β-carotene improves fecal dysbiosis and intestinal dysfunctions in a mouse model of vitamin A deficiency

Abstract

Vitamin A deficiency (VAD) results in intestinal inflammation, increased redox stress and reactive oxygen species (ROS) levels, imbalanced inflammatory and immunomodulatory cytokines, compromised barrier function, and perturbations of the gut microbiome. To combat VAD dietary interventions with β-carotene, the most abundant precursor of vitamin A, are recommended. However, the impact of β-carotene on intestinal health during VAD has not been fully clarified, especially regarding the VAD-associated intestinal dysbiosis. Here we addressed this question by using Lrat^−/−Rbp^−/− (vitamin A deficient) mice deprived of dietary preformed vitamin A and supplemented with β-carotene as the sole source of the vitamin, alongside with WT (vitamin A sufficient) mice. We found that dietary β-carotene impacted intestinal vitamin A status, barrier integrity and inflammation in both WT and Lrat^−/−Rbp^−/− (vitamin A deficient) mice on the vitamin A-free diet. However, it did so to a greater extent under overt VAD. Dietary β-carotene also modified the taxonomic profile of the fecal microbiome, but only under VAD. Given the similarity of the VAD-associated intestinal phenotypes with those of several other disorders of the gut, collectively known as Inflammatory Bowel Disease (IBD) Syndrome, these findings are broadly relevant to the effort of developing diet-based intervention strategies to ameliorate intestinal pathological conditions.

1. INTRODUCTION

Several intestinal disorders, including Crohn’s disease and ulcerative colitis, fall under the umbrella of the inflammatory bowel diseases (IBD) syndrome [1]. They display a commonality of symptoms such as inflammation, increased redox stress and reactive oxygen species (ROS) levels, imbalance in inflammatory and immunomodulatory cytokines, compromised barrier function, and perturbations of the intestinal microbiome [1]. Micronutrient deficiencies (i.e., poor diet) are often linked to these pathological conditions [2], but whether they play a direct causal role in disease etiology and/or progression, or are secondary manifestations, is not fully understood. Nevertheless, improving the dietary intake of these limiting nutrients ameliorates the above-mentioned conditions, including intestinal dysbiosis [3]. Remarkably, many of these intestinal phenotypes, if not all, occur under vitamin A deficiency (VAD). It is long known that vitamin A modulates intestinal morphology, maintains epithelial barrier integrity, regulates gut immunity and curtails inflammation [4–8]. Hence, VAD is generally linked to pathological conditions of the gastrointestinal tract [6] and specific taxonomic and functional changes of the intestinal microbiome [9–13]. Therefore, improving the vitamin A status through dietary interventions bodes well for gut health both in humans and animal models [14–17].

β-carotene from fruits and vegetables is the most abundant source of vitamin A worldwide, and often the sole supply of the vitamin in developing countries where VAD is still more prevalent [18–20]. Upon ingestion and absorption by enterocytes, β-carotene is cleaved symmetrically by the enzyme β-carotene 15–15’ oxygenase (or β-carotene 15–15’ oxygenase 1, BCO1) to yield two molecules of retinaldehyde. The latter can be either oxidized to retinoic acid, the active form of vitamin A, or reduced to retinol which upon esterification with longchain fatty acids generates retinyl esters, the storage form of vitamin A [21–23]. β-carotene is also known for its antioxidant capacity [24, 25] as it quenches singlet oxygen mostly through electron transfer to free radicals and formation of β-carotene radical cations [26]. In general, though, the chemistry governing the interaction of carotenoids with reactive oxidizing species is not fully understood [27]. Consumption of foods rich in β-carotene has been linked to beneficial health effects in humans due to both its provitamin A and antioxidant activities [28], even though β-carotene has also shown pro-oxidant behavior [29]. Whereas supplementation and/or biofortification with β-carotene are common strategies to combat VAD, especially in developing countries [18–20], the impact of β-carotene on intestinal health during VAD has not been fully clarified, in particular regarding intestinal dysbiosis. Dietary β-carotene is poorly bioavailable, the majority of ingested β-carotene transiting to the large intestine for excretion [18, 30, 31]. Thus, the idea that dietary β-carotene can also exert its beneficial effects by perturbing the complex and large microbial community residing in the gastrointestinal tract (intestinal microbiome) is gaining traction [16, 32, 33], as is the case with other carotenoids, including lycopene, astaxanthin and fucoxanthin [34–41].

Here we asked whether, and to what extent, β-carotene dietary intervention can normalize fecal microbial composition and overall improve intestinal health under VAD. We took advantage of a genetic model of VAD, i.e., mice lacking both lecithin-retinol acyltranferase (Lrat) and retinol-binding protein (Rbp or Rbp4), that displays fecal dysbiosis and intestinal dysfunctions associated with their vitamin A deficient status [9]. Unable to store vitamin A due to the lack of LRAT and to mobilize hepatic retinol towards the peripheral tissues due to the lack of RBP [9, 42, 43], the Lrat^−/−Rbp^−/− mice are highly susceptible to develop severe VAD when maintained on a vitamin A-free diet [9, 42, 43]. By using WT and Lrat^−/−Rbp^−/− mice supplemented with β-carotene as the sole dietary source of the vitamin, we showed that β-carotene improves intestinal health and gut microbiome composition depending upon the vitamin A status, with a greater extent under overt VAD.

2. MATERIAL AND METHODS

2.1. Mice, diet and β-carotene supplementation

Wild type (WT) and Lrat^−/−Rbp^−/− mice on a mixed genetic background (C57BL/6J × 129/Sv; [9, 42, 43]) were maintained on a standard vitamin A-sufficient chow diet containing 18 IU vitamin A/g diet (Prolab Isopro RMH3000 5p75). Establishment of the mouse colony and assessment of the genetic background was carried out as described [9]. At six weeks of age, Lrat^−/−Rbp^−/− and WT female mice, raised on the above-described regular chow diet, were placed on the vitamin A deficient diet (Research Diets, VA-def: < 0.02 IU vitamin A/g diet) for four weeks. The macronutrient composition of the regular chow and purified diet was similar (protein 26%, carbohydrate 60%, fat 14% Kcal). At eight weeks of age (after two weeks on the vitamin A deficient diet) mice from both genotypes were randomized to two groups and received either vehicle (corn oil; Veh) or β-carotene in corn oil (50 mg/kg body weight; ~0.9 mg/dose considering an average body weight of ~ 18g), every other day via feeding tube (gavage) between 9 and 10 AM for two weeks. A 10 μg/mL stock solution of β-carotene (Type II, Sigma-Aldrich, St. Louis, MO, USA) was prepared by dissolving crystalline β-carotene with > 98% purity into vehicle (corn oil) by vortexing, and the concentration of the resulting solution was determined using spectrophotometer at 450 nm. The solution was protected from light the entire time. Two-three mice of the same genotype were housed per cage. At ten weeks of age, mice were sacrificed by CO₂ inhalation 4 hours after the last gavage. Serum, liver, adipose tissue (perigonadal), small intestine, colon and colon fecal content were collected, immediately snap frozen on dry ice and stored at −80 °C for further analyses. The day prior to euthanasia, fresh feces were collected, immediately frozen and stored at −80 °C for microbiome analysis. At dissection, the small intestine was divided in three segments of approximately equal length (proximal, medial and distal small intestine). The proximal and medial segments (corresponding to duodenum and most of the jejunum) were opened longitudinally, food content was removed, and the mucosa was gently scraped. The colon was also opened longitudinally to remove feces content, but this segment of the intestine was collected as is, i.e., without scraping the mucosa. Throughout the experiment, mice had constant access to diet and water ad libitum and were housed in a room with a temperature of 24 ± 1°C and a 12:12-h light:dark cycle (7:00 AM - 7:00 PM). Food intake and body weight was measured weekly throughout the experiment. All animal experiments were conducted in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and were approved by the Rutgers University Institutional Committee on Animal Care.

2.2. HPLC analysis of retinoids

Reverse-phase HPLC analyses of retinol and retinyl ester concentrations in serum and tissues were performed as previously described [9].

2.3. RNA extraction and quantitative real-time PCR (QPCR)

RNA extraction and quantitative real-time PCR (QPCR) were performed as described previously [9]. Briefly, total mRNA was extracted from mouse intestine using RNABee according to the manufacturer’s instructions (Tel-test Inc. Friendswood, TX). The concentration and purity of RNA were determined by Nanodrop 2000 Spectrophotometer (ThermoFisher Scientific, Waltham, MA). One microgram of RNA was reverse transcribed to complementary DNA (cDNA) using Verso cDNA synthesis kit according to the manufacturer’s instructions (ThermoFisher Scientific, Dallas, TX). To quantify mRNA, quantitative real-time PCR was performed using an Applied Biosystems QuantStudio 3 Applied Biosystem instrument (ThermoFisher Scientific, Dallas, TX). Primer sequences are: Bco1 – Fwd 5’GAGCAAGTACAACCATTGT3’; Rev 5’AACTCAGACACCACGATTC3’. Srb1 – Fwd 5’TCCCTCATCAAGCAGCAGGT3’; Rev 5’TTCCACATCCCGAAGGACA3’. Isx – Fwd 5’TTCCACTTCACCCATTACCC3’; Rev 5’CTCTTCTCCTGCTTCCTCCA3’; or as previously published [9]. For the QPCR experiments, 300 nM of each specific primer were mixed with 25–50 ng of cDNA equivalent of the total RNA, and 7.5 μL of SYBR Green Master Mix (ThermoFisher Scientific, Dallas, TX) in a total volume of 15 μL. Each sample was run in duplicates. Relative quantification of mRNA expression was calculated using 2^−ΔΔCt method [44], normalized to the TATA-binding protein gene (Tbp – Fwd 5’CAAACCCAGAATTGTTCTCCTT3’; Rev 5’ATGTGGTCTTCCTGAATCCCT3’). Gene expression changes were expressed as mRNA fold change from the WT control group.

2.4. Preparation of hydro-indocyanine green (H-ICG) and in vivo ROS imaging and analysis

H-ICG was prepared from the cyanine dye, indocyanine green, by reduction with NaBH₄ as previously described [9]. Briefly, 1 hour after administration of the last dose of β-carotene or vehicle (according to the dietary/supplementation regimen described above) WT and Lrat^−/−Rbp^−/− mice were gavaged with H-ICG reconstituted in water (0.5 mg/mL) at a dose of 2 mg/kg. One hour after the dye administration, mice were anesthetized with 2% isoflurane and placed in the imaging system. Isoflurane (1–2%) anesthesia was maintained during the imaging procedure. This regimen resulted in the most reliable and reproducible fluorescent images. In vivo ROS imaging and analysis was performed by using a Bruker In-Vivo Multispectral (MS) FX PRO imaging system (Bruker, Ettlingen, German). Fluorescence was quantified as photons/s/mm² using Carestream MI software v5/0.529 (Carestream Health Inc., Rochester, NY). The background intensity of each image was set to zero and identical elliptical regions of interest were drawn on each image (137×145 pixels; interior area = 15605). The mean fluorescence intensity within the ellipse was recorded for each animal.

2.5. Fecal sample collection and DNA extraction

Each mouse was placed in an empty cage without bedding for 10–15 min to allow collection of fresh stool samples that were snap frozen in dry ice and kept at - 80 °C until further processing. Fecal genomic DNA was extracted at the University of Missouri DNA Core facility using PowerFecal kits (Qiagen, Hilden, Germany) according to the manufacturer’s instructions, with the exception that the samples were homogenized in the provided bead tubes using a TissueLyser II (Qiagen, Venlo, Netherlands) for three minutes. DNA yields were quantified via fluorometry (Qubit 2.0, Invitrogen, Carlsbad, CA) using quant-iT BR dsDNA reagent kits (Invitrogen, Carlbad, CA).

2.6. 16S rRNA library construction and sequencing

Library construction and sequencing were performed at the University of Missouri DNA Core facility [45]. Bacterial 16S rRNA amplicons were generated via amplification of the V4 hypervariable region of the 16S rRNA gene using dual-indexed universal primers (U515F/806R) flanked by Illumina standard adapter sequences and the following parameters: 98°C(3:00)+[98°C(0:15)+50°C(0:30)+72°C(0:30)] × 25 cycles +72°C(7:00). PCR was performed in 50 μL reactions containing 100 ng DNA, primers (0.2 μM each), dNTPs (200 μM each), and Phusion high-fidelity DNA polymerase (1U; ThermoFisher Scientific, Waltham, MA). Amplicon pools (5 μL/reaction) were combined, thoroughly mixed, and then purified by addition of Axygen Axyprep MagPCR clean-up beads (ThermoFisher Scientific, Waltham, MA) to an equal volume of 50 μL of amplicons and incubated for 15 minutes at room temperature. Products were then washed multiple times with 80% ethanol and the dried pellet was re-suspended in 32.5 μL elution buffer, incubated for 2 minutes at room temperature, and then placed on the magnetic stand for 5 minutes. The final amplicon pool was evaluated using the Advanced Analytical Fragment Analyzer automated electrophoresis system (Agilent, Santa Carla, CA), quantified using quant-iT HS dsDNA reagent kits (Invitrogen, Carlsbad, CA), and diluted according to Illumina’s standard protocol for sequencing on the MiSeq instrument (Illumina, San Diego, CA), using the V2 chemistry to generate 2×250 bp paired-end reads.

2.7. Informatics analysis

Read merging, clustering, and annotation of DNA sequences was performed at the University of Missouri Informatics Research Core Facility. Paired DNA sequences were merged using FLASH software and removed if found to be far from the expected length of 292 bases after trimming for base quality of 31. Cutadapt (https://github.com/marcelm/cutadapt) was used to remove the primers at both ends of the contig and cull contigs that did not contain both primers. The usearch fastq_filter command (http://drive5.com/usearch/manual/cmd_fastq_filter.html) was used for quality trimming of contigs, rejecting those for which the expected number of errors was greater than 0.5. All contigs were trimmed to 248 bases and shorter contigs were removed. The Qiime [46] 1.9 command split_libraries_fastq.py was used to demultiplex the samples and the command beta_diversity_through_plots.py was used to subsample data to a uniform read count. The outputs for all samples were combined into a single file for clustering. The uparse method (http://www.drive5.com/uparse/) was used to both clusters contigs with 97% identity and remove chimeras. Taxonomy was assigned to selected operational taxonomic units (OTUs) using BLAST [47] against the SILVA database v132 [48] of 16S rRNA gene sequences and taxonomy.

2.8. Statistical analysis

All the data were tested for normal distribution using the Shapiro-Wilk test. When the data were normally distributed, comparisons between genotype and treatment/supplementation (4 groups) were evaluated using a two-way analysis of variance (ANOVA) followed by the Fisher’s least significance difference post hoc test. When only two groups were compared, differences were assessed by t-test. Non-normally distributed data were first log transformed and then analyzed as above. Analyses were performed with SPSS statistical software (IBM SPSS Statistics, version 23: SPSS, Inc.) and p < 0.05 was the cutoff for significance.

3. RESULTS

3.1. β-carotene as the sole dietary source of vitamin A ameliorates retinoid concentrations in serum and tissues of the vitamin A deficient Lrat^−/−Rbp^−/− mice.

To evaluate the effects of dietary β-carotene supplementation on intestinal health during VAD, we fed Lrat^−/−Rbp^−/− and WT mice a vitamin A deficient diet for two weeks followed by oral gavage with β-carotene (50 mg/kg body weight) or vehicle (corn oil; Veh) every other day for another two weeks, prior to sacrifice. No significant changes in food intake (Fig. S1A) and body weight (Fig. S1B) were observed throughout the study among the four experimental groups. With the exception of adipose tissue (concentrations below the limit of detection), intact β-carotene was detected in serum, liver, small intestine (proximal segment) and colon, with no statistically significant differences between genotypes (Fig. 1). Also, regardless of the genotype, the highest concentration of β-carotene was found in feces (colon content), indicating that most of the ingested β-carotene was excreted, as expected [18, 30, 31] (Fig. 1).

Figure 1. Serum and tissue β-carotene concentrations in Lrat⁻/−Rbp⁻/− and WT mice on the vitamin A deficient diet with and without the β-carotene supplementation.

Serum (A), liver (B), small intestine C), colon (D) and feces (E) β-carotene levels were measured by reverse-phase HPLC. Data are mean ± SD with individual data points overlaid; n = 3 – 4 mice/group. Statistical analysis by t-test between the treatment groups; labeled means without a common letter indicate significant difference (p < 0.05) among the groups. BC, β-carotene; ND, not-detectable (i.e., below the HPLC limit of detection of β-carotene: serum < 1 ng/dL, tissues 10 ng/g); Veh, vehicle.

To assess whether dietary β-carotene was properly metabolized, we measured retinol and retinyl ester levels in serum and tissues by HPLC analysis [49]. As expected, due to the lack of RBP [50], serum retinol was significantly reduced in Veh-supplemented Lrat^−/−Rbp^−/− mice compared to WT controls (Fig. 2A). Remarkably, β-carotene administration restored serum retinol concentrations of the mutant mice to WT levels (Fig. 2A), without altering serum retinol or retinyl ester levels in WT mice (Fig. 2A and B). Also, serum retinyl ester levels remained undetectable in the mutant mice despite β-carotene administration, but consistent with the lack of LRAT [51–53] (Fig. 2B). Hepatic retinol and retinyl ester concentrations were significantly lower in the mutant mice compared to WT, regardless of the regimen of supplementation (Fig. 2C and D). Moreover, dietary β-carotene did not alter hepatic retinol levels neither in Lrat^−/−Rbp^−/− nor WT mice relative to their respective Veh-supplemented group (Fig. 2C and D). Carotenoid supplementation, however, significantly elevated retinyl ester levels in the WT liver (Fig. 2D). Adipose tissue is another important storage site of retinoids in mice [23]. Indeed, retinol and retinyl ester concentrations increased significantly in the adipose depot (perigonadal) of Lrat^−/−Rbp^−/− and WT mice, following β-carotene supplementation (Fig. 2E and F). In summary this regimen of dietary β-carotene administration increased serum retinol concentration in the mutants, hepatic retinoids stores in the WT mice and retinoid concentrations in the adipose tissue of both strains.

Figure 2. Serum and tissue retinoid concentrations in Lrat⁻/−Rbp⁻/− and WT mice on the vitamin A deficient diet with and without the β-carotene supplementation.

Serum (A and B), liver (C and D), adipose (E and F), duodenum (G and H) and colon (I and J) retinol (ROH) and retinyl ester (RE) levels were measured by reverse-phase HPLC. Data are mean ± SD with individual data points overlaid; n = 3 – 4 mice/group. Statistical analysis by two-way ANOVA with G (genotype) × T (treatment/supplementation) as factors or by t-test (panel B, D, H and J). Labeled means without a common letter indicate significant difference (p < 0.05) among the groups. In (F), retinyl ester levels in the β-carotene-supplemented Lrat^−/−Rbp^−/− group showed a trend towards significance (p=0.05) compared to the Veh-supplemented Lrat^−/−Rbp^−/− mice. BC, β-carotene; not-detectable (i.e., below the HPLC limit of detection of retinol/retinyl esters: serum < 0.1 ng/dL, tissues < 1 ng/g); Veh, vehicle.

Next, we measured retinoid levels in the gastrointestinal tract. Veh-supplemented Lrat^−/− Rbp^−/− mice showed significantly reduced retinol concentrations in the (proximal) small intestine compared to WT mice, as anticipated [9] (Fig. 2G).

Interestingly, dietary β-carotene significantly increased retinol levels compared to the Veh-supplemented group in both genotypes (Fig. 2G and H), as well as retinyl ester concentrations, but to a much lesser extent in the Lrat^−/−Rbp^−/− mice (Fig. 2H). The retinyl esters detected in the mutant mice upon β-carotene supplementation are likely due to the activity of DGAT 1 [9, 54]. Similar to the small intestine, Veh-supplemented Lrat^−/−Rbp^−/− mice displayed a lower retinol concentration in the colon compared to the WT control group (Fig. 2I). However, in both genotypes, dietary β-carotene supplementation significantly increased colon retinol levels compared to the respective Veh-supplemented group (Fig. 2I). Upon β-carotene supplementation retinyl esters in the colon remained undetectable in the mutants and unchanged in the WT mice compared to Veh-supplemented control mice (Fig. 2J).

As absorption and utilization of dietary β-carotene in the small intestine have been shown to be regulated by SRB1 and BCO1 through a mechanism that involves ISX, a retinoic acid-induced repressor of both genes [21], we measured the expression of these three genes in the proximal segment of the small intestine. In agreement with this feedback regulatory mechanism, the vitamin A deficient Lrat^−/−Rbp^−/− mice prior to carotene supplementation displayed lower expression of Isx, and higher expression of Srb1 and Bco1 compared to the Veh-treated WT controls (Fig. 3). However, upon dietary β-carotene supplementation, when Isx expression increased, Srb1 and Bco1 decreased in both strains (Fig. 3), implying that the β-carotene-derived retinoic acid attenuated both absorption and cleavage of the provitamin A carotenoid in the small intestine.

Figure 3. mRNA expression of Bco1, Srb1, and Isx in the small intestine of WT and Lrat⁻/− Rbp⁻/− mice on the vitamin A deficient diet with and without the β-carotene supplementation.

QPCR analysis of mRNA expression levels of (A) Bco1, (B) Srb1 and (C) Isx in the small intestine (medial segment). Data are mean ± SD, calculated using the 2^−ΔΔCT method, with individual data points overlaid; n = 3 – 4 mice/group. Statistical analysis by two-way ANOVA with G (genotype) × T (treatment/supplementation) as factors. Labeled means without a common letter indicate significant difference (p < 0.05) among the groups. BC, β-carotene; Veh, vehicle.

Altogether, these data indicate that dietary β-carotene was metabolized and affected retinoid concentrations in the gastrointestinal tract of both genotypes.

3.2. β-carotene supplementation alters the taxonomic profile of the fecal microbiome of the vitamin A deficient Lrat^−/−Rbp^−/− mice.

We next asked whether dietary β-carotene supplementation differentially affected the fecal microbiome profile depending on the vitamin A status. We performed high throughput sequencing of bacterial 16S rRNA V4 region by Illumina Miseq platform of fecal samples from the four experimental groups. The rarefaction curves for the four groups of mice reached a plateau (Fig. S2), indicating adequate sequencing depth and species richness in all the samples. β-diversity in the fecal bacteria community was measured by Shannon Index, that takes into consideration both richness and abundance of the bacteria; total amplicon sequence variance (ASV), that measures the richness or the number of the species; and Faith’s phylogenetic diversity (PD) which considers species richness based on the phylogenetic distances in the samples [55]. Both Shannon Index and Faith’s PD were not significantly different among the four experimental groups, regardless of genotype or dietary supplementation (Table 1). Moreover, there was no significant difference in the total number of amplicon sequence variance (ASV) in the β-carotene-supplemented groups compared to the Veh controls of the same genotype (Table 1).

Table 1.

Alpha-diversity indexes of fecal samples of Lrat^−/−Rbp^−/− and WT mice on the vitamin A deficient diet with and without the β-carotene supplementation.

	WT Veh	WT BC	Lrat−/−Rbp−/− Veh	Lraf−/−Rbp−/− BC
Shannon	5.3 ± 0.04	5.5 ± 0.02	5.3 ± 0.02	5.3 ± 0.03
Total ASV	83.3 ± 23.3ab	90.3 ± 23.5a	84.9 ± 20.7ab	81.0 ± 19.9b
Faith’s PD	6.4 ± 2.3	6.5 ± 2.3	6.3 ± 2.3	5.9 ± 2.1

Data are shown as mean ± SD; n = 3–5 mice/group; Statistical analysis by two-way ANOVA. Different letters indicate significant difference (p < 0.05) among groups. ND, not detected; BC, β-carotene; Veh, vehicle

As shown in Fig. 4A and B, the fecal microbial community of both genotypes was mainly composed of Bacteroidetes, Firmicutes, Actinobacteria, and Proteobacteria. Consistent with our earlier characterization of the fecal microbial taxa of the Lrat^−/−Rbp^−/− mice [9], fecal samples of the mutants showed lower abundance of Actinobacteria compared to WT mice (Fig. 4B). Moreover, they also showed a significantly lower abundance of Firmicutes (p < 0.001) (Fig. 4C) but higher abundance of Bacteroidetes (p = 0.002) (Fig. 4D) compared to the WT groups. Thus, the Firmicutes/Bacteroidetes ratio was significantly reduced in the mutant mice irrespective of the supplementation with β-carotene (Fig. 4E). Notably, dietary β-carotene did not alter the fecal bacterial communities at the level of phylum in any genotype, relative to its corresponding Veh-supplemented group (Fig. 4B–E).

Figure 4. Relative abundance of bacterial phyla in fecal samples of WT and Lrat⁻/−Rbp⁻/− mice on the vitamin A deficient diet with and without the β-carotene supplementation.

(A) Relative abundance of bacterial phyla in mouse fecal samples. Total number of the ASVs for (B) Actinobacteria, (C) Firmicutes (D) Bacteroidetes and (E) the Firmicutes/Bacteroidetes ratio. Statistical analysis by two-way ANOVA with G (genotype) × T (treatment) as factors. Different letters indicate significant difference (p < 0.05) among the groups with a main effect of genotype. n = 3 – 5 mice/group. BC, β-carotene; Veh, vehicle.

Principle component analysis (PCA) was performed to define potential differences in community composition, i.e., β-diversity, among the bacterial genera. As shown in Fig. 5A, PC1 accounted for 53.4% of the variation and separated the fecal samples only by genotype. However, unlike WT mice (data not shown), PCA of the fecal microbiome of the Lrat^−/−Rbp^−/− groups also showed distinct clusters depending on the supplementation regimen (Fig. 5B). Specifically, PC1 accounted for 41% of the variation and separated the fecal microbiome of the mutant mice by dietary regimen (Fig. 5B). Only 8 genera displayed a difference in their relative abundance between Veh- and β-carotene-supplemented Lrat^−/−Rbp^−/− mice: Parabacteroides, Alistipes, Millionella, Paraprevotella spp. (Bacteroidetes phylum), Lachnoclostridium and Marvinbryantia spp. (Firmicutes phylum), Brachyspira sp. (Spirochaetes phylum) and Desulfovibrio (Proteobacteria phylum) (Fig. 5C–J). Specifically, exposure to dietary β-carotene significantly lowered the abundance of these genera in the mutant mice, but not in the WT (Fig. 5C–J). Notably, higher taxonomic ranks of these genera were similar between the same two groups. Moreover, compared to Veh-supplemented WT mice, the Lrat^−/−Rbp^−/− mice under the same regimen showed a significantly higher abundance of the above-mentioned genera, except for Parabacteroides and Desulfovibrio (Fig. 5C–J).

Figure 5. Impact of dietary β-carotene on fecal microbial genera of WT and Lrat⁻/−Rbp⁻/− mice.

(A and B) Principal component analysis (PCA) of microbial genera. (A) PC1 separated fecal samples by genotype and accounted for 53.4% of total variance. (B) PC1 separated Lrat^−/−Rbp^−/− fecal samples by the treatment (BC vs. Veh) and accounted for 41.4% of total variance. n = 3 – 5 mice/group. Total number of the ASVs for (C) Paraprevotella sp. (D) Alistipes sp. (E) Millionella sp. (F) Parabacteroides sp. (G) Lachnoclostridium sp. (H) Marvinbryantia sp. (I) Desulfovibrio sp. (J) Brachyspira sp. in the fecal microbiome of WT and Lrat^−/−Rbp^−/− mice. Statistical analysis by two-way ANOVA with G (genotype) × T (treatment) as factors. Labeled means without a common letter indicate significant difference (p < 0.05) among the groups. n = 3 – 5 mice/group. BC, β-carotene; Veh, vehicle.

All together these results support our previous findings [9] that the vitamin A status (sufficient vs. deficient) has a dramatic influence on fecal bacterial diversity. Importantly, they indicate that β-carotene exposure impacts the fecal taxonomic profile only under VAD.

3.4. Dietary β-carotene impacts the physical and functional barrier properties of the colon.

We recently showed that the Lrat^−/−Rbp^−/− mice display abnormalities of the intestinal physical barrier (in the colon) resulting in reduced mucin expression, leaky gut and high ROS luminal concentrations [9]. Given that intestinal bacteria are massed in the colon, we asked whether dietary β-carotene could attenuate the above-mentioned phenotypes in the mutant intestine. We confirmed that the colon of the Veh-supplemented Lrat^−/−Rbp^−/− mice showed the lowest mRNA expression of Muc2 and 3 (Fig. 6A). However, β-carotene availability only modestly increased the expression of these genes, as it did in the WT in the case of Muc3 (Fig. 6A).

Figure 6. Physical and functional intestinal barrier properties in the colon of Lrat⁻/−Rbp⁻/− and WT mice on the vitamin A deficient diet with and without the β-carotene supplementation.

(A) QPCR analysis of mRNA expression levels of mucins (Muc2 and Muc3); Values are mean ± SD, calculated using the 2^−ΔΔCT method with individual data points overlaid; n = 3 individual sample pools/group (3 mice/pool). (B) Serum LPS concentrations measured by a commercial kit (Pierce^™ LAL Chromogenic Endotoxin Quantitation Kit); n = 5 mice/group. (C) Representative overlay of ROS-associated NIRF image and corresponding brightfield image of Veh-treated WT mice (top, left), β-carotene-treated WT mice (top, right), Veh-treated Lrat^−/−Rbp^−/− mice (bottom, left) and β-carotene-treated Lrat^−/−Rbp^−/− mice (bottom, right). Near-infrared fluorescence (NIRF) intensity scale shown on the right. Images were normalized accordingly using Carestream MI software. (D) Quantification of the data in (C). n = 4 – 5 mice/group; values are mean ± SD with individual data points overlaid. (E) QPCR analysis of mRNA expression levels of inflammatory (Il-6, Tnfα and Il1-β) and (F) immunomodulatory cytokines (Il22, Il23 and Il17). Values are mean ± SD calculated, using the 2^−ΔΔCT method, with individual data points overlaid; n = 3 independent sample pools/group (3 animals per pool). Statistical analysis by two-way ANOVA with G (genotype) × T (treatment/supplementation) as factors. In panel E and F, significance of factors and interactions apply to all the genes within the panel. Labeled means without a common letter indicate significant difference (p < 0.05) among the groups. BC, βC-carotene; Veh, vehicle.

To assess intestinal permeability, we measured how much microbial lipopolysaccharide (LPS) transitioned from colon lumen to serum. As expected [9], the Lrat^−/−Rbp^−/− mice had significantly higher concentrations of circulating LPS signifying increased leakiness (Fig. 6B). Importantly, administration of β-carotene reduced leakiness to levels similar to those of WT groups (Fig. 6B).

We next measured intestinal ROS in the four experimental groups of mice by optical in vivo imaging using near-infrared fluorescence (NIRF) light generated by cyanine-based fluorescent dyes. Consistent with our previous observations [9], the Veh-supplemented Lrat^−/−Rbp^−/− mice displayed significantly increased intensity of the fluorescent signal in the intestinal lumen compared to WT mice on the same dietary regimen, indicating elevated ROS levels (Fig. 6C and D). Remarkably, β-carotene supplementation significantly reduced ROS levels in the intestine of the mutants compared to their corresponding Veh control group (Fig. 6C and D). Only a trend towards ROS attenuation was observed in the WT groups (p > 0.05; Fig. 6C and D).

The colon of the vitamin A deficient Lrat^−/−Rbp^−/− mice also showed altered mRNA expression of key inflammatory markers and immuno-modulatory cytokines [9]. Veh-supplemented Lrat^−/−Rbp^−/− mice expressed significantly elevated mRNA levels of the proinflammatory cytokines Il-6, Tnfα and Il-1β, β-carotene administration significantly reduced Il-6, Tnfα and Il-1β expression in the colon of the mutant mice, but it also attenuated the expression of Tnfα and Il-1β in WT mice, compared to their respective Veh-supplemented group (Fig. 6E). Furthermore, confirming our previous findings [9], the expression of Il-22, Il-23 and Il-17 was significantly increased in the colon of the Veh-supplemented vitamin A deficient Lrat^−/−Rbp^−/− mice compared to WT mice on the same dietary regimen (Fig. 6F). β-carotene administration attenuated the expression of the above-mentioned three genes in Lrat^−/−Rbp^−/− mice and of Il-17 and Il-22 in WT mice, compared to the vehicle-treated group of their respective genotype (Fig. 6F).

These findings indicate that dietary β-carotene relieves oxidative stress and improves the physical and functional properties of the epidermal barrier of the colon of the mice under a regimen of dietary vitamin A deprivation but with more pronounced effects in the vitamin A deficient mutant mice.

5. DISCUSSION

Vitamin A deficiency (VAD) remains a public health issue, particularly in developing countries where malnutrition is more prevalent [56, 57]. Even though dietary interventions are effective, subclinical VAD remains an unsolved problem in certain areas of the world [58–60]. Moreover, emerging data suggest that marginal VAD may be more frequent than realized even in industrialized countries, depending upon socioeconomic status, dietary habits and/or race/ethnicity [61–63]. Consumption of foods naturally rich in (or enriched with) β-carotene, along with provitamin A carotenoid supplement, is recommended to normalize vitamin A status and to provide a reliable and sustained dietary safety net for populations at risk [64]. Despite demonstrable beneficial health effects of β-carotene attributed to its retinoid-generating and antioxidant capacities [28], the full impact of β-carotene supplementation during VAD is unclear.

Here we assessed whether β-carotene could ameliorate intestinal disfunctions and dysbiosis linked to VAD by using Lrat^−/−Rbp^−/− mice, a well-characterized mouse model of VAD [9, 42, 43]. When deprived of dietary vitamin A for four weeks Lrat^−/−Rbp^−/− but not WT mice develop severe VAD associated with reduced mucin expression, leaky gut, and high ROS luminal concentrations in the colon as well as fecal dysbiosis [9]. For consistency with our previous study [9], we used Lrat^−/−Rbp^−/− and WT female mice fed a vitamin A deficient diet for two weeks followed by oral gavage of β-carotene at 50 mg/kg body weight (~0.9 mg/dose considering an average body weight of ~ 18g) every other day for two additional weeks. This dose was based on previous studies that investigated the effects of different regimens of β-carotene supplementation on intestinal retinoid homeostasis and functions in various animal models [14, 16, 33, 65]. To our knowledge, this is the first report that also investigated the consequences of β-carotene exposure during VAD on fecal microbiome integrity.

Absorption and utilization of dietary β-carotene in the small intestine have been shown to be regulated by SRB1 and BCO1 via transcriptional control by ISX. The involvement of this retinoic acid-induced repressor of both genes implies dependence on vitamin A status and intestinal retinoid levels [21]. At low retinoic acid concentrations, as in VAD, intestinal Isx expression is attenuated and consequently Srb1 and Bco1 are increased enabling efficient uptake and metabolism of dietary β-carotene. Conversely, under vitamin A sufficiency and/or on provitamin A carotenoid-rich diets the induction of Isx by retinoic acid would prevent vitamin A toxicity [21]. In agreement, maintaining our mice on VAD diet correlated with low Isx expression, concomitantly with higher Srb1 and Bco1 mRNA levels in the small intestine of the Veh-treated mutants compared to WT controls (Fig. 3). Increased Isx together with decreased Srb1 and Bco1 expressions upon dietary β-carotene supplementation (Fig. 3) is also consistent with this feedback mechanism. It is reasonable that retinoic acid production from β-carotene would be attenuated in WT mice that, indeed, are not vitamin A deficient after two weeks of dietary vitamin A deprivation [9]. Surprisingly, however, the β-carotene-induced transcriptional response also occurred in the vitamin A deficient mutant mice (Fig. 3). Based on the notion above, one could have expected enhanced intestinal (small intestine) absorption of β-carotene in the vitamin A deficient Lrat^−/−Rbp^−/− mice compared to the vitamin sufficient WT group [21]. This does not seem to be the case, though, as no differences in β-carotene concentrations were observed in feces (colon content), serum or tissues (including intestine) between the two genotypes (Fig. 1). Ramkumar and colleagues [65] recently proposed that the activity of LRAT - the main enzyme that synthesizes retinyl esters, i.e., retinoid stores, in the small intestine [9, 54] and liver [51, 52] - influences the ISX-mediated negative feedback loop that controls intestinal absorption of β-carotene and its cleavage into retinoids. Thus, it is possible that the absence of Lrat in our VAD model may account for the failure to observe enhanced β-carotene absorption under VAD. Nevertheless, β-carotene was converted into retinoids in both WT and Lrat^−/−Rbp^−/− mice on the vitamin A deficient diet. The β-carotene supplementation dramatically increased serum retinol in the mutants to the same level as that of the WT mice (Fig. 2). In the absence of LRAT, postprandially-derived unesterified retinol can be incorporated into chylomicrons [52]. As mice were sacrificed 4 hours after the last dose of β-carotene, it is likely that the majority of retinol detected in the circulation of the supplemented mutants is associated with postprandial lipoprotein particles. Our data indicate that the supplementation did not change hepatic retinol levels, regardless of the genotype, but increased hepatic retinyl ester concentrations only in the mice expressing Lrat, as expected [51, 52, 54] (Fig. 2). Perhaps, β-carotene-derived retinol was rapidly metabolized into retinoic acid in the liver of the severely vitamin A deficient Lrat^−/−Rbp^−/− mice. This hypothesis however needs confirmation. Given that β-carotene was proposed to be the primary source of retinoids in adipocytes [66], it is not surprising that intact β-carotene was undetectable in the adipose tissue, likely due to its rapid conversion into retinoids. Lobo and colleagues [66] showed that when Lrat^−/− mice were fed a vitamin A deficient diet and gavaged daily with 0.5 mg of β-carotene for 10 days, only trace amounts of the provitamin A carotenoid (~10 pmol/g) were detected in the adipose tissue whereas retinol levels increased. Consistently, in our study the dietary β-carotene supplementation increased retinol levels in the adipose tissue of both genotypes (Fig. 2). Indeed, unlike the liver, the adipose tissue of the Lrat^−/−Rbp^−/− mice can store retinoids [9]. We cannot establish whether adipocytes store β-carotene-derived retinoids that are generated locally or elsewhere in the body (for instance in the liver). In the future, adipose and liver retinoid homeostasis should be assessed by gene expression and/or turnover studies in Lrat^−/−Rbp^−/− and WT mice upon exposure to β-carotene.

Dietary vitamin A is predominantly absorbed in the small intestine (duodenum) and represents most of the retinoids in this segment of the intestine [9]. Thus, we expected that β-carotene administration would enhance retinoid content in the proximal small intestine of both genotypes (Fig. 2), whereas in the absence of LRAT total retinoid (retinol + retinyl esters) concentration in the Lrat^−/−Rbp^−/− small intestine would remain lower compared to WT mice (129 ± 52 ng/g vs. 347 ± 57 ng/g, respectively; p = 0.01; and Fig. 2). β-carotene supplementation also significantly enhanced colonic retinol content in both genotypes, but increased retinyl ester levels only in the mice expressing LRAT (Fig. 2). These findings contrast with our earlier studies [9] that suggested LRAT not be the predominant retinol-esterifying enzyme in the colon, at least while the mice were maintained on a vitamin A-containing diet. We could postulate that the retinol esterification in the colon may be carried out by different enzymes depending upon the vitamin A status and/or the vitamin A content in the diet. This hypothesis and its underlying biological reasons remain to be proven.

We confirmed that VAD in our mouse model correlated with a distinct fecal taxonomic profile, characterized for example by a lower abundance of Actinobacteria (as also shown in [9]) and by a lower F/B ratio compared to the WT controls (Fig. 4). Furthermore, we showed that Alistipes and Paraprevotella genera (Bacteroidetes Phylum) were also more abundant under VAD (Lrat^−/−Rbp^−/− mice on the vitamin A deficient diet) compared to the vitamin A sufficient WT mice, possibly because the intestinal dysfunctions of the VAD intestine (Fig. 6 and [9]) provided a better niche for them to thrive. Indeed, abundance of Alistipes was previously shown to be associated with chronic inflammation and colorectal cancer in human [67, 68] and animal models [69, 70], conditions that have also been linked to higher oxidative stress [71]. Alistipes also positively correlated with the production of IL-6 and TNF-α in the intestine [69, 70]. Similarly, increased abundance of the Paraprevotella bacteria observed in Sod1^−/− mice was attributed to enhanced oxidative stress [72]. More relevant to this study, administration of the carotenoid astaxanthin significantly reduced the relative abundance of the Parabacteroides genus in mice [35].

Importantly, we showed that dietary β-carotene can influence the taxonomic composition of the fecal microbiota by selectively attenuating the abundance of certain fecal bacteria taxa (8 genera) solely in the vitamin A deficient Lrat^−/−Rbp^−/− mice (Fig. 5). These same 8 genera are known to be associated with inflammatory diseases and oxidative stress [35, 39, 69, 70, 72, 73]. Overall, it is reasonable to hypothesize that some of the changes driven by β-carotene in the VAD fecal bacterial communities may be attributable, at least in part, to the antioxidant effects of this carotenoid which travels mainly unabsorbed in the gastrointestinal tract [18, 30, 31]. As β-carotene supplementation did not modulate the abundance of these bacteria taxa in the WT (Fig. 5), we do not assume that β-carotene acts directly on the bacteria. Since β-carotene increased retinoid concentrations in the various segments of the intestine, it is also plausible that its provitamin A activity affected the abundance of certain taxa by repairing the dysfunctional intestinal barrier associated with VAD, for instance by improving the leaky gut phenotype and thus reducing leakage of ROS from the intestinal cells. In support, dietary β-carotene lowered intestinal ROS, reduced LPS levels in the circulation and improved markers of inflammation and immunity (Fig. 6). Trivedi and colleagues [14] showed that β-carotene administration (5, 10 and 20 mg/kg body weight) to WT mice for 28 days attenuated inflammation associated with ulcerative colitis by reducing the colonic levels of IL-17, IL-6 and TNFα proteins in the colon. These authors proposed that the anti-inflammatory action of β-carotene was due to decreased expression of NF-kB, a key regulator of proinflammatory cytokines, in turn linked to the direct effects of ROS on NF-KB signaling [74, 75]. Also, retinoic acid, the cleavage product of β-carotene, acts as an immune modulator [76] which, for example, inhibits the differentiation of naïve T cells to Th17 cells by blocking IL-6, IL-21, and IL-23 production in mice [15]. Furthermore, retinoic acid generated from supplemental β-carotene induced intestinal mucosal IgA production in both mice [16] and chickens [17] which prevented bacteria from translocating to the intestinal mucosa thereby suppressing host inflammation. Our findings indicate that dietary β-carotene attenuated the expression levels of intestinal inflammatory cytokines in VAD and to a certain extent also under vitamin A sufficiency (Fig. 6). Also, β-carotene regulated the expression of immunomodulatory cytokines, such as Il-17 and Il-22, potential drivers of chronic intestinal tract inflammation [4], regardless of the vitamin A status (Fig. 6). We cannot unequivocally ascribe these beneficial effects of β-carotene on gut health to either its provitamin A or antioxidant activity, but most likely to both.

6. CONCLUSIONS

In summary, we demonstrated that β-carotene supplementation does not alter the fecal microbiota composition under conditions of intestinal health, but selectively modifies the abundance of certain fecal microbial taxa and improves the dysbiosis of VAD. β-carotene also ameliorates the intestinal retinoid status and functional dysregulations of VAD. As the VAD-induced intestinal pathobiology is very similar to that observed in other intestinal disorders like the IBD syndrome [1–3], our findings may contribute to developing new therapeutic opportunities not only for VAD but other gut pathologies.

Highlights.

• Vitamin A deficiency (VAD) impairs intestinal health.

• Dietary b-carotene selectively modifies the abundance of certain fecal microbial taxa and improves the dysbiosis of VAD.

• Dietary b-carotene ameliorates the intestinal retinoid status and functional dysregulations of VAD.

• Dietary b-carotene does not alter the gut microbiota composition under conditions of intestinal health.

Data availability

The fecal 16S rRNA gene survey raw data have been deposited in the NCBI’s Sequence Read Archive (SRA) data repository (BioProject ID: PRJNA779384) and can be downloaded without any restrictions. All the remaining data are contained within the article.