Sodium, but not potassium, blocks bitterness in simple model chicken broths

Abstract

Potassium chloride (KCl) has proven useful as a salty taste replacer to help reduce dietary sodium. But unlike sodium, which in simple aqueous solutions blocks the perception of bitterness of selected compounds, KCl does not blocker bitterness. We tested the ability of potassium to block bitterness in a more complex translational system by presenting model chicken broths to healthy adults. Broths were presented in three added salt conditions: (1) no added salt, (2) salted with sodium chloride (NaCl), or (3) salted with KCl. To create a model bitter off-taste, four concentrations of l-tryptophan (l-tryp, present in chicken meat) were added to each broth. In Experiment 1, the base broth consisted of chicken flavor only. In Experiment 2, the base broth was more complex, containing savory (umami) ingredients. In both experiments, subjects rated broths with either added NaCl or KCl as saltier than unsalted broths. Only NaCl, however, suppressed bitterness (by about 30%, across a wide range of l-tryp concentrations). Accordingly, when complex foods have sodium reduced and potassium increased to balance salty taste, the bitterness reducing properties of sodium will need to be replaced independently, since potassium does not share this effect.

Introduction

Over-consumption of sodium contributes to hypertension, a key risk factor for heart disease and stroke (Carey et al. 2018; IOM 2010). According to some estimates, lowering sodium intake in the USA from current levels (~ 3500 mg/day) to 2300 mg/per day (1500 mg/day for those at high risk for hypertension) could prevent up to 100,000 deaths and save billions of dollars in medical costs per year (IOM 2010). Unfortunately, substantially lowering sodium in many foods makes them less palatable (Hoppu et al. 2017). People might acclimate to less sodium in the diet over time (IOM 2010), but changing the diet as a whole would require broad cooperation by government, industry, and consumers.

Most efforts have focused on re-formulating low sodium products to improve palatability (Abimbola and Fouladkhan 2018; Jaenke et al. 2017). Potassium chloride (KCl) is one of the most practical substitutes to replace lost salty taste but has undesirable side tastes which limit the amount of KCl substitution that consumers tolerate (Greiff et al. 2015; Jaenke et al. 2017; Liem et al. 2011; Silva et al., 2018). Regardless of how salty taste is replaced, NaCl also improves overall flavor balance, enhancing desirable mouthfeel, enhancing sweetness, and suppressing off tastes in various foods (Gillette, 1985). Importantly, for the model foods studied, KCl failed to show the same broad enhancement of flavor profiles (Gillette 1985).Thus, reformulation of low sodium products must account for more than just lost salty taste.

Here we focus on differential effects of NaCl and KCl on bitterness. In simple aqueous solutions, sodium reduces bitterness from some compounds (Breslin and Beauchamp 1995, 1997; Green 2003; Keast et al. 2004; Wilkie and Phillips 2014). Two key observations suggested that bitterness reduction depends on Na⁺ rather than salty taste. First, bitterness reduction is independent of the perceived intensity of salty taste and depends only on sodium concentration. Second, salty-tasting KCl does not reduce bitterness (Breslin and Beauchamp 1995; Wilkie and Phillips 2014). Simple aqueous solutions often have limited utility as models of actual foods. Gillette reported that sodium reduced bitterness slightly in various model foods, whereas potassium slightly increased bitterness, also consistent with selective bitterness suppression by salty compounds (Gillette 1985). However, the model foods employed were not strongly bitter. Various subsequent studies have shown that sodium can reduce bitterness in more strongly bitter foods (Sharafi et al. 2013; Wilkie et al. 2014), though many of these studies did not include KCl for comparison.

To further establish differential effects of NaCl and KCl with respect to bitter blocking in food matrixes, we asked healthy adults to taste model chicken broths with no added salt, with added NaCl, or with added KCl. In Experiment 1, the broth was very simple, consisting only of chicken flavoring. In Experiment 2, the broth was slightly more complex, also containing yeast extract and added savory compounds. In all broths, a bitter “off taste” was created by adding various concentrations of l-tryptophan (l-tryp), a bitter α-amino acid present in turkey and chicken. Dependent measures included ratings of perceived saltiness and bitterness intensities.

Materials and methods

Participants

Fifteen healthy adults (9 women) between the ages of 18 and 43 (mean = 32, S.D. = 6.6) participated in Experiment 1. Thirteen healthy adults (8 women) between the ages of 18 and 43 (mean = 30.4, S.D. = 6.9) participated in Experiment 2, including 11 from Experiment 1. Most were employees of the Monell Chemical Senses Center or students from nearby universities. All had extensive experience in rating intensity of stimuli containing KCl and NaCl. Participants provided written informed consent using forms approved by an institutional review board (IRB) at the University of Pennsylvania before testing. Procedures were approved by the IRB (protocol 701-334) and conducted in accordance with the guidelines of the Declaration of Helsinki.

Stimulus materials

The base for all stimuli was chicken broth. The broth included a chicken flavor base (used in both Experiment 1 and Experiment 2) with no added salt and savory ingredients (a yeast extract plus savory peptides, used only in Experiment 2 to create a slightly more complex model food). Ingredients were dissolved in Millipore^®-filtered, deionized water, and resulting broths were heated to 64.4 °C on a hot plate. 10 ml aliquots were served in 30 ml plastic cups at ~ 51 °C. As prepared, the bitterness of the plain broth samples was below “barely detectable” on a general Labeled Magnitude Scale, or gLMS (Bartoshuk et al. 2004) in pilot tests. To create a salient model off-taste, we added l-tryp (Sigma-Aldrich, ≥ 98% purity) at 0.011, 0.020, 0.033, and 0.063 mM, spanning a range of rated bitterness from “very weak” to slightly above “moderate” on the gLMS according to preliminary work. Broths were presented with no added salt, with 128 mM NaCl added, and with 171 mM KCl added (KCl and NaCl were comparable in saltiness based on preliminary testing). Accordingly, for both the simpler broth (Experiment 1) and the more complex broth (Experiment 2), there were 12 stimulus conditions: Three added salt conditions X four concentrations of l-tryp.

Training

Subjects were trained on the gLMS following published procedures (Bartoshuk et al. 2004), though all had previous experience with the scale and demonstrated understanding by producing monotonic intensity vs. concentration functions for salty and bitter stimuli. Subjects were trained on the concept of “saltiness” using NaCl (150 mM), KCl (200 mM), cesium chloride (200 mM), and calcium chloride (200 mM) as exemplars, all moderately salty in preliminary work. Subjects were told that all tended to taste salty to most people, but may also have other tastes. Subjects were trained on the concept of “bitterness” using quinine hydrochloride (0.53 mM).

Tasting and rating procedures

Participants began each test session by rinsing the mouth four times with Millipore^®-filtered, deionized water. During each trial, participants (1) took the entire contents of a sample cup into the mouth, (2) held the sample for at least 3 s to allow taste sensation to develop, (3) made sensory ratings (see next paragraph), (4) expectorated the sample, and (5) rinsed twice with water to initiate an inter-trial rest interval of 45 s. For both Experiments 1 and 2, there were two test sessions to include a replication. In each session, participants sampled and evaluated all 12 stimuli twice in blocked, random order. Participants rated bitterness intensity and saltiness intensity on paper ballots (separate pages for each trial). The gLMS scales (Bartoshuk et al. 2004) were 100 mm long (“no sensation” at 0 to “strongest imaginable sensation” at 100).

Data analysis

Ratings of intensity were averaged across replicate conditions within each session (within-subjects) using the arithmetic mean, then log-transformed before inferential analysis (Green et al. 1996). Effects of independent variables were analyzed using repeated-measures analysis of variance (ANOVA) with a significance criterion of p < 0.05. Log rated saltiness and log rated bitterness were submitted to separate, 3-way ANOVAs: Session (first vs. second) X l-tryp concentration (four concentrations) X Salt (no added salt, NaCl, and KCl). Key main effects were further explored using Bonferroni-corrected post-hoc tests. Significant Interactions were explored using simpler ANOVAs. Analyses were conducted using Statistica software (Version 13, Dell Inc.).

Results

Experiment 1

Ratings of saltiness intensity

The main effect of Salt reached significance, F(2,28) = 36.55, p < 0.001. According to a post-hoc test, [NaCl = KCl] > No salt. Accordingly, averaged across l-tryp concentrations, rated saltiness was comparable between broths salted with KCl and NaCl, but both were rated as more salty than the unsalted broth. The main effect of Tryptophan concentration also reached significance, F(3,42) = 3.49, p < 0.05. Rated saltiness tended to decrease as concentration of l-tryp increased (Fig. 1a). The main effect of Session failed to reach statistical significance (p = 0.81). Most interactions failed to reach significance (0.23 < p < 0.99), with one exception outlined in the next paragraph. Correlations between ratings for corresponding stimuli in the two sessions ranged from 0.43 to 0.95 (mean = 0.67, S.D. = 0.16), indicating short-term reliability.

Fig. 1.

Sensory ratings of the broth samples with both added salts, but no added umami substances (partially flavored broth). Common X-axis: Concentration of l-tryp (mM). a Ratings of salty taste intensity (geometric mean, ± SEM). Labels on the right represent general Labeled Magnitude descriptors: BD = “Barely Detectable,” Wk = “Weak,” Md = “Moderate.” b Ratings of bitterness intensity (geometric mean, ± SEM). Lower case letters indicate statistically homogenous groups of points (according to a Bonferroni test)

The one interaction to reach significance was between Salt and Tryptophan concentration, F(6, 84) = 2.34, p = 0.039. Figure 1a suggests that saltiness may have decreased less sharply as tryptophan concentration increased for NaCl than for KCl or no added salt. Simpler ANOVAs were consistent with a sharper decline in saltiness with increasing tryptophan concentration: For an analysis excluding NaCl, there was no significant Salt X Tryptophan concentration interaction (p = 0.53), but there were significant Salt x Tryptophan concentration interactions for analyses which excluded KCl, F(3,42) = 3.47, p = 0.02, and no added salt, F(3,42) = 2.91, p < 0.05. In brief, the saltiness intensity of NaCl and KCl flavored broths were approximately matched, but may have diverged when paired with higher levels of l-tryp.

Ratings of bitterness intensity

The main effect of Tryptophan reached significance, F(3,42) = 26.38, p < 0.001. Rated bitterness increased with concentration of l-tryp, an expected dose–response relationship (Fig. 1b). The main effect of Salt reached significance, F(2,28) = 15.58, p < 0.001. According to a post-hoc test: [No salt = KCL] > NaCl. Accordingly, broth with NaCl added was rated as significantly less bitter (about 29% less on average) than broth without added salt or with KCl. The main effect of Session failed to reach statistical significance (p = 0.80), as did the interactions (0.31 < p < 0.67). Correlations between ratings for corresponding stimuli in the two sessions ranged from 0.40 to 0.96 (mean = 0.61, S.D. = 0.19).

Experiment 2

Ratings of saltiness intensity

The main effect of Salt reached significance, F(2,24) = 19.48, p < 0.001. According to a post-hoc test, [NaCl = KCl] > No salt. Across l-tryp concentrations, rated saltiness was not significantly different between the broths salted with KCl and NaCl, but both were rated as significantly more salty than the unsalted broth. The main effect of Tryptophan reached significance, F(3,36) = 5.38, p < 0.005. Rated saltiness decreased as concentration of l-tryp increased (Fig. 2a). The main effect of Session failed to reach statistical significance (p = 0.09). None of the interactions reached significance (0.09 < p < 0.61). Correlations between ratings for corresponding stimuli in the two sessions ranged from 0.39 to 0.97 (mean = 0.74, S.D. = 0.18).

Fig. 2.

Sensory ratings of the broth samples with both added salts and added umami substances (full flavor broth). Common X-axis: Concentration of l-tryp (mM). a Ratings of salty taste intensity (geometric mean, ± SEM). Labels on the right represent general Labeled Magnitude descriptors: BD = “Barely Detectable,” Wk = “Weak,” Md = “Moderate.” b Ratings of bitterness intensity (geometric mean, ± SEM). Lower case letters indicate statistically homogenous groups of points (according to a Bonferroni test)

Ratings of bitterness intensity

The main effect of Tryptophan reached significance, F(3,36) = 58.72, p < 0.001. Rated bitterness increased with concentration of l-tryp, an expected dose–response relationship (Fig. 2b). The main effect of Salt reached significance, F(2,24) = 15.30, p < 0.001. According to a post-hoc test: [No salt = KCL] > NaCl. Broth with added NaCl was rated as significantly less bitter (about 30% less on average) than broth without added salt or broth with KCl. The main effect of Session failed to reach statistical significance (p = 0.39), as did most of the interactions (0.39 < p < 0.58). An exception is discussed in the next paragraph. Correlations between ratings for corresponding stimuli in the two sessions ranged from 0.27 to 0.96 (mean = 0.76, S.D. = 0.19).

The one interaction that reached significance was Salt X Tryptophan, F(6,72) = 2.65, p < 0.05. Simpler ANOVAs (Session X Salt for each concentration of l-tryp) yielded significant main effects of Salt for 0.011 mM Tryptophan, F(2,24) = 6.45, p < 0.01, for 0.02 mM Tryptophan, F(2,24) = 8.02, p < 0.01, and for 0.033 mM l-tryp, F(2,24) = 5.71, p < 0.01. In all three cases, according to post hoc tests: [KCl = No salt] > NaCl, consistent with the results of the overall ANOVA. The main effect of Salt failed to reach significance for the highest concentration of l-tryp (p = 0.77).

Discussion

For broth with chicken flavor only, and for a more complex broth with added savory compounds, adding NaCl and KCl imparted a similar salty taste. However, the two salts had different impacts on bitterness. NaCl decreased rated bitterness over a wide range of concentrations of l-tryp, whereas KCl had little or no effect. These results in model foods are consistent with bitterness blocking in aqueous solutions (Breslin and Beauchamp 1995, 1997; Green 2003; Keast et al. 2004; Wilkie and Phillips 2014), and support the idea that replacing lost saltiness with KCl in low sodium foods will not replace all sensory benefits of sodium (Gillette 1985; Greiff et al. 2015; Jaenke et al. 2017; Liem et al. 2011). The concentrations of l-tryp used might not represent off tastes commonly found in broths, soups, and similar foods, but they do show that NaCl is an effective bitterness masker for l-tryp across a wide range of bitterness intensities.

We also observed a decrease in saltiness with increasing bitterness. It was less clear whether saltiness from NaCl and KCl were differentially affected by bitterness. In Experiment 1 (simpler broth), saltiness decreased more sharply with increasing l-tryp concentration for KCl than for NaCl. This observation suggests a possible release from suppression effect on perceived saltiness when sodium suppressed perceived bitterness (Breslin and Beauchamp 1997). However, in Experiment 2, the decrease in saltiness with increasing bitterness was not significantly different between KCl and NaCl. The apparent difference in concentration-dependence could reflect interactions with savory additives not present in Experiment 1. Regardless, suppression of saltiness by bitterness is also broadly consistent with results from simple, aqueous solutions, though the effect tends to be less robust and consistent than suppression of bitterness by sodium (Wilkie and Phillips 2014).

The experiments have limitations. First, we employed small, convenience samples trained in sensory analysis, with all the limitations inherent in such samples. In particular, power may have insufficient to find differences between NaCl and KCl broths in saltiness. Inspection of the figures suggests that NaCl might have tasted more intensely salty, especially at some l-typtophan concentrations, and greater saltiness could have contributed greater bitter masking. Conversely, bitter masking appeared to occur at some l-tryp concentrations at which NaCl and KCl were nominally equal in saltiness. Further, research in aqueous solutions suggests that bitter masking is not strongly dependent on the intensity of saltiness (Wilkie and Phillips 2014). Still, we cannot rule out some effect of perceived saltiness. Second, our model food is simple, and certainly does not include all possible flavor interactions. Third, even within our model system, we focused on suppression of bitterness. The effects of sodium on rated bitterness may be mediated by a flavor interactions we did not measure, e.g., with savory taste. Future studies can explore these issues.

Conclusion

For bitterness blocking by salts, results from simple aqueous solutions appear to translate to simple model foods: NaCl blocks bitterness but KCl does not. Thus, even with a more complex array of taste-taste and aroma-taste interactions, sodium has important flavor effects that potassium does not. Attempts to reduce sodium in foods by replacement with potassium should also consider the inclusion of bitterness blockers, since potassium does not replace this aspect of sodium’s flavor impact.