteins against the often-disliked sensation of astringency. Interestingly, however, prior animal work examining the effects of polyphenol consumption on salivary characteristics has almost exclusively focused on measures of bitter taste rather than astringency (Martin et al., 2018; Torregrossa et al., 2014). Notably, many polyphenols found in commonly-consumed foods activate human bitter taste receptors (Soares et al., 2018). Taken together, this indicates that the saliva/polyphenol interactions described above could have implications for bitter taste as well as astringency. Further, additional animal work has also demonstrated similar responses of saliva and subsequent taste sensitivity and perception after consumption of other bitter (non-polyphenol) phytochemicals (Martin et al., 2020; Martin, Kay, et al., 2019; Martin, Nikonova, et al., 2019). Many foods contain other phytochemicals that elicit taste qualities, particularly bitterness, that saliva may influence, which have yet to be evaluated with these methods.

Thus, we designed this study as an initial investigation into whether taste intensity for bitterness, sourness, and sweetness in juices and two green vegetables would correspond to haze/turbidity in saliva-tannic acid mixtures. We hypothesized that bitterness, like astringency in prior work, would be reduced for participants whose saliva developed more turbidity when mixed with tannic acid. We also measured salivary flow rate and hypothesized that increased salivary flow would relate to reduced bitterness and sourness intensity. We did not expect to see any changes in sweetness intensity related to either turbidity or salivary flow rate.

2. Methods

All methods were approved the Purdue University Institutional Review Board (Purdue IRB 1706019362), and all participants provided written informed consent. This work was completed in person, prior to the COVID-19 pandemic (we include this note as taste/smell abnormalities are now more prevalent due to this pandemic, and that context could be important for future comparisons or interpretations).

2.1. Test stimuli

Juices and vegetables tested in this study are shown in Table 1. All products were purchased at local retailers. The broccoli and kale were chopped the day prior to the sensory test in order to damage the cellular tissues and allow for myrosinase activation, increasing the baseline isothiocyanate and nitrile content of these products (Wieczorek et al., 2018). While the emphasis of this study was on vegetables, we included the cranberry juice to allow some comparison to prior work in polyphenol rich beverages.

Table 1:

Stimuli

Name in article	Full brand name	Relevant ingredients*	Total sugar (g/100g)	Manufacturer, Location
Acai Superblend	Acai 10 Superblend Juice	Apple juice, acai puree, black currant juice, lemon juice, chicory root fiber, raspberry juice, merlot grape juice, bilberry juice, hibiscus juice, blueberry juice, pomegranate juice, yumberry juice, mangosteen puree, noni powder, seabuckthorn powder	10.7	Bolthouse® Farms, Bakersfield CA
Evolution Green Devotion	Organic Green Devotion	Celery juice, cucumber juice, spinach juice, romaine lettuce juice, kale juice, lemon juice, parsley juice	1.7	Evolution Fresh®, Monrovia CA
V8 Healthy Greens	V8® Healthy Greens	Sweet potato juice, yellow carrot juice, spinach juice, cucumber juice, celery juice, kale juice, romaine lettuce juice, green pepper juice, apple juice, pineapple juice, dehydrated spinach, spinach puree, huito juice, watermelon juice	5.4	Campbell Soup Company, Camden NJ
Fresh Thyme Green Blend	Green Thyme Blend	Apple juice, cucumber juice, spinach juice, kale juice, parsley juice, lemon juice (prepared in store, purchased locally)	5.4	Fresh Thyme Market™, Downer’s Grove IL
Cranberry juice	Pure Cranberry	Cranberry juice (diluted in our lab to 25% w/w with water)	0.9 (in diluted test sample)	Ocean Spray®, Middleborough MA
Chopped brussels sprouts	NA	Brussels sprouts (raw, roughly chopped and refrigerated overnight)	2.2**	Local grocer
Chopped kale	NA	Kale (raw, roughly chopped and refrigerated overnight)	1.0**	Local grocer

^*Includes all fruit/vegetable ingredients. Full ingredient list in supplemental file

^**From the USDA National Nutrient Database for Standard Reference; all other sugar contents from nutrition panel on product label

2.2. Participants

Participants were recruited through the Saliva, Perception, Ingestion, and Tongues (SPIT) Lab participant database. Participants were included if they were between the ages of 18 and 65 years, and excluded if they: had food allergies, had known problems with their sense of smell or taste, or had nose/cheek/lip piercings. Individuals who smoked or vaped were required to avoid these activities for 1 hour prior to the experiment. Thirty-four participants (15 men, 19 women, 0 other) completed all study tasks, but only twenty-six (13 men, 13 women, 0 other) had analyzable saliva samples due to errors in saliva collection (n=6) or extremely high absorbance values (n=2). Average participant age for the final 26 included participants was 37 years (range: 20 – 63 years). All participants were compensated for their participation.

2.3. Testing

RedJade^® (Redwood City, CA) sensory software was used to guide participants through saliva collection and sensory evaluation of stimuli. All testing was completed on Purdue University’s campus. Participants were seated at testing stations separated by dividers, under normal lighting. Each participant completed one visit between the hours of 9AM and 3PM. Participants began by answering demographic information about participant gender identity, biological sex, racial/ethnic background, and year of birth. We report the data by biological sex because we gave response options of male/female/other for both sex and gender, which is incorrect (male/female are terms for sex, not gender, which should use terms such as “man” or “woman” as well as other options (Clayton & Tannenbaum, 2016)). Participants were also asked if they were green or black tea drinkers, and if so, whether they drank it sweetened with sugar, sweetened with low-calorie sweeteners, or unsweetened. Participants then completed a warmup to familiarize them with the scale used for the test, which included questions on the brightness of the sun & room, loudness of a shout and whisper, bitterness of black coffee, and sweetness of pure sugar (Hayes et al., 2013; Kershaw & Running, 2019).

Next, participants were prompted to rinse their mouths with spring water (Culligan Spring water, purchased locally in 19 L water cooler jugs), and saliva was collected for analysis. Participants were instructed to chew a 2.5x2.5 cm piece of wax (Parafilm^™ M Wrapping Film, Bemis Company, Inc., Neenah, WI) for 30 seconds, remove the wax, and spit all generated saliva into a pre-weighed 50mL centrifuge tube on ice. Immediately afterwards, collected saliva was weighed. Stimulated salivary flow rate was calculated by multiplying this value by two, yielding a salivary flow rate in grams per minute. 1.5mL aliquots of collected saliva were portioned into microcentrifuge tubes and flash frozen to −80C using a CoolRack M90 (Corning^©, Corning, NY). Saliva samples were stored at −80C until analysis. As a note: while collecting liquid-stimulated saliva would be the most relevant to juice consumption, prior work conducted in our research group (Crawford & Running, 2020) using a water-rinsed salivary sample did not provide enough salivary protein to evaluate, and thus would have been unlikely to work in our tannic acid assay.

After saliva collection, participants tasted and rated the stimuli (shown in Table 1). Stimuli were blinded with unique 3-digit codes and presentation order was randomized. For each sample, participants were asked to taste all the stimulus, chew or swish the stimulus around in the mouth for 10 seconds using a provided timer, and swallow. Immediately after swallowing, they were asked to rate bitterness, sourness, and sweetness intensity of the stimulus using a generalized visual analog scale. Scale labels corresponded to the following points on a 110-pt scale: ‘None’ = 0, ‘Barely detectable’ = 5, ‘Weak’ = 25, ‘Moderate’ = 45, ‘Strong’ = 65, ‘Very strong’ = 85, and ‘Strongest ever’ = 105 (adapted from [Kershaw and Running, 2019], figure in supplemental file). Full text of the instructions for saliva collection and sensory evaluation are in the supplemental file.

2.4. Saliva analysis

Saliva samples were first thawed to 4°C and centrifuged at 20,000 g at 4°C for 20 minutes (Sorvall Legend Micro 21R, Thermo Fisher Scientific, Waltham, MA). After centrifugation, the supernatant (either 1mL or 1.5mL, depending on available volume) was mixed with an equal volume of 0.1% w/w tannic acid solution (Sigma-Aldrich, St. Louis, MO) in a spectrophotometer cuvette. These solutions were left at room temperature for 15 minutes, then light absorbance was read in 50nm wavelength increments between 450–700nm using an Epoch 2 Microplate Spectrophotometer (BioTek Instruments, Winooski, VT). Distribution of absorbance values was broadest at 450nm, and the general order of subjects was maintained across the other wavelengths; thus, measurements at 450nm were used for analysis.

2.5. Statistical analysis

Sensory and salivary data were analyzed using linear mixed models in SAS On Demand, within the Jupyter Lab environment. Our main outcomes of interest were the relationships between salivary characteristics (absorbance of saliva/tannic acid mixtures and salivary flow rate) and taste intensity ratings of test stimuli. There were two outliers for absorbance (greater than 1.5 times the interquartile range), and 6 saliva samples for which absorbance was unable to be measured (participants erred in the saliva collection procedure yielding unusable samples). Thus, these data were removed, leaving 26 participants (13 men, 13 women, 0 other) with analyzable saliva samples.

To understand the influence of potentially relevant participant characteristics on our main analytical variables, we analyzed whether the absorbance of 450nm light differed by participants’ reported sex, tea drinking habits (including interaction with sweetener use), salivary flow rate, and age. However, the only significant effect was sex, so other factors were removed from the model. The model used was:

Absorbance450 = Sex

Where Absorbance450 was the absorbance value at 450nm from the saliva-tannic acid mixture for each participant.

For the main analysis of the absorbance and salivary flow rate related to the sensory ratings, two models were run, estimating 1) overall effects across all stimuli, and 2) effects for individual stimuli. All models were run separately for taste quality (bitterness, sourness, and sweetness intensity). These models also yielded descriptive information on whether the stimuli were significantly different from each other for taste intensity.

The model statement for overall sample effects was:

$$\begin{array}{l}\text{Intensity}=\text{Absorbance450 LogSalivaFlow Stimulus}\\ \text{Stimulus*Absorbance450 Stimulus*LogSalivaFlow}\end{array}$$ Model 1:

Intensity refers to the bitterness, sourness, or sweetness intensity rating collected on our generalized visual analog scale. Absorbance450 was the UV-Vis absorbance value at 450nm from the saliva-tannic acid mixture for each participant. LogSalivaFlow was the salivary flow rate (in g/min) collected from the beginning of the test, log₁₀-transformed to improve distribution. Stimulus was the specific stimulus rated as shown in Table 1.

The model statement for models run separately by stimuli was:

$$\text{Intensity}=\text{Absorbance450 LogSalivaFlow}$$ Model 2:

Variable names are described above.

In all models, the Kenward Roger approximation was used for degrees of freedom, with participant included as a repeated measure and compound symmetry used as the covariance structure. We checked sex as a factor in the models for sensory intensity; however, there were no consistent effects, and our first analysis indicated this factor and absorbance were not independent. Thus, we did not keep sex as a factor in the sensory models. Distributions of residuals were checked in all models and indicated adequate compliance with model assumptions for independence and identical distribution. Post-hoc comparisons were adjusted using the Tukey-Kramer method. OriginPro 2020 (Northhampton, MA) was used to generate scatterplots and boxplots. Statistical code is available in the supplemental files.

3. Results

3.1. Absorbance

Saliva/tannic acid mixtures significantly differed in absorbance of 450nm light by sex (Figure 1), with female participants exhibiting greater absorbance (more haze/turbidity) compared to male participants (p=0.048).

Figure 1:

Saliva samples mixed 1:1 with 0.1% w/w tannic acid developed haze/turbidity, measured by absorbance at 450nm. Saliva from female participants showed greater haze development than from male participants (p=0.048). Boxes are the 25–75%, whiskers are 1.5 the interquartile range, and horizontal lines are medians.

3.2. Sensory ratings

Taste ratings (bitterness, sweetness, and sourness) evaluated for each stimulus are visualized in Figure 2. These data are provided mostly for descriptive purposes, but the significant differences noted are from Model 1 (main effect of stimulus on intensity).

Figure 2:

Boxplots for sensory ratings of taste for all samples. Boxes are the 25–75%, whiskers are 1.5 the interquartile range, and horizontal lines are medians. Samples with different letters were significantly different in taste intensity (p<0.05, adjusted for multiple comparisons using Tukey-Kramer).

3.3. Absorbance and sensory ratings

From Model 1, which included all stimuli, there was a main effect of absorbance on sweetness (p=0.013), and a potential trend (p=0.091) for an interaction effect of absorbance and stimulus type on sourness. To investigate these effects further, we used Model 2, which was run by stimulus type. Scatterplots for all sensory ratings for all stimuli by absorbance are shown in Figure 3.

Figure 3:

Scatterplots of absorbance by sensory ratings for each stimulus and taste quality. Note these are the original, raw sweetness intensity ratings. Where significant associations were observed from Model 2 analyses, they are marked with an asterisk and the p-value. In the statistical model (Model 2), this factor of absorbance is corrected for salivary flow, the other factor in the model.

3.3.1. Bitterness

V8 Healthy Greens showed an inverse relationship between absorbance and bitterness (p=0.039). Thus, saliva/tannic acid mixtures that were more turbid/hazy corresponded to lower bitterness ratings for this beverage. No other stimuli showed any effects.

3.3.2. Sweetness

Negative associations or trends were observed between absorbance of the saliva/tannic acid mixtures and sweetness for the Fresh Thyme Green juice blend (p=0.0077), Chopped Kale (p=0.0024), and V8 Healthy Greens (p=0.087). These are all consistent with the main effect showing less sweetness with greater absorbance of saliva/tannic acid mixture.

3.3.3. Sourness

Sourness ratings for Cranberry Juice were lower for individuals whose saliva/tannic acid mixture absorbed more light at 450nm (p=0.030). Additionally, V8 Healthy Greens showed a potential trend of an inverse relationship between absorbance and sourness (p=0.094). The trend of an interaction effect of sourness and stimuli for the overall Model 1 was likely driven by the difference in slope/effect size for Cranberry Juice compared to all other samples.

3.4. Salivary flow rate and sensory ratings

From Model 1, which included all stimuli, there were no main effects of salivary flow rate on sensation. However, there was an overall interaction effect of salivary flow rate by stimuli for bitterness (p=0.040). To investigate these effects further, we used Model 2, which was run by stimulus type. Scatterplots for all sensory ratings for all stimuli by salivary flow rate are shown in Figure 4.

Figure 4:

Scatterplots of salivary flow rate by sensory ratings for each stimulus and taste quality. Note these are the original, raw sweetness intensity ratings. Where significant associations were observed from Model 2 analyses, they are marked with an asterisk and the p-value. In the statistical model (Model 2), this factor of salivary flow rate (log10) is corrected for absorbance of the saliva/tannic acid mixtures, the other factor in the model.

3.4.1. Bitterness

Only Chopped Brussels Sprouts showed an inverse relationship between salivary flow rate and bitterness (p=0.039). No other stimuli showed any effects. The overall interaction effect was likely driven by this sample showing a different pattern from all the others.

3.4.2. Sweetness

Negative associations or trends were observed between salivary flow rate and sweetness for the Fresh Thyme Green juice blend (p=0.033) and Chopped Kale (p=0.052).

3.4.3. Sourness

Sourness ratings showed no associations with salivary flow rate.

4. Discussion

In this study, we looked at relationships between salivary flow rate and absorbance of 450nm light in saliva/tannic acid mixtures (referred to as “absorbance” from here on, for brevity) with sensory ratings of various commercially available vegetable stimuli, both juices and solids. In summary, our findings indicated:

• Saliva/tannic acid mixtures from women had higher absorbance than from men.

• Higher absorbance was associated with:

  • Less bitterness for only one stimulus, a green vegetable juice blend.
  • Less sweetness overall, particularly driven by 2 of the 3 green vegetable juice blends and by chopped kale.
  • Less sourness from cranberry juice, and potentially from one of the green vegetable juice blends.
• Greater salivary flow rate was associated with:

  • Less bitterness from chopped brussels sprouts
  • Less sweetness for one green vegetable juice blend and for chopped kale

These findings are not well aligned with our hypotheses, as we expected to observe effects for bitterness or sourness (for which we found few effects). Indeed, if anything we found more effects for sweetness, which we did not hypothesize would have a relationship with salivary properties. The fact that the associations of salivary flow and absorbance with sweetness were inverse was also surprising. If anything, we would have expected sweetness to be augmented by saliva binding bitter, astringent, or sour/puckering molecules, reducing mixture suppression for the perception of sweetness. Additionally, the fact that the patterns for any of our sensations are not consistent or compelling across all stimuli is intriguing and warrants further consideration of the different stimuli ingredients.

4.1. Absorbance and sensory ratings

Green vegetable juice blends, chopped kale, and cranberry juice showed some patterns of reduced sensation with increasing absorbance in the saliva/tannic acid mixtures. Bitterness and sweetness were lower with greater salivary absorbance for green vegetable products only, while sourness was lower in cranberry juice as well as one of the green vegetable juice blends.

We were surprised at the very few relationships between bitterness intensity and absorbance of the saliva/tannic acid mixtures. Indeed, only the V8^® Healthy Greens juice showed any relationship between absorbance and bitterness intensity. While this juice contained several ingredients unique from the other juices tested (green pepper, pineapple, sweet potato, and as well as watermelon and huito juices for color), none of these were expected to distinctively influence bitterness intensity. Thus, our hypothesis that less bitterness would correlate with higher absorbance, similar to patterns seen in prior work for astringency and to data collected with rodents relating bitter taste to salivary concentrations of binding proteins for bitterants, is not generally supported by our data. In other work investigating bitterness and salivary composition, similar lack of or mixed effects have been observed. In rats, higher concentrations of proline rich proteins (known to bind bitter polyphenols and alkaloids, which can be observed through higher salivary absorbance of light) have been associated with greater acceptance of bitter diets, lower bitter taste detection thresholds, and lower chorda tympani nerve response (Martin et al., 2018, 2020; Martin, Kay, et al., 2019; Martin, Nikonova, et al., 2019; Torregrossa et al., 2014). Yet, in humans, it has been more difficult to directly link bitterness suppression with salivary protein profiles. From work using dietary interventions designed to change salivary protein expression, mixed effects on corresponding sensory changes have been observed. Some work noted changes in human salivary proline rich proteins and cystatins related to increased polyphenol-rich chocolate milk consumption, but no associated changes in bitterness intensity (Crawford & Running, 2020). Later work observed changes in bitterness intensity after exposure to a bitter and astringent polyphenol, but these changes were not linked to salivary protein changes, as the salivary proteins changed as much or more from the control intervention compared to the polyphenol intervention (Davis & Running, 2021). Others have also noted acute changes in salivary proline rich proteins and cystatins after tannic acid or cranberry juice stimulation (Melis et al., 2017) and that exposure to cranberry polyphenols induces salivary protein changes but not sensory changes (Yousaf et al., 2022). Overall, altering expression of human salivary proteins that can theoretically bind bitter molecules does not appear to consistently alter bitterness intensity in humans. Given the consistent data in rats but inconsistent data in humans, a relationship between bitterant-binding proteins in saliva and bitter intensity may still exist through some intermediate factor that has yet to be discovered. Additionally, humans may be confounding bitterness with astringency or other aversive sensations. Ideally, participants in studies on bitterness and astringency should be trained in order to better ascribe the sensation with the technically accurate attribute. However, this is challenging given the data indicating diet, or even acute taste stimulation, may influence salivary protein composition (Bader et al., 2018; Crawford & Running, 2020; Lorenz et al., 2011; Louro et al., 2021; Neyraud et al., 2006; Simões et al., 2021; Yousaf et al., 2020).

Regarding patterns for sweetness intensity, there are several ingredient similarities between the V8^® Healthy Greens and Fresh Thyme Green blends, which both displayed negative associations between sweetness and absorbance. Notably, both these juice mixtures contained apple juice (presumably for sweetness and palatability), cucumber, kale, and sweet potato. Considering that these two juices and the chopped kale all displayed a negative relationship between sweetness and absorbance, we hypothesize that the kale ingredient may be driving this pattern. On average, kale contains considerably more polyphenols than many other green vegetables (Olsen et al., 2009; Satheesh & Workneh Fanta, 2020; Schmidt et al., 2010). Thus, the polyphenols in kale could be interacting with the same salivary proteins responsible for increased absorbance in saliva/tannic acid mixtures, although we did not quantify polyphenol content of the stimuli in this work. Interestingly, the Evolution Green Devotion vegetable juice blend, which also contained kale, did not display the same pattern of less sweetness with more absorbance. This particular juice blend had the least amount of sugar (only 1.7g/100g), yet the sweetness intensity ratings indicated this juice was perceived no differently than the Fresh Thyme blend, which has sugar content more similar to the V8^® blend. The V8^® blend actually was rated higher for sweetness than any other juice except the Acai Superblend, which had considerably higher sugar content.

Consequently, sugar alone is not fully responsible for the sweetness intensity ratings, and there is likely some difference between the V8^® and Fresh Thyme Green blends compared to the Evolution Green Devotion blend to result in less sweetness associating with greater absorbance for the former two and not the latter. The only ingredient shared between the V8^® and the Fresh Thyme products but not the Evolution product is apple juice. Otherwise, most of the green vegetable, fruit, or herb ingredients are shared among these three products. All three contain cucumber, kale, and spinach juices, although we do not have access to information about specific ingredient quantities in these commercial products. Notably, apples do contain many types of polyphenols, yet commercial products tend to contain less than freshly prepared apple juice samples (Kahle et al., 2005). Additionally, how interactions of apple-specific polyphenols with saliva could be influencing sweetness is unclear. Overall, it’s possible some component of the apple juice, or interaction of apple and kale juice, could be responsible for the observable pattern of more salivary absorbance associated with less sweetness in the V8^® and Fresh Thyme Green blends. Future work could look at these interactions in products where ingredient compositions are known to provide more specific mechanistic insights.

Notably, both the Acai Superblend juice and Cranberry juice should contain the highest concentrations of polyphenols, even more than kale-containing products. Thus, the general class of “polyphenols” is not enough to induce the relationship of less sweetness with greater absorbance. This is not too surprising, as polyphenols have very diverse structures. However, we expected to see patterns for cranberry juice in particular, and the Acai Superblend also contains many polyphenol-rich berry juices. These berry sources should contain higher concentrations of flavanols, which have documented interactions with salivary proteins (Ferruzzi et al., 2012). These interactions lead to the formation of aggregates, which cause the increased absorbance/haze development similar to the saliva/tannic acid assay used in our study. Given the prior work showing increased absorbance associates with lower astringency perception (Horne et al., 2002), we expected this pattern to occur for bitterness in cranberry juice and likely the acai blend. This did not occur. Future experiments could investigate the influence of different polyphenol structures on the salivary haze development assay, as well as whether polyphenol structures that are more astringent (typically the oligomers) rather than bitter (typically the monomers; (Robichaud & Noble, 1990)) induce different relationships between patterns of salivary haze development and sensory properties.

Nonetheless, we did observe reduced sourness from cranberry juice with increasing absorbance, consistent with our hypothesis and prior work on astringency. Potentially, this could be a reflection of sourness overlapping with the puckering subquality of astringency (Lawless & Corrigan, 1994). Additionally, sourness and puckering sensations can be driven by the same groups of molecules. Indeed, acids present in juices cause both astringency and sourness, but sourness differs by molecular structure at the same pH while astringent sensations related to acidic compounds appear to be mostly pH dependent (Lawless et al., 1996). Most work investigating astringency corresponding to salivary characteristics has focused on polyphenols, rather than acids (Dinnella et al., 2009, 2010; Fischer et al., 1994; Fleming et al., 2016; Horne et al., 2002; Kallithraka et al., 1998; Melis et al., 2017; Yousaf et al., 2022).

The lack of an effect for the Acai Superblend, which should also be high in flavanols due to the numerous berry juices, is not too surprising. This product has a very extensive ingredient list and contains almost double the sugar content of any other juice. In general, these additional ingredients and higher concentration of sugar could mask any possible relationships between sensation, especially sourness or bitterness, and salivary absorbance.

4.2. Salivary flow and sensory ratings

Salivary flow negatively associated with bitterness in chopped brussels sprouts, as well as sweetness in chopped kale and in one green vegetable juice blend.

Bitter compounds found in Brassica vegetables such as brussels sprouts and kale include isothiocyanates (hydrolysis products of glucosinolates) as well as several classes of polyphenols (Baenas et al., 2019; Gonzales et al., 2015; Wieczorek et al., 2018). Notably, we expected to see relationships for bitterness and salivary flow more pronounced in kale compared to brussels sprouts (contrary to our findings), as kale usually contains more polyphenols than brussels sprouts, especially when raw (Sikora et al., 2008). Other work has noted relationships with increased flow corresponding to reduced astringency and/or bitterness from polyphenols (Dinnella et al., 2009; Fischer et al., 1994), and we expected to see this relationship for bitterness in our stimuli with traditionally higher polyphenol contents. However, the only sample in which we observed this relationship was the chopped brussels sprouts. This is intriguing and suggests more research with bitter phytochemicals, sensation, and salivary flow may be warranted. Currently, more data are available regarding how both sensation and salivary properties relate to polyphenols than to other compounds such as isothiocyanates. For polyphenols, total salivary protein concentration or flux as well as concentration of specific salivary proteins have been related to both astringency and bitterness (Crawford & Running, 2020; Davis & Running, 2021; Dinnella et al., 2009, 2010; Martin et al., 2018, 2018; Martin, Nikonova, et al., 2019; Torregrossa et al., 2014).

For other compounds such as isothiocyanates, data are available concerning just salivary properties related to these compounds, but little is available that considers both salivary properties and sensation. Nonetheless, research does indicate at least some relationships between isothiocyanates and salivary properties, which could influence sensations. For example, isothiocyanates have higher partition coefficients in saliva than in other solvents or their natural state (Marcinkowska & Jeleń, 2020). This would imply that in saliva, retronasal odor sensation from isothiocyanates would be lower compared to the food prior to entering the mouth. Thus with increased salivary flow we would expect lower sensation due to both dilution effects as well as greater partitioning into the saliva rather than the air. Our study did not ask specifically about odor, but unpleasant odors are often confounded with “bitterness,” especially within a vegetable context (Lim & Padmanabhan, 2013).

Salivary flow rate influences protein composition and concentration in saliva. Increased flow rates from stimulation are largely due to increased flow from the parotid gland, which is more watery, and thus stimulated saliva contains lower total protein concentration than resting saliva (Humphrey & Williamson, 2001). However, increased flow would replenish proteins in saliva at a greater rate, so the effective concentration of a protein could be better maintained with increased flow. This has been a phenomena of interest in other work on saliva’s interaction with bitter or astringent polyphenols, where multiple researchers have noted that better repletion of proteins in saliva is more protective against unpalatable sensations of polyphenolic stimuli (Dinnella et al., 2009, 2010; Fleming et al., 2016; Rinaldi et al., 2012). Thus, adding analysis of total salivary protein or flux of salivary protein in relationship to bitterness and sweetness could increase the understanding of the observations we gathered in this study.

Additionally, other research areas have noted that specific salivary proteins, many of which would be predominantly sourced from stimulated parotid saliva, may correlate with bitterness intensity (Davis & Running, 2021; Dsamou et al., 2012; Martin et al., 2018, 2020, 2020; Martin, Kay, et al., 2019; Martin, Nikonova, et al., 2019; Morzel et al., 2014; Torregrossa et al., 2014). In particular, proline rich proteins and salivary cystatins have been identified as potentially correlating with bitter sensations. Thus, increased flux of protein, with more parotid-sourced proteins secreted during stimulation, could increase available salivary protein to interact with bitter compounds. Nonetheless, why in our current study we only observe a relationship between bitterness and salivary flow for the chopped brussels sprouts is unknown. Perhaps the need to chew this sample resulted in more stimulated saliva with proteins that could suppress bitterness, or perhaps the aroma, rather than bitterness, of the vegetable is actually what was modified.

The negative association between salivary flow and sweetness intensity (observed for kale and Fresh Thyme Green blend juice) was unexpected, but not necessarily inconsistent with prior research from other groups. For example, individuals with lower salivary flow rates have been observed to take longer to experience maximum sweetness intensity (Bonnans & Noble, 1995). Similarly, addition of more acid, stimulating greater salivary flow, reduced total duration of perceived sweetness in a chewing gum (Guinard et al., 1997). These prior reports indicate that sweetness may be less intense when salivary flow rate is higher, and that it may appear and end more rapidly. This may explain our observation of less sweetness for kale and one of the green juices with greater salivary flow. Notably, while the effect was significant for kale, the intensity of sweetness was very low overall for this sample. The Fresh Thyme Green Blend also showed relatively low sweetness ratings, with both this juice and the kale in the lowest Tukey-Kramer adjusted statistical group for sweetness.

4.3. Other considerations

As mentioned briefly above for isothiocyanates in brussels sprouts, retronasal olfaction could have been influenced by salivary properties. Taste and aroma are often confounded in sensory ratings, and consequently differences in partitioning of various compounds into saliva could influence taste ratings indirectly by influencing retronasal aroma intensity. As noted, isothiocyanates more strongly partition into saliva, reducing potential aroma intensity (Marcinkowska & Jeleń, 2020). Other research has shown large individual variability in how saliva induces the release of odiferous, sulfur-containing volatiles from vegetables such as cabbage (Frank et al., 2018). As retronasal olfaction is a crucial component of how individuals report experiencing bitterness and liking for vegetables such as brussels sprouts (Duffy et al., 2020; Lim & Padmanabhan, 2013), further work delineating the taste, aroma, and salivary influences on these chemosensory compounds is warranted.

5. Limitations

While our current study uncovered some potential relationships between salivary properties and taste sensations, the number of effects is relatively small when considering the number of analyses conducted. The observed patterns do seem consistent with prior literature, but the lack of strong associations across multiple sample types likely indicates these effects are not well generalized to a broad variety of foods. Instead, we hypothesize that specific ingredients, or combinations of ingredients, may drive the relationships between lower taste intensities and higher absorbance in saliva/tannic acid mixtures (indicative of more binding proteins for tannic acid and similar structures) or higher salivary flow. These findings emphasize the utility of testing diverse food matrices and ingredient profiles to provide insight into practical application of principles. While using prototypical stimuli enables more mechanistic insight into scientific phenomena, using foods readily acquired and consumed by the general population gives valuable information about the relative impact of said phenomena from a holistic sensory perspective.

That being said, our approach has many limitations. Commercial juices, while nutritionally relevant, have proprietary formulations and processing methods that influence the composition and chemical structures of the chemosensory stimuli originating from the vegetables and fruits. Commercial juices would be processed for safety and to increase shelf life, and each slightly different processing method could influence the concentration of taste or aroma stimuli. For our whole vegetable samples, although we included raw brussels sprouts and kale samples in our study, we also chopped these samples the day prior to the experiment. This was intended to standardize the amount of isothiocyanates released from these vegetables, mitigating the factor of chewing efficiency on the release of these compounds. However, by chopping these products ahead of time, some of the released isothiocyanates or similar chemosensory stimuli may have degraded. This again highlights the challenges of working with fresh produce and these complex bioactive compounds. High levels of variability in bioactives may be observed between batches or lots, growing seasons, and processing methods, and that variability is then again augmented by human factors such as mastication, salivary flow and composition, and more. We did not quantify total polyphenol or isothiocyanate content of our study stimuli. We first wanted to examine the application of this assay in a variety of commercially available stimuli to determine if the findings indicated that future evaluation of the stimuli in this setting was warranted. Given how few effects were observed, it is worth considering whether these proposed effects are actually robust enough to overpower other influences (such as processing and inherent variability from the plants) on chemosensation and perception in humans.

We also did not quantify total protein content in our saliva samples. This metric is most helpful when examining protein changes as a result of stimulation or some intervention, as is collected in studies of protein restoration after polyphenol exposure (Dinnella et al., 2009, 2010; Fleming et al., 2016; Rinaldi et al., 2012). Data on total protein content, and the turnover of total protein content after stimuli exposure, might be valuable to explaining the myriad effects/non-effects we observed.

Lastly, we did not ask participants to rate astringency of our samples. There were several reasons for this. First, as described in Section 1, we were primarily interested in the bitterness ratings of our stimuli due to the findings of prior animal work in this area. Most of our chosen stimuli were not particularly astringent, but nevertheless astringency is a relevant sensation when evaluating polyphenol-containing samples. However, our participants were untrained, which can increase the likelihood of confusion between bitter and astringent sensations described in Section 4.1. We did not train participants, 1) to provide a more realistic evaluation of the taste perception of commercially available foods and beverages, and 2) because of the documented short-term effects of astringent phenolic stimuli on salivary protein composition (Yousaf et al., 2020). With this approach we may have missed some relationships between our samples, salivary characteristics, and astringency, but as these relationships have been demonstrated in many other prototypical and practical foods and beverages (Dinnella et al., 2009, 2010, 2011; Fleming et al., 2016, 2016; Horne et al., 2002; Melis et al., 2017; Monteleone et al., 2004; Yousaf et al., 2020), we prioritized other attributes of interest.

6. Conclusions

In this work, we observed that some, but not all, vegetable and fruit products had lower sensory ratings correlated with greater absorbance of 450nm light in saliva/tannic acid mixtures (indicative of more polyphenol-binding proteins in the saliva) as well as with greater salivary flow rate. Greater absorbance corresponded to less bitterness for one green vegetable juice blend, less sweetness for two green vegetable juices and chopped kale, and to less sourness from cranberry juice. Greater salivary flow corresponded to less bitterness from chopped brussels sprouts, and to less sweetness from one green vegetable juice blend and chopped kale. The inconsistency in findings indicates that greater absorbance/presence of certain binding proteins in saliva is not universally associated with less bitterness from products containing polyphenols or isothiocyanates, nor is greater salivary flow universally associated with lesser taste intensity. We propose that specific ingredients may drive the relationships, and that these ingredients could vary in amount of active chemosensory stimuli due to processing, co-ingredients, and other formulation differences across both commercial as well as fresh samples.

Table 2:

Significant or trending relationships between sensory ratings and absorbance measurements or salivary flow rate

Model 1 (run by taste quality): Intensity = Absorbance Log10SalivaFlow Stimulus (Stimulus * Absorbance)    (Stimulus * Log10SalivaFlow)
Taste Quality	Effect	Overall p-value	NumDF/ DenDF2, F
Bitterness	Stimulus*Log10SalivaFlow	0.040	6/138, 2.27
Sweetness	Stimulus	0.0010	6/138, 3.98
	Absorbance	0.013	1/23, 7.33
Sourness	Stimulus	0.000033	6/138, 5.58
	Stimulus*Absorbance	0.091	6/138, 1.87
Model 2 (run by taste quality and stimulus): Intensity = Absorbance Log10SallivaFlow
Taste Quality	Stimulus	Effect	P-value	NumDF/ DenDF2, F	Estimate ± SE
Bitterness	V8 Healthy Greens	Absorbance	0.039	1/23, 4.80	−240.0 ± 109.8
	Chopped brussels sprouts	Salivary flow rate (log10)	0.0066	1/23, 8.93	−60.9 ± 20.3
Sweetness	Fresh Thyme Green Blend	Absorbance	0.0077	1/23, 8.51	−219.2 ± 75.1
		Salivary flow rate (log10)	0.033	1/23, 5.14	−25.6 ± 11.3
	V8 Healthy Greens	Absorbance	0.087	1/23, 3.19	−153.1 ± 85.8
	Chopped kale	Absorbance	0.0024	1/23, 11.63	−65.1 ± 19.1
		Salivary flow rate (log10)	0.052	1/23, 4.2	−5.9 ± 2.9
Sourness	Cranberry juice	Absorbance	0.030	1/23, 5.3	−268.1 ± 115.9
	V8 Healthy Greens	Absorbance	0.094	1/23, 3.0	−174 ± 100.0

Absorbance: measured at 450nm for 1:1 mixtures of participant saliva with 0.1% tannic acid mixture

Salivary flow rate: measured as collection of all saliva while chewing on wax for 30s

Num DF: Numerator degree of freedom; Den DF: Denominator degree of freedom, SE: Standard Error

Highlights:

• Salivary flow & turbidity of saliva/tannic acid mixtures related to taste intensity

• Greater turbidity sometimes corresponded to less bitterness, sweetness, or sourness

• Greater salivary flow sometimes corresponded to less bitterness or sweetness

• Inconsistent patterns indicate specific ingredients may drive these relationships